Myron G. Best. Igneous and metamorphic 2003 Blackwell Science

Подождите немного. Документ загружается.

Crystal-Melt Equilibria in Magmatic Systems

107

with the curved liquidus surface. A two-dimensional

“topographic map” of the three-dimensional conﬁgu-

ration of the liquidus surface can be created by pro-

jecting the contour lines parallel to the T axis of the

prism onto the triangular compositional base (Figure

5.19c).

A perspective view of the solidus and solvus surfaces

in the ternary is shown in Figure 5.20. The intersection

of these two surfaces is the arcuate line UQ

V. This

line is the locus of points representing the composi-

tions of ternary feldspar solid solutions, such as the

stable pair Q

and F

, at a particular set of intensive

variables. Isothermal tie lines that connect these com-

position points are projected onto the ternary “map” in

Figure 5.21.

Consider the crystallization behavior of an initially

all-liquid system such as Z in Figure 5.19b. Upon cool-

ing down the isopleth (dotted line parallel to the T

axis), the system reaches the liquidus surface at Z

where plagioclase Q

begins to precipitate (Figure

5.21). As the system cools further, the melt follows a

curved path Z

down the liquidus surface, precipi-

tating more plagioclase, which is progressively more

enriched in Na and to a much lesser extent, K. The

curved path on the liquidus results from the fact that

a ternary solid solution feldspar is being withdrawn

from the bulk system; at any particular T, the liquid

moves directly away from the equilibrium composition

of the plagioclase. Ultimately, the liquid reaches the

two-feldspar-liquid cotectic line EM

at Z

where an

alkali feldspar solid solution rich in K, such as F

coprecipitates with plagioclase Q

from the melt.

This coprecipitation must occur, regardless of whether

crystallization occurs through the equilibrium or frac-

tional process, because the amount of the KAlSi

component in the initial, bulk composition of the sys-

tem exceeds the amount that can be dissolved in any

plagioclase. In other words, there is a miscibility gap

between plagioclases, which contain only limited Kf in

solid solution, and alkali feldspars, which contain only

limited An in solid solution. As two feldspars, both in-

creasingly sodic at lower T, precipitate, the melt also

becomes more sodic, until at Z

plagioclase Q

and al-

kali feldspar F

coexist in equilibrium (Figures 5.20

KAlSi

NaAlSi

CaAl

5.20 Perspective view of the three-dimensional solvus and solidus

surfaces in the system KAlSi

-NaAlSi

-CaAl

-H

at a moderately high P and T appropriate to shallow plutonic

conditions. The T-X prism has been rotated clockwise about

120° relative to the view in Figure 5.19b, and the liquidus sur-

face in it has been omitted for clarity. Thus, these two ﬁgures

complement each other. The igloo-shaped solvus surface in-

tersects the overlying spoon-shaped solidus surface (ruled thin

lines) along the line VF

U. (It may help in visualizing this

three-dimensional diagram by thinking of a triangular blan-

ket—the solidus surface—held by three people at points O, A,

and S in a rainstorm so that the water-soaked blanket is stuck

to an underlying, ﬂat-topped boulder—the solvus.) The line of

intersection of the solvus and solidus is the locus of composi-

tions of ternary feldspar solid solutions at a particular set of in-

tensive variables shown in Figure 5.21. The curved cotectic

line EM

lies on the common solidus-solvus surfaces as well as

the liquidus surface and is projected onto the compositional

base in Figures 5.19c and 5.21. Lines Q

, F

, and Z

de-

ﬁne the sides of an isothermal, three-phase triangle (shaded)

parallel to the composition base of the prism that represents a

melt, Z

, coexisting stably with plagioclase, Q

, and K-rich al-

kali feldspar, F

NaAlSi

Wt. %

KAlSi

CaAl

5.21 Projection of the system KAlSi

-NaAlSi

-CaAl

O at moderately high P and T. Same as Figure 5.19c but in-

cludes projected compositional points and the arcuate line of

intersection of the solidus and solvus VF

U from Figure

5.20. (Redrawn from Yoder et al., 1957; Morse, 1970.)

tionating plagioclase. Curved liquid paths on variation

diagrams are characteristic of ternary and more com-

plex multicomponent systems that fractionate compo-

sitionally variable solid solutions.

5.5.5 KAlSi

(Kf )-NaAlSi

(Ab)-SiO

(silica)-

O: The Granite System

Pertinent relations in this water-saturated system at

2 kbar are projected onto the ternary diagram in Figure

5.24. A prominent valley, or thermal minimum, in the

liquidus surface near the center of the diagram is de-

ﬁned by a steep liquidus surface descending from the

silica apex and a more gently sloping liquidus surface

from the KAlSi

and NaAlSi

apices. Melts that

contain more than about 40 wt.% of the silica compo-

nent crystallize quartz as the liquidus phase. For exam-

ple, quartz precipitates from melt A at about 820°C; as

T decreases, continued crystallization of quartz drives

the residual melt down the liquidus directly away from

the silica apex, because quartz has a ﬁxed composition,

along a (dotted) straight line in the diagram. Once

the residual melt reaches the curved boundary line at

and 5.21). These three compositions (melt Z

and

feldspars Q

and F

) are signiﬁcant because under

conditions of equilibrium crystallization the tie line

connecting the two ternary feldspars passes ex-

actly through the bulk composition of the system Z

This means that, except for one drop of remaining

melt Z

, the system is crystalline and composed of

feldspars Q

and F

, in proportions given by the lever

rule. Note that the higher T assemblage of liquid Z

and feldpars F

and Q

form a three-phase, isothermal

triangle, which encloses the bulk composition point

. At that higher T, the modal proportions of the

three coexisting phases could be found by the tech-

nique described in Figure 2.3b, but adapted for a sca-

lene triangle.

A wholly liquid system Y in Figure 5.19b upon cool-

ing ﬁrst crystallizes K-rich alkali feldspar F

as it im-

pinges upon the liquidus at Y

. With continued pre-

cipitation of K-rich feldspar during cooling, the melt

moves along a curved path to Y

on the two-feldspar-

liquid boundary line EM

(Figure 5.21), where a rather

calcic plagioclase coprecipitates. Only in exceptionally

K-rich magmas does an alkali feldspar crystallize be-

fore plagioclase. In most magmas, plagioclase precipi-

tates ﬁrst and continues to do so over a broad range of

T before alkali feldspar coprecipitates, if at all.

The orientation of isothermal, two-feldspar tie lines,

such as Q

in Figure 5.21, as well as the compositions

of the coexisting feldspars represented by their end

points, depend upon P and T (Figure 5.22). If P can be

independently evaluated, then an equilibrium pair of

coexisting feldspars can serve as a geothermometer for

the T of crystallization (e.g., Fuhrman and Lindsley,

1988).

Fractional crystallization in the ternary feldspar-

melt system produces plagioclases and alkali feldspars

that have more albitic compositions at decreasing T.

For the common case of normally zoned feldspars, rims

are more albitic than cores, which are more calcic in

plagioclase and more potassic in alkali feldspars (Fig-

ure 5.22).

Fractional crystallization with decreasing T yields

melts that evolve along curved fractionation paths on

the ternary liquidus surface (Figure 5.23). Ultimately,

the most evolved melt may reach a composition near

. One set of liquid lines of descent (among an inﬁ-

nite number of possible residual melt paths) corre-

sponding to the highlighted fractionation path is

shown in the variation diagram in Figure 5.23b. Liquid

lines of descent are not straight lines but instead show

a smoothly varying curvature where changing plagio-

clase solid solutions are not the only fractionating crys-

talline phase but a sharp inﬂection (in this case at about

66.8 wt.% SiO

) where alkali feldspar solid solutions

begin to cofractionate. Residual melts commonly dis-

play a decreasing Ca/(NaK) ratio in magmas frac-

108 Igneous and Metamorphic Petrology

Rim Core KAlSi

NaAlSi

Rim

Core

810°C

1000°C

CaAl

5.22 Compositions of ternary feldspars and the orientation of tie

lines connecting equilibrium pairs of plagioclase and alkali

feldspar solid solutions depend upon intensive variables.

1000°C tie line connects coexisting equilibrium feldspar com-

positions (open squares) in a trachybasalt whose approximate

T of crystallization was determined from coexisting Fe-Ti ox-

ides. (Data from Smith and Carmichael, 1969.) Other zoned

feldspars in the trachybasalt plot in the shaded bands. 810°C

tie line connects coexisting alkali feldspar and rim plagioclase

(ﬁlled squares) in a block of rhyolite pumice in a tuff whose T

of crystallization—at higher water pressures than the trachy-

basalt—was determined from the model of Fuhrman and

Lindsley (1988). (Data from Best et al., 1995.) Zoned feldspars

in rhyolite pumice shown by ﬁlled squares. Note crossing

810°C and 1000°C tie lines. Normally zoned alkali feldspars

and plagioclases have rims that are enriched in NaAlSi

rel-

ative to their cores. Note that the miscibility gap between pla-

gioclase and alkali feldspar solid solutions is less at higher T

and lower water pressure.

point A

between the ﬁelds of quartz  liquid and al-

kali feldspar solid solutions  liquid, potassic alkali

feldspar, initially about Kf

, begins to coprecipitate

with quartz. Ultimately, as the residual melt almost

reaches the lowest T in the thermal valley, represented

by liquid at point M, it is entirely consumed into crys-

tallizing quartz and feldspar. Under equilibrium crys-

tallization and closed system behavior, the tie-line tri-

angle representing the three-phase assemblage melt

M  quartz  Kf

no longer encloses the bulk com-

position point A for the crystallizing system. Less

siliceous, more feldspathic systems initially precipitate

alkali feldspar as the liquidus phase. As alkali feldspar

solid solutions precipitate, residual melts track along a

curved path on the liquidus surface to the boundary

line where quartz coprecipitates.

Above 3.6 kbar in the water-saturated system the

minimum in the thermal valley of the liquidus surface

is replaced by a eutectic; the liquidus has been de-

pressed in T sufﬁciently to intersect the solvus (com-

pare Figure 5.17). Increasing P in water-saturated sys-

tems also shifts the position of the minimum and

eutectic toward more Ab-rich compositions (Figure

5.25). Other modiﬁcations in intensive variables, such

as addition of CaAl

, F, and CO

to the system,

also change the position of the minimum or eutectic

somewhat (Johannes and Holtz, 1996). The shift in the

minimum away from the silica apex corresponds to an

expansion of the stability ﬁeld of quartz on the liquidus

at higher P. This expanded stability ﬁeld may account

for the common occurrence of partially resorbed

quartz phenocrysts in silicic volcanic rocks, such as that

illustrated in Figure 6.19. For example, consider a

Crystal-Melt Equilibria in Magmatic Systems

109

60 62 64 66 68

Oxide concentration

in melt (wt. %)

(b)

SiO

concentration

in melt (wt. %)

CaO

NaAlSi

KAlSi

CaAl

(a)

5.23 Fractional crystallization in the ternary feldspar system

KAlSi

-NaAlSi

-CaAl

. (a) Schematic representa-

tive fractionation paths of residual melts on the liquidus sur-

face. Curvature of paths results from removal of ternary

feldspar solid solutions from the evolving melts. The ﬁlled tri-

angle and double line leading away from it represent a hypo-

thetical andesitic initial melt that fractionates to an evolved

rhyolitic composition. (b) Liquid lines of descent of the hy-

pothetical fractionating melt in (a). The shaded part indicates

the fractionation of plagioclase only from the melt, the un-

shaded part where two coprecipitating feldspars fractionate

from the melt as it tracks down T along the two-feldspar

boundary EM

. Note that fractionation yields residual

evolved melts that are increasingly more enriched in Si and

Na, but impoverished in Ca.

NaAlSi

Wt. % KAlSi

Alkali Feldspar

800

760

720

760

800

Qtz + L

2 kbar

water-saturated

SiO

5.24 Liquidus relations and phase compositions at P



2 kbar (water-saturated conditions) in the granite system

KAlSi

(Kf)-NaAlSi

(Ab)-SiO

-H

O. As projected here

onto the feldspar-silica composition plane, the system can be

considered to be ternary. Representative isothermal tie lines

connect coexisting phases. Filled squares are crystal composi-

tions; open circles are liquid compositions. Isothermal contour

lines with temperatures in Celsius show form of the liquidus

surface. Thermal valley in the liquidus surface where T 

720°C is shaded. (Reprinted by permission of the publisher

from Tuttle OF and Bowen NL, Origin of granite in the light

of experimental studies in the system NaAlSi

-KALSi

SiO

-H

O, Boulder, Colorado, The Geological Society of

magma system on the boundary line at 2 kbar that con-

sists of quartz  alkali feldspar  melt and whose bulk

composition is represented by the ﬁlled triangle in Fig-

ure 5.25. At lower P as the magma ascends to the

surface, the system becomes stranded in the stability

ﬁeld of alkali feldspar  melt; consequently, the pre-

viously precipitated quartz is now unstable and dis-

solves into the melt, but only partly if insufﬁcient time

is available, as would occur during rapid ascent and

eruption.

Origin of Granite Magmas. The KAlSi

-NaAlSi

SiO

system has been called the granite system because

many granites, and their compositionally equivalent

aphanitic and glassy counterparts, are predominantly

alkali feldspar and quartz (Figure 5.26). Granites con-

tain only a small amount of CaAl

, which appears

as the anorthite end member of plagioclase. However,

in granodiorites and associated dioritic rocks, this com-

ponent is more substantial and its absence in the model

system hinders direct application to such magmas. The

presence of Fe, Ti, Mg, and so on, which stabilize var-

ious maﬁc minerals, again especially in dioritic compo-

sitions, is another difference between this model sys-

tem and real magmas. Nonetheless, this system serves

as a very useful model for crystal-melt equilibria in

felsic magmas that form huge volumes of plutonic

and volcanic rock, particularly along convergent plate

margins.

It was said in ancient times that “all roads lead to

Rome.” In the granite ternary, all fractionation paths

lead liquids into the bottom of the thermal valley in

the liquidus surface—a unique composition that con-

tains more or less equal proportions of the three end-

member components (Figure 5.27a). This convergence

of residual melts during fractionation into the ternary

minimum on the liquidus surface is not limited to

initial compositions that lie within this ternary. Frac-

tionation of some magmas that initially have only

small concentrations of Na and K and modest Si can,

nonetheless, yield residual melts that move into this

ternary system and thence to the thermal valley as

fractionation continues. Subequal proportions of alkali

feldspar and quartz coexisting with melt at this low-T

minimum has, for good reason, been referred to as pe-

trogeny’s residua system. The striking coincidence of

petrogeny’s residua system with the favored composi-

tion of granites portrayed in Figure 5.26 supports the

hypothesis, asserted by Bowen (e.g., 1928) through-

out the ﬁrst half of the 20th century, that granite mag-

110 Igneous and Metamorphic Petrology

NaAlSi

Wt. % KAlSi

+ L

10 kbar

+ L

620°C

720°C

Quartz + L

SiO

5.25 Shift in minima and eutectics on the water-saturated liquidus

surface with respect to P in the system KAlSi

(Kf )-

NaAlSi

(Ab)-SiO

-H

O. Data projected onto the anhy-

drous triangular base. Minima (opposing arrows on boundary

line) and eutectics (triple line intersection) are for 1, 2, 5, and

10 kbar. The minimum becomes a eutectic at about 3.6 kbar.

The minimum T at 1 kbar is 720°C, and the eutectic T at

10 kbar is 620°C. Compare Figure 5.24. For comparison, the

2 kbar minimum and 5- and 10-kbar eutectics for water activ-

ity of 0.25 (undersaturated conditions) are shown as open

squares. (Redrawn from Johannes and Holtz, 1996.)

Ab Or

Normative wt.%

Upper limit

of analyses

Total

analyses

Analyses

as %

3,724

1,805

1,649

481

48.6

23.6

21.6

6.3

Total 7,659 100.1

5.26 Chemical analyses of 7659 granitic rocks from around the

world that have  80 wt.% normative Q  Ab  Or. Note

that about 50% of the analyses fall within only about 5% of

the diagram. Compare with Figures 5.24, 5.25, and 5.27.

(Analyses compiled and plotted courtesy of Roger W. Le

Maitre of the University of Melbourne.)

mas originate by extreme fractional crystallization of

basaltic parent magma. But Bowen’s assertion was re-

peatedly criticized by more ﬁeld-oriented petrologists

who could not ﬁnd any direct evidence for the approx-

imately ninefold greater volume of more maﬁc frac-

tionates that should be necessarily associated with

the huge granitic batholiths of mountain belts. (The

amount of granite components sequestered in a typical

basalt magma is only about 10%.)

Another origin for granite magmas that answers the

ﬁeld-relation criticism is to partially melt sialic rocks in

Crystal-Melt Equilibria in Magmatic Systems

111

Special Interest Box 5.3 Do granites originate

by solid-state granitization or from magmas?

The long-standing and often heated debate

(Pitcher, 1997, chapter 1) between magmatists and

granitizers peaked near the middle of the 20th cen-

tury and was ﬁnally resolved by the classic investi-

gation of Tuttle and Bowen (1958). They showed

that the restricted composition of granitic rocks

(Figure 5.26) corresponds closely with a thermal

valley in the liquidus surface in the model granite

system (Figures 5.24, 5.25, 5.27) where a melt coex-

ists in equilibrium with alkali feldspar and quartz.

Thus, granitoids are easily concluded to be the

product of crystallization of a corresponding magma

that either is a residual melt produced by fractional

crystallization of a parental magma of more maﬁc

composition (basalt) or is generated by partial melt-

ing of sialic continental rock, or is both. Besides this

rational explanation in terms of crystal-melt equilib-

ria, geologists began ﬁnding, in the decades after

1958, vast silicic pyroclastic deposits of mostly glass

fragments. Single eruptions of thousands of cubic

kilometers of pyroclasts could only mean that no

lesser volume of magma existed in a subterranean

preeruption magma chamber.

The granitization viewpoint was based on ﬁeld

relations of granite plutons (Read, 1957). To avoid

the “room problem”—how space was provided for

a magmatic intrusion in the crust—granitizers

claimed that the granitic rock was not intruded as

magma but was simply transformed, or granitized,

preexisting rock, a sort of subsolidus, or solid-state,

metamorphism. Chemical diffusion, or at least mo-

bilization and transport of chemical constituents in

migrating ﬂuids, on a very grand and pervasive

scale would be required. But chemical diffusion

was ﬁrst shown by Bowen (1921) to be much too

slow and other means of element transport lacked

a rational thermodynamic basis. But in support of

the granitization argument is the fact that some

plutons lack sharply deﬁned, intrusive contacts and

instead appear to grade over some distance from

granite into country rock, suiting a transformation

front. Confusing the issue is the fact that magmatic

granitic rocks have a variable and locally pervasive

overprint of subsolidus recrystallization that makes

them appear to have originated through metamor-

phism.

In some respects, the granite debate reﬂected

the mutual distrust of protagonists. Field-oriented

granitizers were highly skeptical of experimental

studies that employed small Pt crucibles in a fur-

nace allegedly mimicking huge plutons, and Bowen

and associates lamented the lack of understanding

of basic chemistry and physics of the granitizers.

Since the 1950s, petrologists have adopted a more

holistic integrated viewpoint that experiments and

theory are both useful in constraining and testing

hypotheses based on observations of real rocks, and

vice versa.

67 69 71 73 75 77 79

SiO

concentration

in melt (wt. %)

Oxide concentration

in melt (wt. %)

(b)

NaAlSi

Wt. % KAlSi

SiO

(a)

5.27 Effects of fractional crystallization in the granite system

KAlSi

(Kf)-NaAlSi

(Ab)-SiO

-H

O. (a) Fractionation

paths of residual evolving melts at 1-kbar water-saturated con-

ditions. (b) Liquid lines of descent for fractionation of the melt

represented by the ﬁlled triangle and the double line path

in (a). (Redrawn from Tuttle and Bowen, 1958.)

5.6 CRYSTAL-MELT EQUILIBRIA

INVOLVING ANHYDROUS

MAFIC MINERALS: OLIVINE

AND PYROXENE

Like plagioclases, olivines constitute a complete solid-

solution series, and their phase equilibria are similar.

The more complex pyroxene solid solutions are con-

sidered next.

Compositions of common pyroxenes, mostly in

subalkaline rocks such as basalt, are conveniently rep-

resented in the pyroxene quadrilateral, which is a part

of the ternary system CaSiO

-MgSiO

-FeSiO

. Low-P

liquidus relations (Figure 5.28a) are complex because

of the coexistence of solids whose compositions lie

well outside the ternary and its three deﬁning compo-

nents. For highly magnesian compositions, an Mg-rich

olivine is the liquidus phase, as it is in the Mg

SiO

system (Figure 5.8). Hence, incongruent melting

of enstatite still prevails with addition of Fe and Ca in

112 Igneous and Metamorphic Petrology

the continental crust. The initial partial melt of a rock

that contains substantial amounts of the components

KAlSi

, NaAlSi

, and SiO

in any proportion is a

melt of the ternary minimum T composition. Thus,

some shales and feldspathic sandstones and preexisting

granitic rocks, and their metamorphosed equivalents,

are potential source rocks for generation of granite

magma by partial melting.

Whether derived by fractionation of more maﬁc

parent magma or by partial melting of sialic crustal

rock, the restricted composition of granites (Figure

5.26) is governed by crystal-melt equilibria in the

KAlSi

-NaAlSi

-SiO

system. These fundamen-

tal phase relations, elucidated by Tuttle and Bowen

(1958) in a classic study, laid to rest the then-popular

argument that granites were created by “granitiza-

tion,” the transformation of solid rocks without in-

volvement of a silicate melt. Granitization cannot ex-

plain why the composition of granites should be so

restricted (Special Interest Box 5.3).

1500

Ortho-

pyroxene + L

1400

1300

Pigeonite

polymorph

silica

L +

Pyroxenoid

+ L

CaFeSi

CaMgSi

LIQUIDUS

Mg-olivine

+ L

Augite + L

(a)

(b)

SOLID

COMPOSITIONS

Aug

700

900

700

900

1100

Opx

1300

1100

Pig

Skaergaard

intrusion

Miharayama

volcano

Ultramafic Mafic Intermediate Silicic

5.28 Phase relations of common pyroxenes in the quadrilateral Mg

-CaMgSi

-CaFeSi

-Fe

at 1 atm. (a) Liquidus surface

with 100°C contours and boundary lines. (Redrawn from Huebner and Turnock, 1980.) (b) T-dependent pyroxene compositions. (Re-

drawn from Lindsley, 1983.) Augites (Aug) occur in a band across the top of the quadrilateral, orthopyroxenes (Opx) in a narrower band

across the bottom, and pigeonites (Pig) just above them. Representative partial contour lines on the solvus are labeled in degrees Celsius

for geothermometry after suitable corrections (Lindsley, 1983) have been made for other components in solid solution. Isothermal tie

lines at 1100°C and 900°C connect coexisting equilibrium compositions of three pyroxenes (shaded triangles). Filled squares are pyrox-

enes from the Skaergaard intrusion, Greenland. Open squares and connecting dashed tie-line are phenocrysts from the 1950–51 basalt

lava ﬂow of the Mihara-yama volcano, Japan. Approximate correlation between rock suites and pyroxene compositions is shown below

quadrilateral.

solid solution. For highly Fe-rich compositions, a large

ﬁeld of Fe-rich melts coexist with a silica polymorph.

This occurs because, unlike magnesian compositions,

an Fe-rich olivine  quartz (or one of its polymorphs)

is more stable than ferrosilite pyroxene (FeSiO

) un-

der most magmatic conditions. Instead of hedenber-

gite (CaFeSi

), a complex pyroxenelike solid, or

“pyroxenoid,” is the liquidus phase in the upper-right

apex of the quadrilateral. The remaining liquidus

phases are pyroxenes. Augite is a Ca-rich clinopyrox-

ene that crystallizes in a wide range of natural magmas.

In alkaline magmas that have low silica activity (Sec-

tion 3.5.3), it is the sole pyroxene and contains sub-

stantial amounts of Al, Ti, Fe

3

(all three partly sub-

stituting for Si in tetrahedral sites), and Na (for Ca) in

solid solution. Uncommon pigeonite is a Ca-poor

clinopyroxene crystallizing at low P in subalkaline

basalt and some andesite magmas. Orthopyroxenes

contain no more than 5 mole % of the calcic compo-

nent and crystallize only in subalkaline magmas, usu-

ally with augite. Fractionating melts migrate down T

toward more Fe-rich compositions and ultimately can

reach near CaFeSi

. Thus, Fe/(Fe  Mg) ratios in-

crease in more evolved, lower T-melts and in their

crystalline products.

Phase relations below the liquidus (Figure 5.28b)

are complicated by polymorphism in enstatites, meta-

stability in subcalcic augites and pigeonites, and sub-

solidus exsolution, none of which is discussed here.

Equilibrium compositions of two- and three-pyroxene

assemblages that can stably coprecipitate from a melt

are indicated by isothermal tie lines and tie-line trian-

gles in Figure 5.28b. Equilibria are strongly dependent

upon T and less upon P. Such equilibrium pyroxene

pairs can therefore serve as geothermometers for crys-

tallization T after their compositions have been appro-

priately projected from multicomponent space onto

the Mg

-CaMgSi

-CaFeSi

-Fe

quad-

rilateral and plotted on the diagram of nearest P (for

details see Lindsley, 1983). For example, coexisting

augite and orthopyroxene phenocrysts from a basalt

lava ﬂow extruded in 1950–51 from Mihara-yama vol-

cano, Japan, indicate crystallization at slightly more

than 1100°C; for comparison, the highest T of lava

measured during the eruption was 1125°C.

Pyroxenes and olivines in the highly fractionated

Skaergaard intrusion in northeast Greenland show the

degree to which Fe enrichment can be produced by a

strongly fractionating tholeiitic basalt magma. The rel-

atively reducing conditions (Frost and Lindsley, 1992)

in the fractionating Skaergaard magma body, to as

much as 2 log units below the QFM buffer (Figure

3.14), undoubtedly played a signiﬁcant role in stabiliz-

ing Fe-rich clinopyroxenes and pure fayalite olivine.

Relatively oxidizing magmas in which the oxygen fu-

gacity is higher have more Fe

3

and consequently less

2

to stabilize Fe-rich pyroxenes and fayalite.

5.7 CRYSTAL-MELT EQUILIBRIA IN

HYDROUS MAGMA SYSTEMS

Volatiles are generally present in terrestrial magma sys-

tems, can profoundly depress liquidi and solidi, and

stabilize volatile-bearing phases. In this section, some

aspects of crystal-melt equilibria in hydrous magma

systems are examined.

5.7.1 Equilibria in the Granodiorite-Water System

Granodiorite is a plagioclase-rich granitic rock (Figure

2.8) that contains more maﬁc minerals, generally bi-

otite and hornblende, than granite. Granodiorite and

its aphanitic to glassy volcanic counterpart, dacite, are

widespread in convergent plate margin settings.

To determine melt equilibria with two feldspars and

quartz in the proportions in which they occur in a typ-

ical granodiorite, Whitney (1988) studied a synthetic

rock free of maﬁc minerals that has weight percent-

ages of components CaAl

19.8, NaAlSi

37.3,

KAlSi

19.8, and SiO

23.1. Several generalities can

be gleaned from a T-X

diagram at constant P (Fig-

ure 5.29). Sequential and overlapping crystallization of

solid phases with decreasing T is once again evident, as

Crystal-Melt Equilibria in Magmatic Systems

113

0246 810

O (wt. %)

600

700

800

900

1000

1100

1200

T (°C)

+ Kf

+ Qtz + L + fluid

+ L + fluid

L + fluid

+ L

+ Kf

+ L

+ Kf

+ Qtz + L

+ Kf

+ Qtz + fluid

5.29 Crystal-melt equilibria for felsic phases in a synthetic granodi-

orite at 2 kbar. (Composition R5; redrawn from Whitney). Re-

gion of water-undersaturated melt is shaded. Dashed line indi-

cates the boundary line for stable quartz ( Pl

 L) at 8

kbar, showing the expansion of its stability ﬁeld at increased P.

L, silicate liquid (melt). (Redrawn from Whitney, 1975 cited in

Whitney, 1988.)

in Figures 5.10 and 5.11. Plagioclase is the liquidus

phase, continues to crystallize all the way to the solidus,

is the sole precipitating solid for a large range of T, and

is only joined by quartz and alkali feldspar (Kf

) near

the solidus. For water contents  3–5 wt.%, appropri-

ate to stabilize biotite and amphibole in natural multi-

component magmas, quartz and alkali feldspar only

precipitate within 20°C of the solidus. Depending

upon how much of the NaAlSi

component is dis-

solved in plagioclase, at least one-half of the system

does not crystallize until quartz and alkali feldspar

precipitate. These relations are compatible with tex-

tural relations in many granodiorites in which plagio-

clases are robust euhedral grains mingled with gener-

ally more anhedral quartz and alkali feldspar grains

that appear to have grown from the interstial melt lying

between the plagioclases. Piwinskii (1968) and subse-

quently many others have shown that in real granodi-

orite magmas, Fe-Ti oxides, biotite, and hornblende

along with common accessory phases zircon and ap-

atite crystallize at high T together with plagioclase; the

maﬁc crystals also tend to be more or less euhedral in

granodiorites. The origin of melts forming widespread,

late-stage dikes of leucocratic alkali feldspar-quartz

aplite in granodiorites can also be appreciated (see

Figure 9.3).

5.7.2 Equilibria Involving Melt

and Micas and Amphiboles

All of the crystalline phases heretofore discussed do

not contain volatiles as essential constituents. The

stability of micas and amphiboles, on the other hand,

depends on the existence of water in the system be-

cause (OH)



anions are essential in their crystalline

structure. Water concentrations  4–5 wt.% stabilize

amphibole in andesitic-dacitic magmas (Merzbacher

and Eggler, 1984); somewhat less water stabilizes bi-

otite. Micas and amphiboles also contain generally

lesser concentrations of F and Cl, which substitute

for (OH)



, and variable concentrations of Fe

2

and

3

. Therefore, their equilibria in magma systems de-

pend on the fugacities of H

O, F, Cl, and O, as well as

P and T.

Stability of Mg-Fe Biotites. Discussed in this section is

how oxygen fugacity, bulk chemical composition of the

magma, T, and P inﬂuence the Fe/(Fe  Mg) ratio and

stability of biotite solid solutions (Figure 5.30). Consid-

ering ﬁrst the pure Fe end-member biotite annite

(KFe

2

AlSi

(OH)

), it may be noted that this phase

is not stable at oxygen fugacities above the HM buffer

curve in Figure 3.14, where sanidine  hematite are

more stable, nor below the WI buffer, where fayalite

(pure Fe-olivine) and potassium aluminum silicates are

more stable. The following reactions constrain the sta-

bility of annite

5.6 KFe

2

AlSi

(OH)

 3O

 KAlSi

annite ﬂuid sanidine

 3Fe

3

 H

hematite ﬂuid

5.7 KFe

2

AlSi

(OH)

 0.5O

 KAlSi

annite ﬂuid sanidine

 Fe

3

2

 H

magnetite ﬂuid

5.8 2KFe

2

AlSi

(OH)

 KAlSiO

annite kalsilite

 KAlSi

 3Fe

2

SiO

 2H

leucite fayalite ﬂuid

These reactions may be viewed as decomposition

(breakdown) reactions in which an assemblage of

other phases takes the place of a single crystalline

phase, annite. The three oxidation-reduction reactions

are represented by the three heavy line segments

bounding the T-f

stability ﬁeld of annite (dark-

shaded) in Figure 5.30 (see also the buffer curves in

Figure 3.14).

114 Igneous and Metamorphic Petrology

−30

−25

−20

−15

−10

−5

Log oxygen fugacity (bars)

400 600 800 1000

T (°C)

ΗΜ

0.3

0.4

0.7

1.0

Fe/(Fe + Mg) of biotite

Sanidine +

hematite + fluid

Annite +

fluid

Kalsilite + leucite

+ fayalite + fluid

NNO

QFM

Sanidine

+ magnetite

+ fluid

5.30 Stability of biotite as a function of oxygen fugacity and T at

2070 bars. Compare with Figure 3.14 for identiﬁcation of

buffer curves (dotted lines) in the Fe-Si-O system. Stability

ﬁeld of the pure iron (Fe/(Fe  Mg)  100) end-member bi-

otite annite is the dark shaded area. Breakdown assemblages

from reactions 5.6 to 5.8 are indicated for annite. Stability lim-

its for Fe-Mg biotite solid solutions (light shade) are indicated

by subhorizontal line segments labeled along the HM buffer

curve with their Fe/(Fe  Mg) values. (Redrawn from Wones

and Eugster, 1965.)

More Mg-rich biotites are stable to higher T and

higher oxygen fugacities. Thus, a biotite in which

Fe/(Fe  Mg)  0.4 is stable (light shaded ﬁeld in

Figure 5.30) to as much as 1000°C and f

as much as

12

bar whereas annite is stable to only a little over

800°C and 10

17

bar. Decomposition reactions involv-

ing real Mg-Fe biotites, which also contain Ti, Mn, F,

and Cl, are more complex than those listed for the an-

nite end member.

Three aspects of biotite stability in magmatic

rocks are explained by the phase relations in Figure

5.30:

1. In many volcanic rocks, such as common calc-

alkaline andesites, dacites, and rhyolites, that have

cooled in the oxygen-rich atmosphere from rela-

tively more reduced extruded magmas, biotites are

typically bronze- or copper-colored. These are com-

monly mistaken for phlogopite, the Mg analog of

annite. However, phlogopite is never stable in such

calc-alkaline magmas and occurs only rarely in

some highly alkaline, ultramaﬁc rocks. Instead, the

bronze or copper color is caused by minute in-

cluded particles of hematite, locally with other Fe-

Ti oxides, in the biotite. Hematite is produced by

partial subsolidus reequilibration of biotite to the

high oxygen fugacity (0.21  10

0.68

bar) of the at-

mosphere via a reaction like reaction 5.6. There

may be no sanidine from the decomposition reac-

tion remaining because of its solubility in the high-

T steam escaping from the cooling rock.

2. In a wide range of magmas, biotite is stable with

magnetite and alkali feldspar, suggesting buffering

by a reaction like reaction 5.7.

3. In high-T magmas, irrespective of oxygen fugacity,

biotite is less stable than anhydrous silicates plus an

aqueous ﬂuid. The speciﬁc thermal breakdown re-

action for the annite end-member is reaction 5.8.

Thermal Stability of Volatile Minerals. The general

form of thermal breakdown (endothermic decomposi-

tion) reactions for volatile-bearing minerals is

5.9 thermal energy  volatile-rich mineral

 volatile-free or volatile-poor mineral(s)

 volatile ﬂuid

Absorption of heat by the volatile-rich mineral liber-

ates the volatiles from it: That is, hydrous minerals are

“dried out” by absorption of heat.

The upper thermal stability limit of a volatile-

bearing mineral lacking solid solution in P-T space

is shown in Figure 5.31a and the breakdown band

for a volatile-bearing solid solution crossing a magma

solidus in Figure 5.31b. From the Clapeyron equation

(3.13), because the high-T decomposition products

have a greater volume and entropy than the volatile-

bearing solid, the breakdown curve has a positive

slope. Because the volatile ﬂuid is more compressible

than the other phases in the reaction, this slope steep-

ens at increasing P and actually becomes negative at

high P. Hence, Mg-rich micas and amphiboles are sta-

ble into the upper mantle, but not deeper (see, for ex-

ample, amphibole stability in Figure 5.11).

It is important to realize the contrasting inﬂuence

of volatiles on the stability of volatile-free minerals,

such as feldspars, on the one hand and that on volatile-

bearing minerals, such as micas, on the other hand.

The upper thermal stability limit of volatile-free miner-

als represented in their melting curve is depressed to

lower T at increasing volatile concentration at increas-

ing P, partly because the mineral components in the

melt are diluted by the dissolved volatiles. On the other

hand, if volatiles are an essential component in a min-

eral, then increasing volatile fugacity stabilizes the

phase to higher T.

Crystal-Melt Equilibria in Magmatic Systems

115

(a)

(b)

Volatile-rich

mineral

Volatile-poor

minerals +

volatile fluid

Volatile-poor minerals

+ volatile fluid

Solidus

Rock containing a

volatile-rich

mineral

Partially

melted rock

containing a

volatile-rich

mineral

Volatile-poor

minerals 

volatile fluid

+ melt

5.31 Schematic phase diagrams for reaction 5.9. (a) Hypothetical

volatile-rich mineral that has no solid solution. At increasing P

the breakdown curve steepens and eventually has a negative

slope. (b) Hypothetical solid solution, such as amphibole, that

decomposes over a range of P and T (shaded). Hypothetical

solidus (dashed line) for a magma in which the solid solution

crystallizes and intersects the decom-position band. Compare

Figure 5.11 for a tholeiitic basalt system. Volatile-poor miner-

als may actually be volatile-free.

Case History from Mount Saint Helens. The impor-

tance of these stability relations in common hydrous

magma systems in which both hydrous and anhydrous

minerals crystallize can be illustrated for the Mount

Saint Helens, Washington, dacite magma episodically

erupted during the 1980–1986 activity. The preerup-

tion magma equilibrated at about 900°C and 2.2 kbar

or a depth of 8 km. Magmas extruded after the cata-

strophic explosive eruption of May 18, 1980, contain

irregularly shaped, partially resorbed amphibole phe-

nocrysts enclosed within a reaction rim aggregate of or-

thopyroxene, clinopyroxene, plagioclase, and ilmenite

(Figure 5.32); these same crystalline phases occur as

phenocrysts and matrix constituents in the dacite. Two

observations indicate that in the preeruption magma

chamber the amphibole partially reacted with the melt

to produce the more stable anhydrous mineral assem-

blage in the reaction rim:

1. The aggregate occurs where the amphibole is sur-

rounded by melt (now glass) and not where it is

in tight contact with another crystalline phase, such

as plagioclase, or in the interior of the amphibole

crystal.

2. The bulk chemical composition of the aggregate is

not equivalent to that of amphibole.

The reaction, as calculated by Rutherford and Hill

(1993), involved seven parts amphibole and three parts

melt on a weight basis.

How was the magma system displaced out of the

stability ﬁeld of amphibole so as to produce the partial

resorption? One possibility is an inﬂux of new, higher-

T, less hydrous magma into the chamber, but there

is no independent evidence for this. Another explana-

tion compatible with the observations of the eruptive

activity and its products involves relatively slow de-

compression of an ascending magma, and possibly, but

116 Igneous and Metamorphic Petrology

Plagioclase

microlites in

glass

Reaction rim

Amphibole

Vesicle

Plagioclase phenocryst

0 0.1mm

5.32 Reaction rim around a partially resorbed amphibole phenocryst. Photomicrograph under plane-polarized light of post-May 18, 1980,

Mount Saint Helens dacite. Anhedral amphibole phenocryst (dark color with characteristic cleavage) is partly surrounded by a reaction

rim aggregate of pyroxene  plagioclase  Fe-Ti oxides. Note that the rim occurs between the amphibole and glassy matrix, which was

a reactive melt at the time the rim grew, but not adjacent to a contacting “armoring” plagioclase phenocryst. (Photomicrograph courtesy

of M.J. Rutherford.)