Myron G. Best. Igneous and metamorphic 2003 Blackwell Science

Подождите немного. Документ загружается.

Boundary of accreted Andean-type

arc terrane or continent–continent

collision; teeth on terrane added

to older craton

0 800

0 500

miles

1920–1790 Ma

Rae

Superior

Slave

Rae

1980–1840 Ma

Hearne

Wyoming

1790–1690 Ma

1710–1620

1500–1340 Ma

1900–1790

Grenville

Appalachian

(Pz)

Cordilleran ( Mz)

Keweenawan continental

rift 1100 Ma

1920–1790

later) were emplaced in the northeastern part of the

orogen and into adjacent Labrador during three pulses

from 1650 to 1270 Ma. But a voluminous burst of in-

trusions occurred at 1170–1080 in an interior (central)

magmatic belt (Figure 19.28) during periods of crustal

extension between contractional pulses.

19.5.2 Massif-Type Anorthosite

Massifs of anorthosite and kindred rocks were em-

placed into shallow- to moderate-depth crustal meta-

morphic rocks during the middle Proterozoic. They

have been found on every continent except Green-

land and Australia (Ashwal, 1993, Fig. 3.1) but their

largest concentration occurs in eastern North America

(Figure 19.28). Nearly all have been deformed and re-

crystallized to some degree, erasing primary magmatic

features and hindering accurate interpretations. None-

theless, most appear to be basin- or dome-like sheets a

few kilometers thick and as much as thousands of km

in area. Least deformed massifs appear to be composite

plutons that consist of many separate intrusions (Fig-

ure 19.30). Some have features transitional to layered

differentiated maﬁc intrusions, such as the Skaergaard

(Figures 7.45, 12.19, and 12.20).

Anorthosite Suite. Anorthosite is generally the prin-

cipal rock type in massifs but there are also variable

amounts of leucogabbro, leuconorite, and leucotroctolite

Precambrian Rock Associations

637

19.27 Laurentia Craton made of amalgamated Archean and early Proterozoic terranes (ages in Ma). Archean terranes (2.5 Ga; shaded)

include Slave, Rae, Hearne, Wyoming, and Superior. The Grenville Orogen (dark shaded) was sutured onto the southeast margin of

Laurentia from about 1.7 to 1.2 Ga by arc accretion and continent–continent collision. Later, during the Paleozoic, the Appalachian

Orogen was formed of the reworked southeastern side of the Grenville orogen (note dark shaded outliers) and of newly accreted crust.

The Cordilleran Orogen was added to the western side of Laurentia and Grenvillia (in Mexico) during the late Paleozoic and Mesozoic.

Dashed line is present coastline of North America. Redrawn from Rivers (1997).

Anorthosite-suite massifs

1170–1040 Ma

1365–1270 Ma

Nain

1320–1270

Atlantic

55°N

45°C

60°W

Shear zone;

ornament on

overriding plate

1500–1420 Ma

1645–1620 Ma

Georgian Bay

Lake Ontario

75°W

Marcy

1135–1115 Ma

Montreal

Ottawa

Morin

1155

Grenville

Lac St.

Jean

1160–

1142

Front

Ocean

0 km 400

0 miles 250

19.28 Massifs of the anorthosite suite in the northeastern Grenville Orogen of eastern Canada (shaded). The Grenville Front is the northwest-

ern-most mylonitic shear zone of ductile thrusting where high-grade metamorphic rocks were pushed from the southeast onto the mar-

gin of the Laurentian Craton. Rocks between the front and the next shear zone to the southeast are related to those in Laurentia but have

a Grenvillian tectonic and metamorphic overprint. An enlarged map of the southern part is Figure 19.29. Redrawn from Hoffman (1989)

and Rivers (1997). E, eclogite locality (Indares, 1993).

(Figure 2.10a) that have 10–35% maﬁc minerals as well

as more maﬁc gabbro, norite, troctolite, and minor

rocks that contain abundant Fe–Ti oxides. Altogether,

these associated comagmatic rock types constitute an

anorthosite suite (Table 19.3 and Figure 19.31). In

many locales, the concentrations of ilmenite and mag-

netite are sufﬁcient to create ore deposits; the largest

Ti deposits in the world are associated with anorthosite

massifs. Possible complementary and cogenetic ultra-

maﬁc rocks are rarely exposed and geophysical data

in the form of negative gravity anomalies suggest no

large ultramaﬁc mass directly underlies massifs.

Overprinting cataclasis (Figure 8.7) and recry-

stallization are widespread in the anorthosite suite,

obscuring primary fabrics and generally reducing grain

size. Where metamorphic overprints are minimal,

vestiges of magmatic fabric resemble that of many

gabbroic rocks. In leucogabbros, ophitic texture is

638 Igneous and Metamorphic Petrology

common. Plagioclase crystals range in size from 1 to

10 cm but locally are 1 m or more across. Unlike equant

megacrysts in Archean anorthosites, plagioclases in

Proterozoic anorthosite massifs have typical tabular

habit, which allows expression of igneous lamination.

Various forms of layering and cumulate textures (Fig-

ure 12.16) found in differentiated maﬁc intrusions are

less common.

Plagioclase compositions in massifs as a whole

generally range between An

and An

(andesine to

labradorite), in contrast to the bytownite of Archean

anorthosites and labradorite–bytownite of layered

maﬁc intrusions. In massifs, more calcic plagioclases

are associated with olivine and the more sodic with

orthopyroxene. The content of An in plagioclases is

generally uniform within a particular intrusion; normal

and oscillatory zoning in individual crystals is unusual

whereas reverse zoning is more common.

Plagioclases typically contain different kinds of in-

clusions that formed by exsolution during slow cooling

from initially homogeneous crystals. Relatively sodic

plagioclases are antiperthites that contain exsolved

blebs and lamellae of K-rich feldspar formed from

initially K-rich plagioclase. Inclusions of very ﬁne

(1 micron) Fe–Ti oxides are responsible for black

and shades of gray, blue, and red coloration. Particularly

characteristic of massif-type anorthosites are iridescent

(“labradorescent” or “schillar”) plagioclases produced

by optical interference from regularly spaced, very ﬁne

exsolution lamellae in the composition range between

Precambrian Rock Associations

639

Metamorphosed syenite,

mostly 1200–1100 Ma

Metamorphosed 1700–1400 Ma

granitoids

Ortho- and paragneiss

Quartzite, marble,

paragneiss

2700 Ma paragneiss

Charnockitic suite

Metamorphosed anorthosite,

minor gabbro

Metamorphosed 1300–1100 Ma

granitoids and felsic-maﬁc

volcanic rocks

Metamorphosed 1090–1050 Ma

nepheline syenite

81°W

79°

77°

75°

Extensional

shear zone

Toronto

Lake

Ontario

New York

44°

45°

Montreal

Ottawa

Quebec

New York

St. Lawrence

River

Frontenac

terrane

Elzevir terrane

Contractional

shear zone

Bay

Georgian

45°N

44°

Huron

Lake

Province

Grenville

Sudbury

Superior

Front

79°

76°

Adirondack

Highland

terrane

19.29 Southern end of Grenville Orogen (shaded) in southern Quebec, Ontario, and New York. Pre-Grenville rocks of the Superior Province

to northwest and younger rocks mainly to the south are unshaded. This northwest–southeast transect across the orogen reveals several

mylonitic ductile shear zones in rocks of mid-crustal depths. Contractional zones mark northwest-directed ductile thrusting and exten-

sional shears mark younger minor crustal stretching. Rocks immediately southeast of the Grenville Front include reworked Superior

Province rocks, such as 2700 Ma paragneiss, that were pushed northwestward onto the foreland. Farther southeast, east of Georgian Bay,

a central zone terrane of gneisses intruded by 1700 and 1400 Ma granitoids was transported northwestward on granulite-facies

mylonite zones. The central zone was subsequently a continental margin against which the Elzevir arc terrane and older crust of the Fron-

tenac terrane were thrust northwestward and accreted probably during the Grenville orogeny. The last metamorphic imprint is about

1100–1000 Ma. Redrawn from Hoffman (1989), in which further details of the orogen are described.

and An

where a narrow solvus exists (not the

same solvus as in Figures 5.18 and 5.21). Metamorphic

recrystallization purges exsolved phases from their

host plagioclase crystals, lightening their color as the

exsolved phase forms discrete separate grains of lower

surface energy. This lightened and green coloration

(Section 19.3.3) and reduction of grain size are useful

ﬁeld guides to recrystallized parts of massifs.

Typical maﬁc minerals in massif-type anorthosites

include olivine, (locally occurring as a cumulus phase),

orthopyroxene (typically forming ophitic textures with

major amounts of euhedral to subhedral plagioclases),

and Fe–Ti oxides. A characteristic phase is a high-Mg–

Al orthopyroxene that is typically seen as megacrysts

10–50 cm across but locally to 1 m. The unusually

high concentration of Al

(3–12 wt.%) is harbored

mostly in ﬁne exsolution lamellae of calcic plagioclase.

Most petrologists (e.g. Longhi et al., 1999) believe

these megacrysts crystallized at high pressures in

the lower crustal or upper mantle before they were

640 Igneous and Metamorphic Petrology

Orthopyroxene

granite and monzonite

Ferrogabbro,

ferrodiorite

Recrystallized

anorthosite

Anorthosite

Troctolite

Gneiss country

rock

Strike and dip

direction of

foliation

0km 10

0 miles 6

74°15′ 74°00′ 73°45′

46°15′

46°00′

19.30 Morin Anorthosite Massif, Quebec emplaced at 1155 Ma. For location see Figure 19.28. Field relations clearly indicate that anorthosite

is the oldest magmatic rock and orthopyroxene granite and monzonite the youngest (Emslie, 1975). All rocks are deformed and recrys-

tallized; this overprint is especially strong in the eastern anorthosite lobe. Redrawn from Martignole (1974).

introduced into the shallower crust where the exsolution

occurred and the anorthosite magma ﬁnally solidiﬁed.

Dark green, brown, or red A-type rocks of the

charnockitic suite (Sections 13.9 and 15.2.5) are

commonly spatially and temporally associated with

anorthosite massifs. Whether they are comagmatic has

been controversial. Close relation in space and time

is necessary but not sufﬁcient evidence for a genetic

relationship. Because charnockitic magmas were nearly

always intrusive into anorthosite (seldom the reverse)

and because of their higher Fe/(Fe  Mg) in bulk rock

and maﬁc mineral compositions, some petrologists

have interpreted them to be evolved residual mag-

mas from parental magma of anorthosite composition.

However, charnockitic rocks tend to lie off the

anorthosite-suite variation trend (Figure 19.31b), cast-

ing doubt on the hypothesized genetic relation. More-

over, charnockitic rocks do not always show depletion

in Eu as expected if the magmas were derived by frac-

tionation of the magma from which anorthosite cry-

stallized (Figure 19.32). Charnockitic rocks also have

higher large-ion-lithophile-element concentrations and

higher initial Sr isotope ratios and lower Nd ratios than

anorthosite, suggesting that they were derived from

crustal partial melts, in contrast to anorthositic magmas

that have a dominant mantle component. The mantle-

derived anorthositic magmas conceivably supplied the

heat necessary for the partial melting of the lower con-

tinental crust that generated the charnockitic magmas.

Origin of the Anorthosite Suite. A variety of petro-

genic models have been advocated for massifs of

anorthosite. In one proposal, espoused as late as the

1970s, large masses of calcic plagioclase were thought

to be created by large-scale metasomatic transforma-

tion of interbedded Al-rich shales and Ca-rich gypsum

or anhydrite evaporites. However, very slow rates of

atomic diffusion make this model untenable. The

magmatic origin of anorthosite is incontestable for the

following reasons:

1. Pyroxene geothermometry indicates magmatic

temperatures.

Precambrian Rock Associations

641

Table 19.3. Chemical Compositions of Rocks from

the Morin Massif-Type Anorthosite Suite

123

SiO

54.33 52.66 46.96

TiO

0.26 1.30 3.07

26.56 22.09 15.88

0.51 2.21 6.21

FeO 1.09 3.46 8.65

MnO 0.02 0.06 0.22

MgO 0.92 2.8 3.37

CaO 9.97 9.72 9.97

O 4.71 4.17 3.01

O 0.74 0.75 0.86

0.04 0.08 1.03

LOI 1.24 1.16 1.52

total 100.59 100.91 100.75

Fe/(Fe + Mg) 0.485 0.522 0.703

% Normative

feldspar 93 79 58

Ab 42.7 44.6 44.0

An 52.6 49.8 47.2

Or 4.7 5.6 8.8

1. Average of 14 anorthosites.

2. Average of 15 leucogabbros.

3. Average of four ferrogabbros.

Data from Ashwal (1993) and Papezik (1965).

Alkali

feldspar

Plagioclase

Quartz Quartz

Qtz

Diorite

Pyroxene, Fe–Ti oxides

and other maﬁc minerals

19.31 Composition of anorthosite-suite rock types in the Morin

Massif, Quebec. (a) Modal composition of charnockitic suite

rocks (open circles) and anorthosite, leucogabbro, gabbro,

and norite (small ﬁlled circles). Redrawn from Emslie (1975).

Dividing lines from Figure 2.8. Shaded area at plagioclase

apex is ﬁeld of diorite, gabbro and anorthosite. (b) Chemical

composition; signiﬁcant modiﬁcation by metamorphism is

possible. Redrawn from Ashwal (1993). Only a crude evolu-

tionary trend from anorthosite to leucogabbro to ferrogabbro

(ferrodiorite) to Fe–Ti-oxide-rich rocks is evident. Charnockitic

rocks lie well off this trend.

(Fe  Mg)

1.0

0.8

0.6

0.4

0.2

SiO

(wt.%)

Silicic

charnockitic

rocks

Ferrodiorite

Fe–Ti-oxide-rich

rocks

Ferrogabbro

Gabbro,

norite

MORIN

MASSIF

20 30 40 50 60 70

Anorthosite

Leucogabbro, leuconorite

2. Fabrics, such as ophitic texture, are consistent with

crystallization from a melt.

3. Positive Eu and Sr anomalies in trace element pat-

terns (Figure 19.32) are consistent with substitution

of Eu and Sr for Ca in plagioclases crystallizing

from maﬁc magmas.

4. Field relations, Fe/(Fe  Mg) ratios, and plagio-

clase compositions (Table 19.3) indicate that the

commonly associated sequence (Figures 19.30 and

19.31) of anorthosite, leucogabbro, ferrogabbro,

and Fe–Ti-oxide rock formed through crystal-melt

fractionation. (Nonetheless, not all ferrogabbroic

rocks, including commonly associated ferrodior-

ites, show a negative Eu anomaly complementary to

the positive anomaly in anorthosite, suggesting the

Fe-rich rocks are polygenetic.)

5. Isotopic data indicate variable amounts of a mantle

component.

But the exact nature of the parental magma of

massif-type anorthosites has been a long-standing

controversy. If basaltic, it must have been capable of

crystallizing large amounts of andesine–labradorite and

only minor amounts of maﬁc minerals, mostly orthopy-

roxene and lesser local olivine. Because of the relatively

high Fe/(Fe  Mg) 0.4 in anorthosite suite rocks

(Figure 19.31b), the parent magmas were more evolved

than typical primitive basalt magmas. That anorthosites

are cumulates is compatible with the virtual absence of

glassy or aphanitic rocks of anorthositic composition

and with the relatively uncommon occurrence of

anorthosite dikes. Hence, magmas of anorthosite com-

position probably do not exist. The cumulate heritage

is supported by transitional occurrences and resem-

blances with differentiated layered basaltic intrusions

(Section 12.4.2; see also Ashwal, 1993). But was this

parent basaltic magma unusually enriched in plagioclase

components, as ﬁne-grained “quenched” margins around

leucogabbro on some massifs suggest, or was it a nor-

mal tholeiitic basalt magma from which plagioclases

accumulated? The cumulate origin, espoused by N. L.

Bowen as early as 1917, raises three major questions:

1. Where are the complementary pyroxene–olivine

fractionates?

2. How was this ultramaﬁc fraction so effectively sep-

arated from plagioclase?

3. If anorthosite massifs were created by plagioclase

fractionation and accumulation, why are they re-

stricted to the middle Proterozoic?

The ﬁrst two questions are addressed in the model

depicted in Figure 19.33 that follows the view of

Bowen (1917), as further developed by Emslie (1978)

and Ashwal (1993), among others. In this model,

mantle-derived basaltic or picritic magmas are buoy-

antly blocked at and underplate the continental crust.

Crystallizing olivines and pyroxenes sink, forming ul-

tramaﬁc cumulates which become part of the mantle.

The residual melt becomes enriched in Al, Ca, Na, and

LILE, while Fe/Mg is increased. Crystals of labradorite–

andesine precipitate, which ﬂoat to the top of the

magma chambers in a suspension of Fe–Al-rich re-

sidual melt. High P not only stabilizes aluminous

orthopyroxenes, some of which are entrapped in the

rising mat of buoyant plagioclase, but also allows

crystallization of more sodic plagioclases that are

signiﬁcantly less dense than the melt (Sections 8.3.3

and 16.4.2). High temperatures in the magma and its

slow cooling allow growth of large crystals. Heat from

the crystallizing magma promotes partial melting of the

overlying sialic crustal rocks, producing charnockitic

magmas. The accumulated layer of ﬂoated plagioclases

becomes gravitationally unstable beneath slightly denser

crustal rock and rises as buoyant diapirs, like those of

granitic magma shown in Figure 9.15. Coalescence of

multiple diapirs in the upper to middle crust forms

composite anorthosite massifs.

Isotopic data (Emslie et al., 1994; Scoates and Frost,

1996) as well as crystal-melt equilibria constraints

642 Igneous and Metamorphic Petrology

Rock/primitive mantle

100

ANORTHOSITE

Margin

Maﬁc

Margin

Quartz-rich

ORTHOPYROXENE

MONZONITE

Ce Nd Sm Eu Tb Yb Lu

19.32 Representative primitive mantle-normalized rare-earth-element

patterns for anorthosite (dashed lines) and charnockitic ortho-

pyroxene monzonite in the Marcy Massif, Adirondack

Mountains, New York (location in Figure 19.28). Data from

Seifert et al. (1977). Anorthosites have variable concentra-

tions of REE and strong positive Eu anomalies except for one

leucogabbro near the margin of the massif (open circles).

Charnockitic rocks that intrude anorthosite are more REE

enriched. Quartz-rich sample has a negative anomaly, maﬁc

sample has a positive Eu anomaly, and a sample from the

margin of the intrusion (open triangles) has little anomaly; this

pattern is consistent with the charnockitic magma differentiat-

ing internally and independent of any genetic relation to the

anorthosite magma.

(Longhi et al., 1999) and ﬁeld relations (Dempster

et al., 1999) indicate variable amounts of contamination

of primitive mantle-derived maﬁc magmas by con-

tinental crustal components (Figure 19.34). Chemical

variation diagrams, such as Figure 19.31b, that depict

considerable scatter in rock compositions are difﬁcult

to reconcile with simple closed-system differentiation

processes in the evolution of anorthosite suites. Meta-

morphic overprints could, of course, contribute to the

scatter.

Petrologists for well over a century (see historical

summary in Bowen, 1928, p. 208) have noted that

contamination of maﬁc magma by aluminous rock can

suppress crystallization of clinopyroxene while enhan-

cing formation of plagioclase and orthopyroxene. This

is the very attribute of anorthosite suites that has made

recognition of parent magmas so problematic. The

contamination reaction, expressed in terms of end-

member mineral components, is

SiO

 CaMgSi

→

pelitic clinopyroxene

contaminant in magma

CaAl

 MgSiO

plagioclase orthopyroxene

in magma in magma

The possible role of immiscible melts in produc-

tion of complementary silicic and Fe–Ti-rich melts

associated with anorthosite magmas has yet to be fully

understood (Loferski and Arculus, 1993).

Tectonic Setting. The third question posed above has

not been satisfactorily resolved. If creation of massif-

type anorthosite is indeed linked to ponding of basaltic

magmas near the continental Moho, then several mod-

ern tectonic settings are possible. One that has been re-

peatedly suggested is the continental rift where A-type

granites are common. A possible analog of Proterozoic

anorthosite massifs is the 1200-km-long Silurian–

Jurassic Nigeria–Niger belt of anorogenic ring com-

plexes (Figure 13.35) where anorthosite is associated

with A-type felsic rocks (Demaiffe et al., 1991). More

anorthosite might exist beneath the present level of

erosion in the Niger association that, interestingly,

is also long-lived (250 My) like the eastern North

America association (500 My). However, not all

Precambrian Rock Associations

643

Silicic partial

melt

Continental

crust

Mantle

(a) Basaltic magma underplates

crust; maﬁc crystals sink,

plagioclase ﬂoats

(b) As magma continues to

crystallize, instabilities develop

in mass of ﬂoated plagioclase

diapirs begin to rise

(d) Ascended diapirs

coalesce into anorthosite

massif; ultramaﬁc cumulates become part of mantle

19.33 Conceptual model for evolution of Proterozoic anorthosite

massifs. Redrawn from Ashwal (1993).

⫺2

⫺4

0.5124

0.5125

0.5126

0.5127

143

Nd/

144

100

0.7140.7120.7100.7080.7060.7040.702

High-Al basalt

parent magma

Proterozoic

granitoids

Archean

gneiss

Archean

metashale

Sr/

19.34 Isotope ratios of anorthositic rocks (ﬁlled circles) from the

Laramie Anorthosite Complex, Wyoming, and possible parent

magma and crustal contaminants. Tick marks on contaminant

curves labeled 5, 10, 15, and so on indicate percent of con-

taminant assimilated into parent high-Al basalt parent magma.

Small amounts of assimilated Archean metamorphosed shale

and granitic gneiss that have

Sr/

Sr  0.75331 and 0.76597

and 

16.2 and 16.3, respectively, are capable of

generating the array of isotope ratios in the anorthositic rocks.

Redrawn from Scoates and Frost (1996).

geologists are willing to believe the Nigeria–Niger belt

is a rift structure. The notion discussed above that the

Grenvillian orogen, where most anorthosite massifs

are lodged, began to evolve during Andean-type plate

convergence and concluded with Himalayan-type

continent–continent collision provides the opportun-

ity for potential rifts during episodes of late orogenic

extensional collapse (Rivers, 1997). In both of these

Cenozoic orogenic systems, thickened crust has experi-

enced spatially and temporally localized episodes of

extension and rifting.

In another model, the apparent absence of anor-

thosite massifs before 1600 Ma and since 1100 Ma

was a result of unfavorable tectonothermal conditions.

The Laurentian continent that was assembled about

2000–1800 Ma (Figure 19.27) is envisaged by Hoffman

(1988) to provide a unique situation during the history

of the Earth. This continent was the ﬁrst of sufﬁcient

size to act as an insulating thermal blanket, reducing

heat loss from the underlying mantle, and promoting a

massive upwelling of mantle that produced a copious

cascade of basaltic magmas underplating the base of

the continent to generate anorthosite massifs. This pos-

tulated upwelling must have had a wall-like conﬁgura-

tion thousands of kilometers long, about an order

of magnitude larger than the diameter of cylindrical

modern mantle plumes beneath Hawaii and Iceland

(Figure 13.6). Since the massif–anorthosite event,

Hoffman (1988) believes the mantle cooled sufﬁciently

to preclude recurrence of this unique thermal anomaly.

19.5.3 Rapakivi Granites

The unusual and classic rapakivi granites of southern

Finland have been recognized for three centuries.

Similar rocks are now known elsewhere, mostly in the

Baltic Shield, in southern Greenland, and in a belt

extending southwesterly from Labrador through mid-

continent to the southwestern USA (Figure 19.35).

These granites, also including notable occurrences in

Brazil and Venezuela, are middle Proterozoic, from

1800 to 1100 Ma. Younger and older rapakivi granites

are widely scattered and of minor volume, according to

Haapala and Rämö (1999). These Finnish petrologists,

whose summary is reviewed below, deﬁne rapakivi

granites as “A-type granites [Section 13.9] character-

ized by the presence, at least in the larger batholiths,

of granite varieties showing the rapakivi texture.”

Rapakivi granites share compositional features with

Cenozoic topaz granites and rhyolites in bimodal asso-

ciations with basaltic rocks.

At least two explanations have been proposed for

the origin of rapakivi texture in which K-rich alkali

feldspars are mantled by sodic plagioclase (Figure

7.18). Wark and Stimac (1992) propose mixing of sili-

cic and more maﬁc magmas so that alkali feldspar

phenocrysts crystallized in the former are rimmed by

plagioclase components from the latter. Nekvasil

(1991), on the other hand, proposes an ascending,

decompressing, closed magma system that stabilizes

plagioclase at the expense of alkali feldspar. Figure 5.25

shows that a model granite magma system at, say, P 

5 kbar in which quartz  albite  K-feldspar are in

equilibrium with melt at a ternary invariant point will,

as it decompresses to lower P, experience a shift in

phase equilibria. The system is stranded in the stability

ﬁeld of albite  melt so albite precipitates on unstable,

resorbing K-feldspar; unstable quartz also dissolves into

the melt. Eklund and Shebanov (1999) ﬁnd evidence

for this scenario in Proterozoic rapakivi batholiths

in the Fennoscandian Craton. They extend the deﬁni-

tion of rapakivi fabric to include not only plagioclase-

mantled anhedral alkali feldspars but anhedral quartz

and euhedral plagioclase megacrysts as well. Mineral

inclusions in the partially resorbed quartz and alkali

feldspars indicate equilibration at 5–6 kbar whereas

inclusions in the plagioclase rims indicate growth at

644 Igneous and Metamorphic Petrology

Mostly rapakivi granite

Mostly anorthositic

suite rocks

Rapakivi and

anorthositic suite

1500–1000 Ma

1600–1200 Ma

1600–1500 Ma1600–1500 Ma

1800 Ma

1750 Ma

19.35 Distribution of Proterozoic rapakivi granites and anorthositic suite rocks. Redrawn from Haapala and Rämö (1999). Compare

Figure 19.28.

1–3.5 kbar. Evidence for a decompression origin of

rapakivi texture is better preserved in shallower plu-

tons than in larger, deep crustal, more slowly cooled

intrusions.

The prominent belt of rapakivi granites in northern

Europe and across North America coincides broadly in

age and eclipses the geographic belt of massif-type

anorthosite suite (Figures 19.28 and 19.35). Hence,

any model explaining the origin of the former must

somehow account for the latter, even though the two

may not be strictly co-magmatic. The two might at least

share a similar tectonic setting.

Though by no means certain, an incipient or

aborted continental rift has been suggested because

this is the tectonic setting of Phanerozoic A-type gran-

ites. Indeed, extensional tectonic features have been

recognized in some of the Finish Proterozoic rapakivi

occurrences. A preferred model (Frost and Frost,

1997; Haapala and Rämö, 1999) for the origin of

the essentially bimodal maﬁc–silicic magmas of the

anorthosite–granite association involves underplating

of continental crust by mantle-derived tholeiitic basalt

magmas. Expelled heat generates granitic partial melts

from sialic country rocks. However, the lack of negat-

ive Nb anomalies in most rapakivi and A-type granites

precludes much crustal contribution. This suggests

that, especially for the more reduced rapakivi magmas,

derivation was by extreme differentiation of the tholei-

itic intrusions or partial melting of them generated by

heat from subsequent and continuing underplating

(Section 13.9.2).

U–Pb chronologic data show that the three-step

westward growth of the Proterozoic Baltic Shield in

southwestern Sweden by arc magmatism was “echoed”

by three slightly younger episodes of also westward mi-

grating intracratonic rapakivi magmatism in southeast-

ern Sweden (Åhäll et al., 2000). How these contrasting

styles of magmatism might be linked in the convecting

mantle is speculative. Indeed, how magmatism in the

8000-km-long belt of 1800–1100 Ma rapakivi granites

and largely coeval anorthositic rocks is linked to man-

tle dynamics is a major unresolved question.

19.6 GRANULITE-FACIES TERRANES

IN ARCHEAN AND PROTEROZOIC

CRATONS

Although Phanerozoic granulite-facies rocks do exist,

such as the famous Hercynian granulites of Europe

formed at about 300 Ma, the majority of these high-

grade metamorphic rocks occur in Precambrian cra-

tons (Harley, 1989). In some places, there is a regular

and continuous prograde zonation from amphibolite-

facies rocks into granulite but more typically a terrane

of wholly granulite-facies rocks is separated from un-

like rocks by an unconformity, post-metamorphic fault,

or shear zone. Some wholly granulite terranes were

metamorphosed at constant P but variable T, whereas

others are the reverse. Many granulite terranes show

partial to complete retrogression to amphibolite- or

greenschist-facies assemblages, particularly along faults

and shear zones where water inﬁltrated. Granulite

protoliths are compositionally diverse sedimentary

and magmatic rocks. Polymetamorphism is typical.

Charnockitic rocks (Section 15.2.5) are widespread.

Many granulite terranes have a subhorizontal planar

fabric expressed by compositional layering that has

been locally isoclinally folded and transposed into root-

less folds (Figure 14.19b). This fabric is consistent with

large amounts of vertical ﬂattening and subhorizontal

extension.

For brevity in this section, granulite-facies rocks and

terranes will simply be referred to as granulites. This

usage does not connote a particular fabric, as in the

older geologic literature.

19.6.1 Mineral Assemblages and Reactions

The transition from hydrous amphibolite-facies to mostly

anhydrous granulite assemblages (Sections 18.3.1 and

18.3.5) is marked by formation of orthopyroxene 

alkali feldspar at the expense of biotite in pelitic and

felsic rocks and of orthopyroxene  clinopyroxene (

almandine–pyrope garnet  quartz) from hornblende

in more maﬁc rocks. Mineral assemblages and thermo-

barometry indicate granulite assemblages equilibrate

over a broad range of temperatures generally from

about 650 to 900°C but to as much as 1050°C and

pressures of generally 5 kbar to as much as 12 kbar

or depths of 20–45 km (Harley, 1989, 1998). Lower

pressures are the same as those at which amphibolite-

facies assemblages equilibrate but temperatures are

higher and water activity is lower. At one time it was

thought that the only difference in equilibrating con-

dition was the lower water activity. Possible factors

lowering water activity include extraction of water

from the prograding system via dehydration reactions,

extraction of hydrous partial melts, or transfer of water

down a concentration gradient into adjacent dry maﬁc

magmas. Another factor in the conversion could be

inﬁltration of CO

into the amphibolite system, dilut-

ing the water concentration and destabilizing hydrous

phases. That this can actually happen is vividly shown

in outcrops in southern India where light-colored

amphibole–feldspar orthogneiss has been converted

into dark pyroxene charnockite in folds and tension

fractures related to overprinting shear deformation

( Janardhan et al., 1982). Fluid inclusions in the

charnockite are dominantly CO

whereas those in the

amphibole gneiss are dominantly H

O. Beginning

with Touret in 1971 (see Touret and Hartel, 1988), this

contrast in ﬂuid inclusion composition between

Precambrian Rock Associations

645

amphibolite- and granulite-facies assemblages has been

found to be a general phenomenon. The CO

is specu-

lated to have come from the upper mantle, perhaps via

exsolution from crystallizing underplating basaltic

magmas. However, fracture-controlled inﬁltration of

to create an anhydrous mineral assemblage from

amphibolite is likely to be only local and not a wide-

spread, pervasive terrane-forming process because

-rich ﬂuids cannot migrate along grain boundaries

(Section 11.2.2).

There has been a persistent debate during the late

decades of the twentieth century, made more complic-

ated by the CO

issue, whether granulites in general

represent the crystalline residua remaining after extrac-

tion of granitic partial melts (Section 11.6; see also

Clemens et al., 1997). Although mineral assemblages of

many granulite-facies rocks satisfy a residual relation-

ship after extraction of granite partial melt, not all

granulites have the expected depletions in LIL and

LREE elements (Vielzeuf et al., 1988).

19.6.2 P T Paths and Tectonic Evolution of

Granulite-Facies Terranes

An additional debate revolves around the tectonic

setting and thermal history of granulites. Partly, the

controversy stems from their widespread occurrence

in Precambrian cratons whose tectonic character is

usually unknown with certainty but also because their

P–T paths that can provide important tectonic con-

straints are difﬁcult to accurately determine and many

published ones are suspect.

Accurate determination of complete P–T paths (Sec-

tion 18.1) in granulites is hindered and usually imposs-

ible because of the relatively rapid rates of diffusion

and, hence, of mineral equilibration at the relatively

high temperatures of the granulite facies (Frost and

Chacko, 1989; Selverstone and Chamberlain, 1990).

Commonly used ion-exchange mineral thermometers

have closure temperatures below those at which granu-

lites equilibrate. Thus, mineral thermobarometry can

provide no information on prograde paths and near-

peak P–T conditions, both of which are overprinted by

subsequent retrograde but still high-T equilibration.

The highest recorded T may not be the peak T of

metamorphism.

More reliable P–T paths can be constructed in some

instances using petrogenetic grids based on mineral

assemblages preserved in undeformed or otherwise

unmodiﬁed reaction textures, such as coronas (Figures

16.6 and 16.8) and inclusions in poikiloblasts.

Harley (1989) documented two general categories

of retrograde granulite paths—essentially isothermal

versus isobaric in Figure 19.36—that are presumed to

correspond to contrasting tectonothermal evolution-

ary regimes. The simplest explanation for the nearly

isothermal decompression paths is that they are retro-

grade segments of clockwise P–T paths created in

thickened orogens (Section 18.1). Such orogens com-

monly experience extension late in their evolution,

such as since the Miocene in the Altiplano–Puna in the

central Andes Mountains and Tibet Plateau north of

the Himalaya Mountains. Such extension is consistent

with the subhorizontal planar fabric of many granulite

terranes.

The lack of pronounced decompression in the

nearly isobaric retrograde paths implies that the peak

metamorphism occurred in continental crust of more

or less normal thickness, or at least one in which near

isostatic equilibrium prevailed so that very little sub-

sequent uplift and unrooﬁng took place. As one example,

retrograde corona reaction textures and chronologic

data indicate the Archean Napier Complex, Enderby

Land, in the East Antarctica shield cooled nearly

isobarically after attaining peak 900–1000°C and 8–

10 kbar at 3070 Ma. Independent evidence indicates

the stabilized craton experienced little uplift and ero-

sion for about 2 Gy while cooling only 300–400°C

(Figure 19.36; see also Ellis, 1987), thus precluding

the metamorphism being related to orogenic crustal

thickening.

The nature of the prograde path segment of these

paths has been far more controversial. Is it clockwise or

counter-clockwise and what is the responsible thermo-

tectonic regime? Critics of the clockwise-path option

have pointed out an absence of eclogite relics or of

kyanite inclusions in poikiloblasts that would indicate

a higher P prograde path. However, Selverstone and

Chamberlain (1990) note the existence of kyanite

inclusions in garnets in the Coast Ranges granulite

terrane in British Columbia. Together with other data,

they construct a clockwise path consistent with thermal

relaxation in a thickened orogen. They also argue that

the nearly isobaric retrograde path for this terrane, and

by implication others, calculated from garnet–biotite

and GASP thermobarometry (Section 16.11.3) is a spu-

rious result of high-T mineral homogenization. Advoc-

ates of counter-clockwise prograde paths point to the

existence of low-P andalusite and cordierite, either as

inclusions or relict grains.

Documentation of Counter-Clockwise P T Paths in

wo Terranes. The northwestern Superior Province

(Figure 19.8) harbors an unusually complete 20-km-

thick section of the Archean crust in which a zoned

greenschist- to amphibolite-facies terrane continues

without structural break into the Pikwitonei granulite

domain. The lower-grade boundary of the granulite

domain is simply an orthopyroxene isograd. Chrono-

logic data, textural observations, and thermobarometry

on zoned garnets and other phases have been integ-

rated to construct counter-clockwise P–T–t paths

for three parts of the amphibolite–granulite domain

646 Igneous and Metamorphic Petrology