Vous êtes sur la page 1sur 23

Mineral.

Deposita 30, 351-373 (1995)

MINERALIUM
DEPOSITA
9 Springer-Verlag 1995

Genesis of the Precambrian copper-rich Caraiba hypersthenite-norite


complex, Brazil
E.P. Oliveira 1, J. Tarney 2
lInstituto de Geociencias, Unicamp, 13081-970 Campinas, Brazil
2Department of Geology, University of Leicester, Leicester LE1 7RH, UK
Received: 24 November 1993/Accepted: 28 September 1994

Abstract. Caraiba, the largest Brazilian copper deposit


under exploitation, consists mostly of disseminated and
remobilised bornite and chalcopyrite hosted in early Proterozoic norite and hypersthenite. The mafic igneous complex comprises multiple intrusions of dykes, veins and
breccias of norites and hypersthenites, with minor proportions of amphibolised gabbronorite and peridotite
xenoliths transported by the m a g m a from deeper levels in
the lithosphere. The country rocks are high-grade
gneisses, granulites and metasediments. Compositions of
plagioclase (An6o- 40) and orthopyroxene (EnTo- 6o) fall in
a narrow range similar to the Koperberg Suite from the
Okiep copper district, South Africa, and to that in many
massif-type anorthosites. Whole-rock major and trace element geochemistry indicate a parental m a g m a enriched in
Fe, LREE, P, K, and Cu. Negative N b anomalies on
multi-element plots and fractionated REE patterns, along
with sulphide sulphur isotopes in the range 634S =
- 1.495 to + 0.643%o, suggest a primary mantle lithosphere source, although a lower crustal source for the
gabbronorite and peridotite xenoliths cannot be excluded.
Geochronological and field evidence indicate that both
norite and hypersthenite are likely to have been emplaced
during a major sinistral transcurrent (partly transpressional) shearing event associated with the waning stage
of evolution of the early Proterozoic Salvador-Curag~
orogen.

The Caraiba mafic-ultramafic complex, Bahia, is an


example of an intrusive hypersthenite-norite-anorthosite
igneous rock association emplaced into a Precambrian
high-grade terrain. Others are found in South Africa (the
Koperberg Suite; L o m b a a r d et al. 1986, Conradie and
Schoch 1986), the Kola Peninsula of Russia (Sal'nyyeTuadash area; Vinogradov 1976) and in East Greenland
(Ammassalik area; Moorlock et al. 1972). Some of these
hypersthenite-norite-anorthosite complexes are notable
for their rich copper mineralisation: the Brazilian, South
African and Russian examples contain over 1% copper
sulphides, with ore concentrated mostly in the ultramafic

rocks. However, the origin of these important bodies is still


imperfectly understood. Here we present mineralogical and
geochemical data for the Caraiba Complex and draw comparisons with the Koperberg Suite of South Africa in an
attempt to understand the petrogenesis of this unusual and
economically important rock association. Our results confirm that this mineralised hypersthenite-norite-anorthosite
association may have some genetic links with massif-type
anorthosites in general, as has been suggested for the South
African example. An origin through mobilisation of the
deeper parts of metasomatised lithospheric upper mantle or
re-melting of subcreted lower crustal rocks is most compatible with the data.

Geologic setting
The Caraiba mafic-ultramafic complex is located in the Curag~ihighgrade gneiss terrain which in turn is part of the Salvador-Curar
orogen, the northern extension of the Atlantic Coast Granulite Belt
in the S~o Francisco Craton (Figs. 1 and 2).
The early Proterozoic Atlantic Coast Granulite Belt along with the
Archaean Jequie Granulite Complex (Fig. 1) represents one of the
larger granulite complexes of the Precambrian cratons. Recent studies
of Barbosa (1986, 1990), Barbosa and Fonteilles (1989), Figueiredo
(1989) and Teixeira and Figueiredo (1991) have attempted to interpret
the evolution of the Atlantic Coast Belt in the light of modem plate
tectonics. The belt is characterised by a strong subvertical NE-SW
trending foliation, and can be divided into three domains:
1. The southeastern domain consists of granulite facies volcanic and
plutonic lithologies with chemical compositions similar to modern island arc rocks, including shoshonitic, calc-alkaline and
tholeiitic types, as well as Fe-Ti rich basalts: KzO contents
increase towards the northwest;
2. The northwestern domain comprises enderbitie to charnockitic
plutonic rocks intruded by gabbro-anorthosite complexes, and
granulite facies metavolcanic-metasedimentary rocks; the mafic
granulites have trace element compositions similar to oceanfloor or back-arc basalts;
3. The third domain, lying between the other two, consists of interleaved amphibolites and felsic amphibolite-facies gneisses.
The spatial distribution of these domains and their lithologies
strongly suggest that the Atlantic Coast Belt was a continental
margin magmatic arc with a west-dipping subduction zone below

352

Fig. 1. Simplified geology of the


$5.o Francisco Craton (adapted
from Inda and Barbosa 1978;
Mascarenhas et al. 1984)

the southeastern domain and a collision zone in between (Figueiredo


1989; Barbosa 1990). The geochronological data discussed by Barbosa (1990) and Marinho et al. (1993) suggested that collision might
have occurred about 2.4 Ga; the metamorphism reached its climax
about 2.0 Ga.
The Curaq/t high-grade gneiss terrain (Figs. 2 and 3) has been
mapped by Sa et al. (1982) and the staff of the Caraiba Mine (Hasui
et al. 1982; Sa and Reinhardt 1984; Silva 1985) who suggested that
the copper-hosting mafic-ultramafic rocks intruded a supracrustal
sequence now represented by banded gneisses (Surubim Gneiss),
graphite gneiss, banded iron formation, calc-silicate rocks and
alumina-rich gneisses (Born Despacho Banded Gneiss), and biotitehornblende-bearing quartz-feldspar gneiss with minor amphibolites
and quartzites (Arapua Banded Gneiss).
This supracrustal sequence suffered three main phases of deformation with associated granitic intrusions, metamorphism and migmatisation. The first, characterized by tight isoclinal folds (F1) was
responsible for the metamorphic banding, and was associated with
intrusion of the tonalitic orthogneiss G1, migmatisation and amphibolite facies metamorphism. The second, characterized by tight
isoclinal folds with E-W trending a x e s (F2) , was followed by the
intrusion of G 2 tonalites, mafic dykes and eventually granulite facies
metamorphism (P-T conditions, 5.5-6.5 kbar, 720-750~
Ackermand et al. 1987). The G~ and G z granite gneisses can sometimes be
distinguished by the presence of dome and basin interference patterns in the former. The third phase produced tight N-S folds (F3)
with sub-vertical axial planes and penetrative foliation over the
whole Cura~/t terrain, and was accompanied by G 3 granitoid sheets
and amphibolite- to granulite-facies metamorphism (Sa et al. 1982;
Silva 1985). Silva (1985) suggested that the Caraiba deposit has
a mushroom-like shape that resulted from the interference between
F 2 and F 3 folds. Later events include the establishment of a NE-

trending fracture system along which a swarm of dolerite dykes and


quartz veins were emplaced, and local retrogression to greenschist
facies. Other minor intrusions of peridotites (serpentinite, lherzolite,
wehrlite) found locally in the Cura~;fi terrain may be broadly contemporaneous with the copper-hosting mafic-ultramafic complex (Silva
1985).
At Caraiba, the supracrustal sequence and the G 1 and G 2 orthogneisses crop out mainly to the west and to the east of the ore-body,
respectively (Fig. 4). The contact between the orthogneisses and the
mineral-hosting mafic-ultramafic rocks may be sharp or gradational. Biotite schists are not uncommon at the sharp contacts.
Gradational ones are represented by banded mafic gneisses, possibly
the result of migmatisation during granite emplacement followed by
deformation. The supracrustals comprise banded gneiss, diopsidite,
forsterite marble, garnetiferous mafic granulite, minor biotite schist
and banded iron-formation. Anhydrite found in marbles and diopsidites led Leake et al. (1979) to propose an evaporitic environment
for this association.
The age of the mafic-ultramafic rocks at Caraiba is not tightly
constrained within the Palaeoproterozoic Curaqfi terrain. There are
U-Pb zircon dates of 2,328-2,235 Ma and 2,050 Ma respectively on
G1-G 2 tonalitic and G 3 granitic orthogneisses (Silva 1985). Thorpe
(pers. comm. in Silva 1985) obtained ages of 1.8 Ga and 2.02 Ga from
bornites. Oliveira (1990a) obtained a Sm-Nd mineral isochron of
1,890 ___60 Ma (eN~ = - 4 . 3 ) on a hypersthenite dyke from the
Caraiba mine; this age is very similar to that (1889 + 64 Ma) found
by Padilha and Melo (1991) for granites further south that are
associated with transcurrent faulting in the Salvador-Cura~/t orogen. The complex appears to be significantly older than its closest
analogue, the Koperberg Suite in South Africa (ca. 1050 Ma; Boer
et al. 1994); however, it is to be noted that model ages for the latter
are much older (1700 Ma TCHUR,2000 Ma TDM:Clifford et al. 1994)

353

Fig. 2. Simplified geology of


part of the Salvador-Curaq/t
Orogen showing the high-grade
terrains of Curaqfi (CT) and
Jacurici (.IT) and their mafic
ultramafic bodies. Geology after
Inda and Barbosa (1978), Gava
et al. (1983) and Mascarenhas et
al. (1984)

indicating that the age of development of the precursor sources of


the two complexes may not have been very different.

Models for the Caraiba mineralisation


The Caraiba mineralisation is basically of disseminated
type. The main ore minerals - chalcopyrite and bornite
- are hosted in centimetre to tens-of-metres thick hypersthenitic and noritic bodies enclosed within the highgrade gneisses. The hypersthenites are by far the most
Cu-rich rocks (Cu = 2-5 wt%; norites < 0.5 wt%). Magnetite, and to lesser extent ilmenite, are the main accessories in the disseminated mineralisation, their abundance
correlating positively with the main ore minerals, i.e., the

richer the ore the more magnetic it is (Sa and Reinhardt


1984). Veins and veinlets of remobilised ore are very
common, and consist mostly of chalcopyrite and bornite
with minor cubanite, magnetite and Ni-tellurides. Some
veins are richer in Ni and carry pentlandite, pyrrhotite,
mackinawite, chalcopyrite and cubanite. These parageneses, along with chlorite, epidote, carbonate, anthophyllite and talc, indicate that the vein sulphides have
been remobilised under amphibolite- to greenschist-facies
metamorphic conditions.
The origin of the Cu-bearing mafic-ultramafic rocks of
the Cura~/~ terrain has long been controversial. Leinz
(1948) and Schneider (1951) described the sulphide-bearing pyroxenites as magmatic. Ladeira and Brockes (1969)
ascribed the mafic-ultramafic rocks to an igneous suite

354
the sulphides. Mandetta (1982) later extended Lindenmayer's model to suggest that the Caraiba deposit is
a layered intrusion composed of cycles of igneous differentiation, with a bottom-to-top sequence of: (i) massive
hypersthenite with minor olivine-pyroxenite, (ii)melanorites, (iii) norites and leuconorites with hypersthenite
streaks, and (iv) norite to leuconorite with interleaved
gabbros to gabbronorites, locally banded.
Subsequent field and mineralogical studies on outcrops as well as drill-cores (Oliveira 1990a,b) failed to
confirm the layered intrusion model of Mandetta (1982).
Instead, crosscutting relationships suggested that the
Caraiba complex may be a multiple sequence of intrusions
of dykes, veins, and breccias. Here, field, mineralogical
and geochemical data are integrated to provide a better
petrogenetic model for the complex and for the mineralisation.

Field and petrographic aspects

Fig. 3. Simplifiedgeologyof part of the Cura~fihigh-gradeterrain


(modified from Delgado and Souza 1981, and the Caraiba Mine
stare

As with many igneous complexes in Precambrian highgrade terrains (cf. Windley et al. 1981), the Caraiba orebearing mafic-ultramafic rocks still preserve a good proportion of their primary igneous features. Cross-cutting
relationships with the country gneisses, and between the
different rock types within the complex, can be observed
in the Caraiba open pit. These clearly suggest emplacement of the complex as multiple intrusions of norite and
hypersthenite as dykes, veins and igneous breccias. The
associated peridotites and gabbroic rocks are interpreted
as xenoliths.
Hypersthenites and norites

representing the initial magmatism of a geosynclinal pile.


Suszczynski (1972) suggested that the mafic-ultramafic
rocks resulted from metamorphism of pure- to impurecarbonates, locally highly magnesian. Delgado and Souza
(1975) presented trace element data for some mafic-ultramafic rocks and concluded that the mean Cr, Ni, Ti, Cu,
Co, and V values found are more compatible with an
igneous origin for these rocks. Townend et al. (1980),
reviewing previous models for Caraiba, proposed an igneous origin, the available data favouring a sequence of
basic to ultrabasic sills metamorphosed at granulite grade.
Figueiredo (1981), noting the similarity between the REE
patterns of an iron-rich hypersthenite from the Caraiba
mine and some regional iron formations, suggested a sedimentary origin for parts of the Caraiba ore-body. Lindenmayer (1981) described relict cumulus textures in some
gabbros associated with the Cu-bearing norites-hypersthenites and found possible clinopyroxene exsolution
lamellae in hypersthene. On the basis of chemical data and
cross-cutting relationships deduced from drill cores, she
concluded that the ore-hosting rocks are intrusive and
evolved through fractional crystallization of a previously
differentiated Fe-Ti-rich tholeiitic liquid, and that the
hypersthenite-norite-gabbro-anorthosite sequence might
well represent the original igneous stratigraphy. She also
suggested that some anhydrite-bearing marble and calcsilicate rocks could have supplied sulphur to form part of

These rocks account for more than 94% of the Caraiba


mafic-ultramafic complex, and are found mostly as dykes,
veins, and magmatic breccias. Observations on
9 boreholes suggest the proportion of hypersthenite to
norite is ca. 60:40. The contact between these rock-types is
mostly sharp, but a complete gradation from hypersthenites to melanorites, norites, and to the less abundant
noritic anorthosites may occasionally be observed. Sometimes a homogeneous rock, either hypersthenite or norite,
can be traced for some tens of metres. Centimetre-thick
folded dykes of hypersthenite or norite may sporadically
be found cross-cutting the ore-grade rocks. Similar situations occur in the southern part of the open-pit where
off'shoots of coarse-grained hypersthenite clearly transect
the metamorphic banding of hornblende-pyroxene
granulite and migmatitic gneiss. The igneous breccias
comprise fragments of norite enveloped by hypersthenite.
This feature, and the offshoots described above, suggest
that some hypersthenites originated from liquids of the
same composition, rather than from another parental
magma.
Field observations on chalcopyrite-bearing hypersthenite dykes from the southern part of the Caraiba open
pit (Oliveira and Lacerda 1993) demonstrate that some
hypersthenites have been emplaced synkinematically during the most prominent deformational event (F3):

355

Fig. 4. Simplifiedgeology of the


Caraiba open-pit with location
of studied outcrops and
boreholes. Geology after Silva
(1985)

1. Drag folds in country-rock granulite-facies gneiss indicate that a steeply dipping hypersthenite dyke intruded
along a dextral transtensional shear zone (Fig. 5a);
2. A set of narrow and parallel melanorite to hypersthenite dykes cut across the steep foliation of the
migmatitic gneiss, whereas another set, conformable
interleaved with the gneiss, has been boudinaged and

broken apart, sometimes clearly showing a sigmoidal


shape indicative of a dextral sense of motion (Fig. 5b);
3. A cm-thick hypersthenite dyke, a branch of a thicker
and adjacent one, cross-cuts the migmatitic gneiss foliation and was subsequently transposed along the
gneiss foliation; the adjacent dyke is nearly conformable in the host gneiss and shows antithetic slip on

356
95
low T

--~
high T

85
X

O75
+

. o ~. + & . w-.,+~.r +
0 % ~ 0o+~" +6
+ +

65

45

Hypersthenite

Melanorite

35

+
o

2
3
Mole % AI203 In Opx

Norite
~

Fig. 6. Mg-number versus alumina content in orthopyroxenes from


Caraiba norites and hypersthenites. SK Skaergaard trend after
Hoover (1989). Also shown is the low- and high-T trend of the
Josephine peridotite after Dick (1977)

85

8O
x
O.

_c
7s

z
IE

70

\
65
Contact
Perldotlte

Fig. 5a--c. Cross-sections, Caraiba open-pit, showing evidence of


hypersthenite emplacement controlled by shearing in (a) and (b) and
concurrent dyking and deformation in (c). Grey hypersthenite; subparallel lines foliation of migmatitic gneiss

fracture planes, suggesting dextral displacement


(Fig. 5c). The transposed dyke contains hypersthene
transected by biotite, which is a typical amphibolitegrade F3 mineral.
The structural features, described by Oliveira and Lacerda
(1993) and summarised here, suggest that at least some of
the Caraiba rocks may have their emplacement controlled
by a regional shear belt that reached senility in the waning

Transition

Coarse
Hypersthenlte

Fig. 7. Variation of orthopyroxene composition across the contact


between a peridotite xenolith and the host hypersthenite. Sample
380-C from outcrop 380

stage of evolution of the Paleoproterozoic SalvadorCura~/t orogen.


The hypersthenites and norites are composed of hypersthene, plagioclase, sulphides, Fe-Ti oxides, phlogopite
and apatite in varying proportions. Under the microscope, hypersthene and plagioclase are very fresh, some
showing triple-junction contacts in response to subsolidus
recrystallization; but several primary features are still
preserved. Plagioclase and hypersthene in the norites are
in proportions consistent with eutectic crystallisation. In
the hypersthenites, however, hypersthene may often contain anhedral to subhedral inclusions of plagioclase in
optical continuity, implying that plagioclase may be both
a cumulus or a postcumulus phase; more rarely prismatic
hypersthene forms an apatite-hypersthene cumulate
or, together with plagioclase, shows cross-bedding like

357
igneous lamination. What may have been trapped or
interstitial noritic liquids are commonly found within the
hypersthenites; in these patches, hypersthene may show
subhedral cross-section and plagioclase is reversely zoned.
Minor sulphides, oxides and phlogopite grains also occur
as inclusions in hypersthene, though most are texturally
interstitial. They have recrystallised to different extents
during deformation/metamorphism so that intergranular
sulphides and oxides display all stages from having equilibrium textures with hypersthene ( = annealed contacts)
to having cross-cutting relationships (secondary growth).
Although phlogopite frequently shows secondary growth,
primary interstitial phlogopite around hypersthene can be
observed. Phlogopite is also found as a primary phase in
the hypersthenite vein (noted in (3) above) as well as in the
peridotite-hosted hypersthenites.
Apatite is a ubiquitous mineral, being granular in shape
in the hypersthenites, where it may reach over 30 modal%, and granular to prismatic in norites. Philpotts
(1967) noted that high concentrations of apatite and Fe-Ti
oxides commonly occur within anorthosite plutons and
explained the association in terms of immiscible liquids. In
some anorthosites the proportion oxide:apatite is 2:1
which Philpotts (1967) interpreted in terms of eutectic
crystallization. In the Caraiba anorthositic rocks, apatite
does not show such a direct relationship with Fe-Ti
oxides; nonetheless the petrogenesis of P +_ (Fe + T0-rich
liquids in massif anorthosites remains enigmatic.

90

80

70

9 Set 1 [

60

50
Edge

Centre

Edge

Fig. 8. Compositional variation in reverse zoned plagioclase of


norite 380-1D
Phlogopite
100

Eastonite

Phlogopite
8O

-H- +o
+

+~lr

& 60

+
.~.o

0~

o oSo

~.

Ultramafic xenoliths
Ultramafic xenoliths are locally common, cropping out as
fine-grained discontinuous rounded to oval shaped balls
in a groundmass of hypersthenite and norite; compositionally they comprise dunite, harzburgite, olivine orthopyroxenite, olivine-hornblende orthopyroxenite and olivine-orthopyroxene hornblendite with minor amount of
brown to black spinel and phlogopite. It is very difficult to
establish whether the xenoliths occur dominantly at one
stratigraphic level because they are absent in some
boreholes, even in those drilled along the same vertical
cross-section through the mafic-ultramafic complex. The
xenolith-groundmass contacts are sharp and, from petrographic evidence, the peridotites clearly crystallized before
the hypersthenites and norites. Detailed examination of
a contact between a peridotite xenolith and the groundmass hypersthenite shows a fine-grained hypersthenite
offshoot injected into the peridotite, a slight decrease of
the hypersthenite grain size towards the peridotite, and
elongated hypersthenes perpendicular to the contact with
the peridotite. This last remarkable feature is consistent
with growth of minerals at a cooled surface. Moreover, the
host hypersthenite is composed of phlogopite, apatite and
interstitial magnetite, bornite and chalcopyrite, which
leaves no doubt that the magma was hydrous and that
two immiscible liquids, viz. silicate and sulphide-oxide,
coexisted.
The peridotite orthopyroxene composition shows a gap
between 8%-18% En compared with pyroxenes from
hypersthenites and norites (Oliveira 1990b; Oliveira and
Tarney 1994), which again rules against deriving the latter

=
9

40

Biotite

20

+ Norltes
o Hypersthenites

0
0.44
Annite

0.46

0.48

0. 0
AI/(Mg+Fe)

0. 2

0.54

0.56
Siderophyllite

Fig. 9. Composition of biotite-phlogopite series in hypersthenites


and norites from the Caraiba mine

from the former by fractional crystallisation. Notably,


calcic pyroxene occurs in the most ffactionated peridotite,
but is absent in the Cu-rich hypersthenites.

Gabbroic xenoliths
Xenoliths of gabbronorites, gabbros, amphibole-gabbros,
and less often amphibolites, are common in the Caraiba
mafic-ultramafic rocks. Like the ultramafic xenoliths they
occur in cm-size rounded oval or elongate blocks. Because
some similar lithologies occur in the supracrustal sequence and/or as dykes in the orthogneisses G1 and G2 it
is potentially possible that a few xenoliths could be local.
The xenoliths are composed of plagioclase and Ca-pyroxene with varied proportions of amphibole, hypersthene,
phlogopite and opaques, and have granoblastic, more
rarely intergranular, texture. In some gabbroic xenoliths,
Ca-pyroxenes are partially to totally replaced by hornblende, which in turn may display symplectic intergrowths

51.10
0.03
3.08
22.28
0.46
21.48
0.14
.
98.57

%SIO2
TiO2
AlzO3
Cr/O3
FeO
MnO
MgO
CaO
Na20

4.001

0.704
0.015
1.209
0.006
-

Fe 2+
Mn
Mg
Ca
Na

0.6255
0.0029
63.2
-

Cr

Sum

1.93
0.001
0.137

Si
Ti
A1

XMg
Xca
# mg
%An

Oxygens

Cation proportions

Total

K20

FC2516
253.8
Hyps
Opx

Sample
Depth
Rock
Mineral

0.6355
0.0082
64.7

4.011

0.69
0.018
1.262
0.016
.

0.088

0.07

0.5563
0.0077
56.5

3.996

0.834
0.015
1.083
0.015
-

1.96

1.954

100.23

99.83

26.25
0.48
19.14
0.37

22.22
0.56
22.80
0.41
-

1.61

51.63
1.96

FC2516
304.94
Norite
Opx

52.63

NO33
54.13
Hyps
Opx

0.6101
0.0091
62.2
-

4.008

0.731
0.019
1.202
0.018
0.005

0.076

1.957

100.14

52.34
1.72
23.39
0.60
21.57
0.45
0.07

NO33
92.02
Norite
Opx

Table 1. Representative mineral compositions from the Caraiba complex

45.2

19.95

0.009
1.788
2.167

5.706

10.281

32

99.24

56.94
26.81
0.06
9.24
6.19

FC2516
253.8
Hyps
Plag

48.6

1.917
2.021
0.009
19.928

5.788

10.192

32

9.89
5.76
0.04
99.16

27.14

56.33

FC2516
269.2
Norite
Plag

56.0

19.903

2.186
1.718

6.087

9.913

32

99.36

11.26
4.89

28.50

54.71

NO33
54.13
Hyps
Plag

43.1

2.118
0.116
19.903

1.69

5.532

10.448

32

8.65
5.99
0.50
98.18

25.74

57.30

NO33
88.33
Norite
Plag

40.0

1.565
2.258
0.095
19.911

5.457

10.536

32

8.08
6.44
0.41
98.82

25.61

58.28

NO33
89.9
Anorth
Plag

72.70

65.70,

0.144
1.781
15.625

3.352

3.62

1.794
15.509

1.748

5.609
0.467
2.524

22
5.524
0.515
2.657
0.04
1.36

22

0.49
9.21
93.57

14.84

16.34

9.46
94.03

13.79

37.01
4.10
14.13

FC2516
282.7
Norite
Phlog

37.17
4.61
15.17
0.34
10.94

NO33
95.25
Hyps
Phlog

Or3

359

Table 2. Representative analyses of sulphide and oxides from hypersthenites and norites in Drillcores FC2516 and NO33, Caraiba mine
Mineral
Fe
Co
Ni
Cu
Zn
S
Total
Ni/Cu
Continued
TiO2
A12Oa
CrzO3
Fe203
FeO
MnO
MgO
Total

Cpyr
30.7
0.01
0.01
33.8
0.03
34.21
98.76

Pyrr
59.76
0.04
0.35
0.02
0.03
38.77
98.97

Born
12.64
0.0
0.02
60.07
0.02
26.7
99.44

Cpyr
26.41
0.06
0.0
39.59
0.02
32.57
98.65

Born
11.60
0.01
0.02
61.15
0.02
26.60
99.40

0.0

15.7

0.0

0.0

0.0

Magn
0.33
1.53
5.63
56.31
29.37
0.13
0.07
93.37

Spin
0.01
53.04
7.51
23.64
0.28
8.66
93.14

Magn
0.24
1.22
5.59
57.38
29.52
0.12
0.09
94.16

Spin
0.04
49.93
7.42
0.79
23.66
0.28
7.86
89.98

0.002
14.728
1.399

0.08
0,464
1.425
14.0
7.96
0.032
0.042

0.006
14.486
1.443
0.148
4.871
0.059
2.891

Cation proportions
Ti
A1
Cr
Fe 3+
Fe 2+
Mn
Mg

0.08
0.587
1.45
13.8
8.0
0.036
0.033

4.657
0.056
3.041

Cpyr, Chalcopyrite; Pyrr, pyrrhotite; Born, bornite; Magn, magnetite; Spin, spinel

of hypersthene and plagioclase at grain boundaries. This


formation of amphibole at the expense of pyroxene and
its further recrystallisation, suggest re-equilibrium of the
hydrated gabbroic xenolithic rock to the higher temperatures of the host noritic-hypersthenitic magma. The
gabbroic xenoliths have few counterparts in the surrounding country rocks (Oliveira and Tarney 1994), but their
distinct negative N b anomalies on mantle normalised
diagrams coupled with their fractionated REE patterns
suggest a continental influence/input.

Pyroxene
Orthopyroxene ranges from En68_54 in hypersthenites
through 1~n67.63 in melanorites to En66_56 in norites. In
the anorthosites there is very little variation in composition (En61-En62). Comparing samples from drillholes
FC2516 and NO33, orthopyroxenes from the former are
slightly more iron rich (En61.56 compared with En66_6o).
The alumina content of pyroxenes falls within the range
3.6-1.1 mole%, those from the hypersthenites being more
aluminous and less fractionated than those from the
melanorites and norites (Fig. 6). Pyroxenes from all rocks
follow a trend perpendicular to the trend observed in
low-pressure igneous intrusion (e.g. Skaergaard) but parallel to the trend defined by low-temperature - high-temperature Alpine peridotites and high-pressure peridotites.
The A1203 content of pyroxenes increases with increasing
temperature and pressure (Green and Ringwood 1967,
Green 1969), hence these parameters m a y have controlled
A1203 entering into orthopyroxene. However, the higher
AlzO3 abundances in pyroxenes from the hypersthenites
m a y be due to other factors. Both high total and water
pressure inhibit the crystallization of plagioclase (Yoder
and Tilley 1962), so if the hypersthenites represent higher
pressure magmas, then Ca and A1 would be available to
enter the pyroxene structure. However, the coexistence of
norite and hypersthenite m a g m a s makes pressure differences an unlikely explanation. Calcic pyroxenes occur in
none of these rocks and the Ca content of the analysed
orthopyroxenes is very low ( < 0.60 mole%). It appears
that the A1203 content of orthopyroxene is determined by
the presence or absence of plagioclase, coupled with
a relatively Al-rich, Ca-poor nature of the parental magmas.

Mineral chemistry
Mineral grains of norites and hypersthenites from
boreholes FC2516 and NO33 (see Fig. 4 for location) were
analysed with a J E O L Superprobe at Leicester University
to ascertain any compositional trends that might support,
or otherwise, the layered sills model of Lindenmayer
(1981) and Mandetta (1982). These two cores were chosen
because together they contain representatives of all rock
types. Extra samples from Outcrop 380 in the open-pit
(Fig. 4) helped clarify the plagioclase and pyroxene compositional relationships in the peridotite-bearing hypersthenite. Representative mineral analyses are given in
Table 1.

Fig. 10. Compositional variation of pyroxene and plagioclase from


the Caraiba complex compared with island-arc cumulates (Burns
1985), layered intrusions (Wager and Brown 1967), massif-type anorthosites (Emslie 1985) and the Koperberg Suite (Mclver et al. 1983;
Conradie and Schoch 1986)

360

Where hypersthenites are in intrusive contact with


peridotite, the orthopyroxene grain size decreases towards
the contact with the peridotite. Mineral analyses across
this contact (Fig. 7) show that the enstatite component of
the orthopyroxene increases from En67 in the host hypersthenite to En8 ~-82 close to the peridotite, indicating that
the hypersthenitic magma has reacted with the more Mgrich peridotite. These data, combined with the field and
petrographic observations, strongly suggest the peridotite
xenoliths are not cogenetic tectonically disrupted layers
within the hypersthenite, but instead must be allochthonous in nature. Moreover, the hypersthenitic magma
must have been hot enough to react with them. These
observations are difficult to reconcile with a model of
a layered complex.

Plagioclase
Plagioclase compositions range from An41-An61 in hypersthenites through An45-An58 in melanorites to An+2-Ans9
in norites (Table 1). The anorthosites have the most sodic
plagioclases (An+o_44). The plagioclases from the two drillholes were not significantly different.
The zoned plagioclases observed in the hypersthenites
and norites have a compositional range of Ansv-Ans7, but
with the highest An values in the rim, and the centres
being more sodic (Fig. 8). Reversely zoned plagioclases
have been produced experimentally at a single crystallization temperature and relatively high degree of supercooling (Lofgren 1974). Several natural occurrences of reversely zoned plagioclases have been described from maficultramafic plutonic rocks (e.g. Maaloe 1976; Berg 1980;
Mathison 1987), and have been interpreted in terms of
supercooling through heat loss to country rocks or abrupt
loss of volatile from the magma. It is unlikely the zoning
results from metamorphic or magmatic reaction between
pyroxenes and plagioclase (or a plagioclase liquid), or
a secondary effect of K-metasomatism, because coexisting
metamorphic pyroxenes would be Ca-rich diopsides and
not the Ca-poor orthopyroxenes observed. Supercooling
seems to be the most reasonable explanation, and is consistent with the mode of occurrence of the hypersthenites
and norites as veins, dykes and relatively narrow bodies,
that could create the sharp temperature gradients favourable to the formation of reversely-zoned plagioclases.

sometimes found as exsolution in magnetite. Representative analyses are given in Table 2.


Experiments carried out by MacLean and Shimazaki
(1976) on the partitioning of Zn, Fe, Co, Ni and Cu
between silicate and sulphide liquids have shown that Ni
and Cu partition most strongly into the sulphide liquid
with partition coefficients (Ko) of 150 and 50, respectively.
Fe has low KD (1.2), but as a major element it is readily
available to form sulphides. Zn, on the other hand, has the
lowest Kv (0.5) and is thus strongly partitioned into the
silicate liquid, eventually to reside in aluminous minerals.
These authors, along with Naldrett and Mason (1968),
have pointed out that the Ni/Cu ratio of ores mirrors that
of the host rock: the more ultrabasic the rock, the higher
the Ni/Cu ratio of the associated sulphides. From Tables
2 and 3, it is evident that Caraiba sulphides and wholerocks have low to very low Ni/Cu ratios, hence the hypersthenites and norites are more likely to have evolved from
a basic source rather than an ultrabasic one.

Mineralogic comparisons with other mafic-ultramafic


complexes
Plutonic mafic-ultramafic complexes in orogens are diverse in composition and origin. In Phanerozoic orogenic
belts they are generally classified into three types:
1. Alpine-type complexes consist of ophiolite associations
(Coleman 1977) that were formed either at mid-ocean
ridges, including ocean plateaus, or formed through
back-arc or fore-arc spreading in young oceanic arc
systems, and that were too hot or buoyant to subduct;
2. Alaskan-type (Irvine 1974) are thought to represent the
root of island-arc volcanics (Burns 1985), though some
may have an en-sialic back-arc origin (Khan et al.
1989), and
3. Minor bodies associated with Cordilleran bathotiths
(Mullah and Bussell 1977), which in general are dykes
genetically linked to continental margin calc-alkaline
magmatism.
30

25
O

Mica
O

2O

~ o "b

High AI

15

A majority of the micas (Fig. 9, Table 1), have Mg/Fe


ratios characteristic of phlogopite. Those from hypersthenites have mg numbers and A1/(Mg + Fe) ratio slightly
higher than their norite counterparts.

+
+

10~

5"

+
+
@o

04

380/365N

FC2516

N033

o~ %

egg +
+

High Fe

Non-silicate minerals
The most common non-silicates present in the Caraiba
hypersthenites and norites are chalcopyrite, bornite, pyrrhotite, magnetite, ilmenite and apatite. Hercynite is

0
0

12

18

24

30

FeO (wt %)
Fig. l l . Plot of A120 3 vs FeO for the Caraiba norites and hypersthenites, drillholes FC2516 and NO33, benches 380 and 356N.
Fields after Olson and Morse (1990)

361
10

12

Hypersthenites
o Norites

e
o

10

0
0

0%*o

0
o

0 0 0

8o o ~ 1 7 6 1 7 6
1 7 69 o
9

O0

0000 0
0

9 o
9

O~ w

eo" .

8
o

080

OO0

O0
9

, O0

O' I

.1

O0~O

@
0

ooO

to
0

e~e.

OoooOo

0 0

O@OO
9 9

.01

10
0

.O_2
I-

O0

0.3
0
0

90

O9

=E 0.2

0.1

9
0

0
0

IO

dl
1000

9 O ~Ia,

000 o

,m
O~

@0

o~o

O0

100

0 0
0

rp

,O 9

O0

O8
0

IO

oeoo
oO~

008
0
O0

.'

O~ 9

0 0

5
9

0 O0 O0 lu~-

99

0t -

9 00
o.
0

0 0~@

~"

, 0~I

0
0 O0

o%

o oa :i".
0.4

0
O0

~0 ~

~'i
,

O0

Ooo% O, o 9~
90

8~

0
0

0
0

0.0
0.2

'

'
0.3

'

'

'

0.4

0.5

'

'
0.6

'
0.7

mg*

0.3

0.4

0.5

10
0.6

0.7

mg*

Fig. 12a-f. Summaryof major and trace element variation versus rag-number(rag*) in hypersthenitesand norites from Caraiba (drillholes
FC2516 and NO33, and outcrops 380 and 365N)

All three types may be variably layered.


In Archaean high-grade metamorphic terranes, the
complexes are layered and can be grouped into a maficultramafic type and a gabbro-anorthosite type (Windley
et al. 1981). The geotectonic environment of formation of
these ancient complexes is uncertain, but various authors
have suggested similarity with island-arc (Windley et al.
1981), ocean floor (Weaver et al. 1981, 1982) or continental
margin (Srikantappa et al. 1984) environments.
In the Proterozoic, the majority of the mafic-ultramafic
complexes have been emplaced on to the cratons along
megafractures (Windley 1984). They are represented by
mafic dyke swarms, large layered tabular bodies such as

the Great Dyke of Zimbabwe, and layered complexes such


as the Bushveld in South Africa and the Niquelgmdia in
Brazil. An important type of plutonism characteristic of
the Mesoproterozoic is that represented by massif-type
anorthosites (Herz 1969, Emslie 1985, Ashwal 1988) which
may have been emplaced into the continental crust in
orogenic or anorogenic environments (Emslie 1985). Can
any links be drawn with Caraiba?
In Fig. 10, pairs of coexisting orthopyroxene and plagioclase from the Caraiba hypersthenites and norites are
compared with island-arc cumulates, layered intrusions
and anorthosites of the massif-type. Data for the South
African Koperberg Suite (Schoch and Conradie 1990), the

362
1500

1200
b

9 Hypersthenites
o Norites

1000

1000

o%

oo

oO

50O

O Oo~
Fe203
0

N N~

10

o ~ oo

~0 0

O0
J

00,

30

20

40

0(% ) ~II'

600 >

q,j

~00

8OO

400

200

Fe203

10

20

30

40
10000

4O
0
9

3O
O

O,I
O
M.

.%o.

~'~IE

1000

"

2O
o
o

10

o
o

oO

''~

)Ooo

100

o
o
. . . . . . . .

10

@@

0
0

0
O0

Cr

Cr
o

100

. . . . . . . .

1000

10000 10

. . . . . . . .

1O0

1000

.......

10
10000

Fig. 13a-d. Major and trace element variationin hypersthenitesand norites from the Caraiba mine
closest analogue of Caraiba, are also plotted for comparison. Most of the Caraiba rocks plot on the fields for
massif-type anorthosites and the Koperberg Suite. The
orthopyroxene compositions of the Caraiba samples varies less than the plagioclase but no marked positive
correlation is seen like that commonly observed in layered
intrusions. The mineral chemistry of the Caraiba hypersthenites and norites then is similar to that observed in
massif-type anorthosites, and unlike that seen in islandarc cumulates and tholeiitic layered intrusions. Analogy
between the Koperberg Suite and massif-type anorthosites has already been drawn (Conradie and Schoch
1986), hence it is useful to test whether models for the
petrogenesis of massif-type anorthosites can be applied to
the Caraiba rocks. First we examine the constraints from
whole rock geochemistry.
Geochemistry
Major, trace and rare-earth element analyses of representative hypersthenites, melanorites and norites from drillholes FC2516 and NO33, and outcrops on benches 380,
365N and 395S (see Fig. 4 for location) are given in Tables
3a and 3b. It is evident that the typical hypersthenite
or norite rocks from the different localities within the
complex are not significantly different from each other,

and that the chemical variations observed are mostly


controlled by the relative proportions of contained minerals. The hypersthenites have higher Fe203, MgO, Cr,
Cu, Mn and Ni, and less A1203, CaO, Na20, Sr and Zr
than the norites. Also, the former are generally more
enriched in Rb, K20 , Ba, Zn and V, and more depleted in
Nb, TiO2, rare earth elements (REE) and P205 than the
latter. Hypersthenites and norites from FC2516 are slightly richer in MgO, Cr, Ni, V, Rb, Ba and K20 than the
equivalent rocks from NO33, reflecting the mineral chemistry.
The calculated CIPW norms (using Fe203/FeO = 0.15)
of all rocks show very low contents of diopside. Indeed,
the hypersthenites and norites contain significant proportions of normative corundum, indicating saturation with
respect to A1203, and further implying derivation from an
Al-rich magma, as also suggested by the pyroxene compositions. The Al-rich characteristic of the Caraiba rocks
is further shown by a plot of A1203 vs tFeO (Fig. 11),
which also emphasises their Fe-rich nature. Plots of ragnumber [MgO/(FeO + MgO) using FezO3/FeO = 0.15]
against major and trace elements (Fig. 12), and the positive correlation between element pairs Zn-Fe, V-Fe, Cr-Fe
and Cr-Ni (Fig. 13) reveal that Ni, Cr, MnO, Zn, V and
Fe203 decrease with fractionation, whereas Nb, TiO2,
P205 and CaO increase.
Good positive correlation (not illustrated here) is

363

40

1000

* FC2516-Hypersthenites I
9 NO33-Hypersthenites
o FC2516-Norites
]
o NO33-Norites
J

30

b
e

800

O0
9

.,,..,..,<>=+: + o:
20

0
0

10

O0
0
0

.~

0
,I

9 ~

o*o"00o
,01,

200

..

t.
S

o+>.~ .

,+

**r162e

o+O. d,.

'

1000

o0

Q 0
0

O0
0

9M

tO0

<~
0

"

""

tO0

~o

+,,41'

"'~0

*s
1000

~'~00

*+>*

@0

>
400

600

0
@

O0

0~0

10

,I

,I

i i ''I

I0

f
@@

.~0

o+:
0

O0

0$

o41,
9

O9

e9
9

O
0

0
O

9
e~ 9

.1

,=
E

9149

+I.

0.6

o.

o .o
.

0
0

0
0
0

9
0

0.5

"g
4

0
~0

0.4

<3

.01

......

IO0

.......

1000

10000

. . . . . . . . . . . . .

1 O0

CU

......

1000

10000

......
0.3
100000

Cu

Fig. 14a-f. Variation of rag-number (mg*), major and trace elements in relation to copper concentration in samples from drillholes FC2516
and NO33

observed between element pairs Nb-Ti, A1-Ti, K-Ba, RbBa, Mg-Fe, K-Rb, Ca-Sr and to less extent also Ti-P. Ti
correlates negatively with V and Zn. All these observations indicate that Nb is mainly held in ilmenite and
possibly also in the brown mica (see high Nb value in
biotitite 395S-B2 in Table 3b). V, Cr and Zn are mostly
partitioned into magnetite, K, Rb and Ba into brown mica,
and Sr into plagioclase (Ba also in norite plagioclases).
The increase of Ti with increasing A1 also suggests that the
amount of ilmenite increases in the most fractionated
rocks, i.e. the norites. The negative correlation between Ti
and V and Zn, on the other hand, indicates that magnetite
is an early fractionating phase along with hypersthene and
sulphides. This agrees very well with the high modal

proportions of sulphides and magnetite in hypersthenites.


Focusing on the behaviour of copper in the hypersthenites
and norites, plots of Cu vs Fe, Ni, Cr, V, K and rag*
(Fig. 14) reveal the following:
1. The highest Cu contents are found in the hypersthenites as expected from field and petrographic evidence;
2. The Cu vs mg* and Cu vs Fe203 plots show the NO33
hypersthenites are geochemically slightly more evolved
and more Cu-rich than their FC2516 counterparts.
This is in agreement with the enhanced sulphur solubility with increasing Fe content in mafic magmas, as
determined experimentally (Haughton et al. 1974).

48.5
0.39
4.9
21.3
0.35
20.0
0.9
0.9
0.61
0.00
97.84

FC2516
260.7
Hyps

47.4
0.35
4.1
22.2
0.35
20.7
0.6
0.1
0.86
0.13
96.78

FC2516
325.3
Hyps

298
1599
912
12390
414
22
23
52
10
63
1.1
314
< 2
4.8
0.6
17.2
0.07

351
1999
1354
18873
478
26
41
11
16
29
3.5
181
19,0
54.0
20.4
20.3
0.07

La
Ce
Pr
Nd
Sm
Eu
Gd
Dy
Er
Yb
Lu
(La/Yb)N

4.06
7.4
0.80
3.22
0.87
0.22
0.77
0.93
0.72
0.99
0.16
2.8

31.4
62.8
7.12
24.3
4.42
0.25
3.45
2.24
1.19
1.21
0.18
17.5

Rare earth elements (determined by ICP-ES)

V
Cr
Ni
Cu
Zn
Ga
Rb
Sr
Y
Zr
Nb
Ba
La
Ce
Nd
Th
Ni/Cu

Trace elements (ppm) determined by XRF

SiO2
TiO2
A120 3
Fe203
MnO
MgO
CaO
NazO
K20
PzO 5
Total

Core
No.
Type

10.2
24.9
3.28
12.8
2.96
0.61
2.48
2
1.06
1.39
0.25
5.0

683
2539
927
47859
688
29
15
76
14
42
0,8
169
4.3
20.3
8.5
4.8
0,02

37.2
0.87
4.5
32.7
0.40
14.6
1.4
0.5
0.50
0.24
92.87

NO33
88.0
Hyps

122.0
317.0
44.5
183.0
42.6
5.2
36.8
21.9
9.2
6.44
0.86
12.8

115
409
348
4827
373
16
6
76
112
13
3
53
113.6
329
204
36
0.07

43.7
0.14
2.6
19.6
0.43
17.6
7.8
0.2
0.16
5.29
97.41

NO33
92.29
Hyps
44.6
0.41
4.4
22.7
0.37
18.6
1.6
0.3
0.31
0.28
93.57

365N
7B
Hyps

4.91
12.3
1.76
7.51
2.03
0.32
1.92
1.53
0.92
1.11
0.18
3.0

1.5
0.03

364
2105
1316
44600
474
25
14
39
13
13
1.5
84
12.8
8.1

Table 3a. Representative analyses of hypersthenites (Hyps) and norites (Norit)

2.71
6.74
1.01
3.90
1.32
0.18
1.02
0.90
0.63
0.81
0.14
2.3

1194
6187
2044
105600
1235
37
3.6
18
10
26
3.7
41
9.7
6.3
0.4
3.6
0.02

29
2.11
2,6
39.7
0.38
13.3
0.9
< 0.2
0.14
0.16
88.29

365N
11.2
Hyps

3.98
9.11
1.25
5.57
1.66
0.38
1.62
1.58
1.13
1.14
0.19
2.4

455
1599
685
9161
316
27
94
35
11
2O
2.1
319
1.6
3.1
0.4
0.3
0.08

44.5
0.79
7.8
20.1
0.29
21.0
1,9
0.3
1.66
0.08
98.42

380
1A
Hyps

16.3
38.7
5.42
23.1
4.99
0.63
4.39
2.67
1.42
1.25
0.19
8.8

399
2414
1256
25096
427
22
48
15
17
2O
1.1
684
13
39
20
1.7
0.05

43.7
0.88
3.6
24.0
0.35
19.5
1.5
0.1
1.12
0.67
95.42

380
1B
Hyps

2.77
5.27
0.67
2.17
0.60
0.09
0.68
0.79
0.55
0.88
0.15
2.1

2.3
0.19

290
2494
2781
14574
454
26
14
17
8
25
1.2
158
5

47.2
0.24
3.9
23.0
0.37
21.4
0.4
0.2
0.36
0.02
97.89

395S
C
Hyps

10.5
24.7
3.35
13.1
3.22
0.84
2.87
2.38
1.58
1.42
0.22
5.0

361
3133
1567
20291
489
27
4
115
12
8
2.6
53
6
24.4
14.7
1.5
0.08

46.2
0.61
6.3
22.0
0.33
17.0
2.3
0.8
0.27
0.32
96.98

FC2516
265.2
Norit

31.9
67.7
9.02
38.2
7.92
2.53
6.42
3.62
1.52
0.99
0.14
21.8

173
29
34
504
120
29
12
790
17
269
5.3
170
31
68
42
5
0.07

49.4
1.13
20.7
10.3
0.11
4.0
8.2
4.0
0.51
0.54
98.96

FC2516
269.2
Norit

,~'

49.0
1.12
18.3
11.6
0.14
9.6
5.6
2.5
2.19
0.16
100.0

SiOz
TiO2
A1203
FezO3
MnO
MgO
CaO
Na20
K20
P205
Total

42.8
1.86
11.1
23.8
0.26
11.5
2.7
1.4
1.24
0.01
96.7

NO33
62
Norit

7.5
14.6
1.8
6.8
1.45
0.78
1.3
0.93
0.66
0.65
0.11
7,7

Rare earth elements (ppm) determined by ICP-ES


La
17.8
12
34.3
Ce
36.4
17.3
73.9
Pr
4.26
1.67
9.59
Nd
14.4
5.8
38.3
Sm
2.67
0.99
8
Eu
1.19
0.82
2.12
Gd
2.02
0,83
6.69
Dy
1.18
0.87
4.1
Er
0.65
0.55
1.86
Yb
0.68
0.81
1.55
Lu
0.14
0.13
0.22
(La/Yb)N
17.7
10.0
14.9

770
163
327
6163
311
42
44
245
9
58
6.4
614
9.2
15.1
2.9
3.8
0.05

48.9
1.09
16.5
13.3
0.24
11.6
7.5
1.5
0.11
0.17
100.9

365N
6A
Norit

221
331
211
1515
169
25
2
252
7
13
3.2
29
9.6
7.6
4.5
< 2
0.13

280
292
227
505
160
29
119
303
9
46
7.1
869
17.8
32.7
11.2
5.4
0.4

47.7
1.27
19.2
13.5
0.15
5.3
7.8
3.3
0.75
0.47
99.4

NO33
101
Norit

272
14
41
2133
135
34
20
556
24
29
5.3
275
30.5
72.9
40.7
0.2
0.01

V
Cr
Ni
Cu
Zn
Ga
Rb
Sr
Y
Zr
Nb
Ba
La
Ce
Nd
Th
Ni/Cu

Trace elements (ppm) determined by XRF

FC2516
315
Norit

Core
No
Type

21.0
46.5
6.12
25.2
5.36
1.79
4.61
2.6
1.19
0.84
0.12
16.9

95
267
232
8632
116
29
5
520
14
109
1
87
24.3
51.1
28.1
6.6
0.02

49.9
0.22
21.6
8.0
0.14
6.0
9.4
3.0
0.21
0.58
99.1

365N
8A
Norit

71.8
161
21.22
85.3
17.16
3.03
13.8
7.21
2.91
1.79
0.24
27.1

272
64
235
3430
133
29
3
474
30
30
8.2
i00
60.3
138.6
79.1
4.8
0.06

47.1
1.76
18.3
11.8
0.14
6.5
9.5
2,8
0.18
1.56
99.6

380
2A
Norit

17.0
35.2
4.37
17.5
3.66
1.03
3.11
2.01
1.13
0.95
0.15
12.1

349
714
326
5644
246
28
13
268
13
29
5.4
178
15.7
26.7
15.4
3.5
0.05

45.4
1.49
13.3
18.9
0.25
11.9
6.2
1.1
0.35
0.32
99.2

380
2C
Norit

23.1
51.0
6.58
26.7
5.71
1.18
4.87
2.96
1.35
1.21
0.18
12.9

369
1158
548
15337
241
26
11
240
16
33
6.8
111
20.1
50.3
26.3
1.5
0.03

45.6
1.29
11
19.2
0.27
13.2
5.0
1.2
0.34
0.61
97.7

380
2D
Norit

Table 3b. Representative analyses: norites (Norit) anorthosites (Ano) peridotite (Peri) amphibolite (Amph) and biotitite (Biott)

37.1
78.6
9.93
40.1
8.65
2.59
7.28
4.33
1.93
1.25
0.16
20.1

25
35
38
6
72
25
5
696
2O
<5
1.2
274
34.3
80.5
41.5
3.9
5.8

51.9
0.04
21.1
4.4
0.1
3.3
8.6
5.0
0.66
1.02
96.2

NO33
91
Ano

3.4
9.5
1.28
4.8
1.27
0.2
1.23
1.2
0.81
0.71
0.11
3.3

74
4357
1784
1315
270
10
10
10
8
51
0.4
24
6
8
5
0.4

42.5
0.32
3.3
16.3
0.17
34.1
1.4
0.2
0.25
0.04
98.6

380
3G
Peri

8.6
23.6
3.53
15.7
3.97
1.17
3.52
2.73
1.52
1.6
0.24
3.7

5.4
11.2
1.45
5.3
0.87
0.19
0.67
0.55
0.26
0.24
0.04
15.3

2.7
0.2

277
39
420
46
6
11
21
1921
6

747
1589
915

40.2
3.18
13.8
17.2
0.12
17.0
0.6
0.5
7,52
0.03
100.1

47.9
0.61
15.6
10.2
0.16
10.2
13.3
1.3
0.41
0.06
99.8

260
568
138
4174
81
19
4
153
17
17
2.5
37
12
19
17
0.3
0.1

395S
B2
Biott

365N
B
Amph

366
F

/ ~

o Norites
9 Hypersthenites

Fig. 15. AFM diagram for hypersthenites and norites of the Caraiba
complex. The peridotites are shown for comparison. Fields after
Kuno (1968) and Irvine and Baragar (1971)

It had been suggested that the Caraiba complex has the


major element chemistry of tholeiitic suites (Lindenmayer
1981). Fig. 15 shows the AFM diagram for hypersthenites
and norites of Caraiba. The rocks clearly plot along
a trend of relatively constant Fe/Mg ratio and increasing
total alkalis, a characteristic of the calc-alkaline series.
The tholeiitic trend obtained by Lindenmayer (1981) results from the over-emphasis given to peridotites and
gabbroic rocks. Although some hypersthenites and norites
have cumulate textures, which might artificially extend
supposed 'liquid' trends, it is important to note that it
would be difficult to explain the dykes and veins of norites
and hypersthenites, and the mobility of the hypersthenite
groundmass to the breccias, with a liquid component.
Current hypotheses cited to explain the origin of calcalkaline suites (or trends), include magnetite fractionation
(Osborn 1959; Gill 1981) and combined assimilation with
fractional crystallization (AFC: DePaolo 1981, Grove and
Baker 1984), while Cawthorn and O'Hara (1976) suggested that early fractionation of amphibole can generate
calc-alkaline trends. The latter seems not to be applicable
to the Caraiba norites-hypersthenites as no primary amphibole has been observed. The relations between Fe, Ti,
V, and Zn suggest that fractionation involving magnetite
(plus sulphide) could account for the calc-alkaline trend
observed on the AFM diagram. However, the Caraiba
complex is not a normal calc-alkaline suite because
the late 'differentiates' are norites and anorthosites, not
granites.
The mantle normalized multi-element diagrams for the
average compositions of hypersthenite and norites and
various selected samples (Fig. 16) emphasise the strong
mineralogical control on the whole-rock chemistry. Thus,
those hypersthenites and norites with high modal abundances of apatite generally contain high contents of REE,
Y and P (e.g. sample NO33-92.29 in Table 3a and
Fig. 16b, d) confirming that apatite concentrates most of
the REE. Similarly, high modal phlogopitic mica corre-

lates with high contents of Rb, Ba, Ti, Nb and K (e.g.


sample 395S-B2 in Table 3b and Fig. 16b). The coarsegrained and intrusive hypersthenite of bench 395S (analyses 395S and 395S-C in Figs. 16a, b) has one of the
lowest modal apatite contents and consequently has low
P and REE abundances on mantle normalized diagrams.
Conversely, it has up to 3% primary phlogopite, reflected
by the high Rb, Ba and K on these diagrams.
In Fig. 16, the plagioclase-poor hypersthenites have
a marked negative Sr anomaly whereas the plagioclaserich norites do not. Although Sr partitions strongly into
plagioclase (Philpotts and Schnetzler 1970), it is not easy
to account for the negative Sr anomaly in the hypersthenites unless plagioclase had been involved in their
petrogenesis, for instance through earlier removal of a plagioclase-rich melt from the hypersthenite source. Both the
hypersthenites and, to a lesser extent, the norites show
marked depletion in Nb relative to the light rare earth
(LREE) and low-field strength elements (LFSE). Such
negative Nb anomalies are characteristic of island-arc
lavas, many continental flood basalts and of the continental crust itself(Saunders et al. 1980; Thompson et al. 1983;
Weaver and Tarney 1983,1984). Proterozoic dolerite
dykes commonly show this anomaly (Weaver and Tarney
1983), as do Phanerozoic flood basalts thought to be
derived from the subcontinental lithospheric mantle, such
as the Paran~ from southern Brazil (Mantovani et al.
1985, Hawkesworth et al. 1986). It is reasonable to suspect
then that the Caraiba ultramafic/mafic hypersthenites and
norites with this geochemical signature could also be of
sub-continental lithosphere derivation. Rare earth elements (REE) can be useful petrogenetic indicators in this
regard.
The REE patterns of hypersthenites and norites are
quite fractionated (Fig. 17, Table 3a, b), but variable, with
(La/Yb)N ratios ranging from 2 to 13 in hypersthenites and
from 7 to 27 in norites. The hypersthenites with the
highest and lowest REE abundances are, respectively, the
apatite-rich samples from borehole NO33 (sample 92.29)
and the coarse-grained and intrusive pyroxenite from
bench 395S in the open-pit (sample 395S-C). The hypersthenite dished REE patterns with consistent negative Eu
anomalies (Fig. 17c) would be consistent with removal of
phases such as plagioclase and apatite which normally
concentrate Eu + LREE and MREE respectively. The
norites (Fig. 17a) are more variable regarding the Eu
anomaly, but it is clear that removal of plagioclase from
either the source (i.e. earlier melting episode, as noted
above) or a magma has played a significant role in the
evolution of the Caraiba rocks.
Interesting comparisons can be made with the Koperberg Suite in South Africa, and massif-type anorthosites
(Fig. 17b, d). REE patterns for the Caraiba and the Koperberg Suite rocks are much more fractionated than for the
massif-types, especially with respect to the middle- to
light-REE. While this may be because the source for the
Caraiba and Koperberg rocks contained residual phases
with low distribution coefficient for these elements, a more
likely explanation is that the Caraiba/Koperberg source
had suffered metasomatism that had left it enriched
in more easily fusible LREE- to MREE-rich mineral
phases such as mica and amphibole +_%apatite. The lack of

367

500
[

Averages of Hypersthenites

200

Representative Hypersthenite- Biotitite

100

==

50

20
m

g:

s
2
1
0.5

395S
--

0.2

365N

380

z~ 395S-B2
o 395S-C
o NO33-92.29

9 NO33
t~ FC2516

0.1

Averages of Norites

Representative Norites

200
100
50

20

lO
g:
2
1
9

0.5

0.2

13

~5N

9 380

--

o N033
0.1

N033-93.26

c= FC2516-265.20

[] FC2516
t

Rb Ba K Nb La C e S r Nd P

Zr Ti

Rb Ba K N b L a Ce Sr Nd P Zr Ti

Fig. 16a-d. Mantle normalised multi-element diagrams of Caraiba Complex rocks: a average compositions of hypersthenites from different
localities, b selected hypersthenites and one biotitite, e averages of norites, and d selected norites. Normalising values from Sun and
McDonough (1989)

368
I

500
200

Caraiba Nodtes

9 FC2516-NO33
o 380-365N

ii

b ~
e.~

Range of
9 Koperberg Diodtes
o Koperberg Anorthosites
~ Adirondack Anorthosites

",..=
~

~__~o~ ~RogalandAnorthosites

100
5O

lo

2
1

o+

,.,,

,~.+,,,.,,o~
_

100

~ ,+
s,+++

I ,,,.t,,

o Koperberg Hypersthenite

9 FC2516-NO33

9 Adirondack Hypersthenite

50

lO

s
2
1

0.5

! i i i
LaCePrNd

I , ~
SmEuGd

I
Dy

,I
Er

I I
I
i + I
Yb Lu L a C e P r N d

,i,,,J

SmEuGd

Dy

Er

Yb Lu

Fig. 17a-d. Chondrite-normalised rare earth diagram for the Caraiba complex, Koperberg Suite (Conradie and Schoch 1988)and massif-type
anorthosites (Simmons and Hanson 1978; Demaiffe and Herlogen 1981)

369
len et al. (1990). A more recent study of the Koperberg
Suite sulphides by Boer et al. (1994) has shown a weak
negative correlation of mean 034S with mean Cu/S ratios
which is interpreted in terms of sulphur devolatilisation
during the high-grade metamorphism. They also recorded
modest 180 enrichment in the mafic host rocks (range in
61SOrock = + 5.9 to + 8.3%0, compared with a 'normal
mantle' value of ca. + 5.7 ___0.3%0), which was taken to
indicate some crustal contamination of mafic magmas.
The correspondence between Caraiba and the Koperberg
suite is close, despite the ca. 1000 Ma difference in age;
Boer et al. (1994) also favour mantle derivation. Crustal
geochemical or isotopic signatures are present in both
suites, and the question remains as to what processes are
responsible for this.

20

Norilsk

Duluth

15

10

(3.

03
~o3

"~
..Q

-5

-10

Fig. 18. Sulphur isotope data for mafic complexes and the upper
mantle in comparisonwith the Caraiba sulphides(data after Kyser
1986; Ohmoto 1986; Von Gehlen et al. 1990)

Discussion
Petrogenetic models for the Caraiba complex are constrained by the following observations:

significant HREE depletion suggests that garnet was not


a residual phase during melting. The Koperberg rocks
have higher abundances of REE but both suites exhibit
similar patterns.

Sulphur isotopes
Sulphur isotopic compositions of the Caraiba sulphide
minerals were used by Oliveira and Choudhuri (1993) to
evaluate the crustal contribution in the genesis of the
sulphides (cf. Lindenmayer 1981). Fourteen samples of
remobilised and disseminated chalcopyrite or chalcopyrite + bornite from the Caraiba open-pit were analysed for sulphur isotopes at the University of Calgary,
Canada. They show 634S in the range - 1.495 to + 0.643
with an average value of - 0.604.
Because the isotopic fractionation factors among all
sulphide species are probably within 0.5%o, magmas formed by partial melting of the mantle or rocks crystallized
from such magmas should have 6348 values similar to
those of the parental mantle material (Ohmoto 1986).
Sulphur derived from mantle has an isotopic composition
very similar to meteorite sulphur and it is assumed that
the upper mantle as a whole has 634S = 0 _ 3%0 and
the primitive upper mantle an average of + 0.5%0 (e.g.
Ohmoto 1986; Chaussidon et al. 1989). However, many
mantle derived mafic igneous rocks have 6a*s outside this
range (Ohmoto 1986, and references therein), such as the
Duluth complex (0 to + 17%o,Noril'sk intrusive ( + 6 to
+ 16%o), and the Bushveld complex ( - 9 to -6%o).
Although the mantle might be slightly heterogeneous
(Chaussidon et al. 1989), assimilation of crustal sulphur
has been evoked to explain such anomalous values (e.g.
Naldrett 1981; Ohmoto 1986).
Figure 18 shows the Caraiba sulphur isotope data fall in
the range predicted for the mantle and appear not to have
inherited significant amounts of heavy 3% from supracrustal rocks. Moreover, the sulphur isotope ratios from
Caraiba are very similar to sulphide sulphur from the
Okiep copper district, South Africa reported by Von Geh-

1. Field evidence indicates that the complex consists of


multiple dyke-like intrusions and igneous breccias of
hypersthenite, norite and minor anorthosite with
peridotite and gabbro xenoliths, and that it is younger
than most high-grade country-rocks: supracrustals,
banded gneisses, migmatitic gneiss, tonalitic orthogneisses, garnet- and pyroxene granulites. It may be
coeval with the youngest granitic to syenitic sheets.
Several hypersthenites and norites are likely to have
been liquids judging from their mode of occurrence.
2. The peridotitic and gabbroic xenoliths have few
counterparts in the surrounding country rocks and are
not linked to the hypersthenites and norites by fractional crystallization, though may conceivably be related to them by some other process/processes at some
earlier stage of evolution of the complex;
3. The similarities with massif-type anorthosites suggests
a common petrogenesis.
4. The hypersthenites and norites display a major element
calc-alkaline signature and may be linked theoretically
by fractionation processes involving plagioclase, orthopyroxene and magnetite ( + sulphide). Local crustal contamination seems not to have been important.
Field, petrographic and geochemical evidence indicates that there have been several episodes of magma
injection, each having different degrees of enrichment
(e.g. hypersthenites associated with the anorthosites
are richer in incompatible trace elements, Fe and Cu);
5. The primary magma(s) must have had a basic composition, low calcium content, high Fe/Mg ratio and relatively high alumina; also hydrous and enriched in
Ti, Cu, S, K, REE and P to account for the presence of
primary copper sulphides, ilmenite, phlogopite and
apatite, sometimes in large modal proportion. The
source must have been enriched in incompatible
elements, most likely the subcontinental mantle or
lower crust as suggested by the fractionated REE
(LaN/YbN = 2-27), and by negative Nb anomalies on
mantle-normalized diagrams;

370
A basic magma impoverished in calcium must necessarily
be derived from a Ca-depeleted mantle source having
few or no calcium-bearing minerals such as garnet,
amphibole or clinopyroxene. Harzburgite or orthopyroxenite from the sub-continental lithosphere would be
compatible with this requirement. The source must be
heterogeneous in order to explain the variable incompatible trace element ratios found in norites and hypersthenites. Minor modal proportions of phlogopite, apatire, Fe-Ti oxides and possibly also amphibole in that
source may explain the geochemical characteristic of the
Caraiba rocks. Lithosphere veined by small degree melts
has commonly been proposed to explain this (Tarney et al.
1980; Foley 1992).
It is useful to examine briefly models proposed for
complexes similar to Caraiba, especially close analogues
such as the Koperberg Suite in South Africa. An origin by
partial melting of prexisting granulite facies gneisses or
lower crustal materials has been suggested for hypersthenites, norites and anorthosites of the Koperberg Suite
(Clifford et al. 1975, 1990, 1994; Stumpfl et al. 1976), from
the Sal'nyye-Tuadash area in the Kola Peninsula of Russia (Vinogradov 1976), and the Ammassalik area in East
Greenland (Moorlock et al. 1972). It has also been suggested that the Koperberg Suite may represent the deep levels
of a massif-type anorthosite (Conradie and Schoch 1986).
The similarity with Cariba in rock compositions might
support this model, but the absence of large scale anorthosite batholiths in nearby areas weakens support for
this model, unless appeal is made to difference in level of
erosion.
Current models for the genesis of massif-type
anorthosites involve derivation from a parent basaltic magma which may have resulted from the following
processes:
1. Melting of basic granulites of the lower crust or underplated mafic material (Simmons and Hanson 1978;
Taylor et al. 1984; Czamanske and Bohlen 1990);
2. Crystal-liquid fractionation from more primitive
mantle-derived magmas whose ultimate source was
at greater depths (Emslie 1985), or from an unusually
Fe-rich mantle low in clinopyroxene (Morse 1982;
Olsen and Morse 1990);
3. Mixtures between a depleted "mantle-like" component
and a crustal component (Ashwal et al. 1986, Menuge
1988, Frost et al. 1989).
On the basis of the relative high K, REE and P abundances in the Koperberg Suite, compared with anorthosites
of the massif-type, Anderoli et al. (1988) suggested that the
precursor of the Koperberg Suite could be an enriched
basalt which was collectively named " K R E E P basalt".
They also suggested that this basalt precursor may have
been generated from extensive regions of metasomatized/contaminated upper mantle. The data on
Caraiba, presented here, largely concur with this type of
model. Independent evidence of enriched lithospheric
mantle sources linked with anorthosites comes from associated mafic dykes (Carlson et al. 1993).
The puzzling association of unusually Ti-, Fe- and Prich magmas with massif-type anorthosites is well known

(Philpotts 1967; Ashwal 1982; Goldberg 1984; Olsen and


Morse 1990; Wiebe 1992; McLelland et al. 1994), and
there is increasing evidence that liquid immiscibility may
play an important role in the petrogenesis of the anorthosite association (Wiebe 1979), particularly during the
later stages of evolution of such bodies when Fe- and
P-rich de-polymerised liquids separate from Al-rich
highly polymerised feldspathic liquids (Loferski and Arculus 1993). The possibility arises that the sulphides, such
as those at Caraiba and in the Koperberg suite, may also
have developed as immiscible liquids at a late stage of
emplacement of the hypersthenite-norite-anorthosite
body. The ultimate source of the sulphide is then dependent on, or linked to, the petrogenetic model for the
hypersthenite-norite-anorthosite association. Subductionaccretion processes are important in the development of
the continental lithosphere and lower crust, and it is
possible that sulphide may have been introduced into the
source as subducted fossil seafloor hydrothermal systems
(along with Fe-rich sediments), or through reduction of
subducted seawater sulphate. Seawater sulphate can usually be recognised isotopically in volcanic arc magmas (Alt
et al. 1993) because of the enhanced 634S (mean + 4;
range + 0 to + 10), although these drop down rapidly to
typical "mantle" values (~34S = ca.0) in the back-arc regions as a result of isotopic re-equilibration with mantle
sulphur. Low total sulphur contents in arc magmas are
usually explained in terms of early removal of sulphur
droplets from the magma (Alt et al. 1993), so source
regions in the lithosphere or lower crust may become
enriched in sulphide, which could be mobilised later.
Caraiba sulphide isotopic compositions, however, are
clearly not arc-type, and have equilibrated or been reequilbrated with mantle sulphur.
A number of features of the anorthosite association
imply a multistage petrogenesis. For instance, whereas
eSr-eyd relationships necessitate a large mantle contribution, or a primitive source with a very short crustal residence time, the fact that Rb-Sr ratios are often too low to
support observed 87Sr/86Sr ratios (Morse and Hamilton
1991), that K/Rb ratios are high, that A1203 and A1/Ca
ratios are too high to result from normal pyrolite mantle
melting (hence An + Hy dominate the mineralogy rather
than Di), all infer some modification of the source before
magmagenesis. Indeed the Koperberg Suite has
Isr > 0.706, ~r~d= - 4.3 and ]A2 ~ 1 0 , which led Clifford
et al. (1994) to invoke an older lower crustal source.
However, it is very difficult to be definitive, because earlier
subduction of sediment into the continental lithosphere
beneath arcs can reproduce much of the stable and
radiogenic isotope and trace element characteristics normally regarded as 'continental' (cf. Hergt et al. 1991; Alt
et al. 1993; Fourcade et al. 1994). Conversely lateral
growth of continents through subduction-accretion (Tarney and Jones 1994) tectonically mixes stripped-off
oceanic mafic rocks with subducted sediment in the deep
crust. In either case, mantle plume interaction or some
other major tectonothermal event is needed to re-mobilise
these sources and generate high temperature hypersthenic,
noritic and anorthositic magmas. The abundant ultramafic xenoliths at Caraiba and the high-Mg lamprophyres linked to the Koperberg suite (Boer et al. 1994)

371
tend to favour a m a n t l e origin. However, in b o t h these
m e c h a n i s m s there is the potential to s u b d u c t mafic ocean
floor with h y d r o t h e r m a l copper e n r i c h m e n t into the deep
crust or u p p e r m a n t l e to create the right source characteristics for a future C a r a i b a deposit.
Finally, the C a r a i b a rocks, especially the hypersthenites m a y have their e m p l a c e m e n t controlled by
a regional shear belt. A coarse-grained hypersthenite
dyke has given a S m - N d m i n e r a l i s o c h r o n of
1890 _+ 60 M a (Oliveira 1990b) which is r e m a r k a b l y similar to fault-controlled granites (1889 _+ 64 Ma) reported
by P a d i l h a a n d Melo (1991) in the Salvador-Curagfi O r o gen south of the C a r a i b a area. M e t a s o m a t i s m in the
m a n t l e or crust is e n h a n c e d within shear zones. It is
possible that the C a r a i b a C o m p l e x m a y result from rem o b i l i s a t i o n of m e t a s o m a t i s e d c o n t a m i n a t e d m a n t l e or
mafic + sediment lower crust d u r i n g the w a n i n g stage of
e v o l u t i o n of the early Proterozoic Salvador-Curaqfi collisional orogen.

Acknowledoements. The Caraiba Mine Ltd., and its staff, are warmly
thanked for field logistic support, and Rob Wilson and Nick Marsh
for laboratory support with microprobe and XRF analyses at
Leicester. The Brazilian CNPq and the British CVCP are gratefully
acknowledged for providing financial support for EPO at Leicester.

References
Ackermand, D., Herd, R.K., Reinhardt, M., Windley, B.F. (1987)
Sapphirine parageneses from the Caraiba complex, Bahia, Brazil.
The influence of Fe 2+-Fe3+ distribution on the stability of sapphirine in natural assemblages. J. Metamorphic Geol. 5 : 323-339
All6gre, C.J., Minster J.F. (1978) Quantitative models of trace element behaviour in magmatic processes. Earth Planet. Sci. Lett.
38:1 25
Alt, J.C., Shanks, W.C., Jackson, M.C. (1993) Cycling of sulfur in
subduction zones: the geochemistry of sulfur in the Mariana
Island Arc and back-arc trough. Earth Planet. Sci. Lett.
119:477-494
Andreofi, M.A.G., Hart, R.J., Andersen, NJ.B., Moore, J. (1988) Th,
U, REE-fich anorthosite from Namaqualand, S. Africa. Implications for metallogenesis and KREEP-basalt. Chem. Geol. 70 : p.69
Ashwal, L.D. (1982) Mineralogy of mafic and Fe-Ti oxide-rich
differentiates of the Marcy anorthosite massif, Adirondacks, New
York. Am. Mineral. 67 : 14-27
Ashwal, L.D. (1988) Anorthosites: classification, mythology, trivia,
and a simple unified theory. J. Geol. Soc. India 31:6-8
Ashwal, L.D., Wooden, J.L., Emslie, R.F. (1986) Sr, Nd, and Pb
isotopes in Proterozoic intrusives astride the Grenville front in
Labrador: implications for crustal contamination and basement
mapping. Geochim. Cosmochin. Acta 50:2571-2585
Barbosa, J.S.F. (1986) Constitution lithologique et metamorphique
de la region granulitique du Sud de Bahia, Br+sil. Univ. Paris,
Mem. de Sciences de la Terre 86-34: 401pp
Barbosa, J.S.F. (1990) The granulites of the Jequi6 complex and
Atlantic Coast mobile belt, southern Bahia, Brazil - and expression of Archean/Early Proterozoic plate convergence. In: Vielzeuf, D., Vidal, P. (eds.) Granulites and crustal evolution. Kluwer,
Amsterdam, pp. 195-221
Barbosa, J.S.F., Fonteilles, M. (1989) Caracterizac~o dos prot61itos de
regi~o granulitica do Sul da Bahia-Brasii. Rev. Bras. Geoc. 19: 3-18
Berg, J.H., (1980) Snowflake troctolite in the Hettasch intrusion,
Labrador: evidence for magma-mixing and supercooling in a plutonic environment. Contrib. Mineral. Petrol. 72 : 339-351
Boer, R.H., Meyer, F.M., Cawthorn, R.G. (1994) Stable isotopic evidence for crustal contamination and desulfidation of cupriferous

Koperberg Suite, Namaqualand, South Africa. Geochim. Cosmochim. Acta 58 : 2677-2687


Burns, L.E. (1985) The Border Ranges ultramafic and mafic complex, south-central Alaska: cumulate fractionates of island-arc
volcanics. Can. J. Earth Sci. 22:1020-1038
Carlson, R.W., Wiebe, R.A., Kalamarides, R.I. (1993) Isotopic study
of basaltic dikes in the Nain Plutonic Suite: evidence for enriched
mantle sources. Can. J. Earth Sci. 30:1141-1146
Cawthorn, R.G., O'Hara, M.J. (1976) Amphibole fractionation in
calc-alkaline magma genesis. Am. J. Sci. 276:309 329
Chaussidon, M., Albarede, F., Sheppard, S.M.F. (1989) Sulphur
isotope variations in the mantle from ion microprobe analyses of
micro-sulphide inclusions. Earth Planet. Sci. Lett. 92 : 144-156
Clifford, T.N., Gronow, L., Rex, D.C., Burger, A.J. (1975) Geochronological and petrogenetic studies of high-grade metamorphic rocks and intrusives in Namaqualand, South Africa. J.
Petrol. 16:154-188
Clifford, T.N., Barton, E.S., Retief, E.A., Rex, D.C. (1990) New
isotope data from the Koperberg Suite and some associated
rocks, O'okiep district, Namaqualand, South Africa (abstract).
Geocongress 90, Geol. Soc. South Africa. Cape Town, pp. 96-98
Clifford, T.N., Barton, E.S., Retief, E.A., Rex, D.C., Fanning, C.M.
(1994) A crustal progenitor for the intrusive anorthosite-charnockite kindred of the cupriferous Koperberg Suite, O'okiep
District, Namaqualand, South Africa. New isotope data for the
country rocks and the intrusives. J. Petrol. (in press)
Coleman, R.G. (1977) Ophiolites. Springer, Berlin Heidelberg, New
York, 229 pp
Conradie, J.A., Schoch, A.E. (1986) Petrographical characteristics of
the Koperberg Suite, South Africa - An analogy to massif-type
anorthosites? Precambrian Res. 31 : 157-188
Conradie, J.A., Schoch, A.E. (1988) Rare earth element geochemistry
of an anorthosite-diorite suite, Namaqua mobile belt, South
Africa. Earth Planet. Sci. Lett. 87:409-422
Czamanske, G.K., Bohlen, S.R. (1990) The Stillwater Complex and
its anorthosites: an accident of magmatic underplating? Am.
Mineral. 75 : 37-45
Delgado, I.M., Souza, J.D. (1975) Projeto Cobre Curar Rel. final.
Goel. Economica do distrito cuprifero do Rio Curar
Brasil. CPRM/DNPM, vol. 30
Delgado, I.M., Souza, J.D. (1981) Projeto Cobre-Cura~fi-Parte 1. In:
Brasil. DNPM-Cobre no Vale do Rio Curaga, Estado da Bahia.
Geologia 20. Se~fio Geologia Economica 3 : 7 149
DeMaiffe, D., Hertogen, J. (1981) Rare earth element geochemistry
and strontium isotopic composition of a massif-type anorthositic-charnockitic body: the Hidra massif (Rogaland, SW
Norway). Geochim. Cosmochim. Acta 45:1545-1561
DePaolo, D.J. (1981) Trace element and isotopic effects of combined
wallrock assimilation and fractional crystallization. Earth
Planet. Sci. Lett. 53:189-202
Dick, H.J.B. (1977) Partial melting in the Josephine Peridotite. I.
The effect on mineral composition and its consequence for
geobarometry and geothermometry. Am. J. Sci. 227 : 801-832
Emslie, R.F., (1985) Proterozoic anorthosite massifs. In: Tobi, A.C.,
Touret, J.L.R. (eds.) The deep Proterozoic crust in the North
Atlantic Provinces. Reidel, Dordrecht, pp. 39-60
Figueiredo, M.C.H. (1981) Geoquimica das rochas metamorficas de
alto grau do nordeste da Bahia-Brasil. In: Inda, H.A.V.,
Marinho, M.M., Duarte, F.B. (eds) Geologia e Recursos Minerais
do Estado da Bahia, Textos Bfisicos, vol. 4, pp. 1-71
Figueiredo (1989) Geochemical evolution of eastern Bahia, Brazil:
a probable Early Proterozoic subduction-relationmagmatic arc.
J. South Am. Earth Sci. 2:131-145
Foley, S. (1992) Vein-plus-wall-rock melting mechanisms in the
lithosphere and the origin of potassic alkaline magmas. Lithos
28 : 435-453
Fourcade, S., Maury, R.C., Defant, M.J., McDermott, F. (1994)
Mantle metasomatic enrichment versus arc crust contamination
in the Philippines: oxygen isotope study of Batan ultramafic
nodules and northern Luzon arc lavas. Chem. Geol.
114:199-215

372
Frost, B.R., Lindsley, D.H., Simmons, C. (1989) Origin and evolution
of anorthosites and related rocks. Penrose Conference Report.
Geology 17:474-475
Gava, A., Nascimento, D.A., Vidal, J.L.B., Ghignone, J.I., Oliveira,
E.P., Santiago Filho, A.L., Teixeira, W. (1983) Geologia. In:
Projeto RADAMBRASIL. Folhas SC.24/25 Aracaju/Recife.
Levantamento de Recursos Naturais 30:27 376
Gill, J.B. (1981) Orogenic andesites and plate tectonics. Springer,
Berlin Heidelberg New York, 390 pp
Goldberg, S.A. (1984) Geochemical relationships between anorthosite and associated iron-rich rocks, Laramie Range, Wyoming, Contrib. Mineral. Petrol. 87:376 387
Green, D.H., Ringwood, A.E. (1967) The genesis of basaltic magmas.
Contrib. Mineral. Petrol. 15 : 103 190
Green, T.H. (1969) High-pressure experimental studies on the origin
of anorthosite. Can. J. Earth Sci. 6:427-440
Grove, T.L., Baker, M.B. (1984) Phase equilibrium controls on the
tholeiitic versus calc-alkaline differentiation trends. J. Geophys.
Res. 89 : 3253-3274
Hasui, Y., Silva, L.J.H.D., Silva, F.J.L., Mandetta, P., Moraes, J.A.C.,
Oliveira, J.G., Miola, W. (1982) Geology and copper mineralization of Curaq~ River Valley, Bahia. Rev. Bras. Geoc. 12:463-474
Haughton, D.R., Roeder, P.L., Skinner, B.J. (1974) Solubility of
sulfur in mafic magmas. Econ. Geol. 69:451-467
Hawkesworth, C.J., Mantovani, M.S.M., Taylor, P.N., Palacz, Z.
(1986) Evidence from the Paran~t of south Brazil for a continental
contribution to Dupal basalts. Nature 322:356-359
Hergt, J.M., Peate, D.W., Hawkesworth. CJ. (1991) The petrogenesis of Mesozoic Gondwana low-Ti flood basalts. Earth Planet.
Sci. Lett. 105:134 148
Herz, N. (1969) Anorthosite belts, continental drift, and the anorthosite event. Science 164:944 947
Hoover, J.D. (1989) Petrology of the marginal border series of the
Skaergaard intrusion. J. Petrol. 30 : 399-440
Inda, H.A.V., Barbosa, J.F. (1978) Mapa geologico do Estado da
Bahia. (Escala 1: 1.000.000). Secretaria de Minas e Energia da
Bahia, Bahia
Irvine, T.N. (1974) Petrology of the Duke Island ultramafic complex,
southeastern Alaska. Geol. Soc. Am. Mere. 138, 240 pp.
Irvine, R.N., Baragar, W.R.A. (1971) A guide to the chemical classification of the common volcanic rocks. Can. J. Earth Sci. 8 : 523~548
Khan, M.A., Jan, M.Q., Windley, B.F., Tarney, J., Thirlwall, M.F.
(1989) The Chilas mafic igneous complex: the root of the Kohistan Island Arc in the Himalayas of N. Pakistan. Geol. Soc. Am.
Spec. Pap. 232 : 75 94
Kuno, H. (1968) Differentiation of basalt magmas. In: Hess, H.H.,
Poldervaart, A. (eds.) Basalts, vol. 2, Wiley New York, pp. 623~88
Kyser, T.K. (1986) Stable isotope variations in the mantle. In: Valley,
J.W. et al. (eds.) Stable isotopes in high temperature geological
processes. Rev. Mineralogy 16:141-164. Mineralogical Society
of America
Ladeira, E.A., Brockes, J.R.H. (1969) Goelogia das Quadriculas de
Poco de Fora, Esfomeado, Tanque Novo e Lajes, Distrito cuprifero do Rio Curaq/t, Bahia, Brasil. Relat. Geosol submetido ao
DNPM, 209 pp
Lambert, D.D., Simmons, E.C. (1988) Magma evolution in the
Stillwater complex, Montana.II. Rare earth element evidence for
the formation of the J-M reef. Econ. Geol. 83 : 1109-1126
Leake, B.E., Farrow, C.M., Townend, R. (1979) A pre-2000 Myr old
granulite facies metamorphosed evaporite from Caraiba, Brazil?
Nature 277 : 49-50
Leinz, V. (1948) Genese da jazida de cobre de Caraiba, Bahia. Miner.
Metal. 12 : 277-278
Lindenmayer, Z.G. (1981) Evoluc~o geol6gica do Vale do Rio Curaqfi
e dos corpos mfifico-ultram~ficos mineralizados a cobre. In: lnda,
H.A.V., Marinho, M.M., Duarte, F.B. (eds.) Geologia e Recursos
Minerais do Estado da Bahia. Textos Bfisicos, vol. 4:72-110
Loferski, P.J., Arculus, R.J. (1993) Multiphase inclusions in plagioclase from anorthosites in the Stillwater Complex, Montana:
implications for the origin of the anorthosites. Contrib. Mineral.
Petrol. 114:63-78

Lofgren, G. (1974) Temperature induced zoning in synthetic plagioclase feldspar. In: MacKenzie, W.S., Zussman, J. (eds.) The Feldspars. Manchester Univ. Press pp. 362-375
Lombaard, A.F., Exploration Department Staff of the O'okiep Copper Company Ltd. (1986) The copper deposits of the Okiep
district, Namaqualand. In: Anhaeusser, C.R., Maske, S. (eds.)
Mineral deposits of southern Africa. Vols. I, II. Geol. Soc. S.
Africa, pp. 1421-1445
Maaloe, S. (1976) The zoned plagioclase of the Skaergaard intrusion,
East Greenland. J. Petrol. 17:398-419
MacLean, W.H., Shimazaki, H. (1976) The partition of Co, Ni, Cu,
and Zn between sulfide and silicate liquids. Econ. Geol.
71 : 1049 1057
Mandetta, P. (1982) Aspectos geologicos e petrogeneticos das associacoes mafico-ultramaficas de regiao de Caraiba, Vale do
Curaq/~. B.A. Unpublished Master thesis, Federal University of
Bahia, Bahia
Mantovani, M.S.M., Marques, L.S., Sousa, M.A., Civetta, L., Atalla,
L., Innocenti, F. (1985) Trace element and strontium isotope
constraints on the origin and evolution of Paran~t continental
flood basalts of Santa Catarina State (Southern Brazil). J. Petrol.
26: 187-209.
Marinho, N.M., Barbosa, J.S.F., Vidal, P. (1993) O embasamento do
craton do S/io Franciso no suteste da Bahia: revis/io geochronol6gica. Proceedings of the II Symposium on the San Francisco Craton, Brazilian Geological Society, Salvador, pp. 12-16
Mascarenhas, J.F., Pedreira, A.J., Mist, A., Motta, A.C., Silva S.A,
J.H. (1984) Prov[ncia Sao Francisco. In: Almeida, F.F.M., Hasui,
Y. (eds.) O Pre-Cambriano do Brasil. Blucher, S~o Paulo, pp.
4(~122
Mathison, C.I. (1987) Pyroxene oikocrysts in troctolitic cumulates evidence for supercooled crystallization and postcumulus modification. Contrib. Mineral. Petrol. 97:228-236
Mclver, J.R., McCarthy, T.S., Packham, B.V. (1983) The copperbearing basic rocks of Namaqualand, South Africa. Mineral.
Deposita 18 : 135-160
McLelland, J., Ashwal, L., Moore, L. (1994) Composition and petrogenesis of oxide-, apatite-rich gabbronorites associated with
Proterozoic anorthosite massifs: examples from the Adirondack
Mountains, New York. Contrib. Mineral. Petrol. 116:225-238
Menuge, J.F. (1988) The petrogenesis of massif anorthosites: a Nd and
Sr isotopic investigation of the Proterozoic of Rogaland/VestAgder, SW Norway. Contrib. Mineral. Petrol. 98:363-373
Moorlock, B.S.P., Tarney, J., Wright, A.E. (1972) K-Rb ratios of
intrusive anorthosite veins from Angmagssalik, East Greenland.
Earth Planet. Sci. Lett. 14:39-46
Morse, S.A. (1982) A partisan review of Proterozoic anorthosites.
Am. Mineral., 67:1087-1100
Morse, S.A., Hamilton, M.A. (1991) The problem of unsupported
radiogenic strontium in the Nain anorthosites, Labrador. In:
Gower, C.F., Rivers, T., Ryan, B. (eds.) Mid-Proterozoic Laurentia-Baltica. Geol. Assoc. Canada, Spec. Pap. 38:373-385
Mullah, H.S., Bussell, M.A. (1977) The basic rock series in batholithic associations. Geol. Mag. 114:265 280
Mysen, B.O., Boettcher, A.L. (1975) Melting of a hydrous mantle: I.
Phase relations of natural peridotite at high pressures and temperatures with controlled activities of water, carbon dioxide, and
hydrogen. J. Petrol. 16:52(~548
Naldrett, A.J. (1981) Nickel sulfide deposits: Classification, composition, and genesis. Econ. Geol. 75th Anniversary Volume, pp.
628-685
Naldrett, A.J., Mason, G.D. (1968) Contrasting Archean ultramafic
igneous bodies in Dundonald and Clergue townships, Ontario.
Can. J. Earth Sci. 5 : 111 143
Ohmoto, H. (1986) Stable isotope geochemistry of ore deposits. In:
Valley, J.W., et al. (eds.) Stable isotopes in high temperature
geological processes. Rev. Mineralogy 16:491 559. Mineralogical Society of America
Oliveira, E.P. (1990a) Novos conceitos sobre as rochas m~tfico-ultramfificas cupnferas da mina Caraiba, Bahia, Rev. Bras. Geoc.
19 : 449-461

373
Oliveira, E.P. (1990b) Petrogenesis of mafic-ultramafic rocks from
the Precambrian Curaq/t terrane, Brazil. Unpublished PhD
thesis, University of Leicester, Leicester, UK, 287 pp
Oliveira, E.P., Choudhuri, A. (1993) Sulphur isotope geochemistry
indicative of a mantle source for the Caraiba copper sulphides,
Brazil. Ext. Abstract, 2nd Symposium on $5+oFrancisco Craton,
21-25 August, Salvador, Brazil
Oliveira, E.P., Lacerda, C.M.M. (1993) Field evidences for the synkinematic emplacement of the Caraiba hypersthenites, Brazil,
Ext. Abstract, 2rid Symposium on S~o Francisco Craton, 21-25
August, Salvador, Brazil
Oliveira, E.P., Tarney, J. (1994) Nature and significance of mafic and
ultramafic inclusions in the early Proterozoic Caraiba Hypersthenite-Norite Complex, Brazil. J. South Am. Earth Sci. (in
preparation)
Olson, K.E., Morse, S.A. (1990) Regional AI-Fe mafic magmas
associated with anorthosite-bearing terranes. Nature 344:
76t~762
Osborn, E.F. (1959) Role of oxygen pressure in the crystallization
and differentiation of basaltic magma. Am. J. Sci. 257 : 609 647
Padilha, A.V., Melo, R.C. (1991) Evoluq~o Geol6gica. In: Melo, R.C.
(ed.) Programa de Levantamentos Geol6gicos B~tsicos do Brasil.
Pintadas-Folha Sc. 24-YD-V Estado da Bahia. DNPM/CPRM,
pp. 129 157
Philpotts, A.R. (1967) Origin of certain iron-titanium oxide and
apatite rocks. Econ. Geol. 62:303-315
Philpotts, J.A., Schnetzler, C.C. (1970) Phenocryst-matrix partition
coefficients for K, Rb, Sr and Ba, with applications to anorthosites and basalt genesis. Geochim. Cosmochim. Acta
34: 307-322
Sa, E.F.J., Archanjo, C.J., LeGrand, J.M. (1982) Structural and
metamorphic history of the part of the high-grade terrain in
Curar valley, Bahia, Brazil. rev. Bras. Geoc. 12:251-262
Sa, E.P., Reinhardt, M.C. (1984) Aspectos metadologicos da prospeq~a mineral no vale do Rio Cura~/t-Bahia. In: Anais Simp.
Bras. Tecn. Explor. Aplic. Geologia, Soc. Bras. Geol., Salvador,
pp. 250-278
Saunders, A.D., Tarney, J., Weaver, S.D. (1980) Transverse geochemical variations across the Antarctic Peninsula: implications
for the genesis of calc-alkaline magmas. Earth Planet. Sci. Lett.
46 : 344-360
Schneider, A. (1951) Piroxenitos cupriferos de Caraiba, Bahia,
Miner. Metal. 15:271 276
Schoch, A.E., Conradie, J.A. (1990) Petrochemical and mineralogical
relationships in the Koperberg Suite, Namaqualand, South Africa. Am. Mineral. 75:27-36
Silva, L.J.H.D. (1985) Geologia e controle estrutural do dep6sito
cuprifero de Caraiba. Vale do Curaq/t-Bahia. In: Sa, P.V.S.V.,
Duarte, F.B. (eds) Geologia e Recursos Minerais do Estado da
Bahia, Textos B/Lsicos 6:51-136
Simmons, E.C., Hanson, G.N. (1978) Geochemistry and origin of
massif-type anorthosites. Contrib. Mineral. Petrol. 66:119-135
Srikantappa, C., Hormann, P.K., Raith, M. (1984) Petrology and
geochemistry of layered ultramafic to mafic complexes from the
Archaean craton of Karnataka, southern India. In: Kr6ner, A.,
Hanson, G.N. Goodwin, A.M. (eds.) Archaean Geochemistry,
Springer, Berlin Heidelberg, New York, pp. 138-160
Stumpfl, E.F., Clifford, T.N, Burger, A.J., Van Zyl, D. (1976) The
copper deposits of the Okiep district, South Africa: new data and
concepts. Mineral. Deposita 11 : 46-70
Sun, S.-s., McDonough, W.F. (1989) Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and
processes. In: Saunders, A.D., Norry, M.J. (eds.) Magmatism in the
Ocean Basins. Geol. Soc. London, Spec. Publ. 42, pp. 313 345

Suszczynski, E.F. (1972) A origem sedimentar-metam6rfica estratiforme do minerio cuprifero do distrito norte da bahia. Anais
XXVI Congr. Bras. Geol., SBG, Bel6m, 1 : 167 172
Tarney, J., Wood, D.A., Saunders, A.D., Cann, J.R., Varet, J. (1980)
Nature of mantle heterogeneity in the North Atlantic: evidence
from deep sea drilling. Phil. Trans. Roy. Soc. Lond
A297:179 202
Tarney, J., Jones, C.E. (1994) Vertical and lateral contributions to
crustal growth: geochemical and tectonic aspects. U.S. Geological Survey Circular 1107, p. 315
Taylor, S.R., Campbell, I.H., McCulloch, M.T., McLennan, S.M.
(1984) A lower crustal origin for massif-type anorthosites. Nature
311 : 372-374
Teixeira, W., Figueiredo, M.C.H. (1991) An outline of the Early
Proterozoic crustal evolution in the S~o Franscisco craton, Brazil: a review. Precambrian Res. 53 : 1-22
Thompson, R.N., Morrison, M.A., Dickin, A.P., Hendry, G.L. (1983)
Continental flood basalts.., arachnids rule OK?. In: Hawkesworth. C.J., Norry, M.J. (eds.) Continental basalts and mantle
xenoliths. Shiva Publ., Nantwich, pp. 158-185
Townend, R., Ferreira, P.M., Franke, N.D. (1980) Caraiba, a new
copper deposit in Brazil. Trans. Inst. Min. Metall. 89 : B159-B164
Vinogradov, I.A. (1976) Palingenic norites and pyroxenites of the
zone of granulite facies of metamorphism. Int. Geol. Rev.
18:1103-1109
Von Gehlen, K, Nielsen, H., Rozendaal, A., Jensen, M.L. (1990)
Sulphur isotopes indicate a homogeneous, probably mantle
source of the Okiep copper ores, Namaqualand. Geocongress,
23rd Earth Sci. Congr. Geol. Soc. South Africa, Capetown, 2-6
July 1990, pp. 96-98
Wager, L.R., Brown, G.M. (1967) Layered igneous rocks. Freeman
San Francisco, 588 pp
Weaver, B.L., Tarney, J. (1983) Chemistry of the sub-continental
mantle: inferences from Archaean and Proterozoic dykes and
continental flood basalts. In: Hawkesworth, C.J., Norry, M.J.
(eds) Continental basalts and mantle xenoliths. Shiva, Nantwich,
pp. 209 229
Weaver, B.L., Tarney, J. (1984) Empirical approach to estimating the
composition of the continental crust. Nature 310:575-577
Weaver, B.L., Tarney, J., Windley, B.F. (1981) Geochemistry and
petrogenesis of the Fiskenaesset anorthosite complex, southern
West Greenland: nature of the parent magma. Geochim. Cosmochim. Acta 45 : 711-725
Weaver, B.L., Tarney, J., Windley, B.F., Leake, B.E. (1982) Geochemistry and petrogenesis of Archaean metavolcanic
amphibolites from Fiskenaesset, S.W. Greenland. Geochim.
Cosmochim. Acta 46: 2203-2215
Wiebe, R.A. (1979) Fractionation and liquid immiscibility in an
anorthosite pluton of the Nain complex, Labrador. J. Petrol.
20 : 239-269
Wiebe, R.A. (1992) Proterozoic anorthosite complexes. In: Condie
K.C. (ed.) Proterozoic Crustal Evolution. chpt. 6. Elsevier, Amsterdam, pp. 215-261
Windley, B.F. (1984) The evolving continents, 2nd edn. Wiley, New
York London, 399 pp
Windley, B.F., Bishop, F.C., Smith, J.V. (1981) Metamorphosed
layered igneous complexes in Archaean granulite-gneiss belts.
Ann. Rev. Earth. Planet. Sci. 9 : 175-178
Yoder, H.S., Tilley, C.E. (1962) The origin of basalt magmas: an
experimental study of natural and synthetic rock sytems. J.
Petrol. 3 : 342 532
Editorial handling: H. Prichard

Vous aimerez peut-être aussi