Segregation vesicles, cylinders, and sheets in vapor...

www.elsevier.com/locate/jvolgeores

Journal of Volcanology and Geotherm

Segregation vesicles, cylinders, and sheets in vapor-differentiated

pillow lavas: Examples from Tore-Madeira Rise and

Chile Triple Junction

Renaud Merlea,*, Martial Caroff b, Jacques Girardeaua,

Joseph Cottenb, Christele Guivela

aUMR 6112, Laboratoire de Planetologie et Geodynamique, UFR des Sciences et Techniques, Universite de Nantes, 2 rue de la Houssiniere,

44322 Nantes cedex 3, FrancebUMR 6538 bDomaines OceaniquesQ, Universite de Bretagne Occidentale, 6 avenue Le Gorgeu, C.S. 93837, 29238 Brest cedex 3, France

Received 24 February 2004; accepted 1 September 2004

Abstract

We conducted a detailed field and laboratory study of internal segregation structures of two hand-size pillow lavas samples.

They were dredged, respectively, on the Josephine seamount, Tore-Madeira Rise (TMR), and on a small quaternary volcanic

edifice located on the continental edge of the trench close to the Chile Triple Junction (CTJ). Both pillows display a

combination of four types of segregation structures (spherical vesicles, pipe vesicles, vesicle cylinders, and vesicle sheets)

observed so far only within subaerial basalt flows typically 2–10 m thick. In particular, the samples offer a remarkable exposure

of the transition between pipe vesicles and cylinders. We show that the vesicle sheets are not generated by the same mechanism

in both occurrences; they do not seem to be connected to cylinders in the CTJ pillow as they are in the TMR pillow. The two

pillows are geochemically distinct, the TMR being alkaline and the CTJ calc–alkaline. Two types of internal differentiation are

proposed. The first one implies the extraction of the residual liquid from the host lava and transport towards the segregation

structures, whereas the other one results from in situ crystallization within one given structure. In the latter case, glass

composition is highly dependant on the nature of the neighbouring crystallizing minerals. The degree of crystallization required

to produce a crystal framework strong enough for generating the segregation structures seems to be lower in pillows (ca. 25%

crystallization) than in vapor-differentiated basaltic lava flows (35% crystallization).

D 2004 Elsevier B.V. All rights reserved.

Keywords: vapor differentiation; segregation vesicles; vesicle cylinders; vesicle sheets; crystal framework

0377-0273/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.jvolgeores.2004.09.007

* Corresponding author. Present address: Laboratoire de Geo-

chronologie, Geosciences Azur, UMR 6526, Parc Valrose, 06108

Nice cedex 02, France. Tel.: +33 4 92 07 65 88; fax: +33 4 92 07 68

16.

E-mail address: [email protected] (R. Merle).

1. Introduction

Four main types of segregation structures have

been recognized in effusive and intrusive magmatic

al Research 141 (2005) 109–122

R. Merle et al. / Journal of Volcanology and Geothermal Research 141 (2005) 109–122110

bodies: spherical segregation vesicles, (rare) pipe

vesicles, vesicle cylinders, and vesicle sheets. Many

mechanisms have been put forward to explain the

origin of these segregation structures. Spherical

segregation vesicles may derive from shrinkage of

gas during cooling (Smith, 1967) or from escape of

gas to form microvesicle chains (Bideau and Heki-

nian, 1984). Pipe vesicles are elongated bubbles

occurring only either near the base of basalt flows

or in the interior part of pillow lavas. Vesicle cylinders

result from solidification of low-density diapirs of gas

and differentiate liquid rising through magma bodies

(Goff, 1977, 1996). Segregation sheets have been

considered to result from thermal contraction (Green-

ough and Dostal, 1992), compaction (Philpotts et al.,

1996), or compositional convection (Helz, 1980; Helz

et al., 1989).

Only gas filter-pressing is able to generate

together the four types of segregation structures in

basalt flows or dykes. This magmatic differentiation

process was described by Anderson et al. (1984) as

migration of residual liquid through a porous and

permeable but rigid network of interlocking crystals

(i.e., crystal framework). Gas filter-pressing is caused

Fig. 1. Location maps of the Tore-Madeira Rise and Chile Triple Junctio

Gibraltar Transform Zone; TMR: Tore-Madeira Rise; Smt: Seamount. (b

(modified after D’Orazio et al., 2003). 1: Trenches; 2: Oceanic fracture z

by the build-up of gas pressure due to second boiling

that is relieved either by expulsion of melt out of the

crystallization zone (Sisson and Bacon, 1999) or by

its migration into the previously formed vesicles

(Anderson et al., 1984). This process has been named

vapor differentiation by Goff (1977, 1996), Sanders

(1986), Puffer and Horter (1993), Rogan et al.

(1996), and Caroff et al. (1997, 2000). In some

basaltic flows, the inflation process may have been an

important factor in formation of segregation struc-

tures (Thordarson and Self, 1998; Stephenson et al.,

2000). The main characteristic of vapor-differentia-

tion-related segregation structures (except spherical

vesicles) is the high vesicularity of enclosed melt.

Caroff et al. (2000) have shown that the morphology

of the segregation structures is highly dependent on

the thickness of the magmatic host. However,

exceptions are the spherical segregation vesicles

which are ubiquitous in the whole flow, and the pipe

vesicles, which occur only at the base of the flows.

According to the classification of Caroff et al. (2000),

lava flows thicker than 10–15 m are characterized by

the occurrence of vesicle sheets 10–40 cm thick,

having generally a pegmatoid texture (S3 type).

n. (a) Sketch map of north-central Atlantic Ocean. AGTZ: Azores-

) Location map of CTJ relative to the Chile Trench and Patagonia

ones; 3: Liquine-Ofqui Fault System; 4: Volcanoes.

R. Merle et al. / Journal of Volcanology and Geothermal Research 141 (2005) 109–122 111

Conversely, basalt flows typically 2–10 m thick are

devoid of such sheets but contain numerous other

structures, such as different types of segregation

vesicles, partly glassy thin upper sheets (S1 type),

10–40 cm thick S2-type sheets in the central part of

the flow, and vesicle cylinders.

The purpose of this paper is to document the

segregation structures within two pillows, respectively

from the Tore-Madeira Rise (TMR) and from the

Chile Triple Junction (CTJ) area. These hand-size

samples, ca. 15 and 30 cm in diameter, respectively,

expose a combination of the four types of segregation

structures (spherical vesicles, pipe vesicles, vesicle

cylinders, and vesicle sheets) up to now observed only

within basalt flows typically of 2- to 10-m thickness.

The only mechanism rapid enough to form these

structures in such small, fastly cooled pillows is vapor

differentiation. Our geochemical study reveals that the

differentiation trends from the host lavas to the

Fig. 2. Pillow lava TMD17b-1 from Tore-Madeira Rise (TMR). (a) Photogr

three samples analysed by ICP-AES is shown. (b) Sketch of the same cros

pillow lava. The curve shows the distribution of olivine phenocrysts throug

olivine cluster. (d) Sketch of the same cross-section.

segregation structures are different in the two cases

because of different crystallization sequences.

2. Studied samples

The Tore-Madeira Rise (TMR) is a seamount chain

ranging from 418 to 328N and from 128 to 188W, 300

km off the Portuguese coast (Fig. 1a; Laughton et al.,

1975). It was built up at first by a magmato-tectonic

event during the Barremo-Aptian time (119–108 Ma;

Tucholke and Ludwig, 1982; Peirce and Barton, 1991;

Olivet, 1996) followed by several magmatic episodes

during the Cenozoic (Wendt et al., 1976; Olivet, 1996;

Geldmacher et al., 2000; Cornen et al., in prepara-

tion). The petrological studies (Merle et al., in

preparation) show that the TMR magmatism is clearly

alkalic (alkali basalts, basanites, and trachytes). The

studied pillow has been dredged from the Josephine

aph of a vertical cross-section of the pillow lava. The location of the

s-section. (c) Photograph of another cross-section through the same

hout the log outlined by the rectangle. Inset: photomicrograph of an


seamount, located in the middle part of the rise (Tore-

Madeira mission, 2001). It was an entire 16-cm-large

pillow, grossly triangular (Fig. 2). The host lava, alkali

basaltic in composition, has a red-brown color

characteristic of low-temperature seawater alteration

under oxidizing conditions. It is olivine–phyric (ca. 10

modal %) and has a vesicular partly glassy ground-

mass, with a diktytaxitic texture in its central part. The

cylinders show the original orientation of the pillow as

illustrated in Fig. 2. From the base to the top of the

pillow, we observe segregation pipe vesicles, which

progressively become the cylinders, then rare thin

sheets (Fig. 2a,b). Spherical segregation vesicles are

found everywhere throughout the sample.

The Chile Triple Junction (CTJ) corresponds to the

area where a segment of the Chile spreading ridge

enters the Chile Trench near the Taitao Fracture Zone

(Fig. 1b; Bourgeois et al., 2000). The studied pillow

was dredged on a small quaternary volcanic edifice

Fig. 3. Details of segregation structures in the pillow lavas from Tore-Made

vertical cross-section of the fragment of the CTJ pillow lava, showing pip

three samples analysed by ICP-AES is shown. (b) Photograph of a vertical

the transition between pipe vesicles and vesicle cylinders. (c) Photomicrog

segregation spherical vesicles of TMD17b-1.

located on the continental edge of the trench close to

CTJ (CTJ28 dredge; CTJ cruise of R/V L’Atalante,

1997; analyses in Guivel et al., 2003). It is a fresh

basal fragment of a pillow of basaltic andesite

composition, with a 2-mm-thick glassy margin (Fig.

3a). The host lava is glomerophyric, with plagioclase-

rich olivine-bearing clusters (ca. 15 modal %). The

partly glassy groundmass has a highly vesicular and

diktytaxitic texture. The segregation structures appear

grossly similar to those in the TMR pillow (Fig. 3a,b).

3. Segregation structures

3.1. Segregation pipe vesicles

Pipe vesicles occur either near the base of basaltic

flows (Walker, 1987; Stephenson et al., 2000) or in

pillow lavas (Philpotts and Lewis, 1987). In the latter

ira Rise (TMR) and Chile Triple Junction (CTJ). (a) Photograph of a

e vesicles, vesicle cylinders, and bubble waves. The location of the

cross-section of the fragment of the TMD17b-1 pillow lava showing

raph of a vesicle cylinder of TMD17b-1. (d) Photomicrograph of a


case, they have a radial distribution. They never occur

in the chilled margin but only in the interior of the

pillow that contains crystals (Philpotts and Lewis,

1987). Two interpretations have been advanced to

relate these two occurrences. In lava flows, pipe

vesicles have been attributed to gas bubbles buoyantly

ascending through lava at a stage when it has acquired

a yield strength sufficient to prevent closure behind

them (Walker, 1987). Philpotts and Lewis (1987) have

proposed an alternate model to explain the radial

position of pipe vesicles in pillow lavas. They point

out that formation of a long pipe of gas requires

continuous exsolution of gas onto bubbles attached to

the solidification zone. During cooling of the lava,

continued exsolution of gas causes the pipes to grow

normal to the solidification front. Thordarson and Self

(1998), then Caroff et al. (2000), have shown for

continental lava flows that pipe vesicles could be

partly filled by glassy segregation material.

In the TMR and CTJ pillow lavas, pipe vesicles are

clearly segregation structures, partly or entirely filled

by vesicular glass. They occur only at the base of the

pillow, perpendicular to the chilled crust (5 mm above

in the TMR pillow, 1 cm above in the CTJ one; Figs. 2

and 3a,b). Although this latter feature has been

described by Philpotts and Lewis (1987), the fact that

such structures are limited to the base of the pillow

lava is consistent with Walker’s (1987) model.

Consequently, we suggest a two-step formation of

pipe vesicles. Pipe vesicles start to form according to

the model proposed by Philpotts and Lewis (1987)

followed by buoyancy-driven rise according to

Walker’s model (1987). In other words, the zone of

solidification might act as a guide to orientate the

bubbles at the beginning of their ascent.

3.2. Vesicle cylinders

Vesicle cylinders are the structures which are

probably the most characteristic of a vapor differ-

entiation processes. They are vertical tubes, 2–20 cm

in diameter, filled with residual liquid and bubbles

(Goff, 1977). According to the classification of

Caroff et al. (2000), they occur typically within 2-

to 10-m-thick lava flows, but they were also

observed in the lower half of some thicker inflated

lava flows (Self et al., 1997; Thordarson and Self,

1998; Stephenson et al., 2000). In continuous

exposures, cylinders extend from ca. 25 cm above

the flow base to the bottom of the upper chilled crust

(Goff, 1996; Caroff et al., 2000). To form these

vertical structures, Goff (1996) has proposed that

residual liquids generated within the lower solid-

ification zone move into vesicle-rich low-density

areas, through a gas filter-pressing mechanisms, then

migrate towards the top of the flow. Once trapped

beneath the crust, cylinders end either in sill-like

sheets or in vesicular pods, from whose the differ-

entiated liquid invades the chilled crust as thinner

vesicle sheets. In some inflated lava flows, Thordar-

son and Self (1998) have observed segregation pipe

vesicles which converge to form cylinders. These

features have also been observed in non inflated lava

flows by Goff (1996) and Caroff et al. (2000).

In the pillows described here, we have also

observed the transition between pipe vesicles and

cylinders. In the TMR pillow, the lowest cylinders

resemble greatly elongated, partly filled pipe vesicles,

1–3 mm wide and 2.5 cm long (Figs. 2 and 3b,c).

They end in rounded bulbs located in the middle part

of the pillow lava. Towards the top of the pillow, the

cylinders become wider (4–5 mm in diameter) and full

of segregated melt. The upper extremity of some

cylinders is connected to sill-like sheets or to diffuse

veinlets (Fig. 2a,b). We observe a clear convergence

of the cylinders towards the central area of the TMR

pillow. This noticeable feature cannot be related to the

mechanism responsible of the pipe orientation, normal

to the basal crust, because some cylinders do not

follow the trend initiated down by the pipes (Fig.

2b,c). We propose that solidification of the two lateral

chilled crusts takes place just before the residual melt

migrates towards the top of the pillow. As its

consequence, cylinders are forced to converge

towards the still molten central part of the pillow.

The vesicle cylinders of the CTJ pillow appear to be

relatively disconnected from the pipe zone (Fig. 3a).

In addition, all the CTJ cylinders have a morphology

similar to the lower TMR cylinders (i.e., small

bulbous structures).

3.3. Vesicle sheets

The vesicle sheets described by Goff (1996) and

Caroff et al. (2000) in 2- to 10-m-thick lava flows are

located either within (or just below) the upper chilled


crust (S1-type) or in the central part of the flow (S2-

type). They are filled by vesicular, (partly) glassy

material, which has a composition grossly similar to

the cylinder-filling material. After Goff (1996), the

S1-type sheets either spread below the crust or invade

incipient joints and cracks created at the beginning of

the crystallization history.

We have observed vesicle sheets (2–5 mm thick)

and veinlets (b2 mm) in the TMR pillow lava, where

they occur in the external part of its upper half (Fig.

1). They are generally horizontal structures diverging

towards the lateral edges of the pillow through

incipient cracks of the outer solidified zone. The

thickest sheets are very vesicular. At the left border of

the pillow, one of these thick sheets connected to a

degassing chimney (Fig. 2c,d). We suggest that gas

escaped outside from the thick sheet towards the left

side of the pillow.

3.4. Segregation spherical vesicles

Segregation vesicles have been studied by Smith

(1967), Baragar et al. (1977), Bideau et al. (1977),

Shibata et al. (1979), Bideau and Hekinian (1984),

and Bacon (1986). Anderson et al. (1984) have

proposed a gas filter-pressing model to explain melt

migration into bubbles within certain subaerial basalt

flows. Caroff et al. (2000) have proposed a morpho-

logical classification of the vapor-differentiation-

derived segregation vesicles. The most widespread

ones are the V1-type spherical vesicles, which are

relatively ubiquitous throughout the vapor-differenti-

ated lava bodies, except in their basal part where they

are absent. They are b1 cm in diameter and partly

filled with glassy, differentiated material.

The TMR and CTJ pillows contain such V1-

vesicles, increasing in size and decreasing in abun-

dance upwards (Fig. 3d). The CTJ pillow displays

waves of bubbles such as described, for instance, by

McMillan et al. (1987) in a N70-m-thick basalt flow.

These layers have been interpreted by Manga (1996)

as the result of an hydrodynamic process. In his

model, suspensions of bubbles initially homogeneous

become unstable and form rising waves of bubbles.

The specificity of bubble waves in the CTJ pillow is

that they are formed by segregation vesicles placed

edge to edge. Such bubble layers are mainly observed

in the lower part of the pillow. Upward layers

resemble to the TMR vesicle sheets. However, it

seems that there is no connection between these upper

layers and cylinders.

4. Petrology and geochemistry

4.1. Textural and mineralogical notes

Microprobe analyses were performed with a

Cameca SX50 automated electron microprobe (Micro-

sonde Ouest, Brest). Analytical conditions were 15

kV, 15 nA, counting time 6 s, correction by the ZAF

method. Concentrations of b0.3% are considered

qualitative.

The pillow from the Tore-Madeira Rise has an

olivine–phyric, partly glassy groundmass. Numerous

microlites of plagioclase (An63-70) occur together

with sparser, partly iddingsitized olivine (Fo80-81)

and Fe–Ti oxide microcrystals. Clinopyroxene was

not detected, probably because of the small size of the

crystals and the slight alteration of the groundmass.

Some olivine phenocrysts are assembled as clusters

b3 mm in diameter. Variation of the olivine phenoc-

ryst abundance (crystals/cm2) from the base to the top

of the TMR pillow is shown in Fig. 2c. The curve is

serrated with two more pronounced peaks at 3.5 and

10 cm from the base, respectively. The lower peak,

with 28 olivines per cm2, corresponds to the transition

zone between pipe vesicles and cylinders. The upper

peak, with 36 olivines per cm2, coincides with the area

where the vesicle cylinders feed into sheets. Between

these two high values, the curve displays a wide

saddle-like depression reaching six olivines per cm2.

Such features (variation of the vertical olivine

phenocryst distribution and occurrence of clusters)

have been previously described by Caroff et al. (1997)

in a 20-m-thick basaltic lava flow. In this case, olivine

distribution was interpreted to be the result of

interference between the upward motion of the

residual melt plus gas and the downward, density-

related settling of olivine. Such a model can be

adapted to the TMR pillow lava. In this view, the

upper peak may be interpreted, on one hand, as a

break of the downward motion of olivines from the

top of the pillow by ascending vesicle cylinders, as

suggested by the olivine depletion near the top of the

pillow (Fig. 2c), or/and, on the other hand, as an

Table 1

ICP-AES major and trace element analyses of TMR and CJT host lavas and microprobe glass analyses of segregation structures

Sample TMD17b-1B TMD17b-1C TMD17b-1G 16 27 35 64 75 83 60 CTJ28B CTJ28C CTJ28G 160 68 113 125 129 130 139 153

Site TMR TMR TMR TMR TMR TMR TMR TMR TMR TMR CTJ CTJ CTJ CTJ CTJ CTJ CTJ CTJ CTJ CTJ CTJ

Structure type HL C HL SP SP C SV SV S S HL C HL CC SP SV SV S S C C

EBD (Am) – – – 10 10 10 5 5 5 5 – – – 10 5 10 10 10 10 10 10

wt.%

SiO2 42.00 45.00 42.25 53.42 45.50 53.55 46.34 50.24 48.59 54.67 53.00 54.20 53.00 53.23 47.03 55.66 50.83 48.05 61.45 62.26 46.38

TiO2 2.47 3.21 2.37 2.60 3.58 2.03 4.40 4.99 2.92 1.94 0.94 1.13 0.92 0.97 1.41 0.22 0.74 1.29 0.58 0.73 1.87

Al2O3 14.85 14.55 15.30 17.75 10.87 16.05 10.62 15.07 13.33 16.56 17.80 15.80 17.75 17.44 9.84 24.20 14.27 7.74 17.51 20.20 9.28

Fe2O3*/FeO* 14.25 13.62 13.74 9.59 12.87 7.47 14.25 13.83 9.91 7.27 6.96 8.15 6.92 6.70 11.17 1.77 11.34 12.20 4.45 2.05 13.36

MnO 0.15 0.18 0.15 0.23 0.21 0.31 0.19 0.17 0.09 0.01 0.12 0.15 0.12 0.26 0.25 0.00 0.17 0.30 0.10 0.01 0.25

MgO 8.70 4.74 8.20 1.42 6.41 3.48 4.02 0.87 5.33 2.45 6.38 6.40 6.43 6.05 14.65 1.34 12.70 13.81 2.59 0.90 12.18

CaO 8.20 11.40 8.60 6.29 15.59 10.35 13.50 5.54 14.19 8.36 9.15 9.00 9.20 9.44 15.90 10.85 6.44 15.32 6.71 6.86 15.87

Na2O 2.56 3.35 2.72 7.00 2.63 5.86 4.43 7.67 4.05 6.36 3.27 3.00 6.10 3.08 0.30 5.04 3.01 1.14 4.33 4.72 0.49

K2O 0.61 0.82 0.59 0.61 0.45 0.55 0.58 0.89 0.30 0.85 0.65 0.80 0.87 0.74 0.02 0.28 0.29 0.09 1.40 0.97 0.07

P2O5 0.38 0.82 0.44 0.54 0.57 0.59 0.70 0.80 0.83 0.51 0.23 0.28 0.24 0.35 0.18 0.03 0.25 0.41 0.29 0.39 0.37

Cr2O3 – – – 0.00 0.00 0.05 0.10 0.08 0.10 0.04 – – – 0.11 0.06 0.00 0.05 0.09 0.00 0.07 0.20

LOI 5.92 2.59 5.66 – – – – – – – 1.32 – 1.26 – – – – – – – –

Total 100.09 100.28 100.02 99.45 98.68 100.29 99.13 100.15 99.64 99.02 99.82 98.91 99.81 98.37 100.81 99.39 100.09 100.44 99.41 99.16 100.32

ppm

Rb 7.1 6.8 5.8 – – – – – – – 21.0 22.5 18.6 – – – – – – – –

Sr 400 425 415 – – – – – – – 206 175 204 – – – – – – – –

Ba 154 242 192 – – – – – – – 118 152 124 – – – – – – – –

Sc 25.0 37.0 26.0 – – – – – – – 23.0 27.5 23.0 – – – – – – – –

V 242 340 258 – – – – – – – 146 183 150 – – – – – – – –

Cr 386 136 412 – – – – – – – 156 182 157 – – – – – – – –

Co 48 35 56 – – – – – – – 25 27 27 – – – – – – – –

Ni 226 41 256 – – – – – – – 90 80 100 – – – – – – – –

Y 22.5 34.0 23.0 – – – – – – – 26.0 32.5 26.5 – – – – – – – –

Zr 175 210 180 – – – – – – – 157 188 155 – – – – – – – –

Nb 30.0 38.0 31.0 – – – – – – – 8.0 10.0 8.0 – – – – – – – –

La 17.0 24.5 16.5 – – – – – – – 12.9 15.6 12.8 – – – – – – – –

Ce 39 51 37 – – – – – – – 28 35 28 – – – – – – – –

Nd 23.0 31.5 23.0 – – – – – – – 15.2 19.2 15.0 – – – – – – – –

Sm 5.5 7.2 5.6 – – – – – – – 3.5 4.4 3.7 – – – – – – – –

Eu 1.91 2.51 1.95 – – – – – – – 1.17 1.38 1.17 – – – – – – – –

Gd 5.45 7.70 5.50 – – – – – – – 4.30 4.35 4.20 – – – – – – – –

Dy 4.60 6.60 4.55 – – – – – – – 4.45 5.40 4.30 – – – – – – – –

Er 2.10 3.10 2.00 – – – – – – – 2.60 3.10 2.50 – – – – – – – –

Yb 1.67 2.55 1.68 – – – – – – – 2.52 3.09 2.50 – – – – – – – –

Th 1.70 2.40 1.80 – – – – – – – 2.70 3.45 2.85 – – – – – – – –

HL: host lava; CC: chilled crust; SP: segregation pipe; C: cylinder; S: sheet; SV: spherical vesicle; EBD: electron beam diameter.

Fe2O3*/FeO*: total iron expressed as Fe2O3 for ICP-AES analyses and as FeO for microprobe analyses.

R.Merle

etal./JournalofVolca

nologyandGeotherm

alResea

rch141(2005)109–122

115


influx of olivines carried away by ascending cylin-

ders, consistent with the saddle-like depression at the

level of the cylinder zone. The lower peak could

correspond to lower limit of this process.

Caroff et al. (1997, 2000) point out that high-

temperature iddingsite (HTI, i.e., magmatic alteration

of olivine phenocrysts as the result of f02 increase) is

commonly observed in vapor-differentiated basaltic

lavas. Although altered, the TMR olivines do not

seem to contain HTI. The main secondary minerals

identified are low-temperature iddingsite, zeolites, and

various clay-minerals.

The pillow lava from the Chile Triple Junction

contains phenocrysts of plagioclase (An88–89), plus

Fig. 4. Al2O3 versus MgO diagrams (ICP-AES and microprobe analyses).

Rise pillow lava. (b) Host lava and glassy segregation structures in the C

scale. The composition of olivine microcrystals in equilibrium with host lav

Fields of microprobe analyses of plagioclase and clinopyroxene microlit

enlarged scale. Composition of olivine microcrystals in equilibrium with h

the TMR pillow lava. Fields of microprobe analyses of plagioclase micro

subordinate olivine (Fo86–87). The groundmass con-

sists of similar phases (plagioclase An63–86 and

olivine Fo84–86) with glass. The phenocrysts are

arranged in clusters (glomerophyric texture). The CTJ

pillow is perfectly fresh and contains no alteration

minerals, including HTI.

The segregation material in both pillows is mainly

glassy with a few microcrystals of plagioclase,

clinopyroxene, and Fe–Ti oxides.

4.2. Geochemical variations

We present in Table 1 six ICP-AES analyses (two

host lava and one cylinder analyses for each pillow).

(a) Host lava and glassy segregations structures in the Tore-Madeira

hile Triple Junction pillow lava. (c) TMR diagram with an enlarged

a has been calculated with the formula of Roeder and Emslie (1970).

es and olivine phenocrysts are indicated. (d) CTJ diagram with an

ost lava have been calculated following the same procedure than for

lites and olivine microcrystals and phenocrysts are indicated.


Analytical methods are described by Cotten et al.

(1995). Relative standard deviations are b2% for

major elements and b5% for trace elements. Numer-

ous microprobe analyses of glass have also been

performed in the CTJ chilled crust and in the

segregation structures of both pillows (Table 1).

Loss on ignition (LOI) values show that the TMR

host lava is relatively altered (TMD17b-1B: LOI=5.92

wt.%; TMD17b-1G: LOI=5.66 wt.%), whereas the

cylinders are fresher (TMD17b-1C: LOI=2.59 wt.%).

CIPW normative nepheline percentages range from

0.37 (host lava) to 3.53 wt.% (cylinders), which

correspond to mildly alkali basalt values. The cylinders

have a mean composition clearly more evolved than the

host lava (e.g., MgO=4.74 vs. 8.45 wt.%, respectively).

Both ICP-AES and microprobe analyses are shown in

the Al2O3 versus MgO diagram of Fig. 4a together with

the field of TMR basalts. The two host lava analyses

(TMD17b-1B and-1G) plot within the TMR basalt

field. Two parallel trends are evident, the first one

including analyses of pipe vesicles, cylinders (microp-

robe and ICP-AES data) and sheets, the other one

including only spherical segregation vesicles. Primi-

Fig. 5. Primitive mantle-normalized trace element patterns. Normalization

and from Sun (1982) for compatible elements. (a) Incompatible element pa

(b) Incompatible element patterns of CTJ28B, CTJ28C, and calculated c

TMD17b-1C, and calculated cylinder values. (d) Compatible element patt

tive-mantle normalized trace element patterns (normal-

ization values from Sun, 1982, for compatible

transition elements, and from Sun and McDonough,

1989, for incompatible elements) are shown in Fig.

5a,b for the samples TMD17b-1G and-1C. The two

samples are enriched in the most incompatible ele-

ments with respect to the moderately incompatible

elements, which is characteristic of alkali basalts.

Cylinders, enriched in all incompatible elements with

respect to the host lava, display a pattern more or less

parallel to the host lava, except for Sr (Fig. 5a). On the

other hand, the cylinders are depleted in compatible

transition elements (Co, Cr, Ni) with respect to the host

lava (Fig. 5b).

The CTJ host lava (CTJ28B: LOI=1.32 wt.%;

CTJ28G: LOI=1.26 wt.%) is very fresh, as well as the

cylinders (LOI of CTJ28C not determined because of

the little quantity of material). The three CTJ

compositions plot within the field of basaltic andesite

in the TAS diagram of Le Bas et al. (1986). They have

calc–alkaline affinities (Guivel et al., 2003). There is

no discrepancy between microprobe and ICP-AES

analyses of the CTJ glassy crust (BV160 and CTJ28B

values for incompatible elements from Sun and McDonough (1989)

tterns of TMD17b-1G, TMD17b-1C, and calculated cylinder values.

ylinder values. (c) Compatible elements patterns of TMD17b-1G,

erns of CTJ28B, CTJ28C, and calculated cylinder values.


in Table 1). The CTJ cylinders also have a mean

composition more evolved than the host lava, but this

is not expressed through the MgO values, which

remain remarkably constant in the three analyses.

TiO2, P2O5, and the incompatible trace elements

values show the extent of differentiation. Thus, it is

possible to estimate the weight percent of residual

liquid from the ratio (Th host lava/Th cylinders), Th

being considered to be the most incompatible element

(see the detailed procedure in Caroff et al., 1993). The

result is 0.80 for the CTJ pillow lava, wihch is close to

that calculated for the TMR sample (0.73). Both ICP-

AES and microprobe analyses are shown in the Al2O3

versus MgO diagram of Fig. 4b, together with the

field of CTJ28 dredge. The two host lava analyses

(CTJ28B and G) plot within the CTJ28 dredge field.

Like for the TMR pillow, spherical vesicles plot

slightly apart from the other segregation structures

(except one analyse), but in this case, in contrast to the

TMR, they form a trend on the right side of the other

data. The incompatible trace element patterns (shown

for CTJ28B and CTJ28C) are less enriched in the

most incompatible elements than the TMR ones and

display a negative anomaly in Nb (Fig. 5c), a feature

characteristic of calc–alkaline series. The compatible

element patterns display a surprising Cr enrichment in

the most differentiated liquid (CTJ28C), whereas Ni

Fig. 6. Cartoon illustrating four stages in the formation of

decreases slightly from the host lava to the cylinders

(Fig. 5d).

5. Geochemical evolution between and within the

segregation structures

Variations of Al2O3 versus MgO (Fig. 4) show

complex differentiation modalities, which can be dealt

in two stages. First, we should consider the transition

from host lava to spherical vesicles through pipes/

cylinders/sheets (Tinter-trendr evolution), then the

differentiation within the trends themselves (Tintra-

trendr evolution). The fundamental difference between

the two types of differentiation is that the Tinter-trendr

evolution involves extraction and transport of residual

liquids from the host, whereas the Tintra-trendr

evolution corresponds to in situ crystallization within

one given segregation structure. In the latter case, glass

compositions are highly dependant on the nature of the

neighbouring crystallizing mineral phases.

To explain the Tinter-trendr evolutions, it is

necessary to determine the groundmass phase assem-

blage crystallizing in the whole host lava. In the TMR

pillow lava, it consists mainly of plagioclase and

olivine, as observed in thin sections. However, the

decrease in Cr from host lava to cylinders (Table 1)

a segregation vesicle (modified after Sanders, 1986).

Fig. 7. AFM diagrams for glassy segregation structures in pillow

lavas. (a) Tore-Madeira Rise pillow lava. Inset: trends controlled by

different mineral assemblages. 1: Plagioclase+olivine+clinopyrox-

ene; 2: clinopyroxene; 3: Fe–Ti oxides+clinopyroxene; 4: Fe–Ti

oxides+plagioclase. The main trend corresponds to the crystalliza-

tion of the Fe–Ti oxides and clinopyroxene assemblage. (b) Chile

Triple Junction pillow lava. Inset: trends controlled by different

mineral assemblages. 1: Plagioclase+olivine; 2: plagioclase+Fe–Ti

oxides; 3: Fe–Ti oxides+olivine+clinopyroxene. This latter assem-

blage corresponds to the prevailing crystallization trend. Symbols as

in Fig. 4.


implies that clinopyroxene must also crystallize.

Mineral proportions have been estimated by using a

graphic method derived from the diagram of Fig. 4c. If

we fix the position of the crystallizing assemblage

somewhere in the AAV segment (point C), together

with the percentage of clinopyroxene, it is possible to

estimate the proportion of plagioclase and olivine

through simple mass balance. Our results are 15 wt.%

clinopyroxene, 33 wt.% olivine, and 52 wt.% plagio-

clase. To validate these estimates, a test has been

performed with trace elements. To recalculate the

cylinder composition, we have introduced into the

Rayleigh fractionation equation the following param-

eters: fraction of residual liquid (Th host lava/Th

cylinders), host lava composition, mineral proportions,

and distribution coefficients from the literature

(incompatible trace elements: Caroff et al., 1997; Co,

Ni, and Cr: Bedard, 1994; Henderson, 1986, except Ni

in olivine estimated from the formula of Hart and

Davies, 1978: D=[124/MgO]�0.9). The results shown

in the diagrams of Fig. 5a,b reproduce satisfactorily

the analysed data, except for Ni and, to a lesser extent,

heavy rare earth elements. A quasi similar mineral

assemblage might explain the mean composition of

spherical vesicles by evolution of the same initial host

lava (Fig. 4c). The mechanism of formation of such

segregation structure is shown in Fig. 6. The cartoons

illustrate how mean compositions of the segregated

liquids are dependant on the nature and the proportion

of minerals crystallizing in the extraction zone. The

position of the spherical vesicle trend relative to that of

the other segregation structures implies a more

important extent of differentiation (Fig. 4a,c).

A similar procedure has been followed for the

CTJ pillow lava. The crystallizing mineral assem-

blage is probably devoid of clinopyroxene, given the

incompatibility of Cr from host lava to cylinders

(Fig. 5d). It is situated at point C in the segment

linking olivine and plagioclase compositions (Fig.

4d). A simple lever-rule method gives the following

proportion: 88 wt.% plagioclase and 12 wt.%

olivine. The trace element test can be used to

validate these estimates, as shown in the diagrams

of Fig. 5c,d. Note in particular the good super-

position of the calculated and measured values for Sr

and Cr, which is consistent with their contrasted

compatibility. The position of the spherical vesicle

trend is not consistent with the crystallization of the

same mineral assemblage, involved in melt evolution

from host lava to the other segregation structures.

Crystallization of Fe–Ti oxides is required to explain

such compositions (Fig. 4d).

The Tintra-trendr evolution is controlled by the

mineral assemblages crystallizing within the segre-

gation structures in the vicinity of the analyzed zone.

Indeed, one microprobe pinpoint cannot deliver an

analysis representative of the bulk chemistry of the

segregated glasses. To identify which type of mineral

is responsible for a given evolution, we have plotted

the glass analyses in the AFM (alkali–iron–magne-


sium) diagrams of Fig. 7. In the TMR pillow, the

AFM diagram shows four trends of differentiation

that each corresponds to a specific groundmass

mineral assemblage. The main trend corresponds to

the crytallization of a Fe–Ti oxides–clinopyroxene

assemblage. One of these two phases is always

present in all trends. In the CTJ pillow, the AFM

diagram displays three trends. The dominant one

involves crystallization of Fe–Ti oxides, clinopyrox-

ene, and olivine. The diagrams show the crucial role

of certain mineral phases during this differentiation,

such as Fe–Ti oxides or clinopyroxene, which have

little influence in the transition from host lavas to

cylinders.

6. Formation of crystal frameworks in basaltic

pillows and lava flows

For Anderson et al. (1984), gas filter-pressing

involves the formation of a rigid and permeable

framework of crystals and locks the size and the shape

of the segregation vesicles. We calculated the degree

of crystallization for the melt evolution from host lava

to cylinder in the CTJ and TMR pillows. For the CTJ

pillow, we obtained 20% crystallization and 27% for

the TMR sample. This means that a crystal framework

strong enough to allow the formation of the segrega-

tion structures, forms earlier in the calc–alkaline rock

than in the alkaline one. In pillows, the crystal

framework occurs when crystallization reaches ca.

25%. In vapor-differentiated basaltic lava flows,

previous studies have estimated degrees of crystal-

lization of melt from host lavas to segregation

structures: from a alkali basalt host to cylinders, the

degree of crystallization is ca. 57% (Caroff et al.,

2000); from host lava to sheets, it ranges from 35% to

53% (Caroff et al., 1997, 2000); and from host to

spherical vesicles, 36% to 74% (Caroff et al., 2000;

Anderson et al., 1984). Thus, in vapor-differentiated

basaltic lava flows, crystallization should reach at

least 35% to form the crystal framework required for

generating vapor-differentiated segregation structures.

In the studied pillows, the melt in segregation

structures is less evolved than that in lava flows

(Caroff et al., 1997; 2000; Goff, 1996), probably as a

result of a lesser extent of differentiation in pillows,

due to their faster cooling.

7. Conclusions

(1) The two studied pillows from Tore-Madeira Rise

and Chile Triple Junction exhibit four types of

vapor-differentiation-related segregation struc-

tures (spherical and pipe vesicles, vesicle cylin-

ders, and vesicle sheets), previously observed

typically within 2- to 10-m-thick basalt flows.

(2) Formation of the segregation structure obeys

slightly different modalities specific for each

pillow lavas. For instance, vesicle sheets are not

generated by similar mechanisms. In the TMR

pillow lava, vesicle sheets spread through incip-

ient cracks of the outer solidified zone. In the CTJ

pillow lava, lower vesicles sheets appear to be

waves of bubbles caused by the formation of

instabilities within suspensions of bubble initially

homogeneous.

(3) Geochemical variations in the segregation struc-

tures are different. The main difference between

the two pillows can be attributed to theirmagmatic

affinity, which influences liquidusmineral phases.

The variations of Al2O3 versus MgO show two

differentiation modalities. The first one considers

the differentiation of melt from host lava to

spherical vesicles through pipes/cylinders/sheets

(Tinter-trendr evolution), and the second one, the

differentiation within the segregation structures

themselves (Tintra-trendr evolution). Within each

pillow, the segregated glass within spherical

vesicles displays a mean composition slightly

different from that in the other structures, which

denotes different extents and/or modalities of

differentiation. Composition of the segregated

liquid in the spherical vesicles is dependant on the

nature and the proportion of minerals crystallizing

in the extraction zone. The Tinter-trendr evolution

is controlled by the mineral assemblages crystal-

lizing within the segregation structures, in the

vicinity of the analyzed zone.

(4) We have calculated the degree of crystallization

corresponding to the transition between host lava

and cylinders in each pillows. We deduced the

degree of crystallization necessary to form a

crystal framework required for generating the

segregation structures. It is lower in pillows (ca.

25% crystallization) than in vapor-differentiated

basaltic lava flows (at least 35% crystallization).


Acknowledgements

The samples were dredged during the Tore-

Madeira Cruise, conducted by G. Cornen, and the

Chile Triple Junction Cruise, conducted by Jacques

Bourgeois. We thank H. Loyen and E. Boeuf for thin

sections of TMR and technical assistance, A.

Cossard for pictures and J.P. Oldra for thin sections

of CTJ. We also acknowledge Y. Lagabrielle who

provided the CTJ sample, the assistance of M. Bohn

for microprobe studies and F. Jourdan and U.

Sch7rer for helpful comments. Critical reviews by

Drs. A.R. Philpotts and F. Goff helped us in

improving this work. This work has greatly benefited

of the comments of Dr. B.D. Marsh as editor.

Contribution no. 933 of the IUEM, European

Institute for Marine Studies (Brest, France).

References

Anderson, A.T., Swihart, G.H., Artioli, G., Geiger, C.A., 1984.

Segregation vesicles, gas filter-pressing, and igneous differ-

entiation. J. Geol. 92, 55–72.

Bacon, C.R., 1986. Magmatic inclusions in silicic and intermediate

volcanic rocks. J. Geophys. Res. 91, 6091–6112.

Baragar, W.R.A., Plant, A.G., Pringle, G.J., Schau, M., 1977.

Petrology and alteratoin of selected units of mid-Atlantic Ridge

basalts sampled from sites 332, DSDP. Can. J. Earth Sci. 14,

837–874.

Bedard, J.H., 1994. A procedure for calculating the equilibrium

distribution of trace elements among the minerals of cumulate

rocks, and the concentration of trace elements in the coexisting

liquids. Chem. Geol. 118, 143–153.

Bideau, D., Hekinian, R., 1984. Segregation vesicles of ocean floor

basalts 1. J. Geophys. Res. 89, 7903–7914.

Bideau, D., Hekinian, R., Francheteau, J., 1977. Orientation of

ocean floor basaltic rocks at time of cooling: A general method.

Contrib. Mineral. Petrol. 65, 19–28.

Bourgeois, J., Guivel, C., Lagabrielle, Y., Camus, T.,

Boulegue, J., Daux, V., 2000. Glacial–interglacial trench

supply variation, spreading-ridge subduction, and feedback

controls on the Andean margin development at the Chile

Triple junction area (458488S). J. Geophys. Res. 105,

8355–8386.

Caroff, M., Maury, R.C., Leterrier, J., Joron, J.L., Cotten, J., Guille,

G., 1993. Trace elements behaviour in the alkali basalt:

Comenditic trachyte series from Mururoa Atoll, French Poly-

nesia. Lithos 30, 1–22.

Caroff, M., Ambrics, C., Maury, R.C., Cotten, J., 1997. From alkali

basalt to phonolite in hand-size samples: Vapor-differentiation

effects in the Bouzentes lava flow (Cantal, France). J. Volcanol.

Geotherm. Res. 79, 47–61.

Caroff, M., Maury, R.C., Cotten, J., Clement, J.-P., 2000.

Segregation structures in vapor-differentiated basaltic flows.

Bull. Volcanol. 62, 171–187.

Cotten, J., Le Dez, A., Bau, M., Caroff, M., Maury, R.C., Dulsky, P.,

Fourcade, S., Bohn, M., Brousse, R., 1995. Origin of anomalous

rare-earth elements and yttrium enrichments in subaerially

exposed basalts: Evidence from french polynesia. Chem. Geol.

119, 115–138.

D’Orazio, M., Innocenti, F., Manetti, P., Tamoni, M., Tonarini, S.,

Gonzalez-Ferran, O., Lahsen, A., Omarini, R., 2003. The

Quaternary calk–alkaline volcanism of the Patagonian Andes

close to the Chile triple junction: Geochemistry and petrogene-

siss of volcanic rocks from the Cay and Maca volcanoes (~458S,Chile). J. South Am. Earth Sci. 16, 219–242.

Geldmacher, J., Van den Bogaard, P., Hoernle, K., Schmincke,

H.U., 2000. The 40Ar/39Ar age dating of the Madeira

archipelago and hot spot track (eastern North Atlantic). Geo-

chem. Geophys. Geosyst. 1 (1999GC000018).

Goff, F., 1977. Vesicle cylinders in vapor-differentiated basalt

flows. PhD thesis, University of California, Santa Cruz, CA,

181pp.

Goff, F., 1996. Vesicles cylinders in vapor-differentiated basalt

flows. J. Volcanol. Geotherm. Res. 71, 167–185.

Greenough, J.D., Dostal, J., 1992. Cooling history and differ-

entiation of thick North Mountain basalt flow (Nova Scotia,

Canada). Bull. Volcanol. 55, 63–73.

Guivel, C., Lagabrielle, Y., Bourgeois, J., Martin, H., Arnaud, N.,

Fourcade, S., Cotten, J., Maury, R.C., 2003. Very shallow

melting of oceanic crust during spreading ridge subduction:

Origin of near-trench quaternary volcanism at the Chile Triple

Junction. J. Geophys. Res. 108 (B7), 2345.

Hart, S.R., Davies, K.E., 1978. Nickel partitioning between olivine

and silicate melt. Earth Planet. Sci. Lett. 40, 203–219.

Helz, R.T., 1980. Crystallisation history of Kilauea Iki lava lake as

seen in drill core recovered in 1967–1979. Bull. Volcanol. 43,

675–701.

Helz, R.T., Kirschenbaum, H., Marinenko, J.W., 1989. Diapiric

transfer of melt in Kilauea Iki lava lake, Hawaii: A quick,

efficient process of igneous differentiation. Geol. Soc. Amer.

Bull. 101, 578–594.

Henderson, P., 1986. Inorganic Geochemistry. Pergamon Press,

Oxford.

Le Bas, M.J., Le Maitre, R.W., Streckeisen, A., Zanettin, B., 1986.

A chemical classification of volcanic rocks based on the total

alkali–silica diagram. J. Petrol. 27, 745–750.

Laughton, A.V., Roberts, D.G., Graves, R., 1975. Bathymetry of the

northeast Atlantic: Mid-atlantic ridge to southwest Europe.

Deep-Sea Res. 22, 791–810.

Manga, M., 1996. Waves of bubbles in basaltic magmas and lavas.

J. Geophys. Res. 101, 17457–17465.

McMillan, K., Cross, R.W., Long, P.E., 1987. Two-stage vesicu-

lation in the Cohassett flow of the Grande Ronde Basalt, south-

central Washington. Geology 15, 809–812.

Olivet, J.L., 1996. La cinematique de la plaque Iberique. Bull. Cent.

Rech. Explor. Prod. Elf-Aquitaine 20, 131–195.

Philpotts, A.R., Lewis, C.L., 1987. Pipe vesicles: An alternate

model for their origin. Geology 15, 971–974.


Peirce, C., Barton, P.J., 1991. Crustal structure of the Tore-Madeira

Rise, Eastern North Atlantic-results of a DOBS wide-angle and

normal incidence seismic experiment in the Josephine Seamount

region. Geophys. J. Int. 106, 357–378.

Philpotts, A.R., Caroll, M., Hill, J.M., 1996. Crystal–mush

compaction and the origin of pegmatitic segregation sheets in

a thick flood-basalt flow in the Mesozoic Hartford Basin,

Connecticut. J. Petrol. 37, 811–836.

Puffer, J.H., Horter, D.L., 1993. Origin of the pegmatitic segrega-

tion veins within flood basalts. Geol. Soc. Amer. Bull. 105,

738–748.

Roeder, P.L., Emslie, R.F., 1970. Olivine–liquid equilibrium.


Rogan, W., Blake, S., Smith, I., 1996. In situ chemical fractionation

in thin basaltic lava flows: Examples from the Auckland

volcanic field, New Zealand, and a general physical model. J.

Volcanol. Geotherm. Res. 74, 89–99.

Sanders, I.S., 1986. Gas filter-pressing origin for segregation

vesicles in dykes. Geol. Mag. 123, 67–72.

Self, S., Thordarson, T., Keszthelyi, L., 1997. Emplacement of

continental flood basalt lava flows. Geophys. Monogr. 100,

381–410.

Shibata, T., DeLong, S.E., Walker, D., 1979. Abyssal tholeiites from

the oceanographer fracture zone 1: Petrology and fractionation.


Sisson, T.W., Bacon, C.R., 1999. Gas filter-pressing in magmas.

Geology 27, 613–616.

Smith, R.E., 1967. Segregation vesicles in basaltic lavas. Am. J. Sci.

265, 696–713.

Stephenson, P.J., Zhang, M., Spry, M., 2000. Fractionation

modelling of segregation in the Toomba Basalt, north Queens-

land. Aust. J. Earth Sci. 47, 291–300.

Sun, S.S., 1982. Chemical composition and origin of the earth’s

primitive mantle. Geochim. Cosmochim. Acta 46, 179–192.

Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic system-

atics of oceanic basalts: Implication for mantle composition and

processes. In: Saunders, A.D., Norry, M.J. (Eds.), Magmatism in

the Ocean Basins, Spec. Publ. - Geol. Soc. Lond. vol. 42. , pp.

313–345.

Thordarson, T., Self, S., 1998. The Roza Member, Columbia River

Basalt Group: A gigantic pahoehoe lava flow field formed by

endogenous processes. J. Geophys. Res. 103, 27411–27445.

Tucholke, B.E., Ludwig, W.J., 1982. Structure and origin of the J

Anomaly Ridge, Western North Atlantic Ocean. J. Geophys.

Res. 87, 9389–9407.

Walker, G.PL., 1987. Pipe vesicles in Hawaiian basaltic lavas: Their

origin and potential as paleoslope indicators. Geology 15, 84–87.

Wendt, I., Kreuzer, H., Muller, P., Von Rad, U., Raschka, H., 1976.

K–Ar age of basalts from Great Meteor and Josephine seamount

(eastern North Atlantic). Deep-Sea Res. 23, 849–862.

Segregation vesicles, cylinders, and sheets in vapor...

Documents

Transcript of Segregation vesicles, cylinders, and sheets in vapor...