American Mineralogist, Volume 78, pages 968-979, 1993

Crystal chemistry of Fe3+ and H+ in mantle kaersutite: Implications formantle metasomatism

M. D.qnnv Dv.q.nDepartment of Geology and Astronomy, West Chester University, West Chester, Pennsylvania, 19383, U.S.A.

STBpHBN J. Ml.crwn'r-rDepartment of Geosciences, The Pennsylvania State University, University Park, Pennsylvania 16802, U.S.A.

AuNn V. McGurnnDepartment of Geosciences, University of Houston, Houston, Texas 77204, U.S.A.

Launa R. Cnoss, J. Dlvro RonenrsoNDepartment of Chemistry and Center for Applied Energy Research, University of Kentucky, kxington, Kentucky 40506, U.S.A.

Ansrru.cr

Chemical and crystal chemical analyses have been performed on a suite of subconti-nental, mantle-derived hornblende (kaersutite) samples. Mdssbauer techniques have beenutilized to investigate Fe valence and site occupancies, U extraction techniques have beenused to determine bulk H contents, proton-induced y-ray emission (PIGE) analysis wasemployed to measure F, and electron microprobe techniques coupled with the abovemeasurements have been utilized to determine major-element contents of hornblende.Similar analyses were performed on a suite of metamorphic amphibole samples fromCosca et al. (1991). Comparison with their wet chemical results on Fe3+/Fe2+ permitteddetermination of C : 1.22, the correction for differential recoil-free fraction effects, whichwas used to correct the mantle sample M0ssbauer data. The results of the analyses for thekaersutite samples show a nearly l:l inverse relationship between the Fe3+ and H+ con-tents. Although the range of Fe3+/H+ in the less oxidized kaersutite samples may beexplained by partial H loss during entrainment and ascent, the nearly total dehydrogena-tion of the Fe3*-rich megacrysts would require time scales significantly longer than whatis expected for transport. Thus, it seems likely that these oxykaersutite samples grew in amore oxidized metasomatic fluid, where incorporation of H was not required for chargecompensation. As megacrysts from the same location show wide variation in Fe3* andH+, it appears likely that significant variations in the oxidation state of the mantle metaso-matic fluid occurred over relatively small temporal or spatial scales.

INrnonucrroN while others lack hydrous phases. Improved understand-

A 1964 paper by Oxburgh was among the first to sug- ing of the role of volatile species in metasomatism and

gest that the Earth's upper mantle must contain td;;- the crystal chemistry of the hydrous phases is needed.

eral amphibole in ordei to provide sufficienr K ";e;il;r

- ̂ l:ttlt^Sineralogical studies bv Dvar et al' (1989'

alkalis to serve as a source region for ururi "or"u"lt.".

1992a' 199.2b) among others including wood' virgo' and

Because the presence of amphibole implies "

,ig"ifl"*t coworkers have focused attention on the presence of Fe3*

volatile component in the mantle, this rugg"rtToi- *i in !o1!

tvpical mantle phases (McGuire et al', 1989; Wood

relaredworkbyvarne(1970)ledronumero"*;l i ;r;" ung virgo, 1989; Dyar et al., 1989; Canil et al. ' 1990)

the influence of volatiles on the melting uetra.,^or orpe- 1na in metasomatized peridotites (McGuire et al'' 1991;

ridotite (e.g., Bailey, lg7}, lg72) and on tfre efects ana Dyar et al., 1992a). The studies on peridotite indicate

characteristicsofmetasomatism(e.g.,Menzies, 1983; Dt"k that contrasting styles of metasomatism are associated

et al., 1984). Field observations (Wilshire and Sh;;;i;, with distinctive sets of Fe3*/Fe,", ratios in component

1975;Harteetal., l975;StoschandSeck, 1980; R"J;; phases; they are also suggestive of an interrelationship

et al., 1984; Wilshire, 1987; Nielson et al., f Sglj h""" between Fer+ and H* in individual minerals'

led to a variety of models for mantle metasomatic pro- The goal of our present research is to examine the crys-

cesses. There is current debate over the mechanismsihat tal chemistry of mantle-derived amphibole by studyingproduce the different styles of metasomatism. For ex- its major-element contents, Fe-site occupancies and va-ample, it is not understood why some styles of metaso- lences, and H* content and partitioning behavior. Datamatism are characterized by the presence of amphibole, on the isotopic behavior of H are presented in brief in

0003-004x/93l09 10-0968$02.00 968

DYAR ET AL.: Fe3+ AND H+ IN MANTLE KAERSUTITE 969

2 0 0

2 5 0

1 .75

0.50

o. 1 .50

$ r.zsB t.oo

1.5 3 .0 4 .5 6 0 7 .5 9 .0 10 .5 12 .0 13 .5 15 .0

Wt% Fe2O3

Fig. 1. FerO. and HrO contents of oxykaersutites and mantleamphiboles. Symbols represent data from Barnes (1930), solidcircles; Aoki (1963), solid squares; Best (1970), solid triangles;Wilshire and Trask (1971), solid diamonds; Aoki (1971), opencircles; Boettcher and O'Neil (1980), open squares; and Grahamet al. (1984), open triangles. The lack ofa closely aligned lineartrend among these variables arises from three main factors: (1)HrO was, in most cases, determined by weight loss with no dis-crimination of possible CO, contribution; (2) FerO, was, in mostcases, detennined by calorimetry, which could not detect con-tributions from Ti and Mn oxidation; and (3) since the truesubstitution here involves oxidation ofFer+ to Fe3+ due to lossof H* ions, the plot of oxide data obscures any evidence of acrystal chemical trend.

Dyar et al. (1992b) and will be pursued, togerher wirh Oisotope data, in a later work (Dyar et al., in preparation).We present here the results of a detailed study of 20 kaer-sutite samples, in order to assess the relationships amongH+, Fe3+, and other major elements within the amphi-bole structure.

B,c,cxcnouNo

Beginning with Barnes (1930) and then Winchell (1945),Aoki (1963, 1970; l97l), Best (1970, 1974), and Ernsrand Wai (1970), geochemists have used atomic absorp-tion and wet chemical analyses to study the dehydrationof amphibole. Incomplete characteization of minor con-stituents such as Cr3* and Tia+ and suspicions that Ti3+might contribute to errors in Fe3+ determinations ham-pered these workers in evaluating the detailed crystalchemistry of their samples. Furthermore, HrO analyseswere generally determined by weight loss, a techniquethat yielded results greatly biased by the contents ofothervolatile phases, especially COr, which are now knownfrom fluid inclusion studies to be abundant in mineralsderived from the upper mantle (see e.g., Roedder, 1965;Bergman, 198 1; Andersen et al., 1984).

More recent geochemical and petrologic studies of am-phibole have focused more closely on major-elementanalyses coupled with studies ofH isotope behavior. Ku-roda et al. (1978) examined 6D ofhornblende from intru-sions in alkali basalts; Boettcher and O'Neil (1980) fol-lowed with a thorough examination of DD, HrO content,and D'8O in amphibole; and Graham et al. (1984) derivedactivation energies for H isotope exchange. Matson et al.

Frg.2. Schematic drawing ofthe octahedral sites adjacent tothe H* atoms in the hornblende structure. O sites are indicatedby large outlined circles and identified by their numbers; cationsites are small, solid circles. The H atom sits close to the 03 sitewith an O-H distance of 0.934 A lPechar et al., 1989), and sothe OH group takes part in the coordination polyhedra of bothMl and M3.

(1984) examined the major-element and volatile com-ponents of amphibole in xenoliths from the Grand Can-yon and found a strong positive correlation between Ticontents and H+ deficiency. An intriguing study (Brynd-zia et al., 1990) investigated the relationship among fo,,OH- content, DD, and Fe3+/Fe,., in samples from Shep-pard and Epstein (1970) and Kuroda et al. (1978) andfound only small ranges of dD and Fe3+/Fe,o,; however,no primary data were given. Recent work by Deloule andcoworkers (e.g., Deloule et al., 1990, l99l) has concen-trated solely on the isotopic variation displayed by horn-blende. Unfortunately, none of these more recent studiesincluded complete (published) Fe3* data on all samples,and so other workers (e.g., Boettcher and O'Neil, 1980)could do little more than note a general inverse correla-tion between FerOr/FeO and HrO content (Fig. l).

Many crystallographic studies of the kaersutite struc-ture using X-ray diffraction have been published, includ-ing Papike and Clark (1968), Papike et al. (1969), Kita-mura and Tokonami (1971), and Hawthorne and Grundy(1973). Powder (Jirdk et al., 1986) and single-crystal (Pe-char et al., 1989) neutron diffraction studies of this min-eral have further improved our understanding ofthe cat-ion distributions and H positions in kaersutite (Table l,Fig. 2).

Additional relevant X-ray data complemented by ex-perimental work has been performed recently by Popp etal. (1990) at Texas A&M University (Phillips et a1., 1988,1989; Clowe et al., 1988). Results showed that dehydro-genation accompanying heat treatment causes preferen-tial ordering oftrivalent cations at the cis-Ml and trans-M3 sites. Some trivalent cations in M2 may relocate toMl or M3 as oxidation occurs. Fe3+-H+ oxysubstitutionand Al substitution were identified as the primary mech-anisms accompanying oxidation and dehydrogenation.Popp et al. (1990) surveyed the literature on Fe3+-rich

970 DYAR ET AL.: FC3* AND H* IN MANTLE KAERSUTITE

Tmle 1. Cation distributions in kaersutite based on structure refinements

Site Hawthorne and Grundy (1973) Pechar et al. (1989) Phillips et al. (1989)

AT1r2M 1M2M3M4H

Na, + Ky and Nao543 , and Ko4s y'0.53Si + 0.47A10.94Si + 0.06A10.343Fe3* + 0.657M90.056Fe3* + 0.502M9 + 0.185At + O.257Ti0.235Fe3* + 0.765M90.009Mn + 0.958Ca + 0.010Fe3* + 0.025Mo

0.265Na + 0.205K0.57Si + 0.43A10.90si + 0.10A10-26Ti + 0.03ca + 0.32FeF* + 0.39M90.17A1 + 0.18Fe3* + 0.65M90.93M9 + 0.01Fe'+ + 0.06Fe3*0.99Ca + 0.01Mn0.51H

0.43Na0.35Ar + 0.65Si0.02Ar + 0.98si0.41sFe + 0.585M90.214Fe + 0.321M90.588Fe + 0.412M90.91SCa + 0.099Ca1.0H

' The variabfes x and y are related by the equation x + 1.8y : 0.802.

calcic and subcalcic amphibole and found that the rela-tion between the Fe3+-oxyamphibole component and theOH- content of amphibole could not be quantified, al-though the lack of large numbers of high-quality amphi-bole analyses greatly limited their conclusions.

Numerous Mdssbauer studies of amphibole samplesexist in the literature, as summarized by Hawthorne(1983). As he noted, the interpretation ofspectra ofcalcicamphibole minerals, in particular, is "not straightfor-ward." Most of the controversy revolves around properinterpretation ofthe doublets resolved in the spectra rel-ative to their assignment to the known crystallographicsites in the structure. In principle, four doublets corre-sponding to Fe2* in each of the metal sites of the structuremight be observed in a spectrum of Fe'?*-rich amphibole.Additional doublets corresponding to Fe3+ in those sitesare also possible, although Fe3* in the M4 site would beunlikely except in Ca or Na-deficient samples. Goodmanand Wilson (1976), in fact, resolved five doublets in asingle hornblende spectrum, which they assigned to Fe2*in Ml-M4 and to Fel;f . Other workers have observedsmaller numbers of doublets (Burns et al., 1970; Burnsand Greaves, l97l; Bancroft and Brown, 197 5; Goldmanand Rossman, 1977) on tremolite-ferroactinolite solu-tions. The question of Fe2+ occupancy in M4 remainscontroversial (e.g., Aldridge et al., 1982); it is apparentthat Fe2+ will occupy the M4 site only in the rare caseswhere Ca2+ + Na+ * K* is insufficient to fill the site.Occupancy of Fe3+ in M4 seems, therefore, extremelyunlikely. One additional conclusion from these previousstudies is germane to the discussion at hand: the distor-tions of the M sites in calcic amphibole, which are knownto decrease in the order M4, most distorted, to M2 toM3 to Ml, least distorted, can be related to the quadruplesplittings observed for Fe2* in the order M4, lowest quad-ruple splitting (A), to M2 to M3 to Ml (Hawthorne, 1981).No Mdssbauer studies of Fe3*-rich kaersutite samplescould be found in the literature, with the exception ofSchwartz and Irving (1978) and McGuire et al. (1989);both of these focused more on the amount of Fe3+ thanon its distribution. However, on the basis of the increas-ing distortion and decreasing A relationship observed byHawthorne (1978, 1981), it would be possible to predictrelative quadrupole splittings of Fe3+ doublets if they couldbe resolved (i.e., in the spectrum of a Fer*-rich amphi-bole). As noted by Burns et al. (1985), although high A

implies low distortion for Fe2* in silicates, high A implieshigh distortion for Fe3+. Therefore we would predict thehighest A for Fe3+ in the M2 site (assuming no Fe3+ inM4) and gradually decreasing A for Fe3+ in the order M3to Ml, which can be used in site assignments for theobserved doublets.

One final comment regarding the interpretation ofMossbauer spectra of amphibole is important within thecontext of the literature. It is well known that the area ofa M6ssbauer doublet (pair of peaks) is a function of peakwidth t, sample saturation G(x), and the Mdssbauer re-coil-free fraction / Bancroft (1973) used the followingformulation for area ratios:

Ar" , , _ ^N(Fe3*)Ar,,:

: t,'ttt1n ,

where N is the amount of each species found by wetchemical analyses. It follows that C may be expressed as

Ie. , - * . f re,* G(x)o. , -c :

*.- * 1",-r C1x1..*'

Bancroft and Brown (1975) determined that saturationcorrections are unnecessary for samples mixed with sug-ar. The Ip.:. irlld fp.:. were also determined to be equalby them. However, when the Miissbauer data on horn-blende reponed by Bancroft and Brown (1975) were com-pared with wet chemical Fe3+/Fe2+ ratios determined byDodge et al. (1968), a C value of l.l3 + 0.02 was derived(on the basis of seven samples). A later study reported Cvalues of 1.36, 1.25, and 1.24 for a hastingite and twokaersutite samples, respectively (Whipple, I 972).

More recently, four samples studied by Schwartz andIrving (1978) yielded Fe3+/Fe2+ values that were higherwhen determined by chemical means than by Mdssbauerspectroscopy (resulting in a C value of < l). They attrib-uted the difference to the presence of Ti3" in their sam-ples. The only other comparison of chemical and Mdss-bauer Fe3*/Fe2+ ratios was performed by Bahgat andFayek (1982), who derived a C value of l.l5 based onone sample. One of the goals of this work was to constrainthis relationship and determine a realistic value for C.

S.lvrpr,r sELEcrroN, pR-EpARATroN, AND ANALvSES

Hornblende megacrysts were selected for this study fromthe extensive collections of Howard Wilshire and Robert

DYAR ET AL.: FC]+ AND H* IN MANTLE KAERSUTITE 971

Coleman of the U.S. Geological Survey. The Colemancollection has since been curated by the National Muse-um of Natural History, Smithsonian Institution. Sourcelocalities for all samples studied are listed in Table 2; allspecimens came from basalt hosts erupted in continentalrift environments. Small pieces for thin sections were re-moved from each megacryst. Additional pieces were bro-ken off and ground by hand under acetone (to preventoxidation) into fragments < I mm to facilitate hand-pick-ing of grains. Only pristine hornblende grains were usedfor this study; any grains containing visible alteration,inclusions, or impurities were rejected. Approximately 680mg of separates were prepared: 150 mg for Mdssbaueranalyses, 100 mg for PIGE, 30 mg for O isotope study,and 400 mg for H+ extraction. In addition, 300-mg ali-quots of metamorphic amphibole samples (Cosca et al.,1991) were obtained from Michael Cosca for the com-parative study of C.

Electron microprobe analyses were performed on theJeol 733 electron microprobe at Southern Methodist Uni-versity, using a Krisel automation package. Routine an-alytical conditions were used: l5-keV acceleration volt-age, 20-nA beam current measured on a Faraday cup,30-s count times, and the Bence-Albee correction routinesupplied by Krisel. At least five points were analyzed oneach megacryst, and only unzoned samples were used.Compositions for AK-Ml, AK-M2, and H366A are de-scribed in McGuire et al. (1989); the analysis used forH366-92 was made using XRFtechniques at WashingtonState University. Analytical errors are +0.5-2o/o for ma-jor elements and + l0-200/o for minor elements.

Mdssbauer analyses were performed in the MineralSpectroscopy Laboratory at the University of Oregon. Asource of 50-20 mCi 57Co in Pd was used on an AustinScience Associates constant acceleration spectrometer.Results were calibrated against an a-Fe foil of 6 pm thick-ness and 990/o purity. Spectra were fitted using a versionof the program Stone modified to run on IBM and com-patible personal computers. The program uses a nonlin-ear regression procedure with a facility for constrainingany set of parameters or linear combination of parame-ters. Lorentzian line shapes were used for resolving peaks,as there was no statisticaljustification for the addition ofa Gaussian component to the peak shapes. Fitting pro-cedures in general followed those described in Dyar et al.(1989) and McGuire et al. (1989). A statistical best fitwas obtained for each model for each spectrum using they2 and Misfit parameters; practical application of theseparameters is discussed elsewhere (Dyar, 1984). Errorsare estimated at +3o/o for doublet areas and +0.02 mm/sfor peak width, isomer shift, and quadrupole splitting.

H contents were determined by means of a procedurefor collecting and measuring all the structural H using avolumetric measurement of HrO vapor extracted fromsilicates (see Bigeleisen et al., 1952; and Holdaway et al.,1986, for details of the technique). In general, clean min-eral separates are degassed under vacuum for at least 8 hat 50-85 oC to evaporate adsorbed atmospheric moisture.

TABLE 2, Hornblende localities represented

DL-s Deadman Lake, Califomia, Hill2237DL-7 Deadman Lake, Califomia, Hill2237DL-g Deadman Lake, Califomia, Hil2284AK-M1 Harrat al Kishb, Saudi Arabia, Coleman locality H271AK-M2 Harrat al Kishb, Saudi Arabia, Coleman locality H271AK-M3 Harrat al Kishb, Saudi Arabia, Coleman locality H271AK-M4 Harrat al Kishb, Saudi Arabia, Coleman localiry H271AK-Ms Harrat al Kishb, Saudi Arabia, Coleman locality H274Ba-s Dish Hill. CaliforniaTm Easy Chair Crater, Lunar Crater, NevadaFr-11 Massif Central, France, just east of Alleyras basalt flowFr-12 Massif Central, France, Montgros, trom oxidized vent agglom-

erate84-BR Dreiser weiher, Germany, Briick-Raders cone, cognateSABM Bullenmeni maar, Victoria, Australia86-LEH' Lake Eacham maar. Queensland, AustraliaH366A Harrat Hutaymah, Saudi ArabiaH366-92 Harrat Hutaymah, Saudi ArabiaS.C. San Carlos, Arizona, USNM 111312.0023K.N.Z. Kakanui, New Zealand, USNM 109647/50Eifel Eifel Volcanic Field, USNM 114856.0006

Note: although the sample numbers listed here may resemble numbersgiven in previous studies, such as Boettcher and O'Neil (1980), they bearno relation to previously studied samples except for locality.

' Sample 86-LEH consists of amphibole grains, rather than a megacryst,separated from a spinel peridotite xenolith found at this locality.

Samples are then fused in an induction furnace to liberatestructural HrO. Distillation processes involving transferof evolved gases through a series of traps employing liq-uid N, and a slush of methanol and dry ice are used toseparate HrO molecules effectively from other condens-able and noncondensable gases. HrO vapor is then passedover hot U (>750 "C) to liberate free H+. A Hg-pistonToepler pump is used to collect vapor in a volumetricallycalibrated reservoir for yield measurement. For this study,every fifth sample was analyzed twice; distilled HrO wasused for calibration. All data are corrected for a sampleblank of 2.5 tor' (roughly 0.5-2o/o of the total wto/o HrO,depending on sample size and volume of HrO vapor).Errors are estimated at =0.1 wto/o HrO based on repli-cations.

F concentrations were determined by external-beamproton-induced 7-ray emission analysis (PIGE) at theUniversity of Kentucky Van de Graaf Accelerator Lab-oratory. The inelastic proton scattering reactions on reF

at proton bombarding eneryies of 3.5 and 2.5 MeV wereutilized. Details of the method are included in Wong etal. (1992). Minimum detection limits of 5 ppm with stan-dard deviations of20-140 ppm were obtained.

Rnsur,rsMiissbauer data

Interpretation of the Mdssbauer data measured for thisstudy proves extremely difficult because ofthe wide rangeof the Fe3+ contents of the samples (from 100 to 27o/oFe3+ ). Both Fe3+ and Fe2+ can occupy (in theory) any ofthe four M sites in the structure, although Fe3+ in M4would be unlikely in this composition range. This wouldgive rise to a total of eight possible doublets (realistically,seven) in each spectrum. In practice, it is impossible toresolve those doublets in a statistically meaningful fash-

972

TABLE 3. Mdssbauer results for six peak fits

DYAR ET AL.: Fe3* AND H+ IN MANTLE KAERSUTITE

Fe3 *

Sample os . r.s. o.s.DL-s 0.38DL-7 0.42DL-g 0.39

0.40AK-M1 0.38AK-M2 0.38AK-M3 0.38AK-M4 0 38AK-Ms 0.38Ba-S 0.38Tm 0.41Fr- l1 0.41Fr-12 0.38

0.400.40

84-BR 0.3886-BM O.3786-LEH 0.40H366A 0.39H366-92 0.39

0.40San Carlos 0.43Kakanui O.37Eifel 0.37Average 0.39s.D. 0.o2ssA-s 0.35HL862C 0.36MR-865A 0.36MrN 864 0.36ssA-10 0.35FA86-1 0.3sH18611 0.34sAS-13 0.37

0.78 0 54 380.69 0 49 461.40 0.s1 430.85 0.43 56o.74 0 49 320.75 0.49 290.73 0.50 32o.71 0.s4 330.74 0 54 410.81 0 55 400.94 0 57 800.90 0.57 761.56 0.44 231 .16 0.46 530 70 0.34 240.84 0.56 490.76 0.55 270.75 0.53 310.79 0.45 281.31 0.59 500.81 0.42 430.63 0.61 370.72 0.57 330.73 0.55 460.87 0.51 41.30.24 0.06 14.20.79 0.35 330.77 0.37 380.76 0.34 250.77 0.37 380.79 0.38 28o.78 0.37 210.81 0.36 360.76 0.41 38

1 .13 2 .561 .12 2 .56

1 .13 2 .521 .13 2 .511 .13 2 .511 .13 2 .481 .14 2 .541. ' t2 2.571.03 2.701.07 2.59

1 . 1 21 . 1 31 . 1 31 . 1 4

1.091 . ' t 21 . 1 31 . 1 21 .060.251.13 2 .701 .13 2.691.13 2 .681.13 2 .691.14 2 .671 .13 2.591.13 2 .701.13 2 .62

0.33 300.34 330.35 330.3s 350.30 230.37 300.27 40.34 12

0.31 24.40.08 10.7o.28 330.29 360.30 41o.29 36o.27 320.31 400.26 290.30 34

1.12 2 .051.08 2:t3

1.06 2.381.12 2 .011 .13 1 .991.12 2 .001 . 1 2 1 . 9 61.12 2 .021. ' t2 2.041.12 2 .061 .10 2.05

1 . 1 0 2 . 1 31.13 2 .O41 . 1 3 1 . 9 91 .15 2 .08

1 .1 1 1 .921.08 1.901.12 2 .091.11 2 .O51 11 2 .050.02 0.101 . 1 6 2 . 1 81 1 7 2 . 1 31 .15 2 .061 1 7 2 . 1 31 1 4 2 . 1 51.13 2 .O71 17 2.201.13 2 .O9

0.30 10.40 380.38 380.39 350.37 320.40 360.41 300.38 160.34 12

0.400.43o.420.45

0230 5 20 4 20.400.39 28.90 06 12.10.43 340.44 26046 34o.44 260.42 400.41 390.40 3s0.40 28

0.03 0.010.04 0.010.00 0.000.12 0.010.13 0.010.19 0.020.01 0.01

-0.01 -0.010.01 0.00

0.000.130.030.090.06

0.00o.020.030.02

0 01 0.010.03 0.010.01 0.010 03 0.01o.02 0.010.02 0.010 01 0.010.04 0.02

0.06 0.010.08 0.01

0.41 320.41 29

0.33 290.32 25

621735

561591507812812819531531532

5 1 5933566o t t

585

524527550525

525537521536522528520546

2.622.s62 5 62.59

2.402.402.602.562.410.57

0.34 290.33 330.35 320.30 32

o.27 40.40 240.33 310.31 24

22403744

J

393630

0.000.010.010.010.01

0.000.010.010.01

Nofe. l.S. : isomer shift, in millimeters per second. Values given are t0 02 mm/s (Dyar, 1984) O.S. : quadrupole splitting, in millimeters per second.Values given are +0.02 mm/s (Dyar, 1984). W : peak width, in millimeters per second. Values given are +0.02 mm/s (Dyar, 1984). A : peak area, inpercent of total area. Values given are +27o of the total area (Dyar, 1984). Other abbreviations: o7"y;s : percent of Misfit (Ruby, 1973); o/.Un : percentof uncertainty of fit (Ruby, 1973); x2: chi squared; s.D. : standard deviation. Mdssbauer parameters are referenced to Fe metal foil calibration.

ion because of their heavily overlapping peak positions.As noted above, previous workers have resolved doubletsassigned to as many as four Fe2+ sites and one Fe3+ site.Such assignments are difficult to determine in our mantlekaersutite suite because none of these hornblende sam-ples has a stoichiometric amount of H+. Some of thecations in the Ml and M3 sites will not have their ex-pected geometries, with bonding to four O atoms and twoOH molecules; instead, they may be bonded to five Oatoms and only one OH molecule or to six O atoms andno OH molecules. Because the M6ssbauer measurementsare sensitive primarily to coordination environment andcharge density, the Mtissbauer spectrum of a H+-deficienthornblende might theoretically yield doublets corre-sponding to Fe atoms in up to eight types of polyhedra:M2,M4, Ml, and M3 with stoichiometric H+, Ml andM3 with one H*, and Ml and M3 with no H+. EitherFe3+ or Fe2+ may occupy any of those sites, giving rise(theoretically) to as many as 16 doublets in the spectrumof even a slightly H+-deficient hornblende. In practice,we were able to resolve no more than seven doublets inany spectnrm because ofthe heavy overlap between peaks;

even these fits are questionable because the peak sepa-ration is less than that required (t/z peak width minimum)for well-resolved peaks (Bancroft, 1973). The assignmentof such doublets is complicated further by the fact thatthe multiplicity of types of environments (as discrimi-nated by the Miissbauer efect) no longer corresponds tothe known site multiplicity of MI:M2:M3:M4 equal to2 ' .2 : l :2 .

In light of these uncertainties in interpretation of mul-tiple doublet fits, the strategy used in the present studywas to fit six peaks initially (a three doublet fit) to eachspectrum. We were able to resolve three doublets in everyspectnrm, as shown in Table 3 and Figure 3. This methodresulted in low values of 12, percent Misfit, and percentuncertainty for each spectrum and yielded a relativelyconsistent set of peak parameters, as shown by the smallstandard deviations in Table 3. The values of percentFe3+ that resulted are thus a valid representation of thetotal percent ofFe3* contributions to the spectra. It shouldalso be noted that the number of doublets fitted to modelFe3+ does not affect the final percent ofFe3+ determined.For these spectra, even six doublet fits yielded exactly the

DYAR ET AL.: Fe3+ AND H" IN MANTLE KAERSUTITE

TaeLE 4. Results of comparative study of amphiboles

U.U. S .M.U. U.U U.O.Sample Wt% HrO Wt% H,O o/oFe3+ ohFe3+ C

9',13

1 0 0 2 0

1 0 0 0 0

9 9 8 0

9 9 6 0

9 9 4 0

99 20

9 9 0 0

98 80

98 60

9 8 4 0

98 20

s 8 0 0

DJA-C

HL862C

MR-865A

MrN 864SSA-10FA86-1HL861 1SAS-13Average

1 .681 .88

1 .78

1.63't.922.261.681.621.731 4 2' '.95

1 .38

29 3335 38

22 25

33 3823 281 8 2 131 3635 38

1.2101 . 135

1 .178

1.2641 3141 2361.2641 . 143

1 .218 + 0 .059

o

F

GF

1 8 01 .751 .231 .641 . 4 1

Fig. 3. Mtissbauer spectrum of sample Ba-5, fitted with athree doublet model. Two doublets corresponding to r6rFe2+ andone doublet corresponding to I6lFe3+ are shown. A three doubletmodel was used to fit all spectra in this study; results are givenin Table 3 and used in the formula recalculations in Table 4.

same percent ofFe3* total peak area as the three doubletfits; i.e., the percent Fe3* determined by M6ssbauer ismodel-independent for these samples.

However, the three doublet model does not allow ac-curate interpretation of the location of the Fe atoms inthe structure because the area contained in each doubletmay represent contributions from Fe in more than onesite (because of the similarity between an Ml or M3 sitein an anhydrous amphibole and a normal M2 site, asdiscussed above). This conclusion is supported by thepeak widths given in Table 3, which are much larger thanthe line widths typically observed for Fe in single sites insilicates: 0.27-0.28 mm/s for Fe2* and 0.28-0.35 mm/sfor Fe3* (Bancroft, 1973). Only in the case of sample Fr-12 were we able to resolve three unique doublets corre-sponding to three sites (for Fe3+ ). We interpret the dou-blets, which have 6 : 0.38, 0.40, and 0.40 mm/s and A: 1.56, 1.16, and 0.70 mm/s, to represent Fe3+ occupan-cy in M2, M3, and Ml sites, respectively, as explainedearlier. For other samples, four, five, and six doubletmodels were also fitted to each spectrum as far as possi-ble. However, the heavy peak overlap that results fromsuch fits renders them mathematically possible but crys-tal-chemically unbelievable.

In order to evaluate the true Fe3+ contents of our man-tle kaersutite samples, it was necessary to establish a vi-able value for C the correction factor. Eight samples froma metamorphic suite analyzed by Cosca et al. (1991) werefitted with six peak fits, as shown in the bottom portionof Table 3. Direct comparison of the results is shown inTable 4. An average correction factor, or C value, of 1.22+ 0.06 was obtained. Previous work by Bancroft andBrown (1975) and Bahgat and Fayek (1982), who derivedvalues of l. l3 + 0.02, and 1.15, respectively, may have

Nofe; analyses performed at University of Utah (U.U.), Southern Meth-odist University (S M.U.), and University of Oregon (U.O.)

* Insufficient sample was available for analysis

been biased by lack of good H analyses for their samples.However, our value of 1.22 agrees well with the two kaer-sutite C values determined by Whipple (1972): 1.25 and1.24. Therefore, we have applied our correction factor of1 .22 to all the Mdssbauer results presented in this paper,and corrected values for percent of Fe3+ have been usedin Table 5 and in related figures.

fI+ contents

H analyses were obtained for seven ofthe eight samplesprovided from the Cosca et al. (1991) study of metamor-phic samples; results are shown in Table 4. The largestdifference between the two laboratories is the 0.33 wo/oHrO difference for sample SSA-10. The remaining anal-yses agree well within the commonly stated error bar of+0. I wto/o HrO. The results suggest that there is no sub-stantive diference between the yields obtained using theapparatus at Utah (as used by Cosca et al., l99l) and theline used at S.M.U.

F contents

PIGE analyses of the kaersutite samples showed onlyminor variations in F contents, ranging from a low of 167ppm for the San Carlos sample to a high of 3054 ppmfor the Eifel specimen. In contrast with conclusionsreached by Kyser (1992) on a suite of kaersutite samples,we find that halogen variation alone cannot explain therange of observed H* contents in our sample suite.

Chemical formulae

Chemical compositions for all samples and for the av-erage of all samples studied were determined from probe,Mtissbauer, PIGE, and H-extraction data and are givenin Table 5. The Fe and FerO, contents were calculatedfrom the probe data using the corrected Mtissbauer mea-surements from the six peak fits given in Table 3 and theC value of 1.22. Data were normalized to formula unitswith 24 O atoms and adiusted to account for the contri-butions of F and Cl.

974 DYAR ET AL.: FEsf AND H] IN MANTLE KAERSUTITE

Tlel-e 5. Hornblende compositions

DL.5 DL.7 DL-g AK-M1 AK-M2 AK-M3 AK-M4 AK-Ms Ba-s t m Fr-11 Ft-12

sio, 38.91Al,o3 14.64Tio, 4.84FeO 7.92Fe.O. 4.43MgO 12.16MnO 0.17CaO 11.02NarO 2.61K.O 1.66Cr,O" 0.03F 0 .14cr 0.01H,O 1.05

Total 99.59

39.85 39.01 39.73 39.9314.39 14.26 14.85 14.834.51 4.52 5.21 5.318.17 0.14 8.20 8.276.36 13.27 3.53 3.07

11 .26 12 31 12.11 12.84o.20 015 0.08 0.09

10.82 1071 9.98 9.872.74 2.61 2.86 2.911.50 1 .64 1 14 1 .100.00 0.02 0.03 0-o2o.17 0 .20 0 .15 0 .170.02 0.02 0.00 0.000.90 0.16 1 .01 1 .03

100.89 99.02 98.88 99.44

5.89 5.85 5.92 5.902.51 2.52 2.61 2.580.50 0.51 0.58 0.59103 0.21 1.02 ' t .020 69 1.31 0.40 0.342.48 2.75 2.69 2.830.03 0.02 0.01 0.011.71 1 .72 1 .59 1.560.79 0.76 0.83 0.830.28 0.31 0.22 0.210.00 0 00 0.00 0.000.08 0.09 0.07 0.080.01 0.01 0.00 0.000.89 0.16 1 .00 1 .02

4028 40.01 39.2014 82 15.20 14.115.32 4.98 5.148.67 8.83 7.953.73 3.97 5.04

12.22 12 01 1 1 .680.13 0 .12 0 .119.79 9.39 9 373.30 3.37 3.101.24 0 .92 1 .120.04 0.00 0.000.16 0 .12 0 .130.00 0.00 0.010.96 0.92 0.84

100.66 99.84 97.80Cations to 24 O atoms minus F and Cl atoms

40.05 39.82 39 2914.48 14.53 13.854.49 4.30 3.273.58 0.00 5.88

10.34 13.95 5.1511.92 12.12 13.770.12 014 0 .10

10.73 10.82 12.102.73 277 1 .961.35 1 .38 2 .18o.02 0.02 0.020.16 0 .11 0 .220.00 0.00 0.000.48 0 .17 1 .32

r 00.45 100.13 99.1 1

5.92 5.96 5.842.52 2.57 2.430.50 0.48 0.370.54 0.00 0.751 06 1.37 0.562.63 2.71 3.050.02 0.02 0.011.70 1.74 1 .930.78 0.80 0.570.25 0.26 0.410.00 0.00 0.000.08 0.05 0.100.00 0.00 0.000.47 0 .17 1 .31

40.41 39.5214.60 14.034.52 5.876.89 2.844.37 10.40

13 .07 11 . 710 .14 0 .17

1 0 . 3 6 1 1 . 1 02.66 2.811 .55 1 .290.01 0.000.16 0.240.00 0.001.03 0.07

99.77 100.05

SiAITiFe,*Fe3*MgMn

NaK

FclH

5.812.580.540.990.492.710 0 21.760.760.320.000.070.001.05

5.92 5.93 5 942.57 2.65 2.520.59 0.55 0.591.06 1 .09 1 .020.41 0.44 0.572 68 2.65 2.640.02 0.02 0.011.54 1 .49 1 .520.94 0.97 0.910.23 017 0.220 00 0.00 0.060 07 0.06 0.060.00 0.00 0.000.94 0.91 0.85

5.95 5 882.54 2.460.50 0.660.86 0.460 48 1.062.87 2.600.02 0.021 .64 1.770.76 0.81o.29 0.240.00 0.000.08 0.110.00 0.001.01 0.06

DrscussroN

Measured vs. calculated Fe3+ and H+

Calculations of amphibole Fe3+ contents based on var-ious charge balance arguments abound in the literature(e.9., Stout, 197 2; Paplke et al., 197 4). Schumacher ( I 99 I )discussed the effects of such empirical corrections on am-phibole geothermometry, and Hawthorne (1983) thor-oughly covered the many reasons why there is "(unfor-tunately) no significant correlation" between Fe3+ valuesthus calculated and those that have been measured. Thisconclusion is also true for our data, as shown in Figure4a, where Fe3+ is calculated solely on the basis ofchargebalance.

Clowe et al. (1988) described a method for calculatingthe oxyamphibole content of amphibole for which bothelectron microprobe and Fe3*/Fe,o, data are available (intheir case, the Fe3+/Fe,". values were determined by cal-orimetry). The process involves correction of an analysisbased on 23 O atoms by recalculating the probe-deter-mined total elemental wto/o Fe to elemental Fe3+ and Fe2+,followed by conversion to weight percent oxide and molesof Fe3+ and Fe2+, and renormalization to 23 O atoms +YzFe3+ pfu. They note that this method makes the as-sumption that Fe3+ is present as an oxyamphibole com-ponent, with one Fe3* for every H* lost.

Their method was tested on the samples in the presentdata set using the Fe3+/Fe,o, values determined by Miiss-bauer and corrected for recoil-free fraction; the results areshown in Figure 4b. Although this figure shows some gen-

eral agreement between the H+ calculated in this way andthe measured H+, the calculated values are consistentlytoo high. In addition to the problems associated with as-sumptions of perfect stoichiometry and charge balancediscussed by Hawthorne (1983), such a recalculationscheme is also strongly biased by its dependence on mea-sured Si content, an element that is particularly difficultto measure accurately by microprobe. Note also that thisrecalculation depends on accurate assessment of Fe3*contents.

Mechanisms for Fe3* substitution

As discussed above, Popp et al. (1990) have identifiedthree likely mechanisms for the accommodation of Fe3+in the amphibole structure, which can be expressed as thefollowing substitution vectors: (l) Fe'z+H+Fe1l, oxysub-stitution (or deprotonation) with H+ providing chargebalance for the Fe2+ , (2) t6lAl3+Fe3+, substitution of Al3*for Fe3+, (3) tAr(K,Na)Fe2+-tAlE_rFe1], replacement of anFe3* and a vacant A site by a Fe2* and either a K or Nain the A site. Each of these relationships is plotted inFigure 5. It is immediately obvious from the nearly l:linverse correlation between H* and Fe3+ that the bulk ofthe Fe3+ variation can be explained by the relationshiprepresenting the oxysubstitution mechanism l. A linearregression fit to the plot of H* vs. Fe3* yields a slope of-0.80 + 0.10, with a correlation coefficient of 0.88. Thedeviations from a l:l inverse relationship (slope of - l)are probably indicative of other less important substitu-tions and of the limitations of the electron microprobe

DYAR ET AL.: Fe3* AND H+ IN MANTLE KAERSUTITE 975

Tlp,te S.-Continued 1 0 0

90

80

70

60

50

40

30

20

1 0

!to(u

o(do+ol!s

86-BM 86-LEH H366A H366-92 S.C. K.N.Z.

41.91 41.52 39.7614.46 15.05 14.043.23 2.50 4.346.1 1 7.51 9.972.06 3.07 3.37

15.16 14 .70 11 .290 1 1 0 . 1 0 0 . 1 5

10.74 10.17 't0.24

3.18 2.88 2.901 .21 0.96 1.52o.77 012 0.000.08 0.07 0.150.00 0.00 0.001 . r 3 1 . 2 3 1 . 1 8

100.15 99 88 98.91

39.61 40.87 39 3214.77 14.76 13 545.19 3.29 4 89o.72 7.50 7.85

12.34 4.01 3.5312.41 12.90 12.310 j2 0 .17 0 .14

10.43 10.94 9.962.82 2.97 2.561.44 1 .05 2.060.00 0.00 0.000.16 0 .o2 0 .180.00 0.01 0.020.13 0 .69 1 .14

100.14 99 .18 97 .50

38.0013.69

+ . o I

8.096.30

11 430 3 1

10.272.182.370.000.310.071 . 1 8

98.57

5 7 42.440.501.040.70z-510.041.680.640.460.000.150.021 . 1 9

Cations to 24 O atoms minus F and Cl atoms6.10 6 07 5 98 5.85 6.10 5.952.48 2 59 2.49 2.57 2.60 2.420.35 0.27 0.49 0.58 0.37 0.560.74 0.92 1.25 0.09 0.94 0.99o.23 0.34 0.39 1.37 0.45 0.403.29 3.20 2.53 2.73 2.87 2780.01 0.01 0.o2 0.02 0.02 0.021 .67 1 .59 1 .65 1.75 1 .62 1 .660.90 0.82 0.85 0.81 0 86 0.750.22 0 18 0 29 0.27 0.20 0.400 09 0.01 0.00 0.00 0.00 0.000.04 0.03 0.07 0.07 0.01 0.090.00 0.00 0.00 0.00 0.00 0.011 . 1 0 1 . 2 0 1 . 1 8 0 . 1 3 0 . 6 9 1 1 5

and wet chemical techniques in performing such sensitivemeasurements.

Examination of Figure 5 yields two further observa-tions relating to H+ contents. At one extreme, where Fe3*reaches its theoretical minimum of zero, H+ contents donot equal the stoichiometric 2 H+ pfu; instead, the inter-cept is 1.32 + 0.18. This observation underscores theidea that bulk composition, particularly substitution oftetravalent ions such as Tio* on trivalent sites and tri-valent ions on divalent sites, probably controls the amountof H+ that can be contained in the amphibole structureat mantle pressures and temperatures. At the other ex-treme, the intercept at the hypothetical zero H+ is veryclose to three data points representing almost H+-freeamphibole. Thus, true oxykaersutite probably exists andis certainly stable at surface pressures and temperatures.

Irrpr-rcnrroNs FoR MANTLE METASoMATTSM

As indicated above, the H+/Fe3+ data plotted in Figure5 display a near l: I inverse relationship between the H+and Fe3+ contents in the kaersutite samples. The data fallinto two clusters: the first, Fe3+ contents ofabout 0.4 pfuand H* contents of about 1.1 pfu, generally show <250lodehydrogenation; the second, with much higher Fe3* (near1.3 pfu) and lower H+ (near 0. I pfu), show about 900/ohydrogenation. There are four possible interpretations ofthese data: (l) the amphibole samples were partially de-hydrogenated during ascent from the upper mantle; (2)dehydrogenation occurred after the amphibole sampleswere erupted onto the Earth's surface; (3) the amphibole

10 20 30 40 50 60 70 80 90 100o/oFe3+ (measured/corrected for c)

0.2 0.4 0.6 0.8 1 .0 1.2 1.4 1 .6H+ Measured, pfu

Fig. 4. Calculated vs. measured Fe3" (a) and H+ (b) contentsof amphiboles. Calculations follow the procedure described inClowe et al. (1988). Measured and corrected Fe3* data pointsare Mrissbauer data corrected for C : 1.22.

samples were dehydrogenated during a metasomatic eventin the upper mantle subsequent to amphibole growth butprior to entrainment and transport; or (4) the amphibolesamples crystallized in regions of the upper mantle thatwere chemically distinct (in terms of O and/or H activi-ty). Each ofthese interpretations will be discussed belowin light of petrographic observations of the megacrystsand their host rocks, as well as experimental constraintson dehydrogenation of the hornblende.

Kinetics of dehydrogenation

The correlation between H* and Fe3* in these horn-blende samples may be compared with observations from

TI

2.1

1 . 9

1 . 7f

d o 1 . 5(Dat= 1 . 3(,(u()+ 1 . 1x

0.7

0.5 L-0.0

0

976 DYAR ET AL.: Fe3+ AND H* IN MANTLE KAERSUTITE

T

Fe3+ pfuFig. 5. Fe3* substitution mechanisms for amphiboles, as sug-

gested by Popp et al. (1990).

experiments investigating the hydrogenation of olivinecrystals at temperatures from 800 to 1000 oC. From theirresults, Mackwell and Kohlstedt (1990) concluded that Hwas incorporated as protons in the crystal structure ofolivine through a reaction involving the consumption ofpolarons (the excess charge on the Fe3+ ):

*[H,]*" + Felf + HJr + Fe3l

resulting in a predicted l:l inverse correlation betweenH* and Fe3+. At higher temperatures (1300'C), Bai andKohlstedt (1992\ have shown that defect relaxation onthe major element sublattices permits a decoupling of theH+ and Fe3*. A similar process might be expected tooccur in reverse in hornblende, with dehydrogenation re-sulting in a parallel increase in Fe3+ at moderate temper-atures (Dyar eL al., 1992b).

Graham et al. (1984) have measured H diffusivities inhornblende experimentally from isotope exchange mea-surements at temperatures in the range 350-850 "C. Wehave replotted the data ofGraham et al. (1984) for par-gasitic hornblende 322 and ferroan pargasitic hornblende6099 in Figure 6. We see no reason why there should besignificant differences in difftrsivities for hornblende sam-ples of such similar compositions and therefore have gen-erated linear regression fits to their data for both horn-blende samples assuming plate and cylinder geometries.Also plotted on the figure is a data point calculated fromthe infrared observations of Skogby and Rossman (1989)for the sample that they dehydrated at room pressure and700 "C in air. This result is in good agreement with thoseofGraham et al. (1984).

As the major element compositions of the hornblendesamples investigated in this study are quite similar tothose of Graham et al. (1984), we extrapolated their re-sults to higher temperatures to determine the times nec-

essary for partial or total dehydrogenation of the kaer-sutite crystals. Althoueh they determined their diffirsivitiesfrom isotope exchange experiments, as opposed to de-hydrogenation experiments, the general agreement of theirdata with the results from the experiments of Skogby andRossman (1989) suggests that the kinetics of the two pro-cesses are similar. We modeled the dehydrogenation pro-cess assuming the reaction given above for olivine andusing a simple calculation of the diffusivity for H neces-sary to produce partial dehydrogenation of hornblendecrystals of a particular grain size. In this calculation, weassumed that diffusion is essentially one-dimensional andobeys a relationship determined for difusion in a finiteslab with an infinite sink for H at the grain edges (seee.g., Schmalzried, l98l;p. 85). It is assumed in this cal-culation that dehydrogenation is driven by the partition-ing of H/HrO from the crystalline hornblende into thesurrounding melt phase, and that the rate of defect re-equilibration on the major ion sublattices in the crystalsis slow. Although no data exist for reequilibration of pointdefects in amphibole, Mackwell et al. (1988) have shownthat, for olivine crystals at 1200 'C, defect reequilibrationon the octahedral cation sublattice takes on the order ofl0 d for crystals with a grain size of about l0 mm.

Dehydrogenation of kaersutite?

The kaersutite samples investigated in this study weregenerally crystals on the order of l0 mm, although thisfigure should probably be taken as an underestimate, asmost of the crystals showed evidence of surface break-down during transport. The partial breakdown of thehornblende also provides some constraints on the tem-peratures (probably 1100-1250 'C) and time scales(probably between several hours and a month) of entrain-ment and ascent (e.g., Spera, 1984). In Figure 6, the hor-izontal extent ofthe boxes in the upper left indicates thistemperature range, while the vertical extent of each boxindicates the time span from 3 h (bottom) to 30 d (top)for l0 mm of hornblende to reach 10, 25, or 900/o dehy-drogenation. Although the cluster of H+-Fe3+ data withlesser degrees (<25o/o) ofdehydrogenation (upper left, Fig.5) might be explained by partial dehydrogenation duringascent, the data that show almost total dehydrogenation(lower right, Fig. 5) would require difftrsivities that arefar in excess ofthose predicted by the results ofGrahamet al. (1984). This latter observation may be interpretedas indicating that the Graham et al. (1984) data are notvalid at the higher temperatures of the hornblende sourceregion. However, it seems unlikely that the diffirsivitywould change by 3-4 orders of magnitude over the rangefrom 850 "C in the experiments to 1000-1200'C in theEarth, particularly given the similar chemical composi-tions of the two suites of hornblende specimens.

The right axis for Figure 6 shows the amount of timenecessary for 900/o dehydrogenation ofa hornblende witha l0 mm grain size. If we extrapolate the data of Grahamet al. (1984) to 1200'C, the time required for significantloss of H (900/0) from the kaersutite samples is on theorder of 3-30 yr. Thus, despite our earlier speculation

1 . 5

1 . 4

t . 3

1 . 2

1 . 1

1 . 0

r 0.9o-o 0.8

3 0.7C'o 0.6

0.1 FIf

0-0 L0.0 1 . 41 . 21 . 0o.4

(Dyar et al.,1992b), it seems unlikely that the entire rang-es of H* and Fe3* observed in this hornblende suite canbe explained by dehydrogenation during transport, as onlylimited H loss is possible by this mechanism.

The second possible explanation for the observed rangein H+/Fe3+ would be that the kaersutite samples main-tained a sufficiently high temperature over a protractedperiod subsequent to emplacement on the Earth's surface,during which time dehydrogenation could occur. Al-though this possibility may be plausible for megacrystsentrained in a thick sequence of lava deposits or em-placed adjacent to a long-lived lava lake, it is unable toaccount for the low H* observed in samples that showevidence of eruption as ejecta.

The third possibility, that of a long-term interactionwith a mantle fluid after crystallization but prior to en-trainment is also unsupported by petrographic analysis.For a silicate melt to produce nearly total dehydrogena-tion of kaersutite would require an interaction over a pe-riod of years that should leave clear petrographic evi-dence and zoning in the major element compositions.Such evidence has been looked for but is not observed.

The final explanation requires distinct chemical difer-ences in the metasomatic fluids from which the varioushornblende samples crystallized in the subcontinentalmantle, in accord with the recent observations of Ionovand Wood (1992), and Amundsen and Neumann (1992).Thus, for the hornblende samples in the lower right inFigure 5, the high Fe3* contents might reflect relativelymore oxidizing conditions in the metasomatic fluid, withthe low H+ contents occurring because H incorporationwould not be required for charge balance during growth:the hornblende would grow as oxykaersutite. The suite ofhornblende samples in the upper left of Figure 5 couldthen be explained as normal kaersutite specimens, someof which partially dehydrogenated during entrainment andascent. Although both types of kaersutite may be foundin the same geographic locality (e.g., the Saudi and Dead-man Lake samples), the entraining basaltic melt traversedlarge regions of the upper mantle during its ascent to thesurface and may have sampled hornblende megacrystsfrom a number of quite distinct chemical reservoirs. Sucha model requires either that the metasomatic fluids variedchemically over time within the region sampled by thebasalt, or that local heterogeneities in the chemical sig-nature of the metasomatic fluids occur over a spatial scaleof a few tens of kilometers or less.

The experimental data clearly indicate that significantdehydration of the hornblende is unlikely on the timescales of mobilization and transport of hornblende mega-crysts with crystal sizes near l0 mm. Thus, we concurwith Graham et al. (1984) that "amphibole crystals inrapidly quenched lavas will, in the absence of extensiveoxidation of alteration, preserve their high-temperature(magmatic) H isotope composition, and they shouldtherefore preserve useful information on the H isotopecompositions and origins of 'magmatic' waters." Conse-quently, we would argue that, at least for the kaersutitesamples investigated in this study, high Fe3* contents are

DYAR ET AL.: FC3* AND H* IN MANTLE KAERSUTITE

1200 900T (.C)600 300

90%

25%

LO%

Hornblende (GHS)

Platesc 322^ 6099

Cylinders

x BZZ

+ 6099

A

\ o

X \

Hornblende (SR)

a Plates

X\

t (

977

- 8

- 1 0

_ 1 9

(h

F

-74

o0

- 1 6

- r t t

_ 9 6

4 a

+)

blt

8 -

1 0

0 . 6 0 . 8 1 . 0 1 . 2 1 . 4 1 . 6 1 . 8ro3/r (/K)

Fig. 6. A plot ofH diffusivity vs. temperature for pargasitichornblende using the data ofGraham et al. (1984) (GHS) and avalue inferred from the observations of Skogby and Rossman(1989) (SR). The boxes in the upper left ofthe figure show thediffusivities necessary to produce the percentage indicated ofHloss for l0-mm hornblendes over the temperature range fromI100 to 1250 'C on a time scale of 3 h to 30 d.

probably indicative of chemical heterogeneity of the

mantle source regions.

CoNcr-usroNs

l. Mdssbauer spectra measured on a suite of 20 mantlekaersutite samples reveal a range of Fe3+ contents from1000/0 Fe3+ to only 2lo/oFe3+ . Spectra can be interpretedwith simple three doublet models. More complicated fitsare mathematically possible but lack physical relevancebecause heavy overlap between peaks renders the resul-tant fits nonunique.

2. Comparison with wet chemical results of Fe3+/Fe2*determinations on a suite of metamorphic amphibolesamples permits calculation of an improved value of C: 1.22 for the recoil-free fraction correction factor inkaersutite.

3. H+ contents as determined by U extraction tech-niques range from 0.07 to 1.32 wto/o HrO. Careful deter-mination of F by PIGE spectroscopy and Cl by electronmicroprobe shows that these elements are present in onlyminor amounts; thus, H+ variation cannot be explainedby halogen contents.

978 DYAR ET AL.: Fe3* AND H+

4. H+ contents of the kaersutite samples vary in a nearlyl: I inverse relationship with Fe3* contents. Although therange of Fe3+/H+ in the less oxidized kaersutite mega-crysts may be explained by partial H loss during entrain-ment and ascent, the near total dehydrogenation of theFe3*-rich megacrysts would require time scales signifi-cantly longer than is expected for transport. Thus, it seemslikely that these oxykaersutite samples grew in a moreoxidized metasomatic fluid, where incorporation of H wasnot required for charge compensation. As megacrysts fromthe same location show wide variation in Fe3+ and H+,it appears likely that significant variations in the oxida-tion state of the mantle metasomatic fluid occurred overrelatively small temporal or spatial scales.

AcxNowr,nocMENTS

Supported by National Science Foundation grants EAR-90-17167(A.v.M.), EAR-93-05 187 (S.J.M ), and EAR-93-04304 (M D.D.). We thankHoward Wilshire for the loan of samples from his personal collection,Michael Cosca for use of samples from Cosca et al. ( I 99 l), and the staffat the National Museum of Natural History for assistance in obtainingsamples from the Coleman collection We also acknowledge Kipp Bajaj,Tim Day, Mike Harrell, Ma{orie Taylor, and Heather Wilcox for assis-tance with sample preparation, and M.T. Colucci, R. Gregory, and C.B.Douthitt for logistical support of the isotope data. We are grateful forthoughtful reviews by Roger Burns, Michael Cosca, Etienne Deloule, Col-in Graham, and Claude Herzberg. Mdssbauer and H* analyses were fund-ed by the Mineral Spectroscopy Laboratory at the University ofOregon.

RprsnrNcns crrEDAldridge, L.P., Tse, J.S., and Bancroft, G M. (1982) The identification of

Fe'?+ in the M4 site of calcic amphiboles: Discussion. American Min-eralogist, 67 , 335-339.

Amundsen, H.E.F., and Neumann, E.-R. (1992) Redox control duringmanfle/melt interaction. Geochimica et Cosmochimica Acta. 56.2405-24r6

Andersen, T., O'Reilly, S.Y., and Griffin, W.L. (1984) The trapped fluidphase in upper mantle xenoliths from Victoria, Australia: Implicationsfrom mantle metasomatism. Contributions to Mineralogy and Petrol-ogy, 88, 72-85.

Aoki, K.-L (1963) The kaersutites and oxykaersutites from alkalic rocksofJapan and surrounding areas. Journal ofPetrology, 4,198-210.

- (l 970) Petrology of kaersutite-bearing ultramafic and mafic inclu-sions in Iki Island, Japan. Contributions to Mineralogy and Petrology,25,270-283.

-(1971) Petrology of mafic inclusions from Itinome-gata, Japan.Contributions to Mineralogy and Petrology, 30, 3 l4-33 l.

Bahgat, A.A , and Fayek, M.K. (1982) Absolure evaluation offerrous andferric concentration in Ca amphibole. Physica Status Solidi A, 7 l, 575-5 8 1 .

Bai, Q, and Kohlstedt, D.L. (1992) Substantial hydrogen solubility inolivine and implications from water storage in the mantle. Nature, 357,672-67 4.

Bailey, D.K. (1970) Volatile flux, heat focusing, and generarion of magma.Geological Joumal Special Issue, 2, 177-186.

-(1972) Uplift rifting and magmatism in continental plates. LeedsJournal of Earth Sciences, 8. 225-239.

Bancroft, G.M. (1973) Mtissbauer spectroscopy: An introduction for in-organic chemists and geochemists, p 34 McGraw-Hill, New york.

Bancroft, G.M., and Brown, J.R. (1975) A M6ssbauer study of coexistinghornblendes and biotites: Quantitative Fe3+/Fe,+ ratios. AmericanMineralogist, 60, 265-272.

Barnes, V.E. (1930) Changes in hornblende at about 800qC. AmericanMineralogist, 15, 393-417.

Bergman, S.C. (1981) Fluid inclusions in xenoliths: SamDles of the mantle

IN MANTLE KAERSUTITE

metasomatic fluid? Geological Society of America Abstracts with Pro-grams, 13,408

Best, M G. (1970) Kaersutite-peridotite inclusions and kindred mega-crysts in basanitic lavas, Grand Canyon, Arizona. Contributions toMineralogy and Petrology, 27, 25-44.

- (1974) Mantle-derived amphibole within inclusions in alkalic-ba-saltic lavas. Journal of Geophysical Research, 79 (14), 2107-2113.

Bigeleisen, J., Perlman, M.L., and Prosser, H.C. (1952) Conversion ofhydrogenic materials to hydrogen for isotopic analysis. AnalyticalChemistry, 24, 1356-1357.

Boettcher, A.L., and O'Neil, J.R. (1980) Stable isotope, chemical, andpetrographic studies of high pressure amphiboles and micas: Evidencefor metasomatism in the mantle source regions of alkali basalts andkimberlites. American Journal of Science. 280-A. 594-62 -.

Bryndzia, L.T., Davis, A.M., and Wood, B.J. (1990) The isotopic com-position (6D) ofwater in amphiboles and the redox state ofamphibole-bearing upper mantle spinel peridotites Geological Society of AmericaAbstracts with Progams, 22 (7), A254.

Burns, R G., and Greaves, C.J. (1971) Conelations of infrared and M6ss-bauer site populations ofactinolites. American Mineralogist, 56, 2010-2033.

Bums, R.G., Greaves, C.J , Law, A.D., Tew, M J., and Prentice, F.J.(1970) Assessment of the reliability of the Mdssbauer and infraredmethods in site-population studies of amphiboles. Geological Societyof America Abstracts with Programs, 2,509-511.

Burns, R.G., Bums, Y.M , Dyar, M.D , Ryan, V.L., and Solberg, T. (1985)Iron coordination symmetries in silicates: Correlations from Mdss-bauer parameters ofFer* and Fer+. Geological Society ofAmerica Ab-stracts with Programs, 17 (7), 535.

Canil, D., Virgo, D., and Scarfe, C.M. (1990) Oxidation state of mantlexenoliths from British Columbia, Canada. Contributions to Mineralogyand Petrology, 104, 453-462.

Clowe, C.A, Popp, R.K., and Fritz, S.J. (1988) Experimental investiga-tion of the effect of oxygen fugacity on ferric-ferrous ratios and unitcell parameters of four natural clinoamphiboles. American Mineralo-gist,73, 487-499.

Cosca, M.A., Essene, E.J., and Bowman, J.R. (1991) Complete chemicalanalyses of metamorphic homblendes: Implications for normaliza-tions, calculated HrO activities, and thermobarometry. Contributionsto Mineralogy and Petrology, 108,472-484.

Deloule, E., Albardde, F., and Sheppard, S.M F (1990) D/11 anatysis ofamphiboles from mantle lherzolite xenoliths by ion probe. Internation-al Conference on Geochronology, Cosmochronology, and Isotope Ge-ology,7 , 26.

-(1991) Hydrogen isotope heterogeneities in the mantle from ionprobe analysis ofamphiboles from ultramafic rocks. Earth and Plane-tary Science lrtters, 105, 543-553

Dick, H.J.B., Fisher, R.L., and Bryan, W.B. (1984) Mineralogic variabilityofthe uppermost mantle along mid-ocean ridges. Earth and PlanetaryScience ktters. 69. 88-106.

Dodge, F.C.W., Papike, J.J., and Mays, R.E. (1968) Hornblendes fromgranite rocks ofthe Central Sierra Nevada Batholith, Califomia. Jour-nal of Petrology, 9, 378-410

Dyar, M.D. (1984) Precision and interlaboratory reproducibility of mea-surements of the Mdssbauer effect in minerals. American Mineralogist,69, t t27-t t44.

Dyar, M.D., McGuire, A.V., and Ziegler, R.D. (1989) Redox equilibriaand crystal chemistry of coexisting minerals from spinel lherzolite man-tle xenoliths. American Mineralogist, 7 4, 969-980.

Dyar, M.D., McGuire, A.V., and Harrell, M.D. (1992a) Redox behaviorin contrasting styles of metasomatism in the upper mantle. Geochimicaet Cosmochimica Acta. 56. 2579-2586.

Dyar, M.D., McGuire, A.V., and Mackwell, S.J. (1992b) Fe.* /H+ and D/11 in kaersutites: Misleading indicators of mantle source fugacities. Ge-ology, 20, 565-568

Ernst, W.G., and Wai, C.M. (1970) Mdssbauer, infrared, X-ray, and op-tical study ofcation ordering and dehydrogenation in natural and heat-treated sodic amphiboles. American Mineralogist, 5 5, 1226-1 258

Goldman, D.S., and Rossman, G.R. (1977) The identification of Fer* inthe M(4) site of calcic amphiboles. American Mineralogist, 62, 205-2 t 6 .

DYAR ET AL.: Fe3+ AND H+ IN MANTLE KAERSUTITE 979

Goodman, B.A., and Wilson, M.J. (1976) A Mtissbauer study of theweathering of hornblende. Clay Minerals, 2, 153-163.

Graham, C.M., Harmon, R.M., and Sheppard, S M.F. (1984) Experimen-tal hydrogen isotope studies: Hydrogen isotope exchange between am-phibole and water American Mineralogist, 69, 128-138.

Harte, B., Cox, K.G., And Gurney, J.J (1975) Petrography and geologicalhistory of upper mantle xenoliths ftom Matsoku Kimberlite pipe. Phys-ics and Chemistry ofthe Earth, 9,477-506

Hawthorne, F.C. (1978) The crystal chemistry of the amphiboles VI Thestereochemistry ofthe octahedral strip. Canadian Mineralogist, 16, 37-

--11f eSf 1 Crystal chemistry of the amphiboles. In Mineralogical So-ciety of America Reviews in Mineralogy, 9A, l-102.

-(1983) The crystal chemistry of the amphiboles. Canadian Min-eralogist, 21, 173-480.

Hawthorne, F.C., and Grundy, H D. (1973) The crystal chemistry of theamphiboles. II. Refinement of the crystal structure of oxy-kaersutite.Mineralogical Magazine, 39, 390-400

Holdaway, M.J, Dutrow, B.L., Borthwick, J., Shore, P., Harmon, R.S.,and Hinton, R.W. (1986) H content of staurolite as determined by Hextraction line and ion microprobe. American Mineralogist, 71, I I 35-I 1 4 1

Ionov, D.A., and Wood, B.J. (1992) The oxidation state ofsubcontinentalmantle: Oxygen therrnobarometry ofmanfle xenoliths from central Asia.Contributions to Mineralogy and Petrology, I I I , I 79- 1 93

lirirk,2., Pechar, F., and Vratislov, S. (1986) Distribution ofcations andthe proton location in kaersutite Crystallographic Research and Tech-no logy ,21 , l 5 l 7 -1519 .

Kitamura, M., and Tokonami, M. (1971) The crystal structure of kaer-sutite. Scientific Repons ofthe Tohoku University, Series 3,11,125-141 .

Kuroda, Y., Hariya, Y., Suzuoki, T., and Matsuo, S. (1978) Hydrogenisotope study on biotite and homblende from Finnish granitic rocks.Geochemical Journal, 12, 259-263.

Kyser, T.K. (1992) Hydrogen isotope systematics of mafic magmas andmantle minerals (abs.). Eos, 73, 348

Mackwell, S J., and Kohlstedt, D.L. (1990) Diffusion of hydrogen in ol-ivine: Implications for water in the mantle. Journal of GeophysicalResearch. 95. 5079-5088.

Mackwell, S.J., Dimos, D., and Kohlstedt, D.L. (1988) Transient creepof olivine: Point defect relaxation times. Philosophical Magazine, 57 ,7'79-789

Matson, D.W., Muenow, D.W., and Garcia, M.O. (1984) Volatiles inamphiboles from xenoliths, Vulcan's Throne, Grand Canyon, Arizona.Geochimica et Cosmochimica Acta. 48. 1629-1636.

McGuire, A.V , Dyar, M.D., and Ward, K.W. (1989) Neglected Fe.*/Fe,*ratios: A study ofFer* content of megacrysts from alkali basalts Ge-ology, 1 7, 687 -690

McGuire, A.V., Dyar, M.D., and Nielson, J E (1991) Metasomatic oxi-dation of upper mantle peridotite. Contributions to Mineralogy andPetrology, 107, 252-264.

Menzies, M.A. (1983) Mantle ultramafic xenoliths in alkaline magmas:Evidence for mantle heterogeneity modified by magmatic activity InC.J. Hawkesworth, and M.J Norry, Eds., Continental basalts and man-tle xenoliths, p. 92-110. Shiva, Nantwich, U.K.

Nielson, J.E., Budahn, J.R., Unruh, D.M., and Wilshire, H.G. (1993)Actualistic models of mantle metasomatism documented in a compos-ite xenolith from Dish Hill, California. Geochimica et CosmochimicaActa,57, 105-121.

Oxburgh, E.R. (1964) Petrologic evidence for the presence of amphibolein the upper mantle and its petrogenetic and geophysical implications.Geological Magazine, l0l, 1-19

Papike, J.J, and Clark, J.R. (1968) The crystal structure and cation dis-tribution of glaucophane. American Mineralogist, 5 3, 1 1 56- I I 73.

Papike, J.J., Ross, M., and Clarke, J.R. (1969) Crystal-chemical charac-terization of clinoamphiboles based on five new structure refinements.Mineralogical Society ofAmerica Special Paper,2,1 156-1 173.

Papike, J J., Cameron, K.L., and Baldwin, K. (1974) Amphiboles andpyroxenes: Characterization of OTHER quadrilateral components andestimates of ferric iron from microprobe data. Geological Society ofAmerica Abstracts with Programs, 6, 1053-1054.

Pechar, F., Fuess, H., and Joswig, W. (1989) Refinement of the crystalstructure of kaersutite (Vlditbra, Bohemia) from neutron diffraction.Neues Jahrbuch {iir Mineralogie Monatshefte, 89 (3), 137-143.

Phillips, M.W., Popp, R.K., and Clowe, C.A. (1988) Structural adjust-ments accompanying oxidation-dehydrogenations in amphiboles.American Mineralogist, 73, 500-506.

Phillips, M.W., Draheim, J.A., Popp, R.K., Clowe, C.A., and Pinkerton,A.A. (1989) Effects of oxidation-dehydrogenation in tschermakitichornblende. American Mineralogist, 7 4, 7 64-7 7 3.

Popp, R.K., Phillips, M.W., and Harrell, J.A. (1990) Accommodation ofFe3* in natural Fer*-rich calcic and subcalcic amphiboles: Evidencefrom published chemical analyses. American Mineralogist, 75, 163-I 69.

Roden, M.F., Frey, F.A., and Francis, D.M. (1984) An example of con-sequent mantle metasomatism in peridotite inclusions from NunivakIsland, Alaska. Journal of Petrology, 25, 546-577 .

Roedder, E. (1965) Liquid CO, inclusions in olivine-bearing nodules andphenocrysts from basalts. American Mineralogist, 50, 1746-1782.

Ruby, S.L. (1973) Why Misfit when you already have 1'?? In I.J. Gruver-man and C.W. Seidel, Eds., Miissbauer effect methodology, p. 263-276 Plenum, New York.

Schmalzried, H. (1981) Solid state reactions, 254 p. Verlag Chemie, Wein-heim, Germany.

Schumacher, J.C. (1991) Empirical ferric iron corrections: Necessity, as-sumptions, and effects on selected geotherrnobarometers. MineralogicalMag;azine,55, 3-18.

Schwartz, K.B., and lrving, A.J. (1978) Cation valence determinations inkaersutite and biotite megacrysts from alkalic basalts: Evidence for Tia*in the upper mantle. Eos, 59, 1214-1215.

Sheppard, S.M.F., and Epstein, S. (1970) D/ H and '80/'60 ratios of min-erals of possible mantle or lower crustal origin. Earth and PlanetaryScience l€tters. 9. 232-239.

Skogby, H., and Rossman, G.R. (1989) OH in pyroxene: An experimen-tal study of incorporation mechanisms and stability. American Min-eralogist, 74, 1059-1069.

Spera, F.J. (1984) Carbon dioxide in petrogenesis. III. Role ofvolatilesin the ascent of alkaline magmas with special reference to xenolith-bearing mafic lavas. Contributions to Mineralogy and Petrology, 88,2t7-232.

Stosch, H.-G., and Seck, H.A. (1980) Geochemistry and mineralogy oftwo spinel peridotite suites from Dreiser Weiher, West Germany. Geo-chimica et Cosmochimica Acta, 44, 457 -47O.

Stout, J.H. (1972) Phase petrology and mineral chemistry ofcoexistingamphiboles from Telemark, Norway. Journal ofPetrology, 13,99-145.

Varne, R. (1970) Hornblende lherzolite and the upper mantle. Contri-butions to Mineralogy and Petrology, 27, 45-51.

Whipple, E.R. (1972) Quantitative Mtissbauer specua and chemistry ofiron, 205 p. Ph.D. thesis, Massachusetts Institute of Technology, Cam-bridge, Massachusetts.

Wilshire, H.G. (1987) A model of mantle metasomatism. Geological So-ciety ofAmerica Special Paper, 215,47-60.

Wilshire, H.G., And Shervais, J.W. (1975) Al-augite and Cr-diopside ul-tramafic xenoliths in basaltic rocks from westem United States: Struc-tural textural relationships. Physics and Chemistry ofthe Earth, 9, 257-272.

Wilshire, H.G., and Trask, N.J. (1971) Structural and textural relation-ships of amphibole and phlogopite in peridotite inclusions, Dish Hill,California. American Mineralogist, 56, 240-253.

Winchell, A.N. (1945) Variation in composition and properties of calcif-erous amphiboles. American Mineralogist, 30, 27 -5O.

Wong, A.S., Robertson, J.D., and Francis, H.A. (1992) Determination offluorine in coal and coal fly ash by proton-induced gamrna-ray emissionanalysis. Joumal ofTrace and Microprobe Techniques, 10, 169-181.

Wood, B.J., and Virgo, D. (1989) Upper mantle oxidation state: Ferriciron contents of lherzolite spinels by "Fe Mtissbauer spectroscopy andresultant oxygen fugacities. Geochimica et Cosmochimica Acta, 53,1277-1291.

Maxr;scnrpr RECETvED Ocrosen 21, 1992MlNuscrrsr AccEPTED MeY 13, 1993