On Li-Bearing Micas: Estimating Li from Electron Microprobe Analyses and an Improved Diagram for Graphical Representation

Home , Hendricksite, Siderophyllite

On Li-bearing micas: estimating Li from electron microprobe analyses and an improved diagram for graphical representation

GERHARD TISCHENDORF Neumannstr. 106, 13189 Berlin, Germany BXRBEL GOTTESMANN,HANS-JORGEN FORSTER, AND ROBERT B. TRUM]]ULL GeoForsehungsZentrum Potsdam, Dept. 4.2, Telegrafenberg A50, 14473 Potsdam, Germany

Abs~a~ Lithium may constitute an essential element in micas, yet it cannot be detected by the electron microprobe. Since Li is critical for correctly classifying micas and properly calculating their formulae, several methods have been proposed to overcome this analytical deficiency. We offer empirical relationships between Li20 and SiO2, MgO, F, and Rb in trioetahedral micas, and between Li20 and F as well as Rb in dioctahedral micas. The resultant regression equations enable lithium contents to be sufficiently well estimated from EPM analyses within the range of validity discussed. Secondly, we introduce an easy to handle, new diagram with the axis variables [Mg-Li] and [Fetot + Mn + Ti-AIva] for graphical representation and discuss its scientific rationale. Being based on absolute abundances of cations in the octahedral layer, the diagram provides a simple means to classify micas in terms of composition and oetahedral site occupancy, and it also allows compositional relationships between Li-bearing and Li-free mica varieties as well as between trioctahedral and dioetahedral micas to be displayed on a single, two-dimensional diagram.

KEVWORDS: mica, lithium, correlation, classification, electron microprobe.

Introduction because of their importance in evolved granites and related pegmatites and aplites. The basis of the work THE importance of micas for petrologic or metallo- is a compilation of several hundred Li-mica analyses, genetic studies of metamorphic and igneous rocks is many from sources in the literature and many others well established and has been the subject of a great from published and unpublished data sets of our own. deal of research (see Bailey, 1984, for an overview). Most of the latter represent micas from the The reason for their popularity is that micas form one Erzgebirge province, hosted in Varisean granites, or of the most common mineral groups, stable over a in a few cases from associated pegrnatites as well as wide range of pressure and temperature in rocks of from greisens in tin deposits. Analyses of mica from many kinds. They have a highly variable chemical metamorphic rocks of the Erzgebirge and of nearby composition and the ability to exchange components granodioritic and granitic complexes of Cadomian readily with fluids or solid phases as external and Varisean ages are also represented in the data conditions change. Being hydrous phases, micas are compilation (Fig. 1). sensitive to the fugacities of water, oxygen, fluorine Our objective is twofold. Firstly, we expand on the and other volatile species. The micas, therefore, are works of Stone et al. (1988) and Tindle and Webb useful as monitors of the pbysico-chemieal environ- (1990), and explore ways to indirectly estimate Li ment in which they grew. This aspect is particularly concentrations in both trioctahedral and dioetahedral important in the ease of granites, in many of which micas from electron microprobe analyses by micas are the only marie and hydrous phases present. empirical approaches based on element correlations. This paper is concerned with Li-bearing micas Secondly, we introduce a new diagram with the axis Mineralogical Magazine, December 1997, 1Iol. 61, pp. 809-834 Copyright the Mineralogical Society 810 G. TISCHENDORF ETAL.

+ + plutonic rocks ~-,,.Harzbur( ~....~B rogcken \ . + o p,ex

Atb Altenberg~ ~/4~~)" Eib Eibenstock Nbz Niederbobritzsch Pob Pobershau SIh Schellerhau Znw Zinnwald

FIG. 1. Simplified geological sketch map showing the sample localities of trioctahedral micas listed in Table 1.

variables [Mg-Li] and [Fetot 4- Mn 4- Ti-AlVI], of ferric and ferrous iron by stoichiometry because which provides a simple means to classify micas in site vacancies are common, and because manganese terms of composition and octahedral site occupancy can be present in the divalent and trivalent state. (trioctahedral and dioctahedral series), and which Therefore Fe2+ and Fe3+ must be treated together as also allows the compositional relationships between Fetot. The inability of the microprobe to analyse Li-bearing and Li-free mica varieties to be displayed lithium is a serious problem in evolved granites, on a single, two-dimensional diagram. aplites and pegrnatites, where Li is an essential Today, the vast majority of chemical analyses of component in both dioctahedral and trioctahedral mica composition for petrologic studies are micas (Tischendorf et al., 1969; Lapides et al., 1977; performed in situ by electron microprobe. Bulk Stone et al., 1988). Failure to analyse Li in micas analysis of mineral separates can be misleading or from this environment means a loss of important non-representative because more than one generation petrogenetic information as well as systematic errors of micas may be present in a given sample, and in calculating structural formulae (Tindle and Webb, because individual mica grains are commonly zoned, 1990). There are established methods of micro- especially in fractionated granites and related aplites analysis capable of analysing lithium, SIMS (FonteiUes, 1987; Monier et al., 1987; Hecht, 1993; (Henderson et al., 1989; Charoy et al., 1995) and Gottesmann et al., 1994a; Charoy et al., 1995). laser-ablation ICP-MS (Bea et aL, 1994), but they are Intergrowths, overgrowths and inclusions of other expensive and out of reach for most researchers; thus minerals in mica are other factors which pose no they do not represent practical alternatives to the great problem for the microprobe method but can electron microprobe at present. seriously affect the results of bulk chemical analysis. Fortunately, it turns out that lithium correlates One disadvantage of electron microprobe analyses strongly with several other elements in micas which of micas is that ferric and ferrous iron cannot be can be measured routinely by electron microprobe, distinguished. Micas are poorly suited for calculation and this offers a way to estimate Li contents from a Li-BEARING MICAS 811

microprobe analysis. Tindle and Webb (1990) Li utilized this in a study of element correlations in Li micas from a literature compilation, and they suggested the following equation to estimate Li from the Si content: Li20 = (0.287 x SiO2) - 9.552. nite~ These authors limit application of their equation to trioctahedral micas with less than 8 wt.% MgO. In this paper we take the same approach and examine element correlations in both trioctahedral and dioctahedral lithian micas separately, making a particular effort to find workable solutions for the ..~ \ ...... \ cases which Tindle and Webb (1990) found problematical, namely, the Mg-rich trioctahedral and the dioctahedral micas. OctahedralR 3+ (AI, 103+)+14+ R2+ (Ie 2+, In 2+, Mg)

Mica compositions and covariances with lithium Fro. 2. Relationship between Li and octahedral cations R 2+ (Fe 2+, MI12+, Mg) and R 3+ (A1TM, Fe3+) + Ti 4+ Trioctahedral micas. The most frequently applied (according to Foster, 1960b). nomenclature for trioetahedral micas was introduced by Foster (1960a, b), and is based on the octahedral site occupancy. Li-bearing micas are represented in a triangular diagram with the comer variables Li; R 2+ large and irregular, with a weak negative correlation (Fe2+, Mn 2+, Mg); and octahedral R 3+ (A1, Fe3+) + and considerable overlap among the three composi- Ti4§ (Fig. 2). tional series which is partly caused by the presence of We examine below the correlations of Li20 with A1TM and A1xa. The Li-AI series shows the greatest other elements in the octahedral layer (TiO2, A1203, range in composition, with A1203 contents from less FeOtot, MnO, MgO) as well as with SIO2, F, and than 15 wt.% (Bargar et al., 1973; Skosyreva and Rb20. As recognized by Tindle and Webb (1990), Vlasova, 1983; Uhlig, 1992) to over 25 wt.% SiO2 alone does not permit estimation of lithium over (Heinrich, 1967). Total iron-silica distributions the entire range of mica compositions. Thus we (Fig. 3b) are more regular than alumina-silica, but focussed on the search for elements alternative or there is complete overlap of the Mg-biotite-Fe- supplementary to silica. The data used were taken biotite and siderophyllite varieties. The diagram of from our own analyses and from the literature which MgO and SiO2 (Fig. 3c) shows the most regular met the following criteria: (a) Li was determined, (b) variation and good discrimination among the three the analyses are commonly not older than 25 years mica series, which has been recognized before by (to ensure comparable data quality), and (c) other workers (e.g. Tindle and Webb, 1990). Note compositions of concentrates appeared to represent that we use a logarithmic scale for MgO to better pure trioctahedral compositions not contaminated by illustrate the correlation with SiO2 over the extreme dioctahedral micas. compositional range of MgO. The important feature Figures 3-6 plot analyses of trioctahedral micas of this diagram is that the correlation between MgO (screened by the above criteria a-c) which were and SiO2 is negative in the Li-Fe and Li-A1 series taken from the literature, our own published data (< 6 wt.% MgO), and positive in the Mg-Fe series. (Tischendorf et al., 1969, n = 66; Gottesmann et al., This transition in behaviour has an effect on the Li 1994b, n = 4) and about 100 unpublished analyses distribution in the low-Mg and high-Mg micas, and (see Appendix). Several representative analyses from must be born in mind when one considers the the unpublished data are given in Table 1 (see Fig. 1 correlations of Li with Mg and Si (see below). for sample localities). Details on the petrology and The variation of Li20 with other chemical geochemistry of the rocks from which the micas were components of trioctahedral micas is examined in separated are given by Tischendorf et al. (1987) and Fig. 4a-f In most of these, there are ranges of Frrster and Tischendorf (1996). composition with different degrees of correlation or Figures 3a-c show the main compositional even reverses in the sense of correlation. Table 2 lists features of the micas. They form three compositional regression equations describing what we found to be series, labelled on the figures as: Mg-Fe series the best chemical correlations, and the range of (phlogopite, Mg-biotite, Fe-biotite, siderophyllite, compositions for which they are valid. The excellent lepidomelane), Li-Fe series (Li-bearing siderophyl- positive correlation of SiO2 and Li20 (Fig. 4a) was lite, protolithionite) and Li-AI series (zinnwaldite, already noted and used by Tindle and Webb (1990) lepidolite). The SiO2 and A1203 variation (Fig. 3a) is for calculating Li from a regression equation. The 812 G. TISCHENDORF ETAL.

35 AI20 3 (wt.%) G [] 30

[] [] Li-Fe ~ ~ series ~ ~ 25

\ ~E Leor • [][] ~o~~'~ ~ Li-AI" o,o 20

15 Mg-Fe [] .'1~ ~series \ [][] [] [] [] SiO 2 (wt.%) []

10 .... I ' ' '(~ I .... I .... I .... I .... I .... 30 35 40 45 50 55 60 65

35 FeOto t (wt.%) b 30

20 o~0 Li-Feseries

15 Mg-Fe %~ ~"~ •• series ~ [] 10 [] Li-AIseries

Orq ~ \ [] [] •\ [] 0 25 35 45 55 65 Li-BEARING MICAS 813 50 .0o c

10 [] ~ Mg-Fe series

-- ~ .... recommended cut-off line

0.1 t~ ~ ~ff~ t3 Li-AI Sel~eS

D O e EEIt]~ [] [] SiO z (wt.%) 0.01 .... , .... , .... , , , , D,l~rrlTl-1, , .... ~ .... 30 35 40 45 50 55 60 65

Fro. 3a-c. Variation of SiO2 with A1203 (a), FeOtot (b), and MgO (c) in trioctahedral micas. Dots mark analyses from the authors (n = 169), open squares mark analyses from the literature (n = 262 - 265).

figure demonstrates that this approach can only work Within the Li-A1 series, micas relatively poor in Li for micas with Li20 values greater than about 0.6 are rich in A1, whereas those strongly enriched in Li wt.%. For micas with less lithium, the positive are depleted in both A1vr and AITM. correlation breaks down and turns negative in the The variation of Li20 with total Fe is shown in concentration range below about 0.5 wt.% Fig. 4d, and this is a further excellent illustration of the (corresponding to approximately 35 wt.% SiO2 or 6 switch in sense and degree of correlation with changing wt.% MgO). In terms of MgO concentration (see Li concentration (of. Fig. 4a). Above about 0.8 wt.% discussion below), a workable value for the cut-off is Li20 there is a very good negative correlation of total 6 wt.%. Using this value, our data in the range of iron and lithium; at values between 0.8 and 0.2 wt.% MgO < 6 wt.% yields an essentially identical there is virtually no correlation, and below 0.2% a regression equation as the previous authors: positive correlation is observed. We have examined the correlations of Li20 with FeO and Fe203 separately Li20 = (0.289SIO2) - 9.658 (not shown) and found that the diagrams are similar to (R 2 = 0.912, n = 232) [tri 1] that of Fig. 4a but show more scatter. Furthermore, of Formally, this equation is valid only down to a course, microprobe analyses must be cast in terms of SiO2 value of 33.4 wt.%, where it cuts the abscissa. total iron and so the separate relationships of lithium In practice, we recommend application of this with ferric and ferrous iron are of no significance to the equation to micas with SiO2 exceeding 34 wt.%. question at hand. Micas with lower values of SiO2, of course, contain Titanium and magnesium both show a steady some finite amount of lithium. inverse correlation with lithium (Figs. 4e-j). In the The variations of Li20 with A1203 and with MnO TiO2-Li20 diagram this correlation breaks down (Figs. 4b-c) are quite irregular, particularly in the below about 0.1 wt.% Li20. In the case of MgO, case of MnO, and therefore of no value for predicting curve-fitting leads to the following expression which Li contents from microprobe analyses. Li20 and is valid for the full range of mica compositions: A1203 are positively correlated for members of the Li20 = [2.7/(0.35 + MgO)] - 0.13 Mg-Fe and Li-Fe series. In contrast, for micas of (R ~ = 0.880, n = 434) [tfi 2] the Li-A1 series two different trends are apparent. 814 G. TISCHENDORF ETAL.

~o O0 ~q~l ~" IC~" ~ ~rl .

m t..q 1~1

N ~g

~o~ ~r-i ~cSd~dcSc~q c5c5,~c5c5~ c5c5 c5~: ~ ~-~:O O

O m ~:O O O

tx:l ~~ ~; ~,~. r~

O q.., O

~ O et0oo e-, t~ N

O 2; o*

-o o

~~d~d~d~dd~~ o

[..., o .~ =~.~ -~ E E ~ O q ~ Li-BEARING MICAS 815

~ ~ ~~ .. ~.~ ~.. 8 g =

o s=

I=I 9^ ='8 ~ t"q "~ &

I=I ~d ,:5 ,:5 .. L ~ 2 +~,

+ o

etO0 .~ 0

~# o E ;..q

,,=P

,.~ o

~8 .s =~,.~ E E o. o o ~ 816 G. TISCHENDORF ET AL.

10 a Li 2 0 (wt.%) ~ -10-~176176176 _~d~ ~"~T~t~ Li-AI sedes

D D D .

- - ~ .... recommended cut-off line o~JE2~ [] Li-Fe series

0.1

~m\o~ D [] Mg-ge series 0.01 o\o []

SiO 2 (wt.%)

0.001 .... I ' ' ,17 I .... L .... I .... I .... I .... 30 35 40 45 50 55 60 65

10 b

Li20 o_ (wt.%) [] ~ Li-AI series r176 o [] :o # ~ [] Li-Fe senes

0.1 ~ nD o o OETO[] [] [] i~/ 0 [] 0 Mg-Fe series 0.01 ~/ [] / [] o [] [] [] AI203 (wt.%)

0.001 ,.,,,,,, [D. D , .... , .... , .... 10 15 20 25 30 35

FIC. 4a--f. Variation of Li20 with SiO2 (a), A1203 (b), MnO (c), FeOtot (d), TiO2 (e), and MgO (j) in trioctahedral micas. Dots mark analyses from the authors (n = 169), open squares mark analyses from the literature (n --- 256 - 267). Li-BEARING MICAS 817

C F[3~ -t-tZl~p-C" --D r-ID L",~o(.~) wt o ~,~149 ~o u-A, %N=~ ==~ '__ o[] senes

n ~~-.:lh Li-Fe [] n ~ q~ ~ .o series

0.1 []~[].%~.~

[] Mg-Fo [] / [][]o~} 0.01 sedes []/ o I [] [] / MnO (wt.%)

0.001 ...... 0.01 0.1 1 10

10 • D Li-AI series [] d

Li20 [] "~_~...~_~ []

Li-Fe ~j~, series D

0.t o

0.0t D ~ / [] Mg-Fe / [] series [] []

/ 6"~ FeOto t (wL%) 0.001 .13.['3, .... , .... ~ .... ~ .... , .... , .... 0 5 10 15 20 25 30 35

Fig. 4. Contd. 818 G. TISCHENDORF ETAL.

lO e B H~-* I%~-= D [] Li-AI ~%~.~o~,--,-, ~-i~;~l P. . ~o.o~ Li20 (wt.%) [] u ~~

Li-Fe senes. D ~

0.1

0.01 [] [] [] D I [] TiO2 (wt.%) / [] 0.001 ...... [m ...... 0.01 0.1 1 10

Li20 (wt.%) ~,B~:l

1 LiFe sefie~s~~~.D.

Mg-Feseries D~

0.01 qn y = [2. 7/(0.35 + x)] - O. 13 -~ DO MgO (wt.%) 0.001 0.001 0.01 0.1 1 10 100

Fig. 4. Contd. Li-BEARING MICAS 819

TABLE 2. Compilation of regression equations for estimating Li20 contents in tri- and dioctahedral micas

Equation Li20 (wt.%) Number of Goodness of fit Range of validity analyses (n) parameter (R 2) (wt.%)

[tri 1]* = (0.289"SIO2) - 9.658 232 0.912 SiO2 > 34; MgO < 6 [tri 2]* = [2.7/(0.35 +3.MgO)] - 0.13 434 0.880 MgO = 0.01 to 20 [tri 3]** 155*MgO 191 0.58 SiO2 > 34; MgO > 6 [tri 4] 0.237"F L544 501 0.852 F = 0.1 to 9 [tri 4a] = (0.697"F) + 1.026 62 0.70 aplites, pegmatites [tri 4b] 0.177"F 1"642 439 0.906 granitoids a.o. [tri 5] = 3.5"Rb20 L55 470 0.865 Rb20 = 0.01 to 1.7 [di 1]* = 0.3935"F 1'326 199 0.843 F = 0.01 to 8 [di 2] = 1.579"Rb20 L45 209 0.71 Rb20 = 0.01 to 2

* - recommended ** - recommended in combination with eq. [tri 1]

We note that some mica analyses from a few into two groups we obtain, for pegmatites and aplites occurrences (not plotted in Fig. 4j0 deviate from this using data from Cem~ et al. (1970, 1995), Rieder relationship. In these, at a given MgO content, the (1970), Bargar et al. (1973), Chaudhry and Howie concentration of Li20 is displaced towards either (1973), Skosyreva and Vlasova (1983), Edmunds et higher (Cern~, et al., 1970; CereS' and Trueman, al. (1985), Jolliff et al. (1987), Henderson et al. 1985; Neiva, 1981a; Henderson et al., 1989) or lower (1989), Neiva and Gomes (1991) and from our data values (Goeman, 1972; Rub et al., 1986; Nash, 1993; set: du Bray, 1994; Cern~" et al., 1995; Rieder et al., Li20 = (0.697F) + 1.026 (R 2 = 0.70, n = 62)[hi 4a] 1996) than would be calculated by equation [hi 2]. The reason for the deviations is not clear. It may and for the micas in granitoids and other rocks: include analytical problems (for example, the data of Li20 = 0.177F 1"642 (R 2 = 0.906, n = 439) [hi 4b] Henderson et al., 1989, are from SIMS analyses, which systematically overestimate Li, see Ottolini et Finally, Fig. 5b displays the positive correlation of al., 1993); and/or particular conditions of formation lithium and rubidium in trioctahedral micas. The (F-rich aunitic biotites from the Honeycomb Hills correlation is excellent except for micas with very rhyolite, Utah, apparently formed or reequilibrated at low values of Li20 and Rb20. The data can be fit by high F and O fugacity; see Nash, 1993 and Icenhower the following regression equation: and London, 1997). Li20 = 3.5"Rb20155 (R 2 = 0.865, n = 470) [hi 5] One of the problems encountered by Tindle and Webb (1990) was the poor correlation between Li Dioctahedral micas. Compared with the trioctahe- and Si at higher MgO compositions. For micas with dral micas, there are much fewer published analyses MgO greater than 6 wt.%, we obtained the following of dioctahedral Li micas and also fewer elements in expression relating Li20 and MgO: the micas for which a correlation with lithium can be tested. Our data compilation for the dioctahedral Li20 = 155MgO -31 (R 2 = 0.58; n = 191) [hi 3] micas includes, in addition to unpublished data from The chemical correlation between Li and F in our own and R. Naumann, published analyses for hioctahedral micas is good. This is well known, and micas designated as muscovite, Li-muscovite, Li- Tindle and Webb (1990) presented an equation phengite and AI-Li micas including mixed forms linking the two variables with R 2 of 0.617. Figure (see Appendix). 5a shows the relations of Li and F in trioctahedral No systematic variation of Li20 with either SiO2 micas. Use of all analyses results in the following or A1203 (not shown) was found, and we present equation: diagrams here only for Li-F and Li-Rb (Fig. 6a-b). Fluorine shows a good positive correlation with Li20 = 0.237F 1"544 (R 2 = 0.852, n = 501) [hi 4] lithium over the full compositional range, as was the Close examination of Fig. 5a shows that micas case for hioctahedral micas. The Li-F relationship from pegmatites and aplites define an array which can be defined by the following relation: differs in slope from that of the micas contained in Li20 = 0.3935F 1"326 (R 2 = 0.843, n = 199) [di 1] granitoids and other rocks. If we divide the data set 820 G. TISCI{ENDORF ET AL.

G Li20 (wt.%) Li-AI series [] [] / y = (0.697"x) + 1.026~• x

• LkFe xN x series X

0.1

-----.._ y = O. 177"xE1.642 • Mg-Fe series X/~ X"Xx..X>,~..~'" X F (wt.%) 0.01 4- 0.1 1 10

10 [] Li20 (wt.%) b _~~Li-AI series

1 Lei_rieFes" ~~u

0.1

Mg-Fe ~ ~ [] series

0.01 [] y = 3.5"xE1.55 [] [] [] [] [] Rb20 (wt.%) 0.001 , [] [] ...... 0.01 0.1 1 10

FIG. 5a-b. Variation of Li20 with F (a) and Rb20 (b) in trioctahedral micas. (a): Open squares denote aplites and pegmatites (n = 62). Crosses denote granitoids and other rocks (n = 439). (b): Dots mark analyses from the authors (n = 178), open squares mark analyses from the literature (n = 292). Li-BEARING MICAS 821

10 / a Li 20 (wt.%) Li muscovites H~~

Li phengites

0.1

muscovites/ .~ ~ o8 [] U IlllllltQ phengites / 0.01 O[ll t 'I IIII Illl I

/,~y = 0.3935"xE1.326

F (wt.%) 0.001 ..... o,, . 0.001 0.01 0.1 1 10

10 b Li muscovites ~

Li phengites 2 ~ =o 0.1 oy/@

149 [] muscovites/ / phengites 0.01 /~ y = 1.579"xE1.45 [] Rb20 (wt.%) 0.001 [] ...... 0.01 0.1 I I0

FIG. 6a-b. Variation of Li20 with F (a) and Rb20 (b) in dioctahedral micas. Dots mark analyses from the authors (n = 12), open squares mark analyses from the literature (n = 187 and 197, resp.). 822 G. TISCHENDORF ET AL.

The Li-Rb correlation is slightly worse, and can The classical triangular diagrams of Foster be expressed as follows : (1960a, b) meet the above criteria for most purposes, Li20 = 1.579Rb20145 (R 2 = 0.71, n = 209) [di 2] and they are widely used. However, for the Li-micas, Foster's representation has the disadvantage that it Summary. Tindle and Webb (1990) concluded that separates the Mg-Fe micas (phlogopite, biotite, Li can best be estimated from microprobe analyses of siderophyllite and lepidomelane) from the lithian lithium-bearing trioctahedral micas using the correla- varieties (protolithionite, zinnwaldite, lepidolite and tion with SiO2. Our compilation of Li-mica analyses Li-muscovite), and requires the use of two classifica- confirms this relationship and provides additional tion triangles with: Mg2+; Fe2+ + Mn2+; A1vi + Fe 3+ + information. Table 2 compiles all the equations we Ti; and Li; Mg 2+ + Fe 2+ + Mn2+; A1vI + Fe 3+ + Ti have discussed above which relate Li to other (see Fig. 2), respectively. Tr6ger (1962) also treated components which (with the exception of Rb), are the Li micas and the Mg-Fe micas as two separate routinely and reliably analysed by electron microp- series. This is a problem when one studies micas in robe. Although we have used a data set different from composite granite plutons, whose whole-rock compo- that of Tindle and Webb (1990), the equations for the sitions vary from fairly primitive to highly evolved. Li20-SiO2 relation are essentially identical and Both Li micas and Mg-Fe micas coexist in many of point to their universal applicability. In addition to these granites and there is appreciable Li even in the this, we offer other correlation equations, particularly latter, so that the treatment of these mica groups in for high-Mg micas with compositions outside the separate diagrams makes it difficult to see the range dealt with by Tindle and Webb. relationships between them. Rieder (1970) was the In our experience, equation [tri 1] gives the best first to combine the Li micas with the Fe-Mg micas results for micas with greater than 36 wt.% SiO2 and in one classification diagram, a tetrahedron with the less than 6 wt.% MgO. For micas outside this range comer points Mg (phlogopite), Fe 2+ + Mn2+ (annite), one can use either [tri 3] (for MgO-rich micas) or A1w + Fe 3+ + Cr 3+ (muscovite) and Li/R 3+;vI [Iri 2], which is applicable over the whole range of (trilithionite). In practice, this requires using four 2- MgO. Equation [tri 2] is recommended for micas dimensional projections. Also, taenolite and poly- with less than 36 wt.% SiO 2. lithionite lie outside the range of the diagram and Ti We present the first equation which quantifies the is not accounted for. Monier and Robert (1986b) also relationships between Li and F for Irioctahedral introduced a mica tetrahedron with the components micas alone, and is based on a comparatively large M 2+, Li, A1 and Si. Their scheme has the important number of analyses. The Li-F regression calculated disadvantage of combining Mg and Fe 2+ into one by Tindle and Webb (1990) incorporated data from variable. Gottesmann and Tischendorf (1978) both trioctahedral and dioctahedral micas, which is proposed a classification diagram based on a double the main reason for the relatively poor 'goodness of triangle with the following comer variables: Mg-Li fit' recognized. A good approximation for Li in (phlognpite), Li-Mg (near-lepidolite), AIw + Fe 3+ + trioctahedral micas can also be obtained if data for Ti (muscovite) and Fe 2+ + Mn2+ (annite). Drawbacks Rb are available. of this classification scheme are that Ti and Fe 3+ are Furthermore, we present equations applicable to assigned to the muscovite component, as in Foster dioctahedral micas, for predicting Li contents based (1960a, b), which is not geochemically justified, and on the F or Rb concentration, a group for which that ferric and ferrous iron are plotted separately. Tindle and Webb (1990) found insufficient data to CereS, and Burt (1984) proposed a vector representa- make a reliable assessment of element correlations. tion of mica composition which also incorporated Both equations can be used, however, the Li-Rb vacancies in the octahedral layer. Their emphasis was correlation is less good than Li-F. on micas in granitic pegmatites, and they considered only the components K-Li-Fe-A1-Si. This limits Mica classification and graphical representation application of this scheme to mica varieties with negligible amounts of Mg, Mn and Ti. Principles. Graphical representations of mineral We propose a new diagram which has particular compositions should be simple and practical to use, advantages for representing the composition of Li even for the non-specialist. They should allow micas. The diagram is based on the octahedral classification of the mineral species and also show cations in micas, and these are cast into four the important compositional changes of the minerals components according to geochemical reasoning in response to evolution of the host rocks (e.g. and the element correlations in natural micas which prograde metamorphism, igneous fractionation). There were discussed above: Mg, Li, A1vI, and Fetot + Ti + are a number of graphical schemes in use for the Mn. We draw attention to the fact that titanium is micas, but we f'md that none of these is ideally suited treated here as a member of the iron-group and not, for the study of Li micas, for the reasons given below. as usual, added to A1vI. The reason for this is that Li-BEARING MICAS 823

r-u-r," r-etot+tvtn+/, annite R v' cations of ideal VI / Fe6 ~ micas, end members - AI / I ~ and of micas of tran- /; NNNNIK sitiona, position

[ sidero- Fe biotite(l~ [ phylUte (I) ag2Fe4 J~proto.F~sAI / lithionite(I) Fe biotite(N) / LiFe4AIZ Mgl .sFe3.rSAI.... N

[ sidero- Mg biotite(N) Mg biotite(I) / phylUte(N) Ug3Fe2.sAlo.s Mg4Fo2 Fo4AI2 _~ protom- / lithionite (N) [ LIFo3AI2 pnlogopite 0 -- Mg6 -- / e ferriphengite taerl|ollta / Li2Fe2AI2 (MgFoIFe"AI2 Mg4Li2 8~ -1 Jl~ cryo - MgsAI I~ / phyllito Li phengite(p) phengite(B) alumino- alumino- / / Li3FeAI2 LiFeAI2 MgFeAI2 celadonite phlogopite / /polylJthionite Mg2AI2 Mg4AI2~ " -2 phengite (A) - ~i4AI2 Li phengite(z) (MgFe)AI3 ' ~" LiFetl 5AI2_~ ; trilithionite ' / LiaAI3 = -3 "~1"~[~ Li muscovite I / u,5~35 m'.l~gvite Mg - Li -4 i I i i J i i ~ / / / / / "4 "3 -2 -1 0 ' 1 ' 2 = 3 I ' ' 5

Fro. 7. Position of ideal mica end-members and selected intermediate compositions in the diagram [Mg-Li] vs. [Fetet + Mn + Ti-AlW]. See text for further details.

titanium correlates positively with total iron (R = siderophyllJte and protolithionite, two compositions 0.676, n = 201, own data set), but negatively with the are shown; one (I) is the ideal formula; the other (N) six-fold coordinated alumina (R = -0.895; n = 201). represents an average natural composition (vacancy- The variables on both the x- and y-axes are free). For phengites, two formulae are given (A, B). combinations of two components each, making it We have also added three intermediate compositions, possible to display the effects of four components Li phengite(p) at LiFeAI2, Li phengite(z) at (albeit not separately) on a two-dimensional diagram. LiFeo.~A12.5, and Li muscovite at LiEs,A13.5 to The axis variables are ~Mg minus Li] and [Fetot plus emphasize the natural transitions between muscovite Mn plus Ti minus AI*']. These take account of the and pmtolithionite, zinnwaldite and tri- to polylithio- need to show Mg and Fe variations separately and are nite, respectively. The composition of taoniolite based on the strong, constant negative correlations in (MgaLi2) is anomalous. It is a trioctahedral mica mica which we observe between Mg and Li on the which, because of the lack ofFe, Mn, Ti and A1, plots one hand, and Fetot and A1vz on the other. The at zero on the Y-axis. The same point would be parameter [Mg-Li] reflects the proportions of the occupied by a mica with Mg2Fel.25Alt.25 and Rw = phlogopite and polylithionite components, and the 4.5 (see Fig. 8 below), which has not been previously parameter [Fetot + Mn + Ti-A1vI] reflects the deseribed. Taeniolite apparently requires exceptional proportions of the annite and muscovite components. conditions of form~ation where lithium-rich alkaline This diagram has a number of useful features. fluids interact with high-Mg rocks (such as fenites and First, the full range of micas from Li-bearing to Li- carbonatites, see Ganzeeva, 1973 and Cooper et al., free can be plotted on the same diagram. Second, in 1995). Note that the hypothetical end-member east- contrast to triangular plots, it shows true and not onite is put in quotation marks in the figure because relative concentrations. Third, because it takes into natural 'eastonite' actually consists of a mixture of account all of the important octahedral cations, the phlogopite and serpentine (Livi and Veblen, 1987). total occupancy of the octahedral sites earl be read off It has long been known that the trioctahedral micas the diagram and relations between the trioctahedral can cover the whole range in composition from the and dioctahedral micas are easy to see. Mg-rich to the Li-rich varieties, but there are also Figure 7 shows the position of ideal mica transitions with the dioctahedral micas (Monier and compositions in the diagram. For Fe and Mg biotites, Robert, 1986b), which are favoured by the presence 824 G. TISCHENDORF ET AL.

k Fetot+Mn+Ti /~r[~ + theoreticaloctahedral - AI vI occupancy and R w Fe5iAI0.5 MgFes~ ,sollnes (~)

AI MgFe4,5 Mg2Fe4 AI0"5jI* "~"

F+O4A. LI~ Mg~e4 M~21:.:3-5 Mg3Fe3~

2 /LiFe325iFe35LiosMgo5 MgFe3 Mg2Fe3 Mg3Fa25 Mg4Fe2_.

1 /LiFedd~/" a 2 ~'o.~M~o:~ IdgA~e~.s' M~IFo2 I, % Mg~oFer~75% Mg~loFO,.s MgsFa

0 Li2 2 13 o2AI2 Lio.sMgo.5 MgFel.5 Fe1~25Mg3Fo Mg4Feo.75MgsFo0. 5 MO~

-1 Li3FeAI21~ILFe/Li2FoA I LiFeALiFeAI2Lio sMgo5 MgFe Mg2Feo5Mg3Fe025 Mg4AI MgsAI

-2 '-i4AI2LiaFoo.s]Li:tFe0 I LiFeos LiosMgosMOFeos M~AI2 Mg'3AI3 Mg'4AI2

-3 3 2 ~ ~'3\ '0.~1,o 0.5.",o,, V 3

-4 , -4 -3 -2 -1 0 1 2 3 4 5 6

FIG. 8. Theoretical octahedral occupancy in micas (Fe in octahedral occupancy = Fetot + Mn + Ti) and lines of constant octahedral sum (Rxa) based on the unit cell X2Y4_tZsOEo(OH,F)4 in the [Mg-Li] vs. [Fetot + Mn + Ti-A1vI] diagram.

of high Li (and/or perhaps F). To aid in recognizing below, and the reader is referred to Table 3 for a transitions between tri- and dioctahedral micas, compilation of the boundary values. Figure 8 shows all theoretical cation occupancies of Most phlogopites ([Mg-Li] > 4) are relatively the octahedral layer in the diagram, and we have aluminum-poor and plot in quadrant I, although there added lines of constant octahedral,vzSUm (RW). It is are also some Al-enriched phlogopitcs, which plot in customary to consider micas with R from 5.0 to 6.0 the field labelled alumino-phlogopite. The fields of as trioctahedral, and micas with R w near 4.0 as Mg and Fe biotites are divided by the value [Mg-Li] dioctahedral. We fmd it appropriate to fix a value for = 2 and, additionally, by a dotted line which R vi = 4.4 as the division between dioctahedral and corresponds to Foster's (1960a) criterion of whether trioctahedral micas (see also Monier and Robert, Fe or Mg is the dominant octahedral cation. The 1986a). lower end of the biotite field is defined by the Figure 9 shows the compositional range of trioctahedral/dioctahedral boundary at R w = 4.4. naturally-occurring, common mica species on the Lepidomelane was defined by Foster (1960a) at Fe 3+ proposed diagram. The compositional ranges for the > A1w and Mg < 0.3. The lepidomelane field in our trioctahedral mica species follow the suggestions of diagram is bounded by the condition [Fetot + M/l+ Foster (1960a, b). We have simply transposed the Mg Ti-A1 vT] > 4. The siderophyllite field crosses and Li values to fit our axis variables, so the existing quadrants I and IV, with a range of [Mg-Li] nomenclature remains unchanged. The lines at x = 0 values from -0.4 to 0.6. Like biotite, the lower end and y = 0 divide the diagram into four quadrants, of the siderophyllite field is defined by the which correspond to the Mg-Fe micas (quadrant I), trioctahedral/dioctahedral boundary. Protolithiortite Mg-A1 micas (quadrant II), Li-A1 micas (quadrant (Winchell, 1942; Foster, 1960b) is a mica species III) and Li-Fe micas (quadrant IV). Also shown is with relatively high lithium contents and Fetot + Mn + the line of R vI = 4.4, which divides the fields of the Ti > AIvI. Rieder (1970) suggested that the term be dioctahedral and trioctaheral micas. Details about the disused as a mica variety because there are no natural range of specific mica species are discussed briefly micas with the ideal composition of LiFe4A1. We Li-BEARING MICAS 825

-1

-2

-3

-4 -4 -3 -2 -1 0 1 2 3 4 5 6

Fro. 9. Compositional fields of natural trioctahedral and dioctahedral micas on the diagram of [Mg-Li] vs. [Fetot + Mn + Ti-AlW], with R w = 4.4 as the boundary between di- and trioetahedral micas. The dotted line shows the discrimination boundary between Mg and Fe biotite from Foster (1960a). Quadrants I, II, llI and IV correspond to the mica series Mg-Fe, Mg-AI, Li-A1 and Li--Fe, respectively. believe it is useful to retain a protolithionite field (see Mn + Ti-A1 ~] = 0 to 0.5, and some protolithionites also Gottesmann and Tischendorf, 1980) and suggest where this value is as low as -0.5. Lepidolites, which that protolithionite is better represented by the can be considered as a collective name for the most composition LiFe3AI2 (see Fig. 7). Zirmwaldites Li-rich micas including polylithionite, trilithionite (= are not only Li-richer than protolithionites, but they paucilithionite), and cryophyllite plot exclusively in are also generally Al-richer and plot in quadrant III. quadrant HI and are bounded from zirmwaldite by the There are, however, some ziunwaldites with [Fetot + condition [Mg-Li] < -2.4.

TABLE 3. Compositional range of mica varieties with respect to [Mg--Li] and [Feto t + Mn + Ti'AlW], and other important element relations

Variety Mg-Li Fetot + Mn + Ti-A1 w Element relations

Phlogopite 6.0 ... 4.0 2.0 ... 0 Mg >> Fetot Mg biotite 4.0 ... 2.0 2.0/4.0 ... 0 Mg > Fetot Fe biotite 2.0 ... 0.6 4.0 ... 0.4 Lepidomelane 2.0 ... -0.6 6.0 ... 4.0 / Fetot > Mg Siderophyllite 0.6 ... -0.4 4.0 ... 0.4(0) Protolithionite -0.4 ... - 1.4 4.0(1.75) ... - 1.0(0) Fetot > Li Zinnwaldite - 1.4 ... -2.4 1.75(-0.375) ... -3.0 Li ~ Fetot Lepidolite -2.4 ... -4.0 -0.375 ... -3.0 Li > Fetot Li phengite -0.4 ... - 1.4 (0)- 1.0 ... -3.0 Li ~ Fetot Li muscovite -0.4 ... -2.4 -3.0 ... -3.8 Li >> Fetot Muscovite 1.0 ... -0.4 -3.0 ... -4.0 A1vI >> Fetot 826 G. TISCHENDORF ETAL.

We fred it appropriate to define a field for Li analyses, quadrant U will not be treated here in more phengites in quadrant IH, to include dioctahedral micas detail. Theoretically, trioctabedral micas composition- with values of [Mg-Li] between -1.4 and -0.4 ally resembling alumino-Mg biotite should form in (analogous to the protolithionite boundaries). Micas of nature. However, according to the authors' knowledge, this composition are fairly common (see Lapides et al., natural occurrences are not yet known. 1977; Ublig, 1992). Note that the line at [Mg-Li] = Lapides et al. (1977) and Sun Shihua (1984) have -1.4, which divides the Li phengite field from already emphasized that transitions between diocta- zinnwaldite, coincides exactly with the value of R xa bedral and trioetabedral micas exist. We can best = 4.4 for the dioctabedral/trioctahedral boundary. This show the extent of this by plotting the sum of confirms the usefulness of both boundary values. The octabedral cations (R vI) against the axis variables Li pbe~ites are succeeded, at a value of [Fetot + Mn + [Mg-Li] and [Fetot + Mn + Ti-A1 vI] (Figs. 10 and Ti-AI ] < -3, by the field of Li muscovite. Micas of 11). Figure 10 shows that most of the trioctahedral the structurally 'mixed forms', which Foster (1960b, micas have R xa values between 5 and 6, with the Fig. 39) attributed to immiscibility between dioctahe- lowest values and the greatest spread at intermediate dral and trioctahedral mica, plot in this field. The values of [Mg-Li] between -2 and 2. Very 'mixed forms' cover the same range of Li concentra- prominent in the figure is a continuous series tions as zinnwaldite, and therefore we extend the field between lepidolite/zinnwaldite ([Mg-Li] = -3 to of Li muscovite out to a value of [Li-Mg] = -2.4. The -2, R vI = 6 to 5) and muscovite ([Mg-Li] = 0, R w = muscovite field is defined at [Mg-Li] > -0.4 and 4), which is explained by the experimental [Fetot + Mn + Ti-A1vt] < -3. It is bounded upward in demonstration of solid solution between muscovite the diagram by the large phengite field, which covers and zirmwaldite by Monier and Robert (1986b). nearly the entire dioctahedral portion of the Mg-A1 Figure 11 shows the extent of transitions in natural mica field (quadrant II) and extends slightly into mica compo_sitions between dioctahedral muscovite/ quadrants III and 1. Due to a lack of sufficient Li phengite (R w = 4, [Fetot+ Mn + Ti-A1 w] = -3 to

6.5 Ipl zi pro sid Fe bi Mg bl 7" phi 1 / o ~ .. o o

o ~176 5.5

o o

4.5 I0 trilliterature I dilunpubl, data [ [] di/literature 4

Ipl LI mu/Li phe mulphe Mg - Li 3.5 -4 -2 0 2 4 6

FIG. 10. Compositions of natural micas in terms of [Mg-Li] and octahedral occupancy (RVl). The continuous transition of dioctahedral micas between muscovite ([Mg-Li] = 0) and zinnwaldite/lepidolite ([Mg-Li] = -2 to --4) is discussed in the text. The number of trioctahedral micas plotted is 730, dioctahedral micas: 212. Abbreviations here and in Figs. 11 and 12:A1 phi = aluminophlogopite, Fe bi = Fe biotite, Li mu = Li muscovite, Li phe = Li phengite, lpm = lepidomelane, lpl = lepidolite, Mg bi = Mg biotite, mu = muscovite, phe = phengite, phi = phlogopite, pro = protolithionite, sid = siderophyllite, tae = taenolite, zi = zinnwaldite. Li-BEARING MICAS 827

5.5 u

4.5 /. ./***, D/ I,,~;~L~ ~ ~i-7+ ~-~ /-~ ., 4- -'-+T+/' S

4 + Fetot + Mn + Ti muscovite I phengite - AIV I 3.5 ...... , ...... -4 -3 -2 -1 0 e mu, partly phe I~ trans-lpl trend (L) o trans-zi trend (Z)] •trans-pro trend (P) +trans-sid trend (S)

Fro. 11. Compositions of natural micas in terms of [Fetot + Mn + Ti-A1w] and octahedral occupancy (RW). We emphasize the extent of intermediate compositions between dioetahedral muscovite/phengite at Rw = 4 and trioctahedral micas lepidolite (trend L), zinnwaldite (trend Z), protolithionite (trend P) and siderophyllite (trend S).

-4) and the trioctahedral micas lepidolite (trend L; Figure 12 shows the position of a wide variety of data from Foster 1960b; Lapides et al., 1977; Luecke natural micas. The groups of mica analyses are coded 1981; Jolliff et al., 1987), zinnwaldite (trend Z; data by a number to the following data sources: from Lapides et al., 1977; and our data, see Fig. 13), [1] Phlogopites and Mg biotites from marie igneous protolithionite (trend P; data from Lapides et al., rocks (mantle xenoliths, kimberlites, gabbroic rocks, 1977; Uhlig, 1992; du Bray, 1994; and our data) and lamprophyres, nephelinites, melilites), and from siderophyllite (trend S; data from Lapides et al. metamorphic rocks (M&ais et al., 1962; Boettcher 1977). We designate micas plotting along trend Z as and O'Neil, 1980; De Fino et al., 1983; Basu et al., Li-phengites (z), and those along trend P as Li 1984; De Kimpe et al., 1987; Wagner et al., 1987; phengites(p). Additional sources of data for Fig. 11 Malyshonok, 1989; Grew et al., 1990; Edgar, 1992; are given in the Appendix (column of Fig. 6). Kramer and Seifert, 1994; Harlow, 1995). [2] Mg biotites from granodioritic rocks (Neiva, 1981a; Bigi and Brigatti, 1994). Discussion and applications [3] Fe biotites from granitic rocks and contact Use of the proposed diagram (Fig. 9) requires, of metamorphic rocks (Miiller, 1966; Barri~re and course, that Li concentrations are either determined Cotton, 1979; Bea, 1980; Neiva, 1980, 1981b). directly or can be estimated by correlation with other Some of the analyses from Barri&e and Cotton constituents of a microprobe analysis (using the (1979) plot in the lepidomelane field. correlations outlined in this paper and by Tindle and [4] Lepidomelanes from nepheline syenite pegma- Webb, 1990). Note that very Mg-rich or Li-rich micas tites (Foster, 1960a), theralites (Mokhtari et al., will be correctly classified by the diagram even without 1985: 'annite-ferriannite') and from granitic rocks Li or Mg data, respectively, because the dominant of the Brocken pluton (Harz) and the K6nigshain cations defme the position of the mica varieties. massif (Lausitz) (see analysis 906 and 520 in The diagram can be used for the full range of mica Table 1) from the authors' unpublished data. compositions, except for those in which other [5] Siderophyllite-protolithionite-zinnwaldite-lepi- octahedral cations such as Zn (hendricksite) and dolite micas from evolved granitic bodies and tin Mn (masutomilite) play a dominant or essential role. deposits (Rieder, 1970; Rub et al., 1983; Sun Shihua, 828 G. TISCHENDORF ET AL. 6

-2

-4 -4 -2 0 2 4 6 10111 +[2] • At4] *[5] O[6] v[7] I 0[8] e[9] ~[101 *[11] #[12] 11113]

Fro. 12. Positions of the full range of natural micas on the diagram of [Mg-Li] ~'s.[Feto t Jr Mn -I- Ti-AlV[]. The numbers in the legend refer to data sources and are explained in the text.

1984; Henderson et al., 1989; Stone et al., 1988; du [9] Li phengites from highly fractionated granites Bray, 1994). (Lapides et al., 1977) and from rocks of the tin [6] Ziunwaldite and lepidolite from evolved granites deposit Ehrenfriedersdorf, Erzgebirge (Uhlig, 1992). and associated aplites and pegmatites (Chaudhry and [10] Muscovites from two-mica granites and Howie, 1973; Fonteilles, 1987; Jolliff et al., 1987; pegmatites (Mtiller, 1966; Lapides et al., 1977; Charoy et al., 1995; Charoy and Noronha, 1996). Henderson et al., 1989; Silva and Neiva, 1990; Note that the micas designated as trilithionite (= Neiva and Gomes, 1991; Neiva, 1992; (~ern~, et al., lepidolite) by Charoy et al. (1995) have [Mg-Li] = 1995; Grew et al., 1995; Rieder et al., 1995). -2.06 to -2.35 and therefore plot here as [11] Phengites from highly fractionated granites zinnwaldites. (Lapides et aL, 1977 'phengite-muscovites'). [7] Lepidolites from a rare-metal albite leucogranite [12] Phengitic muscovites from two-mica para- (Beauvoir granite), rare metal-enriched alkaline gneisses (Naumann's and authors' unpublished complexes, and pegmatites (Cern~, et al., 1970, data) and phengites from jadeite and mica-albite 1995; Cern~;, and Trueman, 1985; Monier et al., rocks (Harlow, 1995). 1987; authors' unpublished data and analysis 1015 in [13] Taeniolite from alkali-metasomatic zones Table 1). (fenites) (Ganzeeva, 1973; Cooper et al., 1995). [8] Li muscovites from two-mica granites, and The goals of a particular study may make it useful pegrnatites (Foster, 1960b; Lapides et al., 1977; to use only a portion of the proposed diagram or Luecke, 1981; Charoy et al., 1995; Charoy and change the axis scaling. As an example, Fig. 13 Noronha, 1996). Charoy et al. (1995) classified some shows the Li-rich portion of the diagram to illustrate of these as phengites (cores in zoned grains) but we the variation of trioctahedral and dioctahedral micas would term them Li muscovites because they have in the multi-phase Variscan Eibenstock granite, [Fetot + Mn + Ti-A1vx] below -3.2. western Erzgebirge, Germany (el. Fig. 1). The Li-BEARING MICAS 829

4 Fe4.,,++Mn+Ti-AI vl / I ..I L~,L / I staero- Fe / I phyllite~ bio- / : .', ,,,. /" p.fot.o-..roto--- 2 ~ lithionite ~

zinn- +" Lt pnengite I waldite ~:~:~ ~,6~ I phengite lepi- "~ .a~b~"lCQ I I -2 dolite ~J~~

-4 , ~ : ~~Mg-Li -3 -2 -1 0 1 t EIB02/tri q-EIB1/tri • r162 #A-EIBItri I EIB02/di O EIBlldi ~3EIB2/di [2 EIB31di ~tA-E B/di FIG. 13. Composition of micas from the Variscan Eibenstock granite, westem Erzgebirge, shown in a portion of the diagram [Mg-Li] vs.[Fetot+ Mn + Ti-AlXa]. The samples are coded according to the granite subintrusions in which they occur (EIB0 to EIB3 are least evolved to most evolved facies and A-EIB is aplite), and according to trioctahedral (tri) or dioctahedral (di) composition. The arrows indicate evolution of mica compositions due to magma fractionation. Analyses by electron microprobe, Li-eontents estimated using equations [tri 1] and [di 1] in Table 2.

micas were analysed by electron microprobe and mica compositions from muscovite/phengite through their Li contents were estimated from equations Li phengite towards zinnwaldite observed in the [tri 1] and [di 1] of Table 2. The Eibenstoek granite Eibenstock granites serves as a natural example for constitutes one of the most evolved tin granites in the the experimentally proved continuous solid solution province and it contains a number ofpetrographically between muscovite and zinnwaldite (Monier and and chemically distinct subintrusions. These are Robert, 1986b). designated in the legend of Fig. 13 and range in Finally, we note that the X-axis variable [Mg-Li] order from the least-evolved facies (EIB0) to the is a good parameter to use as a parameter for mica most evolved (EIB3), and to the aplites (A-EIB). The composition when plotting other chemical variables trioctahedral micas show a regular progression from (e.g. Sn, Cs) or physical properties (e.g. optical siderophyllite to zinnwaldite in the least evolved refraction). The full range of [Mg-Li] values from 6 facies to the aplites. Petrographic evidence suggests to -4 encompasses the trioctahedral micas from that these micas are of magmatie origin and have phlogopite to lepidolite, and the dioctahedral micas been formed late in the crystallization history. Late are covered in the range from 1 to -1.4. magmatic, and early metasomatic (autometasoma- tism) processes in these granites produced dioctahedral micas, and these show a parallel development, in Acknowledgements the different subintrusions, from Li-poor phengites Our database of mica analyses is the product of many and muscovites to Li phengites(z) which nearly reach years of collaborative work. We wish to acknowledge the zinnwaldite field. The Li-rich phengites in the in particular G. Friese (wet chemistry) and R. EIB3 intrusion are hard to distinguish microscopi- Schindler (XRF) from Berlin for their contributions cally from the trioctahedral zinnwaldites with which in the years 1968 to 1981. R. Naumann (Potsdam) they coexist. The continuous series of dioctahedral performed or supervised XRF analyses from 1982 to 830 G. TISCHENDORF ET AL.

1996 and also provided numerous unpublished amphiboles and micas: evidence for metasomatism analyses of muscovites. L. Hecht (M/inchen) and in the mantle source regions of alkali basalts and M. Geisler (Leipzig) are thanked for INAA data. D. kimberlites. Amer. J. Sci., 280-A, 594-621. Rhede and O. Appelt (Potsdam) assisted with Bokonbaev, K.D. (1976) Pecularities of metasomatic electron microprobe work. U. Heitmarm (Berlin) alteration of biotite in granites of the Sukhodol'sky helped with statistical analysis. D. Wolf (Freiberg) massif (southeast Kirgizia) (Russ.). Zap. kirg. Otd. granted permission to use unpublished mica analyses vses. mineral Obshch., 9, 96-100. from the Ph.D. thesis of J. Uhlig. Finally, the senior Borodanov, V.M. (1983)Peeularities of biotite composi- author wishes to thank the director of the tion in granitoids associated with tungsten miner- GeoForschungsZentrum Potsdam, R. Emmermann, alization (Russ.). Izvest. AN SSSR, Ser. geol., 7, for granting him a research stipend and office space 76-81. while this study was completed. An anonymous Brigatti, M.F. and Davoli, P. (1990) Crystal-structure Mineralogical Magazine referee provided construc- refmements of 1M plutonic biotites. Amer. MineraL, tive comments which led to an improvement of the 75, 305-13. manuscript. Cem#, P. and Burt, D.M (1984) Paragenesis, crystal- This work is dedicated to the memory of Hertha lochemical characteristics, and geochemical evolu- Tischendorf, whose careful preparation of mica tion of micas in granite pegmatites. In Reviews in separates during several years of collaboration laid Mineralogy, 13, Micas (S.W. Bailey, ed.), Mineral. the foundation for much of the analytical work on Soe. Amer., 257-97. which this paper is based. Hertha passed away in CereS,, P. and Tmeman, D.L. (1985) Polylithionite from 1986 as a result of political repression in the German rare-metal deposits of the Blachford Lake alkaline Democratic Republic. complex, N. W. T., Canada. Amer. Mineral., 70, 1127-34. (~em~,, P., Rieder, M. and Povondra, P. (1970) Three References polytypes of lepidolite from Czechoslovakia. Lithos, A1-Saleh, S., Fuge, R. and Rea, W.J. (1977) The 3, 319-25. geochemistry of some biotites from the Dartmoor CereS,, P., Stanek, J., Nov~tk, M., Baadsgaard, H., granite. Proc. Ussher Soc., 4, 37-48. Rieder, M., Ottolini, L., Kavalovh, M. and Bailey, S.W. (Ed.) (1984) Micas. Reviews in Chapman R. (1995) Geochemical and structural Mineralogy, 13, Mineralogical Society of America, evolution of micas in the Rozml and Dobr~i Voda Chelsea. pegmatites, Czech Republic. Mineral. Petrol., 55, Bargar, K.E., Beeson, M.H., Fournier, R.O. and Muffler, 177-201. L.J.O. (1973) Present-day deposition of lepidolite Charoy, B. and Noronha, F. (1996) Multistage growth of from thermal waters in Yellowstone National Park. a rare-element, volatile-rich microgranite at Amer. MineraL, 58, 901-4. Argemela (Portugal). J. Petrol., 37, 73-94. Barri~re, M. and Cotton, J. (1979) Biotite and associated Charoy, B., Chaussidon, M. and Noronha, F. (1995) minerals as markers of magmatie fractionation and Lithium zonation in white micas from the Argemela deuteric equilibration in granites. Contrib. Mineral. microgranite (central Portugal): an in-situ ion-, Petrol., 70, 183-92. electron-microprobe and spectroscopic investigation. Basu, A.R., Rubury, E., Mehnert, H. and Tatsumoto, M. Eur. J. Mineral., 7, 335-52. (1984) Sm-Nd, K-Ar and petrologic study of some Chaudhry, M.N. and Howie, R.A. (1973) Lithium- kimberlites from Eastern United States and their aluminium micas from the Meldon aplite, implication for mantle evolution. Contrib. Mineral. Devonshire, England. Mineral. Mag., 39, 289-96. Petrol., 86, 35-44. Cooper, A.F., Paterson, L.A. and Reid, D.L. (1995) Bea, F. (1980) Geochemistry of biotites in an Lithium in carbonatites - consequence of an enriched assimilation process. An approach to recognition of mantle source. Mineral. Mag., 59, 401-8. metamorphic biotites from magmatic occurrence. De Fino, M., La Volpe, L. and Piccarreta, G. (1983) Krystalinikum, 15, 103-24. Marie minerals from Punta delle Pierre Nere Bea, F., Pereira, M.D. and Stroh, A. (1994) Mineral/ subvolcanites (Gargano, Southern Italy): Their leucosome trace-element partitioning in a peralumi- petrological significance. Tscherm. Mineral. nous migmatite (a laser ablation-ICP-MS study). Petrogr. Mitt., 32, 69-78. Chem. Geol., 117, 291-312. De Kimpe, C.R., Miles, N., Kodama, H. and Dejou, J. Bigi, S. and Brigatti, M.F. (1994) Crystal chemistry and (1987) Alteration of phlogopite to corrensite at microstructures of plutonic biotite. Amer. Mineral., Sharbot Lake, Ontario. Clays Clay Mineral., 35, 79, 63-72. 150-58. Boettcher, A.L. and O'Neil, J.R. (1980) Stable isotope, du Bray, E.A. (1994) Compositions of micas in chemical and petrographic studies of high pressure peraluminous granitoids of the eastern Arabian Li-BEARING MICAS 831

Shield. Implications for petrogenesis and tectonic sizing Li, Be, B and F contents. J. Petrol. 31, setting of highly evolved, rare-metal enriched 1025-70. granites. Contrib. Mineral. PetroL, 116, 381-97. Grew, E.S., Hiroi, Y., Motoyoshi, Y., Kondo, Y., Edgar, A.D. (1992) Barium-rich phlogopite and biotite Jayatileke, S.J.M. and Marquez, N. (1995) Iron-rich from some Quaternary alkali marie tavas, West Eifel, komerupine in sheared pegmatite from the Wanni Germany. Eur. J. Mineral., 4, 321-30. Complex, at Homagama, Sri Lanka. Eur. J. Mineral., Edmunds, W.M., Kay, R.L.F. and McCarl~ey, R.A. 7, 623-36. (1985) Origin of saline groundwaters in the Harlow, G.E. (1995) Crystal chemistry of barian CarnmeneUis Granite (Cornwall, England): Natural enrichment in micas from metasomatized inclusions processes and reaction during Hot Dry Rock in serpentinite, Motagua Fault Zone, Guatemala. reservoir circulation. Chem. GeoL, 49, 287-301. Eur. J. Mineral., 7, 775-89. Fiala, J., Vejnar, Z. and Kucerovh, D. (1976): Hecht, L. (1993) Die Glimmer als Indikatoren ftir die Composition of the biotites and the coexisting magmatische und postmagmatisehe Entwicklung der biotite-hornblende pairs in granitic rocks of the Granite des Fichtelgebirges (NE-Bayem). Miinchner Central Bohemian Pluton. Krystalinikum, 12, Geol. Hefte, 10, 1-221. 79-111. Heinrich, E.Wm. (1967) Micas of the Brown Derby Fonteilles, M. (1987) La composition chimique des pegmatites, Gunnison County, Colorado. Amer. micas lithiniFeres (et autres min6raux) des granites Mineral., 52, 1110-21, 1578. d'l~chassi~res comme image de leur 6volution Henderson, C.M.B., Martin, J.S. and Mason, R.A. magmatique. G~ologie de la France, n ~ 2-3, (1989) Compositional relations in Li-mieas from 149-78. S.W.England and France: an ion= and electron- F/Srster, H.-J. and Tisehendorf, G. (1996) Compositional microprobe study. Mineral. Mag., 53, 427-49. heterogeneity of silicie magmatic rocks from the Icenhower, J. and London, D. (1997) Partitioning of German Variscides. Z. geol. Wiss., 24, 467-82. fluorine and chlorine between biotite and granitic Foster, M.D. (1960a) Interpretation of the composition melt: experimental calibration at 200 MPa 0-120). of trioctahedral micas. U.S. Geol. Survey Prof. Contrib. Mineral. Petrol., 127, 17-29. Paper, 354-B, 11-49. Jolliff, B.L., Papike, J.J. and Shearer, C.K. (1987) Foster, M.D. (1960b) Interpretation of the composition Fractionation trends in mica and tourmaline as of lithium micas. U.S. Geol. Survey Prof. Paper, indicator of pegmatite internal evolution: Bob 354-E, 115-47. Ingersoll pegmatite, Black Hills, South Dakota. Ganzeeva L.V. (1973) Taeniolite of alkali metasomatic Geochim. Cosmochim. Acta, 51, 519-34. rocks from Byelorussia (Russ.). Dokl. Ak. Nauk Kramer, W. and Seifert, W. (1994) Mica-lamprophyres Byeloruss. SSR., T XVII, 560-62. and related volcanics of the Erzgebirge and their Goeman, U.E.H. (1972) Untersuchung und Berechnung metallogenic signirieanee. In Metallogeny of einiger Biotite aus Graniten des nordwestlichen Collisional Orogens (R. Seltmann, H. K~mpf and Fichtelgebirges/NE-Bayern. Schweiz. Min. Petr. P. M611er, eds.) Czech Geol. Surv., Prague, 159-65. Mitt., 52, 317-29. Lapides, I.L., Kovalenko, V.I. and Koval', P.V. (1977) Gottesmann, B. and Tischendorf, G. (1978) The Micas of Rare-Metal Granitoids (Russ.). Ed. Klassirikation, Chemismus und Optik trioktae- Nauka, Sibirskoje Otd., Novosibirsk, 103 pp. drischer Glimmer. Z. geol. Wiss., 6, 681-708. Livi, K.J.T. and Veblen, D.R. (1987) 'Eastonite' from Gottesmann, B. and Tischendorf, G. (1980) 0ber Easton, Pennsylvania: A mixture ofphlogopite and a Protolithionit. Z. geol. Wiss., 8, 1365-73. new form of serpentine. Amer. Mineral., 72, Gottesmann, B., Tischendorf, G. and F/~rster H.-J. 113-25. (1994a) Trioctahedral micas as indicators of the Lueeke, W. (1981) Lithium pegmatites in the Leinster compositional evolution of Sn-Li-Rb-Cs-F granites, granite (southeast Ireland). Chem. Geol., 34, an example: The Eibenstock pluton (Western 195-233. Erzgebirge/Germany). 16th General IMA Meeting, Malyshonok, Yu V. (1989) Pecularities of the chemical Pisa, 4.-9. Sept. 1994, Abstracts, p.152. composition of micas from the Murun Massif Gottesmann, B., Tischendorf, G., Wand, U., Bielicki, (Russ.). Miner. Zhurn., 11/6, 38-52. K.-H., Fi~rster, H.-J., Haase, G. and Thomas, R. M&ais, D., Ravier, J. and Phan, K.D. (1962) Nature et (1994b) Die granitoiden Gesteine des Siichsischen composition chimique des micas de deux lampro- Granulitmassivs -Petrographie, Geochemie und phyres. Bull, Soc. franc. Min~r. Crist., 85, 321-28. Altersstellung. Hallesches .lb. Geowiss., 16, 23-55. Mokhtari, A., Wagner, C. and Velde D. (1985) Presence Grew, E.S., Chernosky, J.V., Werding, G., Abraham, K., of late crystallizing ferriannite-rich annite in basic Marquez, N. and Hinthorne, J.R. (1990) Chemistry eruptive rocks from Morocco. N. Jb. Miner., Mh., 11, of kornerupine and associated minerals, a wet 513-20. chemical, ion microprobe, and X-ray study empha- Monier, G. and Robert, J.-L. (1986a) Muscovite solid 832 G. TISCHENDORF ET AL.

solutions in the system K20-MgO-FeO-AI203- 1960-68. SiO2-H20: an experimental study at 2 kbar Pn2o Pechar, F. and Rykl, D. (1992) Die Vergleichung der and comparison with natural Li-free white micas. kristallchemischen Parameter der Fe-Li-Glimmer aus Mineral Mag., 50, 257-66. den Lokalitiiten Cinovee, Vysolc~, K,~len trod des Monier, G. and Robert, J.-L. (1986b) Evolution of the Biotits aus dem Mittelb6hmischen Pluton (Czech., miscibility gap between muscovite and biotite solid Germ Res.). (]steck~ Muzejnl Se~ity, (Jsti, 4, 56-70. solutions with increasing lithium content: an Pomarleanu, V. and Movileanu, A. (1977/78) experimental study in the system K20--Li20-MgO- Contributii la geoehimia biofitelor din Romania. D. FeO-A1203-SiO2-H20-HF at 600~ 2 kbar Primo: S. Inst. GeoL Geofiz., 1. Miner.-PetroL-Geochim., comparison with natural lithium mieas. Mineral Bucuresti, LXV/1, 101-20. Mag., 50, 641-51. Rieder, M. (1970) Chemical composition and physical Monier, G., Charoy, B., Cuney, M., Ohnenstetter, D. and properties of lithium-iron micas from the Kruw Robert, J.-L. (1987) l~volution spatiale et temporelle hory Mts. (Er-zgebirge). Contrib. Mineral Petrol., de la composition des micas du granite albitique ~t 27, 131-58. topaze-16pidolite de Beauvoir. G~ol. France, 2-3, Rieder, M., Povondra, P. and Fr~da, J. (1995) Coexisting 179-88. biotite and muscovite; an example from a Moinian MOiler, G. (1966) Die Beziehungen zwischen der mica schist at Glenfinnan, Scottish Highlands. chemischen Zusammensetzung, Liehtbrechung und Mineral Petrol., 53, 63-74. Dichte einiger koexistierender Biotite, Muskowite Rieder, M., Haapala, I. and Povondra, P. (1996) und Chlorite aus granitischen Tiefengesteinen. Mineralogy of dark mica from the Wiborg rapakivi Contrib. Mineral Petrol., 12, 173-91. batholith, southeastern Finland. Eur. J.. Mineral., 8, Nash, W.P. (1993) Fluorine iron biotite from the 593 -605. Honeycomb Hill rhyolite, Utah: The halogen record Rub M.G., Rub, A.K. and Loseva, T.I. (1971) Micas as of decompression in a silicic magma. Amer. guides for ore presence detection in granitoids Mineral., 78, 1031-40. (Russ.)./zv. Ak. Nauk SSSR, Ser. geol., 10, 73-85. Neiva, A.M.R. (1976) The geochemistry ofbiotites from Rub, M.G., Pavlov, V.A., Rub, A.K., Stemprok, M., granites of northern Portugal with special reference Drabek, M. and Drabkova, E. (1983) Vertical to their tin content. Mineral. Mag., 40, 453-66. zonality of elements in Li-F granites of the Neiva, A.M.R. (1980) Chlorite and biotite from contact Cinovec massif (CSSR) (Russ.) In Korrelyaziya metamorphism of phyllite and metagraywacke by magmaticheskikh porod Chekhoslovakii i nekotorikh granite, aplite-pegmatite and quartz veins. Chem. rayonov SSSR (O.A. Bogatikov and A.M. Borsuk, GeoL, 19, 49-71. eds.) Ed. Nauka Moskva, 108-37. Neiva, A.M.R. (1981a) Geochemistry of hybrid Rub M.G., Rub, A.K. and Akimov, V.M. (1986) Rare- granitoid rocks and of their biotites from central metal bearing granites of Central Sikhote-Alin' northern Portugal and their petrogenesis. Lithos, 14, (Russ.). Izv. Ak. Nauk SSSR, Ser. geol., 7, 33-46. 149-63. Schmidt, W. and Pietzsch, C. (1990) Iron distribution Neiva, A.M.R. (1981b) Geochemistry of chlorite and and geochemistry of pegmatitic dioctahedral 2M1 biotite from contact metamorphism of phyllite by micas. Chem. Erde, 50, 27-38. granites. Mem. Noticias, PubL Lab. MineraL, Geol., Silva, M.M.V.G. and Neiva, A.M.R. (1990) Univ. Coimbra, No. 91/92, 113-34. Geochemistry of the granites and their minerals Neiva, A.M.R. (1992) Geochemistry and evolution of from Paredes da Beira-Penedono, northern Portugal. Jales granitic system, Northern Portugal. Chem. Chem. Geol., 85, 147-70. Erde, 52, 225-41. Skosyreva M.V. and Vlasova E.V. (1983) First Neiva, A.M.R. and Gomes, M.E.P. (1991) Geochemistry occurrence of polylithionite from rare-metal granite of the granitoid rocks and their minerals from Lixa pegmatites (Russ.). DokL Ak. Nauk SSSR, 272/3, do Alvao-Alfarela de Jales-Toureneinho (Vila Pouca 694-97. de Aguiar, northern Portugal). Chem. Geol., 89, Stone, M., Exley, C.S. and George, M.C. (1988) 305-27. Compositions of trioctahedral micas in the N~mec, D. (1969) Glimmer der regionalmetamorphen Comubian batholith. Mineral Mag., 52, 175-92. Skarne Westmiihrens. Tscherm. Miner. Petrogr. Sun Shihua (1984): The subdivision of lithium micas Mitt., 13, 55-89. and their significance in the study of granitoids. In N~mec, D. (1983) Zinnwaldit in moldanubischen Geology of Granites and their Metallogenetic Lithium-Pegmatiten. Chem. Erde, 42, 197-204. Relations (Xu Keqin, Tu Guangchi, eds.), Proc. of Ottolini, L., Bottazzi, P. and Vannucci, R. (1993) the Int. Symp., Nanjing, China, Oct. 26-30, 1982, Quanitification of Lithium, Beryllium, and Boron 379-93. in silicates by secondary ion mass spectrometry Tindle, A.G. and Webb, P.C. (1990) Estimation of using conventional energy filtering. Anal Chem., 65, lithium contents in trioctahedral micas using microp- Li-BEAR1NG MICAS 833

robe data: application to micas from granitic rocks. Granitoid- und Greisenglimmer aus Zinnlagerstiitten Eur. J. Mineral., 2, 595-610. des Siichsischen Erzgebirges und der Mongolei. Tischendorf, G., Friese, G. and Schindler, R. (1969) Die Thesis, Bergakademie Freiberg, 129 p. Dunkelglimmer der westerzgebirgiseh-vogtl[in- Volkov, V.N. and Gorbacheva, S.A. (1980) Variations dischen Granite und ihre Bedeutung als petrogen- of crystallization environment of granites in a etische und metallogenetisehe Indikatoren. Geologie, vertical section of an intrusive body according to 18, 384-99, 1024-44. data on the composition of rock-forming biotite Tischendorf, G., P~ilchen, W., R611ig, G. and Lange, H. (Russ.). Geokhimiya, 147-53. (1987) Formationelle Gliederung, petrographisch- Wagner, C., Velde, D. and Mokhtari, A. (1987) Sector- geochemische Charakteristik und Genese der zoned phlogopites in igneous rocks. Contrib. Granitoide der Deutschen Demokratischen Mineral. Petrol., 96, 186-91. Republik. Chem. Erde, 46, 7-23. Winchell A.N. (1942) Further studies of the lepidolite Tr6ger, W.E. (1962) Uber Protholithionit und system. Amer. Mineral., 27, 114-30. Zinnwaldit - Ein Beitrag zur Kenntnis von Chemismus und Optik der Lithiumglimmer. Beitr. Mineral. Petrol., 8, 418-31. [Manuscript received 8 January 1997." Uhlig J. (1992) Zur Mineralogie und Geochemie der revised 2 April 1997] 834 G. TISCHENDORF ET AL.

Appendix Sources of data on which the figures and regression equations are based

Author Fig. 3-4 Fig. 5a Fig. 5b Fig. 6a, b Eq. tri 1-3 Eq. tri 4 Eq. tri 5 Eq. di 1-2

Foster, 1960b x Miiller, 1966 x x x x Heinrich, 1967 x N~mec, 1969 x x CereS, et al., 1970 x x x Rieder, 1970 x x Rub et al., 1971 x x Goeman, 1972 x Bargar et al., 1973 x x Chaudhry and Howie, 1973 x x x Sun Shihua, 1974 x Bokonbaev, 1976 x x Fiala et al., 1976 x Neiva, 1976, 1980, 1981b x x Al-Saleh et al(, 1977 x x Lapides et a/.,\1977 x Pomarleanu and Movileanu, 1977/78 x Barri6re and Cotton, 1979 x x x Bea, 1980 x x x Volkov and Gorbacheva, 1980 x x Luecke, 1981 x x x Borodanov, 1983 x x N~mec, 1983 x Rub et al., 1983 x x x Skosyreva and Vlasova, 1983 x x CereS, and Trueman, 1985 x Edmunds et al., 1985 x x Rub et al., 1986 x Fonteilles, 1987 x x x Jolliff et al., 1987 x x Monier et al., 1987 x x x Stone et al., 1988 x x x Henderson et al., 1989 x x Malyshonok, 1989 x x Brigatti and Davoli, 1990 x Grew et al., 1990 x Schmidt and Pietzsch, 1990 x Silva and Neiva, 1990 x x x x Neiva and Gomes, 1991 x x x Neiva, 1992 x Pechar and Rykl, 1992 x x x Uhlig, 1992 x x x x Hecht, 1993 x x x Bigi and Brigatti, 1994 x du Bray, 1994 x x x CereS, et aL, 1995 x x x Charoy et al., 1995 x x x Grew et aL, 1995 x Rieder et al., 1995 x x x Charoy and Noronha, 1996 x x x Rieder et al., 1996 x x Authors' and Naumann's data base x x x x