American Mineralogist, Volume 76, pages 548-560, 1991

Magmatic processes in titanite-bearing dacites, central Andes of Chile and Bolivia

Snrsuvl Nlx,l,olDepartment ofEarth and Planetary Sciences, Faculty ofScience 33, Kyushu University, Hakozaki, Fukuoka 812, Japan

Ansrnc.cr

Dacites in the central Andes contain euhedral phenocrysts of titanite. In the studieddacites, pargasite phenocrysts coexist with hornblende phenocrysts, neither of which shownotable compositional zoning, implying mixing of pargasite-bearing mafic magma andhornblende-bearing rhyolitic magma just before eruption. The rhyolitic magma was highlyoxidized, cool, and hydrous at shallow levels in the crust and was underlain by the hotter,less-oxidized mafic magma. Titanite crystallized from the rhyolitic magma.

Titanite phenocrysts record evidence for reduction and oxidation during crystallizationaccording to a reaction such as titanite + Ti-rich magnetite + quartz : ilmenite * calcicsilicate + O. The change of /o, probably resulted from devolatilization of magma at thetop of the chamber. Heating of titanite phenocrysts during mixing or unloading duringeruption caused vesiculation of their melt inclusions, fragmenting the phenocrysts.

INrnooucrroN

Although it is a common mineral in plutonic and meta-morphic rocks, titanite is rare in volcanic rocks (e.g., Deeret al., 1982). Compared with titanite in plutonic andmetamorphic rocks (Paterson et al., 1989; Franz andSpear, 1985), volcanic titanite has not been described indetail. Ewart (1979) summarized occurrences of titanitephenocrysts in silicic volcanic rocks from the Andes,westem United States, and Mediterranean regions. Thestability of titanite in magmas has been discussed by sev-eral authors (Verhoogen, 1962; Carmichael and Nicholls,1967;Lipman, l97l; Wones, 1989), who noted that it isstable in oxidized magmas. Because the occurrence oftitanite phenocrysts is sensitive to fr, its presence as aphenocryst allows us to place constraints on /o, in themagma chamber. However, experimental results of par-tial melting at high pressure and temperature (Yoder andTilley, I 962; Helz, 197 3; Hellman and Green, I 979; Greenand Pearson, 1986a, 1986b, 1987) show that titanite isstable in hydrous silicic melts of high-Ti content regard-less of /or. Thus, important questions are whether or nottitanite in volcanic rocks crystallized directly from sili-cate melts, and what conditions stabilize magmatic tita-nite.

Volcanic rocks from the central Andes, including thetitanite-bearing dacites from the central Andes which Ihave studied, commonly show evidence for magma mix-ing (Thorpe et al., 1984; O'Callaghan and Francis, 1986);therefore, clarifying the mode of mixing is an importantquestion regarding the stability oftitanite. In this paper,I describe in detail titanite-bearing dacites from the cen-tral Andes and discuss the stability of titanite and mag-matic processes in the silicic magma chambers.

0003-004x/9 l/0304-0548$02.00 548

Gnor,ocrclr. sETTTNG AND SAMPLTNG

The central volcanic zone of the Andes (Thorpe andFrancis, 1979) is situated on continental crust with athickness of 70 km or more (James, l97l). Sajama, Pari-nacota, and Porquesa Volcanoes stand near the borderbetween Chile and Bolivia (Fig. l). They are andesite-dacite stratovolcanoes rising upward to 5200-6500 m onan extensive ignimbrite plateau of 4000 m elevation. Nei-ther Sajama nor Porquesa Volcano has been studied geo-logically or petrologically. Parinacota Volcano was stud-ied by Katsui and Gonzalez (1968) and Wirrner et al.( l e88).

The summit of Sajama Volcano (Nevado de Sajama)is covered with a glacier. Parasitic lava domes and flowsoccur on its deeply glaciated slopes. The parasitic conesare arranged in a N70'E-S70'W direction and appear tobecome older with distance from the summit. ParinacotaVolcano is one of a pair of volcanoes (Nevados de Paya-chata); Pomerape Volcano stands just to the north. Theactivity of Parinacota Volcano started in the Pleistoceneand continues to the present (Wiirner et al., 1988). Thisvolcano records two stages ofgrowth; glacial erosion andcollapse of the volcano took place between the two stages.Porquesa Volcano consists oftwo adjacent cones ofthicklava flows and domes (M. Takahashi, personal commu-nication, 1989). The relationship between age and conemorphology in the Payachata volcanic field (Wdrner etal., 1988) indicates that Sajama and Porquesa Volcanoesprobably formed during a period extending from the lateMiocene to the Pleistocene.

Titanite-bearing dacites were collected from lava flowsand domes produced by different stages of eruption foreach volcano: three samples from different parasitic cones

,) ']pSojomo

BOLIVIA

taqa

a a

aa

NAKADA: TITANITE-BEARING DACITES 549

66"

Fig. l. Map showing distribution of Pliocene-Recent vol-canic centers in the central Andes volcanic zone (dots and stars: Thorpe et al., 1982). Stars are volcanoes from which titanite-bearing dacites were sampled (Sajama Volcano, 18t7,968"53,W;Parinacota Volcano, 18"10'S-69"09'W; Porquesa Volcano,19"58'S-68"44'W).

at Sajama Volcano (706,708, and 801), two samples fromthe second and third stages at Parinacota Volcano (M62and M8l, respectively Katsui and Gonzales, 1968), andtwo samples from different cones at Porquesa Volcano(POR-01 and POR-02). The modal compositions are list-ed in Table l. All are dacites with abundant phenocrystsof plagioclase, amphibole, and biotite. Titanite pheno-cryst content reaches 0.4 volo/o; that of quartz is charac-teristically low (<2olo). Mg-rich olivine phenocrysts co-exist with quartz and sanidine phenocrysts in the Sajamavolcanic rocks. All rocks have a glassy groundmass that

TABLE 1. Modal compositions of titanite-bearing dacites

contains various sizes of vesicles. The groundmass isvariably devitrified, and spherulitic aggregates and crys-tobalite pods may be present.

AN^lr,vrrc,ll, METHoDS

Whole-rock composition was determined by X-ray flu-orescence (XRF: Rigaku-GF3063P) at Kyushu lJniver-sity following the methods of Nakada et al. (1985) andNakada (1987). Analytical errors for trace elements are:50 ppm for Ba; l0 ppm for Sr, Zr, and V; 5 ppm for Rb;and 2 ppm for Nb and Y. Sr isotopic ratios of two sam-ples from Sajama Volcano were obtained at OkayamaUniversity with the method of Kagami et al. (1982).

Chemical compositions of minerals and glass were de-termined at Kyushu University by a scanning electronmicroscope (JEOL-SEM35-CFII) equipped with an en-ergy-dispersive spectrometer (LINK 860-2-500), using l5kV accelerating voltage, a specimen current of 1.5 nA oniron metal, and the ZAF program (Shinno and Ishida,1983). Glass and exsolved Fe-Ti oxide minerals were an-alyzed by sweeping the focused electron beam over asquare area as large as possible in order to prevent sodi-um loss from glass and to approximate the integratedbulk composition of Fe-Ti oxide grains. Relative abun-dances of trace elements in titanite are considered reliablealthough absolute values may not be. Detection limits are0. 15, 0.35, 0.25,0.15, and 0. l5 wto/o for Y2O3, La2O3,CerOr, Nd2Oj, and SmrOr, respectively. Backscatteredelectron images of the titanites were observed on the sameinstrument. For comparison, titanites from Cretaceousgranitoids in northern Kyushu (e.g., Tsusue et d., 1984)were also analyzed.

Wrror,r-nocK coMPosrrroNs

Whole-rock analyses (Table 2) show that the dacitesare calcalkalic. Sr isotopic ratios of the Sajama volcanicrocks are as high as those ofthe Parinacota rocks (Table2; Notsu et al., 1985). Wdrner et al. (1988) have notedenrichment in large-ion-lithophile elements in the Pari-nacota volcanic rocks compared to other volcanic rocksin the central Andes. It is likely that volcanic rocks fromthis region around l8'S and far from the trench (Sajamaand Parinacota) are more enriched in these elements and

Nos. g.m. qz san plag ampn ap

1.0 7 .2 1 .7 0 .2 1 .00.6 5.3 2.7 0.1 0.71.3 2.1 3.5 0.1 1.0

- 0.5 4.1 0.1 0.50.2 7.1 1.9 0.1 0.5

- 4.1 0.7 + 0.5- 6.8 0.4 + 1.0

+ - 11 .5 0 .20.7 1.9 18.0 0.3+ 0.7 17.9 0.3

1.4 4.3 19.4 0.40.1 1.5 13.7 0.3

0.1 0.5 22.3 0.31.3 - 19.7 0.1

+

+

77.169.673.1

69.274.3

71.370.5

Saiama801706708

ParinacotaM62M81

PorquesaPOR-01POR-02

/Vote.'9.m.:groundmass,qz:quartz,san:sanidine,plag:plagioclase,tit:titanite,ol :olivine,cpx=clinopyroxene,amph:amphibole,bt=biotite,ap : apatite, ore : Fe-Ti oxide minerals; + : <0.1%. All in volo/". 2000 points were counted for each sample except M62 and M81 . Data for M62 andM81 are from Katsui and Gonzalez (1968). M81 contains 0.3 vol% of orthopyroxene.

550 NAKADA: TITANITE-BEARING DACITES

Tlele 2. Chemical analyses of titanite-bearing dacites

Sajama Parinacota Porquesa

708 M81 M62 POR-02

sio,Tio,Alro3FerO.*MnOMgoCaONaroKrOP,OuBaNbZr

SrRb

I'Sr/sSr

63.900.92

15.395.040.072.373.964.293.790.27

1 260a

2341 1

8351241 1 5

0.70665(4)

67.950.72

14.793.870.061.322.654.124.29o.24

1 2301 4

233o

716165770.70664(5)

68.040.68

14.873.760.061.312.624.094.35o.22

12901 3

2348

69515272

n.d.- '

64.370.96

15.824.870.061.853.724.483.500.37

1 1901 1

2156

87998

109

69.490.57

14.673.670.061.022.464.083.820.16

10001 0

1574

57612564

63.390.93

15.815.390.092.384.264.442.990.32

992I

1697

78680

1 1 00.70555t

69.350.62

14.633.540.061 . 1 32.534.273.670.19

10508

1493

480106640.70595i0.70681 -O.70691 t

A/ote.'Major-element contents in wt7o, recalculated on Hro-free basis. Trace-element contents in ppm.- Total Fe as FerO".

" The symbol n.d. : not determined.f Notsu et al. (1985).

Zr than rocks from elsewhere in the central volcanic zonerocks. The depletion of Y in these dacites may supportthe idea of Aramaki et al. (1984) and Onuma and Mon-toya (1984) that magmas on the continental side of thecentral volcanic zone evolved mainly by clinopyroxenefractionation.

PrrnNocnysrs IN DAcrrES

Titanite

The occurrence of titanite phenocrysts as euhedral dis-crete grains up to I mm long is a common feature of theseven samples in this study (Figs. 2 and 3). Sometimes,aggregates of titanite phenocrysts are observed (Fig. 3C).Titanite is rarely included in hornblende phenocrysts.Some titanite phenocrysts (POR-02) are surrounded byaggregates of fine, slender ilmenite crystals (Fig. 3D). In-clusions of ilmenite, hematite, rhyolite glass, and apatiteare common near the core of titanite phenocrysts (Figs.2, 34, and 3B). In some samples, inclusions of ilmeniteand hematite are concentrated along a zone in the phe-nocryst that represents an earlier crystal surface (M62;Fig. 2). A thin film ofglass is observed around the inclu-sions (Figs. 2A and 2B), indicating that the Fe-Ti oxideswere trapped together with melt. Titanite phenocrysts areoften fragmental (Figs. 2,{ and 3C) and contain cracksfilled with glass that appear to originate at glass inclusions(Fig. 2B). The cracks and fragmentation are likely to haveresulted from expansion of the melt inclusions.

Backscattered electron images of titanite phenocrystsshow zoning patterns, in which the bright zones are en-riched in REE. The zoning patterns in phenocryst rimsare concordant with the crystal outlines (Figs. 2C, 38,

and 3C). However, their inner parts typically have com-plex zoning patterns. Phenocryst cores or inclusion areaswith complex zoning patterns (Figs. 2C and 3B) recallhopper crystal forms, which are commonly found in ol-ivine phenocrysts formed at high growth rate (Donald-son, 1976). Bacon (1989) suggested that zones enrichedin accessory phases or melt inclusions within phenocrystsreflect episodes of rapid crystal $owth and are distinctfrom zones formed by resorption of the host; rapid growthrate is thought to create a phenocryst-melt boundary lay-er differentiated enough for accessory-phase saturation.Although the inclusion zones of the titanite phenocrystsindeed may have formed at rapid gfowth rates, cores ofthe phenocrysts typically show corrosion textwes; earliersurfaces are irregularly rounded and the concentric zon-ings are cut by the corrosion surfaces (Figs. 2C and 3C).Rounded and irregular forms of ilmenite inclusions sur-rounded by glass indicate that resorption of ilmenite tookplace before rapid nucleation of titanite on th€ ilmenitesubstrata. Thus, the textures of titanite phenocrysts in-dicate the following episodes during crystallization: (l)precipitation and later resorption of titanite; (2) ilmenite(and hematite) precipitation at titanite margins; (3) re-sorption of ilmenite and resumption of precipitation oftitanite at rapid growth rates; (4) titanite crystal growthat relatively slow rates; and (5) fragmentation of pheno-crysts.

The titanite contains small amounts ofAl, Fe, and REE(Table 3, Fig. 4). Titanite in the Porquesa volcanic rockshas lower Fe, Al, and REE contents than that in the Sa-jama and Parinacota volcanic rocks (Fig. 4). Titanite iszoned in Ti, Ca, Al, Fe, and REE, but there are no sys-tematic chemical changes from core to margin (Table 3).

NAKADA: TITANITE.BEARING DACITES 5 5 1

Fig. 2. Titanite phenocrysts in dacite from Parinacota Vol-cano (M62). Scale bar : 100 pm. (A) Photomicrograph of eu-hedral titanite phenocryst in reflected light. The phenocryst (lightgray-colored) lost its left half prior to eruptive quenching andthe broken-edge is in contact with groundmass glass. Large boxedinclusion is rhyolitic glass. (B) Photomicrograph ofboxed areain A. Cracks extending from a glass inclusion (gl) are also filledwith rhyolite glass. Ilmenite inclusions (ilm) are surrounded in

part by a glass filrn. (C) Backscattered electron image showinggrowth zoning. White, rounded inclusions (ilmenite and hema-tite) parallel the edges of the phenocryst. The growth zoningpattern is concordant with the edge ofthe phenocryst and withthe arrangement of Fe-Ti oxide inclusions. (D) Backscatteredelectron image of Fe-Ti oxide inclusions. All ilmenite grains showexsolution, although hematite (hem) is homogeneous.

The zones with ilmenite and melt inclusions, or zonessurounding the resorbed cores, are usually enriched inREE (Figs. 2C and,3C). Fe content increases with Al con-tent (Figs. 4 and 5) and with decrease in Ti content. )REEcontents increase with decrease in Ca, and do not cor-relate with Ti content. A poor positive correlation existsbetween REE and Fe contents. Such variations were alsoobserved in titanites from granodiorites (Paterson et al.,1989). Probably, Al (not Fe) substitutes mainly for Ti,whereas the REEs substitute for Ca.

Paterson et al. (1989) reported sector zoning oftitanitefrom granodiorites and showed that REE contents ofthefast-growing sectors are higher than those of the slow-growing sectors. This may imply that the diffusivity ofCa is lower than those of REE in silicate melts, REE wereincorporated by the fast-growing titanite more easily thanCa. High REE contents in the inclusion-rich zones sur-

rounding the corroded cores of the titanite phenocrystssupport rapid growth ofthese zones.

Amphibole

Amphibole is represented in all thin sections as euhe-dral large crystals of edenitic- to magnesio-hornblende,as much as 2 mm long, and as skeletal small crystals ofpargasite, less than 0.5 mm long. Both amphibole sam-ples are reddish brown in color, but some phenocrysts areopacitized. Hornblende sometimes contains Ti-rich mag-netite (commonly referred to as titanomagnetite) and ti-tanite inclusions. It occurs in aggregates with biotite, andwith biotite and clinopyroxene in the Sajama volcanicrocks. Pargasite is rarely aggregated with olivine in theSajama volcanic rocks. Amphibole grains are slightlyzoned; the rims of hornblende are richer in Mg than thecore, whereas the rims of pargasite are richer in Fe than

552

Fig. 3. Titanite phenocrysts in dacites from Porquesa andSajama Volcanoes. Scale bar : 100 pm. (A) Photomicrographin reflected light (POR-01). Glass (gray) and ilmenite (white)included in titanite phenocryst (ight gray). (B) Backscatteredelectron image of A. Zoning pattern is complex in the core andsimple in the rim. (C) Backscattered electron image of aggregate

NAKADA: TITANITE-BEARING DACITES

of titanite phenocrysts (706); mt : Ti-rich magnetite. The grainin the lower part ofthe photograph is fragrnented and the bro-ken edge is in contact with goundmass glass. A titanite fragmentis present in the upper right. All titanite phenocrysts have cor-roded cores. (D) Backscattered electron image of titanite phe-nocryst fringed with ilmenite needles (POR-02).

TABLE 3. Representative analyses of titanite phenocrysts

Saiama706

ParinacotaM62

PorquesaPOR-01

sio,Tio,Alro3FeO-MnOMgoCaONaroYro.LarogCerooNdrosSmrO.

TotalFelAl

29.436.51 . 1 91.450.130.09

26.70.06

o.440.430.17

28.935.41.331.820.06

25.5

0.270.741 . 1 0

95.10.96

29.436.61 .231.500.27

26.7

1 .061.O70.56

29.037.01 .251.410.240 . 1 1

26.40.27

0.671.O70.20

97.60.80

29.236.31.361.480.30

27.1

29.137.71 .011.200.09

26.s

30.037.70.991.31

27.4

0.37

0.290.66

98.70.95

98.30.86

96.50.85

0.500.860.42 0.490.33

97.8 96.00.78 0.83

Note.'Dash indicates less than the lower detection limit of the element (see text).'Total Fe as FeO.

NAKADA: TITANITE-BEARING DACITES

0.10

Fe

0.05

o gronitoids . ,1;^ metomorphic r. ,/rS;iiid;r;:i';Jillo C.Andes volc. r .

i9

aii^e^^^ ^^ ̂

xx xX xx

553

0 0

oes T i 10 o .es cq 1 0

Fig. 4. A1, Fe, and ZREE contents plotted against Ti and Cacontents for titanite phenocrysts in Andean dacites; atomicabundances calculated on a five-O atom basis. Symbols: circle,Sajama; square, Parinacota; triangle, Porquesa. Dotted area isfor titanite in Japanese ganitoids (Nakada, unpublished data).Crosses indicate analytical error.

the core (Table 4). However, pargasite is always moreenriched in Mg/(Mg * Fe,o,), t4rAl, Ti, and Na than horn-blende (Fig. 6). M/(Mg + Fe,.,) ratio ranges for pargasiteare 0.75-0.68 in the Sajama volcanic rocks, 0.68-0.65 inthe Parinacota volcanic rocks, and 0.74-0.62 in the Por-quesa volcanic rocks. For hornblende, these ratios are0.6 6-0. 5 8, 0.64-0.62, and 0.64-0. 5 6, respectively.

Other mafic silicates

Biotite occurs in all thin sections as flaky reddish-brownphenocrysts, less than 0.5 mm long, which do not shownotable chemical zoning (Table 5). The Mg/(Mg + Fe.,)ratio is 0.66-0.58 in the Sajama volcanic rocks, 0.60-0.58 in the Parinacota volcanic rocks (M62), and 0.54-0.53 in the Porquesa volcanic rocks (POR-02).

Olivine and pyroxene were observed only in the Saja-ma volcanic rocks. Euhedral olivine phenocrysts, less than0.5 mm across, show no reaction relation with pyroxene.The olivine is normally zoned in Mg; the Mgi(Mg + Fe,o,)ratio of olivine is 0.82-0.78 (Table 6). Both diopside andFe-rich diopside (formerly termed salite) phenocrysts arepresent in the dacites from Sajama Volcano. Diopside islarger (as much as 0.5 mm long) and less abundant thanFe-rich diopside; the former shows normal zoning in Mgbut the latter hardly shows chemical zoning (Table 6).Wo content ranges from 42 to 44 molo/o in diopside andfrom 45 to 47 molo/o in Fe-rich diopside, whereas Mg/(Mg * Fe,..) ratios range from 0.83 to 0.78 and from 0.77to 0.72, respectively. The Mg/(Mg + Fe,",) ratio of di-

0.10 0.15

Fig. 5. Fe,., plotted against Al contents for titanite pheno-crysts in Andean dacites. The compositions of titanite crystalsin granitoids (Gromet and Silver, 1983; Ayuso, 1984; Sawka andChappell, 1988; Nakada, unpublished data), metarnorphic rocks(Ernst and DalPaiz,1978; Ghent and Stout, 1984; Franz andSpear, I 985; Nishiyama, unpublished data), and products ofhigh-temperaturc and high-pressure experiments (Hellman and Green,1979; Green and Pearson, 1986b) are shown for comparison.

opside is similar to that of olivine, indicating equilibriumbetween olivine and diopside.

Fe-Ti oxide minerals

In all specimens, Ti-rich magnetite occurs as micro-phenocrysts, less than 0.5 mm across, or as inclusions inhornblende phenocrysts. It always shows exsolution tex-tures with fine lamellae of hematite-ilmenite solid solu-tion. Ti-rich magnetite rarely occurs in composite grainswith ilmenite. Ilmenite rarely occurs as small crystals inthe groundmass, which show no exsolution textures. Nogroundmass ilmenite could be observed in the Parinacotasamples. Partitioning of Mg and Mn between Ti-richmagnetite and ilmenite grains in these dacites (Table 7)indicates equilibrium between these coexisting phases(Bacon and Hirschmann, 1988).

Other minerals

Plagioclase is the most abundant phenocryst and mi-crophenocryst in all samples. Phenocrysts, as much as 2.5mm long, can be either clear or dusty. The latter showthe same texture as the dusty plagioclase described byTsuchiyama (1985); a corroded crystal with a partiallydissolved margin (ca. 0.1 mm thick) is mantled by aclear,thin rim (as much as 30 m thick). The chemical compo-sitions of feldspar phenocrysts were determined only forthe samples from the Sajama Volcano. In the Sajamavolcanic rocks, the cores of clear microphenocrysts arethe richest in Ca (An55) (Fig. 7); the most sodic compo-sitions (An25) are found in the cores of dusty pheno-crysts. The compositions of rims in both types of plagio-clase approach An30. Euhedral crystals of sanidine, asmuch as 1.5 mm long, occur in SiOr-rich samples; they

0.05AI

0 9 0

0

T

554 NAKADA: TITANITE-BEARING DACITES

0.

2

1.5

3

.1

5

0

7

tnl41

\')

Ti

No

M9l(M9+f,sz'1

Ab 50 -0rFig. 7. Ternary An-Ab-Or plot for plagioclase and sanidine

phenocrysts in dacites from Sajama Volcano. Solvi and tie linesfor feldspars coexisting in equilibrium are shown for 750'C and825 qC at I kbar (Fuhrman and Lindsley, 1988).

are sometimes corroded, and zoned from 0166 to 0160(Fig. 7). Tie lines (Fig. 7; Fuhrman and Lindsley, 1988)indicate that the sanidine is in equilibrium with the coresof dusty plagioclase. In all rocks, quartz is euhedral butcorroded, and apatite is a microphenocrystic phase form-ing needles as much as 0.3 mm long.

Sulfide could not be identified even as inclusions ofphenocrysts in this study.

INcr,usroNs rN TrrANrrE pHENocRysrs

Glass inclusions in titanite are rhyolitic and enrichedin KrO (Table 8); KrO/NarO ratios are much higher thanin whole-rock samples. TiO, contents range from 0.34 to0.15 wto/o. The normative anorthite contents are less than4 wto/o, and small amounts of normative corundum arepresent. High KrO/NarO ratios and corundum-norna-tive compositions may be attributed to Na migration awayfrom the electron beam during analysis. Glass inclusions

0.8

0.7

0 .612

Totol AlFig. 6. Tetrahedral A1, Ti, Na, and Mg/(Mg + Fe2*) plotted

against total Al of amphibole phenocrysts in Andean dacites.Atomic abundances calculated on a 23-O atom basis. Fe3+/Fe,o,was assumed as 0.3 accordilg to the experimental redox con-ditions ofSpear (1981). Symbols as in Figure 4.

TABLE 4, Representative analyses of coexisting amphibole phenocrysts

a c 0 r eo r i m

s ty p tog , core

Salama (708) Parinacota (M62) Porquesa (POR-01)

Hb Par

core core core core @re core

sio,Tio,Al2o3FeO*MnOMgoCaONa2OKrO

Totalmg

46.51.757.47

14.430.29

13.71 1 . 62.011.09

98.862.8

48.3 42.81.36 3.746.35 10.3

13.4 10.60.31 0.08

14.5 15.111.7 1 1 .62.06 2.720.79 0.94

98.7 97.965.8 71.8

42.4 47.23.79 1.56

11.0 6.8711.5 14 .60.13 0.36

14.1 13.31 1 .6 11.72.66 1.280.97 0.76

98.1 97.668.7 61.8

47.8 41.91.29 3.586.20 12.3

14.2 12.00.36 0.00

13.4 ',t3.711.7 1 1 .61 .18 2 .880.71 0.89

96.8 98.962.7 67.1

41.3 47.13 . 1 5 1 . 1 1

12.2 6.7512.3 14.60.50 0.50

13.3 13.011 .2 11 .62.69 1.250.87 0.67

97.5 96.565.9 61.4

46.4 41.6 43.11.21 3.77 3.286.88 11.8 10.9

14.2 12.2 12.70.49 0.09 0.06

13.1 13.4 13.411 .5 11 .4 11 .61.21 2.42 2.440.76 0.54 0.65

95.7 97.2 98.162.1 66.3 65.3

Note: Pat: pargasite, Hb: edenitic-homblende to hornbtende, mg: 100 x Mg/(Mg + >Fe).. Total Fe as FeO.

NAKADA: TITANITE-BEARING DACITES

TABLE 5. Chemical analyses of biotite phenocrysts in titanite-bearing dacites

555

Salama Porquesa

706 708 M62

sio,Tio,Al2o3FeO-MnOMgoNa"OK"O

Totalmg

37.24.43

13.017.10 .15

14.31.069.05

96.2s9.8

37.44.86

13.417.0o.02

14.10.839.22

96.859.7

36.14.63

13.317.4o.24

13.70.439.16

94.958.3

36.15 .11

13.717.20 .14

13.50.609 .15

95.558.3

36.94.30

13.116.20.09

13.51 .018.85

93.959.7

35.93.99

13.018.80.28'12.4

0.989.15

94.453.9

35.74.09

12.818.30.24

12.20.989.08

93.354.3

" Total Fe as FeO.

coexisting with titanite in hornblende phenocrysts aresimilar in composition to those in titanite (Table 8), al-though they are diopside normative. Postentrapmentcrystallization of the host crystal (titanite) on the inclu-sion wall causes a decrease in the TiO, and CaO contentsof the melt inclusions, so that normative corundum con-tents increase in compensation for the decrease in nor-mative anorthite content.

Ilmenite grains in titanite show subhedral to irregular,rounded forms. The ilmenite is always exsolved, suchthat fine lamellae of a Ti-rich phase (probably rutile) oc-cur in the host ilmenite (Fig. 2D). The RrO, content ofgrains of ilmenite inclusions is higher than that ofgroundmass ilmenite (Tables 7 and 9). Discrete, small,rounded grains of homogeneous hematite are rarely in-cluded in titanite and do not touch the ilmenite inclusions(Fig. 2D).

DrscussroxComposition and stability of titanite in silicate melts

The temperatwe-forrelations for the reaction

titanite + magnetite + quartz: ilmenite * calcic silicate + O (l)

were reported in Wones (1989). He concluded that higher

/6, is implied by association of the assemblage titanite +magnetite + quaftz with Ca-rich pyroxene (or calcic am-phibole) of higher Mg/(Mg * Fe,o,), and that somewhatlower fo. is implied by the same quartz-free assemblage.The absence of magnetite also reduces the f6r limit ontitanite stability. The latter two conditions were em-ployed in the experiments of Helz (1973) and Green andPearson (1986a, 1986b, 1987), allowing titanite to crys-tallize from quartz- and magnetite-undersaturated meltsat relatively low for. Nevertheless, relatively oxidizedconditions may be more typical in natural silica-richmagmas, which usually contain quartz and Ti-rich mag-netite phenocrysts.

High-pressure experiments (Yodor and Tilley, 1962;Helz, 1973; Hellman and Green, 1979) indicate that ti-tanite crystallizes from hydrous partial melts of Ti-richmafic rock only at temperatures near the wet solidus. Thisalso implies instability of titanite in melts at atmosphericpressure. Experiments of Green and Pearson (1986a)showed that the TiO, content of melts saturated with Ti-rich accessory phases decreases with decreasing temper-ature, increasing pressure, and increasing SiO, content inthe melt. Thus, Ti-rich ac@ssory phases may become sta-ble in cooled SiOr-rich melts at crustal pressures, even ifthe melt is not enriched in TiOr. In a magma chamber,residual melt will become saturated with HrO on cooling

TABLE 6. Chemical analyses of olivine and clinopyroxene phenocrysts in titanite-bearing dacites from Sajama Volcano

801Olivine

801Diopside

708Fe-rich diopside

sio,Tio,Al2o3FeO"MnOMgoCaONaro

Totalmg

38.8

o.2717.450.10

43.20.09

99.981.5

38.1

0.0418.290.17

41.80 .11

51.40.072.995.89

16.320.60.54

97.883.1

51.40.613.336.180.27

15.921.30.41

99.482.1

52.40.270.848.59o.44

13.122.90.94

99.573.1

98.580.3

52.60.210.628.570.53

13.023.00.79

99.373.1

'Total Fe as FeO.

556 NAKADA: TITANITE-BEARING DACITES

TABLE 7. Chemical analyses of Fe-Ti oxide minerals in titanite-bearing dacites

Sajama Parina Porquesa801 708

mt ilm3 3

mt ilm3 3

mt3

mt ilm8 6

mt3

mt ilmN 4 3

sio,Tio,Al2o3FerO"FeOMnOMgoCaO

TotalMn/MgX,"oXt^rfc)Alog fo,

0.36o.co1.42

53.533.9

0.281.940.06

98.00.0820.187

0.4039.30.06

27.131.60.651.980.03

1 0 1 . 10 .19

0.715.051 .01

55.434.40.650.670 .12

98.00.540.150

2.054.702.16

52.334.00.851.580.58

98.20.300.145

0.9342.50.26

18.738.s1.O72.54o.17

99.7o.24

0.8100.779815

2.4

0.69 0.5739.8 5.100.13 1.35

21.1 55.631.3 33.70.37 0.512.68 0.940.11 0.05

96.2 97.80.078 0.31

0.149

0.615.68't.32

54.834.00.491.450 . 1 1

98.50.200.164

0.732810

2.7

0.51 0.,134.63 38.91 .01 0 .12

56.6 25.333.5 31.80.52 0.900.73 1.470.32 0.09

97.8 99.00.41 0.340.136

o.746768

2 .6780

2.9

Notej m! : Ti-rich magnetite, ilm : ilmenite, N: number of area analyses. The Alog fo, = fo, relative to FMe. FerO", FeO, ulvdspinel (X*e), andilmenite (Xil contents were calculated according to Stormer (1983). The Alog fo. and tempeiaturswere calculated according to Andersen and Lindsley(1 e88).

toward the wet solidus. If f" is suitable for titanite sta-bility, it easily crystallizes from these residual melts. Thus,titanite phenocrysts may form in low temperature silica-rich magmas that are hydrous and relatively oxidized.

The Fe/Al ratios ofthe Andean titanite phenocrysts areapproximately I (Fig. 5, Table 3), as is typical for vol-canic titanites (e.g., Cundai, 1979; Giannetti and Luhr,1983; Luhr et al., 1984; Wdrner and Schmincke, 1984).Fe/Al ratios in titanite from granitoids range from 0.5 to1.0 (Fig. 5). Titanite precipitated from melts in high tem-perature-pressure experiments (Hellman and Green. 1979:

Tmu 8. Chemical analyses of glass inclusions in titanite andh"r.bl"rd" ph""".

Sajama Parinacota Porquesa

POR{14

Green and Pearson, 1986b) has Fe/Al ratios less than 0.5.Finally, FelAl ratios in metamorphic titanite are muchIess than 0.5 (Fig. 5).

The incorporation of Al and Fe in titanite can be ex-plained by the following substitutions (e.g., Ribbe, 1980),

(Al, Fe)3+ + REE3+ : Ca2+ + Ti4+(Al, Fe)3* : Ti4+ + (OH, F, CD

Fel+ * REE3':Ca3tr +TiX+

Equation 2 cannot explain the chemical variations ob-served (Fig. 4) because Ca content has no correlation withFe or Al contents and )REE content shows no correlationwith Ti content. The observed variations are likely con-trolled by a combination of Equations 3 and 4.

Titanite from granitoids, which is typically less en-riched in REE than the volcanic titanite studied, has Alcontents as high as volcanic titanite, but lower Fe con-

TABLE 9. Chemical analyses of Fe-Ti oxide minerals included intitanite phenocrysts

Saiama Parinacota Porque

M62

(2)(3)(4)

M621

706 708N 1 2

M62(hb)2

sio,Tio,Al2o3FeO.MnOMgoCaONaroKrO

TotalCIPW normoorabancdihyilmmt

75.6 76.30.15 0.31

12.6 12.40.59 0.750.06 0.120.24 0.130.75 0.522.96 3.655.07 5.26

98.0 99.4

73.80.32

12.30.84

0.250.203.455.47

96.6

33.133.430.21 . 00.3

75.50.34

12.40.3s0.04o.240.443.22s.24

97.7

36.731.627.81 . 01 . 1

37.630.625.63.80.8

33.831.33 1 . 11 . 9

75.80.22

12.40.610.150.150.682.824.50

97.3

39.728.225.33.61 .8

0.4434.6o .12

33.029.00.570.98o.27

99.00.320.669

i lm6

ilm hm7 2

ilm ilmN 2 2

0.60.40.4

0.60.30.60.5

i, os o,0.6 0.7 0.30.6 0.3 0.4

sio,Tio,Al2o3FerO.FeOMnOMgoCaO

TotalMn/Mgx,,^

0.31 0.4433.7 38.00.21 0.13

33.0 37.327.2 27.20.61 0.451 .19 1 .270.55 0.24

96.8 100.0

0.44 0.2711.1 37 .201 .6 0.13

78.0 28.47.26 31 .00.69 1.011.01 0.840.60 0.19

99.3 99.00.38 0.690.195 0.716

Notei (hb): glass together with titanite included in hornbtende. Fe,O.was assumed as 0.4 x >FeO., for ClpW norm.- Total Fe as FeO.

0.300.660

o.210.629

Note.'hm : hematite.

NAI(ADA: TITANITE-BEARING DACITES 557

tents (Fig. 4). This fact indicates that REEs were incor-porated into these titanite samples according to Equation4 rather than Equation 2. Hellman and Green (1979) andFranz and Spear (1985) showed that Al rather than Fe3+preferentially enters into the titanite structure at higherpressures. Green and Pearson (1986b) showed experi-mentally that substitution according to Equation 2 oc-curred in titanite crystallized from silicate melts at pres-sures greater than 7 kbar. These observations suggest thatREEs are coupled with Fe2+ at relatively low pressure andwith Al at high pressure.

Magrna mixing

The coexistence ofpargasite and hornblende in all da-cites studied and ofdiopside and Fe-rich diopside in theSajama voleanic rocks suggests mixing of two magmas.The time interval between mixing and eruption was short,as these phenocrysts do not show notable compositionalzoning. This is also supported by the coexistence of Mg-rich olivine and quartz without reaction rims. The twoequilibrium phenocryst assemblages before mixing can beestimated from Mg-Fe partitioning among mafic pheno-crysts, phenocryst aggregation textures, and inclusion re-lationship: (l) hornblende + biotite + titanite + Ti-richmagnetite + Fe-rich diopside + sanidine + q\artz +sodic, dusty plagioclase, and (2) pargasite a diopside +olivine + plagioclase. These assemblages represent rhyol-itic and mafic magma, respectively. Probably, the rims ofphenocrysts grew slightly in the mixed magma or weremodified chemically. The outermost rims on dusty pla-gioclase phenocrysts may have grown in the mixed mag-ma after rapid growth on the original crystal margin rath-er than after dissolution ofit.

Experimental results on dissolution of plagioclase at Iatm (Tsuchiyama, 1985) showed that the rate of partialdissolution depends on magma temperature and is ex-tremely slow at low temperature; 100-1000 yr are nec-essary for a dusty zone of 100 pm width to form even at1000'C in a dry melt. If this is so, a more complicatedscenario of magma mixing must be considered for thedacites. However, fu2o ma! be of great importance. Themorphologies of dusty plagioclase closely resemble thereaction rims on plagioclase that were obtained experi-mentally by Johannes (1989). His experiments on meltingof plagioclase-qwrtz assemblages at2-kbar HrO pressureshowed that a wide reaction rim consisting of glass andAn-rich plagioclase developed around unchanged plagio-clase at low temperatures during a short time; a reactionrim of approximately 50 pm width formed at 875 "C in4 d .

T, P, and f- conditions

Equilibration temperatures of the mafic magma at Sa-jama may be estimated at roughly 1000 'C, based onexperimental studies of the partitioning of Fe-Mg be-tween diopside and olivine (e.g., Grove et al., 1982). Ex-

perimental results on amphibole stability (Gilbert et al.,1982) demonstrate that the upper stability limit of par-gasite in basaltic to andesitic melts is approximately 950"C at a few kilobars and near 1050 "C at higher pressures.Coexistence of plagioclase with pargasite in a mafic mag-ma may indicate that the magma was hydrous and atrelatively low pressure, since the stability field ofplagio-clase is drastically reduced with increasing pressure (e.g.,Green, 1982). Therefore, the equilibration temperatureof the mafic magma was unlikely to have been higherthan 950'C.

A core of dusty plagioclase and sanidine in a Sajamadacite gives a temperature of approximately 750'C withthe two-feldspar geothermometer (Fuhrman and Linds-ley, 1988; Fig. 7). This value is consistent with the esti-mated temperatures of 700-740 "C obtained from a dustyplagioclase core and hornblende with the geothermome-ter of Blundy and Holland (1990). Coexistence of he-matite and ilmenite in the titanite phenocrysts gives in-formation on temperature arl.d forof the rhyolitic magma(Spencer and Lindsley, l98l; Burton, 1984). The /", isapproximately I log unit below the HM buffer, and thetemperature is as low as 700 "C (Fig. 8). The rhyoliticmagma is thought to have had stable hornblende, biotite,qtJartz, plagioclase, sanidine, titanite, and Ti-rich mag-netite, all of which are required to use the geobarometerbased on the Al content of hornblende (Johnson andRutherford, 1989). The estimated pressure ranges from1.2 to 3.3 kbar.

Magrna chamber model

That all of the dacites studied show evidence of mixingbetween rhyolitic and mafic magmas implies that rhyolit-ic and mafic magmas commonly were present simulta-neously beneath these volcanoes. A possible magmachamber model for the central Andean volcanoes is sim-ilar to that proposed for large-scale pyroclastic flow de-posits with compositional gradients (e.g., Hildreth, 198 l)or a magma chamber that is replenished by denser, moremafic magma (e.g., Huppert et al., 1982); in either casemagma was stratified compositionally within the cham-ber, with mafic magma overlain by rhyolitic magma. Insuch a magma chamber, mixing between the two magmascould occur in the chamber just before eruption (e.g.,Sparks et al., 1977) or inthe conduit during eruption (e.g.,Koyaguchi, 1985).

Temperatures deduced from the compositions of Ti-rich magnetite microphenocrysts and groundmass ilmen-ite in the Sajama and Porquesa volcanic rocks are 770-810 lC at an fo2 of approximately 2log units above theNi-NiO buffer (Table 7, Fig. 8). These temperatures arehigher than those estimated for the rhyolitic magma andprobably lower than that of the mafic magma. Becausethe Fe-Ti oxide minerals are the most sensitive to physio-chemical change (Bacon and Hirschmann, 1988), it ispossible that the Tirich magnetite microphenocrystschanged composition following magma mixing. Thus,

558 NAKADA: TITANITE-BEARING DACITES

these temperatures and ,6, values may represent those ofthe mixed magmas.

It would be expected that heating of the rhyolitic mag-ma during magma mixing would cause rhyolitic melttrapped in titanite phenocrysts to be oversaturated in HrO,because HrO solubility in silicate melts decreases withincreasing temperature (e.g., Burnham and Jahns, 1962).As a result, the melt could have vesiculated and expandedduring mixing and eruption so that the titanite pheno-crysts were broken. Alternatively, depressurization dur-ing eruption might have been adequate to vesiculate andexpand the melt inclusions.

Evidence for reduction and oxidationBecause the time interval between mixing and eruption

was short in the Andean dacites, textures within the ti-tanite phenocrysts should record physiochemical changesin the rhyolitic magma before mixing. Resorption of thetitanite phenocrysts implies instability during crystalli-zation. Titanite becomes unstable in silicate melts by thefollowing mechanisms: (l) decreasing fo,; (2) increasingtemperature; and (3) decreasing SiO, or TiO, contents inthe melt (Green and Pearson, 1986a). These conditionscould have been brought about by a previous magmamixing event. However, phenocrysts coexisting with thetitanite phenocrysts in these dacites show little evidenceof drastic changes in the composition of coexisting meltsduring their crystallization before mixing.

The occurrence of resorbed crystals of titanite sur-rounded by ilmenite coronas in the studied dacites (Fig.3D) implies reaction between titanite and ilmenite beforeor during eruption, as suggested by Czamanske (1988).Because hornblende, Ti-rich magnetite, and quartz werepresent in the rhyolitic magma from which titanite crys-tallized, a reduction reaction such as Equation I wasprobably responsible for resorption oftitanite and growthof ilmenite. The ilmenite inclusion zones in titanite phe-nocrysts (Fig. 2) record a similar event, although ilmeniteinclusions are more enriched in RrO, than the coronailmenite. Overgrowh oftitanite on the ilmenite inclusionzones indicates a return to a more oxidizing condition(Eq. l). Czamanske (personal communication, 1989) hassimilarly interpreted identical textures found in titanitein granodiorites of the high-level granitoids at Questa,New Mexico. Ilmenite was partly resorbed and enclosedalong with melt within newly precipitated titanite. TheFeTiO. component was preferentially extracted from theilmenite inclusions once oxidation resumed, so that theirbulk compositions shifted toward FerOr, leading to pre-cipitation of hematite as they reached the ilmenite-he-matite solvus.

Mafic silicates may also reflect oxidation and reductionin the magma. Hornblende and biotite were abundantmafic silicates in the silicic magmas, and these mineralsshow little chemical zoning in these Andean dacites. Bothminerals contain Fe3+ as well as Fe2+, so that Mgl(Mg +Fe,.,) may not have changed measurably during the ex-cursion in /", that was critical to titanite stability.

ry-it ic mogmo"Imss-Hemss

% PorquesoSojomo

a a

NNO

700 900T ( 'C)

Fig. 8. Oxygen fugacity relative to FMQ (Alog f,) plottedagainst temperature for Andean dacites. Oxygen fugacities andtemperatures were estimated from Ti-rich magnetite microphe-nocrysts and groundmass ilmenite according to Andersen andLindsley (1988) using the Stormer (1983) recalculation ofanal-yses. Dots for titanite-bearing volcanic rocks from the literatureof Lipman (1971) and Whitney and Stormer (1985), whose datawere also recalculated by the above methods. The field labeledIlm*-Hem"" represents the coexisting ilmenite and hematite solidsolutions (Spencer and Lindsley, l98l).

Reduction and oxidation in the magma chamber

The forvalues deduced for the mixed magmas are ap-proximately 2 log units above the Ni-NiO buffer, andthose for the rhyolitic magmas are roughly I log unithigher (Fig. 8), implying differences in /o, between therhyolitic and mafic magmas. Czamanske and Wones(1973) reported increasing "fo, with crystallization in theFinnmarka Intrusive Complex. The fr-temperature re-lations for titanite-bearing dacites-rhyolites from the SanJuan Mountains, Colorado (Lipman, l97l; Whitney andStormer, 1985) show a similar trend; relative to buffercurves, 6, increased with cooling (Fig. 8). Calculation ofchemical mass transfer based on thermodynamic rela-tions (Ghiorso and Carmichael, 1985) showed that fo,ofresidual liquids undergoes a relative increase during crys-tallization in a system closed to O. Oxidation in therhyolitic nugma may be caused by the loss of H, gasduring devolatilization of magma (Czamanske and Wones,1973; Burnham, 1979; Czamanske, 1988) or by contam-ination with ground water and pyrite-bearing countryrocks (Gerlach and Nordlie, 1975; Whitney and Stormer,l 983) .

As SO, is the principal S gas species in highly oxidizedmagmas, escape of SOr-rich gas from magma nray causereduction of the residual magmas (Gerlach and Nordlie,1975; Whitney, 1984; Carmichael and Ghiorso, 1986).Loss of S-rich gas from hydrous magma can suppress the

a=4g

(\lo

cn2o

OL500

NAKADA: TITANITE-BEARING DACITES 559

oxidizing trend due to crystallization, devolatilization, orcontamination; as a result, titanite may be resorbed andan ilmenite corona may form. Because S contents in high-ly oxidized silicic melts are very low (e.g., Carmichaeland Ghiorso, 1986), S in the magma is consumed byescape of SO, gas in the early stage of degassing (Gerlachand Nordlie, 1975; Burnham, 1979). The successive de-gassing will cause the magma to oxidize rapidly due toH, gas escape. The rapid growth of titanite in inclusionzones may be promoted by such oxidation.

Suprvr.lv

Titanite-bearing dacites from the central volcanic zoneof the Andes commonly show evidence of magma mix-ing. Hornblende and pargasite phenocrysts coexist in allthe studied dacites as a result of mixing between pargas-ite-bearing mafic magma and hornblende-bearing rhyo-litic magma. Two populations of clear (less sodic) and dusty(more sodic) plagioclase phenocrysts are present. In somesamples, unreacted olivine and quartz phenocrysts co-exist. The time intervals between mixing and eruptionare thought to have been short, such that reaction rimswere slightly formed on the amphibole and plagioclase.

The rhyolite magmas were relatively cool, hydrous, andhighly oxidized at shallow levels in the crust, whereas themafic magmas were hotter and less oxidized. In the mag-ma chambers, the rhyolitic magmas may have been un-derlain by the mafic magmas. Titanite crystallized fromthe rhyolitic magma. Reduction events in the magmachambers were recorded in titanite phenocrysts by Fe-Tioxide and melt inclusion zones and irregular zoning pat-terns. Overgrowth of titanite on the inclusion zones re-cords oxidation that occurred after the reduction events.The changes from reduction to oxidation may have re-sulted from escape of S-rich gas from the rhyolite mag-mas in the early stages of devolatilization and of S-defi-cient gas in the later stages. Finally, titanite phenocrystswere fragmented because of expansion of melt inclusionscaused by heating during mixing or unloading duringeruption.

AcxNowr,nncMENTS

Valuable comments and discussions for this study by C.R. Bacon, G.K. Czamanske, T. Yanagi, and T. Nishiyama are appreciated. Criticalreviews by G.K. Czamanske, C.R. Bacon, J.F. Luhr, J.A. Whitney, andP.A. Candela are greatly appreciated. I wish to thank I. Shinno and K.Ishida for providing the facility for the EDS analyses; H. Honma and Y.Okamoto for Sr isotopic analyses; E. Soria for his assistance in collectingdacite samples from Sajama; Y. Katsui for the Parinacota samples; M.Takahashi, the late N- Onuma, K. Notsu, A. Lahsen, and H. Moreno-Lfor the Porquesa samples and their geologic implications; and T. Nishi-yama for his permission to use unpublished dala on metamorphic titan-ites. I thank also N. Shimada and A. Sugaki for giving me a chance tostay in h Paz and JICA for financial support.

RprnnBNcrs CITED

Andersen, D.J., and Lindsley, D.H. (19E8) Intemally consistent solutionmodels for Fe-Mg-Mn-Ti oxides: Part I. Fe-Ti oxides. American Min-eralogist, 73,7l+726.

Aramaki, S., Onuma, N., and Portillo, F. (1984) Petrography and majorelement chemislry of the volcanic rocks of the Andes, southern Peru.Geochemical Journal, 18, 2 l7 -232.

A1'uso, R.A. (198a) Field relations, crystallization, and petrology of re-versely zoned gmnitic plutons in the Botfle Iake Complex, Maine. U.S.Geological Survey Professional Paper 1320, 58 p.

Bacon, C.R. (1989) Crystallization ofaccessory phases in magmas by local

saturation adjacent to phenocrysts. Geochimica et Cosmochimica Acta,

53, 1055-1066.Bacon, C-R., and Hirschmann, M.M. (1988) Mg/Mn partitioning as a test

for equilibrium between coexisting Fe-Ti oxides. American Mineralo-g1st,73, 57-61.

Blundy, J.D., and Holland, J.B. (1990) Calcic amphibole equilibria and

a new amphibole-plagioclase geothermometer. Contributions to Min-

eralogy and Petrology, 104, 208-224.Burnharn, C.W. (1979) Magmas and hydrothermal fluids. In H.L. Barnes,

Ed., Geochemistry ofhydrothermal ore deposits (2nd edition), p. 7l-

136. Wiley, New York.Burnham, C.W., and Jahns, R.H. (1962) A, method for determining the

solubility of watfi in silicate melts. American Joumal of Science, 260,721-745.

Burton, B.P. (1984) Thermodynamic analysis of the system Fe'Or-Fe-TiOr. Physics and Chemistry of Minerals, ll,132-139.

Carmichael. I.S.E., and Ghiorso, M.S. (1986) Oxidation-reduction rela-

tions in basic magma: A case for homogeneous equilibria. Earth andPlanetary Science Letters, 7 8, 2OO-210.

Carmichael, LS.E., and Nicholls, 1.0967) Iron-titanium oxides and ox-ygen fugacities in volcanic rocks. Joumal ofGeophysical Reseatch,72,4665-4687.

Cundari, A. (1979) Petrogenesis of leucite-bearing lavas in the Roman

volcanic region, Italy. The Sabatini lavas. Contributions to Mineralogy

and Petrology, 70, 9-21.Czarnanske, G.K. (1988) Interplay between REE-bearing accessory min-

erals and silicic magma, Questa, New Mexico, U.S'A. Fifth interna-tional symposium on tinltungsten granites in southeast Asia and west-

ern Pacific (Matsue). IG'CP 220 Abstracts with Program, 25-29.Czamanske, G.K., and Wones, D.R. (1973) Oxidation during magmatic

differentiation, Finnmarka Complex, Oslo area, Norway: Pan 2, The

mafic silicates. Joumal ofPetrology, 14, 349-380.Deer, W.A., Howie, R.A., and Zussman, J. (1982) Sphene. In Rock-form-

ing minerals, vol 1A: Orthosilicates (2nd edition), p. 443-466. Long'man, London.

Donaldson, C.H. (1976) An experimental investigation of olivine mor-phology. Contributions to Mineralogy and Petrology, 57, 187-213.

Ernst, W.G., andDalPaiz, G.V. (1978) Mineral parageneses of eclogiticrocks and related mafic schists ofthe Piemonte ophiolite nappe, Breuil-St. Jacques area, Italian western Alps. American Mineralogist, 63, 621-640.

Ewart, A. (1979) A review of the mineralogy and chemisry of Tertiary-Recent dacitic, latiric, rhyolitic and related salic volcanic rocks. In F.Barker, Ed., Trondhjemites, dacites, and related rocks, p. I l-121. El-

sevier, Amsterdam.Franz, G., and Spear, F.S. (1985) Aluminous titanite (sphene) from the

eclogite zone, south-central Tauem Window, Austria. Chemical Geol-

ogy, 50, 33-46.Fuhrman, M.L., and Lindsley, D.H. (1988) Ternary-feldspar modelling

and thermometry. American Mineralogist, 7 3, 201-21 5.Gerlach, T.M., and Nordlie, B.E. (1975) The C-O-H-S gaseous system,

part III: Magmatic gases compatible with oxides and sulfides in basalticmagmas. American Journal of Science, 27 5, 39 5 -4 l0 -

Ghent, E.D., and Stout, M.Z. (1984) TiO, activity in metamorphosedpelitic and basic rocks: Principles and applications to metamorphismin southeastem Canadian Cordillera. Contributions to Mineralogy andPetrology, 86, 248-25 5.

Ghiorso. M.S., and Carmichael, I.S.E. (1985) Chemical mass-transfer in

magmatic processes, II. Applications in equilibrium crystallization,fractionation and assirnilation. Contributions to Mineralogy and Pe-

trology,90, l2l-141.Giannetti, 8., and Luhr, J.F. (1983) The White trachytic tuff of Rocca-

monfina Volcano (Roman Region, Italy). Contributions to Mineralogyand Petrology, 84, 235-252.

Gilbert, M.C., Helz, R.T., Popp, R.K., and Spear, F.S. (1982) Experi-

mental studies of amphibole stability. In Mineralogical Society of

America Reviews in Mineralogy, 98,229-353.Green, T.H. (1982) Anatexis of mafic crust and high pressure crystalli-

560 NAKADA: TITANITE-BEARING DACITES

zation ofandesite. In R.S. Thorpe, Ed., Andesites: Orogenic andesitesand related rocks, p. 465-487 , Wiley, New York.

Green, T.H., and Pearson, N.J. (1986a) Ti-rich accessory phase saturationin hydrous mafic-felsic compositions at high P,T. Chemical Geology,54 ,185 -201 .

- (1986b) Rare-earth element partitioning between sphene and co-existing silicate liquid at high pressure and temp€rature. Chernical Ge-o logy ,55 ,105 -119 .

- (1987) An experimental srudy ofNb and Ta partitioning betweenTi-rich minerals and silicate liquids at high pressure and temperature.Geochimica et Cosmochimica Acla. 5 l. 55-62.

Gromet, L.P., and Silver, L.T. (1983) Rare earth element distributionsamong minerals in a granodiorite and their petrogenetic implications.Geochimica et Cosmochimica Act?', 47,925-939.

Grove, T.L., Gerlach, D.C., and Sando, T.W. (1982) Origin of calc-al-kaline series lavas at Medicine Iake Volcano by fractionation, assim-ilation and rnixing. Contributions to Mineralogy and petrology, 80,160- l 82.

Hellman, P.L., and Green, T.H. (1979) The role of sphene as an accessoryphase in the high-pressure partial melting of hydrous mafic composi-tions. Earth and Planetary Science ktters, 42, l9l-201.

Helz, R.T. (1973) Phase relations of basalts in their melting range atPn,o : 5 kb as a function ofoxygen fugacity. Journal ofPetrology, 14,249-302.

Hildreth, W. (1981) Gradients in silicic magma chambers: knplicationsfor lithospheric magmatism. Joumal of Geophysical Research, 86,l0 I 53-101 92.

Huppert, H.8., Turner, J.S., and Sparks, R.S.J. (1982) Replenished mag-ma chambers: Efects ofcompositional zonation and input rates. Earthand Planetary Science Irtters, 57 , 345-357.

James, D.E. (1971) Andean crustal and upp€r mantle structure. Joumalof Geophysical Research, 7 6 , 3246-327 | .

Johannes, W. (1989) Melting of plagioclase-quartz assemblages at 2 kbarwater pressure. Contributions to Mineralogy and Petrology, 103,27O-276.

Johnson, M.C., and Rutherford, M.J. (1989) Experimenral calibration ofthe aluminurn-in-homblende geobarometer with application to InngValley caldera (California) volcanic rocks. Geology, 17, 837-841.

Kagami, H., Okano, O., Sudo, H., and Honma, H. (1982) Isotopic anal-ysis of Rb and Sr using a full automatic thermal ionization mass spec-trometer. Papers of Institute for Thermal Spring Research (OkayamaUniversity), 52, 5l-70 (in Japanese).

Katsui, Y., and Gonzalez, F.O. (1968) Geologia del area neovolcanica delos Nevados de Payachata. Pablicacion, Facultad de Ciencias Fisicas yMatematicas, Depanamento de Geologia (Universidad de Chile) 29,1 -6 l .

Koyaguchi, T. (1985) Magma mixing in a conduit. Journal of Volcanologyand Geothermal Research, 25, 365-369.

Lipman, P.W. (1971) Iron-titanium oxide phenocrysts in compositionallyzoned ash-flow sheets from southern Nevada. Journal of Geology, 79,4t8-456.

Luhr, J.F., Carmichael, I.S.E., and Varekamp, J.C. (1984) The 1982 erup-tions of El Chichon Volcano, Chiapas, Mexico: Mineralogy and pe-trology of the anhydrite-bearing pumice. Journal of Volcanology andGeothermal Research, 23, 69- lO8.

Nakada, S. (1987) X-ray fluorescence determination oftrace elements insilicate rocks (part 2). Science Repons ofDepartment ofGeology (K1u-shu University), 15, 37-44 (in Japanese).

Nakada, S., Yanagi, T., Maeda, S., Fang, D., and Yamaguchi, M. (1985)X-ray fluorescence analysis ofmajor elements in silicate rocks. ScienceReports of Department of Geology (Kyushu University), 14, 103-l l5(in Japanese).

Notsu,IC, Takahashi, M., Onuma, N., Ui, T., Moreno, H.R., and l:hsen,A. (1985) Strontium isotope compositions of late Cenozoic volcanicrocks from northern Chile between latitudes l 7"S and 20"S. In M. Ta-giri, Ed., Geochernical investigation of the southem Andes volcanicbelt, 1982-1984. Overseas Scientific Research (No. 59043009) p. 195-216, Ibaraki University-Universidad de Chile.

O'Callaghan, L.J., and Francis, P.W. (1986) Volcanological and petrolog-ical evolution ofSan Pedro Volcano, Provincia El Loa. North Chile.Journal of Geological Society, London, | 43, 27 5-286.

Onuma, N., and Montoya, M. (1984) Sr/Ca-Ba,rCa systematics of volca-nic rocks from the central Andes, southern Peru, and its implicationfor Andean magmatism. Geochemical Joumal, 18, 251-262.

Paterson, B.A., Stephens, W.E., and Herd, D.A. (1989) Zoning in gran-itoid accessory minerals as revealed by backscattered electron imagery.Mineralogical Mzgazine, 53, 56-6 l.

Ribbe, P.H. (1980) Titanite. In Mineralogical Society of America Reviewsin Mineralogy, 5, 137-154.

Sawka, W.N., and Chappell, B.W. (1988) Fractionation of uranium, tho-rium, and rare earth elements in a venically zoned granodiorite: Im-plications for heat production distributions in the Sierra Nevada batho-lith, California, U.S.A. Geochimica et Cosmochimca Acta, 52, ll3l-t143.

Shinno, I., and Ishida, K. (1983) Quantitative electron microprobe anal-ysis of silicates and sulfides using energy-dispersive X-ray spectrome-try. Reports on Earth Science, College of General Education (KyushuUniversity), 23, 25-33 (in Japanese).

Sparks, R.S.J., Sigurdsson, H., and Wilson, L. (1977) Magma mixing: Amechanism oftriggering acid explosive eruptions. Nature, 267, 315-3 1 8 .

Spear, F.S. (1981) An experimental study of homblende stability andcornpositional variability in amphibolite. American Journal of Science,281,697-734.

Spencer, K.I., and Lindsley, D.H. (1981) A solution model for coexistingiron-titanium oxides. American Mineralogist, 66, I 1 89- 1 20 1.

Stormer, J.C.J. (1983) The effects of recalculation on estimates of tem-perature and oxygen fugacity from analyses of multicomponent iron-titanium oxides. American Mineralogist, 68, 586-594.

Thorpe, R.S., and Francis, P.W. (1979) Variations in Andean andesitecompositions and their petrogenetic significance. Tectonophysics, 57,53-70.

Thorpe, R.S., Francis, P.W., Hammill, M., and Baker, M.C.W. (1982)The Andes. In R.S. Thorpe, Ed., Andesites: Orogenic andesites andrelated rocks, p. 187-205. Wiley, New York.

Thorpe, R.S., Francis, P.W., and O'Callaghan, L.J. (1984) Relative rolesof source composition, fractional crystallization and crustal contami-nation in the petrogenesis of Andean volcanic rocks. PhilosophicalTransactions of Royal Society, London, A310, 67 5-692.

Tsuchiyama, A. (1985) Dissolution kinetics of plagioclase in the melt ofthe system diopside-anorthite-albite, and origin of dusty plagjoclase inandesite. Contributions to Mineralogy and Petrology, 89, l-16.

Tsusue, A., Mizuta, T., and Hashimoto, K. (1984) Granitic rocks innorthern Kyushu. Mining Geology (I apan), 3 4, 3 8 5-399.

Verhoogen, J. (1962) Distribution oftitanium between silicates and ox-ides in igneous rocks. American Joumal ofScience,260,2ll-22o.

Whitney, J.A. (1984) Fugacities ofsulfurous gases in pyrrhotite-bearingsilicic magmas. American Mineralogist, 69, 69-7L

Whitney, J.A., and Stormer, J.C.J. (1983) Igneous sulfides in the FishCanyon Tuff and the role of sulfur in calc-alkaline magmas. Geology,tt, 99-102.

- (1985) Mineralogy, petrology, and magmatic conditions from theFish Canyon Tuf, central San Juan volcanic freld, Colorado. Journalof Petrology, 26, 7 26-7 62.

Wones, D.R. (1989) Significance of the assemblage titanite + magnetite+ quartz in granitic rocks. American Mineralogist, 74,744-749.

Wijmer, G., and Schmincke, H.U. (1984) Mineralogy and chemical zo-nation ofthe Iaacher See tephra sequence (East Eifel, Germany). Jour-nal ofPetrology, 25, 805-835.

Wdrner, G., Harmon, R.S., Davidson, J., Moorbath, S., Tumer, D.L.,McMillan, N., Nye, C., Lopez-Escobar, L., and Moreno, H. (1988) TheNevados de Payachata volcanic region (18:S/69"W, N., Chile). I. Geo-logical, geochemical, and isotopic observations. Bulletin of Volcanol-ogy,50,287-303.

Yoder, H.S.J., and Tilley, C.E. (1962) Origin of basaltic magmas: Anexperimental study of natural and synthetic rock systems. Journal ofPetrology, 3, 342-532.

MeNuscnrrr REcETvED Apm l7. 1990MeNuscnrgr AccEprED Jexueny 14, l99l