American Mineralogist, Volume 77, pages 34-43, 1992

Miissbauer spectroscopy of tetrahedral Fe3* in trioctahedral micas

D. G. R-lNcounrOttawa-Carleton Geoscience Centre and Ottawa-Carleton Institute for Physics, Department of Physics,

University of Ottawa, Ottawa, Ontario KIN 6N5, Canada

M.-2. DeNc*Ottawa-Carleton Institute for Physics, Department of Physics, University of Ottawa,

Ottawa, Ontario KIN 6N5, Canada

A. E. Lar-oNnnOttawa-Carleton Geoscience Centre, Department of Geology, University of Ottawa,

Ottawa, Ontario KtN 6N5, Canada

Arsrnlcr

Six trioctahedral true micas are studied by Mtissbauer spectroscopy and microprobeanalysis. Three biotite samples, with a wide range of octahedral Fe3*/Fe2* ratios, chemicalcompositions that recalculate to full tetrahedral occupancies without requiring Fe3*, andno trace of tetrahedral spectral components, are used for comparison. Both near end-member phlogopite and near end-member annite samples exhibit the same distinctivetetrahedral spectral feature as that ofa reverse pleochroic phlogopite sarnple that is known(Hogarth et al., 1970) to contain t4lFe3+. At room temperature, the latter spectral featureis a distinct shoulder occurring al +0.41 a 0.02 mm/s with respect to a-Fe-on the high-energy side of the mainly 1ol1isz+ p€ak at -0.1 mm/s. It arises from a relatively narrowdistribution of quadrupole doublet splittings and can be used to assess unambiguously thepresence of t41Fe3+ from room-temperature Mdssbauer spectra. It becomes significantlymore distinctive at liquid N, temperatures-where it appears at - *0.5 mm/s.

All previous reports of I4lFe3+ from Mcissbauer spectra are reviewed and compared withour results. A pervasive mistake has been to assume that line positions for the hiddenlow-energy lines of tetrahedral doublets can be obtained reliably by fitting. When only thehigh-energy lines are considered, the true and brittle trioctahedral micas separate naturallyinto two fields that are consistent with the different interlayer charges of true and brittlemicas. Another mistake is to believe in tetrahedral parameters lying outside these fieldsand corresponding to completely hidden doublets. In particular, such spectral contribu-tions recently reported by Dyar (1990) are argued to be an arllfact of improper analysisand interpretation.

IlmnooucrroN

In natural and synthetic micas the biggest problem withMtissbauer sp€ctroscopy arises from the superposition ofdifferent spectral lines. This makes discriminating thespectral components that result from different ions (Fe2*and Fe3*) or from ions in different crystallographic sitesdifficult, when not impossible. In this context, it is im-portant to establish which information can be confidentlyobtained from the spectra and which arises as an artifactof improper analysis and overfitting.

The only separate spectral contributions that can bebelieved and used in quantitative measurements are thosethat give rise to distinct and resolved spectral featuresseen by visual inspection of high-quality spectra. Hidden

* On leave from: Department of Physics, Hebei Normal Uni-versity, Shi Jia Zhuang, Hebei, People's Republic of China.

0003-004x/92l0 1 02-0034$02.00

spectral contributions buried under more intense lines-even such contributions that have reasonable or expectedhyperfine parameters and that do not conflict with for-mula recalculations based on chemical analyses-shouldnot be believed or used. In such situations, many otherfits are statistically equivalent to the chosen one, and onecan only conclude that the chosen model is not inconsis-tent with the data and that it cannot be distinguishedfrom several other possible models.

The purpose ofthe present paper is to show that certainobserved spectral features in the room-temperature andliquid-N, temperature Milssbauer spectra of trioctahedralmicas unambiguously indicate the presence of t4lFe3+ in-dependently of any assumed line-shape fitting model.When these features are present, they can be used to es-timate amounts of t41Fe3+ that otherwise are missed byboth the usual assignment method during formula recal-

34

RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS

TlaLe 2. Composition of trioctahedral micas

35

Trele 1. Sample descriptions

ROM- HEP. BIS- MOC.ANN BIO BIO BIO

MCL-PHL

b

MCL-PHL

Sample Description

HEA.PHL

Reverse pleochroic phlogopite mica from the Mccloskeycarbonatite, Quebec. Small flakes (-1-2 mmr) sepa-rated from rock. Previously studied (Hogarth et al.,1970; Faye and Hogarth, 1969). This phlogopite sam-ple has now been deposited in the mineral collectionof the Royal Ontario Museum (sample no. M44527).

Phlogopite mica from the Headley mine near Old Chel-sea, Quebec. Same single-crystal wafer as studiedpreviously (Hargraves et al., 1990): approximately 4cm x 5cm x 335rm.

Annite from Mont. St. Hilaire, Ouebec. Specimen fromthe mineral collection of the Royal Ontario Museum,Toronto, Canada. Sample no. M42126. Magnetismpreviously studied (Rancourt et al., 1990). Single crys-tal -1 cm x 1 cm x 350 sm.

Biotite from the Hepburn intrusive suite, Northwest Terri-tories, Canada. Sample L341A (Lalonde, 1989). Verysmall flakes separated from rock. About 5 mg used in0.s-in. diameter absorber.

Biotite from the Bishop intrusive suite, Northwest Terri-tories, Canada. Sample L531 (Lalonde, 1989). Largeflakes (-2 mm diameter) manually separated fromrock and used to make a mosaic absorber.

Biotite from the Silver Crater mine near Bancroft, Ontar-io. Sample MOC2661 from the Canadian Museum ofNature, Mineral Sciences Division. Large single-crystalwaler -4 cm x 5 cm x 150 pm. This sample hasbeen the subject of an extensive singl+crystal andvariable-temperature Mdssbauer study (Rancourt etal., in preparation).

ROM-ANN

HEP-BIO

BIS.BIO

MOC-BrO

culation from microprobe analysis and not overly carefulMiissbauer spectroscopy.

We compare our findings with all available Mdssbauerdata, including a study of a large number of specimensby Dyar (1990), and conclude that the usual analysis interms of a talFe3+ quadrupole doublet is misleading. Weobserve that the lalFe3* hyperfine parameters of brittletrioctahedral micas are well resolved from those of truetrioctahedral micas and that the spread of parameter val-ues for trioctahedral micas is thereby not as wide as sug-gested (Dyar, 1987), in that it is probably composed oftwo separate fields. We argue that parameters falling out-side these fields, especially those also corresponding tohidden lines only, are incorrect.

M,lrrRrALs AND METHoDS

The six mica samples used as Mdssbauer absorbers inthe present study are described in Table l. Chemicalanalyses for all samples were obtained by microprobeanalysis, in the same way as described elsewhere (Har-graves et al., 1990). The analyses and correspondingstructural formulas are given in Table 2.

Transmission 57Fe Mdssbauer spectra were obtained,calibrated, and folded in the usual way (e.g., Hargraveset al., 1990) with a range of +4 mm/s. Folding is essentialto produce a flat background with parameters that do notinteract with those of the absorption lines when least-squares fitting. Thickness corrections (Rancourt, 1989)were not performed since accurate amounts of ionic pop-ulations were not required. All line positions and centershifts are given with respect to the center shift ofa 57Fe-

enriched Fe foil at room temperature.

42.44 39.51 34.06 34.24 36.540.12 1.25 1.00 2.93 3.139.72 14.30 8.91 19.16 14.646.51 3.67 39.69 21.87 19.870.01 0.03 2.15 0.29 0.34

25.47 24.27 0.47 6.37 11.100 0.01 0 0.01 0.030.10 0.20 0.86 0.14 0.06

10.58 9.55 8.56 9.26 9.764.45 4.19 0.45 0.63 3.590.01 0.03 0.02 0.32 0.69't.87 1.78 0.20 0.34 1 .67

97.54 95.23 95.97 94.88 98.08Structural formulae based on 22 O atoms6.121 5.728 5.882 5.333 5.5541.652 2.272 1.814 2.667 2.4460 - 0 1 3 0 0 . 1 3 0 0 07.786 8.000 7.826 8.000 8.0000 0.170 0 0.850 0.1760 0.137 0 0.343 0.3580.785 0.445 5.732 2.848 2.5270.002 0.004 0.315 0.038 0.0445.478 5.245 0.120 1.478 2.5146.265 6.001 6.168 5.558 5.6100.001 0.001 0 0.003 0.0040.028 0.058 0.289 0.044 0.0171.948 1.765 1.886 1.839 1.8931 .977 1 .824 2.175 1 .885 1 .9142.029 1.922 0.246 0.308 0.3300.004 0.007 0.007 0.084 0.023

Sample-

No. ofanalyses**

HEA-PHL

4

sio,Tio,Al,o3FeO,o,MnOMgoCaONaroKrOF

O = F , C lTotal

38.652.22

11.0918.1 I0.90

13.6900.529.453.870.081.65

97.00

5.9402.0080.0538.00000.2052.3370 . 1 1 83.1365.79400.1561.8532.0091.8780.020

Sir+;Al14tTi

>dretAl16rTi

FeP-MnMg>dCaNaK2",r

- According to sample descriptions in Table 1." Number of analyses included in mean.

In fitting these spectra, as few Voigt lines as needed togive statistically ideal fits were used, without worryingabout how the various Voigt lines are coupled into quad-rupole doublets, etc. Voigt lines are better suited thanLorentzian lines for two reasons: (l) they are better ableto handle spectral distortions from thickness effects (Ran-court, 1989), and (2) they are a natural choice when dis-tributions of static hyperfine parameters (i.e., quadrupolesplittings and center shifts in this application) are impor-tant (Rancourt and Ping, l99l). Distributions are impor-tant in mica and are the main reason that Lorentzian linescannot be used (Hargraves et al., 1990).

Rrsur,rs AND DrscussroN

Recognizing I4lFe3+

Room-temperaine (22-24 "C) Mdssbauer spectra ofthe six micas described in Tables I and 2 are shown inFigure l. The fits corresponding to the parameters givenin Table 3 are purposely not shown so that the spectraldata can be scrutinized without interference or bias.

Samples HEP-BIO, BIS-BIO, and MOC-BIO have typ-ical M0ssbauer spectra of textured or single-crystal sam-ples that do not contain ?oy [+lpsr+. Such spectra, in gen-eral, have three main absorption peaks at --0.1, +1.0,and +2.3 mm/s. The peak at 2.3 mm/s is due only tohigh-energy lines of I5lFe2* quadrupole doublets. The peakat 1.0 mm/s is due only to high-energy lines qf trllrer+

RANCOURT ET AL.: TETRAHEDRAL FE3* IN TRIOCTAHEDRAL MICAS

Velocily (--/r) Velocity (--/9

- 2 - 1 0 1 2 3 4 - 2 - 1 0 1 2 3 4

Velociiy (--/9 Velocily (.-/9

0 .706

36

oC r z t z

o-cO

(Do-

1.293@

cfo

O

$ r.zosc)

6 .42c)cco-c 6.76OLo

o- 6.70aEf

8 6.64d

o,q)

6.52

BIS-BIO RT

2.221

q)cc 2.181o_cO

^o 2.141uU)tf ^ . ^ .o z . t v l(_)oO)o ^ ^ - .

2.021

occo{ 1.7375

bo_at 1.717sfo

O

fl r.oszs

occ

: o .7soO

oo_

t u . / o ofoO

o) n zrn( D - " - -

1.7575

1.6775

2.255q)ctro r r t c

_c

Loo- 2.i7sa

Elo c r z <

O

oO)o)

2.055- 2 - 1 0 1 2 3 4 - 2 - 1 0 1 2 3 4

Velocily (--/t) Velociiy (--/9

HEP-BIO RT

- 2 - 1 0 1 2 3 4- 2 - 1 0 1 2 3 4

HEA-PHL RT

MOC-BIO RTROM_ANN RT

Fig. 1. Room-temperature (folded) Mdssbauer spectra of the six natural micas (as labeled) described in Table 1.

RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS 5 l

[o]Fe2*

-?o+2+4

DOPPLER VELOCITY (mm/s)

Fig. 2. Schematic representation of a room-temperatureM<issbauer spectrum of a true trioctahedral mica containingI6lFe2+, tslFer+, and t4lFe3*. Each type of ion gives rise to a distri-bution of quadrupole doublets. Each such family of doublets isrepresented by a "bar doublet" showing the approximate cen-troids ofthe high- and low-energy absorptions resulting from thedistributions. Visible absorption line a is due to the low-energycontributions ofall three ionic types; the three centroids are notactually at the same velocity as shown but usually cannot beresolved. Visible shoulder b is due to the high-energy doubletlines of the [alFe3+ contribution. Visible bump c is due to thehigh-energy doublet lines of the I61Fe3+ contribution, and visibleline d is due to the high-energy doublet lines ofthe t6rFe2+ con-tribution. Shoulder b is the object of the present study.

quadrupole doublets. The peak at -0. I mm/s, on theother hand, is due to the low-energy lines of quadrupoledoublets corresponding to both the Fe2* and the Fer* inoctahedral sites. This is depicted schematically in Figure2, where the tetrahedral doublet is also shown. From theseassignments, it can be seen that sample HEP-BIO has asmall octahedral Fe3*/Fe2* ratio, BIS-BIO has a relativelylarge ratio, and sample MOC-BIO has an intermediatevalue. The three spectra illustrate the range of octahedralFe3*/Fe2* ratios encountered in natural biotite.

The spectrum for sample MCL-PHL has a prominentshoulder at -0.4 mm,/s that is believed to be the high-energy line of a quadrupole doublet (Figs. I and 2) cor-responding to lalFe3+ (Hogarth et al., 1970). Since Fe-bearing richterite coexists with sample MCL-PHL in therock, great care was taken to establish (l) that sampleMCL-PHL was pure to better than 97 volo/o and (2) thateven a large amount of this amphibole could not causethe spectral feature at 0.4 mm/s.

0 .95

MCL-PHL(HEP-Bro)

RT

0 .89- 2 - 1 0 1 2

Velocity (../9

Fig. 3. Direct comparison (on expanded velocity scale) be-tween spectra for samples MCL-PHL (data points) and HEP-BIO (solid line), showing the high-energy tetrahedral shoulder asa significant difference.

The shoulder at 0.4 mm/s in the spectrum for sampleMCL-PHL is shown most clearly in an expanded anddirect comparison with that of sample HEP-BIO (Fig. 3).It is too narrow and at too low an energy to correspondto t5lFe3*, which usually gives absorption centered at 0.8-1.2 mm/s. Also, and most importantly, work done on asynthetic ferriphlogopite having all Fe in the foln sf trtps:+shows a quadrupole doublet with its high-energy line at-0.4 mm/s (Annersten et al., l97l). We conclude thatour feature is indeed due to I4lFe3+ that must be presentin sample MCL-PHL.

The same shoulder is also present in spectra of samplesHEA-PHL and ROM-ANN (Fig. l). That the spectrumof sample HEA-PHL indeed contains such a contributionis shown in expanded comparisons with spectra of sampleMOC-BIO at both room temperature and liquid N, tem-perature (Fig. a). The liquid N, temperature (actually 107+ 2 K) spectra themselves are shown with Table 3 fits inFigure 5. At this temperature, the t4lFe3+ shoulder hasmoved to -0.5 mm/s (2, in Table 3).

That samples HEA-PHL and ROM-ANN have talFe3+components is strongly supported by efforts to fit theirspectra. Statistically ideal fits (x?.o - I or less) are onlyobtained if the shoulder is allowed its own absorptionline. Table 3 is a complete compilation of the Mdssbauer-fit results that are discussed below.

Fitting the spectra

The main absorption line at -0.1 mm/s, in general,contains the low-energy quadrupole components gf t+tFar+,telpsr+, ard t0tpsz+ (line a in Fig. 2) and is modeled byVoigt lines I and 2 (Table 3). These two lines have noparticular physical meaning. They mathematically modelthe effective absorption caused by all the true constituentsof line a. The Fe3* components are not resolved because

1 . 0 1

=ococ

o.zq)

E.aFz:foC)

0.92

38 RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS

q)cco_cO

oo_aEf

Ooo

=vlca)co

oo

M

0.952

'^:.^'

HEA-PHL(Moc-Bro)

RT

- 1 0 1

Velocity (mm/s)

0 982

0.952

0.922

0 .892- 2 - 1 0 1 2

Velocity (--/.)

Fig. 4. Direct comparisons (on expanded velocity scales) ofspectra for samples HEA-PHL (data points) and MOC-BIO (sol-id lines) at room temperature (RT) and liquid N, temperature(LI.|. At the lower temperature, the tetrahedral component ismuch more highly resolved and occurs at -0.5 mm/s.

1.8972

1.8672

1 .A372

1.4072

ocqo

_cOLoo_acf

OoO)o

' a

c(!)

c

o

=oov.

0.714

0 .684

0.654

o.624

0 594

0 1 2

Velocity (--/9

MOC-B IO LN

- 2 - 1 0 1 2 3 4

Velocity (rn-,/.)

Fig. 5. Liquid N, temperature (actually lO7 ! 2 K) M6ss-bauer spectra of the Headley phlogopite sample (HEA-PHL) andthe MOC biotite sample (MOC-BIO). Solid lines correspond tothe fits described in Table 3. Broken lines show the six individualVoigt components in the fit of the spectrum of sample HEA-PHL. These are shifted up to be more visible.

TABLE 3. Mossbauer fit results using either five or six Voigt lines-

v3nsv2h2g1Sample

MCL.PHLHEA-PHLROM-ANNHEP-BIOBIS.BIOMOC-BrOHEA-PHLMOC-BtO

RT 0.105 354.6 -0.159 0.277 155.5RT 0.041 32.6 -0.196 0.144 81.0RT 0.066 140.2 -0.188 0.115 1s7.3RT 0.082 104.6 -0.182 0.174 57.5RT 0.072 118.4 -0.181 0.185 225.8RT 0.094 87.0 -0.171 0.183 117 .0LN 0.102 123,6 -0.133 0.063 27.4LN 0.'112 1225 -0.178 0.174 103.5

0.191-0.030

o.o420.073

-0.0110.0330.163o.o22

0.185 61 .40 .166 11 .30.176 50.20.120 27.10.180 73.10.237 36.30.000 2.20.183 26.9

2.235 0.084 230.12.174 0.088 36.42.233 0.080 108.12.154 0.080 59.32.167 0.091 99.62.253 0.130 66.72.246 0.120 67.12.435 0.135 98.1

' Here f is the absorber temperature (RT : 22-24 "C, LN : 107 + 2 Kl, o, is the Gaussian width for the rth Voigt line (in mm/s), 4 is the Lorentzianheight for the /th Voigt line (in kilocounts per channel), 4 is the ith Voigt line position (in mm/s with respect to metallic Fe at room temperature), ? isthe Lorentzian FWHM (in mm/s) for all Voigt lines, and BG is the background level in megacounts per channel. Since common ?'s are used, ratios ofVoigt heights are equal to corresponding ratios of Voigt areas for a given spectrum.*'This parameter was frozen durino fit.

RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS 19

they are relatively weak. The fact that two Voigts areneeded for line a does not imply that Fe3* componentsare present. For example, two Voigts are also needed forthe purely octahedral Fe2* absorption ofline d (Fig. 2).

The absorption at *2.3 mm/s (line d in Fig. 2) is dueonly to the high-energy quadrupole doublet componentsof I6lFe2+ and is modeled by Voigt lines 3 and 4. Here theneed for two Voigt lines at different velocities implies anintrinsically asymmetric absorption line caused by askewed distribution of Fe2* elemental line positions. Suchdistributions are common and arise from the expectedcoupling between quadrupole splittings and center shifts(Rancourt and Ping, l99l).

Voigt line 5 (Table 3) is used to model the broad ab-sorption band at 0.8-1.2 mm/s (line c in Fig. 2) arisingfrom the high-energy quadrupole doublet components ofr61Fe3+. The Voigt line labeled "1" in Table 3 models thehigh-energy quadrupole doublet line arising from t4lFe3+(line b in Fig. 2) and is required to fit shoulders in thespectra of samples MCL-PHL, HEA-PHL, and ROM-ANN (Fig. l).

Because of overlapping lines, it is impossible to obtainquantitative site populations from single-crystal, orient-ed, or textured samples unless enough is known about thehyperfine electric-field gradient magnitudes and direc-tions (Rancourt, 1989; Hargraves et al., 1990). Given re-cent single-crystal work (Hargraves et al., 1990), hs/> hiand hr/(h, + fto), where fr is the Lorentzian height for theparticular Voigt line, can be used as lower and upperbounds, respectively, to the true ratio of total I6lFe3+ tototal spectal area. Also, given the negative sign ofthe Fer*electric field gradient (e,qQ < 0), the lower bound is clos-er than the upper bound for highly textured or orientedsamples, as in this study. Similarly, h,/2 h, and h,/(h, +ho) are, respectively, taken to be lower and upper boundsto the true ratio of total t4lFe3+ to total spectral area. Thesequantities (Table 3) allow us to conclude that (l) samplesMCL-PHL, HEA-PHL, and ROM-ANN have compa-rable ratios of t4lFe3+ to Fe,", of -5-l0o/o and (2) thatsamples HEA-PHL and ROM-ANN have comparableamounts of tetrahedral and octahedral Fe3*, whereassample MCL-PHL probably has no I61Fe3+.

Table 3-Continued

The latter conclusions cannot easily be made based onlyon visual inspection of the spectra. Indeed, sample MCL-PHL seems to have a much larger I4lFe3+ to Fe,". ratiothan any other sample, and sample HEA-PHL seems tohave much more I6lFe3+ than I41Fe3+. This illustrates thedifficulties related to strongly overlapping lines. The tarFe3'

shoulder in the spectrum of sample MCL-PHL is moreprominent visually only because the -0.1 mm/s line ismore narTow.

In conclusion, the fit results (Table 3) show a talFe3+

shoulder or line at 0.41 + 0.02 mm/s in the three samplesshowing such a contribution. This is significant in thatalmost ideal end-members phlogopite and annite are in-cluded, suggesting that any trioctahedral true mica con-taining t41Fe3t should have a room-temperature contri-bution at this well-defined velocity.

Another result from the fits is that the I+tFer* distribu-tion of quadrupole splittings is very nalrow comparedwith the octahedral sites. In fact, our results indicate anabsence of distribution broadening (q = 0 with a normal"y - 0.2 mm/s). This is consistent with the much smallerlocal-environment viability in tetrahedral sites comparedwith octahedral sites. On tetrahedral sites the near neigh-bors are four O atoms, whereas in octahedral sites theyare four O atoms and 2 OH- groups in either cis or transarrangements and with defects such as H vacancies andOH- being replaced by F or Cl-. The further neighborenvironments are also more variable in the octahedralsites, with a great variety of possible octahedral cationscompared with almost only Sio*, Al3*, and Fe3* in thetetrahedral sites.

This sharpness makes the I4lFe3+ shoulder at 0.41 mm/smore observable than it otherwise would be. It makes thehigh-energy tetrahedral line distinct and recognizable de-spite its overlap with the absorption at --0.1 mm/s.Excellent statistics (e.g., S/N > 50) and a high channel-wise resolution (e.g., channel-width < 0.015 mm/s) arenonetheless desirable if the presence of t4lFe3r is to beascertained unambiguously.

Finally, we comment on "fit quality." The reduced chisquared (1|j values are given in Table 3. The fits to theroom temperature spectra (Table 3 but not shown) were

hl(h3 + h4)v5h5v1 BG

h"lx?n hJ> h, (h" + h.) h,l2 h,

2.4252.3762.4502.4032.3862.4112.5812.610

0.949o.7400.84*0.8410.8301.0670.906

0.205 13.50.278 28.50.388 9.80.249 51.90.192 14.30.254 16.40.185 17.8

13.5 0.495

0.241 6.8340.262 1.7570.280 2.2600.214 1.3370.220 2.2180.236 0.8250.261 1.9130.237 0.730

1.22o.s2 0.073 0.2830.59 0.055 0.1800.27 0.038 0.114o.s7 0.091 0.3000.49 0.045 0.1390.49 0.066 0.2360.53 0.048 0.142

0.03{1 0.0920.057 0.2230.065 0.213

0.054 0.195

26.9 0.42610.6 0.39733.7 0.391

0.0000.0000.052

0.000

40 RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS

TAELE 4. Previously reported t4rFe3+ in room temperature spectra of micas

%tet. tet.3./Dyar's no. layer total AJ,\. 6, 4 l. H, oct.2+ oct.3+ Reference

0.40o.420.04

0.500.290.26

2P3P4L6P7P

33854P55P59C60c62C6iilc78479A80A81482P87P88P

13BB145B

4.03.50.32.55.03.03.60.91 . 31 . 53.46.54.5

10.616.413.85.9

0.40o.420.050.500.500.390.370.090.500.550.770.870.080.210.230.180.38

0.540.550.760.870.030.160.200.180.621.00

0.200.24o.250.260.150.260.190.210.270.24o.26o.240.200.19o.21o.200.19o.170.190.13o.32

0.520.470.350.620.630.450.440.440.620.680.650.800.370.390.400.450.560.500.570.540.58

-0.06-0.005+0.075-0.05-0.165+0.035-0.03-0.01-0.04-0 .10-0.065-0 .16+0.015-0.005+0.01-0.025-0.09-0.08-0.09s-0 .14+0.03

0.460.475o.4250.570.4650.4850.410.430.580.580.5850.640.3850.3850.41o.4250.47o.420.4750.400.61

yesyesyesnoyesyesyesyesyesyesyesnoyesyesyesyesyesnoyesyesyes

nononoyesnononoyesyesyesnoyesyesyesyesyesnononoyesno

Shinno and Suwa (1981)Shinno and Suwa (1981)Levillain et al. (1977)lshida and Hirowatari (1980)lshida and Hirowatari (1980)Sanz et al. (1978)Hogarth €t al. (1970)Hogarth et al. (1970)Annersten and Olesch (1978)Annersten and Olesch (1978)Annersten and Olesch (1978)Annersten and Olesch (1978)Dyar and Burns (1 986)Dyar and Burns (1 986)Dyar and Bums (1986)Dyar and Bums (1986)Dyar and Burns (1986)Annersten et al. (1971)Huggins (1976)Pollak and Bruyneel (1974)Vertes et al. (1981)

of the same visual and statistical quality as the fits (Fig.5) to the liquid N, temperature spectra. The somewhatlarger 1fu value for sample MCL-PHL at room temper-ature may be due to a very small amount of t6lFe3+. Weare investigating this possibility in a detailed study ofsample MCL-PHL (Rancourt et al., in preparation).

Comparing the Miissbauer results toformula recalculations

Our Mdssbauer analysis shows that samples MCL-PHL,HEA-PHL, and ROM-ANN have -5-l0o/o of their totalFe as talFe3*, whereas the biotite samples HEP-BIO, BIS-BIO. and MOC-BIO have no t4lFe3+.

By comparison, our formula recalculations from themicroprobe analyses, based on all Fe as FeO, stronglyindicate larFe3+ only in samples MCL-PHL and ROM-ANN, which have the largest total amounts. These twosamples have octahedral sums significantly larger than 6and tetrahedral sums significantly smaller than 8 (Table2), as is expected from ignoring larFe3+. The three biotitesamples that have 1e l+lfsr+ detectable in their Mdss-bauer spectra have tetrahedral sums of 8 and octahedralsums that are significantly smaller than 6. The latter isusual and probably due to octahedral vacancies and oc-tahedral ions that were not analyzed.

Sample HEA-PHL has 2*, : 6.0 and 2*, : 8.0. Thissample does not have enough t4lFe3+ to make its presenceobvious in the recalculation; however, its ideal octahedralsum is, in hindsight, suspect and might be used as anindicator of talFe3+. This shows that, whereas the presenceof I6lFe3+ is usually impossible to ascertain from formularecalculations, the presence of I4lFe3+ is more easily rec-ognized.

On the other hand, in room-temperature Mdssbauerspectra, Irlps:+ sntt more easily be missed than I6lFe3*. For

example, Hargraves et al. (1990) studied the same single-crystal wafer of sample HEA-PHL and did not recognizethe tetrahedral shoulder. This is due to small experimen-tal differences (better MCS electronics and smaller chan-nel width in the present study) and to not knowing whatto look for. The present paper shows exactly what to lookfor. Compare the spectrum for sample HEA-PHL in Fig-ure I to Figures 2 and 4 of Hargraves et al. This suggeststhat true t4lFe3+ may have been missed frequently in paststudies.

CovrpanrsoN wrrH PREvIous woRK

Dyar (1987) has reviewed the room-temperature MOss-bauer work on trioctahedral micas. Of the 130 samplesreported by Dyar, 3l samples are claimed by the respec-tive authors to show a distinct spectral component duetO talFe3+.

On close examination, we find that in at least five ofthe latter samples (samples lB, 98, 2lB, 77B., and I l9Bin Dyar's list) the t41Fe3* component actually corresponds1s lolfs. Also, a group of spectra from samples studied byManapov and Sitdikov (1974) and Manapov and Krinari(1976) (samples 48B., 498,51B, 128B, and l29B) seemto have a systematic calibration error in absolute velocitywith respect to metallic Fe (or in any case are of verypoor quality) and will therefore be omitted from furtherconsideration.

Information concerning the remaining 2l samples isgiven in Table 4. Here each sample is labeled with Dyar'ssample number that contains a letter indicating the typeof mica as classified by the original authors (P : phlog-opite, B : biotite, A : annite, C : clintonite, and L :

lepidolite). The other quantities in Table 4 are defined asfollows: 0/o tet. layer is the claimed percentage of the tet-rahedral sites that contain Fe3*: tet.3*/total is the claimed

a aa a o a

a a -

a oo

I

o o o o

@ @ @@ o o

o @ o o o

o o

o o o

o o

@

RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS

o.2

-o.2

4 l

0 .10.24

G-\trtr 0.18

qLF 0.o

jd o.''z

0.06

0.000.10 0.40 0.55 0.70 0.85

Ar (-./")

Fig. 6. Previously reported values of room-temperature cen-ter shifts (with respect to a-Fe) vs. quadrupole splittings of sup-posed tetrahedral doublets in trioctahedral micas. Such d-A plotsare often used but do not justly reflect the large uncertaintiesrelated to hidden lines. Points in our brittle mica field are rep-resented as filled triangles. Those in our tme mica field are rep-resented by filled squares. The points taken from Dyar's recentstudy (Dyar, 1990) are represented by open symbols. Not all ofDyar's points are visible because of overlap. The filled symbolparameters are given in Table 4.

fraction of the Feb, that is t41Fe3+ (from chemical analysisor combined chemical analysis and Miissbauer); A,.,/A,o,is the ratio of the spectral area of the t4lFe3+ quadrupoledoublet to the total Mdssbauer spectral area; 6, and 4 are,respectively, the center shift with respect to metallic Feand the quadrupole splitting of the assume6 t+lps:+ quad-rupole doubleq I, and H, are, respectively, the low- andhigh-energy line positions of the quadrupole doubletscorresponding to D, and A, (i.e., L,: 6, - L,/2 and Hr :6, + Ar/2); oct.2+ indicates whether the spectrum showedt61Fe2+' and oct.3+ indicates whether the spectnrm showedany I6lFe3+.

Whereas a surprising diversity of (0,, A,) values is re-ported (Table 4 and Fig. 6), much of this diversity is seento be an artifact when the corresponding line positionsthemselves (I, and H,) are examined (Fig. 7). This is be-cause, most often, only 11, corresponds to a visible spec-tral feature (line or shoulder), whereas Z, corresponds toa contribution that is hidden under the strong t61Fe2+ andt61Fe3+ absorption at - -0. I mm/s. Indeed, 11, is found tobe meaningful, whereas Z, has a broad range of valuesfrom approximately -0.2 to *0. I mm/s.

This is seen in Figure 7, which shows two narrow fieldsof fI, values: one at 0.38-0.49 mm/s corresponding totrioctahedral true micas in agreement with our results (-Fl,: V,:0.41 + 0.02 mm/s; Table 3) and one at 0.56-0.65mm/s corresponding to trioctahedral brittle micas (clin-tonite). The latter field contains all the clintonite samplesreported in Dyar's review (samples 59C, 60C, 62C, and63C) and two other samples identified by the original

0 .1 o.2 0.3 0.4 0.5 0.6 0.7

Ht (mmls)

Fig. 7. I"-H, plot, as an alternative to the 6t-4 plot (Fig. 6).Here I, is the position of the low-energy tetrahedral line (l : 6,- L,/2) and Il, is the position of the high energy component (fI,: 6. + A,/2).I1, is well separated into true and brittle mica fields.The symbols have the same meanings as in Figure 6. The new"field," at fI, - 0. 15-0.35 mm/s, rhat appears on plotting Dyar'sdata (Dyar, 1990) is most probably not real, as explained in thetext.

authors as tnre micas (6P and l45B). In a d-A plot (Fig.6), the true and brittle fields have widely overlapping val-ues ofboth 6, and A,.

In addition, on examining Table 4 and Figure 7, wefind that other trends also seem to exist: (l) annite seemsto have lower 11, values than phlogopites, especiallyphlogopite that does not containt6lFe3*, and (2) biotitecontaining talFe3* seems to be relatively rare comparedwith annite and phlogopite. Both the above points aresupported by the present study of six micas.

Figures 6 andT also show points (open symbols) of therecent work by Dyar (1990) involving 52 biotite samplesfrom metapelites, all reported to contain I4lFe3+. OnIy forusamples are within our true mica field (samples O-K-I5,O-K-53, O-C-26, and OBl6);one is not visible becauseof overlap of points. One sample (Ra-d37-66) lies directlybetween orrr true and brittle mica fields. The spectra forthe latter samples are not shown by Dyar. Only one spec-trum is shown (sample O-Ll0), which in our opinionhas no detectable I4lFe3+ since it lacks the characteristicshoulder at -0.4 mm/s or higher.

The tetrahedral doublet in this spectmm (O-L-10) isseen as consisting of two weak and highly overlappinglines that are completely buried in the main absorptionline centered at - -0.1 mm,/s. They do not give rise to anoticeable spectral feature that might be used as evidencefor their existence.

In all, 47 of Dyar's (1990) samples have tetrahedraldoublet parameters similar to those of O-L-10 and pre-sumably have similarly hidden lines. This group of sam-ples gives rise to a new field (Figs. 6 and 7, open circles)that is as far removed from that of the true micas as is

- u . I

oo

o oo o

q o

o \0. q-

o l o o I

o o o o o I^ h

o o o ! :o i

t ,

r l

42 RANCOURT ET AL.: TETRAHEDRAL FE3* IN TRIOCTAHEDRAL MICAS

that of the brittle micas-only in the opposite direction.In our opinion, this new field is an artifact ofoverfittingand is not real.

Dyar (1990) claims that the new tetrahedral parametersare in agreement with previous work; however, they areclearly and systematically different from all previouslyreported values (Figs. 6 and 7). Whereas quadrupolesplittings of -0.40-0.55 mm/s have mainly been report-ed for trioctahedral true micas, Dyar's biotite samples arereported to have an average of (A,) : 0.25(12) mm/s (Fig.6; Dyar, 1990). Such a difference is difficult to under-stand. The quadrupole splitting of an Fe3* ion arises fromthe lattice point-charge contribution only, and Dyar shouldexplain how the latter can be so small in her biotite sam-ples, compared to all other trioctahedral micas.

In addition, Dyar (1990) describes how, given presentunderstanding of the crystal chemistry of micas, her bi-otite samples should not contain I4lFe3+. This unusualt+lfs:+, whose presence is not directly detected in thespectra (unlike the t61Fe2+ and t61Fe3*), was not corrobo-rated by an independent measurement. Also, a distinc-tion is not made between the 47 samples that have theunusual parameters and the other five whose parametersare stepwise different and more usual.

Dyar (1990) has fitted each spectrum in several differ-ent ways ("In some samples 30-50 different models weretested . . .") and has found a group of fits that give aconsistent interpretation in the sense that the differentlines are always present in about the same positions. Thisis, in our experience, no reason to believe in weak andhidden lines, especially when such lines are inconsistentwith present knowledge and thereby require an ad hocand specialized explanation.

Dyar's new tetrahedral parameters are probably an ar-tifact ofimproper analysis and interpretation. A plausiblemechanism for this artifact is as follows. Slight absorbertexture, as is virtually impossible to avoid with micas,will shift intensity from the high-energy I6lFe2* line to thelow-energy telFe2* line, thereby requiring extra intensityin a broad neighborhood of --0.1 mm/s when sym-metric quadrupole doublet models are used (as was thecase). This error will occur in about the same way insimilar samples and will lead to false lines in about thesame positions and of about the same relative intensities.This is what was observed by Dyar, with a tetrahedraldoublet corresponding to a nearly constant fraction (-80/0)of the Fe,",. When possible texture effects, thickness ef-fects, and non-Lorentzian line shapes resulting fromquadrupole splitting distributions are all ignored, thensuch small and systematic false spectral contributionsshould be expected.

CoNcr,usroNs

fhs [+lps:+ in true trioctahedral micas gives a room-temperature quadrupole doublet having center shift 6, =0.17 mm/s (with respect to a-Fe) and quadrupole split-ting A, * 0.50 mm,/s (Annersten et al., 197l). In naturalsamples, this leads to a low-energy line whose position(tr, : 6, - L,/2 = -0.08 mm/s) is such that it is com-

pletely buried under the large absorption line mainlycaused by tolpsz+ and centered at --0.1 mm/s (Fig. 2).Consequently, in most cases reliable information cannotbe obtained concerning the low-energy tetrahedral lineexcept that it is hidden, i.e., it lies somewhere between-0.2 and *0.1 mm/s (Fig. 7). On the other hand, thehigh-energy tetrahedral line has a position (11, : 6, + A,/2= 0.42 mm/s) that potentially makes it resolvable from[hs lelpsz+ and r6lFe3+ lines. The 11, line is indeed resolv-able (largely thanks to its relatively small width) and canbe used to recognize unambiguously the presence of (andto quantify the amounts of) tatF":+ in natural samples.

Brittle trioctahedral micas have 11, = 0.6 mm/s (Fig.7) such that, assuming about the same center shift as intrue trioctahedral micas (6, = 0.2 mm/s), we expect Ar =

0.8 mm/s and L, = -0.2 mm/s-again making the lowenergy line unobservable in most cases. The larger d val-ue for brittle micas is consistent with the larger interlayercharge.

In comparing different natural samples, only f/, valuesare dependable; separate 6, and A, values are not sincethese are uncertain to the extent that L,is uncertain. Twofields of 11, values seem to exist: a true trioctahedral micafield at 0.38-0.49 mm/s and a brittle trioctahedral micafield at 0.56-0.65 mm/s. Our samples, including near end-members phlogopite and annite, gave H,: 0.39-0.43mm/s. It is difficult to assess how much of the field widthsare actually due to lab-toJab calibration differences.

Supposed tetrahedral contributions having H, > 0.7mm/s are suspect because of the known I6lFe3+ line at 0.8-1.2 mm/s. They are most probably incorrectly identifiedcontributions. We do not expect that the tetrahedral sitescan produce such large quadrupole splittings.

Supposed tetrahedral contributions having H, < 0.35mm/s, such as recently reported by Dyar (1990) for bio-tite samples, are also suspect because (l) in this case, bothtetrahedral doublet lines are essentially hidden by thebroad and mainly t61Fe2* absorption at - -0.1 mm/s and(2) such values are significantly outside the above-de-scribed fields that are based on several samples in whichthe presence sflalfgr+ is often corroborated by indepen-dent observations such as the occurrence ofreverse ple-ochroism. We have shown that Dyar's new tetrahedralparameters are most likely an artifact of improper anal-ysis and interpretation.

We stress that it is important to rely on visible spectralfeatures (the sharp room temperature shoulder at -0.41

mm/s, e.g., Figs. 3 and 4), rather than numerical resultsfrom multiline fits using arbitrary constraints and as-sumptions, to identify positively the presence or absenceof t41Fe3+. The visible shoulder can be enhanced by suchaids as obtaining spectra at low temperatures, using thesmallest practical channel width (or region of interestscans), and acquiring the best possible statistics.

AcxNowlnncMENTs

We thank C. Robinson of the Canadian Museum of Nature for supply-ing specimen MOC266I and J.A. Mandarino of the Royal Ontario Mu-seum for specimenM42l26. We thank D.D. Hogarth for help in locating

RANCOURT ET AL.: TETRAHEDRAL Fe3* IN TRIOCTAHEDRAL MICAS 43

the McCloskey sample. Financial support from the Natural Sciences andEngineering Research Council of Canada to D.G.R. and A.E.L. is grate-fully acknowledged.

Rnrnnrxcns crrnoAnnersten, H., and Olesch, M. (1978) Dstribution of fenous and ferric

iron in clintonite and the Mdssbauer characteristics of ferric iron intetrahedral coordination. Canadian Mineralogisr, 16, 199-203.

Annersten, H., Devanarayanan, S., Hiiggstrdm, L., and Weppfing, R. (1971)Miissbauer study of synthetic ferriphlogopite KMg,Fes*SirO,o(OlI)r.Physica status Solidi B, 48, K137-K138.

Dyar, M.D. (1987) A review of Mtissbauer data on trioctahedral micas:Evidence for tetrahedral Fe3* and cation ordering. American Mineral-og;st,72, 102-112.

-(1990) M<issbauer spectra of biotite from metapelites. AmericanMinerafogist, 7 5, 656-666.

Dyar, M.D., and Burns, R.G. (1986) Miissbauer spectral srudy of femr-ginous one-layer trioctahedral micas. American Mineralogist, 7 l, 955-965.

Faye, G.H., and Hogarth, D.D. (1969) On the origin of 'reverse pleoch-roism' of a phlogopite. Canadian Mineralogist, lO, 25-34-

Hargraves, P., Rancourt, D.G., and Ialonde, A.E. (1990) Single-crystalMdssbauer study ofphlogopite mica. Canadian Journal ofPhysics, 68,128-144.

Hogarth, D.D., Brown, F.F., and Pritchard, A.M. (1970) Biabsorption,Mdssbauer spectra, and chemical investigation of five phlogopite sam-ples from Qu6bec. Canadian Mineralogist, l0,7lO-722.

Huggins, F.E. (1976) Mdssbauer studies ofiron minerals under pressuresof up to 200 kbars. In R.G.J. Sterns, Ed. The physics and chemistry ofminerals and rocks, p. 613-640. Wiley, New York.

Ishida, K., and Hirowatari, F. (1980) On the chemical composition andMiissbauer spectra of manganoan phlogopite with reverse pleochroism.Kobutsugaku Tasshi, 14 (Tokubet sugo 3), 54-61.

lalonde, A.E- (1989) Hepburn intrusive suite: Peraluminous plutonismwithin a closing back-arc basin, Wopmay orogen, Canada. Geology,17. 26t-264.

I-evillain, C., Maurel, P., and Menil, F. (1977) Localization du fer, par

voie chimique et par spectrometrie M6ssbauer, dans la l6pidolite dugranite de Beauvoir (Massif central frangais). Bulletin de la Soci6t6Frangaise de Min6ralogie et de Cristallographie, 100, l3'l-142.

Manapov, R.A., and Krinari, G.A. (1976) Gamma-r€sonance spectros-copy of layer silicates and some problems of their crystal chemistry. InV.M. Yinokurov, Ed., Fiz Svoistva Miner Gom Porod, p. ll0-122.lzdanja Kazat Universiteta, Kazan, USSR.

Manapov, R.A., and Sitdikov, B.S. (1974) Metamorphic rocks of the Pre-cambrian basement in the Tatar Dome. Geochemistry International,rr,976-980.

Pollak, H., and Bruyneel, W. (1974) Saut d'6lectrons et le rapport Fe'?+/Fe3* dans deux silicates. Joumal de Physique, Colloque, 35, C6-571-c6-s74.

Rancourt, D.G. (1989) Accurate site populations from M6ssbauer spec-troscopy. Nuclear Instruments and Methods in Physics Research, B44,199-210.

Rancourt, D.G., and Ping, J.Y. (1991) Voigf-based methods for arbitrary-shape static hyperfine parameter disributions in Miissbauer spectros-copy. Nuclear Insruments and Methods in Physics Research, B58, 85-97 .

Rancoun, D.G., I-amarche, G., Tume, P., Ialonde, A.E., Biensan, P., andFlandrois, S. (1990) Dipole-dipole interactions as source ofspin glass

behaviour in random ferromagnetic-layer compounds. Canadian Jour-nal of Physics, 68, 1134-1137 .

Sanz, J., Meyers, J., Vielvoye, L., and Stone, W.E.E. (1978) The locationand content ofiron in natural biotites and phlogopites: A comparisonof several methods. Clay Minerals, 13,45-52.

Shinno, I., and Suwa, K. (1981) Miissbauer spectrum and polytl'pe ofphlogopite with reverse pleochroism. Sixth Preliminary Report of Af-rican Studies, p. 15l-157. Nagoya University, Nagoya, Japan.

Vertes, A., Jonas, K., Czako-Nagy, I., and Nemecz, E. (1981) A combinedapplication of Miissbauer spectroscopy and chernical transformationfor chemical analysis. Radiochemical and Radioanalytical IJtters, 48,93-100.

MeNuscnrrr RECEIvED Ocronsn 4, 1990MeNuscnrrr AcCEPTED Aucusr l, 1991