American Mineralogist, Volume 78, pages 1056-1065, 1993

Barian titanian phlogopite from potassic lavas in northeast China:Chemistry, substitutions, and paragenesis

MrNc ZnaNc*Department of Geology, Imperial College, University of London, London SW7 2BP, U.K.

and Department of Geological Sciences, Southern Methodist University, Dallas, Texas 75275, U.S.A.

P,rur Suoo.LnvDepartment of Geology, Imperial College, University of London, London SW7 2BP, U.K.

RonBnr N. TrrorrpsoNDepartment of Geological Sciences, University of Durham, Durham DHI 3LE, U.K.

MrcH.q.nL A. DuNc,lNDepartment of Geological Sciences, Southern Methodist University, Dallas, Texas 75275, U.S.A.

Ansrnl,cr

Magmatic barian titanian phlogopite (BaO : 4.2-11.2 wto/o and TiO, : l0.l-13.1 wt0/0)occurs in the groundmass and in a magmatic inclusion of olivine leucitites from northeastChina. In addition, titanian phlogopite (BaO < 0.3 wto/o and TiO, < 6.4 wto/o) formed byreaction between olivine phenocrysts and late fluids also occurs as coronas rimming olivinephenocrysts in associated leucite basanites. Compositions of the barian titanian phlogopitevary systematically; AlrO. and TiO, increase and SiOr, MgO, FeO, and KrO decrease withincreasing BaO. Substitutions deduced from such variations include Ba * (Al,Fe3* ) : K+ Si and 2(Mg,Fe'?* ) : Ti + n. Ti-O and Ti-Tschermak's substitutions may also bepresent in subordinate amounts. The Ba + Al : K + Si substitution is common to mostmagmatic barian micas, except for some lamproitic micas. In contrast, no general Tisubstitution scheme is applicable, as Ti substitutions vary with melt compositions andintensive variables.

These phlogopite samples crystallized as a late phase at low pressures and temperatures.Differences in mineral assemblages between olivine leucitites and leucite basanites contrib-ute significantly to the observed disparities in chemical composition between the bariantitanian phlogopite and the corona phlogopite and, in general, among phlogopite fromother alkalic magmatic rocks.

INrnoouc:ttoN

Titanian phlogopite occurs in lamproites and other po-tassic rocks (e.g., Mitchell, 1985; Wagner and Velde, 1986;Mitchell and Bergman, l99l), as well as in some alkalicbasaltic rocks (e.g., Ryabchikov et al., 198 I ; Qi and Xiao,1985), and varies widely in composition. Magmaticphlogopite enriched in both Ba and Ti is rare. The oc-currences ofbarian titanian phlogopite have been report-ed in some potassic rocks (e.g., Thompson, 1977; Wend-landt,1977;- Birch, 1978, 1980; Holm, 1982; O'Brien etal., 1988) and in a few alkalic rocks such as nephelinites,olivine melilitites, and aillikites (e.g., Mansker et al., 1979;Velde, 1979; Mitchell and Platt, 1984; Barnett et al., 1984;Boctor and Yoder, I 986; Cao and Zhu, 1987 ; Rock, I 99 l).There is no general agreement regarding the substitutionschemes in barian titanian phlogopite (e.g., Mansker et

t Present address: School of Earth Sciences, Macquarie Uni-versity, Sydney, NSW 2109, Australia.

0003-004x/93109 l 0- l 056$02.00

al.,19791, Bol et al., 1989; Guo and Green, 1990). Litt leattention has been paid to their chemical variations andparagenetic implications. In this paper, we report chem-ical data for barian titanian phlogopite samples in potas-sic lavas and a magmatic inclusion from northeast China,compare them with other magmatic phlogopite samples,and discuss their substitution schemes and parageneses.

Gnolocrc.rl sETTTNG AND pETRoLocy

Three adjacent volcanic fields ofpotassic volcanic rocks,Wudalianchi, Erkeshan, and Keluo (WEK), in northeastChina (125"30' -126'45'E, 48'00'-49"30'N), are locatedbetween the Songliao Basin (a late Mesozoic basin) andthe Xing'an Mountains (a Paleozoic fold belt) (Hsii, 1989).These lavas, which cover an area of > 1400 km2, wereerupted during three main episodes: late Miocene (9.6-7.0 Ma), late Pleistocene (0.56-0.13 Ma), and Recent (AD1719-1721) (Zhans et al., l99l).

The WEK potassic rocks are porphyritic, with ground-mass textures varying from vitric to holocrystalline. They

1056

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE lo57

Fig. 1. Backscattered electron images of the WEK barian titanian phlogopite (a) in an olivine leucitite sample and (b) in DZ2n.Abbreviations: phl : phlogopite; o1 : olivine; cpx : clinopyroxene; af: hyalophane; ne : nepheline; lc : leucite; sp : Fe-Tioxides.

are divided into three types; olivine leucitite, leucite bas-anite, and trachybasalt. Both olivine leucitite and leucitebasanite contain phlogopite. Olivine (8-130/o), clinopy-roxene (2-4o/o), and leucite (14-20o/o) phenocrysts in ol-ivine leucitites are dispersed in a matrix of olivine, cli-nopyroxene, leucite, Fe-Ti oxides, nepheline, phlogopite,and rare sodalite. Leucite basanites have a phenocrystassemblage of olivine (4-9o/o),leucite (l-7o/o), and clino-pyroxene (0-l lol0). Matrix phases are potassium feldspar(up to 500/o) and clinopyroxene (up to 350/o), plus minorleucite, olivine, and Fe-Ti oxides.

Olivine leucitites have lower SiO, but higher MgO andMg/(Mg + Fe2*) ratios (43-45 wt0/0, 14-10 wto/0, and0.78-0.71) than leucite basanites (47-55 wto/0, 3.8-8.0wt0/0. and 0.68-0.52). Most olivine leucitites and someleucite basanites are primary or nearly primary, formedby partial melting of the mantle sources at different pres-sures (Zhang et al., 1991). K.O content is high and vari-able (3.6-7.l wto/o), and most samples have KrO/NarO> l. TiO, and Ba concentrations are 2.0-3.1 wt0/o and1300-2200 ppm, respectively. Olivine leucitites and leu-cite basanites are not systematically different in Ba,whereas the former generally contain lower TiO, and KrOthan the latter. For additional details, see Zhang et al.( 1 9 9 1 , 1 9 9 3 ) .

OccunnsNcEs oF PHLocoPrrE

Phlogopite occurs in the WEK potassic lavas as (l) amatrix phase in olivine leucitites, (2) a constituent of amagmatic inclusion (DZ2n), and (3) coronas rimming ol-ivine phenocrysts in a leucite basanite (WZKI6). Phlog-opite samples found in the first two occurrences are Ba-rich (BaO > 4.2 wo/o), whereas the phlogopite coronascontain much lower BaO (<0.3 wto/o).

Interstitial phlogopite platelets (0.01-0.1 mm, Fig. la)constitute up to 3 modalo/o of olivine leucitites. A fewlarger anhedral grains, up to 0.2 mm, enclose earlygroundmass phases such as clinopyroxene and Fe-Ti ox-ides. They show very strong, distinctive red-brown topinkish yellow pleochroism, which may be attributed tohigh Tio, contents (Mitchell, 1985).

DZ2n \s a small inclusion (l cm) composed of bariantitanian phlogopite, hyalophane (9-20 molo/o celsian),leucite, and sodalite. Subhedral phlogopite grains (0.2-0.3 mm, Fig. lb) exhibit the same distinctive pleochro-ism as those from olivine leucitites.

Phlogopite coronas rimming olivine phenocrysts arefound in a holocrystalline leucite basanite (WZKI6) sam-pled from the central portion of a thick lava flow (>20m thick). These mica flakes are <0.05 mm in the direc-

1058 ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE

4 00 0 2 0 4 _ 0 6 0 8 0 r o o r z o 0 0 2 0 4 0 6 0 8 0 1 0 0

wt% BaOwt',6 BaO

Fig.2. (a-f) BaO vs. orher elements for the WEK phlogopite.Solid triangles : olivine leucitite; solid circles : DZ2n; opentriangles : corona.

Traue 1, Representative phlogopite analyses

tion perpendicular to (001) and show pale red to lightyellow pleochroism.

AN,q.Lyrrc,{L TEcHNIeUE

Analyses of mica chemistry were mainly performed atImperial College (IC), University of London, using aCambridge Instruments Microscan 5 electron microscopefitted with a Link energy-dispersive detector. Analyticalconditions were accelerating voltage : 15 kV, fixed spec-imen current on a Co standard : 4.0 nA, and countingtime : 100 s. ZAF corrections were used, and syntheticBaTiO, was analyzed to monitor interference betweenBaZa, and TiKa, lines, which proved to be negligible.

A subset of barian titanian phlogopite samples has alsobeen analyzed using a Jeol 733 Superprobe equipped withwavelength-dispersive detectors at Southern MethodistUniversity (SMU), in order to analyze F and Cl contentsand to crosscheck the Ba-Ti interference. Analytical con-ditions were accelerating voltage : 15 kV, current : 20nA, and beam spot diameter : l0 pm. The X-ray count-ing rates were corrected using the a matrix correction(Bence and Albee, 1968). Synthetic fluor-phlogopite andnatural scapolite were used as standards for F and Cl.Benitoite (BaTiSi.On) analyses gave BaO 37.15-38.38wto/0, TiO, 18.35-18.86 wt0/0, and SiOr 43.4l-43.51 wt0/0,in good agreement with the theoretical values of 37.08,19.32, and 43.58, respectively. The two sets of analysesare generally comparable.

wt SiOz

^l l^

^ l . ^^^

|

(a) l'4;" (d) o.^^i*&.,

^l i^t^ ;f.ir. { f r .

(0

r t 0

r 6 0

320

3 0 01 5 0

l 4 0

l 4 0

t20

l 0 01 6 0

1 4 0

t 2 0

1 0 0

8 0

6 09 0

8 0

1 0

6 0

5 0

13.0

1 2 0

wt% N,OJ ^ A^^ .t

^ "iffi"r ^ r r ll ^

(b)

^l l 0

l 0 01 4 0

t20

r 0 0

8 0

6.0

''*''o' ,r^i*fto

i .60

i{^Tfi,.

2 5 2OZ2n WZK16

2dDZ2n

94DZ6

82DZ6

2aDZ4

62DZ4

otoz4

No.Host

sio,Alro3Tio,FeOMnOMgoCaOBaONaroKrOf

Total-Mg'

J I

AITiFE

MnMg

TotalCaBaNaK

TotalF

Charge'.

34.9412.7310.4410.64n.a.

13 .760.55o . b /

0.936.89

97.540.698

5.2592.2581.1821.340n.a.3.088

12.9820.0890.3930.2721.3222.O77n.a.

46.871

32.8713.431 1 . 0 910.54n.a.

12.87n 1 R

9.36U . O J

6 0 8n.a.

97 030.685

5.0752.4441.2881.361n.a.2.962

13.1 300.0250 5660 1901 . 1981.979n.a.

46.916

34.1612.5610.5410.570 1 2

14 570.237.700.83o.5c1.34

98.610.711

5 1 8 12.24s1.2031.3410.0153.294

13.2790.0370.4s80.2441.2672.0060.643

46.411

36.7512.3510.281 1 . 6 4n.a.

t J . c o

0.354.480.907.70

98.01u . o / c

o : 2 25.4392.1541.1441.441n.a.2.992

13.1690.05s0.2600.2581.4542.027n.a.

46.777

33.2413.5711.471 1 . 3 60.01

12.460.349 6 90.486 . 1 6n.a

98 780.662

5.0662.4371 . 3 1 41.4480.0012.831

13.0970.056u . 5 / v0.1421 . 1 9 81.974n.a.

47.033

31.5213.8612.4810 090 . 1 1

12.470 . 1 29.490.875.200.93

96.360.688

4.9052.5421.4611 .3130.0142.893

13.1280.0200.5790.2621.0321.8930.458

46.934

31.6014.4713.0610.240.09

11.820.13

10.911 . 0 15 . 1 8n.a.

98.510.673

4.8492.6171.5071 . 3 1 40.0122.704

13.0020.0210.6560.3001.0141.991n.a.

47.377

39.5910.316.259.030.09

17.990.040.240.908.55n.a.

92.990.78

5.8961 .8100.7001.1250.0113.994

13.5300.0060.0140.2601.6241.905n.a.

45.507

A/ote: numbers 2aand2d analyzed at SMU; others at lC. Febt as FeO; n.a. : not analyzed.- L e s s O = F .

" Calculated with t€trahedral + octahedral: 14.

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE 1059

Mrcl crrnrvnsrnv

This study is based on about 80 phlogopite analyses.Representative results are given in Table l. Figure 2 il-lustrates chemical variations of the phlogopite samples ina series of variation diagrams that show BaO vs. othercompositional variables.

Barian titanian phlogopite

Barian titanian mica samples found in WEK olivineleucitites and in DZ2n conlain 4.2-11.2 wto/o BaO andl0.l-13.1 wto/o TiOr. Mg' [: atomic ratio Mg/(Mg +Fe,",)l values are 0.627-0.71l. Thus, most of them arephlogopite (Deer et al., 1962). These phlogopite sampleshave moderate but variable SiO, (31.7-36.8 wto/o) andAlrO3 (l1.5-14.6 wto/o), whereas MgO and FeO (Fe,.,) arell.3-14.2 and,9.2-14.2 wt0/0, respectively. KrO is lowand variable (5.2-7 .7 wto/o). NarO and CaO are below l.land 0.4 wt0/0, respectively, and do not correlate with BaO.They contain <0.06 wto/o Cl, but moderate amounts of F(0.50-1.39, av. 0.90 wtolo). Although many of the ana-lyzed phlogopite grains are heterogeneous, no systematiczoning patterns have been detected.

SiOr, MgO, FeO, and KrO decrease whereas AlrO, andTiO, increase with increasing BaO in the barian titanianphlogopite (Fig. 2). Phlogopite samples inDZ2n are gen-erally more enriched in BaO than those in olivine leuci-tites; thus they have higher AlrO. and TiO, but lowerSiOr, MgO, FeO, and KrO. However, the coherent ele-ment trends and overlapping Mg' values demonstratesimilar substitution schemes and suggest a genetic affinitybetween the two groups of phlogopite, despite their dif-ferent occurrences.

Si and Al cations (O : 22) are <8.00 (7.38-7.73; av.7.5 + 0.06) pfu, implying that Al is confined exclusivelyto the tetrahedral occupancy. The sums ofoctahedral cat-ions are <6.0 (5.29-5.93, av. 5.6 + 0.09), lower thanthose for an ideal trioctahedral mica. In contrast, the oc-cupancy in the l2-fold-coordinated interlayer site is al-most stoichiometric (av. 1.99 t 0.006).

Phlogopite coronas

Phlogopite coronas in WEKI6 are chemically distinctfrom the barian titanian phlogopite. They have very lowBaO (<0.24 wto/o) and moderate TiO, (6.38 wto/o), buthigh Mg' value (0.78), similar to the olivine phenocrysts(For,-,n, Zhang, unpublished data). The phlogopite coro-nas contain higher SiO, (39.5 wto/o) and K,O (8.6 wto/o)but lower AlrO3 (10.4 wto/o) than the barian titanianphlogopite. They have fewer vacancies in the tetrahedraland octahedral sites (t4rR :7.72, I6rR : 5.79) but do nothave sufficient interlayer ions (A site : 1.90).

Comparison with other barian phlogopite

The compositions of barian phlogopite (arbitrarily setas BaO > 2 wto/o) differ significantly among three cate-gories ofhost alkalic rocks: (1) nephelinites, (2) aillikites,melilitites, and carbonatites, and (3) maflc potassic rocks.Figure 3 shows variations of BaO, AlrOr, and Mg' vs.

TiO, for the WEK phlogopite samples and the other sam-ples from the literature.

Mica samples in Hawaiian nephelinites are the mostenriched in BaO (14.3-20.4 wto/o), TiO' (13.0-14.5 wto/o),and Al,O. (15.8-17.6 wto/o) but have the lowest Mg' (aslow as 0.40) among all the samples concerned. This is inagreement with their reported occurrence as a Iategroundmass phase. On the other hand, those in aillikitesand melilitites show wide variation in BaO (3-15 wto/o)but relatively low TiO, (<4.0 wto/o). Their Al'O, contentsare variable, but generally high (up to 19.3 wto/o). HighMg' values (>0.84) reflect the ultramafic nature of thehosts (Rock, l99l). The composition of barian phlogo-pite in the silicate bands ofJacupiranga carbonatites (Bra-zil) is characterized by extremely high Mg' (0.94-0.96)and low TiO, (<0.2 wto/o). The high Mg' values are con-sistent with the dolomite-dominant host carbonatites(Gaspar and Wyllie, 1987). Al,O3 is high (16-20 wto/o),but BaO is variable (<1.0-10.2 wto/o). Chemical charac-teristics of barian titanian phlogopite in potassic rocksare apparently related to tectonic settings (e.g., Barton,1979). The phlogopite from potassic rocks associated withsubduction (e.g., Roman Province, Italy, and Montana)shows a wide range in BaO contents (2-9 wto/o) and Mg'(0.61-0.89) but, as expected from the host magmas, con-tains lower TiO, (<6.0 wto/o) than the phlogopite frompotassic rocks of intraplate settings (e.g., WEK and EastAustralian leucitites, >8.0 wto/o). The Australian phlog-opite samples are similar to the WEK ones in terms ofTiOr, BaO, and AlrO. contents but are lower in Mg' (0.56-0.61) than the latter (Fig. 3). Mica samples in lamproitesrarely have BaO > 3.0 wto/0, despite strong enrichment ofBa in the hosts.

Comparison with phlogopite from other potassic rocks

TiO2, Al2O3, and FeO contents of the WEK phlogopitesamples are plotted in Figure 4 for comparison with thosefrom other potassic rocks, including lamproites, minettes,Italian and Ugandan potassic rocks, and Mongolian leu-cite basanites. The WEK phlogopite samples can be readilydistinguished by their generally higher Al'O. and TiO,contents from those hosted by lamproites and the Ugan-dan kamafugites. The latter generally evolve toward verylow AlrO, at various TiO, and FeO contents, forming alate-stage tetraferri phlogopite. This chemical signature isapparently imparted by the peralkaline nature of hostmagmas. Although most of the WEK samples plot in thefield distinguished by micas from minettes and the Italianpotassic rocks in the AlrO. vs. FeO diagram (Fig. 4a),they have much higher TiO, contents (Fig. 4b). The com-bination of a negative correlation between AlrO, and FeOand a positive correlation between AlrO, and TiO, forthe WEK samples (Fig. a) is distinctive and has not beenobserved elsewhere. The WEK phlogopite samples aresimilar to the Mongolian mica megacrysts in terms ofTiO, content (Fig. ab) and Mg' (0.63-0.71 and 0.59-0.67,respectivel/, but WEK phlogopite samples lower in AlrO,and FeO than the mica megacrysts (Fig. 4a).

;t* o 'sro

U (a)V

1060

1 6

t 2

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE

(\l

oirgiRl

4

r6

l 2

(\l

ol'. 8!R=

4

l 6

t 2

(\l

oF - 8

si

4

0.60 0.70 0.80 0.90 1.00Ms/(Mstft)

SussrrrurroN scHEMES

Assigning a unique set of substitutions to the WEKbarian titanian phlogopite is hindered by (l) the com-plexity of substitution schemes in micas, which can berepresented in terms of various linear combinations ofend-members (Hewitt and Abrecht, 1986), (2) the uncon-strained Fe and Ti valencies and HrO contents, and (3)

Fig. 3. TiO, vs. (a) BaO, (b) Al,O3, and (c) Mg' for magmaticbarian micas. For the sake of clarity, one-half to two-thirds ofthe WEK data points have been omitted from Fig. 3, 4,6, and7. Dala sources: nephelinite-Hawaii, Mansker et al. (1979);southeast China, Cao and Zhu (1987); aillikite and melilitite-South Africa, Velde (1979), Boctor and Yoder (1986); Green-land, Rock ( I 99 I ); carbonatite - Brazil, Gaspar and Wyltie ( I 9 8 7);potassic rocks-East Australia, Birch (1978; 1980); RomanProvince, Thompson (1977), Holm (1982); Montana, Wend-landt (1977), O'Brien et al. (1988); lamproite-Mitchell andBergman (1991).

ts

the possibility that Ti, Fe, and even Mg (Foley, 1990)may occur in tetrahedral coordination. However, as theWEK phlogopite samples have genetic affinities, the sub-stitution schemes can be evaluated by their coherentchemical departures from end-member phlogopite (Hew-itt and Abrecht, 1986) in light of relevant experimentaldata.

Ba substitutions

There are two alternative substitutions involving Baand the interlayer sites. One, suggested by Shmakin (1984)and Wagner and Velde (1986) and implied from coupledsubstitutions proposed by Mitchell (1981) and Guo andGreen (1990), is

r r2tBa + r r2tE: 2rr2tK. ( l )

Another, proposed by Wendlandt (1977), Mansker et al.(1979), and Bol et al. (1989), is

rr2tBa + r4iAl: rt2tK + t4rsi. (2)

The WEK phlogopite samples have almost stoichio-metric interlayer occupancy and plot along the line of l: Isubstitution of Ba for K + Na + Ca (Fig. 5a), whichfavors Substitution 2 over l. As Ba cations per formulaunit (0.25-0.70) are >r4rAl - 2.0 (0.15-0.62), Al is in-sufficient to compensate for the charge imbalance. Thus,even ifSubstitution 2 describes the general form ofcou-pled Ba substitution, another cation, probably Fe3+, mustalso be present in tetrahedral coordination to compensatefor both the tetrahedral site deficiency and the chargeimbalance.

Chemical data for the other barian trioctahedral micasfrom the literature also define a similar negative corre-lation between Ba and K + Na + Ca, as shown by theWEK samples. Tho sums of the interlayer cations arenearly 2.0 pfu for all but a few samples from the LeuciteHills and Smoky Butte lamproites, whose interlayer cat-ions are as low as 1.8 pfu and whose BaO contents are<3 wto/o (Ba < 0.2 pfu) (Mitchell and Bergman, l99l).Therefore, Substitution 2 may be applicable to the ma-jority of magmatic barian micas, except for those fromlamproites.

wt% BaO

o

(b) "V0 5 1 0 1 5 2 0

wt% NOs

0.398n " " n

x

o 9oo

oo(c) o

tr

, VT

o""* 'V

0 r _0.50

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE

Z O 0 5 l 0 1 5

wt%TiOz

t . 2

1 . 01 3 . 8

13.6

13.4

13.2

1 3 . 0

12.8

1 0 6 1

oAA

inclusion DZ2nol-lcucititcolivine corona

TBENDS

+ kamafugitee mincttc &

Italian high-K.

KEYSMO Mongolian bas.U Ugandan high-K.M minettcI Italian high-K.LC core, lamproite

0.7

I

RRto;et

(a)

0 5 l 0 1 5wt% tuOt

Ti substitutions

The role of Ti in barian titanian phlogopite dependsupon Ti valence and occupancy. The relationship be-tween Ti cations and excess positive charge (Fig. 6a) onthe basis of 14 tetrahedral * octahedral cations (e.g., Bolet al., 1989) lends strong support for Tia+, as most of theWEK samples plot close to or above the line defined by2 ionic charge - 44 :2Ti. Values of /o, calculated fromFe-Ti oxides in the WEK olivine leucitites are 0.5 logunits above the FMQ buffer (Zhang, unpublished data),further suggesting that all Ti in the potassic magma couldbe quadravalent.

Another controversy concerns Ti occupancy in micas.In light of the tetrahedral deficiencies in all WEK sam-ples, Substitutions

ra lS i : ra tT i

and

rrMg + 2t41{l: t6rE + 2t4rTi (4)

are theoretically possible.Optical absorption and Miissbauer spectral results fail

to provide conclusive evidence in favor of t4rTi in micas(see summaries of Edgar and Arima, 1983; Dymek, 1983;Abrecht and Hewitt, 1988; Brigatti et al., l99l). In par-ticular, the presence of torTi in barian titanian phlogopitewould increase lattice energy by electrostatic repulsionbetween t4lTi and t"lBa, and hence decrease phlogopitestability (Bol et al., 1989). With regard to the WEK mi-cas, Substitution 3 cannot explain observed variations inthe sum of tetrahedral * octahedral cations (about 0.5pfu), nor does the positive correlation between Al and Ti

Gig. ab) support Substitution 4. For these reasons, oc-

(a)

1",t"s ? a

"N-:-^+or^'o;

a\a

Fig. 4. AlrO. vs. (a) FeO and (b) TiO, for phlogopite from potassic rocks. Data sources: lamproite, Mitchell (1981), Mitchell

and Bergman (1991); kamafugites, Edgar (1979); Italian potassic rocks, Thompsot(1977), Holm (1982); minettes, Bachinski and

Simpson (1984); Mongolian leucite basanites, Ryabchikov et al. (1981).

2.4

* 2 .0$(!

; 1 . 8

{ 1 . 6+

Q r .4

(3)14

o

Iq).=U)

+b

0.0

Fig. 5. Ba vs. (a) K + Na + Ca and (b) tetrahedral + octa-hedral site cations for the WEK phlogopite. Calculated with O :

22. Symbols as in Fig. 2.

0.1 0.3 0.4 0.6Ba2*

(b)

M O

)"^s^ o

(b) (T+o)-site = 13.G

\062

4.0

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE

\tT 3.0og$

6 z.oo

B 1.0a.

rA^ )z'(a)

(f'6'{q'

0.03.5

3 .0

2.O0.0 0.5 1.0 l .5 2.0

Ti4+Fig. 6. Ti vs. (a) excess positive charge, (b) tar(Al * Fe3+),

and (c) t6t(Ivtg + Fer*) for the WEK phlogopite; a, calculatedwith tetrahedral + octahedral : l4:b and c, with O : 22. Suf-ficient Fe3+ is calculated from Fe,", and added to Si + Al to fillthe T sites and is subtracted from Fe,". on the O sites. Symbolsas in Fig. 2.

tahedral occupancy ofTi, and thus tetrahedral occupancyofFe3*, is strongly preferred.

Assuming that Ti occurs as t6lTi4+, three basic substi-tutions involving Ti are frequently proposed (e.g., Dy-mek, 1983). They are

(c)

Ti-Tschermak's:

t0l11ylg,Fe2*) + 2t4rsi: I6rTi + 2t4r(Al,Fe3+) (5)

Ti vacancy:

2tol(Mg,Fe2+): I6lTi + I6rE (6)

Ti-o:t6t(Mg,Fe2+) + 2(OH) : t61Ti + 2O2- + H2. Q)

In addition, Foley (1990) suggests another substitutionfor lamproitic micas formed under reducing conditions,

r6tTi + r61Fe2+ + r6tE + r41si + r4rE

:3r61Mg + 2totMg. (g)

This substitution is, however, inapplicable to the WEKmicas because of the antipathetic correlation between Siand Ti (ref. Fig. 2a,2c).

Although the data for the WEK barian titanian phlog-opite define a slope broadly parallel to the line repre-senting Ti-Tschermak's substitution [Ti : 2(Al + Fe3+)- 21, the amount of Ti is much too high for this to bethe only or even dominant substitution (Fig. 6b). Dis-tinction between Ti vacancy (Tronnes et al., 1985) andTi-O (Bol et al., 1989) is equivocal because the correla-tion shown in Figure 6a is consistent with both interpre-tations, and because the importance of an oxy substitu-tion cannot be evaluated without independent Fe3+ andHrO determinations (Dymek, 1983; Abrecht and Hewitt,1988). A well-defined negative correlation between Ti andMg * Fez+ for the WEK samples (Fig. 6c), which plotparallel and close to the line at zTi + (Mg + Fe'z+; : 6,is strong evidence for the Ti-vacancy to be a dominantsubstitution. The tetrahedral * octahedral site deficiency(Fig. 5b) also points to the same conclusion. An alterna-tive explanation is that the deficiency is an artifact ofnormalizing to O : 22.The Ti-O substitution could pro-duce a positive correlation between O and Ti and, as O> 22,higher calculated tetrahedral * octahedral cations.If this were the case, and a fixed-cation normalizationtetrahedral * octahedral : 14 were used instead, the cal-culated interlayer cations per formula unit would be onaverage >2. 13 (up to 2.35), rather than being stoichio-metric. The extra interlayer cations, however, must beused with great caution, as evidence against Ti-O substi-tution because it may result from an effect of substitutionof F for OH upon the normalization procedure. This sub-stitution results in O < 22 and thus increases the sum ofthe interlayer occupancy calculated from fixed cations.Since the maximum F content in the WEK phlogopite is1.4 wto/o, corresponding to [(OH)3rrFor,]ooo, the calculat-ed interlayer cations can only be increased by 0.04 pfurelative to an F-free stoichiometric phlogopite, far lessthan necessary to compensate for the apparent excess ininterlayer cations. In other words, ifthe calculated excessof the A-site occupancy were to solely reflect the presenceof F, the WEK barian titanian phlogopite should contain

F

$ 2's

ss 2.o

1 . 56.0

5 .0

eg40={9.

3.0

(b)

. r lA\

IA

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE 1063

at least 4 w{/oF. Therefore, although no conclusive choicefrom the alternatives can be made, the data favor a Tivacancy substitution.

The above arguments are based on coherent chemicalvariations of the WEK phlogopite using simple parame-ters such as the amounts of cation(s) or site occupancy.Some plots with complex parameters representing linearcombinations of various substitutions have been em-ployed (e.g., Mansker et al., 1979; Wendlandt, 1977; Guoand Green, 1990) in an attempt to identify specific com-binations of the basic substitutions for barian titanianphlogopite. Nevertheless, the results fail to provide betterdiscrimination and sometimes could be misleading be-cause the apparently linear relationships might be an ar-tifact derived from correlations among simple parameters(Foley, 1990). Three cases will be considered below toelucidate the problems associated with this approach.

1. In Figure 7 2(Al + Fe3+) * Ti is plotted vs. 2Si +(Mg + Fe'?+ ) for the WEK phlogopite. Wendlandt (1977)interpreted a relationship similar to that shown in Figure7 as evidence for the Ti-Tschermak's substitution in bar-ian phlogopite from the Montana potassic rocks. In fact,compositions that are parallel to the join defined byphlogopite and Ti-Tschermak's end-member can be de-rived from a linear combination of Substitutions 2 and7. Moreover, the offset of data points below the join re-quires some type of vacancy substitution. Even Substi-tution 6 cannot significantly change the slope ofthe trendbecause Si and Al, which assume greater weight than Tiand Mg in the plot, apparently control the trend. Thus,the slope of the trend in Figure 7 is not a sensitive indi-cator for particular substitutions.

2. From a near ideal linear trend with a slope of - 1between Ba + 2Ti + 3Al and K + 3(Mg,Fe) + 3Si, Man-sker et al. (1979) argued for a combined substitution of

rr2rBa + t6r! + 2r6rTi + 3t4lAl

: rr2tK _u 3ror(Mg,Fer+) + 3ratSi (g)

for the Hawaiian barian titanian biotite. As Substitution9 involves an O-site vacancy, the l:1 replacement shouldindicate a deviation from the proposed substitution rath-er than an accordance with it. This trend does not parallelthe join linking phlogopite and an end-member createdby Substitution 9, KBaMg3Ti2Si3Al5O,o(OH)", which is alinear combination of Substitutions 2, 5, and 6 in theproportions l:1:1. Whereas a constant Ba:Ti ratio of 0.5is specified by Substitution 9, the Hawaiian biotite sam-ples have variable Ba:Ti ratios (0.54-0.81, Mansker et al.,1979). Therefore, the data indicate that all the three sub-stitutions operate in the Hawaiian biotite samples at vari-able ratios.

3. Recently, Guo and Green (1990) proposed anothersubstitution,

rr2lBa + rr2tD + 3t61Ti + t61[] + 4t4rAl

:2' l2tK 1 4rel(Mg,Fe2+) + 4r4rsi (10)

10.0

8 . 0

0.01 0 . 0 r2 .0 14 .0 16 .0 18 .0 20 .0

23i+{Mg+ft2.)

Fig.1. Aplotof 2Si + (Mg + Fe'?t)vs.2(Al + Fer*) + Tifor the WEK phlogopite. Calculated with O : 22. Symbols as inFig.2.

as a best fit for natural and synthetic phlogopite frompotassic rocks (mainly lamproites). It is a linear combi-nation ofSubstitutions l, 6, and 5 in the proportions ofl: l:2. Therefore, it cannot be verified for the same reasonthe Hawaiian biotite cannot be. As Substitution l0 in-volves a significant proportion of the Ti-Tschermak's end-members, it is very unlikely to be applied to phlogopite

samples in lamproites because of their Al-deficient char-acter (e.g., Mitchell, 198 l) and a negative correlation be-tween AlrO, and TiO, (Fig. ab). The Hawaiian data(Mansker et al., 1979) fit both Substitutions 9 and l0well, further illustrating that complex parameters cannotdiscriminate sensitively among alternative substitutions.

Ti substitution schemes vary among micas crystallizedfrom diverse magma compositions and under differentphysical conditions (Edgar and Arima, 1983). Therefore,no general applicability of a particular scheme can beassumed. For instance, Ti vacancy and talFe3+ : t4lAl sub-stitutions (Mitchell, l98l), but not Ti-Tschermak's, maybe predominant in lamproitic micas for the reasons dis-cussed above. In contrast, Tschermak's substitution islikely to prevail, accompanied by minor Ti-Tschermak's,in phlogopite in potassic rocks related to subduction (e.9.,

Roman Province).

Pln-tcnNnsrs AND MICA coMPosrrloN

Textural relations indicate that the phlogopite in theWEK lavas is a late phase crystallized from residual po-

tassic melts at near-surface pressures and low tempera-tures, e.g., <960'C in olivine leucitites (Fe-Ti oxide tem-perature, Zhang, unpublished data). The presence ofleucite in inclusion DZ2n also suggests that it formed atlow pressures (probably <4 kbar, Taylor and Mackenzie,1975). As Mg' values for the phlogopite inDZ2n (62.7-

Ti-Tsch

A

Phl

$uo,tR

oo

2.0

to64 ZHANG ET AL,: BARIAN TITANIAN PHLOGOPITE

70.1) are slightly lower than for the phlogopite in thegroundmass of olivine leucitites (66.5-71.1), DZ2n mayhave crystallized from a more evolved potassic magma.

Differences in composition between the barian titanianphlogopite in olivine leucitite and the corona phlogopitein leucite basanite cannot be attributed to the differenthost magma compositions. The corona phlogopite showslower BaO, AlrOr, and TiO, but higher Mg'than the bar-ian titanian phlogopite, whereas the host for the former(leucite basanite wZKl6) in higher in Ba, Al,O., andTiO, but lower in Mg' than for the latter (olivine leuci-tites). Differences in paragenesis between WEK olivineleucitites and leucite basanites may provide an explana-tion for the contrast. Neither leucite nor nepheline, bothof which are present in the olivine leucitites, is the re-pository for Ba. Ba is hardly above 0.06 wto/o in leuciteand even lower in nepheline (Zhang, unpublished data).Therefore after crystallizing >850/o Ba-free phases, theresidual olivine leucititic melts are so enriched in Ba thatbarian titanian phlogopite can crystallize (Zhang et al.,1993). In contrast, potassium feldspar, a major ground-mass phase in the leucite basanites, incorporates a mod-erate amount of BaO (0.2-1.8 wto/0, av. 0.5 wto/o; Zhang,unpublished data). Thus, the residual liquids in WZKl6,with which olivine phenocrysts interacted to form phlog-opite, must have undergone Ba depletion. Higher Mg' inthe phlogopite coronas is attributed to the forsteritic ol-ivine phenocrysts that dominate the MgO budget of thereaction, whereas low AlrO. content reflects the com-bined effects of AlrO.-absent olivine phenocrysts plusAl,O.-depleted residual liquids caused by potassium feld-spar crystallization. Therefore, the chemistry of magmat-ic phlogopite depends on not only the composition ofhost magmas and intensive variables but also mineralassemblages (Foley, 1990) and crystallization sequences.The latter factors can be employed to explain the lack ofbarian phlogopite in extremely Ba-rich lamproites be-cause Ba-rich accessory phases, such as priderite, jep-peite, and shcherbakovite, compete for Ba.

AcxNowr,nncMENTS

Financial support for this study was generously provided to M.Z. bythe late Stephen Hui through the Anna and Stephen Hui fellowship atIC. Support at SMU was provided by the Department ofGeological Sci-ences and by NSF grant EAR-90-04812 to M.A.D. Discussions with MikeHoldaway are appreciated. We thank N. Royall and D. Deuring for theirassistance in microprobe analysis. The manuscript was improved by thereview of D.A. Hewitt and R.F Wendlandt

RrrnnnNcps crrEDAbrecht, J , and Hewitt, D.A. (1988) Experimental evidence on rhe sub-

stitution of Ti in biotite. American Mineralogist, 73, 1275-1294.Bachinski, S W , and Simpson, E L. (1984) Ti-phlogopites of the Shaw's

Cove minette: A comparison with micas of other lamprophyres, po-tassic rocks, kimberlites, and mantle xenoliths. American Mineralogist,69 ,4 t - 56 .

Barnett, R.L., Arima, M., Blackwell, J.D., Winder, C.G, palmer, H.C.,and Hayatsu, A (1984) The Picton and Varty Lake ultramafic dikes:Jurassic magmatism in the St. Lawrence platform near Belleville, On-tario. Canadian Joumal ofEarth Sciences. 21. l46o-14i2.

Barton, M (1979) A comparative study of some minerals occurring inthe potassium-rich alkaline rocks of the Leucite Hills, Wyoming, theVico Volcano, Western Italy, and the Toro-Ankole Region, Uganda.Neues Jahrbuch fiir Mineralogie Abhandlungen, 137, ll3-134.

Bence, A.E, and Aibee, A L. (1968) Empirical correction factors for theelectron microanalysis of silicates and oxides Joumal of Geology, 76,382-403

Birch, W.D. (1978) Mineralogy and geochemistry of the leucitite at Cos-grave, Victoria. Joumal ofthe Geological Society ofAustralia, 25,369-38 5 .

-(1980) Mineralogy of vesicles in an olivine leucitite at Cosgrave,Victoria, Australia. Mineralogical Magazine, 43, 597 -603.

Boctor, N.2., and Yoder, H.S. Jr (1986) Petrology of some melilite-bear-ing rocks from Cape Province, Republic ofSouth Africa: Relationshipto kimberlites. American Journal ofScience, 286, 513-539.

Bol, L.C.G.M., Bos, A., Sauter, P C.C , and Jansen, J.B H. ( 1989) Barium-titanium-rich phlogopites in marbles from Rogaland, southwest Nor-way. American Mineralogist, 74, 439-447.

Brigatti, M.F., Galli, E , and Poppi, L. (1991) Effect of Ti substiturion inbiotite-/M crystal chemistry. American Mineralogist, 7 6, I 17 4-l 183.

Cao, R.L., and Zhu, SH. (1987) Mantle xenoliths and alkali-rich hostrocks in eastern China. In PH. Nixon, Ed., Mantle xenoliths, p. 167-180. Wiley, Chichester, U K.

Deer, W.A., Howie, R.A., and Zussman, J. (1962) Rock-forming miner-als. III. Sheet silicates, p. 42-54. Longmans, London.

Dymek, R.F. (1983) Titanium, aluminum and interlayer cation substi-tutions in biotite from high-grade gneisses, West Greenland. AmencanMineralogist, 68, 880-899.

Edgar, A.D. (1979) Mineral chemistry and petrogenesis ofan ultrapotas-sic-ultramafrc volcanic rock Contributions to Mineralogy and Petrol-ogy ,7 l , 17 l - 175 .

Edgar, A.D., and Arima, M. (1983) Conditions of phlogopite crystallizationin ultrapotassic volcanic rocks. Mineralogical Magazine, 47, ll-19.

Foley, S.F (1990) Experimental constraints on phlogopite chemistry inlamproites. II Effect of pressure-temperature variations EuropeanJournal of Mineralogy,2, 327-341

Gaspar, J.C., and Wyllie, P.J (1987) The phlogopites from the Jacupi-ranga carbonatite intrusions. Mineralogy and Petrology, 36, 121-1 34.

Guo, J., and Green, T.H. (1990) Experimental study ofbarium partition-ing between phlogopite and silicate liquid at upper-mantle pressure andtemperature. Lithos, 24, 83-95.

Hewitt, D.A., and Abrecht, J. (1986) Limrtations on the interpretation ofbiotite substitutions from chemical analyses of natural samples. Amer-ican Mineralogist, 7 l, | | 26- 1 128.

Holm, P.M. (1982) Mineral chemistry of perpotassic lavas of the Vulsi-nian district, the Roman Province, Italy. Mineralogical Magazinq 46,379-386.

Hsii, K.J. (1989) Origin of sedimentary basins of China. In X. Zhu, Ed.,Chinese sedimentary basins: Sedimentary basins ofthe world, p. 207-227 . Elsevier, Amsterdam.

Mansker, W L., Ewing, R.C., and Keil, K (1979) Barian-titanian biotitesin nephelinites from Oahu, Hawaii. American Mineralogist, 64, 156-I 59.

Mitchell, R.H. (1981) Titaniferous phlogopites from the Leucite lam-proites of the West Kimberley area, Western Australia. Contributionsto Mineralogy and Petrology, 7 6, 243-25 l.

-(1985) A review of the mineralogy of lamproites Transactions ofthe Geological Society of South Africa, 88, 411-437 .

Mitchell, R.H., and Bergman, S.C. (1991) Petrology oflamproites, p. 169-217. Plenum, New York.

Mitchell, R.H., and Platt, R.G. (1984) The Freernans Cove volcanic suite:Field relations, petrochemistry, and tectonic setting of nephelinite-bas-anite volcanism associated with rifting in the Canadian Arctic Archi-pelago Canadian Joumal ofEarth Sciences, 21,428-436.

O'Brien, H.E., Irving, A, and McCallum, I.S ( I 988) Complex zoning andresorption of phenocrysts in mixed potassic mafic magmas of the High-wood Mountains, Montana American Mineralogist, 73, l0O7-1O24.

Qi, J.Y., and Xiao, Z.Y. (1985) On titanium-phlogopite megacrysts inalkali basalts. Acta Mineralogica Sinica, 5, 351-358 (in Chinese).

Rock, N.M S. (1991) Lamprophyres, p. 64-65. Blackie, Glasgow, Scot-land.

ZHANG ET AL.: BARIAN TITANIAN PHLOGOPITE 1065

Ryabchikov, I D., Kovalenko, V.I., Dikov, Yu P , and Vladykin, N V.(1981) Titaniferous micas from the mantle: Composition, structure,formation conditions, and possible role in the production of potassicalkali magmas. Geokhimiya, 6, 873-888

Shmakin, B.M (1984) Causes and consequences ofhigh contents ofbar-ium in sheet muscovite and phlogopite. Proceedings ofthe 27th Inter-national Geological Congress. XV. Non-metallic mineral ores, p 261-271. VNU Science. Utrecht. The Netherlands.

Taylor, D , and Mackenzie, W S. (19?5) A contribution to the pseudoleu-cite problem. Contributions to Mineralogy and Petrology, 49,321-333.

Thompson, R.N. (1977) Primary basalts and magma genesis. III. AlbanHills, Roman comagmatic province, central Italy Contributions toMineralogy and Petrology, 60, 9l-108.

Tronnes, R G , Edgar, A.D., and Arima, M. (1985) A high pressure-hightemperature study of TiO, solubility in Mg-rich phlogopite: Implica-tions to phlogopite chemistry. Geochimica et Cosmochimica Acta,49,z ) L J - Z t Z '

Velde, D. (1979) Trioctahedral micas in melilite-bearing eruptive rocks.

Camegie Institution of Washington Year Book, 78,468-475.Wagner, C., and Velde, D. (1986) The mineralogy of K-richterite-bearing

lamproites. American Mineralogist, 7 1, l'7 -37 .Wendlandt, R.F. (1977) Barium-phlogopite from Haystack Butte, High-

wood mountains, Montana. Carnegie Institution of Washington Year

Book.76. 534-539Zhang,M, Menzies, M.A., Suddaby, P., and Thirlwall, M.F. (1991) EMl

signature from within the post-Archaean subcontinental lithosphericmantle: Isotopic evidence from the potassic volcanic rocks in NE Chi-na. Geochemical Journal, 25, 329-340.

Zhang,M., Suddaby, P., Thompson, R.N., and Dungan, M A (1993) The

origin of contrasting zoning patterns in hyalophane from olivine leu-citites, Northeast China Mineralogical Magazine, 57, in press.

MeNuscprm RECEIVED Jurv 13, 1992MaNuscnrsr ACCEPTED Mnv 10, 1993