It '2 c

.315. .n..\.d-o C

81%” m5 .2 Law...” u .ou

«u «a .5.
Ml, Ti”... 1 a... I.

New.” Vi n. w:

l
I L . u ‘ “1 us 6
HUMME- . 1 up 1.5qu. Add”
.to .\| . w I “up
‘3”: fi\ 0“ 0...,“ w... '
9; sh- .. a,“ an»;
d o . .. v I

.. Ann»
«2..» . x p.

m a} a .c ..!....-H An”.

. o ‘03- . .0
ll n 4 .h .5. . b as .
Av?” .14... aflafi.’ ”J...“
T. L \. . . l ‘
..v.. r. .
fl. ,.

.3. b JCu-x -.
‘ . .... u .

Asa. on... mflé‘. 3 r..
“c...“ ._.W .n.... ..
. u. a o a: u to
u... . ”.7. . a «Au
.4 - a». a.
My f... .0 0.! .5 1W.ol“
P... —
.‘C. B.
”n.1,“

I.)

ply.

 

PETROGENESIS OF OCEANIC GRANITFS FROM THE
AVES RIDGE IN THE CARIBBEAN BASIN

by

Bruce M, Walker

A THESES~
Submitted to
iMichigan State University
in partial fUlfillment of the requirements
. for the degree of
iMASTER OF SCIENCE
Department of Geology

1972

ACKNOWLEDGMENTS

I sincerely thank Dr. Thomas A. Vogel for providing me with the
opportunity to research the Aves specimans. His stimulating discussions
and patience with my procrastinations have been Herculean.

Also, I am indebted to Dr. Robert Ehrlich for his genius and to Dr;
Paul J. Fox for the Aves specimens he provided.

Thank you, Pat Kaneshiro, for'your strength in times of despair.

EARTH AND PLANETARY SCIENCE LETTERS 15 (1972) 133-139. NORTH-HOLLAND PUBLISHING COMPANY

[E

PETROGENESIS OF OCEANIC GRANITES
FROM THE AVES RIDGE IN THECARIBBEAN BASIN

Bruce M. WALKER, Thomas A. VOGEL and Robert EHRLICH
Department of Geology, Michigan State University, USA

Received 9 December 1971
Revised version received 22 February 1972

A genetic model proposed for granitic rocks dredged from the Aves Rise involves a sodium-enriched hydrothermal
fluid phase, shallow emplacement, and significant derivation of components from the upper mantle. Megascopic observa-
tions of serially slabbed sections of the rock samples show that aggregates of plagioclase phenocrysts concentrate into
channel-like mats. The presence of a late-stage, hydrothermal fluid is suggested by crystal-lined cavities in the plagioclase

aggregates.

With progressively decreasing pressure, this low density, sodium-rich, hydrous fluid phase unmixed from the silicate
melt and entrained the plagioclase aggregates as it rose. Reaction between this sodium-rich hydrous phase and the plagio-
clase produced albite rims. Genetic independence of the plagioclase aggregates and the granitic groundmass which sur-
rounds them is implied by the results of a chi square test. Thus, the system consisted of three distinct components — the
early-fanned calcic plagioclase, a sodium-rich hydrous fluid phase, and a granitic-silicate fluid phase.

Shallow intrusion (less than 15 km) is indicated by the structural state of K-feldspars and presence of miarolitic cavities.
8-Sr/8‘58r values, which correspond to ratios obtained from oceanic basalts, suggest that the granitic rocks are primitive

and most likely from the upper mantle.

1. Introduction

In 1969, Dr. Bruce C. Heezen of Columbia Univer-
sity, aboard Duke University’s R/V Eastward, at-
tempted to sample an anomalous 6.0—6.3 km/ sec
crustal layer [1] that underlies most of the Caribbean
Basin area. Samples were obtained by dredging across
an escarpment near the southern end of the Aves Rise,
which separates the Venezuela Basin from the Grena-
da Trough 450 km north of Venezuela [2]. Two
dredge hauls, located within 10 km of each other, con-
tained a combined total of 3500 kg of granitic rocks.

These rocks range in size from small cobbles to a
600 kg boulder. The granitic rocks are coarse-grained
and fractured. Some of the larger fragments appear
to have been broken directly from outcrop by the
dredge. Fossil encrustations and up to several centi-
meters of manganese oxide coatings [2] , indicate that
they were exposed to sea water for considerable time.

Tests on the rock samples under a confining pressure
equivalent to 6-10 km depth gives a compressional
wave velocity ranging from 6.0 to 6.4 kin/sec, and
the velocity versus pressure curves are similar to those
of continental granites [2] .

The present study on these dredge samples was
begun during the latter part of 1970. This paper pre-
sents a genetic model of the Aves Rise granitic rocks,
based upon a synthesis of the textural and composi-
tional variations observed from the dredge fragments.

2. Megascopic features

Slabbed sections of the rock samples from both
dredge hauls showed specific compositional and tex-
tural variation not observable on a smaller scale. The
largest sample available (30 X 20 X 15 cm) from the
two dredge hauls was slabbed serially in each of two

 

134 B.M. Walker et aI., Petrogenesis of oceanic granites

directions at right angles to one another in order that
variation in three dimensions could be observed. Ten
smaller samples were also slabbed and were studied to
test the consistency of textural relationships from
sample to sample.

Four petrogenetically important features were
identified on the slab level:(i) Aggregates of plagio-
clase forming bleb—like particles, (ii) Crystal-lined vugs,
(iii) Groundmass material surrounding the aggregates
of plagioclase, and (iv) Granophyric matrix material.

The first three features are common to all clasts.
Granophyric matrix is confined to two samples.

2.1. Plagioclase aggregates

All of the rock samples that were slabbed exhibit
bleb—like aggregates of plagioclase. These aggregates
have varying spatial relationships Gigs] and 2), and in
some instances they approximate a close-packed ar-
rangement with interstitial groundmass, whereas in
other instances they appear to be suspended in

 

Fig. 2. Slab exhibiting varying spatial relationships of out-
lined plagioclase aggregates. Scale is in centimeters.

 

groundmass (see below). The size and shape of the
aggregate vary considerably throughout each slab.

The combined effect of the above textural char-
acteristics is to produce aggregate-rich zones separa-
ted by aggregate-poor zones. The rocks were slabbed
serially to study the spatial character of these zones.
In all cases segregations of plagioclase aggregates oc-
cur in channel-like features roughly circular in cross-
section, a few centimeters wide extending through
the sample. In some instances these channel-like
features were observed to bifurcate. In fig. 3.a chev-
ron-like effect within a very dense mat of plagioclase
represents an extreme condition of such a channel-
way. The presence of chevron-like lineaments in this
very dense clot of plagioclase aggregates provides di-
rect evidence for relative movement along these
channels.

These features, plus the association of plagioclase

Fig. l. Plagioclase aggregates (outlined) surrounded by ground- aggregates With VUSS, as discussed belOW, indicate
mass material. Scale is in centimeters. that the channel-like, aggregate-rich masses accumu-

 

 

r!

-

II'

 

RM. Walker et aI., Petrogenesis of oceanic granites 135

Fig. 3. Photograph displaying mat-like features of plagioclase.
Plagioclase aggregates are poorly defined in the light area in
the center of the photo, which is almost entirely plagioclase.
The dark zone just to the right of this light area has chevron
lineations of plagioclase suggesting relative movement of the
aggregates in the rock before crystallization.

lated in actual channelways during the movement of
solid and fluid phases.

2.2. Vugs

Scattered through the larger rock samples are vugs
lined with crystals of feldspar and epidote that com-
prise less than one percent of the rock volume. The
vugs are always found in association with aggregates
of plagioclase. In most cases they occur within the
aggregate; in a few cases they occur near the point of
contact of two plagioclase aggregates. In general, the
number of vugs in a sample is proportional to the size
of the sample and no particular importance is placed
on an apparent absence of vugs in some of the smaller
sized samples.

2.3. Groundmass

Surrounding the plagioclase aggregates (except in
the two granOphyric samples) is a groundmass which
is comprised principally of various preportions of ge-
nerally finer-grained quartz, potassium feldspar, am-
phibole-chlorite, and plagioclase not included in the
aggregate population. It is possible that, due to the
irregular surfaces of the plagioclase within the aggre-
gates, some of the plagioclase in the groundmass may
be part of an aggregate. Volumetrically, however, this
effect must be minimal, since the single grains of pla-
gioclase in the groundmass in many cases can be ob-
served as separate entities and not comers of plagio-
clase within the aggregates. The shapes of the aggre-
gates would have to be extremely irregular in order to
produce the amounts of groundmass plagioclase that
are present in the rock samples. Such is not the case.

2.4. Granophyric matrix

Two samples have a fine-grained, granophyric ma-
trix. Information obtained from microprobe and pet-
rographic data is more pertinent than can be obtained
on this slab scale and will be discussed below.

3. Selected mineralogic relations

The dredged samples are composed principally of
plagioclase, quartz, and K-feldspar, with lesser
amounts of hornblende, chlorite and Opaques. Modal
analyses of five samples are given in table 1. As will
be demonstrated in a later section, the plagioclase
aggregates were not derived from their immediate
groundmass; and, therefore, in table 1 each analysis
is presented in two ways: column 1 gives the total
modal analysis of the thin section; column 2 gives the
modal composition of only the matrix. According to
the classification scheme of Streckeisen [3] the plots
of the modal analysis of the whole rock fall near the
boundary of granodiorites and granites, but in the
granodiorite field. The matrix of these rocks would
fall in the middle of the granite field.

The plagioclase are extremely zoned in all cases.
Well over fifty zones were counted in large plagio-
clase phenocrysts from the aggregates and an estima-
ted 100 zones appear in a few cases. Plagioclase com-

 

136

RM. Walker et aI., Petrogenesis of oceanic granites

Table 1

Modal analyses of five samples dredged from the Aves Rise. Column (1) for each sample is the modal analysis of all the minerals

present in the sample. Column (2) is the modal analysis of the matrix; that is, the plagioclase in the blebs have been subtracted
from the bulk modal analysis. The mineral category labeled ‘Other’ is chlorite, hornblende and Opaques.

 

 

 

7A8 7—Cb1 7A1 Cb15 319—C9
l 2 w *1 2 l 2 1 2 1 2
Plagioclase S6 20 43 16 44 21 50 13 4O 14
K-Feldspar 18 31 23 34 19 27 16 27 22 32
Quartz 20 36 22 33 30 42 26 44 29 41
Other 7 13 ll 17 7 10 9 15 8 12

 

 

Fig.4. Plagioclase aggregate phenocryst showing oscillatory

zoning from core to rim. Light area bounding the extinct

zone of the phenocryst in an albite rim. Short dimension of
the photograph is 1.6 mm.

position in the aggregates ranges from An 70 in the
core and decreases continuously, or in an oscillating
manner to An 20 - 25, and is mantled by pure al-
bite. These mantles generally show a sharp discontin-
uity in composition with respect to the inner zones

(fig. 4). In contrast, the groundmass plagioclase lacks ‘

this clearly defined albite rim. Clearly, these plagio-
clase were subjected to many changes in pressure,
temperature, and composition.

Microprobe analyses of the granophyre shows
abundant albite in the matrix and that nearly pure
K-feldspar is surrounded and embayed by this albite
(fig. 5). Perthitic development within single K-feld-
spar crystals varies from large areas that contain no
albite to areas that are nearly completely replaced by
this albite with only small remnants of K-feldspar.

Using X-ray techniques described by Wright and
Stewart [4] and Wright [5] , structural state and
compositional analyses of potassium feldspar were
made on feldspars from the same thin section sites
used in the chi square study (see below) and on
granophyric samples. The samples yielded potassium
feldspar having a structural state and composition
analogous to the Spencer B type (low sodium-inter-
mediate microcline) used by Wright [5] . The grano-
phyric samples also yielded Spencer U or Spencer B
type low sodium-intermediate microclines. The low
amounts of albite in solid solution in these interme-
diate K-feldspars is further evidence that much of the
sodium present is a late event rather than associated
with silicate melt and the crystallizing K-feldspars.

The pattern of late-stage crystallization of albite
and the presence of vugs implies the existence of a

 

lbs

all

all

§A~

ml
all

B.M. Walker et aI., Petrogenesis of oceanic granites 137

 

(A)

. ,-a '.
a .‘ h . 'I'.‘ ‘
' ' I -~'.' .v-hi'a' . ' - I '
' . - a .I
c ‘2' C?" .
-. . . . _ ..

5......“

{=64 .

h .' a
I u
e ,_' ‘ o
' ‘ f‘J-

I

“IV-r - . '
3‘
u.

““H‘!

 

Fig. 5. (A) Microprobe sample current photograph of granophyric material and a perthitic K-feldspar. Each division is 10 microns.
(B) Microprobe photograph of sodium KaX-ray of the same area as in fig. 5 (A). Albite (light areas) appears as a matrix in
the granophyre and as included areas in the K-feldspar phenocryst.

sodium-rich, hydrous fluid phase which has separated
from the silicate liquid. The solubility of water in a
granitic liquid has been shown by Luth and Tuttle
[6] to be greatly increased upon the addition of so-
dium disilicate and that the solubility of albite in the
fluid phase decreases markedly with decreasing pres-
sure and temperature. The Aves Rise granitic rocks
were affected by a late, highly sodic fluid as shown

by the albite rims on the plagioclase within the bleb-

like aggregates and by the K-feldspar-albite relation-
ships. It is concluded that the late stage emplacement
of the albite probably is a response of the fluid phase
to decreasing pressure and/ or temperature.

4. Relationship of plagioclase aggregates to ground-
mass

In a system where there is a significant temperature

-. range between the first-formed phases and later phases

it is most likely that these early phases will be separa-
ted from the liquid from which they crystallized.
Most shallowly emplaced granitic rocks probably ex-
hibit relationships similar to those discussed here, but

I". have not been subjected to the same type of textural

rs.
it:

3

analysis.
In the Aves Ridge granites the independence be-
tween plagioclase aggregates and groundmass appears

to be obvious because of the channelizations and in-
homogeneities of the plagioclase aggregates. This in-
dependence was tested quantitatively by a chi square
test of a contingency table array. The specific clas-
sification tested in the contingency table was the
plagioclase aggregate to groundmass ratio per thin
section versus total plagioclase counts per thin section.
The results of this test, xfgdjje 24.7 < x2(P.05),
imply independence between the plagioclase aggre-
gates to groundmass ratio and total plagioclase counts
in thin sections. Thus, the observed variation in pla-
gioclase counts can occur at any and all of the obser-
ved plagioclase aggregate to groundmass ratios. This
implies, as would be predicted from the nature of the
phases, that the plagioclase aggregates were not deri-
ved from their immediate groundmass.

5. A genetic model of the Aves Rise granites

The following observations are significant for con-
struction of a genetic model for the Aves Rise gran-
ites.

(1) Volumetric concentrations of plagioclase aggre-
gates forrning channel-like features within the
rock samples;

(2) Presence of crystal-lined vugs within plagio-
clase aggregates;

 

138 RM. Walker et aI., Petrogenesis of oceanic granites

(3) Genetic independence of plagioclase aggregates
and their immediate groundmass;

(4) Presence of albite rims around highly zoned cal-
cic cores on plagioclase crystals in the aggrega-
tes;

(5) Albite mantling potassium feldspar grains and
forming replacement perthites;

(6) Presence of granophyric matrix in some samples;

(7) Structural state, X-ray diffraction analyses of
K-feldspar showing them to be a low sodium,
intermediate microcline;

(8) The rocks are dredged from the anomalous
6.0—6.3 km/sec layer in the Caribbean Basin
[2] ;

(9) Absence of muscovite and biotite.

These results must also be interpreted in the light
of the 87Sr/868r ratios of these samples determined by
Spooner (personal communication) to be discussed
more fully, with the rare-earth element abundance pat-
terns, in a later paper. Values for 87Sr/B‘SSr of 0.7038,
0.7046 and 0.7080 (all i 0.001) were found for sam-
ples of fresh granite, an inclusion in the granite and a
granophyre, respectively. In view of the young ages
encountered (K/Ar = 57—89 my) [2] , the isotopic ra-
tios measured are effectively initial ratios, especially
since the Rb/Sr ratios for these specimens are low
(0.31 i 018*"). The initial 87Sr/868r values are lower
than those generally encountered in crustal granites
and are similar to values found in oceanic basalts pre-
sumably derived from the upper mantle. Further, all
the ratios measured are lower than that of present-day
sea water indicating little or no equilibration with
common strontium.

The following genetic conditions can be derived
directly from the above observations. (1) The existence
of a three component system consisting of early form-
ed calcic plagioclase, a granitic silicate melt phase and
a sodium-rich, hydrous fluid phase. (2) All textural
components — aggregates, groundmass and granophyre
— have 8"Sr/36Sr ratios reflecting an upper mantle
origin. (3) Early formation of basic plagioclase pheno-
crysts. (4) Disequilibrium conditions existing between
calcic plagioclase and granitic groundmass. (5) Aggre-
gate formation, channelization and transport of early
formed plagioclase by the sodium-rich hydrous fluid.
(6) A progressive decrease in pressure implied by pro-
gressive late-stage sodium enrichment. (7) Shallow
depth of crystallization (less than 15 km) is indicated

by the presence of intermediate structural state of
K-feldspar and absence of muscovite. The presence
of vugs probably indicates crystallization at a shal-
lower level.

Taking these observations and interpretations into
account, the following conclusions can be made. The
Aves Rise granites represent a primitive granite. The
groundmass crystallized at its present location (less

than 15 km depth) and the plagioclase aggregates pro”:

ably crystallized at a greater depth.

Although the 87Sr/86Sr data indicate all of the
components are derived from the upper mantle, the
chi square results indicate that the plagioclase aggre-
gates are not derived from their immediate ground-
mass. The plagioclase may either have formed earlier
and been separated from its parent melt, or the pla-
gioclase and the granitic groundmass reflect two sep-
arate genetic events. Because of the highly calcic
cores of the plagioclase in the aggregates, the authors
favor the latter alternative. A possible model is sug-
gested if one considers the sequence of events causing
heat transfer from the mantle to the crust.

It is generally agreed that the material overlying t
mantle consists oflow melting point products of the
mantle. If a molten fluid derived from the mantle
were injected into this material, the overlying mate-
rial would melt along with crystallization of some of
the injected material. In the Aves Rise rocks the pla-
gioclase aggregates could arise from the mantle-derio
ved liquid whereas the granitic groundmass could be
produced by partially melting basaltic lower crust.
The composition of the highly calcic plagioclase cor.
is consistent with crystals in equilibrium with a liqu;.
derived by partial melting under upper mantle cond:
tions, and similarly the granitic groundmass is con-
sistent with liquid derived by partial melting under
crustal conditions. The initial melt produced in the
lower crust was probably under saturated with respvc.
to water [7]. As this liquid, along with the early pla-
gioclase crystals, was emplaced into higher levels in
the Caribbean crust, the silicate liquid became satur:

ted and a hydrous (sodium-rich) fluid phase separatJl

Total crystallization occurred soon after separation .1
the hydrous phase. The present textures observed at;
a result of the tendency for the early plagioclase cry
tals to agglutinate into aggregates and the entrainmei
of these plagioclase aggregates by the low density
fluid phase as it rose in the silicate liquid.

|

 

 

RM. Walker et aI., Petrogenesis of oceanic granites l 39

In light of F ox et al.’s [2] velocity determination
for these rocks, it is possible that the 6.0—6.3 km/sec
velocity layer that underlies the Caribbean is com-
posed, in part, of granitic rocks derived in a similar
fashion to those from the Aves Rise. For this reason
other granitic rocks from the Caribbean should be
subjected to a similar geochemical-textural analysis.
If the results are in accord with those presented here,
then perhaps these rocks and the processes which
formed them are analogous to primitive continental
crusts, in which case the Caribbean crust is a proto-
continent.

Acknowledgements

We would like to thank Dr. Bruce C. I-leezen and
Dr. Paul J. Fox (Lamont-Doherty, Columbia Univer-
sity) who kindly made all of their samples from the
Aves Rise available for our study. The strontium iso-
topic measurements were made by Charles M. Spooner
(Michigan State University) at the Massachusetts In-
stitute of Technology with the cooperation of Profes—
sor P.M. Hurley. Although we, of course, take full
responsibility for the views expressed here, we thank
the individuals, especially Dr. Peter H. Mattson, who
reviewed earlier drafts of the manuscript.

References

[1] QB. Officer, J.I. Ewing, J.F. Hennion, D.G. Harkrides and
D.E. Miller, Geophysical investigations in the eastern Ca-
ribbean: Summary of 1955 and 1956 cruises, in: Physics
and Chemistry of the Earth, L.H. Ahrens, F. Press, K. Ran-
kama and S.K. Runcom, eds. (Pergamon Press, London,
1959) 17.

[2] RI. Fox, E. Schrieber and B.C. Heezen, The geology of
the Caribbean crust: Tertiary sediments, granitic and basic
rocks from the Aves Ridge, Tectonophysics 12 (1971) 89.

[3] A.L. Streckeisen, Classification and nomenclature of ig-
neous rocks, NJb. Miner. Abh. 107 (1967) 144.

[4] T.L. Wright and DB. Stewart, X-ray and optical study of
alkali feldspar: 1. Determination of composition and
structural state from refined unit cell parameters and 2V,
Am. Mineralogist 53 (1968) 38.

[5] T.L. Wright, X-ray and optical study of alkali feldspar:

II. An X-ray method for determining the composition
and structural state from measurement of 2 0 values for
three reflections, Am. Mineralogist 53 (1968) 88.

[6] W.C. Luth and O.F. Tuttle, The hydrous vapor phase in
equilibrium with granite and granitic magmas, Geol. Soc.
Am. Mem. 115 (1969) 513.

[7] W.S. Fyfe, Some thoughts on granitic magma, inzMecha-
nism of Igneous Intrusion, G. Newall and N. Rast, eds.,
Geol. J. Spec. 135. no. 2 (1970) 201.