You are on page 1of 45

4.

14 The Lower Oceanic Crust


LA Coogan, University of Victoria, Victoria, BC, Canada
ã 2014 Elsevier Ltd. All rights reserved.

4.14.1 Background 497


4.14.1.1 Introduction 497
4.14.1.1.1 Aims and approach 498
4.14.1.2 Setting the Scene 498
4.14.1.2.1 History of models of lower oceanic crust processes 498
4.14.1.2.2 MORB crystallization 501
4.14.2 Observations 503
4.14.2.1 MORB Differentiation 503
4.14.2.1.1 Phenocrysts, xenocrysts, and xenoliths in MORB 503
4.14.2.1.2 Global basalt differentiation trends in Mg# and chlorine: the influence of spreading rate 505
4.14.2.1.3 Do MORB differentiation trends record variable crystallization pressure? 506
4.14.2.2 The Mechanical and Thermal Structures of the Lower Oceanic Crust 507
4.14.2.2.1 Slow-spreading ridges 507
4.14.2.2.2 Fast-spreading ridges 508
4.14.2.2.3 Thermal history of the lower oceanic crust 511
4.14.2.3 The Composition of the Lower Oceanic Crust 512
4.14.2.3.1 Lithologies 512
4.14.2.3.2 Mineral compositions 513
4.14.2.3.3 Bulk-rock compositions 518
4.14.2.3.4 Chemical profiles through the oceanic crust 522
4.14.3 Generating the Lower Oceanic Crust 524
4.14.3.1 How Large and How Frequent Are Magma Additions to the Lithosphere? 525
4.14.3.2 How Well Aggregated Are Melts Added to the Lithosphere? 526
4.14.3.3 Where and Why Does Melt Pond in the Lithosphere? 526
4.14.3.3.1 Evidence for magma ponding at high pressures at slow-spreading ridges 527
4.14.3.3.2 Where does magma pond at fast-spreading ridges? 527
4.14.3.4 Thermal Constraints on Lower Crustal Processes 527
4.14.3.5 Crystallization, Mixing, Assimilation, and Magmatic Metasomatism 530
4.14.3.6 Near-Solidus Processes 531
4.14.3.7 Hydrothermal Circulation in the Lower Oceanic Crust 531
4.14.3.8 Working Models and Future Questions 531
Acknowledgments 534
References 534

4.14.1 Background about the compositions of melts extracted from the mantle
and their differentiation prior to eruption. The cooling of
4.14.1.1 Introduction
magma in the lower crust provides the heat to drive axial
The formation and recycling of oceanic lithosphere is the prin- hydrothermal systems (See Chapter 4.16), which, in turn,
cipal mechanism of planetary cooling on Earth and is probably play an important role in controlling seawater chemistry (see
responsible for much of the chemical heterogeneity in the Chapter 8.7). The lower crust constitutes the majority of the
mantle. Plate spreading at mid-ocean ridges (MORs) leads to oceanic crust (roughly 70% by mass, which makes it 20% of
near-adiabatic mantle upwelling which, in turn, leads to partial Earth’s entire crust), and thus, upon recycling into the mantle,
melting (See Chapter 4.13). Some of the melt generated in the it potentially becomes an important mantle ‘reservoir.’
mantle ascends faster than the surrounding mantle and crys- Because the lower oceanic crust forms at depth in a marine
tallizes within the thermal boundary layer near the surface, environment, its study is frustrated by inaccessibility. Detailed
forming oceanic crust. These melts do not erupt on the Earth’s sampling has occurred in a limited number of places (Figure 1;
surface where they would be quenched by seawater. Instead, Table 1) through dredging, drilling, and submersible opera-
they initially cool slowly in a (or multiple) magma chamber(s) tions that generally rely on tectonic dismemberment of the
leading to magma differentiation and forming the plutonic crust to expose lower crustal rocks at the seafloor. This tectonic
rocks that make up the lower crust. After variable amounts of dismemberment may, in some places, bias the sampling sub-
differentiation, a portion of the magma is extracted from these stantially. For example, most sampling of lower crust formed at
magma chambers to form the dikes and lavas of the upper slow-spreading ridges comes from regions where low-angle
oceanic crust. The lower oceanic crust contains information normal faults have exposed the lower crust in ‘core complexes.’

Treatise on Geochemistry 2nd Edition http://dx.doi.org/10.1016/B978-0-08-095975-7.00316-8 497


498 The Lower Oceanic Crust

30 60 90 120 150 180 -150 -120 -90 -60 -30

25 My isochron

60 60

DSDP Hole 334


Atlantis Massif
40 40
(IODP 1309D)
15N(ODP
MARK area 1268–1275)
Oman ophiolite 9° N EPR (ODP920–924)
20
Hess Deep
0 (ODP 894) 0
ODP 1256
Garrett
Vema
20 Transform 20
Transform
Atlantis Bank Pito Deep
(ODP 735B) & Terevaka
40 40
Transform

60 60

30 60 90 120 150 180 -150 -120 -90 -60 -30


Figure 1 Location map showing the areas discussed in this chapter (Table 1). The dashed lines show the 25 My isochrons as an indicator of the
spreading rate. Base map from Smith and Sandwell (1997) and isochrons from Muller et al. (2008).

Lower oceanic crust formed at slow-spreading ridges that ophiolites. Finally, as the chemical stratigraphy of the lower
produce ‘normal’ volcanic seafloor has not been sampled in oceanic crust is poorly known, evidence from ophiolites is also
detail. Geophysical studies have provided critical insights discussed. Most ophiolite data used here come from the Oman
into the structure of the ridge axis where a new crust is formed, ophiolite, as the lower oceanic crust is superbly exposed there
but detailed studies have only been carried out in a few loca- and has been studied in detail. This ophiolite is interpreted to
tions, such as at 9 –10 N on the East Pacific Rise (EPR; have formed at an intermediate- or fast-spreading ridge, al-
Figure 1; Crawford and Webb, 2002; Detrick et al., 1987; though the actual spreading rate is unknown. As more is learnt
Dunn et al., 2000). about modern oceanic crust, insights from ophiolites should
be critically evaluated.
4.14.1.1.1 Aims and approach
This chapter has two principal aims. Firstly, to summarize what
4.14.1.2 Setting the Scene
is known about the lower oceanic crust and, in particular, its
composition. These are the data that must be explained by The aim of this section is to provide the nonspecialist with the
models of crustal accretion and magma differentiation at ridges background required to understand the following sections –
as well as providing the basis for understanding large-scale much other critical information is covered in Chapters 3.3,
chemical processes (e.g., exchange with the ocean and recy- 4.13, 4.16, and 8.7.
cling into the mantle). This is the focus of Section 4.14.2. The
second aim is to define the processes that control the accretion 4.14.1.2.1 History of models of lower oceanic
and chemical differentiation of the lower oceanic crust, which crust processes
are discussed in Section 4.14.3. Prior to describing the obser- The rate of magma supply to a MOR plays a key role in
vational data, a brief introduction is provided for the non- controlling the processes operating in the lower oceanic crust
specialist (Section 4.14.1.2). principally because magma supply rate largely determines the
Because most ophiolites formed in suprasubduction zone thermal structure, which, in turn, influences rheology,
environments and the spreading rate of the ridges that they crystallization, and hydrothermal circulation. The principal
formed at are unknown, this review minimizes information control on the rate of magma supply is the spreading rate,
from ophiolites. However, ophiolites cannot be ignored with- and thus, spreading rate is a critical variable in models for the
out biasing the discussion unacceptably in three areas. Firstly, formation of the lower oceanic crust (e.g., Sinton and Detrick,
in discussing models for the formation of the lower oceanic 1992; Sleep, 1975). Mantle temperature and composition also
crust, it is impossible to avoid models developed from ophio- impact magma supply rates but generally to a lesser extent than
lite studies without giving an unbalanced view. Secondly, so spreading rate. There is a wide range of spreading rates for the
little is known about the structure of the lower oceanic crust at global ridge system (Figure 1), but for simplicity and due to
fast-spreading ridges that it is necessary to use insights from lack of data, the following discussion divides models for lower
Table 1 Information on the main sampling areas of lower oceanic crust discussed in the text

Location Full Maximum Principal sampling Dominant lithologies Notes


spreading height of
rate section
(mm year1) sampled
(m)

Atlantis Bank, 13 1510 (Hole ODP Holes 735B and 1105A (Legs Mainly evolved gabbros and abundant oxide This is a 3 km by 8 km platform of gabbro adjacent to the
31 S, SWIR 735B) 118, 176, and 179), wireline gabbros. Only two dikes were recovered in Atlantis II Transform exposed in a core complex. No
drilling, submersible, and dredging Hole 735B – both near the top intermixed peridotite has been reported
MARK area 23 <100 ODP Leg 153, Sites 921–924, short Evolved and primitive gabbros
(23 N, MAR) cores
Atlantis Massif 24 1410 IODP Site 1309D Very primitive gabbros and troctolites, rare Drill hole into a core complex. Structural relationship to upper
(30 N, MAR) evolved lithologies. Dikes throughout, but crust is unknown
many more at top
15 N MAR 25 200 ODP Sites 1268–1275, short cores. Peridotite intruded by a mixture of evolved and Multiple drill holes either side of 15 N transform. Site 1275 is
Wireline drilling primitive gabbro. Some cores only gabbro on top of a core complex and contains the most gabbro
Hess Deep, 130 900 (on ODP Site 894 (Leg 147) on the North wall and samples from near the top of the On the northern scarp the upper gabbros are continuous into
2 N, 101 W northern intrarift ridge and submersible intrarift ridge: mainly evolved gabbros. the sheeted dike complex, and their structural position is
scarp) operations and dredges on the Samples from deeper on the intrarift ridge well constrained. On the intrarift ridge tectonic
intrarift ridge and the north wall include more primitive gabbros dismemberment makes reconstruction of crustal position
uncertain
Pito Deep 144 900 Submersible Troctolites and troctolitic gabbros, more evolved Well-established structural context from submersible studies
23 S, 112 W gabbros in upper few hundred meters
ODP Hole 1256D 200 100 ODP/IODP Legs 206, 309, and 312 Evolved gabbros Good structural context with overlying thin sheeted dike
7 N, 92 W complex and lava section
Oman ophiolite Intermediate Full crust Field Upper few hundred meters of gabbros are Southern part of ophiolite appears to be most relevant to

The Lower Oceanic Crust


or fast and upper generally evolved and deeper gabbros mainly studies of ridge-related processes (minimum subduction
mantle moderately primitive influence)

499
500 The Lower Oceanic Crust

crustal accretion into end-member fast- and slow-spreading models envisaged a large, ever-present, molten chamber
rates (roughly 80 mm year1 and 30 mm year1, respec- (Figure 2(a)) that fed the dikes and lavas of the upper crust
tively). The focusing of melt extraction from the mantle in and from which the plutonic rocks of the lower crust
space and time (e.g., to segment centers) can also impact the crystallized – the so-called infinite onion model (Cann,
supply of magma to MOR. 1974). In this model, the upper plutonic rocks form by down-
Models for the formation of the lower oceanic crust have ward freezing from the roof and the lower plutonic rocks form
changed dramatically over the past 30 years (Figure 2). Early from crystals settling out of the magma. Later thermal

Infinite onion
Lavas
Dikes

Isotropic
gabbro

(a) Layered gabbro

Slow-spreading ridges Fast-spreading ridges


Infinite leek Lavas
Dikes
Lavas
Dikes
Plutonic
rocks
Plutonic
Plutonic
(e) rocks
rocks
(b) Gabbro glacier
Axial heat flux Axial magma
0 ~30 MW km-1 chamber
Plutonic
rocks Scale (approximate)
except part (c) Lavas
0 2 4 Dikes
Depth (km)

15 Plutonic
Km Lateral heat flux
rocks
Magma ~40 MW km-1

Crystal mush
30 (f)
Solidified pluton
Mantle Sheeted sill
Axial magma
(c) 45
Isochron Axial heat flux chamber
Fault ~15 MW km-1
Plum pudding
Upper-crust Lateral heat flux
~55 MW km-1
Seismic
lower-crust

(d) (g)

Figure 2 Historical view of models for the formation of the lower oceanic crust: (a) the ‘infinite onion’ model (Cann, 1974) in which an ever-present
magma chamber generates the lower oceanic crust through crystals plating its margin; (b) the ‘infinite leek’ model (Nisbet and Fowler, 1978) in which
the lower oceanic crust is constructed from numerous small magma chambers; (c) a deep crystallization model in which the lower oceanic crust
crystallizes at depth and is brought upward and smeared out (Elthon et al., 1982); (d) the ‘plum pudding’ model (Cannat, 1993, 1996) in which the lower
oceanic crust at slow-spreading ridges is constructed from a number of nested plutons that may crystallize within the mantle or crust; (e) a ‘large magma
chamber’ model (Pallister and Hopson, 1981) based on structural and chemical data from the Oman ophiolite; (f) a ‘gabbro glacier’ model (Henstock,
2002; Henstock et al., 1993; Phipps Morgan and Chen, 1993a,b; Quick and Denlinger, 1993) in which the lower oceanic crust crystallizes in a small sill at
the base of the sheeted dike complex from which cumulates subside down to form the lower crust; (g) a ‘sheeted sill’ model (Boudier et al., 1996;
Kelemen et al., 1997) in which the lower oceanic crust forms through the crystallization of multiple sills. Note the dramatically different distribution of
hydrothermal heat extraction in the latter two models due to the difference in where the latent heat of crystallization of the lower crust is released.
The Lower Oceanic Crust 501

modeling showed that this situation was unlikely and certainly suggested lower crustal flow, but did not model it. An alterna-
could not be the case at slow-spreading ridges, where such a tive model was proposed by Boudier et al. (1996), building on
body would freeze (Kuznir and Bott, 1976; Sleep, 1975). the models of Reuber (1990) and Bedard (1991), in which
Thermal constraints led to the development of models for some crystallization occurs in situ in the lower crust. This
lower crustal accretion at slow-spreading ridges that allowed model was developed on the basis of the occurrence of sill-
the accretion of the lower oceanic crust from magma chambers like plutonic bodies within the upper mantle section of the
that are either smaller or crystallize deeper than in the infinite Oman ophiolite, which were interpreted to have crystallized
onion model. In the ‘infinite leek’ model, small magma in situ. By analogy, they suggested that some of the lower
batches are added to the crust, forming many small ‘nested’ gabbros may have formed in situ too. Kelemen and coworkers
intrusions (Figure 2(b); Nisbet and Fowler, 1978). An alterna- (Kelemen and Aharanov, 1998; Kelemen et al., 1997; Korenaga
tive model proposed that crystallization could occur at depth, and Kelemen, 1997, 1998) extended this model, suggesting
where temperatures are higher, with the cumulates being that almost the entire lower oceanic crust crystallizes in situ
‘dragged’ up by mantle flow to form the lower oceanic crust in a ‘sheeted sill’ complex (Figure 2(g)). These models require
(Figure 2(c); Elthon et al., 1982). Based largely on observa- extensive hydrothermal cooling of the lower oceanic crust
tions of gabbroic bodies hosted in peridotites on the seafloor at along the sides of the mush zone in order to remove the latent
slow-spreading ridges and thermal constraints, a model inter- heat of crystallization on axis and prevent a large molten
mediate between these has become popular. In this model, the region from forming (Chen, 2001).
crust at slow-spreading ridges is made up of small magma
bodies emplaced into mantle peridotite at a variety of depths –
a model sometimes referred to as a ‘plum pudding’ (Figure 2 4.14.1.2.2 MORB crystallization
(d); Cannat, 1993, 1996). In this model, the ‘crust’ is no longer At least since O’Hara (1965, 1968a) demonstrated that mid-ocean
composed entirely of crystallized melt but includes screens of ridges basalts (MORBs) have compositions that lie on low-
mantle peridotite, making it difficult to use seismic measure- pressure cotectics and that crystallization trends follow these
ments to determine crustal thickness (Cannat, 1993, 1996). cotectics, low-pressure magma chambers have been known to
Another variant in these themes is described by Schwartz et al. be important in controlling basalt compositions and, thus,
(2005), where the lower crust is constructed both from the in forming the lower oceanic crust. A combination of experimental
nested shallow-level plutons and from the products of studies, textures of basalts and plutonic rocks, and geochemical
deeper-seated crystallization. The crystallization of plutonic studies provides evidence for a general low-pressure crystallization
rocks at depth within the mantle (Figure 2(c) and 2(d)) re- sequence in MORB of olivine ! olivine þ plagioclase ! olivine
quires the mantle temperature to be below the liquidus of the þplagioclase þ clinopyroxene ! plagioclase þ clinopyroxene
melt passing through it. þ orthopyroxene þ FeTi oxides (e.g., Grove et al., 1992 and
Thermal constraints on generation of the lower oceanic references therein). Cr spinel is a common but minor
crust took longer to influence magma chamber models for high-temperature phase that may be replaced by Cr-rich
fast-spreading ridges because the original thermal models con- clinopyroxene through a melt–cumulate reaction. Typically,
sidered only conductive cooling. In this case, large magma low-pressure experiments using primitive MORB as starting
chambers could exist at fast-spreading ridges. For example, materials become olivine saturated at  1220  C, plagioclase
Pallister and Hopson (1981) and Smewing (1981) proposed saturated at 1210  C, and clinopyroxene saturated a fur-
models based on observations in the Oman ophiolite that ther 30–40  C cooler; these temperatures and temperature in-
included 20-km-wide and 4-km-high magma chambers tervals differ between studies. The mass crystallized per degree
(Figure 2(e)). These models were developed largely to explain cooling is highest when plagioclase  clinopyroxene join the
the observed limited compositional variation and systematic liquidus (Kelemen and Aharanov, 1998). This means that
steepening of foliation and layering orientations with height the amount of heat that must be removed per degree of cooling
in the lower crust in the Oman ophiolite. However, such is greatest at this stage. Late-stage phases include apatite, zircon,
models were disproved by seismic data, which shows that a quartz, and amphibole (e.g., DeLong and Chatelain, 1990; Gillis
small (1-km-wide and  50-m-deep) magma chamber overlies et al., 2003; Natland et al., 1991). Immiscible sulfides precipitate
a crystal-rich mush zone at fast-spreading ridges (e.g., Detrick over a wide range of temperatures.
et al., 1987; Sinton and Detrick, 1992; Section 4.14.2.2.2). The Simplified phase diagrams for MORBs suffer from significant
‘gabbro glacier model’ was developed to reconcile the forma- loss of information through projecting ten or so significant
tion of 4 km of gabbros from such a thin magma chamber. In components onto a flat surface and ignoring the solid solution
this model, most crystallization occurs within the shallow within mineral phases; however, useful basic insight into
magma chamber, losing the latent heat of crystallization of MORB crystallization can still be gained from them (Figure 3).
the plutonic complex to the overlying hydrothermal system, As indicated by the crystallization sequence outlined above,
and a crystal mush subsides downward and outward, forming most MORBs start in the olivine field and fractionate to the
the lower oceanic crust (Figure 2(f); Henstock et al., 1993; plagioclase cotectic and follow this to clinopyroxene saturation.
Phipps Morgan and Chen, 1993a; Quick and Denlinger, A number of factors can influence the saturation temperature of
1993; see also Nicolas et al., 1988, 1993). These models repro- these principal phases, for example, by changing the phase
duce the foliation orientations and the lack of compositional equilibria or by causing the melt to follow a differentiation
variation with height in the crust that the ‘large magma cham- path that does not track the cotectics.
ber’ models (Figure 2(e)) were developed to explain. Gabbro Variations in the major element composition of the paren-
glacier models built on the thermal model of Sleep (1975) tal melt will alter the fractionation path by changing both the
502 The Lower Oceanic Crust

Parental melt composition


Clinopyroxene

High sodium
Low sodium

Cpx

Silica

Plagioclase

(a)
Olivine

Crystallization pressure
Clinopyroxene

Low pressure
High pressure

Cpx

Silica

Plagioclase

(b) Olivine
Crystallization process
Clinopyroxene
[Plagioclase]

Fractional crystallization
Replenishment/mixing

In situ crystallization
Mi ecti s
co gma
xin c
ma
t

Parental magma
g

Olivine Orthopyroxene Silica


(c)

Figure 3 Schematic phase diagrams based on the parameterizations of Yang et al. (1996) and Longhi (1991), illustrating the roles of (a) parental melt
composition, (b) pressure, and (c) crystallization process (projected from plagioclase) in controlling the crystallization sequence and modal proportions
during MORB crystallization. Increased melt Na content tends to depolymerize the melt, leading to an increase in the primary phase volumes of less
polymerized minerals (e.g., olivine) relative to those of more polymerized minerals (e.g., plagioclase). Increased pressure expands the clinopyroxene
stability field. Mixing of two olivine–plagioclase–clinopyroxene saturated magmas can lead to a hybrid magma in the clinopyroxene field (Dungan and
Rhodes, 1978). In situ crystallization and magma chamber replenishment can lead to melts that have precipitated olivine–plagioclase–clinopyroxene
plotting in the olivine (plagioclase) field (Langmuir, 1989).

position of the phase boundaries and the compositions of the Kushiro, 1975; Longhi, 1991). Alternatively, increasing the
crystallizing phases. For example, a higher Na content of the melt Ca/Al of a melt causes it to reach the clinopyroxene–
shrinks the plagioclase primary phase volume relative to that of olivine–plagioclase cotectic after a smaller total amount of
olivine and makes the plagioclase that crystallizes more albitic crystallization (e.g., at a higher melt Mg# ¼ molar (Mg/
(Figure 3(a); Biggar and Humphries, 1981; Grove et al., 1992; (Mg þ Fe2þ))). Both the global range in average primary
The Lower Oceanic Crust 503

MORB composition (e.g., Dick et al., 1984; Klein and Langmuir, signature could inadvertently be interpreted as indicating
1987) and the local variations in melt compositions will, thus, crystallization at high pressure.
lead to subtle differences in the crystallization path followed by Irrespective of the processes controlling the initial accumu-
MORBs and lead to changes in mineral compositions and their lation of crystals, an element of ‘in situ’ crystallization is re-
proportions in the cumulates. For example, troctolite cumulates quired for solidification. This is because once the crystallinity
of high Na melts will be more plagioclase-rich than those of reaches 50%, magmas are likely to become rheologically
Na-poor melts (Grove et al., 1992). The minor element compo- solid (e.g., Marsh, 1981). However, geochemical data, such as
sition of the melt can also play a role in controlling the crystal- the abundance of incompatible elements, indicate that most
lizing assemblage, especially when the minor component is plutonic rocks do not ‘trap’ this amount of melt. Instead, a
strongly partitioned into one phase. Cr may be an important portion of this pore-filling melt must be removed by compo-
minor element in the MORB system as it is strongly partitioned sitional convection (e.g., Tait et al., 1984) or compaction (e.g.,
into both spinel and clinopyroxene and will increase the stability Hunter, 1996). Replenishment, eruption, and assimilation are
of these phases over the Cr-free system (e.g., Onuma and Tohara, all processes that occur in MOR magma chambers. These open
1983). Although MORBs are dry relative to most other terrestrial system processes can lead to a diverse array of end products,
magmas, a role for water in controlling differentiation paths has dependent on the relative magnitudes of the mass fluxes and
been proposed (e.g., Michael and Chase, 1987) due to its sup- their relative timings (Figure 3(c); O’Hara and Herzberg,
pression of plagioclase crystallization (e.g., Feig et al., 2006; 2002), and can affect the heat budget of and temperature
Kushiro, 1975). within MOR magma chambers.
Pressure also plays an important role in controlling phase
equilibria. With increasing pressure, the clinopyroxene pri-
mary phase volume expands with respect to both olivine and 4.14.2 Observations
plagioclase (Grove et al., 1992; Kushiro, 1969; O’Hara, 1968b;
4.14.2.1 MORB Differentiation
Stolper, 1980; Figure 3(b)). An increased pressure also leads to
the crystallization of more albitic plagioclase, although this MORBs partially crystallize prior to being erupted. Their Mg#s
may be substantially offset by changes in the SiO2 content (Section 4.14.2.1.2) are almost always too low to be in equilib-
of the melt (Grove et al., 1992). Interpretation of MORB dif- rium with mantle olivine, and their low-pressure cotectic com-
ferentiation trends and mineral compositions in the lower positions are far from saturation with all mantle phases at any
oceanic crust as recording evidence for high-pressure crystalli- pressure (O’Hara, 1965, 1968a; Stolper, 1980). The numerous
zation is largely based on the increased stability of clinopyrox- detailed studies of MORBs (compared to relatively few studies of
ene at higher pressure. For example, the interpretation that oceanic plutonic rocks) provide important insights into the
early clinopyroxene crystallization requires high-pressure crys- crystallization that generates the lower oceanic crust. More
tallization was the principal reason for the development of the specifically, they provide insight into that stage of the crystalli-
model shown in Figure 2(c) (see Sections 4.14.2.1.3 and zation process at which eruptible melts are present. Inverting the
4.14.3.3.1). compositional trends observed in MORB suites to understand
Crystallization processes play an important role in control- lower crustal processes is complicated by the range of melt
ling the composition of both the differentiating melt and result- compositions generated beneath MOR, both globally and lo-
ing cumulates (Figure 3(c); e.g., O’Hara and Herzberg, 2002). cally (e.g., Dick et al., 1984; Johnson et al., 1990; Klein and
As magma bodies crystallize, many different processes can be Langmuir, 1987). Here, two observations are used to investigate
involved in crystal–melt separation (e.g., Jaupart and Tait, 1995; lower crustal processes that are largely unaffected by variations
Marsh, 1996; Naslund and McBirney, 1996), but for brevity, in parental melt composition. Firstly, the crystals carried by
only end-members are considered here. Magma chambers cool MORBs as phenocrysts, xenocrysts, glomerocrysts, or xenoliths,
largely by heat loss from their margins, and thus, crystallization which escaped incorporation into the lower oceanic crust, are
will occur along the boundaries, forming mush zones. Whether described. Secondly, the global variations in MORB Mg# and
crystallization also occurs throughout the interior of the magma chlorine content as a function of spreading rate are discussed,
body is uncertain (Hort et al., 1999; Marsh, 1996; Worster et al., based on the assumption that both are primarily controlled by
1990). If crystallization occurs largely in the main body of differentiation in the lower crust. Finally, the evidence for
magma, with crystals separating out (e.g., by settling), then the MORB differentiation and, hence, lower crustal formation oc-
eruptible melt will follow a differentiation trend close to perfect curring at high pressure (up to 1 GPa) at slow-spreading ridges
fractional crystallization. Alternatively, if crystallization occurs are considered in Section 4.14.2.1.3.
mainly along the magma chamber margins, with the interstitial
melt returned to the main magma body, then the magma will 4.14.2.1.1 Phenocrysts, xenocrysts, and xenoliths in MORB
follow an ‘in situ’ crystallization pathway (Langmuir, 1989). In Crystallization products that escape incorporation into the
this case, phases can be saturated in the magma chamber lower oceanic crust and end up in lavas provide snapshots of
boundary layer but not in the eruptible magma volume, leading the physicochemical conditions during lower crustal forma-
to differentiation trends that do not follow the cotectics tion. Plutonic xenoliths are rare, having only been described in
(Figure 3(c); Klein, 2003). For example, in situ crystallization a few studies from intermediate- to fast-spreading ridges. Glo-
could produce a suite of basalts that show evidence of clinopyr- merocrysts and phenocrysts are more common and have been
oxene saturation, such as decreasing CaO with decreasing MgO, observed in lavas from ridges of all spreading rates. Crystal
that lie within the low-pressure olivine þ plagioclase phase vol- abundances are generally low in MORBs, with slightly higher
ume and have relatively high MgO contents; this geochemical abundances in basalts from slow-spreading ridges (Figure 4);
504 The Lower Oceanic Crust

plagioclase is generally the most abundant phenocryst phase, Meyer and Shibata, 1990; Nisbet and Fowler, 1978; Pan and
and olivine is the second most abundant, with clinopyroxene Batiza, 2003; Rhodes et al., 1979). There is a wide scatter in the
being relatively rare. Because most MORBs have lost  50% of ratio of plagioclase to olivine crystals in MORB suites, demon-
their initial mass prior to eruption (see Section 4.14.2.1.2), the strating that these are not all simple phenocrysts formed in a
low abundance of phenocrysts in most MORBs indicates that closed system, where the ratio would be defined by cotectic
the crystals are efficiently separated from the residual melt. proportions (e.g., Bryan, 1983). Instead, open system processes,
Phenocryst abundances record the probability of a magma such as mixing and/or crystal–melt separation, must have oper-
being erupted at a given crystallinity (Marsh, 1981). This, in ated. An overabundance of large crystals with respect to smaller
turn, is the product of the probability of a magma existing at any ones in at least some EPR lavas has also been ascribed to open
given crystallinity and the probability of a magma at any given system processes (Batiza and Niu, 1992).
crystallinity being eruptible (Marsh, 1981). As the latter seems Evidence for disequilibrium between crystals and their host
unlikely to vary as a function of spreading rate, the higher MORBs comes from their compositions. Both olivine and
proportion of phenocrysts in MORBs from slow-spreading plagioclase crystals in MORBs are commonly in equilibrium
ridges suggests that there is a higher probability of crystal-rich with melts more primitive than their host glass (Figure 5).
eruptible melts existing beneath slow-spreading ridges. Forolivine, in which Mg–Fe diffusion is rapid, major element
Xenoliths of incompletely crystallized lower crustal mate- equilibrium with the melt does not prove that that the crystal
rial provide rare insights into the lower crust beneath the ridge grew from the host; significant reequilibration between the
axis. The crystals making up xenoliths are not in equilibrium
with either their host basalt or their (different composition)
interstitial melt within the xenolith. This small-scale disequi-
librium likely results from magma mixing both within the
6
crystal mush zone and within the axial magma chamber 0.27
(AMC) at intermediate- to fast-spreading ridges (Davis and k d=

MgO/FeO olivine
5
Clague, 1990; Dixon et al., 1986; Hekinian et al., 1985; Ridley 0.30
k d=
et al., 2006). 4 0.33
Individual crystals in MORBs may be either phenocrysts k d=

or xenocrysts and these can be difficult to distinguish. Many 3


studies have emphasized the abundance of xenocrysts, based on
both compositional and textural evidence for disequilibrium MAR
2 EPR
between crystals and melts, and used this as evidence for
magma mixing in the crustal plumbing system at both fast-
and slow-spreading ridges (e.g., Costa et al., 2009; Dick and 0.6 0.8 1.0 1.2 1.4

Bryan, 1978; Dungan and Rhodes, 1978; Dungan et al., 1978; (a) MgO/FeO melt

ium
0.9 ibr
< 60 mm year -1 (n = 188) uil
70 Eq
> 60 mm year -1 (n = 241)
An observed

0.8
60
Percentage of samples

50 0.7

40 0.6
MAR
EPR
30
0.6 0.7 0.8 0.9
20 (b) An model

Figure 5 Comparison of the composition of (a) olivine and (b)


10 plagioclase crystals found in MORBs with the composition calculated to
be in equilibrium with their host glass using the data from the PetDB
database for crystals reported as phenocrysts. Olivine equilibrium
5 10 15 20 25 30 35 40 compositions are calculated assuming three different values of the
Mg–Fe exchange coefficient (eqn [2]) and that 88% of the Fe in the melt
Phenocryst abundance (%)
was FeO. Equilibrium plagioclase compositions are calculated using the
Figure 4 Histogram showing the abundance of crystals in MORBs from approach described in Appendix 2 of Grove et al. (1992), with their
fast- and slow-spreading ridges with an arbitrary split at 60 mm year1 uncertainties (gray field). Crystals in MORBs are generally either in
full spreading rate. MORBs erupted at fast-spreading ridges are generally equilibrium with their host or more primitive than the equilibrium
less crystal-rich than those from slow-spreading ridges (data from Dick composition. Plagioclase from the MAR is slightly further from
and Bryan, 1978; Hodges, 1978; le Roux et al., 2002; Pan and Batiza, equilibrium with their host (9% An on average) than those from
2003; Sato et al., 1978; Schilling et al., 1982; and the PetDB database). the EPR (6%).
The Lower Oceanic Crust 505

melt and crystal can occur in years (Chakraborty, 1997; Dohmen 4.14.2.1.2 Global basalt differentiation trends in Mg# and
et al., 2007). This is not the case for plagioclase, due to sluggish chlorine: the influence of spreading rate
coupled CaAl–NaSi exchange. Disequilibrium between crystals In a global compilation of MORB glass analyses, Melson et al.
and their host melt (Figure 5) reinforces the conclusion that (1976) noted that evolved compositions are more common at
magma mixing is important in the lower oceanic crust. However, fast- than at slow-spreading ridges, a result subsequently con-
random mixing of crystals and melt would not lead to the firmed (e.g., Sinton and Detrick, 1992) and reiterated here in
tendency for crystals in MORBs to be more primitive than Figure 6. The Mg# of melts generated within the mantle is
those in equilibrium with the host melt. Assuming that differen- likely to be almost invariant along the global MOR system
tial dissolution rates do not lead to rapid dissolution of evolved due to equilibrium with mantle olivine of approximately con-
crystals in primitive melts and substantially slower dissolution stant Mg#. This means that the observed change in average
of primitive crystals in evolved melts, this suggests that mixing Mg# with the spreading rate is due to greater extents of differ-
is generally between a more evolved phenocryst-poor magma entiation prior to eruption at faster spreading ridges. The var-
and a more primitive phenocryst-rich magma/mush. iability of glass Mg#s also generally increases with spreading
The time interval between the last magma-mixing event and rate (Figure 6). The decrease in Mg# with increased spreading
magma eruption can potentially be determined by modeling the rate could be explained by an 20  C temperature decrease
extent of diffusive reequilibration between crystals and magma. (Figure 6) leading to  20% difference in extent of crystalliza-
Although these studies are hindered by uncertainty in the initial tion (40% vs. 60% assuming fractional crystallization) of
and boundary conditions (e.g., whether the crystal was zoned the mantle-derived magma (cf. Rubin and Sinton, 2007). Al-
before mixing and whether growth or dissolution accompanied ternatively, the variation in average Mg# could reflect mixing
diffusion after mixing), it appears that mixing precedes eruption between melts of different end-member compositions or of
by days to years (Costa et al., 2009; Humler and Whitechurch, same end-member compositions but in different proportions
1988; Nableck and Langmuir, 1986; Pan and Batiza, 2002). at different spreading rates. For example, the lower average
To summarize, irrespective of the spreading rate, the phe- Mg# of MORBs from fast-spreading ridges can be explained
nocrysts, xenocrysts, and xenoliths found in MORB demon- by mixing of a larger proportion of evolved aphyric magma
strate that open system processes such as magma mixing, and a smaller amount of primitive phyric magma, consistent
crystal–melt separation, and mush–melt interaction are com- with the lower abundance of crystals in MORBs from fast-
mon in the lower oceanic crust. In general, during mixing, the spreading ridges (Figure 4). If the end-member compositions
crystal-poor end-member is more evolved than the crystal- are similar at all spreading rates, then the greater variability in
rich end-member (Figure 5). Higher abundances of pheno- MORB compositions at fast-spreading ridges suggests more
crysts in MORB from slow-spreading ridges suggest that either variable mixing proportions and a greater contribution from
crystal–melt separation is less efficient and/or there is a greater the more evolved end-member in the erupted magma. Alter-
proportion of the primitive, phyric magma in the mixture at natively, the evolved end-member may be more evolved at fast-
slow-spreading ridges. spreading ridges. In this scenario, the increased variability in

Mg# Temperature (°C)

0.61

0.59
1180
0.57

0.55 1170
0.53

0.51 1160

0.49
Mixing model
1150
0.47

20 40 60 80 100 120 140


Full spreading rate (mm year -1)
Figure 6 Average basalt Mg# vs. spreading rate for basalts from the Smithsonian database, filtered to exclude any samples with K2O/TiO2 > 0.15
(n ¼ 3344). The point that lies at low Mg# for its full spreading rate (50 mm year1) is controlled by a large number of evolved samples from the Juan de
Fuca ridge. Samples from this spreading rate bracket from the Indian Ocean fall on the trend defined by the other spreading rates (shown as an
open symbol). The temperature is calculated from the Mg#, based on the correlation of MgO and temperature given by Wilson et al. (1988), and is shown
as a guide rather than as an accurate temperature. One standard deviation of the basalt Mg# is also shown. Basalt compositions become systematically
more evolved but also more variable with increased spreading rate. The gray background shows an example of how this variability could be
explained by mixing between evolved and primitive melts. The primitive end-member in the model has a Mg# of 70 and the evolved end-member varies
from 50 to 32 with increasing spreading rate. The proportion of the evolved end-member is 50  15% (1s) at all spreading rates. The increased
variability at the more evolved end of the spectrum simply reflects the more evolved composition of the end-member in the mix.
506 The Lower Oceanic Crust

Mg# at fast-spreading ridges could simply be a consequence of begins at high pressure beneath slow-spreading ridges but at low
the larger difference in composition between the endmember pressure at fast-spreading ridges (Elthon, 1993; Elthon et al.,
compositions (Figure 6). 1982, 1992; Grove et al., 1992; Herzberg, 2004; Kelemen
The second global signal that can be observed in MORB glass et al., 2004; Michael and Cornell, 1998; Villiger et al., 2007).
compositions, without concern for systematic variation in pa- Quantitative estimation of crystallization pressure from MORB
rental magma compositions, is the overenrichment of chlorine compositions or MORB differentiation trends requires that both
in MORB glasses from fast-spreading ridges compared to that the phase equilibria are well enough constrained and that the
expected for the extent of crystallization (Jambon et al., 1995; crystallization process is understood; it is not clear that these
Michael and Cornell, 1998; Michael and Schilling, 1989; Perfit requirements are currently met (see also Section 4.14.3.3).
et al., 1999). MORB glasses from slow-spreading ridges do not The basic premise behind using MORB compositions to
show this chlorine overenrichment; thus, it cannot simply be determine the pressure of crystallization revolves around the
due to interaction of the magma with seawater during dike or increased stability of clinopyroxene with increasing pressure
lava emplacement. Instead, this observation is consistent with (Figure 3(b); Grove et al., 1992; Herzberg, 2004; Michael and
assimilation of one or more components that have interacted Cornell, 1998; Villiger et al., 2007). Estimates of the crystalli-
with hydrothermal fluids (brine, altered roof, and wall rocks) zation depths of MORBs erupted at slow-spreading ridges
into magma chambers. This interpretation is supported by the range up to 1 GPa. This has been interpreted to indicate
observation that the extent of chlorine overenrichment is great- crystallization within a thick, cool lithosphere, even at the
est at propagating ridge tips, where it is expected that the most ridge axis (Grove et al., 1992; Herzberg, 2004; Michael and
assimilation of hydrothermally altered material into magma Cornell, 1998; Villiger et al., 2007).
chambers will occur (Michael and Schilling, 1989). Addition- Figure 7 shows that calculated crystallization pressure for a
ally, enrichments in chlorine in magmatic amphibole in oceanic global suite of MORBs correlates strongly with major element
gabbros from fast- but not slow-spreading ridges fit well with parameters that are thought to record the extent of mantle
this interpretation (Section 4.14.2.3.2.2). This indicates that melting required to produce the basalt (Na8, which is the basalt
assimilation of material that has interacted with hydrothermal Na2O content corrected for fractionation to 8 wt% MgO; Klein
fluids is more common at fast- than at slow-spreading ridges. It and Langmuir, 1987, and basalt Ca/Al; notably, Fe8 does not
also demonstrates that hydrothermal circulation occurs close correlate with calculated crystallization pressure; cf. Michael
enough to the margins of magma chambers at fast-spreading and Cornell, 1998). High crystallization pressures are calcu-
ridges to allow assimilation of altered material. lated for melts that are apparently generated by small extents of
melting. There are two explanations for this observation.
The simplest is that crystallization pressure and extent of
4.14.2.1.3 Do MORB differentiation trends record variable melting could be correlated. For example, thick, cold litho-
crystallization pressure? sphere may cause cessation of melting at high pressure, gener-
The pressure at which MORBs begin to crystallize is also the ating smaller percent melting and more Na-rich parental melts.
depth at which the lower oceanic crust begins to form and The same thick, cold lithosphere could lead to high-pressure
provides a minimum pressure for the top of the mantle melting crystallization. However, Fe8 and crystallization pressure might
column. Numerous studies have suggested that crystallization be expected to correlate if this were the case. Alternatively, the

Increasing extent of melting? Increasing extent of melting?

Slow-spreading ridges
Calculated crystallization pressure (GPa)

1.0 Fast-spreading ridges

0.8

0.6

0.4

0.2

-0.2

1.5 2.0 2.5 3.0 0.6 0.7 0.8 0.9


Na8 CaO/Al2O3

Figure 7 Correlated Na8 (Na at 8 wt% MgO) and CaO/Al2O3 and calculated crystallization pressure for the same large MORB glass dataset shown in
Figure 6. Calculated crystallization pressure is determined using the approach of Villiger et al. (2007), which does not take into account the Na content of
the melt, but a similar (slightly weaker) correlation is found using the approaches of Herzberg (2004) or Yang et al. (1996). The histograms on the y-axis
show the crystallization pressures arbitrarily grouped by the spreading rate (slow <40 mm year1 and fast >60 mm year1).
The Lower Oceanic Crust 507

correlated parental melt composition and calculated crystalli- that are >1 km high and >10 km in extent along a flow line
zation pressure may be an artifact of either the effect of melt with little or no intermixed peridotite (Allerton and Tivey, 2001;
composition on phase equilibria and/or the differentiation Dick et al., 1991a; Expedition Scientific Party, 2005a,b;
process, which is not (fully) accounted for in the geobarom- Kinoshita and Dick, 2001). These observations are consistent
eters. For example, low-degree mantle melts with high Na8 and with magmatic accretion of the lower crust occurring in rela-
low Ca/Al may contain a smaller clinopyroxene component tively large and continuous bodies, at shallow levels, at least at
(and thus lower Ca and Ca/Al) when they are saturated some times, and in some places. However, microseismicity at
with clinopyroxene þ plagioclase þ olivine than higher-degree depths of up to 8 km at the ridge axis indicates that, elsewhere
melts. Indeed, the global correlation of MORB Ca/Al with axial on slow-spreading ridges, the lithosphere is cool (e.g., Toomey
depth (Klein and Langmuir, 1987) suggests that the Ca/Al of et al., 1988). This spatial variability suggests that episodic and/or
a clinopyroxene–plagioclase–olivine saturated melt is not sim- spatially focused (e.g., to the segment center) magma supply
ply controlled by crystallization pressure. In this second sce- is important at slow-spreading ridges. Seismic velocities
nario, most MORBs may have last equilibrated with olivine þ indicative of largely unaltered mantle peridotite are reached by
plagioclase þ clinopyroxene at low pressures (0.3 GPa). 10 km depth, suggesting that most hydrothermal circulation
and melt crystallization is constrained to less than this depth at
the ridge axis, even at very slow-spreading ridges (e.g., Sleep and
4.14.2.2 The Mechanical and Thermal Structures of the
Barth, 1997).
Lower Oceanic Crust
Uranium–lead ages of zircons from plutonic rocks of the
4.14.2.2.1 Slow-spreading ridges lower oceanic crust provide information about the timing of
Thermal models for (very) slow-spreading ridges predict that the solidification of these rocks (because zircon crystallizes near to
thermal boundary layer above the adiabatically upwelling mantle or below its closure temperature for Pb loss). Recent studies at
melting column is thicker than the crust (i.e., the thickness of melt three locations (Atlantis Bank, Atlantis Massif, and the Vema
produced; e.g., Bown and White, 1994; Reid and Jackson, 1981; fracture zone; Baines et al., 2009; John et al., 2004; Grimes
Sleep and Barth, 1997). A thick thermal boundary layer is also et al., 2008; Lissenberg et al., 2009; Schwartz et al., 2005) have
suggested by models for the formation of the axial graben at slow- demonstrated that the majority (90%) of zircons have about
spreading ridges, which require a thick, strong lithosphere at the the same age (þ0.25/0.5 Ma) as the crust (Figure 8). This
axis (Chen and Morgan, 1990; Phipps Morgan and Chen, 1993b). suggests that most of the gabbroic rocks that have been dated
If there is a thermal boundary layer above the upwelling mantle (zircon-bearing ones) crystallized at shallow crustal levels, at
melting column that is thicker than the crust, then melts will the ridge axis, and were transported horizontally at the half
begin to cool and, hence, crystallize within mantle lithosphere spreading rate from their time of formation. Those 6% of the
(e.g., Cannat, 1993, 1996). Three observations suggest that this samples that contain zircons that are substantially older than
occurs. Firstly, mantle peridotites at slow-spreading ridges com- the crustal age (by 0.6–2.5 Ma) can be explained either by
monly contain gabbroic and dolerite intrusions (e.g., Cannat,
1993, 1996; Cannat and Casey, 1995; Kelemen et al., 2004).
Secondly, the seismic Moho is less well defined at slow- compared
to fast-spreading ridges, suggesting a more gradual transition from
gabbro to peridotite (Mutter and Karson, 1992; Sleep and Barth, 60
1997). Thirdly, lower crustal seismic velocities systematically de-
crease with increasing crustal age at slow-spreading, but not at
Percentage of ages

50
fast-spreading, ridges. Assuming that the fast-spreading ridge
+ One age at −2.5 My

lower crust is representative of plutonic lower crust, then these 40


data can be explained by progressive serpentinization of  10%
Anomalously young

mantle peridotite intermixed with plutonic rocks (Sleep and


Anomalously old

30
Barth, 1997). How much crystallization occurs within the mantle
lithosphere depends on the thermal structure, the rate of melt 20
transport through the lithospheric mantle, and the extent of melt
focusing to the axis. Hot lithosphere, rapid melt transport, and 10
focused magma supply to the axis will all lead to less crystalliza-
tion within the lithospheric mantle and vice versa. Recent ad-
vances in understanding the crystallization depth of plutonic -1.5 -1.0 -0.5 0.0 0.5 1.0
rocks at slow-spreading ridges have come both from geological Age of crust − age of zircon
and geophysical field studies and from dating zircon crystals from
plutonic rocks. Figure 8 Histogram of the differences between ‘crustal age’ and zircon
age for oceanic gabbros from slow-spreading ridges. ‘Crustal age’ is
Crustal level melt sills that are a few kilometers wide and
defined by magnetic isochrons for zircons from the Atlantis Bank (Baines
long and mush zones that may be several kilometers high and
et al., 2009; John et al., 2004; Schwartz et al., 2005), a calculated
may have pipe- or dike-like forms have been identified in mag- spreading rate based on correlated zircon ages and distance from the ridge
matically robust regions of the Mid-Atlantic Ridge (MAR) using axis for zircons from the Vema fracture zone (Lissenberg et al., 2009) and
geophysical techniques (Dunn et al., 2005; Magde et al., 2000; an estimate based on distance off axis and spreading rate for Hole 1309D
Seher et al., 2010; Singh et al., 2006a; Sinha et al., 1997). (Atlantis Massif; Grimes et al., 2008). One age of 2.5 Ma older than the
Likewise, geological studies have identified gabbroic bodies crust has been left off the plot to avoid compressing the scale.
508 The Lower Oceanic Crust

crystallization at depths of up to 18 km, followed by tectonic suggested that some very olivine-rich troctolites from IODP
unroofing before being incorporated into younger crust or by Hole 1309D may be peridotites that have reacted with large
portions of shallow-level lower crust being trapped at the axis amounts of basaltic melt (Drouin et al., 2010; Suhr et al.,
by subsequent intrusions (cf. Elthon et al., 1982; Schwartz 2008). For evolved magmas to intrude peridotites, there must
et al., 2005). The 4% of samples that contain zircons that be large volumes of more primitive plutonic rocks elsewhere
are substantially younger than the crustal age (by 0.4–1.1 Ma) and the mantle rocks must have been below the solidus tem-
most likely were intruded off axis (Schwartz et al., 2005). perature (900  C for the most evolved apatite- and zircon-
Plutonic rocks from slow-spreading ridges generally show bearing rocks). These observations suggest that basaltic
weak or no magmatic foliation and little modal layering, al- magmas commonly crystallize within the lithospheric mantle
though grain size layering, which can be sharp or gradational at slow-spreading ridges but generally in regions warm enough
and fine or coarsen upward, is common (Bloomer et al., 1991; that centimeter- to meter-scale bodies cool slowly enough to
Cannat et al., 1995; Expedition Scientific Party, 2005a,b; form coarse-grained cumulates and, in some cases, that strain
Kelemen et al., 2004; Robinson and Von Herzen, 1989). can be partitioned into these bodies.
Crystal plastic deformation, starting in the granulite facies,
and subsolidus strain-induced foliations are common in 4.14.2.2.2 Fast-spreading ridges
some oceanic gabbros from slow-spreading ridges (e.g., ODP Direct observation and sampling of the seafloor via submersible,
(Ocean Drilling Project) Hole 735B) but much less so in others dredging, and drilling shows that at fast-spreading ridges the
(e.g., IODP (Integrated Ocean Drilling Program) Hole 1309D). seafloor is composed almost entirely of lavas. Dikes, plutonic
Crystal plastic deformation ranges from undulose extinction of rocks, and mantle peridotites are only exposed in ‘tectonic
magmatic silicates through subgrain formation and, in some windows’ where the crust has been tectonically dismembered
cases, mylonite formation (e.g., Cannat, 1991; Cannat et al., (e.g., Hess Deep and Pito Deep; Francheteau et al., 1990; Karson
1995, 1997; Dick et al., 1991b, 2000; Expedition Scientific et al., 2002) or at transform faults (e.g., Constantin, 1999).
Party, 2005a,b; Kelemen et al., 2004). In the 1500-m-deep Tectonic windows reveal a relatively simple layered structure
drill core at Atlantis Bank (ODP Hole 735B), the extent of for oceanic crust formed at fast-spreading ridges.
crystal plastic deformation is highest at the top, consistent In contrast to slow-spreading ridges, at intermediate- to
with the most extreme deformation being related to the tec- fast-spreading ridges, an AMC is imaged by seismic reflection
tonic unroofing of this large gabbro massif. However, the studies along much of the ridge axis, for example, at >60% of
extent of plastic deformation in the deep drill core from the regions between 9 –13 N and 13 –21 S, EPR (Detrick
IODP Hole 1309D is only slightly higher at the top, suggesting et al., 1987; Hooft et al., 1997). This is a minimum value
that unroofing of the lower oceanic crust can occur in different because the AMC may go undetected in some locations due
ways in different places. Cannat et al. (1997) report that the to either the ship wandering off axis or the AMC not lying
most extreme deformation is observed in moderately evolved directly beneath the axis. The AMC is 1-km wide and 50-m
plutonic rocks (gabbros, gabbronorites, and oxide gabbros), deep and can be continuous along axis for tens of kilometers
with the most primitive (troctolites) and most evolved (Detrick et al., 1987; Hooft et al., 1997; Kent et al., 1990; Singh
(dioritic) lithologies being less deformed in the MARK area et al., 1998). The AMC is underlain by a region with low shear
(Mid-Atlantic Ridge south of the Kane transform). These and compressional wave velocities (the low-velocity zone),
observations have been interpreted as suggesting that crystal interpreted to be a partially molten region containing 20%
plastic strain is localized where late-stage evolved melts migrate melt (e.g., Dunn et al., 2000). A Moho reflection is observed at
through a crystal mush, chemically modifying it and forming the ridge axis at approximately the same depth as in the off axis,
more evolved lithologies (Cannat et al., 1997). However, indicating that the majority of crustal accretion occurs within a
by the time the most evolved silicic lithologies form, crystal narrow region at the ridge axis (Detrick et al., 1987; Singh et al.,
plastic deformation has ceased. Similar observations have been 2006a,b). Assuming that the low-velocity zone all ends up as
made in ODP Hole 735B, except that this core shows more plutonic rocks, the ratio of the depth to the AMC to the thick-
evidence for hydrothermal circulation accompanying amphib- ness of the crust provides the ratio of upper (basaltic) to total
olite facies deformation (Bloomer et al., 1991; Cannat, 1991; crust, by volume. For a 6–7-km-thick crust (e.g., Grevemeyer
Dick et al., 1991b; Stakes et al., 1991). et al., 1998; White et al., 1992), this means that 60–85% of the
Gabbroic bodies enclosed within peridotite are common oceanic crust is plutonic at fast-spreading ridges.
in some areas along slow-spreading ridges, comprising as As shown in Figure 9, both the maximum and minimum
much as 20–40% of the uppermost ‘crust’ in the magma- depths of the AMC decrease with increasing spreading rate, but
starved 15 N area on the MAR, based on drilling results from there is 1500 m of depth variation at a given spreading rate.
ODP Leg 209 (Kelemen et al., 2004). These intrusions can be The shallowing of both the maximum and minimum depths of
either more or less plastically deformed than the surrounding the AMC with increasing spreading rate at full spreading rates
peridotites (e.g., Cannat et al., 1995; Tartarotti et al., 1995) and >60 mm year1 is consistent with a thermal control on the
sometimes appear to have induced recovery processes in average depth of this body (Phipps Morgan and Chen, 1993b).
the surrounding peridotite (Ceuleneer and Cannat, 1997). The large range in AMC depth at a given spreading rate suggests
Gabbroic bodies within peridotites are commonly highly frac- that spatial and temporal variabilities in magma supply and/or
tionated (e.g., Cannat and Casey, 1995; Kelemen et al., 2004; hydrothermal cooling lead to variations in thermal structure as
Tartarotti et al., 1995), although very primitive troctolitic great as those induced by large changes in spreading rate.
intrusions into mantle peridotites have also been observed in Although some of this depth variation may relate to decreased
some places (e.g., Kelemen et al., 2004). It has also been magma supply toward segment ends (Macdonald et al., 1988;
The Lower Oceanic Crust 509

ODP Hole 1256D


ODP Hole 504B

Hess Deep
Pito Deep
Gabbro at Hess
Deep, Pito Deep
and ODP Site
1256
1000 Drill hole in
Depth below seafloor (m)

lavas and dikes

Depth (range)
2000 of AMC in a
given area
Tomographically
3000 identified top of
mush zone
Earthquake
4000 depth
Depth range of
40 80 120 160 200 microseismicity

Fullspreading rate (mm year-1)


Figure 9 Summary of geophysical and geological constraints on the depth of magma chambers at mid-ocean ridges and how this varies with
spreading rate, updated from Purdy et al. (1992). The stippled area highlights the range of depths of the AMC. The range of depth of seismic reflections
interpreted as the roof of a magma chamber comes from Babcock et al. (1998), Bruguier and Livermore (2001), Canales et al. (2005), Carbotte et
al. (1998), Collier and Sinha (1990), Detrick et al. (1987), Harding et al. (1989, 2000), Hooft et al. (1997), Kent et al. (1993a,b), Mutter et al. (1995), Navin
et al. (1998), Singh et al. (1998, 2006a, 2006b), Sinha et al. (1997), Van Ark et al. (2007), and Vera et al. (1990). Depths of mush zones come from
Dunn et al. (2005), Magde et al. (2000), Menke et al. (2002), and West et al. (2001). Microseismicity depths, which are assumed to generally be
shallower than any magma body, come from Barclay et al. (2001), Huang and Solomon (1988), Kong et al. (1992), Murray et al. (1984), Sohn et al.
(1998), Tolstoy et al., 2008 (depth range of majority of the 7000 events is shown), Toomey et al. (1985, 1988), Wilcock et al. (2002, 2009), and
Wolfe et al. (1995). Geological observations are for the northern scarp at Hess Deep (Karson et al., 2002), the Pito Deep (Perk et al., 2007) and for
drill core depths from Alt et al. (1993) and Expedition 312 Scientists (2006). Below 60 mm year1 full spreading rate, magma chambers become
less common and deep seismicity more abundant. The overlapping depths of seismicity and magma bodies at slow-spreading ridges (in different
locations) suggest episodic magmatism, with magma chambers occurring in some places (times), and the crust being cool enough for faults to propagate
to 7–8-km depth in other places (at other times). The wide range in AMC depth at fast-spreading rates also suggests episodic changes in the thermal
state of the crust.

Tolstoy et al., 1997), the ends of some small-scale segments are (the deformation of the seafloor under ocean waves;
magmatically robust (e.g., Kent et al., 2000), and mantle tomog- Crawford and Webb, 2002). Even in this area, there is a debate
raphy shows no indication of decreased melt supply at small- about how well the data constrain the thermal structure of the
scale segment boundaries (Dunn et al., 2001). Variations in lower crust (Webb, 2008). Figure 10 summarizes the results of
AMC depth can occur over short spatial scales along the ridge these studies in cartoon form. There are a number of difficulties
axis away from segment ends. For example, at 13 N on the EPR, in interpreting these geophysical data in terms of geological
the AMC depth varies by  400 m over  2 km along axis, with parameters. A significant uncertainty comes from the variation
no corresponding variation in surface topography (Babcock in compressional wave velocity with varying melt distribution.
et al., 1998). Likewise, Mutter et al. (1995) observed a >150 m For example, a small melt fraction distributed in high aspect
shoaling of the AMC over a 15-km section of the EPR, where ratio discs leads to the same decrease in P-wave velocities as a
recent eruptions are thought to have occurred. This suggests much larger amount of melt distributed in separated spheres
a significant component of temporal variability in AMC depth (e.g., Mainprice, 1997). This means that the increase in P-wave
at fast-spreading ridges. Combined with the evidence for velocity with depth in the crust could be due to either decreas-
assimilation based on chlorine overenrichment in basalts ing average melt fraction or changing melt distribution. Based
(Section 4.14.2.1.2) and gabbros (Section 4.14.2.3.2.2), these on both the tomography model of Dunn et al. (2000) and the
AMC depth variations suggest a dynamic magma chamber, which results of their compliance study, Crawford and Webb (2002)
commonly migrates upward through roof assimilation, presum- suggest that the increase in velocity with depth could be
ably after replenishment. During magma-starved periods, the explained at constant melt fraction with the melt being held
AMC will likely deepen due to crystallization at its roof. in isolated bodies at depth but being interconnected at shal-
Detailed constraints on the internal structure of the low- lower levels. However, they note that the ratio of shear to
velocity zone require extensive geophysical surveys. Because of compressional wave velocity suggests that the melt in the
this, there is currently only one region of the EPR (9–10 N) lower crust is mainly distributed in high aspect ratio cracks
where the low-velocity zone is constrained by both seismic rather than in isolated sills (Crawford and Webb, 2002). Like-
tomography (Dunn et al., 2000) and seafloor compliance wise, Singh et al. (2006a,b) found no evidence for melt sills as
510 The Lower Oceanic Crust

Based on seafloor compliance seismic experiments is either highly fortuitous or suggests


AMC warm crust and slow solidification.
Axis Fault Gabbros from fast-spreading ridges provide another per-
spective on the structure of the lower oceanic crust; however,
1 the current sample suite is very limited. Samples from both

r e
Hess Deep and Pito Deep (Figure 1) display magmatic fabrics

wat
3 Partial
largely defined by the alignment of plagioclase (e.g., Gillis
Depth below seafloor (km)

Gabbros

Sea
melt
5 et al., 1993a; MacLeod et al., 1996; Perk et al., 2007). In both
Moho areas, the fabric is oriented approximately parallel to the EPR
Partial melt
7 with a lineation that dips steeply (MacLeod et al., 1996; Perk
-5 0 5 10
et al., 2007). At Pito Deep, modal layering parallels this folia-
(a) Based on seismic tomography tion (Perk et al., 2007). Crystal plastic deformation in both of
AMC Axis these sample suites is weaker than in samples from slow-
Fault spreading ridges.
The magmatic fabrics in gabbros formed at the EPR are
1 similar to those in the Oman ophiolite, where the magmatic
Partial
melt fabric has been mapped in detail. In the Oman ophiolite, the
3 foliation in the upper gabbros is generally subparallel to the
5 overlying sheeted dike complex, with a near-vertical lineation.
Gabbros
In the lower gabbros, the foliation is subparallel to the crust–
7 Moho
mantle boundary, with a lineation that is subparallel to the
-5 0 5 10 lineation in the underlying mantle harzburgites (e.g., Nicolas
(b) Distance across axis (km) et al., 1988, 2009). There is little evidence for crystal plastic
deformation in most of the plutonic section in Oman, except
Figure 10 Cartoons of the structure of the lower oceanic crust at in the basal few 100 m of the crust, suggesting that the main
9 300 N on the EPR based on (a) compliance data, and (b) seismic
foliation and lineation formed by flow alignment in the pres-
tomography; redrawn with modifications and to the same scale from
ence of melt (Nicolas and Ildefonse, 1996; Nicolas et al., 1988,
Crawford and Webb (2002) and Dunn et al. (2000), respectively.
Compliance is the deformation of the seafloor under ocean wave loading 1993, 2009). This magmatic deformation is also observed on a
and is highly sensitive to the shear wave velocities and, hence, melt larger scale, with mineralogical layering being commonly dis-
distribution in the crust. Note the difference in size and shape of the turbed by folding (including isoclinal and sheath folds),
partially molten zone in the two different models. Deeper shades of red shearing, and boudinage (e.g., Figure 16 in Nicolas, 1992).
indicate higher melt fractions. Folds can be up to 100 m in size (Boudier et al., 1996),
indicating that a crystal mush at least this large existed, with
sufficient interstitial melt to allow large-scale folding without
large as the AMC within the low-velocity zone at 9 N, although crystal deformation. Thus, although it is commonly said that
smaller sills could exist. the deformation intensity in the lower gabbros is inconsistent
Geophysical data indicate that melt can pool beneath the with gabbro glacier models (Figure 2(f)), the evidence from
Moho, both on and off axis, and also within the lower crust Oman suggests that this interpretation is premature.
off axis (Canales et al., 2009; Crawford and Webb, 2002; Plutonic rocks encased in mantle harzburgite, which repre-
Durant and Toomey, 2009; Garmany, 1989). Additionally, sent the frozen residues of sub-Moho magma bodies, have
sub-Moho reflectors, interpreted as plutons, have been been described in a number of locations along the EPR (e.g.,
observed off axis, suggesting that at least some of the melt in Constantin, 1999; Constantin et al., 1996; Dick and Natland,
sub-Moho bodies crystallizes in place (Nedimovic et al., 1996). Plutonic rocks enclosed in peridotites at the Terevaka
2005). Sub-Moho magma bodies are rare, occurring in only and Garrett transform faults are between 1 mm and 10 cm in
10% of areas studied by Crawford and Webb (2002), and width, range in composition from troctolite to ferrogabbro, and
sub-Moho plutons occur in only  5% of the off-axis areas record evidence of crystal plastic deformation (Constantin
studied by Nedimovic et al. (2005). The scarcity of evidence et al., 1996). An apparently simple fractionation history, with
for sub-Moho magma reservoirs may indicate either that they more evolved intrusions cutting less evolved ones, but not vice
are not common or that most melt bodies and/or their crys- versa, is preserved (Constantin et al., 1996). Dick and Natland
tallization products (which may be of a very different size if (1996) studied troctolitic and gabbroic bodies encased in du-
these are open systems) are too small to resolve. Likewise, the nite and harzburgite recovered from ODP Site 895 at Hess
rarity of off-axis magma bodies, along with the lack of any Deep. These intrusions also show local evidence for high-
substantial thickening of the crust away from the axis, temperature crystal plastic deformation. In the Oman ophio-
indicates that most of the lower crust accretes in a narrow lite, a systematic study of sub-Moho plutonic bodies has shown
axial zone at intermediate- to fast-spreading ridges. Impor- that these are formed from two separate parental magmas – one
tantly, the observation of off-axis intrusions suggests that the MORB-like and one that was richer in H2O and silica (Benoit
crust must be relatively warm off axis; if the crust was cool, et al., 1999; Ceuleneer et al., 1996; Python and Ceuleneer,
then the small amounts of magma intruded off axis would 2003). Only the former, which are similar in composition to
rapidly freeze and such bodies would be present in only a very the majority of lower crustal gabbros, are considered here.
small fraction of the time. To observe any intrusions during Many gabbroic bodies are found just beneath and subparallel
The Lower Oceanic Crust 511

to the Moho (e.g., Boudier et al., 1996). These near-Moho summarized in Figure 11(a). The temperature–time path de-
gabbros have a strong magmatic and/or crystal plastic foliation rived for this hole matches a near-conductive cooling history at
and lineation, both of which are parallel to those in the sur- high temperatures (magmatic to  350  C). In detail, the high-
rounding mantle and overlying basal crustal section (Boudier temperature cooling rate of  1000–5000  C My1 can be
et al., 1996). These are commonly surrounded by dunites, explained in different ways, such as conductive cooling of a
which may have formed either through dissolution of pyroxene kilometer-scale pluton, cooling of the lithosphere during
from harzburgite (Kelemen et al., 1995) or as cumulates. uplift, or cooling as part of a conductive boundary layer (a
Whether any of these intrusions were emplaced in their current ‘half space’ model; Coogan et al., 2007a; John et al., 2004).
geometry at the ridge axis is unclear. Alternatives include em-
placement as (1) dikes in the upwelling mantle, followed by
transposition by corner flow beneath the axis; (2) part of the
main lower crust, followed by downward entrainment into the
mantle, as has been proposed for mantle-hosted dunites Zircon - U/Pb
(Jousselin and Nicolas, 2000); or (3) sills intruded off axis. 800

Closure temperature (°C)


dT/dt from Ca-in-olivine
4.14.2.2.3 Thermal history of the lower oceanic crust
600
Crystallization of melt is driven by heat loss, and at MORs, the Magnetic age
majority of the heat loss driving crystallization of the lower
crust is due to hydrothermal circulation. The efficiency with 400 Biotite
which hydrothermal circulation can remove heat from magma Titanite - fission tracks Ar–Ar
bodies depends on both the flux of cool seawater-derived fluid Zircon - fission tracks
and how close the fluid gets to the magma. Heat must be 200 3 km
Zircon - U–Th–He
transported across a boundary layer between the magma and
1 km
hydrothermal fluid by conduction, and the thicker the conduc-
tive boundary layer, the smaller the heat flux into the fluid. The 7 8 9 10 11 12
thermal history of samples of the lower oceanic crust provides (a) Age (Ma)
insight into temperature structure of the axial region and the
temperature drop between magma solidification and the
ingress of hydrothermal fluids.
The solidus temperature of oceanic plutonic rocks can be 500
Depth below the sheeted dike complex (m)

determined from the application of geothermometers to the


last crystallizing phases. Amphibole–plagioclase thermometry 1000 Pacific Ocean:
shows that magmatic amphibole (defined using trace element Hess Deep
criteria such as high Nb/La) formed at  860  30  C in gab- Pito Deep
1500
bros from the MARK area (Coogan et al., 2001a) and Oman ophiolite:
850–925  C in samples from Hess Deep (Gillis et al., 2003). Crustal section

The Ti-in-zircon thermometer (Watson et al., 2006) gives 2000 Gabbroic


bodies hosted
somewhat lower temperatures ( 680–760  C) for the most in peridotite
evolved lithologies, irrespective of spreading rate (Coogan 2500
and Hinton, 2006; Grimes et al., 2009). These low-temperature
Oman Moho
magmas crystallized at similar temperatures to the first inva- 3000
sion of hydrothermal fluids into the crust ( 700  C; see
below). It should be noted that these thermometric approaches
have only been applied to those samples that contain highly 3500
evolved interstitial phases. These are probably representative of
solidus temperatures of most plutonic rocks at slow-spreading
2 3 4 5
ridges and the uppermost gabbros at fast-spreading ridges.
(b) Cooling rate (log [°C My-1])
However, there is no amphibole or zircon in most of the
plutonic rocks from Pito Deep, indicating substantially higher Figure 11 Summary of constraints on the cooling history of the lower
solidus temperatures there. oceanic crust at (a) slow- and (b) fast-spreading ridges. (a) Closure
The thermal history of the lower crust at slow-spreading temperature versus measured age for ODP Hole 735B. Also shown is the
ridges has been investigated both by dating minerals with cooling rate, between 1000 and 600  C, derived from Ca-in-olivine
different closure temperatures to provide a temperature–time geospeedometry. Models for conductive cooling in a semi-infinite half
space (initial T ¼ 1300  C; surface T ¼ 0  C; thermal
history and by determining cooling rates using Ca-in-olivine
diffusivity ¼ 106 m2 s1) at 1- and 3-km depths are shown by the blue
geospeedometry. The latter approach is based on the
and red lines, respectively (data from Baines et al., 2009; Coogan et al.,
temperature-dependent partitioning of Ca between olivine 2007a; John et al., 2004; Schwartz et al., 2009). (b) Ca-in-olivine-derived
and clinopyroxene, with slower cooling allowing diffusive cooling rates as a function of depth in the crust at Hess Deep, Pito Deep, and
loss of Ca from olivine to occur down to lower temperatures the Oman ophiolite. Note that the upper gabbros cool much more rapidly
(Coogan et al., 2007a). The best-studied location is ODP Hole than the deeper ones, indicating that heat extraction from shallow levels is
735B, and the cooling rate constraints for this location are much more efficient than at depth (data from Coogan et al., 2007a).
512 The Lower Oceanic Crust

Near-conductive cooling rates for samples that have zircon oceanic crust forms, and how MORBs differentiate, to be
ages indicating shallow-level crystallization (Figure 8) are dif- tested. They also provide direct information on the composi-
ficult to reconcile with models that involve deep crystallization tion of lower oceanic crust that gets recycled into the mantle at
because this requires extensive, deep, hydrothermal circula- subduction zones. This section briefly describes the lithologies
tion, which would lead to rapid cooling rates at shallow levels. that make up the lower oceanic crust and is followed by a
At lower temperatures, the crust sampled in Hole 735B stayed more detailed discussion of their mineral and bulk-rock
warm (200–300  C) for longer than expected if the Atlantis compositions.
Bank was unroofed near the ridge axis (John et al., 2004;
Schwartz et al., 2009). This discrepancy can be explained either 4.14.2.3.1 Lithologies
by late unroofing or by off-axis magmatism and associated A wide range of lithologies that match the predictions for the
hydrothermal circulation reheating the crust (John et al., crystallization of a tholeiitic basalt at low pressure occur in the
2004; Schwartz et al., 2009). lower oceanic crust (Miyashiro et al., 1970; Section 4.14.1.2.2).
Models for hydrothermal circulation on axis at fast- These include (Cr-spinel bearing) troctolites, olivine gabbros,
spreading ridges typically have the base of the hydrothermal gabbros, gabbronorites, FeTi oxide gabbros (magnetite and il-
system separated from the roof of the AMC by a thin menite), amphibole gabbros, diorites, quartz diorites, tonalites,
conductive boundary layer (e.g., Cann et al., 1985; Wilcock and trondhjemites. Pyroxenites and wehrlites have not been
and Delaney, 1996). Recent observations at both Hess Deep described from the oceanic crust, although they do occur as
and Pito Deep (Gillis, 2008) and ODP Hole 1256D veins in the oceanic mantle. This is consistent with the crystal-
(Koepke et al., 2008; Wilson et al., 2006) have confirmed lization of olivine and plagioclase before pyroxene in MORBs at
ophiolite studies (e.g., Gillis and Roberts, 1999) that the low pressure. At higher pressure, pyroxene crystallizes earlier,
basal sheeted dikes are commonly amphibole–hornfels and perhaps explaining the occurrence of pyroxenite and wehrlite
pyroxene–hornfels with granular textures. Two-pyroxene veins in the oceanic mantle (Figure 3(b)). Cr-spinel is a rela-
and amphibole–plagioclase thermometry gives equilibration tively rare phase in oceanic plutonic sequences, only occurring
temperatures of 700–1050  C for these hornfels (Gillis, in trocolites, but is a relatively common microphenocryst in
2008; Koepke et al., 2008). These rocks are interpreted to MORBs. This suggests that it is replaced by Cr-rich clinopyrox-
be dikes that were altered by hydrothermal circulation ene, through a peritectic reaction, during melt–cumulate reac-
under greenschist facies conditions and were subsequently tion. Limited sampling makes it difficult to draw firm
heated by the underlying magma chamber. These rocks conclusions about the distribution of lithologies within the
constitute relics of a conductive boundary layer that sepa- lower oceanic crust, although it does appear that evolved lithol-
rated hydrothermal fluids and magma, supporting the pre- ogies are more commonly found in the upper few 100 m of the
dictions of models (e.g., Cann et al., 1985; Wilcock and plutonic complex than deeper in the crust at fast-spreading
Delaney, 1996). Reheating to temperatures up to >1000  C ridges (Coogan et al., 2002a; Gillis et al., 1993a; Hekinian
is sufficient to lead to partial melting and the production of et al., 1993; Natland and Dick, 1996; Perk et al., 2007; Wilson
felsic melts (Koepke et al., 2005a). et al., 2006).
At fast-spreading ridges, hydrothermal fluids penetrate into Troctolites are relatively rare (Figure 12), with clinopyrox-
the lower crust along a network of fractures that are filled by ene generally occurring as a major phase, even in samples with
amphibole. Thermometric studies suggest that these veins seal primitive mineral compositions (Coogan et al., 2000b; Elthon,
between  600 and 800  C, with a strong peak at 700  C 1987, 1993; Perk et al., 2007; Ross and Elthon, 1997). This
(Coogan et al., 2002a,b; Manning et al., 1996). Preliminary contrasts with low-pressure crystallization experiments in
work suggests that individual veins may be open for which clinopyroxene does not crystallize until  40  C below
<100 years (Coogan et al., 2007b), which, combined with the plagioclase and olivine cosaturation. This has been interpreted
sharp peak in metamorphic temperatures, is consistent with a as evidence that most oceanic gabbros crystallized at high
relatively short-lived hydrothermal event (Manning et al., pressure, where the clinopyroxene phase field expands at the
1996). The thermal history of samples of the lower oceanic expense of plagioclase and olivine (Figure 3(b); Elthon, 1987,
crust formed at fast-spreading ridges have also been investigated 1993). However, primitive cumulates containing magnesium-
using Ca-in-olivine geospeedometry (Figure 11(b)). There is a rich clinopyroxene have been sampled from within <1 km of
dramatic decrease in cooling rate with depth in the crust over the the base of the sheeted dike complex at Pito Deep (Perk et al.,
upper 1 km of the plutonic complex in the Oman ophiolite. 2007), indicating that primitive clinopyroxene can crystallize
Much more limited data from Hess Deep and Pito Deep match at shallow pressures. Abundant troctolitic rocks have also re-
the cooling rates determined in Oman and indicate very rapid cently been sampled at the MAR at ODP Site 1275 and IODP
cooling rates in the uppermost plutonic rocks and slower cool- Site 1309D. Another surprise is the abundance of orthopyrox-
ing rates at depth (Coogan et al., 2007a). This is consistent with ene in lower crustal samples from Hess Deep. Here, gabbro-
efficient hydrothermal cooling of the upper gabbros, but not the norites make up >25% of the samples recovered (Coogan
deeper gabbros in the near-axis region. et al., 2002a) – more than expected based on experimental
studies and considering the rarity of orthopyroxene xenocrysts
and phenocrysts in MORBs.
4.14.2.3 The Composition of the Lower Oceanic Crust
Evolved lithologies have been found in all lower crustal
In the last two decades, in situ sampling of the lower oceanic sequences that have been studied in detail, demonstrating
crust has increased dramatically, providing direct insight into that protracted crystal–melt separation occurs down to rela-
its composition. These data allow models of how the lower tively cool temperatures, allowing these lithologies to form.
The Lower Oceanic Crust 513

Pito Deep
and, hence, provide critical information about parental melt
Hess Deep compositions and differentiation processes in the lower crust.
SWIR For both major and trace elements, the partition coefficient for
Juan de Fuca xenoliths
MAR any element depends on the pressure, temperature, and min-
eral and melt compositions (Wood and Blundy, 2003).
Clinopyroxene Because the latter is difficult to accurately constrain from plu-
tonic rocks, it is generally not possible to invert mineral com-
positions to determine the melt major element composition
with certainty. Some melt composition parameters, such as the
Mg#, can be calculated from olivine and clinopyroxene com-
positions using relatively simple partition coefficients (Grove
et al., 1992; Hess, 1992; Langmuir et al., 1992; Roeder and
Emslie, 1970) such as
Mg#crystal Kd
Mg#melt ¼ [1]
1 þ ðKd  1ÞMg#crystal
where

Mgmelt Fe2þ
crystal
Plagioclase Olivine Kd ¼ [2]
Mgcrystal Fe2þ
melt
Figure 12 Modal data for oceanic gabbros. All data are from point
counting (99 gabbros and 20 xenoliths). Data sources: MAR: Coogan and Mg and Fe are molar proportions.
et al. (2000b); Tiezzi and Scott (1980); Hess Deep: Gillis et al. (1993a,b); Trace elements present a potentially simpler system, under
SWIR: Robinson and Von Herzen (1989); Pito Deep: Perk et al. (2007); certain situations (Klein, 2003; Wood and Blundy, 2003):
and Juan de Fuca gabbroic xenoliths: Dixon et al. (1986). For
comparison, Bloomer et al. (1991) estimate that in the upper 500 m of
t
Xmelt ¼ Xat =Dia [3]
Hole 735B, the average mode for olivine gabbros, the dominant lithology, t
where Xmeltis the concentration of trace element i in the melt;
is 8% olivine, 57% plagioclase, and 33% clinopyroxene.
Xai is the concentration of trace element i in mineral phase a;
and Dai is the equilibrium partition coefficient of element i
Amphibole-rich oxide gabbros are common in the upper por- between the melt and phase a.
tion of the lower crust sampled at Hess Deep (e.g., Gillis et al., Before using either the major or trace element composition
2003). The most evolved rocks are commonly coarse-grained of a mineral to calculate the composition of the melt from
to pegmatitic (Kelley and Malpas, 1996). Volumetrically minor which it grew, it must be demonstrated that the element in
dioritic veins are common in drill cores from slow-spreading question has been immobile since crystal growth. In addition
ridges (e.g., 0.45% of ODP Hole 735B is composed of ‘felsic to chemical changes due to hydrothermal alteration, element
veins’; Dick et al., 2000). Oxide gabbros, rich in both ilmenite mobility can occur either because: (1) the element was origi-
and magnetite, are common at slow-spreading ridges, making nally zoned in the crystal in which case diffusion will tend to
up 18% of the 1500 m drilled at ODP Hole 735B (e.g., Dick homogenize the composition or (2) during cooling the equi-
et al., 2000) and 7% of Hole 1309D (Expedition Scientific librium partitioning of the element between the coexisting
Party, 2005b). These may form when evolved interstitial phases changed, providing a driving force for element migra-
magmas are ‘sucked’ into shear zones during tectonic deforma- tion from one phase into another. Whether a mineral com-
tion of the crystal mush (Dick et al., 1991b; Natland et al., position changes postcrystallization also depends on the
1991) and/or migrate downward into a mush due to their high diffusion coefficient of the element of interest in that phase.
density (Ozawa et al., 1991). However, oxide gabbros are For example, calculating a parental melt Mg# from the forster-
commonly undeformed (Expedition Scientific Party, 2005b), ite (Fo) content of an olivine in an oceanic gabbro makes the
and even when they are, it is sometimes unclear if deformation implicit assumptions that the olivine crystal did not interact
was focused in these zones due to the high ductility of oxide with any residual (e.g., trapped) melt, that it was not originally
minerals (Agar and Lloyd, 1997) or because of synmagmatic zoned, and that it did not exchange Fe and Mg with coexisting
deformation (Dick et al., 1991a,b; Natland et al., 1991). Evi- phases during cooling. If any of these assumptions are invalid,
dence for pervasively distributed evolved late-stage magmas is it is very likely that the olivine composition will have
observed in many samples, especially from slow-spreading been modified because Mg–Fe diffusion in olivine is rapid
ridges, in the form of small interstitial amphibole, apatite, (Chakraborty, 1997; Dohmen et al., 2007). By contrast, ele-
and ilmenite (e.g., Coogan et al., 2001a; Gillis, 1996; Koepke ments that diffuse very slowly, such as rare earth elements
et al., 2005b; Maeda et al., 2002). These occur in even the more (REE) in clinopyroxene (Van Orman et al., 2001), are unlikely
primitive samples from slow-spreading ridges but only in the to undergo any resolvable subsolidus diffusion and can be
more evolved samples from fast-spreading ridges. used to track melt composition evolution. In the following
sections, the major and trace element compositions of oceanic
4.14.2.3.2 Mineral compositions gabbros are reviewed and the roles of variable parental melt
Mineral compositions in plutonic rocks can be used to track composition, melt differentiation processes, and subsolidus
the compositional evolution of the melt that they grew from elemental redistribution are discussed.
514 The Lower Oceanic Crust

4.14.2.3.2.1 Major elements olivine from the MARK area and IODP Hole 1309D relative
Variations in the major element composition of the parental to phenocrysts in MORBs and olivine in gabbroic rocks from
melts that crystallize to form the lower oceanic crust can be Pito Deep.
seen from the correlation of the anorthite content of plagio- Postcumulus reaction of crystals with interstitial melt can
clase and the Mg# of clinopyroxene (Figure 13(a)). Areas in also be traced through the correlations between major and
which the parental melts have high Na contents (Atlantis Bank incompatible minor elements in clinopyroxene. For example,
and MARK area) have lower plagioclase An contents at a given a small decrease in clinopyroxene Mg# can be accompanied by
clinopyroxene Mg# and vice versa (Dick and Natland, 1996; variable increases in Ti content, especially in gabbros from
Elthon et al., 1992; Longhi, 1982; Meyer et al., 1989, see slow-spreading ridges (Figure 13(c)). This variability is incon-
caption to Figure 13 for details). There is substantial scatter sistent with simple crystallization processes, which would lead
in the plagioclase An content at any given clinopyroxene Mg#, to gradual increases in clinopyroxene Ti content with decreas-
which may be due to at least three different factors: (1) hetero- ing Mg#, as depicted by the dotted gray line in the figure (e.g.,
geneity in the parental melt compositions feeding the crust; (2) Dick et al., 2002; Ross and Elthon, 1997). A nearly fourfold
variable magma chamber processes, including postcumulus increase in Ti concentration is observed in clinopyroxene from
reaction with interstitial melt; and (3) biases related to how both ODP Hole 735B and Pito Deep, between Mg#s of 88 and
zoning in crystals is accounted for when reporting mineral 85. This could be explained either by highly variable parental
compositions. It is notable that some plagioclase crystals in melt compositions or crystal–melt reaction as a melt migrates
plutonic rocks from the uppermost gabbros at Hess Deep are through a crystal mush zone – a process termed ‘magmatic
more calcic than expected based on the Na content of basalts in metasomatism’ (Mathez, 1995). However, the variation in
this area (see inset in Figure 13(a)). This is especially true for REE ratios in high Mg# clinopyroxene is too limited to explain
some plagioclase crystal cores that range up to An95 (Natland this variation in Ti concentration through parental melt het-
and Dick, 1996). As discussed below, these data are most erogeneity (see Section 4.14.2.3.2.2 and Figure 14). Instead,
readily explained by lowering of the clinopyroxene Mg#, but magmatic metasomatism better explains the data. During mag-
not plagioclase An content, due to reaction with large amounts matic metasomatism, the concentrations of compatible ele-
of interstitial melt. ments in the melt are buffered to relatively high levels by
Mineral major element compositions also reflect variable reaction with cumulus crystals while incompatible elements
efficiencies of crystal–melt separation within magma cham- increase in concentration and undergo fractionation from
bers. The wide range in mineral compositions shown in one another (Coogan et al., 2000a; Mathez, 1995).
Figure 13(a) demonstrates that, overall, crystal–melt separa- The correlation between Na and Mg# in clinopyroxene re-
tion is relatively efficient at generating a wide range of melt cords both variations in parental melt composition (Na8) and
compositions within the lower oceanic crust. Figure 13(b) magmatic metasomatism. The average Na content of clinopyr-
shows the correlation between olivine Fo and Ni contents in oxene correlates with the Na8 content of MORBs from the same
oceanic plutonic rocks. Although there is much scatter, which area, but there is a large amount of scatter in clinopyroxene Na
may be partially analytical, especially at low Ni contents, most content at any Mg# (Figure 13(d)). However, unlike Ti, which
olivine in the lower oceanic crust follows a similar composi- is likely to diffuse slowly in clinopyroxene due to its high
tional trend to olivine crystals in MORBs, although shifted to charge, sodium is a singly charged cation that may diffuse
somewhat lower Fo contents. Samples from IODP Hole 1309D rapidly in clinopyroxene and also diffuses rapidly in melt
extend to higher Ni contents than those from other areas (Lundstrom, 2000). Thus, the details of the crystal–melt
consistent with the suggestion that these may be impregnated reaction process may differ for these two elements (e.g.,
mantle rocks rather than cumulates (Drouin et al., 2010; Suhr dissolution–reprecipitation vs. diffusion). Sodium may also
et al., 2008). Samples from shallow levels in the lower oceanic be mobile within subsolidus clinopyroxene, further complicat-
crust at Hess Deep contain olivines that are significantly more ing its distribution. The correlation of Mg#, Na, and Ti in
Ni-rich than would be expected from their Fo contents. Be- clinopyroxene demonstrates that magmatic metasomatism is
cause Ni partitions much more strongly into olivine than any important in controlling mineral compositions in the lower
of the other major phases, especially at low temperatures, post- oceanic crust, especially at slow-spreading ridges (e.g., Coogan
crystallization reactions with interstitial melt are unlikely to et al., 2000a; Dick et al., 1991a,b, 2002; Lissenberg and
significantly modify the Ni content of olivine (except via Dick, 2008).
dilution if olivine grows). By contrast, reaction of olivine Subsolidus processes can lead to the redistribution of ele-
with interstitial melt decreases its Fo content along a trend ments between phases. For example, the equilibrium partition-
such as that shown in Figure 13(b) (e.g., Barnes, 1986). This ing of Ca between olivine and clinopyroxene is temperature
suggests that the Ni-rich olivine in shallow-level gabbros at dependent (Kohler and Brey, 1990), providing a constant driv-
Hess Deep formed from mixtures of relatively primitive crystals ing force for Ca to migrate out of olivine into clinopyroxene
and substantial amounts of interstitial melt. This interpretation during cooling. Comparison of the Ca content of olivine crys-
is consistent with the high An content of some plagioclase tals in MORBs and oceanic plutonic rocks (Figure 13(e)) sug-
cores (Figure 13(a)) and the high Cr content of some clino- gests that subsolidus diffusion leads to a large decrease in the
pyroxene crystals in these samples (Pedersen et al., 1996; see Ca content of olivine during cooling (Coogan et al., 2002b).
Section 4.14.2.3.4) that both require an initial stage of crystal- Figure 13(f) shows the correlation of MgO and anorthite
lization from a primitive magma. The same process of reaction content in plagioclase for oceanic gabbros and crystals in
of primitive crystals with an evolved interstitial melt, although MORBs; plagioclase crystals in the gabbros have much lower
to a lesser extent, may explain the lower Fo for a given Ni in Mg contents for a given An content. Based on the correlation in
The Lower Oceanic Crust 515

Plag-An80 Olivine NiO


70
Plagioclase An
90
65

80 60 Crystals in MORB
55 0.2
MORB Na8
70 2.5 3.0

60 Reaction with
Na8 = 1.78
trapped melt
0.1
50
Na8 = 2.5
40
Na8 = 3.9 Mg# cpx Fo olivine
(a) 60 70 80 90 (b) 40 60 80

TiO2 cpx Oxide Na2O cpx


1.2 saturation 0.7

Magm

Magmatic
1.0 0.6 Na8 = 3.9
atic m
0.8 0.5

metasom
etasom

0.6 0.4
Na8 = 2.5
atism

atism
0.4 0.3

0.2 0.2
?metamorphic? Mg# cpx Mg# cpx
(c) 60 70 80 90 (d) 60 70 80 90

CaO olivine MgO plagioclase


SWIR 0.3
0.4 MARK area
Ca loss on cooling

Pito Deep
Mg loss on cooling

Hess Deep
0.3 IODPHole 1309D
ODP Hole 1256D 0.2
MORB crystals
0.2
0.1
0.1
Fo olivine Plagioclase An

(e) 60 70 80 90 (f) 30 40 50 60 70 80 90
Figure 13 Variation in the major element composition of minerals in oceanic gabbros (see legend in panel (e)). (a) Plagioclase An content versus
clinopyroxene Mg#. The modeling trends (small gray symbols) were calculated in MELTS (Ghiorso and Sack, 1995) using the average of 40 primitive
MORBs compiled by Presnall and Hoover (1987) as a starting composition and assuming perfect fractional crystallization. The Na2O content of this
starting melt was varied (as labeled) to demonstrate that the different data suites can be broadly explained by different Na contents of the parental melts.
Differences in the Na content in primitive MORBs are associated with variations in other elements too, so these models are not meant to be accurate but
simply show the relative changes in mineral compositions expected for different parental melt Na contents. The inset shows the correlation between
plagioclase composition at a clinopyroxene Mg# of 80 (Plag An80), determined by linear regression through the data, and the Na8 content of spatially
associated MORBs. The Na8 of MORB glass from the area surrounding ODP Hole 1256D is unknown and is based on the Na2O–MgO correlation in
whole-rock lava samples (Neo et al., 2009); (b) Variation in olivine Fo and Ni contents. The gray line shows the trend followed by ‘phenocrysts’ in
MORBs, based on average Fo and Ni contents in phenocryst data from PetDB, binned at 5% Fo intervals to remove the scatter. High Ni olivines from the
upper gabbros at Hess Deep are circled; (c) Mg#–Ti in clinopyroxene with a crystallization trend calculated using MELTS as in (a). Note the large range in
Ti concentration at high Mg# and the characteristic drop in Ti after oxide saturation, which occurs at a clinopyroxene Mg# of 77 and a clinopyroxene
TiO2 content of 1 wt%; (d) Mg#–Na in clinopyroxene showing two crystallization trends calculated using MELTS for the two more Na-rich melts as
shown in (a). Clinopyroxene Na–Mg# covariation can be used as a rough guide to the parental melt Na8; (e) CaO versus Fo of olivine showing that olivine
in oceanic gabbros is highly Ca depleted compared to olivine phenocrysts in MORBs, due to subsolidus diffusion of Ca out of olivine into clinopyroxene
(Coogan et al., 2002b); (f) MgO versus anorthite content of plagioclase showing that plagioclase in oceanic gabbros is highly MgO depleted compared to
plagioclase phenocrysts in MORBs, again suggesting subsolidus reequilibration. The generally lower Fo and An contents of minerals in oceanic gabbros
than in basalts shown in (e) and (f) suggests that evolved minerals crystallize within an uneruptible crystal mush. All oxide abundances are given as wt%
(data are from Constantin et al., 1996; Coogan et al., 2000a, 2002a; Dick et al., 2002; Meurer and Gee, 2002; Natland and Dick, 1996; Pedersen et al.,
1996; Perk et al., 2007; Ross and Elthon, 1997; Suhr et al., 2008; Yamazaki et al., 2009).
516 The Lower Oceanic Crust

plagioclase anorthite content are common; Coogan et al.,


Ce (ppm) = 2
2000a; Dick et al., 2002; Natland and Dick, 1996; Pedersen
/ Yb
SWIR Ce .5 et al., 1996). These differences in the extent of zoning may
MARK area
b =0 largely reflect differing extents of homogenization via diffusion
10 Pito Deep
C e/Y during and after crystal growth. Normal zoning is more com-
Hess Deep
DSDP Site 334 mon than reverse zoning, but both have been reported (e.g.,
Dick et al., 2002). Complex zoning patterns have also been
Ce/Yb
1 5 EPR basalts reported, including small-scale oscillations in clinopyroxene
compositions (Coogan et al., 1999), but await detailed study.
4

4 6 MgO 4.14.2.3.2.2 Trace elements


Both the range of parental melt compositions feeding the crust
1 10
and the degree of differentiation of melts within the crust can
(a) Yb (ppm)
be deduced using the trace element composition of minerals in
plutonic rocks from the lower oceanic crust. The range of melt
Zr (ppm) 5 compositions feeding the crust provides information about the
Y=
Zr/ mechanism of melt extraction from the mantle. Mantle melting
100
.3 is thought to be near fractional, with melt extracted from the
Y =0 mantle continuously at low mantle porosity (Johnson et al.,
Zr/
10 1990). This leads to the generation of a wide range of parental
Zr/Y EPR basalts melt compositions. Because MORBs are relatively composi-
4 tionally uniform, these diverse melt compositions must be
1 mixed together, a process known as aggregation (e.g., Grove
3 et al., 1992), either within the asthenospheric mantle during
4 6 MgO melt extraction or within the lithospheric magma reservoirs.
The trace element compositions of minerals in plutonic rocks
10 100
record the range of melt compositions within lithospheric
(b) Y (ppm)
magma reservoirs. If plutonic rocks recorded evidence of crys-
Figure 14 Variation in trace element abundances in clinopyroxene in tallizing from a wider range of parental melt compositions
oceanic gabbros. (a) Ce versus Yb and (b) Zr versus Y. Note the general than are erupted, this would provide evidence for melt aggre-
increases in more-to-less incompatible element ratios (Ce/Yb and Zr/Y) gation within the crust.
with increasing incompatible element abundances. The lack of a large Two studies have interpreted the trace element signature of
increase in Zr/Y with increasing incompatible element concentrations in
minerals in oceanic gabbros as recording crystallization from
the samples from the SWIR may be because these are mainly oxide-rich,
incompletely aggregated melts. Ross and Elthon (1993) docu-
and some Zr will partition into the oxide phases. The insets show that the
same fractionation of more-to-less incompatible element abundances mented highly incompatible element-depleted clinopyroxene
occurs in MORBs (data from 10 300 N EPR; Regelous et al., 1999). Data in a suite of lherzolites, olivine gabbronorites, gabbronorites,
sources: Hess Deep: Coogan et al. (2002a); Dick and Natland (1996); and noritic anorthosites drilled at DSDP Site 334, 10 km west
Gillis, (1996), Natland and Dick (1996); ODP Site 334: Ross and Elthon of the MAR (Figure 14). If these rocks are cumulates, then they
(1993); SWIR: Coogan (unpublished data – all oxide gabbros) Gao et al. must have precipitated from very depleted melts in magma
(2007); Pito Deep: Perk et al. (2007); MARK area: Coogan et al. (2000a), chambers beneath this slow-spreading ridge. This would indi-
Coogan (unpublished data), Ross and Elthon (1997). cate inefficient mixing during melt extraction from the mantle.
However, these rocks have also been interpreted as melt im-
pregnated residual peridotites (Girardeau and Francheteau,
Figure 13(a), this is not due to the gabbros crystallizing from 1993), drawing into question the broader significance of
a Mg-poor melt. Instead, this is an evidence that Mg migrates these samples. Gabbros from the MARK area contain plagio-
from plagioclase into the surrounding mafic phases during clase that have cores and rims with different trace element
cooling, as described above for Ca migration from olivine ratios, which Coogan et al. (2000b) interpreted as evidence
into clinopyroxene. These last two examples demonstrate the of growth from incompletely aggregated melts. Since then,
problems that may be encountered in using mineral composi- Cherniak (2003) has shown that REE diffuse relatively rapidly
tions to calculate melt compositions if subsolidus diffusion has in plagioclase, and thus, this interpretation may need revisit-
occurred. However, they also demonstrate that the distribution ing. Other studies of the trace element composition of minerals
of rapidly diffusing elements can be used as geospeedometers in oceanic gabbros have found little or no evidence for sub-
in oceanic gabbros to determine cooling rates. This is because stantial variation in parental melt trace element compositions
the extent of subsolidus elemental exchange during cooling is (e.g., Coogan et al., 2002a; Natland and Dick, 1996; Perk et al.,
dependent on the cooling rate (Coogan et al., 2002b, 2007a; 2007). In summary, the trace element compositions of min-
Dodson, 1973; Ozawa, 1984; Section 4.14.2.2.3). erals in oceanic gabbros have not, as yet, provided conclusive
There have been few detailed studies of major element evidence of crystallization from diverse unaggregated melts.
zoning in minerals from the lower oceanic crust. Current data This suggests either that the majority of aggregation occurs
show that zoning is generally minor or undetectable in olivine within the mantle or that processes within the crust, such as
but can be quite large in plagioclase (>10% changes in mixing and melt–rock reaction, erase the chemical signature of
The Lower Oceanic Crust 517

the diversity of parental melt compositions. This interpretation


contrasts with studies of melt inclusions in MORB phenocrysts Cr (ppm)
1200
that suggest that a wide range of parental melt compositions are
present during the early stages of crystallization (e.g., Nielsen 1000
et al., 1995; Sobolev and Shimizu, 1993).
Melt differentiation within the crust can be tracked using the 800
trace element composition of clinopyroxenes in oceanic gab-
600
bros; these contain a wide range of incompatible element abun-
dances (Figure 14). At low incompatible element abundances, 400
there is little variation in incompatible elemental ratios. This
suggests that melts are largely aggregated by the time the clin- 200
opyroxene grows. Even plutonic rocks intruding peridotites (a)

within the upper mantle at Hess Deep contain clinopyroxene Ce (ppm) 153 923A 8R2-81
in equilibrium with fully aggregated MORBs (Dick and Natland, 5 153 923A 10R2-116F
1996). With increasing incompatible element abundances, the 153 923A 10R2-116C
153 923A 5R1-19
variation in incompatible element ratios increases, with highly 4 153 923A 5R1-28
incompatible elements generally being enriched relative to mod- 153 923A 5R1-46
erately incompatible elements (Figure 14). These data demon- 735B 880 mbsf
3
strate that incompatible trace elements can be fractionated from
one another within magma chambers – something generally 2
ignored in the interpretation of basalt geochemistry. Although
much of this fractionation probably occurs within crystal mush 1
zones, the impact on MORB compositions is shown by the (b)
variations in incompatible element abundances during fraction-
ation (decreasing MgO) in a suite of basalts from 10 300 N on Zr/Y
the EPR (Figure 14, inset; Regelous et al., 1999). 2.5
The spatial variation in trace element concentration in sin-
gle clinopyroxene crystals from six gabbros from the MARK 2.0
area and one gabbro from the Atlantis Bank (Figure 15) pro-
1.5
vides further evidence for the importance of magmatic meta-
somatism in crystal mush zones. These crystals show limited
1.0
zoning in compatible elements, such as Cr and Mg, but com-
plex incompatible trace element zoning. Crystal cores are
0.5
generally relatively homogeneous in both trace element abun-
dances and ratios, suggesting that they grew from a melt that (c)

changed little in composition during core growth. By contrast, Rim 0.2 0.4 0.6 0.8 core
the rims of all crystals show large enrichments in incompatible Normalized distance into crystal
element abundances and increases in the ratio of more-to-less Figure 15 Trace element zoning in clinopyroxene in gabbros from the
incompatible element (e.g., Zr/Y, Zr/Nd). Trapped melt crys- MARK area (Coogan et al., 2000a and unpublished) and a single crystal
tallization cannot explain the increases in incompatible ele- from ODP Hole 735B (Gao et al., 2007). There is a very limited zoning in
ment ratios or the lack of decrease in Cr or Mg#; instead, Cr (a) despite large variations in incompatible element abundances (b)
magmatic metasomatism of a crystal mush is required and ratios (c) in individual crystals. Note the different patterns in Zr/Y
(Coogan et al., 2000a; Gao et al., 2007). zoning with most crystals showing a large increase in the outer third,
The halogen content of magmatic amphiboles provides but two showing more subdued and progressive increases. 153 923A
information about the role of assimilation that is complemen- 10R2-116 F/C: fine and coarse areas of same thin section. The outermost
edges of two crystals show small decreases in Ce abundances,
tary to that discussed in Section 4.14.2.1.2, based on the
suggesting late-stage saturation of an accessory phase that concentrates
chlorine content of MORBs. Chlorine is massively enriched REE, most likely apatite.
in magmatic amphibole in plutonic rocks from the EPR com-
pared to those from the MAR (Gillis et al., 2003). This strongly
supports the conclusion, drawn from basalt compositions, that this is one possible explanation for some data (Coogan et al.,
chlorine is enriched in MORBs at fast-spreading ridges through 2000b; Ross and Elthon, 1993). The trace element composi-
assimilation (Michael and Schilling, 1989). Plutonic rocks tions of clinopyroxenes provide strong evidence for a wide
containing chlorine-rich magmatic amphibole occur to at range of incompatible element concentrations in interstitial
least 800 m below the base of the sheeted dike complex at melt within the lower oceanic crust (Figures 13–15). This can
Hess Deep, requiring assimilated chlorine to be transported be best explained by magmatic metasomatism, with interstitial
to at least this depth, most likely via subsidence of a crystal melt migrating through a crystal mush and reacting with it.
mush (Figure 2(f); e.g., Quick and Denlinger, 1993). This supports previous suggestions of extensive interstitial melt
In summary, the trace element compositions of minerals in migration within crystal mush zones in the lower oceanic crust
oceanic gabbros have yet to provide unambiguous evidence for at slow-spreading ridges (Coogan et al., 2000a; Dick et al.,
crystallization from incompletely aggregated melts, although 1991b, 2002; Gao et al., 2007; Lissenberg and Dick, 2008;
518 The Lower Oceanic Crust

Natland et al., 1991; Suhr et al., 2008). The upper gabbros Table 2 Average and median compositions of the lower oceanic
formed at the EPR and sampled at Hess Deep show evidence crustal sections sampled to date
for both extensive assimilation and reaction with large
wt% Average Median
amounts of interstitial melt (Figure 13(b)).
SiO2 50.6 (1.6) 51.0
4.14.2.3.3 Bulk-rock compositions Al2O3 16.7 (2.3) 16.6
Bulk-rock compositions provide another way to investigate the FeO 7.5 (3.0) 6.1
processes operating in the lower oceanic crust and allow the MgO 9.4 (1.1) 9.0
bulk composition of the lower oceanic crust to be estimated, MnO 0.14 (0.05) 0.1
which is essential for understanding both: (1) the average CaO 12.5 (1.0) 12.6
composition of Moho-crossing melts and (2) the role of the Na2O 2.35 (0.45) 2.4
K2O 0.06 (0.03) 0.1
lower oceanic crust in global geochemical cycles. Unfortu-
TiO2 0.78 (0.81) 0.4
nately, sampling of the lower oceanic crust in most regions
P2O5 0.02 (0.02) 0.0
has generally been too limited to allow its bulk composition to Mg# 69 (10) 73
be accurately determined. Only at Atlantis Bank and, perhaps, ppm
at the Atlantis Massif (Figure 1; Table 1) has sufficient sam- Sc 37 (12) 40
pling occurred to allow any confidence in the lower crustal V 209 (136) 162
compositional estimates. The Mg/Fe ratio of the bulk compo- Cr 308 (129) 285
sition of the plutonic section of the crust at Atlantis Bank is too Co 50 (26) 41
low to be in equilibrium with the mantle. This requires either Ni 138 (45) 159
crystallization in the subcrustal mantle lithosphere or the pres- Cu 71 (19) 65
Zn 38 (9.6) 39
ence of primitive cumulates in unsampled areas of the crust
Ga 14 (1.5) 14
(Coogan et al., 2001b; Dick et al., 2000). Despite these prob-
Sr 115 (30) 97
lems, an estimate of the bulk composition of the lower oceanic Y 13.9 (7.4) 13
crust in those areas that have been sampled in detail is com- Zr 28.4 (13.6) 28
piled in Table 2. This should be considered a very approximate Nb 0.93 (0.65) 1
value (note the large standard deviations about the average) La 0.86 (0.68) 0.83
and is unlikely to be representative of the global average com- Ce 2.75 (1.95) 2.76
position of the lower oceanic crust as its Mg# is too low. Pr 0.52 (0.38) 0.53
In the following sections, the bulk composition of samples Nd 2.78 (1.86) 2.61
of the lower oceanic crust are first described, then the data are Sm 1.10 (0.68) 1.09
Eu 0.58 (0.25) 0.58
used to calculate the fractionation of elements between the
Gd 1.60 (0.93) 1.60
lower and upper oceanic crust. Finally, the stable and radio-
Tb 0.31 (0.18) 0.30
genic isotopic composition of the lower oceanic crust is briefly Dy 2.09 (1.17) 2.06
discussed. Ho 0.46 (0.29) 0.48
Er 1.34 (0.77) 1.32
4.14.2.3.3.1 Major elements Yb 1.27 (0.74) 1.23
Bulk-rock major element analyses of the lower oceanic crust Lu 0.19 (0.11) 0.18
form broad fields rather than tightly defined differentiation
These were calculated as follows: The median composition of gabbros from each of the
trends (Figure 16). This reflects both the variation in parental
seven best-studied areas of the modern oceans (Table 1) was calculated (with the
magma composition and the crystal accumulation processes.
major elements normalized to 100%). The median rather than the mean was used
For example, bulk-rock sodium, at a given MgO content, cor- because petrologists love of the unusual can make the proportion of analyses of extreme
relates inversely with plagioclase An80 (plagioclase An content compositions unrepresentative of their abundance. The average (1s in parenthesis) and
at a clinopyroxene Mg# of 80; Figure 13(a) inset) and posi- median of these was then calculated. The low average Mg# of the gabbro suites
tively with the associated basalt Na8 (Figure 16(a)). The role of demonstrates that this bulk composition is not representative of the bulk lower oceanic
variable mineral proportions can be seen in the correlation crust. Space does not permit a more rigorous assessment of this here.
between whole-rock MgO and CaO (Figure 16(b)). Rocks
that are comprised of near-cotectic proportions of plagioclase,
clinopyroxene, and olivine form a dense cloud. While varia- proportions on the bulk-rock composition can be seen in
tions in both mineral compositions and abundances play a the correlation of MgO and Al2O3, with a small fraction of
role in the scatter in this cloud, the data are consistent with samples being troctolites and the vast majority being gabbroic
most samples having similar modal compositions. The (Figure 16(c)).
range in modal abundances needed to explain the scatter is The relative paucity of dunitic and troctolitic samples relative
illustrated by modeling the variation in bulk composition of to gabbros (Figure 16(b) and 16(c)) is surprising for two reasons.
different mineral mixtures (Figure 16(b)). Moderate changes Firstly, 20% of the mass of a primary mantle melt is expected to
in modal proportions (30–45% clinopyroxene and 5–20% be crystallized prior to clinopyroxene saturation, suggesting
olivine) explain the bulk of the data well, although a small that there should be large amounts of clinopyroxene-poor cumu-
proportion of samples are olivine enriched. These are cumu- lates in the oceanic crust. Secondly, in many kilometer-scale
lates produced during high-temperature olivine and olivi- continental mafic intrusions, modal layering is common, lead-
ne þ plagioclase crystallization. The same control of modal ing to large variations in the modal proportions on a meter to
The Lower Oceanic Crust 519

Oxide accumulation
Na2O 8 Na2O TiO2/Y SWIR IODP Hole 1309D
5 6 0.6 MARK area ODP Hole 1256D
Pito Deep 15° N MAR
4
Hess Deep MORB
4 2 MgO
10 20 30 40 0.4
3

2
0.2
Mg#
1 Increasing parental melt Na8
MgO
(a) 5 10 15 (d) 0.3 0.4 0.5 0.6 0.7 0.8

CaO CPX Cr (ppm)


0.3c,0.5p,0.2ol
0.4c,0.5p,0.1ol
15 1000 Reaction with
Plagioclase

0.45c,0.5p,0.05ol interstitial melt

on
ati
10 100

lliz
sta
cry
Reaction with

al
10 interstitial melt

on
5

cti
Fra
Olivine Mg#
(b) 10 20 30 MgO 40 (e) 0.3 0.4 0.5 0.6 0.7 0.8

Al2O3 Y (ppm) =1
Plag /Y
Zr
25
Tro 100
20 cto
lite
15 s
10
10

5
Zr (ppm)
CPX Olivine
(c) 10 20 30 MgO 40 (f) 10 100 1000
Figure 16 Bulk-rock compositions of samples from the lower oceanic crust: (a) MgO–Na2O, (b) MgO–CaO, (c) MgO–Al2O3, (d) Mg#–TiO2/Y, (e) Mg#–
Cr, and (f) Zr–Y; see text for discussion. A general field for normal MORBs is shown in the figures (except (d) for clarity) as a black outline. Three
models are shown in (b) that delineate the composition of bulk rocks composed of different proportions of clinopyroxene (c), plagioclase (p) and
olivine (o). These compositions were calculated using the correlation of clinopyroxene Mg# and plagioclase An content shown in Figure 13(a) and
stoichiometric constraints (the Mg# of olivine and clinopyroxene are assumed to be identical) to determine the change in mineral compositions with
decreasing bulk MgO. The gray arrow in (e) shows a schematic fractional crystallization trend that indicates that most cumulates have a lower Mg#
for their Cr content than expected, suggesting reaction with interstitial melt, as in Figure 13(b) (data from Barling et al., 1997; Cannat et al., 1995; Casey,
1997; Coogan et al., 2001a,b; Expedition Scientific Party, 2005a,b; Gaggero and Gazzotti, 1996; Gillis et al., 1993a,b; Hart et al., 1999; Hekinian
et al., 1993; Hertogen et al., 2002; Holm, 2002b; Kelemen et al., 2004; Neo et al., 2009; Niu et al., 2002; Pedersen et al., 1996, and unpublished data;
Perk et al., 2007; Pettigrew et al., 1999; Ross and Elthon, 1997).

decimeter scale. (ODP samples are small relative to most samples unsampled regions (e.g., deeper) and/or that primitive cumulates
collected in the field, and so their bulk composition can record are generally compositionally modified by reaction with large
small-scale variations in modal proportions.) This rarity of well- amounts of migrating interstitial melt.
developed modal layering in the oceanic crust, at least from slow-
spreading ridges, can be explained if the size and thermal regime 4.14.2.3.3.2 Trace elements
of magma reservoirs are different to those in the continental crust. Trace element abundances in oceanic plutonic rocks depend
The paucity of primitive (dunitic and trocolitic) cumulates can on the concentration of the element in the melt from which the
be explained by either these lithologies occurring in, as yet, plutonic rocks crystallized, the cumulus phases (especially
520 The Lower Oceanic Crust

accessory phases), and the amount of postcumulus crystal– reflect a combination of the following: (1) variations in paren-
melt reaction. Cumulus accessory phases can strongly influ- tal melt composition, (2) mineral accumulation, and (3) large
ence the partitioning of trace elements between crystals and variations in extent of reaction with interstitial melt.
melts, and thus, their accumulation can be readily identified
using bulk-rock compositions. Ilmenite and titanomagnetite
accumulation in a rock can be identified by enrichments in Ti 4.14.2.3.3.3 Element fractionation between the upper and lower
over similarly incompatible elements. Primitive MORBs have oceanic crust
a TiO2/Y of 0.045 (Sun and McDonough, 1989) and the Although the bulk composition of the lower oceanic crust in
bulk partition coefficient of these elements during silicate each study area is not well constrained, it is still informative to
phase crystallization is approximately the same. Most samples compare the relative fractionation of elements between the
of oceanic gabbros have TiO2/Y between 0.02 and 0.055, upper and lower oceanic crust. The ‘effective partition coeffi-
but samples that have accumulated FeTi oxides have dramati- cients’ for incompatible elements, defined as the bulk lower
cally elevated TiO2/Y, of up to  0.6 (Figure 16(d)). Becau- crust composition divided by the bulk upper crust composi-
se 1 wt% cumulus ilmenite will increase the TiO2/Y of a tion, are similar in the three best-studied areas from slow-
typical gabbro (with 0.4 wt% TiO2 and 12 ppm Y; TiO2/ spreading ridges, with the crust recovered from IODP Hole
Y ¼ 0.035) to 0.075, the amount of cumulus ilmenite in 1309D a little more depleted in incompatible trace elements
most oceanic gabbros must generally be less than this. than the other areas (Figure 17). However, at fast-spreading
Titanomagnetite has a lower Ti concentration, allowing greater ridges, there is much more variability in the fractionation of
accumulation of this phase, but the basic premise remains the elements between the upper and lower crust. At Pito Deep, the
same. Samples with high abundances of accumulated oxides, plutonic rocks are quite depleted in incompatible elements,
as indicated by both high Ti/Y and high Ti concentrations, can whereas at Hess Deep, the plutonic rocks are much less de-
have bulk-rock Mg#s much lower than the Mg# of the silicate pleted in incompatible elements. In IODP Hole 1256D, where
phases in the rock (by >30% in some cases). Likewise, elevated only the upper <100 m of gabbroic rocks have been sampled,
P/Y can be used to identify apatite accumulation, but the plutonic rocks are only slightly depleted in incompatible
much of the existing P data is of insufficient quality to use elements relative to the upper crust. The difference between the
this fingerprint. different areas can be explained by differences in the efficiency
The highly compatible elements Cr and Ni are only broadly of crystal–melt separation, with gabbros from slow-spreading
correlated with bulk-rock Mg# (Figure 16(e)), likely reflecting ridges reflecting  20% melt mixed with cumulate crystals, on
variable extents of reaction of primitive (high Ni, Cr, and Mg#) average, and gabbros from ODP Hole 1256D requiring 70%
cumulates with evolved interstitial melt, leading to a large melt mixed with cumulate crystals.
decrease in the crystal Mg#s but small decreases in Ni and Cr
contents, as discussed for olivine compositions (Figure 13(b)).
The extent of enrichment in Cr for a given Mg# relative to that La Ce Pr Nd Zr Sm Eu Gd Tb Dy Ho Y Er Yb Lu
1.0
expected for cumulates produced by fractional crystallization
varies systematically by sampling location. Troctolites from 70%
Bulk lower crust/bulk upper crust

deeper in the crust at Pito Deep appear to be cumulates of


near-perfect fractional crystallization with little reaction with 20%
interstitial melt, consistent with their very low incompatible
element abundances (Perk et al., 2007). Samples from the
MARK area, the Atlantis Bank, and IODP Hole 1309D, which 0.1
all formed at slow-spreading ridges, are more enriched in Cr for rium
Equilib
a given Mg#, suggesting more reaction with interstitial melt. SWIR
ional MARK area
The samples with the highest Cr for a given Mg# come from the Fract Pito Deep
uppermost gabbros formed at fast-spreading ridges (Hess Hess Deep
IODP Hole 1309D
Deep, Pito Deep, and ODP Hole 1256D), suggesting that ODP Hole 1256D
they experienced the most reaction with interstitial melt.
The two most incompatible elements for which there is La Ce Pr Nd Zr Sm Eu Gd Tb Dy Ho Y Er Yb Lu
extensive bulk-rock data are Zr and Y. These elements Figure 17 Effective bulk partition coefficients for incompatible trace
correlate with each other (Figure 16(f)) and increase, along a elements between the lower and upper oceanic crust. These are
scattered trend, with decreasing Mg#. There is a general calculated based on the median composition of the lower and upper crust
increase in Zr/Y with increasing concentrations of these ele- in each well-studied area (Table 1; excluding the Oman ophiolite). Lower
ments in all of the sample suites, as observed in clinopyroxene crustal data sources from Figure 16 and additional upper crust data from
analyses (Figures 14(b) and 15). There is a little overlap in PetDB. The black dashed lines show the compositions of perfect
cumulate rocks formed by fractional crystallization (‘fractional’) and
Zr and Y concentrations between gabbros and MORBs,
equilibrium crystallization (‘equilibrium’) after 70% crystallization. Also
indicating that incompatible elements are efficiently frac-
shown are the composition of cumulates formed by fractional
tionated between the upper and lower oceanic crust (see crystallization mixed with 20% (‘20%’) and 70% (‘70%’) interstitial melt.
Section 4.14.2.3.3.3). The bulk partition coefficients used for the modeling are shown by the
In summary, the general trends observed in bulk-rock ‘equilibrium’ model and were selected to produce a perfect cumulate
major and trace element compositions support the conclusions from fractional crystallization that fit the composition of the Pito Deep
drawn from mineral chemistry that cumulate compositions lower crust.
The Lower Oceanic Crust 521

The highly incompatible elements (Zr and light rare earth there is no record in the plutonic rocks of parental melts with a
elements (LREE)) are more depleted in the plutonic rocks than broad range of Nd isotopic compositions being added to the
the moderately incompatible elements (Y and heavy rare earth lower oceanic crust. This can be explained in two ways. Most
elements (HREE)). The calculated effective partition coeffi- simply, the range of isotopic composition of melts added to
cients support the suggestion (Coogan et al., 2000a) that Zr is the crust may match the range erupted. Alternatively, a broad
more incompatible than any of the REE during lower crustal range of melt isotopic compositions may have been added to
crystallization. This leads to higher Lu/Zr and, by analogy, the crust but then efficiently homogenized within the lower
higher Lu/Hf in the lower crust than upper crust that will crust prior to solidification of the plutonic rocks. The larger
cause it to evolve to a substantially more radiogenic Hf isotopic range in basalt than gabbro Nd isotopic compositions most
composition over time relative to the associated upper crust. likely reflects the slightly wider area over which the basalts were
This may provide a means to recognize recycled lower oceanic sampled, although all basalts come from the same spreading
crust in magmas from mantle plumes. segment as the plutonic rocks. More complicated scenarios
involving isotopically heterogeneous melts feeding the crust,
4.14.2.3.3.4 The isotopic composition of the lower oceanic crust with this isotopic variability preserved in eruptives but
The radiogenic isotopic compositions of samples of the lower homogenized in lower crustal magma reservoirs, are also
oceanic crust potentially provide insight into the composi- possible.
tional variability of melts feeding the crust, the extent of their The Pb isotope composition of coexisting plutonic rocks
homogenization within the crust, and the extent of subsequent and basalts has only been measured in samples from the MARK
hydrothermal alteration. Figure 18 compares the Nd isotopic and Atlantis Bank areas. There is a wider range in Pb isotope
composition of the lower and upper crust from the three areas composition in plutonic rocks than in basalts in the Atlantis
where data are available. In all cases, there is less variability in Bank area, but plutonic rocks and basalts show a similar range
Nd isotopic composition of the plutonic rocks, even though in composition in the MARK area. Since the basalts come from
more samples have been analyzed. Three gabbro intrusions a wider geographical area than the gabbros, this is unlikely to
encased in peridotite in the MARK area also show a similarly represent a sampling bias. The origin of the wider range in
restricted range of Nd isotopic compositions, suggesting that Pb isotopes in plutonic rock at Atlantis Bank is unclear.
their parental melt was similar to the melts that generated the Kempton et al. (1991) suggested that this may, in part, be
larger gabbro intrusions (Kempton and Hunter, 1997). Thus, analytical due to the very low levels of Pb in some samples.
Holm (2002a) found that leaching produced large shifts in the
Pb isotope composition of gabbros from Hole 735B. Because
0.51295 0.51300 0.51305 0.51310 0.51315 0.51320
the shift in Pb isotope composition during leaching does not
0.8 Gabbros correlate with shifts in Sr isotopic composition, Holm (2002a)
interpreted this as due to magmatic contamination. At the
MARK n = 24 sample scale, Kempton and Hunter (1997) observed small
0.6
Hess n = 7 differences between the Nd and Pb isotopic compositions of
SWIR n = 117 coexisting plagioclase and clinopyroxene in two gabbros from
0.4 the MARK area. Without further studies, the global significance
of the Pb isotope variation in oceanic gabbros remains
Fraction of analyses

0.2 uncertain.
The strontium isotopic composition of oceanic plutonic
(a) rocks potentially provides insights into both the compositions
of the magmas that these rocks crystallized from and the
0.4 Basalts
amount of seawater-derived hydrothermal fluid (with high
87
MARK n = 15 Sr/86Sr) that has passed through them. In ODP Hole
0.3 Hess n = 5 735B, 65% of gabbros have 87Sr/86Sr within the range of
SWIR n = 19 fresh basalts from that spreading segment and only 12% have
87
Sr/86Sr >0.7030 (Bach et al., 2001; Hart et al., 1999; Holm,
0.2
2002a; Kempton et al., 1991). The most elevated 87Sr/86Sr
values are all in the upper 500 m of the core, where mineral-
0.1 ogical alteration at high temperatures is greatest and the
O isotopic composition of the crust is lowest (Alt and Bach,
(b)
2006; Bach et al., 2001; Gao et al., 2006; Hart et al., 1999;
0.51295 0.51300 0.51305 0.51310 0.51315 0.51320 Kempton et al., 1991; Stakes et al., 1991; ). There is less Sr
143Nd/144Nd isotopic data for lower crustal samples from the MARK area
(Kempton and Hunter, 1997) and Hess Deep (Lecuyer and
Figure 18 Comparison of the Nd isotopic composition of plutonic rocks
Gruau, 1996), but the data from these areas also suggest lim-
(including mineral separates) and basalts from the same mid-ocean ridge
segment. There is less range in Nd isotopes in gabbros than basalts in all
ited increases in the 87Sr/86Sr of the lower crust due to hydro-
areas (data sources: Atlantis Bank: Coogan et al. (2004); Hamlin and thermal alteration.
Allegre (1985), Hart et al. (1999), Holm (2002a), Kempton et al. (1991), The average bulk-rock O isotope ratio of most ocean gab-
Snow (1993), MARK area: Dosso et al. (1993); Kempton and Hunter bros falls close to or slightly below that of fresh MORBs,
(1997) and Hess Deep: Pedersen et al. (1996)). although the dataset is heavily biased toward the well-studied
522 The Lower Oceanic Crust

with the Oman ophiolite, where compositional variability


0.4 with depth in the crust is better constrained. Owing to space
Fresh limitations and lack of data, the vertical variation in just four
Hess (n = 31)
compositional parameters is discussed here: bulk-rock Mg#
SWIR (n = 224)
Fraction of analyses

and Cr content, plagioclase An content, and clinopyroxene


0.3
Cr content.
The compositional variation with depth in several well-
studied kilometer-scale sections is shown in Figure 20. Crust
0.2
formed at slow-spreading ridges shows more small-scale com-
positional diversity than crust formed at fast-spreading ridges,
with primitive and evolved samples within close proximity.
0.1 There is little systematic variation in composition with depth
in either ODP Hole 735B or IODP Hole 1309D, suggesting
that the crust may have been built by many small magma
batches rather than from one large magma reservoir. This is
2 3 4 5 6 7 8 9 10
most clearly demonstrated by the wide distribution of Cr-rich
Whole rock δ18O samples throughout IODP Hole 1309D, which suggests that
Figure 19 Histogram of bulk-rock O isotopic composition of lower primitive parental melts were emplaced in multiple different
crustal samples from the SWIR and Hess Deep. See text for discussion. places within this core (Godard et al., 2009).
Some evidence for general upward differentiation of the
crust on a  500-m scale can be seen in both ODP Hole
ODP Hole 735B (Figure 19; Hess Deep d18Oaverage ¼ 4.8%; 735B and IODP Hole 1309D. Dick et al. (1991b, 2000,
Atlantis Bank d18Oaverage ¼ 5.6%; average plagioclase and clin- 2002) have used the bulk rock and mineral chemistry of Hole
opyroxene separates from the MARK area d18Oaverage ¼ 6.2 735B to divide this core into a number of magmatic cycles.
and 5.7%, respectively; Alt and Bach, 2006; Bach et al., 2001; These are defined by general increases in the extent of differ-
Gao et al., 2006; Hart et al., 1999; Kempton and Hunter, 1997; entiation upward, followed by a step to more primitive compo-
Kempton et al., 1991; Lecuyer and Gruau, 1996; Stakes sitions. The number of apparent cycles depends on what
et al., 1991). In ODP Hole 735B, the upper, more deformed, compositional parameter is examined. Two clear breaks in
500 m is depleted in d18O and the lower  1000 m is both bulk rock and mineral compositions are evident (at
enriched in d18O. In general, gabbros with d18O < 5.7% record  530 mbsf and 940 mbsf; mbsf ¼ meters below seafloor)
reaction with seawater-derived fluids at high temperatures and and other subtler variations of unclear significance are also
those with high d18O > 5.7% record either reaction with low- observed (Figure 20). It is beyond the scope of this study to
temperature fluids or reaction with high-temperature fluids evaluate all possible origins of the downhole compositional
that have elevated d18O due to previous high-temperature variations – the key point in the current context is the
fluid–rock interaction.  500-m-scale chemical cycles in this core suggesting a possible
In summary, radiogenic isotope data suggest that either rough height scale for plutons at slow-spreading ridges. This
there is relatively little variability in the radiogenic isotopic scale of variability is less obvious in IODP Hole 1309D, which
composition of melts feeding the crust or this variability is is much more primitive than Hole 735B, but Suhr et al. (2008)
efficiently homogenized before the crystallization of the plu- suggest that there is an upward differentiating ‘magmatic unit’ of
tonic crust. This is consistent with the interpretation of the similar scale between 1100–1200 and 700–800 mbsf.
trace element compositions of clinopyroxene within oceanic Detailed studies of ODP Hole 923A from the MARK area
gabbros (Figure 14). The strontium and oxygen isotopic com- and a short section of IODP Hole 1309D (from 1190 to 1245
positions of the lower crust are only slightly, although hetero- mbsf) provide a smaller-scale view of the compositional het-
geneously, modified by reaction with seawater-derived erogeneity within lower crust formed at a slow-spreading ridge
hydrothermal fluids. Whether this reflects limited fluid flow (Coogan et al., 2000a; Meurer and Gee, 2002; Ross and Elthon,
through the lower crust, reaction with fluids that are close to 1997; Suhr et al., 2008). Sharp steps in mineral compositions
isotopic equilibrium with the crust due to previous fluid–rock in both of these sections are consistent with crust being built
reactions or fluid flow being focused in spatially restricted from many small intrusions (e.g., Figure 5 in Suhr et al., 2008).
zones (e.g., faults) remains to be resolved. However, minor and trace element fractionation trends
(Coogan et al., 2000a) and similar cooling rates throughout
(Coogan et al., 2007a) suggest that the intrusions formed a
4.14.2.3.4 Chemical profiles through the oceanic crust single mush zone. Similar small-scale intrusions have been
Compositional variation with depth in the crust provides crit- suggested to be important in accreting the gabbros in Hole
ical constraints on models of lower crustal accretion. In partic- 735B (Dick et al., 2000). Thus, it appears that magma replen-
ular, the distribution of compatible elements can elucidate ishment at slow-spreading ridges may occur on a much smaller
where primitive melts are added to the crust and thus help to scale than the overall 500-m bodies suggested by broader-
understand the scale of magmatic replenishment and the scale differentiation trends.
mechanisms of differentiation and crustal accretion. Because The plutonic complex formed at a modern fast-spreading
of the limited sampling of in situ oceanic lower crust ridge (the EPR) has been sampled at Pito Deep, Hess Deep, and
that formed at fast-spreading ridges, comparison is drawn IODP Hole 1256D (<100-m-depth range). In all locations, the
The Lower Oceanic Crust 523

200

400
Depth in drill core (m)

600

800

1000

1200
Slow-spreading
1400 ODP Hole 735B
IODP Hole 1309D
Depth beneath sheeted dike complex (m)

500

1000

Fast-spreading
1500 Pito Deep
Hess Deep
ODP Hole 1256D
2000

2500
Normalized depth beneath sheeted

Oman ophiolite
0.2 Kafifah
Abyad
Namarah
dike complex

0.4

0.6

0.8

0.3 0.5 0.7 1000 2000 3000 40 60 80 0.4 0.8 1.2 1.6
Bulk-rock Mg# Bulk-rock Cr (ppm) Plagioclase An content Clinopyroxene Cr2O3 (wt%)

Figure 20 Depth variation in bulk-rock Mg#, bulk-rock Cr content, plagioclase An content, and clinopyroxene Cr content in well-sampled regions of
lower oceanic crust from slow-spreading ridges (ODP Hole 735B and IODP Hole 1309D), fast-spreading crust (Hess Deep, Pito Deep, and ODP Hole
1256D), and the Oman ophiolite. Sample depths >800 m at Hess Deep are based on sampling water depth and thus do not necessarily represent the
depth at which these formed (only one third of these come from outcrops, the remainder from rubble). Sample depths for the three lower crustal
sections of the Oman ophiolite (Wadi Namarah, Wadi Abyad, and Wadi Kafifah) are normalized to the thickness of the plutonic section with 1 ¼ base of
crust, 0 ¼ base of the sheeted dike complex. The ‘best estimates’ of the thickness of these gabbro sections range from 2.5 to 5 km. Dashed lines indicate
possible pluton margins in Hole 735B and solid lines possible pluton margins in IODP Hole 1309D. See text for discussion (data from Barling et al., 1997;
Browning, 1982; Cannat et al., 1995; Casey, 1997; Coogan et al., 2002a, 2006; Dick et al., 2002; Expedition Scientific Party, 2005a,b; Gaggero and
Gazzotti, 1996; Garrido et al., 2001; Gillis et al., 1993a,b; Hekinian et al., 1993; Hertogen et al., 2002; Neo et al., 2009; Niu et al., 2002; Pallister and
Hopson, 1981; Pedersen et al., 1996, and unpublished data; Perk et al., 2007; Ross and Elthon, 1997; Suhr et al., 2008; Yamazaki et al., 2009).

upper few hundred meters of the plutonic section of the crust primitive rocks at depth would lead to an unreasonably Mg-
are moderately evolved (Figure 20). At deeper crustal levels, rich bulk crustal composition. However, despite the general
rocks from Hess Deep and Pito Deep provide contrasting decrease in bulk-rock Mg# upward at Hess Deep, the Cr con-
views, with far more primitive rocks recovered from Pito tent of both clinopyroxene and of the bulk rock does not
Deep than Hess Deep at the same crustal level. The Hess decrease upward, suggesting that the decrease in Mg# is not
Deep samples show a general decrease in bulk-rock Mg# due to precipitation from a more evolved melt. Instead, this
upward. This cannot be the case for Pito Deep, as more may simply be due to samples from shallower crustal levels at
524 The Lower Oceanic Crust

Hess Deep comprising cumulate crystals that have reacted with lithosphere, they pond and cool slowly to form the plutonic
large amounts of interstitial melt, producing evolved bulk rocks that constitute the lower oceanic crust. Once the magmas
compositions (Figures 13(b) and 16(e)). The occurrence of pond, their chemical and thermal evolution depend on factors
high-Cr clinopyroxene in the shallow plutonic rocks in all such as the temperature and composition of the wall rocks, the
locations demonstrates that primitive melts are added to the size of magma body, whether they are open or closed systems,
shallowest portion of the crust (the AMC) at fast-spreading the size and frequency of replenishments, the crystallization
ridges. The difference in composition of the lower crust at mechanism, and how efficient hydrothermal circulation is in
Hess Deep and Pito Deep suggests that spatial and/or temporal removing the latent and specific heat carried to the lithosphere
variability in magmatic processes along the EPR are probably by the magma. Key observations that models for the lower
important in controlling lower crustal processes. For example, crust should explain include:
it has been suggested that the Hess Deep crust formed near a
segment end (Lonsdale, 1988), which may explain the more • Magma chambers are generally steady-state features at fast-
but not at slow-spreading ridges. At intermediate- to fast-
evolved plutonic rocks found here, based on analogy with the
spreading ridges, there is a general shoaling of the AMC
changes in basalt chemistry toward segment ends (e.g., Christie
with increasing spreading rate. However, there are large
and Sinton, 1981; Langmuir et al., 1986).
variations in the depth of the AMC at a given spreading
The Oman ophiolite allows a more detailed investigation of
rate (Figure 9).
the extent of compositional variability with depth in the lower
crust and how this varies between locations. Three different • At fast-spreading ridges, crystallization occurs primarily at
shallow levels and there is a sharply defined Moho, whereas
wadi sections are compared in Figure 20. These show little
at slow-spreading ridges, plutonic rocks can crystallize
intrasection variability, with all three sections showing relatively
within the lithospheric mantle. High-temperature crystal
minor compositional variability with depth in the crust, except
plastic deformation of the lower crust is also more common
within the uppermost few hundred meters. These three lower
at slow-spreading ridges, suggesting that magma supply
crustal sections are more comparable with the two EPR lower
does not always accommodate spreading.
crustal sections than with the crust from slow-spreading
ridges. The lack of any obvious steps in mineral compositions • Ages of zircons from evolved plutonic rocks in the lower
oceanic crust at slow-spreading ridges suggest that 90% of
or extreme differentiation, except within the uppermost
the lower crust forms at shallow levels at the axis (Figure 8).
plutonic rocks, contrasts with the crust recovered from
The remaining 10% either crystallizes (1) off axis (giving
slow-spreading ridges. This is consistent with the occurrence of
ages younger than the crustal age) or (2) on axis, either at
a steady-state mush zone at fast-spreading ridges from which
depth or at shallow levels, but is not rafted off axis at the
evolved interstitial melt can be efficiently extracted into
spreading rate (giving ages older than the crustal age).
the overlying melt sill rather than freezing in place to produce
evolved plutonic rocks. By contrast, at slow-spreading • Differentiation trends of MORBs at fast- and slow-
spreading ridges differ in (1) the average degree of differen-
ridges, solidification of multiple small magma bodies would
tiation and how variable this is (Figure 6); (2) the occur-
allow crystallization over a wider range of temperatures
rence of chlorine overenrichment in magmas from fast- but
throughout the crust, producing intermixed primitive and
not slow-spreading ridges, which reflects assimilation of
evolved lithologies.
altered crust in the former; and (3) the greater abundance
of phenocrysts in MORBs from slow- than fast-spreading
4.14.3 Generating the Lower Oceanic Crust ridges (Figure 4).
• Crystals in MORBs are commonly too primitive to be in
It is helpful to precede a detailed discussion of the processes equilibrium with the melt in which they are carried to the
operating in the lower oceanic crust by describing a basic surface (Figure 5), and there is common evidence for mix-
working model for its generation and reviewing the key obser- ing having occurred days to years before eruption.
vational data. Melt generation occurs due to near-adiabatic • Most samples collected from the lower oceanic crust are
decompression of the mantle as it ascends beneath the ridge. composed of (olivine) gabbros with near-cotectic propor-
Because melting is near fractional, a wide variety of melt com- tions of clinopyroxene, olivine, and plagioclase (Figures 12
positions are generated in the mantle. However, MORBs are and 16(b)). Troctolites and cumulate dunites are much less
relatively homogeneous. This means that melts must mix ei- common.
ther within the melting column or within the lithosphere. The • Sections through lower crust formed at slow-spreading
thickness of the lithosphere, which determines the depth at ridges show small-scale steps in composition that are
which cooling and crystallization will occur, is dependent most readily interpreted as evidence for the differentiation
principally on the spreading rate, as this largely controls the of individual magma bodies that produced intrusions with
rate at which hot mantle material ascends toward the cold heights of  1 to  500 m (Figures 20 and 21).
surface. At full spreading rate 20 mm year1, the lithosphere • At fast-spreading ridges, there is little evidence for system-
is significantly thicker than the crust, whereas at faster spread- atic changes in cumulate compositions with depth, consis-
ing rates, the lithosphere is probably little or no thicker than tent with the occurrence of steady-state magma chambers
the crust (Bown and White, 1994; Reid and Jackson, 1981). imaged using geophysical methods, which can buffer the
These thickness estimates are based on conductive cooling melt composition (Figure 20).
alone – additional heat loss due to hydrothermal circulation • Differentiation trends of oceanic gabbros are not readily
will increase lithospheric thickness. Once magmas enter the modeled by simple crystallization processes; instead,
The Lower Oceanic Crust 525

Plagioclase An content Plagioclase An content


50 60 70 50 60 70

ODP Hole 923A ODP Hole 1309D

25 1200

Depth (mbsf) 35 1210

45 1220

55 1230

65 1240

Gabbro Troctolite or troctolitic gabbro

Figure 21 Variation in plagioclase An content with depth in ODP Hole 923A (Coogan et al., 2000a; Meurer and Gee, 2002; Ross and Elthon, 1997) and a
short section of IODP Hole 1309D that has been studied in detail (Suhr et al., 2008) showing small-scale variability in the composition of the lower
oceanic crust formed at slow-spreading ridges. These data are consistent with crustal accretion occurring through the injection of numerous small sills
building a larger crystal mush zone; mbsf, meters below seafloor.

extensive reactions with trapped and migrating interstitial length, which is enough to accommodate 20 000 years of
melt are implicated (Figures 13–16). spreading (MacGregor et al., 1998; Sinha et al., 1997). Making
• The upper gabbros formed at fast-spreading ridges com- reasonable assumptions about rates of hydrothermal heat
monly show a steeply dipping magmatic foliation sub- extraction MacGregor et al. (1998) predict that this magma
parallel to the sheeted dike complex, suggesting the flow body will freeze in less than a tenth of this time, requiring
of a crystal mush to align the crystals. magmatism to be episodic. Likewise, in the 35 N area of the
• Cooling rates at slow-spreading ridges are near conductive, MAR, seismic data suggest a somewhat smaller amount of melt
but at fast-spreading ridges they are very high in the upper- in the lower crust (1 km3 per kilometer of ridge length; Dunn
most plutonic rock, requiring extensive hydrothermal heat et al., 2005) but still more than is likely to exist at steady state.
extraction, and decrease drastically with depth (Figure 11). These observations suggest episodic addition of significant vol-
umes of magma to the crust, followed by intervals of much more
limited magma supply. The chemical stratigraphy of ODP Hole
4.14.3.1 How Large and How Frequent Are Magma Additions
735B (and perhaps IODP Hole 1309D) is consistent with this
to the Lithosphere?
model, with magma bodies of  500-m height that apparently
Understanding the size and frequency of magma additions to differentiated in isolation from each other (Figure 20). Smaller-
the lithosphere, along with the compositional diversity of scale chemical variations in drill cores from slow-spreading
melts added to the lithosphere (Section 4.14.3.2) and where ridges (Figure 21) may record shorter timescale episodic infla-
these melts pond (Section 4.14.3.3), provides a starting point tion of larger magma bodies by sill emplacement into a crystal
for understanding lower crustal processes. For a given crustal mush. Taken together, these data are consistent with a model of
thickness and spreading rate, the frequency of magmatic re- episodic (10 ky timescale) emplacement of 100–1000-m-high
plenishment and the size of each magma batch are inversely magma bodies at shallow levels that are dominantly mush
correlated. Continuous flow of magma into a lithospheric zones that are gradually inflated by much smaller intrusions.
chamber would provide a very slow rate of magma recharge At intermediate- to fast-spreading ridges, magma is proba-
and, hence, heat addition. By contrast, highly episodic magma bly added to the crust much more frequently than at slow-
supply would lead to episodic heat supply to the crust and spreading ridges. A steady-state AMC requires magma addition
intermittent magma chambers. frequently enough to prevent it from freezing (Cann et al.,
At slow-spreading ridges, several lines of evidence suggest 1985; Henstock et al., 1993; Maclennan et al., 2004; Phipps
intermittent magma supply to the shallow crust. Geophysical Morgan and Chen, 1993a). Simple thermal balances suggest a
studies suggest that magma bodies are only present episodically maximum of 100-year intervals between replenishments if
(Figure 9), but where they have been detected, they contain the AMC is to avoid freezing. High compatible element abun-
significant amounts of melt (Dunn et al., 2005; Magde et al., dances in shallow-level intrusions formed at the EPR
2000; Seher et al., 2010; Singh et al., 2006a; Sinha et al., 1997). (Figure 20) demonstrate that primitive melts are added to the
For example, the magma chamber imaged beneath the Rey- AMC and that it is not simply formed from a residual melt
kjanes ridge contains  3 km3 of melt per kilometer of ridge compacted out of the underlying cumulates. Regular addition
526 The Lower Oceanic Crust

of primitive magma to the crust is also required to buffer 1986); and (3) a level where thermal conditions stall magma
MORB compositions (O’Hara, 1977) and to produce gabbros transport through either crystallization, leading to porosity and
with relatively uniform composition (Figure 20). This last permeability decreases (Kelemen and Aharanov, 1998; Phipps
point was fundamental to the development of the model Morgan et al., 1994), or change in rheology, such as the brittle–
shown in Figure 2(e) (Pallister and Hopson, 1981). Finally, ductile transition.
1-m-wide dikes must be emplaced, on average, every 10 years Magma density has long been suggested to play a critical
for a 100 mm year1 full spreading rate, and since mixing role in lower crustal processes at MORs. The evolved composi-
appears to precede most eruptions by at most a few years, tion of MORBs, compared to any mantle-derived melt
AMC replenishment on this timescale is likely. (Figure 6), requires that they pond and differentiate before
In summary, it appears likely that as the spreading rate eruption. In 1980, it was proposed that this ponding was re-
decreases, the size of magma additions to the crust increases quired for MORBs to differentiate to a density minimum prior
but becomes more episodic. If this inference is correct, under- to eruption (Huppert and Sparks, 1980; Sparks et al., 1980;
standing how magma addition is regulated at depth is critical Stolper and Walker, 1980). However, this model is based on
in linking mantle to crustal (including tectonic and hydrother- the assumption that Moho-crossing magmas are picritic, for
mal) processes. which there is generally no evidence (e.g., Langmuir et al.,
1992). Instead, MORB densities stay approximately constant
during troctolite crystallization and then increase slightly during
4.14.3.2 How Well Aggregated Are Melts Added to the
gabbro crystallization; the lack of primitive lavas therefore can-
Lithosphere?
not be attributed to these being too dense to erupt. Eruption of
The range of melt compositions produced during near fractional high-density FeTi basalts also demonstrates that high density
melting of the heterogeneous mantle is thought to be large (e.g., does not prevent eruption. The level of neutral buoyancy,
Johnson et al., 1990). Although some studies have found evi- which depends on magma density, has been suggested to coin-
dence for incomplete mixing of these melts before they are cide with the depth of the AMC (Ryan, 1993) or to be much
added to crustal magma chambers (Coogan et al., 2000b; Ross shallower (Hooft and Detrick, 1993), depending on the as-
and Elthon, 1993), the trace element (Figure 14) and isotopic sumptions made in the calculation. Uncertainties in vesicle
(Figure 18) data compiled here suggest that oceanic gabbros and crystal content will probably prevent this debate being
crystallized from melts were similar to MORBs. It appears that conclusively resolved. Nevertheless, the shoaling of the AMC
most of the heterogeneity generated within the melting column with increased spreading rate (Figure 9) is not predicted by a
is removed through mixing prior to the solidification of the model in which magma ponds due to attaining neutral buoy-
lower oceanic crust. This conclusion contrasts with the diversity ancy. This suggests that magma density is not the principal
of melt compositions observed within melt inclusions trapped control on the depth of magma ponding at MORs.
within phenocrysts from MORBs, perhaps suggesting that the Vertical changes in the stress field can also lead to magma
latter do not record the composition of widespread melts (e.g., ponding. When an ascending magma reaches a level at which the
Michael et al., 2002). Alternatively, extensive melt–rock reaction horizontal compressive stresses are higher than the vertical stres-
within a crystal mush may erase the signature of diverse parental ses, a sill will form (Gudmundsson, 1986). This change in stress
melts present in the initial cumulates. However, using clinopyr- regime may be due to vertical differences in the elastic properties
oxene trace element compositions to track melt aggregation of the rocks, such as those that occur due to changes in lithology
misses any compositional heterogeneity that existed during troc- or temperature. Alternatively or additionally, temporal changes
tolite crystallization. in the stress field may determine whether magmas will tend to
pond or migrate vertically. This mechanism could potentially
explain magma ponding at the Moho at fast-spreading ridges and
4.14.3.3 Where and Why Does Melt Pond in the Lithosphere?
the base of the lithosphere at slow-spreading ridges.
Magma must pond beneath the surface to form the lower Magma ascent may be stalled by decreasing temperatures if
oceanic crust, and where it ponds controls the depth (range) the melt is moving slowly enough (e.g., via porous flow) to
of crystallization. This, in turn, is important for understanding begin to crystallize, thus decreasing the porosity and hence
such features as the pressure of MORB differentiation, the permeability (Kelemen and Aharanov, 1998; Phipps Morgan
location of latent heat released into the axial hydrothermal et al., 1994). In a texturally equilibrated crystal mush, the melt
system, the depth of the AMC, and the relative thickness of will remain connected down to low porosities (e.g., Hunter,
upper and lower oceanic crust. Despite this importance, it is 1987), but decreasing permeability will decrease the rate at
uncertain why magma ponds at ridges. Once a magma cham- which melt can migrate upward. This mechanism of melt
ber has formed, further upward transfer of magma will be stagnation requires subhorizontal and closely spaced isotherms
stalled, as magma from below feeds into the existing chamber. to allow efficient removal of the latent heat of crystallization and
Continued melt addition from below will eventually cause the formation of a horizontal permeability barrier. Such a thermal
pressure in the chamber to increase, eventually overcoming the structure is likely at the roof of the AMC but is unlikely within
tensile strength of the roof rocks, at which stage magma will a crystal mush zone (see Section 4.14.3.4 and Figure 22),
continue to migrate upward (e.g., Blake, 1981). For a given where widely spaced isotherms will lead to low conductive
volume of magma, a sill has the most efficient shape for stalling heat fluxes and, thus, slow crystallization. A thermal control
vertical magma transport. But why does magma pond in the first on the depth of the AMC at intermediate- to fast-spreading
place? Possible reasons include reaching: (1) a level of neutral ridges is consistent with the general decrease in AMC depth
buoyancy (Ryan, 1993); (2) a stress barrier (Gudmundsson, with increasing spreading rate (Figure 9; Phipps Morgan and
The Lower Oceanic Crust 527

Chen, 1993b). This suggests that thermal structure is the Section 4.14.3.4, this model requires temperatures below the
primary control on magma ponding in the AMC. zircon saturation temperature at up to 18-km depth beneath
the ridge axis, which is difficult to reconcile with thermal
4.14.3.3.1 Evidence for magma ponding at high pressures at constraints and, in particular, the near-conductive cooling
slow-spreading ridges rates for associated samples that crystallized at shallow levels
Geophysical studies have demonstrated that magma can pond (Figure 11(a)). In summary, unambiguous evidence for sub-
within the crust at relatively shallow levels, even at slow- stantial amounts (>10%) of high-pressure (>0.3 GPa) crystal-
spreading ridges (Figure 9). This is consistent with the obser- lization at slow-spreading ridges is lacking.
vation that 90% of zircon crystallization ages are the same as
the crust age (Figure 8). However, three lines of evidence have 4.14.3.3.2 Where does magma pond at fast-spreading
been used to suggest that melts commonly pond and crystallize ridges?
at elevated pressures at slow-spreading ridges. Firstly, compar- The competing models of lower crustal accretion at fast-
ison of MORB compositions with experimentally synthesized spreading ridges predict initial crystallization at different pres-
melts in equilibrium with olivine, plagioclase, and clinopyrox- sures (Figure 2(f) and 2(g)). In the gabbro glacier model
ene at various pressures have been interpreted as indicating (Figure 2(f)), most melt is added directly from the mantle to
that some MORBs crystallize at 0.5–1.0 GPa at slow (but gen- the shallow AMC. There, crystallization occurs at the base of the
erally not at fast) spreading ridges (Figures 3 and 7; Grove sheeted dike complex, with melts stalled at a thermal boundary.
et al., 1992; Herzberg, 2004; Villiger et al., 2007). Secondly, Cumulates formed in the AMC subside downward and outward
mineral compositions in oceanic gabbros, in particular high to form the lower oceanic crust. By contrast, in the sheeted sill
Mg# clinopyroxene, have been used to infer high-pressure model (Figure 2(g)), melts pond repeatedly as they are trans-
crystallization at slow-spreading ridges (Elthon, 1987; Elthon ported through the lower crust, with crystallization occurring in
et al., 1992). Thirdly, rare zircons in oceanic gabbros with situ from the Moho to the AMC. Melt stalling in this model has
U–Pb ages greater than the crustal age have been interpreted been suggested to occur beneath permeability barriers within
as the products of deep crystallization (Schwartz et al., 2005). the low-velocity zone (Kelemen and Aharanov, 1998). A better
Each of these possibilities is discussed below. understanding of why and where melts pond within the axial
The first two lines of evidence for high-pressure crystalliza- magma plumbing systems would provide important constraints
tion at slow-spreading ridges are both based on the earlier on how the lower oceanic crust forms.
saturation of clinopyroxene relative to plagioclase and olivine
as pressure increases (Figure 3(b)). However, olivine gabbros
4.14.3.4 Thermal Constraints on Lower Crustal Processes
that formed at the EPR and that were collected from within
<1 km of the base of the sheeted dike complex at Pito Deep Crystallization and differentiation processes in the lower oce-
contain clinopyroxene with Mg#s up to 89 (Perk et al., 2007; anic crust are dependent on the rate of heat loss from the
Figure 13(a)), demonstrating that clinopyroxene can cry- magma and, hence, the thermal structure of the region through
stallize ‘early’ during MORB differentiation, even at low pres- which the magmas travel and pond. Heat is supplied to MORs
sures. There are several possible explanations for this in a number of ways (e.g., Cannat et al., 2004). Magma pro-
discrepancy. Primitive clinopyroxene in oceanic gabbros may vides both specific heat and the latent heat released during
grow from melts that are different to those used in the exper- crystallization. The latent heat is roughly equivalent to an extra
iments (e.g., higher Cr or Ca/Al; see Figures 3 and 7 and 400  C of magma temperature. Heat is also supplied by con-
Section 4.14.2.1.3). Alternatively, difficulties in nucleating duction from the mantle and, to a limited extent, by hydration
clinopyroxene in experiments could lead to an apparently reactions within the lithosphere, which are generally exother-
lower saturation temperature for clinopyroxene than the mic. Loss of heat from the system occurs through conduction,
equilibrium temperature. Finally, the crystallization process hydrothermal circulation, and (to a very minor extent) dehydra-
in nature and the lab may be different, with experiments tion reactions (generally endothermic). Heat can be redistribu-
approaching equilibrium crystallization, whereas natural ted within the crust through the movement of melts, magmas,
melts may experience near fractional or in situ crystallization crystal mushes, and assimilated blocks. Conductive heat transfer
and extensive melt–rock reaction in a crystal mush (Kvassnes depends on temperature gradients, with steep temperature gra-
and Grove, 2008; Lissenberg and Dick, 2008; Perk et al., 2007). dients required for large conductive heat fluxes. Hydrothermal
Other characteristics of oceanic gabbros cited as evidence heat transport depends on the permeability within the system,
of high-pressure crystallization, such as high Na in clinopyrox- and this decreases dramatically with depth in the crust (Fisher,
ene, and the coexistence of ilmenite and primitive silicates 1998). Constraints on the thermal structure of the lower oceanic
(Elthon, 1987, 1993) can be explained by magmatic metaso- crust come from geophysical data, the distribution of hydrother-
matism in a crystal mush (Section 4.14.2.3.2). mal vents, and the cooling rate of samples from the lower
The third observation suggesting high-pressure crystalliza- oceanic crust. Each of these is discussed, in turn.
tion at slow-spreading ridges are the ‘old’ ages of 6% of Low seismic velocities and the occurrence of an AMC
zircons in the lower crust from slow-spreading ridges reflector demonstrate that the lower crust is partially molten
(Figure 8); Notably, 90% of zircon ages require shallow- at fast-spreading ridges (Figures 9 and 10; Crawford and
level crystallization. The simplest explanation for these old Webb, 2002; Detrick et al., 1987; Dunn et al., 2000). The
zircons is that they crystallized at depth and were then trans- topography of ridge axes also provides information about
ported upward and amalgamated into the lower crust both the thermal and mechanical structure of the ridge axis
(Figure 2(c); Schwartz et al., 2005). As discussed in (e.g., Cochran and Buck, 2001; Phipps Morgan and Chen,
528 The Lower Oceanic Crust

Slow: 20 mm year –1 full rate Fast: 120 mm year –1 full rate

No latent heat or hydrothermal cooling No latent heat or hydrothermal cooling


450
650 2
450
5 850
650 950
1050 4
850 1150
10
950
6
1250
1050
15
8
1150
1250 km
km
(a) 5 10 15 km (d) 2 4 6 8 km
5 0.28 Wm–2 100
Nonconductive heat distribution (Wm–2) 50 Nonconductive heat distribution (Wm–2)
{

Some deep latent heat release Shallow hydrothermal cooling


and latent heat release
450
450 Steep thermal gradient 2
5 650
at AMC roof
650 850
950
4
850 1050
10
950 1150
6
0.41 Wm–2

1050
{ Supra- 15
1250

solidus 1150 8

km
1250
km
100
50
5

5 10 15 km 2 4 6 8 km
(b) (e)
5 −2
100
Nonconductive heat distribution (Wm ) 50 Nonconductive heat distribution (Wm−2)

Extensive deep hydrothermal cooling Deep hydrothermal cooling


and latent heat release
2
Latent heat = 3 Wm–2

5
450
4
Steep thermal gradient
solidus
Supra-

650 10 at side of molten zone

6
850
15 450
950
650 8
850
1050 km 950
1250 1150 1250 1150 1050 km
100
50
5

5 10 15 km 2 4 6 8 km
(c) (f)

Figure 22 Steady-state thermal models for mid-ocean ridges based on using heat sources and sinks to describe latent heat release and hydrothermal
cooling, respectively (based on Maclennan et al., 2004; Morton and Sleep, 1985; Sleep, 1975). (a) Conductive cooling only for a slow-spreading ridge;
(b) as (a) but with 30% of the total latent heat of crystallization released (0.9 kW m1 along axis) added between 13- and 15-km depth and 0–3 km off
axis; (c) as (a) with no latent heat release but with 15 kW m1 along axis of hydrothermal cooling evenly distributed within 3 km of the axis and down to
15-km depth; Note that cooling below zircon saturation temperatures (generally 900  C) at depth requires that the fluid gets to these depths and has
cooled the overlying crust; (d) conductive cooling only for a fast-spreading ridge; (e) all latent heat of the plutonic complex released within a melt sill at
2000-m depth and 16.7 kW m1 along axis of hydrothermal heat extraction in the upper crust dissipating this heat representing a possible gabbro
glacier model; (f) latent heat of the plutonic complex released throughout the lower crust up to 1.5 km off axis and sufficient hydrothermal circulation
throughout the lower crust to cool the axial region completely. Note that even with dramatic off-axis hydrothermal cooling throughout the lower crust,
The Lower Oceanic Crust 529

1993b; Shah and Buck, 2001). The axial high at fast-spreading below zircon saturation temperatures (probably 850  C) at
ridges and at magmatically robust areas of slow-spreading >15-km depth and no subsequent hydrothermal circulation
ridges, such as the Reykjanes ridge near Iceland, suggests a to allow conductive cooling at shallow levels during the main
hot lower crust. By contrast, the lower crust must be cool and phase of crustal accretion.
strong for major faults to form and for the development of an Constraints on the thermal history of lower crust formed at
axial valley, as observed at most slow-spreading ridges (Chen fast-spreading ridges have come from modeling the extent of
and Morgan, 1990; Phipps Morgan and Chen, 1993b). down-temperature diffusion of Ca from olivine into clinopyr-
The number of hydrothermal plumes per unit ridge length oxene (Figure 13(e)). Coogan et al. (2002b, 2007a,b) demon-
correlates linearly with spreading rate, suggesting that the long- strate a large variation in cooling rate with depth in the
term heat supply controls the extent of hydrothermal activity plutonic section in the Oman ophiolite, which is similar to
(e.g., Baker and German, 2004). Long-term venting requires that calculated for samples from Hess Deep and Pito Deep
either a magma body separated from the hydrothermal fluid by (Figure 11(b)). The far more rapid cooling rates at the depth
a thin conductive boundary layer (Cann et al., 1985) or a of the AMC, compared with those deeper in the crust, require a
region of hot rock that hydrothermal fluids gradually penetrate greater crystallization rate in the AMC than at deeper levels.
along a cracking front (Lister, 1974). Along fast-spreading This suggests that crystal subsidence from the AMC must occur
ridges, hydrothermal venting seems to be focused at the ridge to prevent this body from solidifying.
axis; the only detailed off-axis survey carried out, to date, shows Thermal models for the lower oceanic crust have become
no evidence of significant off-axis hydrothermal venting (Baker more sophisticated since it was shown that steady-state magma
et al., 2010). The size of sulfide deposits on the seafloor chambers cannot exist at full spreading rates of 10 mm
changes with spreading rate, with abundant small systems at year1, even in the absence of hydrothermal cooling (Kuznir
fast-spreading ridges and fewer but larger sulfide mounds at and Bott, 1976; Sleep, 1975). Subsequent models incorporate
slow-spreading ridges. Although this difference could be hydrothermal heat removal by either adding heat sinks
explained by differences in permeability or different eruption (Morton and Sleep, 1985) or increasing the thermal conduc-
rates leading to varying probability of sulfide mounds being tivity of the wall rocks by a given factor (known as the Nusselt
buried, a simple explanation has been proposed. Episodic number; Machetel and Garrido, 2009; Phipps Morgan and
magma emplacement at slow-spreading ridges drives episodic Chen, 1993a). Unfortunately, neither approach satisfactorily
large systems, while continuous magmatism and hydrothermal treats heat transport because hydrothermal circulation neither
activity at fast-spreading ridges leads to more but smaller sul- acts as a simple heat sink nor acts to increase the rate of
fide mounds (Wilcock and Delaney, 1996). conductive cooling (e.g., Cannat et al., 2004). The most so-
Cooling rates calculated for samples from the lower oceanic phisticated current models are those of Cherkaoui et al.
crust can also be used to determine the thermal structure of the (2003), Maclennan et al. (2004, 2005), and Machetel and
crust. Relative and absolute cooling rates have been deter- Garrido (2009). Cherkaoui et al. (2003) explicitly include
mined using geothermochronology (John et al., 2004), geos- fluid flow in the crust below a given temperature but treat
peedometry (Coogan et al., 2002b; Ozawa, 1984), crystal size crystallization and crustal accretion simply. In contrast,
distributions (Garrido et al., 2001), and variations in disloca- Maclennan et al. (2004, 2005) use heat sinks to simulate
tion densities (Kennedy et al., 1996). Only the first two hydrothermal cooling but allow the depth of crystallization
methods have yielded quantitative cooling rates, to date. Ther- and the extent of crystal subsidence from the AMC to vary.
mochronology and Ca-in-olivine studies of samples from ODP Thermal models for slow-spreading ridges show that at
Hole 735B suggest a cooling rate of  1000–5000  C My1 at spreading rates below 15–20 mm year1 full rate, the con-
high temperature, which is similar to the cooling rate calcu- ductively cooled lithosphere is substantially thicker than the
lated by assuming pure conductive cooling (Figure 11(a); thickness of melt produced (Figure 22(a); Bown and White,
Coogan et al., 2007a; John et al., 2004). Low-temperature 1994; Reid and Jackson, 1981). This means that melt can
thermochronology suggests that cooling rates continue at potentially begin to crystallize at depth, leading to a mixed
near-conductive rates down to  200  C, although off-axis region of mantle rocks intruded by plutonic rocks. However,
reheating may have occurred due to off-axis magmatism melt crystallization releases latent heat, and this will offset the
(Figure 11; John et al., 2004; Schwartz et al., 2009). This temperature decrease driven by conductive cooling to the sur-
near-conductive cooling of most lower crustal samples from face (Sleep and Barth, 1997). Deriving a melt with the average
the Atlantis Bank area is somewhat paradoxical because a few Mg# of MORBs erupted at a slow-spreading ridge requires
samples from this area have zircon ages substantially greater 30% crystallization of a primary mantle melt. This will re-
than the crustal ages, suggesting deep crystallization lease 30% of the latent heat of the magma, impacting the
(Figure 8; Schwartz et al., 2005). This would require both thermal structure of the ridge axis region (Figure 22(b)). Even
deep hydrothermal circulation to cool the lithosphere to if the steady-state release of latent heat is limited (<1 kW per

heat conduction from the axis is too slow to allow crystallization to occur there. If hydrothermal cooling was as efficient as is shown in this model, then
any sills intruded off axis would freeze very rapidly and it would be highly improbable that any geophysical survey would be performed in an area
containing a partially molten off-axis sill. Input parameters for the models: thermal conductivity: 2.5 WK1 m1; crust density: 2900 kg m3; specific
heat capacity: 1000 J Kg1 K1; latent heat of crystallization: 500 kJ kg1; lithospheric thickness: 100 km; temperature at base of lithosphere: 1300  C;
total thickness of melt extracted from the mantle at slow- and fast-spreading rates: 6 and 7 km (5 km of plutonic rocks), respectively. Left and upper
axes: rate of hydrothermal heat extraction (blue line; W m1) and latent heat addition (red line; W m1) integrated horizontally and vertically,
respectively. Right and lower axis: distance. Note that both the heat and distance scales are different between slow- and fast-spreading ridge models.
530 The Lower Oceanic Crust

meter along axis in the model shown), the relatively high geochemical stratigraphy (MacLeod and Yaouancq, 2000)
temperatures and shallow thermal gradients at depth mean and cooling rates (Coogan et al., 2002b) observed in the
that small additions of heat lead to the generation of a large Oman ophiolite and the axial topography of the EPR (Shah
region that is above the liquidus (i.e., an even larger partially and Buck, 2001). They showed that at least the upper 25% of
molten region would exist). Since large magma chambers have the plutonic section must form in a gabbro glacier style
not been imaged at slow-spreading ridges using geophysical (Figure 2(f)); they could not determine how the deeper por-
techniques, either the majority of MORBs do not crystallize to tion of the crust formed. This minimum value is consistent
this extent at depth or the hydrothermal circulation must with the minimum amount of crystal subsidence required to
efficiently extract heat to a great depth. A model in which explain the elevated chlorine concentrations in the upper gab-
hydrothermal cooling extracts heat from such a deep magma bros at Hess Deep (Gillis et al., 2003; Section 4.14.2.3.2.2).
body is shown in Figure 22(c). In this scenario, temperature
gradients steep enough to drive large amounts of crystallization
4.14.3.5 Crystallization, Mixing, Assimilation, and
at 15-km depth require hydrothermal circulation to penetrate
Magmatic Metasomatism
almost to this depth. If hydrothermal fluids only penetrate to
shallow levels, then heat must still be conducted from the The evolved compositions of MORBs compared to primary
magma chamber to the base of the hydrothermal system, mantle melts (Figure 6) and the fractionation of incompatible
leading to limited crystallization. Even in the model shown elements between the upper and lower crust (Figure 17) dem-
in Figure 22(c), where the hydrothermal heat flux is  15 times onstrate that crystal–melt separation is an important process in
larger than the latent heat released by the magma, temperatures generating MORBs and forming the lower oceanic crust. There
low enough for zircon saturation are only achieved within is strong evidence that assimilation, mixing, and magmatic
10 km of the surface. In this case, the shallow crust would metasomatism are also important processes.
be very cool and shallow-level intrusions would cool more The overenrichment of chlorine with respect to other sim-
rapidly than has been observed, unless they were very large ilarly incompatible elements in both the upper and lower crust
(Figure 11). While the steady-state thermal models shown in at fast-spreading ridges suggests that assimilation is an impor-
Figure 22 are clearly simplifications, they provide some insight tant process (Sections 4.14.2.1.2 and 4.14.2.3.2.2). The lack
into the basic physics. Latent heat release means that large of a similar overenrichment in chlorine at slow-spreading
amounts of crystallization at depths below 10 km and within ridges can be interpreted either as evidence that assimilation
a few kilometer of the ridge axis require the penetration of is a less important process or that the material assimilated is
substantial fluid fluxes to depths within a few kilometer at less chlorine-rich (Michael and Cornell, 1998). Assimilation at
which crystallization is occurring. the base of the sheeted dike complex causes the magma com-
Thermal models for fast-spreading ridges are, perhaps, bet- position to move toward the average composition of the dikes.
ter constrained by observations than those for slow-spreading For example, for a 1-km-wide magma chamber roof, the base
ridges, and the thermal structure is likely to be closer to a steady of a thousand, 1-m-wide, dikes could be assimilated. This
state, meaning that the observational constraints are more would integrate the magma composition emplaced over
likely to be generally applicable. For purely conductive cooling 10 Ky into the AMC, decreasing the variability in basalt com-
with no latent heat release, the axial region is much hotter than positions. This provides a further mechanism, in addition to
at slow-spreading ridges (Sleep, 1975; Figure 22(d)). Figure 22 simple magma mixing, to explain the smaller diversity in trace
(e) and (f) shows the general difference in the distribution of element compositions of basalts erupted at fast-spreading
hydrothermal heat sinks required by the two endmember ridges compared to those at slow-spreading ridges, as reported
models of crustal accretion. For a gabbro glacier model by Rubin and Sinton (2007).
(Figures 2(f) and 22(e)), all hydrothermal circulation can be Commonly observed disequilibrium between crystals and
(but does not have to be) confined to the upper crust (lavas host glass in MORBs (Figure 5) suggests that magma mixing is
and dikes) with most of the latent heat extracted through the a common process in the lithosphere at MORs. Two features of
roof of the AMC. Off-axis hydrothermal circulation can remove this mixing stand out. Firstly, crystals are generally too primi-
the specific heat of the solidified lower crust, leading to a tive to be in equilibrium with their host glass, suggesting that
wide range of possible thermal structures. For a sheeted sill the more evolved mixing end-member is (nearly) aphyric and
model (Figures 2(g) and 22(f)), off-axis hydrothermal circu- the more primitive one is phyric. Secondly, crystals in MORBs
lation is required throughout the lower crust in order to extract from slow-spreading ridges are more abundant (Figure 4) than
the latent heat. This requires that the lower crust is cool close to those at fast-spreading ridges. These observations, along with
the ridge axis. Additionally, to remove the latent heat of crys- the more evolved MORB compositions at fast-spreading ridges
tallization from directly beneath the axis, hydrothermal fluid (Figure 6), are consistent with a greater ratio of aphyric,
must either penetrate to within  1 km of the axis throughout evolved magma to primitive, phyric magma at fast-spreading
the lower crust to allow horizontal heat conduction or sills ridges. Most likely, the evolved aphyric melt is derived from
must be thick enough that magma convection can transport interstitial melt compacted out of mush zones. This process
heat laterally. This is shown in Figure 22(f) where, even in a will be more efficient at fast-spreading ridges, where there is a
model in which the crust >2 km off axis is hydrothermally steady-state AMC in which these evolved melts can pond, and
cooled to 0  C, latent heat cannot be efficiently conducted an underlying mush zone from which interstitial melt can be
from directly on axis leading to a wide axial region that is continuously compacted (e.g., Henstock, 2002; Sinton and
fully molten. Maclennan et al. (2005) calibrated a conceptually Detrick, 1992), as indicated by the chemical stratigraphy of
similar but more sophisticated thermal model against the the lower crust (Figure 20).
The Lower Oceanic Crust 531

A number of lines of evidence suggest that melt–crystal phases are 700–900  C where magmatic amphibole and zircon
reaction and interstitial melt migration play important roles are present. Thus, small amounts of melt can exist in the crust
in the chemical evolution of lower crustal crystal mush zones. down to temperatures as low as 700  C. By contrast, lower
These include the following: (1) the occurrence of a large range crustal rocks can be completely solid at temperatures above the
in incompatible element abundances in clinopyroxene at an liquidus of the parental melt if they are cumulates with no
almost constant Mg# (Figure 13(c) and 13(d)); (2) greater trapped interstitial melt. Thus, a seismically determined melt
fractionation of more-to-less incompatible elements in clino- fraction of a few percent is consistent with a wide range of
pyroxene than is predicted by fractional crystallization temperatures (<800 to >1200  C), depending on the petrol-
(Figure 14); (3) clinopyroxene crystals that show extreme zon- ogy of the lower crust, as emphasized by Maclennan et al.
ing in incompatible element abundances and more-to-less (2004). Lower crustal samples formed at slow-spreading ridges
incompatible element ratios, with limited variations of com- generally contain small amounts of interstitial, low crystalliza-
patible elements (Figure 15); (4) poorly correlated major ele- tion temperature, accessory phases, whereas those from fast-
ment composition of plagioclase and olivine in some sections spreading ridges do not (except in the uppermost portion of
of drill core from ODP Hole 735B (Dick et al., 2002); (5) the the crust). Thus, low seismic velocities in the lower crust,
concentration of oxides in layers in gabbros from slow- indicative of small melt fractions, may reflect much higher
spreading ridges and evidence for disequilibrium between the temperatures at fast- than at slow-spreading ridges.
melts that precipitated them and the silicate phases (Dick et al.,
1991a,b; Natland et al., 1991; Ozawa et al., 1991); (6) the
4.14.3.7 Hydrothermal Circulation in the Lower Oceanic
occurrence of interstitial melt in lower crustal xenoliths that
Crust
is not in equilibrium with the crystals (Section 4.14.2.1.1);
and (7) structures in ophiolites (e.g., Bedard, 1993). This Hydrothermal circulation is critical for the removal of latent
process may also be responsible for generating U-series dise- heat from the lower crust but is poorly understood
quilibria in erupted MORBs (e.g., Van Orman et al., 2006). (Section 4.14.3.4). Hydrothermal circulation is initiated at
temperatures just below the solidus of gabbros that contain
evolved interstitial melt but is mainly limited to a relatively
4.14.3.6 Near-Solidus Processes
narrow temperature interval, at least at fast-spreading ridges
The temperature interval between the crystal mush becoming (600–750  C; Coogan et al., 2002a; Gillis, 1995; Manning
sufficiently rigid so that it behaves like solid and hydrothermal et al., 1996). However, the fluid flux at these temperatures is
fluids penetrating into the system is important for at least two unknown and may be small. It has even been suggested that
reasons. Firstly, seismic refraction data suggest that at most hydrothermal fluids penetrate rock at high enough tempera-
fast-spreading ridges and some slow-spreading ridges, the tures to lead to hydration melting (Koepke et al., 2005a),
lower crust is near its solidus, with just a small melt fraction although this process cannot explain the existence of
(Section 4.14.2.2). The implication of this physical state for chlorine-poor, fluorine-rich amphibole distributed along
the temperature within the crust depends upon the processes grain boundaries in many oceanic gabbros from slow-
operating in the near-solidus crystal mush. Secondly, it pro- spreading ridges (Coogan et al., 2001a). The temperature
vides insight into the temperature drop across the thermal drop across the thermal boundary layer may be greater at
boundary layer between magma and hydrothermal fluids. deeper levels in the crust at fast-spreading ridges, where
The magnitude of this temperature drop, along with the thick- evolved rocks appear to be rare. At slow-spreading ridges,
ness of the boundary layer, controls the heat flux from the mineralogical evidence for high-temperature hydrothermal al-
magma to the hydrothermal system. teration is generally associated with deformation (Dick et al.,
Theoretical relationships between the melt fraction in a 1991a,b; Gillis et al., 1993b; Stakes et al., 1991). On average,
crystal mush, rheology of the mush, and temperature are the plutonic rocks from both fast- and slow-spreading ridges
complex, depending on a number of poorly constrained vari- are less mineralogically altered than samples of the overlying
ables. During simple solidification of a basaltic melt, the crys- sheeted dike complex. Although this may be taken as evidence
tallinity as a function of temperature, even without physical of more limited fluid flow, two points must be kept in
crystal–melt separation, depends on the crystallization process. mind. Firstly, thermodynamic models suggest that very high-
For example, solidification occurs over a much smaller temperature fluid flow will leave little imprint on the mineral-
temperature interval during equilibrium than during ogy of oceanic gabbros (McCollom and Shock, 1998). Indeed,
fractional crystallization. Simple models that assume that the even mineralogically unaltered samples can show shifts in their
solidus temperature can be extrapolated from experimentally d18O (Alt and Bach, 2006; Gao et al., 2006). Secondly, if fluid
determined (approximating equilibrium crystallization) flow is focused into widely separated (100s of meter to
temperature–melt fraction trends at high melt fractions (e.g., kilometer) channels, chemical and mineralogical alteration
Figure 7 in Sinton and Detrick, 1992) do not account for may be confined to these zones while the thermal impact
processes such as densification through compaction (Hunter, extends beyond these zones (Coogan et al., 2006).
1996) and the low solidus temperature of differentiated melts
with elevated volatile contents. An alternative way to deter-
4.14.3.8 Working Models and Future Questions
mine the solidus temperature is from geothermometry of
late-stage phases in oceanic gabbros. Figure 23(a) depicts the formation of lower oceanic crust at
As described in Section 4.14.2.2.3, solidification tempera- fast-spreading ridges. Parental magma is extracted from the
tures determined using geothermometry of late magmatic mantle and may either be ponded at a rheological boundary
532 The Lower Oceanic Crust

Fast-spreading Evolved
upper

MC
Increased magma ?Pito Deep? gabbros

A
supply, increased

er
spreading rate,

ow
towards segment

all
Sh
center, decreased
off-axis hydrother- A
mal heat extraction

released at this depth


% of total latent heat
released at this depth
% of total latent heat

C
AM
MC
A

SH
er

MU
ep
?Hess Deep?

De

AL
Dikes

ST
Lavas Hornfels

Y
CR
upper
Dikes

Replenishing primitive melt


gabbros

interacts with mush zone


Sub-
Gabbros vertically
foliated
gabbros

1 km A
Evolved interstitial melt is compacted
1 km
out of subsiding mush
(a) Primitive Evolved
Magma composition

Slow spreading
n
released at this depth

ei
% of total latent heat

ng
c ha gime
al l re
ati
Sp erma
th Crystallization
distributed
st over a
r cru greater depth
pe
Up range

s te
luton oti
d
rop Peri
bb
Ga
Mantle ‘flow’

Drawn for Melt impregnation


axial region into peridotite

Time 2

Extensive
Temporal change 2.5 km interstitial
Time 1
in thermal regime melt–rock
reaction
(b) 2.5 km

Figure 23 Cartoons summarizing plausible models of the processes operating in the lower oceanic crust at (a) fast-spreading ridges and (b)
slow-spreading ridges. At fast-spreading ridges, the upper portion of the lower crust forms via crystal subsidence from the AMC and evolved interstitial
The Lower Oceanic Crust 533

at the Moho, as suggested by some geophysical data, or be AMC is compositionally zoned with more evolved melt at the
transported directly into the crust. Either way, some primitive cooler off-axis edges, then crystal subsidence will lead to more
magma replenishes the AMC at the base of the sheeted dike evolved shallow-level plutonic rocks (from the chamber mar-
complex frequently enough (<100s of years) to prevent this gins) and more primitive deeper cumulates (from the chamber
body from freezing. Primitive phenocrysts may either grow center). Secondly and perhaps a better explanation of the data
from these melts or be entrained into them from the crystal for samples from Hess Deep, evolved compositions of shallow
mush zone in the lower crust. Since most magmas are erupted gabbros may be due to extensive reaction with interstitial melt
in a narrow zone at the axis (e.g., Hooft et al., 1996), they must that was not completely removed from the crystal mush as it
generally pass through the AMC. Here, the replenishing prim- subsided. There are several lines of evidence suggesting that
itive melts mix with evolved, near-aphyric melts compacted this is the case at Hess Deep based on bulk rock (Figure 16(e))
out of the underlying mush zone to produce the basalts. Sig- and mineral (Figure 13(b)) compositions. The difference be-
nificant variation in the depth of the AMC at a given spreading tween Hess Deep and Pito Deep likely reflects faster cooling at
rate (Figure 9) suggests temporal and/or spatial variations in Hess Deep, with less time for expulsion of the interstitial melt
the heat budget of the AMC. Whether these variations are due from the compacting crystal mush. This could be due to Hess
to variations in magma supply or variations in the efficiency of Deep gabbros forming near a segment end (front Figure 23(a))
hydrothermal heat extraction or both is unclear. The AMC is and the Pito Deep gabbros forming in a ridge segment center
overlain either by evolved gabbros, which crystallized along (rear Figure 23(a)).
the roof of a subsiding magma chamber, or by hornfelsic rocks Figure 23(b) depicts a working model for magmatic pro-
metamorphosed by an ascending magma body. When heat cesses in the lower oceanic crust at a slow-spreading ridge.
supply is high relative to heat extraction, partial melting and Magma probably ponds near the base of the lithosphere (e.g.,
assimilation of these roof rocks is efficient and the AMC mi- Cannat et al., 1999; Nisbet and Fowler, 1978) and begins to
grates upward while the melt within it becomes contaminated crystallize slowly. The depth of initial crystallization is likely to
with altered roof rock. be spatially and temporally variable, dependent on the effi-
Crystal subsidence from the AMC is required for this body ciency of hydrothermal heat extraction. Small amounts of
to exist at a steady state. Much more rapid cooling rates at primitive cumulates form at depth, but the latent heat released
shallow levels (Coogan et al., 2007a) mean that crystallization buffers the lithospheric mantle temperature to near the liqui-
here must be faster than at depth. However, what proportion of dus, preventing magmas from becoming substantially fraction-
the plutonic lower crust is constructed through crystal ated and preventing formation of large plutonic bodies (Sleep
subsidence and what proportion is constructed through crys- and Barth, 1997). Melts from this region are probably period-
tallization in situ are important questions, as is the mechanism ically emplaced to shallower levels. Primitive cumulates are
of crystal subsidence (pervasive flow? distributed shear zones? precipitated in these deep magma chambers, leading to an
vertical slurry flow?). Crystallization in situ in the lower oce- evolved bulk composition of the overlying crust (Coogan
anic crust (e.g., Boudier et al., 1996; Kelemen et al., 1997) et al., 2001b; Dick et al., 2000). Each magma batch added to
requires hydrothermal circulation through the plutonic section the crust forms one or more thin sill(s), such as those seen in
to remove the latent heat of crystallization. If magmas pond ODP Holes 923A and 1309D (Figure 21; Coogan et al., 2000a;
and crystallize at more than one level as they are transported Natland and Dick, 2002; Suhr et al., 2008). These crystallize
through the lower crust, then this will lead to a stratified lower rapidly to form a crystal mush, but repeated sill emplacement
crust, with more evolved rocks at shallower levels. The very builds a mush column on the order of 500 m high during any
primitive shallow gabbros at Pito Deep and the general lack of magmatic event (e.g., Dick et al., 2000, 2002). These mush
any systematic compositional variability with depth in the zones cool slowly because they are too viscous to convect and
lower crust of the Oman ophiolite (Figure 20) are not consis- because they have a low thermal diffusivity (e.g., Snyder et al.,
tent with this model (see Figure 10 in Perk et al., 2007). 1994). Slow cooling of the mush allows extensive interaction
Notably, the inverse is not true; more evolved shallow plutonic between the interstitial melt and crystals in the mush zone.
rocks do not require a sheeted sill model and can be explained Melt–rock reaction may almost completely eradicate any evi-
in at least two ways within a gabbro glacier model. Firstly, if the dence for how the mush zone was originally constructed.

melt is compacted back into the AMC. What proportion of the deeper crust crystallizes in situ is unclear and is probably variable spatially (e.g., toward
segment ends) and temporally (e.g., more when large faults allow deep hydrothermal circulation). Primitive melts feed the AMC and mix with evolved
melts resident within it. Assimilation of the roof of the AMC is common, leading to contamination of the lower crust. The latent heat of crystallizing the
gabbroic section of the lower crust is lost predominantly through the AMC roof with lesser amounts lost at deeper levels. The variation in bulk-rock Mg#
with depth in the crust depends on the extent of crystallization of melts prior to their reaching the AMC, whether the AMC is compositionally zoned, and
the amount of interstitial melt that reacts with the cumulate crystals. Off-axis melt sills are only likely to be geophysically observed in the lower crust
where it is hot and these cool slowly. At slow-spreading ridges, melt starts to pool at depth (10 km) and small amounts of crystallization occur, but
most crystallization and lower crust formation occurs at shallow levels where cooling is more efficient. Small intrusions are episodically emplaced,
gradually building a mush zone that erupts episodically after replenishment and mixing events. Large amounts of interstitial melt remain trapped within
the crystal mush, which, in addition to small amounts of crystallization at depth, leads to a relatively evolved bulk composition for the lower crust.
Temporal and spatial variation in magma supply and hydrothermal circulation causes some areas to have cooler lithosphere, leading to more
crystallization at depth and a large amount of peridotite being mixed into the crust over a large depth range. Note that the cartoon includes a low-angle
detachment fault simply to emphasize that these lead to exposure of the lower oceanic crust and that the sampling has been biased toward these. No
attempt is made to depict the tectonics accurately, including mantle uplift, which is simply shown as ‘flow’ rather than including faulting.
534 The Lower Oceanic Crust

Ductile strain may also partition into this partially molten Babcock JM, Harding AJ, Kent GM, and Orcutt JA (1998) An examination of along-axis
zone, leading to crystal plastic deformation. Felsic veins form variations in magma chamber width and crustal structure on the East Pacific Rise
between 13 300 N and 12 200 N. Journal of Geophysical Research 103(B12):
when an almost solid mush fractures and the last increments of
30451–30467.
interstitial melt are sucked into the fractures. Primitive magmas Bach W, Alt JC, Niu Y, Humphries SE, Erzinger J, and Dick HJB (2001) The
and their phenocrysts mix with more evolved melts derived geochemical consequences of late-stage low-grade alteration of lower ocean crust at
from within the crystal mush zone during transport to the the SW Indian Ridge: Results from ODP Hole 735B (Leg 176). Geochimica et
surface, explaining the disequilibrium between crystals and Cosmochimica Acta 65(19): 3267–3287.
Baines AG, Cheadle MJ, John BE, Grimes CB, Schwartz JJ, and Wooden JL (2009)
their host melt (Figure 5). Temporal and spatial variations in SHRIMP Pb/U zircon ages constrain gabbroic crustal accretion at Atlantis Bank on
the supply of magma from the mantle and hydrothermal heat the ultraslow-spreading Southwest Indian Ridge. Earth and Planetary Science
extraction will lead to fluctuations in the thickness and tem- Letters 287: 540–550.
perature structure of the lithosphere. When magma supply is Baker ET and German CR (2004) On the global distribution of hydrothermal vents.
In: German CR, Lin J, and Parson LM (eds.) Mid-Ocean Ridges: Hydrothermal
lower or hydrothermal heat extraction higher, the whole sys-
Interactions Between the Lithosphere and Oceans, Geophysical Monograph Series,
tem will be cooler, and melts will pond deeper and crystallize vol. 148, pp. 245–266. Washington, DC: American Geophysical Union.
more at depth. The ‘crust’ will then be composed of small Baker ET, Martinez F, Resing JA, Walker SL, Buck NJ, and Edwards MH (2010)
plutons encased in large amounts of peridotite (e.g., Cannat, Hydrothermal cooling along the Eastern Lau Spreading Center: No evidence for
1993, 1996). More robust magma supply will lead to melt discharge beyond the neovolcanic zone. Geochemistry, Geophysics, Geosystems
11: Q08004.
ponding at shallower levels and a near continuous plutonic Barclay AH, Toomey DR, and Solomon SC (2001) Microearthquake characteristics and
layer of lower crust. Tectonic deformation style will also crustal Vp/Vs structure at the Mid-Atlantic Ridge, 35 N. Journal of Geophysical
change with magma supply. The interplay between variations Research 106: 2017–2034.
in magma supply and tectonic process may lead to mixing of Barling J, Hertogen J, and Weis D (1997) Whole-rock geochemistry and Sr-, Nd-, and
Pb- isotopic characteristics of undeformed, deformed, and recrystallized gabbros
‘slices’ of lower crust of different ages (Schwartz et al., 2005).
from Sites 921, 922, and 923 in the MARK area. In: Karson JA, Cannat M, Miller DJ,
It is appropriate to end with a word of caution. Sampling of and Elthon D (eds.) Proceedings of the Ocean Drilling Program, Scientific Results
the lower oceanic crust, to date, is limited and has occurred in vol. 153, pp. 351–362. College Station, TX: Ocean Drilling Program.
places where the opportunity presents itself – not all of these Barnes SJ (1986) The effect of trapped liquid crystallization on cumulus mineral
may be representative of ‘average’ lower oceanic crust. For compositions in layered intrusions. Contributions to Mineralogy and Petrology
93: 524–531.
example, the most detailed sampling of crust from slow- Batiza R and Niu YL (1992) Petrology and magma chamber processes at the East Pacific
spreading ridges has been in areas where gabbros are exposed Rise  9 300 N. Journal of Geophysical Research 97(B5): 6779–6797.
at the surface through detachment faulting. Whether the crustal Bedard JH (1991) Cumulate recycling and crustal evolution in the Bay of Islands
(and mantle) magma plumbing system is fundamentally dif- ophiolite. Journal of Geology 99: 225–249.
Bedard JH (1993) Oceanic crust as a reactive filter: Synkinematic intrusion,
ferent during these times versus during times of ‘normal’
hybridization, and assimilation in an ophiolitic magma chamber, western
crustal accretion is unclear. Likewise, at fast-spreading ridges, Newfoundland. Geology 21: 77–80.
sampling of the lower oceanic crust has been focused on the Benoit M, Ceuleneer G, and Polve M (1999) The remelting of hydrothermally altered
upper <1000 m leaving uncertainty about the nature of the peridotite at mid-ocean ridges by intruding mantle diapirs. Nature 402: 514–518.
deeper crust. In the long term, sampling will have to concen- Biggar GM and Humphries DJ (1981) The plagioclase, forsterite, diopside, liquid
equilibrium in the system CaO–Na2O–MgO–Al2O3–SiO2. Mineralogical Magazine
trate on determining the spatial and temporal variability in 44: 309–314.
lower crustal accretion processes and how this relates to spatial Blake S (1981) Volcanism and the dynamics of open magma chambers. Nature
and temporal variability in tectonic and hydrothermal pro- 289: 783–785.
cesses in a limited number of places. This will require a com- Bloomer SH, Meyer PS, Dick HJB, Ozawa K, and Natland JH (1991) Textural and
mineralogic variations in gabbroic rocks from Hole 735B. In: Von Herzen RP and
bination of deep drilling, offset drilling, and studies of tectonic
Robinson PT (eds.) Proceedings of the Ocean Drilling Program, Scientific Results
windows. Coupling of these studies to colocated investigations vol. 118, pp. 21–39. College Station, TX: Ocean Drilling Program.
of the upper crust (lavas and dikes) and upper mantle will Boudier F, Nicolas A, and Ildefonse B (1996) Magma chambers in the Oman
allow greater understanding of the MOR system. ophiolite: Fed from the top and the bottom. Earth and Planetary Science Letters
144: 239–250.
Bown JW and White RS (1994) Variation with spreading rate of oceanic crustal
thickness and geochemistry. Earth and Planetary Science Letters 121: 435–449.
Acknowledgments Browning P (1982) The Petrology, Geochemistry, and Structure of the Plutonic Rocks of
the Oman Ophiolite. PhD Thesis, The Open University.
Formal and informal reviews by Henry Dick, Roberta Rudnick, Bruguier NJ and Livermore RA (2001) Enhanced magma supply at the southern East
Scotia Ridge: Evidence for mantle flow around the subducting slab? Earth and
Kathi Faak, and Timo Kirchner are gratefully acknowledged.
Planetary Science Letters 191: 129–144.
Bryan WB (1983) Systematics of modal phenocryst assemblages in submarine
basalts: Petrological implications. Contributions to Mineralogy and Petrology
References 83: 62–74.
Canales JP, Detrick RS, Carbotte SM, et al. (2005) Upper crustal structure and axial
Agar SM and Lloyd GE (1997) Deformation of Fe-Ti oxides in gabbroic shear zones topography at intermediate spreading ridges: Seismic constraints from the southern
from the MARK area. In: Karson JA, Cannat M, Miller DJ, and Elthon D (eds.) Juan de Fuca Ridge. Journal of Geophysical Research 110: B12104.
Proceedings of the Ocean Drilling Program, Scientific Results, vol. 153, Canales JP, Nedimovic MR, Kent GM, Carbotte SM, and Detrick RS (2009) Seismic
pp. 123–141. College Station, TX: Ocean Drilling Program. reflection images of a near-axis melt sill within the lower crust at the Juan de Fuca
Allerton S and Tivey MA (2001) Magnetic polarity structure of the lower oceanic crust. ridge. Nature 460: 89–93.
Geophysical Research Letters 28(3): 423–426. Cann JR (1974) A model for oceanic crustal structure developed. Geophysical Journal
Alt JC and Bach W (2006) Oxygen isotope composition of a section of lower oceanic of the Royal Astronomical Society 39: 169–187.
crust, ODP Hole 735B. Geochemistry, Geophysics, Geosystems 7: Q12008. Cann JR, Strens MR, and Rice A (1985) A simple magma-driven thermal balance model
Alt JC, Kinoshita H, and Stokking LB (1993) Proceedings of the Ocean Drilling for the formation of volcanogenic massive sulphides. Earth and Planetary Science
Program, Initial Reports, vol. 148. College Station, TX: Ocean Drilling Program. Letters 76: 123–134.
The Lower Oceanic Crust 535

Cannat M (1991) Plastic deformation at an oceanic spreading ridge: A microstructural Coogan LA and Hinton RH (2006) Do the trace element compositions of detrital zircons
study of the site 735 gabbros (Southwest Indian Ridge). In: Von Herzen RP and require Hadean continental crust? Geology 34: 633–636.
Robinson PT (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, Coogan LA, Howard KA, Gillis KM, et al. (2006) Chemical and thermal constraints on
vol. 118, pp. 399–408. College Station, TX: Ocean Drilling Program. focussed fluid flow in the lower oceanic crust. American Journal of Science
Cannat M (1993) Emplacement of mantle rocks in the seafloor at mid-ocean ridges. 306: 389–427.
Journal of Geophysical Research 98(B3): 4163–4172. Coogan LA, Jenkin GRT, and Wilson RN (2002b) Constraining the cooling rate of the
Cannat M (1996) How thick is the magmatic crust at slow spreading ridges? Journal of lower oceanic crust: A new approach applied to the Oman ophiolite. Earth and
Geophysical Research 101(B2): 2847–2857. Planetary Science Letters 199: 127–146.
Cannat M, Cann JR, and Maclennan J (2004) Some hard rock constraints on the supply Coogan LA, Jenkin GRT, and Wilson RN (2007a) Contrasting cooling rates in the
of heat to mid-ocean ridges. In: German CR, Lin J, and Parson LM (eds.) Mid- oceanic lithosphere at fast- and slow-spreading mid-ocean ridges derived from
Ocean Ridges: Hydrothermal Interactions Between the Lithosphere and Oceans. geospeedometry. Journal of Petrology 48: 2211–2231.
Geophysical Monograph Series, vol. 148, pp. 111–150. Washington, DC: Coogan LA, Kempton PD, Saunders AD, and Norry MJ (2000a) Evidence from
American Geophysical Union. plagioclase and clinopyroxene major and trace element compositions for melt
Cannat M and Casey JF (1995) An ultramafic lift at the Mid-Atlantic ridge: Successive aggregation within the crust beneath the Mid-Atlantic Ridge. Earth and Planetary
stages of magmatism in serpentinised peridotites from the 15 N region. Science Letters 176(2): 245–257.
In: Vissers RLM and Nicolas A (eds.) Mantle and Lower Crustal Exposures in Ocean Coogan LA, MacLeod CJ, Dick HJB, et al. (2001a) Whole-rock geochemistry of gabbros
Ridges and in Ophiolites, pp. 5–34. Dordrecht: Kluwer. from the Southwest Indian Ridge: Constraints on geochemical fractionations
Cannat M, Ceuleneer G, and Fletcher J (1997) Localization of ductile strain and the between the upper and lower oceanic crust and magma chamber processes at (very)
magmatic evolution of gabbroic rocks drilled at the Mid-Atlantic Ridge (23 N). slow-spreading ridges. Chemical Geology 178(1–4): 1–22.
In: Karson JA, Cannat M, Miller DJ, and Elthon D (eds.) Proceedings of the Ocean Coogan LA, Manning CE, Wilson RN, and EIMF (2007b) Oxygen isotope evidence for
Drilling Program, Scientific Results, vol. 153, pp. 77–98. College Station, TX: short-lived high-temperature pervasive fluid flow in the lower oceanic crust at
Ocean Drilling Program. fast-spreading ridges. Earth and Planetary Science Letters 260: 524–536.
Cannat M, Rommevaux-Jestin C, Sauter D, Deplus C, and Mendel V (1999) Formation Coogan LA, Saunders AD, Kempton PD, and Norry MJ (2000b) Evidence from oceanic
of the axial relief at the very-slow spreading Southwest Indian Ridge (49 to 69 E). gabbros for porous melt migration within a crystal mush beneath the Mid-Atlantic
Journal of Geophysical Research 104(B10): 22825–22843. Ridge. Geochemistry, Geophysics, Geosystems 1: 1044.
Cannat M, Karson JA, Miller DJ, et al. (1995) Proceedings of the Ocean Drilling Coogan LA, Thompson G, MacLeod CJ, et al. (2004) A combined basalt and peridotite
Program, Initial Reports, vol. 153. College Station, TX: Ocean Drilling Program. perspective on 14 million years of melt generation at the Atlantis Bank segment of
Carbotte S, Mutter C, Mutter J, and Ponce-Correa G (1998) Influence of magma supply the Southwest Indian Ridge: Evidence for temporal changes in mantle dynamics?
and spreading rate on crustal magma bodies and emplacement of the extrusive Chemical Geology 207: 13–30.
layer: Insights from the East Pacific Rise at lat 16 N. Geology 26: 455–458. Coogan LA, Wilson RN, Gillis KM, and MacLeod CJ (2001b) Near-solidus evolution of
Casey JF (1997) Comparison of major and trace-element geochemistry of abyssal oceanic gabbros: Insights from amphibole geochemistry. Geochimica et
peridotites and mafic plutonics with basalts from the MARK region of the Cosmochimica Acta 65(23): 4339–4357.
Mid-Atlantic Ridge. In: Karson JA, Cannat M, Miller DJ, and Elthon D (eds.) Coogan LA, Wilson R, Grimes G, MacLeod CJ, and Irving DHB (1999) Constraining the
Proceedings of the Ocean Drilling Program, Scientific Results, vol. 153, dynamics of reactive porous melt migration through slow spreading ridge gabbros.
pp. 181–241. College Station, TX: Ocean Drilling Program. Eos, Transactions, American Geophysical Union Fall Meeting Supplement, Abstract
Ceuleneer G and Cannat M (1997) High-temperature ductile deformation of Site 920 T11AA-1133.
peridotites. In: Karson JA, Cannat M, Miller DJ, and Elthon D (eds.) Proceedings of Costa F, Coogan LA, and Chakraborty S (2009) The time scales of magma mixing and
the Ocean Drilling Program, Scientific Results, vol. 153, pp. 23–34. College Station, mingling involving primitive melts and melt-mush interaction at mid-ocean ridges.
TX: Ocean Drilling Program. Contributions to Mineralogy and Petrology 159: 371–387.
Ceuleneer G, Monnereau M, and Amri I (1996) Thermal structure of a fossil mantle Crawford WC and Webb SC (2002) Variation in the distribution of magma in the lower
diapir inferred from the distribution of mafic cumulates. Nature 379: 149–153. crust and at the Moho beneath the East Pacific Rise at 9 -10 N. Earth and Planetary
Chakraborty S (1997) Rates and mechanisms of Fe-Mg interdiffusion in olivine at Science Letters 203: 117–130.
980-1300 C. Journal of Geophysical Research 102(B6): 12317–12331. Davis AS and Clague DA (1990) Gabbroic xenoliths from the northern Gorda Ridge:
Chen YJ (2001) Thermal effects of gabbros accretion from a deeper second melt lens at Implications for magma chamber processes under slow-spreading centers. Journal
the fast spreading East Pacific Rise. Journal of Geophysical Research 106(B5): of Geophysical Research 95(B7): 10885–10905.
8581–8588. DeLong SE and Chatelain C (1990) Trace-element constraints on accessory-phase
Chen Y and Morgan WJ (1990) A nonlinear rheology model for mid-ocean ridge axis saturation in evolved MORB magma. Earth and Planetary Science Letters
topography. Journal of Geophysical Research 95: 17583–17604. 101: 206–215.
Cherkaoui ASM, Wilcock WSD, Dunn RA, and Toomey DR (2003) A numerical model of Detrick RS, Buhl P, Vera E, et al. (1987) Multi-channel seismic imaging of a crustal
hydrothermal cooling and crustal accretion at a fast-spreading mid-ocean ridge. magma chamber along the East Pacific Rise. Nature 326(6108): 35–41.
Geochemistry, Geophysics, Geosystems 4: 8616. Dick HJB and Bryan WB (1978) Variation of basalt phenocryst mineralogy and rock
Cherniak DJ (2003) REE diffusion in feldspar. Chemical Geology 193: 25–41. composition in DSDP Hole 396B. In: Dmitriev L and Heirtzler J (eds.) Initial Reports
Christie DM and Sinton JM (1981) Evolution of abyssal lavas along propagating segments of the Deep Sea Drilling Project, vol. 46, pp. 215–225. Washington, DC: US
of the Galapagos spreading center. Earth and Planetary Science Letters 56: 321–335. Government Printing Office.
Cochran JR and Buck WR (2001) Near-axis subsidence rates, hydrothermal circulation, Dick HJB, Fisher RL, and Bryan WB (1984) Mineralogical variability of the uppermost
and thermal structure of mid-ocean ridge crests. Journal of Geophysical Research mantle along mid-ocean ridges. Earth and Planetary Science Letters 69: 88–106.
106(B9): 19233–19258. Dick HJB, Meyer PS, Bloomer SH, Kirby S, Stakes D, and Mawer C (1991a)
Collier J and Sinha M (1990) Seismic images of a magma chamber beneath the Lau Lithostratigraphic evolution of an in situ section of oceanic layer 3. In: Von
Basin back arc spreading center. Nature 346: 646–648. Herzen RP and Robinson PT (eds.) Proceedings of the Ocean Drilling Program,
Constantin M (1999) Gabbroic intrusions and magmatic metasomatism in harzburgites Scientific Results, vol. 118, pp. 439–540. College Station, TX: Ocean Drilling
from the Garrett transform fault: Implications for the nature of the mantle-crust Program.
transition at fast spreading ridges. Contributions to Mineralogy and Petrology Dick HJB and Natland JH (1996) Late-stage melt evolution and transport in the shallow
136: 111–130. mantle beneath the East Pacific Rise. In: Mével C, Gillis KM, Allan JF, and Meyer PS
Constantin M, Hekinian R, Bideau D, and Hebert R (1996) Construction of the oceanic (eds.) Proceedings of the Ocean Drilling Program, Initial Reports, vol. 147,
lithosphere by magmatic intrusions: Petrological evidence from plutonic rocks pp. 103–133. College Station, TX: Ocean Drilling Program.
formed along the fast-spreading East Pacific Rise. Geology 24(8): 731–734. Dick HJB, Natland JH, Alt JC, et al. (2000) A long in situ section of the lower oceanic
Coogan LA, Gillis KM, MacLeod CJ, Thompson G, and Hekinian R (2002a) Petrology crust: Results of ODP Leg 176 drilling at the Southwest Indian Ridge. Earth and
and geochemistry of the lower ocean crust formed at the East Pacific Rise and Planetary Science Letters 179: 31–51.
exposed at Hess Deep: A synthesis and new results. Geochemistry, Geophysics, Dick HJB, Ozawa K, Meyer PS, et al. (2002) Primary silicate mineral chemistry of a
Geosystems 3(11): 8604. 1.5-km section of very slow spreading lower ocean crust: ODP Hole 735B,
Coogan LA, Hain A, Stahl S, and Chakraborty S (2005) Experimental determination of Southwest Indian Ridge. In: Natland JH, Dick HJB, Miller DJ, and Von Herzen RP
the diffusion coefficient for calcium in olivine between 900 C and 1500 C. (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, vol. 176,
Geochimica et Cosmochimica Acta 69(14): 3683–3694. pp. 1–13. College Station, TX: Ocean Drilling Program.
536 The Lower Oceanic Crust

Dick HJB, Schouten H, Meyer PS, et al. (1991b) Tectonic evolution of the Atlantis II Gao Y, Hoefs J, Hellebrand E, von der Handt A, and Snow JE (2007) Trace element
Fracture Zone. In: Von Herzen PR and Robinson PT (eds.) Proceedings of the Ocean zoning in pyroxene from ODP Hole 735B gabbros: Diffusive exchange or
Drilling Program, Scientific Results, vol. 118, pp. 359–398. College Station, TX: synkinematic crystal fractionation. Contributions to Mineralogy and Petrology
Ocean Drilling Program. 153: 429–442.
Dixon JE, Clague DA, and Eissen JP (1986) Gabbroic xenoliths and host ferrobasalt Gao Y, Hoefs J, Przybilla R, and Snow JE (2006) A complete oxygen isotope profile
from the Southern Juan de Fuca Ridge. Journal of Geophysical Research 91(B3): through the lower oceanic crust, ODP Hole 735B. Chemical Geology 233: 217–234.
3795–3820. Garmany J (1989) Accumulations of melt at the base of young oceanic crust. Nature
Dodson MH (1973) Closure temperature in cooling geochronological and petrological 340: 628–632.
systems. Contributions to Mineralogy and Petrology 40: 259–274. Garrido CJ, Kelemen PB, and Hirth G (2001) Variation of cooling rate with depth in the
Dohmen R, Becker H-W, and Chakraborty S (2007) Fe-Mg diffusion in olivine I: lower crust formed at an oceanic spreading ridge: Plagioclase crystal size
Experimental determination between 700 and 1200  C as a function of composition, distributions in gabbros from the Oman ophiolite. Geochemistry, Geophysics,
crystal orientation and oxygen fugacity. Physics and Chemistry of Minerals 34(6): Geosystems 2: 1041.
389–407. Ghiorso MS and Sack RO (1995) Chemical mass transfer in magmatic processes IV.
Dosso L, Bougault H, and Joron J-L (1993) Geochemical morphology of the North A revised and internally consistent thermodynamic model for the interpolations of
Mid-Atlantic Ridge, 10 -24 N: Trace element-isotope complementarity. Earth and liquid-solid equilibria in magmatic systems at elevated temperatures and pressures.
Planetary Science Letters 120: 443–462. Contributions to Mineralogy and Petrology 119: 197–212.
Drouin M, Ildefonse B, and Godard M (2010) A microstructural imprint of melt Gillis KM (1995) Controls on hydrothermal alteration in a section of fast-spreading
impregnation in slow spreading lithosphere: Olivine-rich troctolites from the Atlantis oceanic crust. Earth and Planetary Science Letters 134: 473–489.
Massif, Mid Atlantic Ridge, 30 N, IODP Hole U1309D. Geochemistry, Geophysics, Gillis KM (1996) Rare earth element constraints on the origin of amphibole in gabbroic
Geosystems 11: Q06003. rocks from Site 894, Hess Deep. In: Mével C, Gillis KM, Allan JF, and Meyer PS
Dungan MA, Long PE, and Rhodes JM (1978) The petrography, mineral chemistry, and (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, vol. 147,
one-atmosphere phase relations of basalts from Site 395. In: Melson WG and pp. 59–75. College Station, TX: Ocean Drilling Program.
Rabinowitz PD (eds.) Initial Reports of the Deep Sea Drilling Project, vol. 45, Gillis KM (2008) The roof of an axial magma chamber: A hornfelsic heat-exchanger.
pp. 461–477. Washington, DC: US Government Printing Office. Geology 36: 299–302.
Dungan MA and Rhodes JM (1978) Residual glasses and melt inclusions in basalts Gillis KM, Coogan LA, and Chaussidon M (2003) Volatile behavior in the roof of an
from DSDP Leg 45 and 46: Evidence for magma mixing. Contributions to axial magma chamber from the East Pacific Rise. Earth and Planetary Science Letters
Mineralogy and Petrology 67: 417–431. 213: 447–462.
Dunn RA, Lekic V, Detrick RS, and Toomey DR (2005) Three-dimensional seismic Gillis KM, Mével C, and Allan J (eds.) (1993a) Proceedings of the Ocean Drilling
structure of the Mid-Atlantic Ridge (35 N): Evidence for focused melt supply Program, Initial Reports, vol. 147, College Station, TX: Ocean Drilling Program.
and lower crustal dike injection. Journal of Geophysical Research 110: B09101. Gillis KM and Roberts MD (1999) Cracking at the magma-hydrothermal transition:
Dunn RA, Toomey DR, Detrick RS, and Wilcock WSD (2001) Continuous mantle melt Evidence from the Troodos ophiolite, Cyprus. Earth and Planetary Science Letters
supply beneath an overlapping spreading center on the East Pacific Rise. Science 169: 227–244.
291: 1955–1959. Gillis KM, Thompson G, and Kelley DS (1993b) A view of the lower crustal component
Dunn RA, Toomey DR, and Solomon SC (2000) Three-dimensional seismic structure of hydrothermal systems at the mid Atlantic ridge. Journal of Geophysical Research
and physical properties of the crust and shallow mantle beneath the East Pacific Rise 98(B11): 19597–19619.
at 9 300 N. Journal of Geophysical Research 105(B10): 23537–23555. Girardeau J and Francheteau J (1993) Plagioclase-wehrlites and peridotites on the East
Durant DT and Toomey DR (2009) Evidence and implications of crustal magmatism on Pacific Rise (Hess Deep) and the Mid-Atlantic Ridge (DSDP Site 334): Evidence for
the flanks of the East Pacific Rise. Earth and Planetary Science Letters magma percolation in the oceanic upper mantle. Earth and Planetary Science Letters
287: 130–136. 115: 137–149.
Elthon D (1987) Petrology of gabbroic rocks from the Mid-Cayman Rise Spreading Godard M, Awaji S, Hansen H, et al. (2009) Geochemistry of a long in-situ section of
Center. Journal of Geophysical Research 92(B1): 658–682. intrusive slow-spread oceanic lithosphere: Results from IODP Site U1309 (Atlantis
Elthon D (1993) The crystallization of mid-ocean ridge basalts at moderate and high Massif, 30 N Mid-Atlantic Ridge). Earth and Planetary Science Letters
pressures. European Journal of Mineralogy 5: 1025–1037. 279: 110–122.
Elthon D, Casey JF, and Komor S (1982) Mineral chemistry of ultramafic cumulates Gregory RT and Taylor HPJ (1981) An oxygen isotope profile in a section of Cretaceous
from the North Arm Massif of the Bay of Islands ophiolite: Evidence for oceanic crust, Samail ophiolite, Oman: Evidence for d18O buffering of the oceans by
high-pressure crystal fractionation of oceanic basalts. Journal of Geophysical deep (>5 km) seawater-hydrothermal circulation at mid-ocean ridges. Journal of
Research 87(B10): 8717–8734. Geophysical Research 86(B4): 2737–2755.
Elthon D, Stewart M, and Ross DK (1992) Compositional trends of minerals in oceanic Grevemeyer I, Weigel W, and Jennrich C (1998) Structure and ageing of oceanic crust at
cumulates. Journal of Geophysical Research 979(B11): 15189–15199. 14 S on the East Pacific Rise. Geophysical Journal International 135: 573–584.
Expedition 312 Scientists (2006) Superfast spreading rate crust 3: A complete in situ Grimes CB, John BE, Cheadle MJ, and Wooden JL (2008) Protracted construction of
section of upper oceanic crust formed at a superfast spreading rate. Integrated gabbroic crust at a slow spreading ridge: Constraints from 206Pb/238U zircon ages
Ocean Drilling Program Preliminary Report, 312. from Atlantis Massif and IODP Hole U1309D (30 N, MAR). Geochemistry,
Expedition Scientific Party (2005a) Oceanic core complex formation, Atlantis Geophysics, Geosystems 9(8): Q08012.
Massif – Oceanic core complex formation, Atlantis Massif, Mid-Atlantic Ridge: Grimes CB, John B, Cheadle MJ, et al. (2009) On the occurrence, trace element
Drilling into the footwall and hanging wall of a tectonic exposure of deep, young geochemistry, and crystallization history of zircon from in situ ocean lithosphere.
oceanic lithosphere to study deformation, alteration, and melt generation. Contributions to Mineralogy and Petrology 158: 757–783.
Integrated Ocean Drilling Program Preliminary Report, 304. Grove TL, Kinzler RJ, and Bryan WB (1992) Fractionation of mid-ocean ridge basalt
Expedition Scientific Party (2005b) Oceanic core complex formation, Atlantis (MORB). In: Phipps Morgan J, Blackman DK, and Sinton JM (eds.) Mantle Flow and
Massif – Oceanic core complex formation, Atlantis Massif, Mid-Atlantic Ridge: Melt Generation at Mid-Ocean Ridges. Geophysical Monograph Series, vol. 71,
Drilling into the footwall and hanging wall of a tectonic exposure of deep, young pp. 281–311. Washington, DC: American Geophysical Union.
oceanic lithosphere to study deformation, alteration, and melt generation. Integrated Gudmundsson A (1986) Formation of crustal magma chambers in Iceland. Geology
Ocean Drilling Program Preliminary Report, 305. 14: 164–166.
Feig ST, Koepke J, and Snow JE (2006) Effect of water on tholeiitic basalt phase Hamlin B and Allegre CJ (1985) Large-scale regional units in the depleted upper mantle
equilibria: An experimental study under oxidizing conditions. Contributions to revealed by an isotope study of the Southwest Indian Ridge. Nature 315: 196–199.
Mineralogy and Petrology 152: 611–638. Harding AJ, Kent GM, and Collins JA (2000) Initial results from a multichannel seismic
Fisher AT (1998) Permeability within basaltic oceanic crust. Reviews of Geophysics survey of the Lau Back Arc basin. Eos, Transactions, American Geophysical Union
36(2): 143–182. 81: T61C-16. Fall Meeting Supplement.
Francheteau J, Armijo R, Cheminee JL, Hekinian R, Lonsdale P, and Blum N (1990) Harding AJ, Orcutt JA, Kappus ME, et al. (1989) Structure of young oceanic crust at
1 Ma East Pacific Rise oceanic crust and uppermost mantle exposed by rifting in Hess 13 N on the East Pacific Rise from expanding spread profiles. Journal of
Deep (equatorial Pacific Ocean). Earth and Planetary Science Letters 101: 281–295. Geophysical Research 94(B9): 12163–12196.
Gaggero L and Gazzotti M (1996) Primary and secondary oxides, sulphides and Hart SR, Blusztajn J, Dick HJB, Meyer PS, and Muehlenbachs K (1999) The fingerprint
accessory minerals in Mid-Atlantic gabbros: Mineralogy and petrology. Ofioliti of seawater circulation in a 500-meter section of ocean crust gabbros. Geochimica
21(2): 105–116. et Cosmochimica Acta 63(23/24): 4059–4080.
The Lower Oceanic Crust 537

Hekinian R, Bideau D, Francheteau J, et al. (1993) Petrology of the East Pacific Rise Karson JA, Klein EM, Hurst SD, et al. (2002) Structure of uppermost fast-spread oceanic
Crust and Upper Mantle Exposed in Hess Deep (East Equatorial Pacific). Journal of crust exposed at the Hess Deep Rift: Implications for subaxial processes at the East
Geophysical Research 98(B5): 8069–8094. Pacific Rise. Geochemistry, Geophysics, Geosystems 3: 1002.
Hekinian R, Hebert R, Maury RC, and Berger ET (1985) Orthopyroxene-bearing Kelemen PB and Aharanov E (1998) Periodic formation of magma fractures and
gabbroic xenoliths in basalts from the East Pacific Rise axis near 12 500 N. Bulletin generation of layered gabbros in the lower crust beneath oceanic spreading ridges.
de Mineralogie 108: 691–698. In: Buck WR, Delaney PT, Karson JA, and Lagrabrielle Y (eds.) Faulting and
Henstock TJ (2002) Compaction control of melt distribution at fast-spreading Magmatism at Mid-Ocean Ridges, Geophysical Monograph Series, vol. 106,
mid-ocean ridges. Geophysical Research Letters 29: 1137. pp. 267–289. Washington: American Geophysical Union.
Henstock TJ, Woods AW, and White RS (1993) The accretion of oceanic crust by Kelemen PB, Kikawa E, Miller DJ, et al. (2004) Proceedings of Ocean Drilling Program,
episodic sill intrusion. Journal of Geophysical Research 98(B3): 4143–4161. Initial Reports, vol. 209, College Station, TX: Ocean Drilling Program.
Hertogen J, Emmermann R, Robinson PT, and Erzinger J (2002) Lithology, Kelemen PB, Koga K, and Shimizu N (1997) Geochemistry of gabbro sills in the
mineralogy, and geochemistry of the lower ocean crust, ODP Hole 735B, Southwest crust-mantle transition zone of the Oman ophiolite: Implications for the origin of the
Indian Ridge. In: Natland JH, Dick HJB, Miller DJ, and Von Herzen RP (eds.) oceanic lower crust. Earth and Planetary Science Letters 146: 475–488.
Proceedings of the Ocean Drilling Program, Scientific Results, vol. 176, pp. 1–82. Kelemen PB, Shimizu N, and Salters VJM (1995) Extraction of mid-ocean-ridge basalt
College Station, TX: Ocean Drilling Program. from the upwelling mantle by focused flow of melt in dunite channels. Nature
Herzberg C (2004) Partial crystallization of mid-ocean ridge basalts in the crust and 375: 747–753.
mantle. Journal of Petrology 45(12): 2389–2405. Kelley DS and Malpas J (1996) Melt-fluid evolution in gabbroic rocks from Hess Deep.
Hess PC (1992) Phase equilibria constraints on the origin of ocean floor basalts. In: Mével C, Gillis KM, Allan JF, and Meyer PS (eds.) Proceedings of the Ocean
In: Phipps Morgan J, Blackman DK, and Sinton JM (eds.) Mantle Flow and Melt Drilling Program, Scientific Results, vol. 147, pp. 213–226. College Station, TX:
Generation at Mid-Ocean Ridges. Geophysical Monograph Series, vol. 71, Ocean Drilling Program.
pp. 67–102. Washington, DC: American Geophysical Union. Kempton PD, Hawkesworth CJ, and Fowler M (1991) Geochemistry and isotopic
Hodges FN (1978) Petrology and chemistry of basalts from DSDP Leg 46. In: Dmitriev L composition of gabbros from Layer 3 of the Indian Ocean crust, Hole 735B. In: Von
and Heirtzle J (eds.) Initial Reports of the Deep Sea Drilling Project, vol. 46, Herzen PR and Robinson PT (eds.) Proceedings of the Ocean Drilling Program,
pp. 227–233. Washington, DC: US Government Printing Office. Scientific Results, vol. 118, pp. 127–143. College Station, TX: Ocean Drilling
Holm PM (2002a) Sr, Nd and Pb isotopic composition of in situ lower crust at the Program.
Southwest Indian Ridge: Results from ODP Leg 176. Chemical Geology Kempton PD and Hunter AG (1997) A Sr-, Nd-, Pb-, O-isotope study of plutonic rocks
184: 195–216. from MARK, Leg 153: Implications for mantle heterogeneity and magma chamber
Holm PM (2002b) Data report: On the composition of the lower ocean crust - major processes. In: Karson JA, Cannat M, Miller DJ, and Elthon D (eds.) Proceedings of
and trace element analyses of gabbroic rocks from Hole 735B, 500-1500 mbsf. the Ocean Drilling Program, Scientific Results, vol. 153, pp. 305–320. College
In: Natland JH, Dick HJB, Miller DJ, and Von Herzen RP (eds.) Proceedings of the Station, TX: Ocean Drilling Program.
Ocean Drilling Program, Scientific Results, vol. 176, pp. 1–13. College Station, TX: Kennedy LA, Kronenberg AK, and Dick HBJ (1996) Structural history and significance of
Ocean Drilling Program. gabbroic rocks in the uppermost mantle: Hess Deep, EPR (Site 895). In: Mével C,
Hooft EE and Detrick RS (1993) The role of density in the accumulation of basaltic melts Gillis KM, Allan J, and Meyer P (eds.) Proceedings of the Ocean Drilling Program,
at mid-ocean ridges. Geophysical Research Letters 20(6): 423–426. Scientific Results, vol. 147, pp. 357–370. College Station, TX: Ocean Drilling
Hooft EEE, Detrick RS, and Kent GM (1997) Seismic structure and indicators of magma Program.
budget along the Southern east Pacific Rise. Journal of Geophysical Research Kent GM, Harding AJ, and Orcutt JA (1990) Evidence for a smaller magma chamber
102(B12): 27319–37340. beneath the East Pacific Rise at 9 300 N. Nature 344: 650.
Hooft E, Schouten H, and Detrick RS (1996) Constraining crustal emplacement Kent GM, Harding AJ, and Orcutt JA (1993a) Distribution of magma beneath the East
processes from the variation in seismic layer 2A thickness at the East Pacific Rise. Pacific Rise between the Clipperton Transform and the 9 170 N Deval from forward
Earth and Planetary Science Letters 142: 289–309. modeling of common depth point data. Journal of Geophysical Research 98(B8):
Hort M, Marsh BD, Resmini RG, and Smith MK (1999) Convection and crystallization in 13945–13969.
a liquid cooled from above: An experimental and theoretical study. Journal of Kent GM, Harding AJ, and Orcutt JA (1993b) Distribution of magma beneath the East
Petrology 40(8): 1271–1300. Pacific Rise near the 9 030 N overlapping spreading center from forward modeling of
Huang PY and Solomon SC (1988) Centroid depths of mid-ocean ridge common depth point data. Journal of Geophysical Research 98(B8): 13971–13995.
earthquakes: Dependence on spreading rate. Journal of Geophysical Research Kent GM, Singh SC, Harding AJ, et al. (2000) Evidence from three-dimensional seismic
93: 13445–13477. reflectivity images for enhanced melt supply beneath mid-ocean-ridge
Humler E and Whitechurch H (1988) Petrology of basalts from the Central Indian Ridge discontinuities. Nature 406: 614–618.
(lat. 25 23’s, long. 70 040 E): Estimates of frequencies and fractional volumes of Kinoshita J and Dick HJB (2001) Yokosuka Shipboard Scientific Party, MODE ’98 Leg 4
magma injections in a two-layered reservoir. Earth and Planetary Science Letters Cruise Report. Yokosuka: Japan Marine Science and Technology Center.
88: 169–181. Klein EM (2003) Geochemistry of the igneous oceanic crust. In: Rudnick RL (ed.)
Hunter RH (1987) Textural equilibrium in layered igneous rocks. In: Parsons I Treatise on Geochemistry, Vol. 3: The Crust pp. 433–463. Oxford: Elsevier-
(ed.) Origins of Igneous Layering, NATO ASI Series, pp. 473–503. Dordrecht: Pergamon.
D. Reidel. Klein EM and Langmuir CH (1987) Global correlation of ocean ridge basalt chemistry
Hunter RH (1996) Texture development in cumulate rocks. In: Cawthorn RG (ed.) with axial depth and crustal thickness. Journal of Geophysical Research
Layered Intrusions. Developments in Petrology, vol. 15, pp. 77–101. Amsterdam: 92: 8089–8115.
Elsevier. Koepke J, Christie DM, Dziony W, et al. (2008) Petrography of the dike-gabbro
Huppert HE and Sparks RSJ (1980) Restrictions on the compositions of mid-ocean transition at IODP Site 1256 (equatorial Pacific): The evolution of the granoblastic
ridge basalts: A fluid dynamical investigation. Nature 286: 46–48. dikes. Geochemistry, Geophysics, Geosystems 9: Q07O09.
Jambon A, Deruelle B, Dreibus G, and Pineau F (1995) Chlorine and bromine Koepke J, Feig ST, and Snow JE (2005a) Hydrous partial melting within the lower
abundance in MORB: The contrasting behaviour of the Mid-Atlantic Ridge and East oceanic crust. Terra Nova 17(3): 286–291.
Pacific Rise and implications for chlorine geodynamic cycles. Chemical Geology Koepke J, Feig ST, and Snow J (2005b) Late stage magmatic evolution of oceanic
126: 101–117. gabbros as a result of hydrous partial melting: Evidence from the Ocean Drilling
Jaupart C and Tait S (1995) Dynamics of differentiation in magma reservoirs. Journal of Program (ODP) Leg 153 drilling at the Mid-Atlantic Ridge. Geochemistry,
Geophysical Research 100(B9): 615–636. Geophysics, Geosystems 6: Q02001.
John BE, Foster DA, Murphy JM, et al. (2004) Determining the cooling history of in situ Kohler TP and Brey GP (1990) Calcium exchange between olivine and clinopyroxene
lower oceanic crust – Atlantis Bank, SW Indian Ridge. Earth and Planetary Science calibrated as a geothermometer for natural peridotites from 2 to 60 kb with
Letters 222: 145–160. applications. Geochimica et Cosmochimica Acta 54: 2375–2388.
Johnson KTM, Dick HJB, and Shimizu N (1990) Melting in the oceanic upper mantle: Kong LSL, Solomon SC, and Purdy GM (1992) Microearthquake characteristics
An ion-microprobe study of diopsides in abyssal peridotites. Journal of Geophysical of a mid-ocean ridge along-axis high. Journal of Geophysical Research
Research 95(B3): 2661–2678. 97: 1659–1685.
Jousselin D and Nicolas A (2000) The Moho transition zone in the Oman Korenaga J and Kelemen PB (1997) Origin of gabbro sills in the Moho transition zone of
ophiolite-relation with wehrlites in the crust and dunites in the mantle. Marine the Oman ophiolite: Implications for magma transport in the oceanic crust. Journal
Geophysical Researches 21: 229–241. of Geophysical Research 102(B12): 27729–27749.
538 The Lower Oceanic Crust

Korenaga J and Kelemen PB (1998) Melt migration through the oceanic lower crust: (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, vol. 176,
A constraint from melt percolation modeling with finite solid diffusion. Earth and pp. 1–56. College Station, TX: Ocean Drilling Program.
Planetary Science Letters 156: 1–11. Magde LS, Barclay AH, Toomey DR, Detrick RS, and Collins JA (2000) Crustal magma
Kushiro I (1969) The system forsterite–diopside–silica with and without water at high plumbing within a segment of the Mid-Atlantic Ridge, 35 N. Earth and Planetary
pressures. American Journal of Science 267A: 269–294. Science Letters 175: 55–67.
Kushiro I (1975) On the nature of silicate melt and its significance in magma genesis: Mainprice D (1997) Modelling the anisotropic seismic properties of partially molten
Regularities in the shift of the liquidus boundaries involving olivine, pyroxene, and rocks found at mid-ocean ridges. Tectonophysics 279: 161–179.
silica minerals. American Journal of Science 275: 411–431. Manning C, Weston PE, and Mahon KI (1996) Rapid high temperature metamorphism
Kuznir N and Bott MHP (1976) A thermal study of oceanic crust. Geophysical Journal of of the East Pacific Rise gabbros from Hess Deep. Earth and Planetary Science Letters
the Royal Astronomical Society 47: 83–95. 144: 123–132.
Kvassnes AJS and Grove TL (2008) How partial melts of mafic lower crust affect Marsh BD (1981) On the crystallinity, probability of occurrence and rheology of lava and
ascending magmas at oceanic ridges. Contributions to Mineralogy and Petrology magma. Contributions to Mineralogy and Petrology 78: 85–98.
156: 49–71. Marsh BD (1996) Solidification fronts and magmatic evolution. Mineralogical Magazine
Langmuir CH (1989) Geochemical consequences of in situ differentiation. Nature 60: 5–40.
340: 199–205. Mathez EA (1995) Magmatic metasomatism and the formation of the Merensky reef,
Langmuir CH, Bender JF, and Batiza R (1986) Petrological and tectonic segmentation of Bushveld Complex. Contributions to Mineralogy and Petrology 119: 277–286.
the East Pacific Rise, 5 300 -14 300 N. Nature 322: 422–429. McCollom TM and Shock EL (1998) Fluid-rock interactions in the lower oceanic crust:
Langmuir CH, Klein EM, and Plank T (1992) Petrological systematics of mid-ocean Thermodynamic models of hydrothermal alteration. Journal of Geophysical
ridge basalts: Constraints on melt generation beneath ocean ridges. In: Phipps Research 103(B1): 547–575.
Morgan J, Blackman GK, and Sinton JM (eds.) Mantle Flow and Melt Generation at Melson WG, Vallier T, Wright TL, Byerly G, and Nelen J (1976) Chemical diversity of
Mid-Ocean Ridges. Geophysical Monograph Series, vol. 71, pp. 183–280. abyssal volcanic glass erupted along Pacific, Atlantic, and Indian Ocean sea-floor
Washington DC: American Geophysical Union. spreading centers. In: Sutton GH, Manghnani MH, Moberly R, and McAfee EU (eds.)
le Roux PJ, le Roux AP, and Schilling J-G (2002) Crystallization processes beneath the The Geophysics of the Pacific Ocean Basin and Its Margin. Geophysical Monograph
southern Mid-Atlantic Ridge (40-55 S), evidence for high-pressure initiation of Series, vol. 19, pp. 351–367. Washington, DC: American Geophysical Union.
crystallization. Contributions to Mineralogy and Petrology 142: 582–602. Menke W, West M, and Tolstoy M (2002) Shallow-crustal magma chamber beneath the
Lecuyer C and Gruau G (1996) Oxygen and strontium isotope compositions of Hess axial high of the Coaxial segment of the Juan de Fuca Ridge at the source site of the
Deep gabbros (Hole 894F and 894G): High temperature interaction of seawater with 1993 eruption. Geology 30(4): 359–362.
oceanic crust layer 3. In: Mével C, Gillis KM, Allan JF, and Meyer PS (eds.) Meurer BP and Gee J (2002) Evidence for protracted construction of slow-spread
Proceedings of the Ocean Drilling Program, Scientific Results, vol. 147, oceanic crust by small magmatic injections. Earth and Planetary Science Letters
pp. 227–234. College Station, TX: Ocean Drilling Program. 201: 45–55.
Lissenberg CJ and Dick HJB (2008) Melt-rock reaction in the lower oceanic crust and Meyer PS, Dick HJB, and Thompson G (1989) Cumulate gabbros from the Southwest
its implications for the genesis of mid-ocean ridge basalt. Earth and Planetary Indian Ridge, 54 S-7 160 E: Implications for magmatic processes at a slow
Science Letters 271: 311–325. spreading ridge. Contributions to Mineralogy and Petrology 103: 44–63.
Lissenberg CJ, Rioux M, Shimizu N, Bowring SA, and Mével C (2009) Zircon dating of Meyer PS and Shibata T (1990) Complexzoning in plagioclase feldspar from ODP site
oceanic crustal accretion. Science 1048–1050. 648. In: Detrick R, Honnorez J, Bryan WB, and Juteau T (eds.) Proceedings of the
Lister CRB (1974) On the penetration of water into hot rock. Geophysical Journal of the Ocean Drilling Program, Scientific Results, vol. 106/109, pp. 123–142. College
Royal Astronomical Society 39: 465–509. Station, TX: Ocean Drilling Program.
Longhi J (1982) Effects of fractional crystallisation and cumulus processes on mineral Michael PJ and Chase RL (1987) The influence of primary magma composition, H2O
compositional trends of some lunar and terrestrial rock series. Journal of and pressure on mid-ocean ridge basalt differentiation. Contributions to Mineralogy
Geophysical Research 87(S01): A54–A64. and Petrology 96: 245–263.
Longhi J (1991) Comparative liquidus equilibria of hypersthene-normative basalts at Michael PJ and Cornell WC (1998) Influence of spreading rate and magma supply on
low-pressure. American Mineralogist 76: 785–800. crystallization and assimilation beneath mid-ocean ridges: Evidence from chlorine
Lonsdale P (1988) Structural pattern of the Galapagos microplate and evolution of the and major element chemistry of mid-ocean ridge basalts. Journal of Geophysical
Galapagos Triple Junctions. Journal of Geophysical Research 93(B11): Research 103(B8): 18325–18356.
13551–13574. Michael PJ, McDonough WF, Nielsen RL, and Cornell WC (2002) Depleted melt
Lundstrom CC (2000) Rapid diffusion infiltration of sodium into partially molten inclusions in MORB plagioclase: Messages from the mantle or mirages from the
peridotite. Nature 403: 527–530. magma chamber? Chemical Geology 183: 43–61.
Macdonald KC, Fox PJ, Perram LJ, et al. (1988) A new view of the mid-ocean ridge from Michael PJ and Schilling J-G (1989) Chlorine in mid-ocean ridge magmas: Evidence
the behaviour of ridge-axis discontinuities. Nature 335: 217–225. for assimilation of seawater-influenced components. Geochimica et Cosmochimica
MacGregor LM, Constable SC, and Sinha MC (1998) The RAMESSES experiment – III. Acta 53: 3131–3143.
Controlled-source electromagnetic sounding of the Reykjanes Ridge at 57 450 N. Miyashiro A, Shido F, and Ewing M (1970) Crystallization and differentiation in abyssal
Geophysical Journal International 135: 773–789. tholeiites and gabbros from mid-oceanic ridges. Earth and Planetary Science Letters
Machetel P and Garrido CJ (2009) A thermomechanical numerical model for crustal 7: 361–365.
accretion of medium to fast spreading mid-ocean ridges. Geochemistry, Morton AC and Sleep NH (1985) A mid-ocean ridge thermal model: Constraints on the
Geophysics, Geosystems 10: Q03008. volume of axial hydrothermal heat flux. Journal of Geophysical Research
Maclennan J, Hulme T, and Singh SC (2004) Thermal models of oceanic crustal 90: 11345–11353.
accretion: Linking geophysical, geological and petrological observations. Muller RD, Sdrolias M, Gaina C, and Roest WR (2008) Age, spreading rates, and
Geochemistry, Geophysics, Geosystems 5(2): Q02F25. spreading asymmetry of the world’s ocean crust. Geochemistry, Geophysics,
Maclennan J, Hulme T, and Singh SC (2005) Thermal models of oceanic crustal Geosystems 9: Q04006.
accretion: Linking geophysical, geological and cooling the lower oceanic crust. Murray MH, Kong L, Forsyth DW, Solomon SC, and Hussang DM (1984) An OBS
Geology 33: 357–360. survey of microearthquakes in the median valley on the Mid-Atlantic Ridge near
MacLeod CJ, Boudier F, Yaouancq G, and Richter C (1996) Gabbro Fabrics from Site 35N. Eos, Transactions American Geophysical Union 65: 1009.
894, Hess Deep: Implications for magma chamber processes at the East Pacific Rise. Mutter JC, Carbotte SM, Su W, et al. (1995) Seismic images of active magma
In: Mével C, Gillis KM, Allan JF, and Meyer PS (eds.) Proceedings of the Ocean systems beneath the East Pacific Rise between 17 050 and 17 350 S. Science
Drilling Program, Scientific Results, vol. 147, pp. 317–328. College Station, TX: 268: 391–395.
Ocean Drilling Program. Mutter JC and Karson JA (1992) Structural processes at slow-spreading ridges. Science
MacLeod CJ and Yaouancq G (2000) A fossil melt lens in the Oman ophiolite: 257: 627–634.
Implications for magma chamber processes at fast spreading ridges. Earth and Nableck PI and Langmuir CH (1986) The significance of unusual zoning in olivines
Planetary Science Letters 176(3–4): 357–373. from FAMOUS area basalt 527-1-1. Contributions to Mineralogy and Petrology
Maeda J, Naslund HR, Yang YD, Kikawa E, Tajima T, and Blackburn WH (2002) 93: 1–8.
High-temperature fluid migration within oceanic layer 3 gabbros, Hole 735B, Naslund HR and McBirney AR (1996) Mechanisms of formation of igneous layering.
Southwest Indian Ridge: Implications for the magmatic-hydrothermal transition at In: Cawthorn RG (ed.) Layered Intrusions. Developments in Petrology, vol. 15,
slow-spreading ridges. In: Natland JH, Dick HJB, Miller DJ, and Von Herzen RP pp. 1–43. Amsterdam: Elsevier.
The Lower Oceanic Crust 539

Natland JH and Dick HJB (1996) Melt migration through high-level gabbroic cumulates Pan Y and Batiza R (2003) Magmatic processes under mid-ocean ridges: A detailed
of the East Pacific Rise at the Hess Deep: The origin of magma lenses and the deep mineralogical study of lavas from the East Pacific Rise 9 300 N, 10 300 N and
crustal structure of fast-spreading ridges. In: Mével C, Gillis KM, Allan JF, and 11 200 N. Geochemistry, Geophysics, Geosystems 4(11): 8623.
Meyer PS (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, Pedersen RB, Malpas J, and Falloon T (1996) Petrology and geochemistry of gabbroic
vol. 147, pp. 21–58. College Station, TX: Ocean Drilling Program. and related rocks from Site 894, Hess Deep. In: Mével C, Gillis KM, Allan JF, and
Natland JH and Dick HJB (2002) Stratigraphy and composition of gabbros drilled in Meyer PS (eds.) Proceedings of the Ocean Drilling Program, Scientific Results,
Ocean Drilling Program Hole 735B, Southwest Indian Ridge: A synthesis of vol. 147, pp. 3–19. College Station, TX: Ocean Drilling Program.
geochemical data. In: Natland JH, Dick HJB, Miller DJ, and Von Herzen RP (eds.) Perfit MR, Ridley WI, and Jonasson IR (1999) Geologic, petrologic, and geochemical
Proceedings of the Ocean Drilling Program, Scientific Results, vol. 176, pp. 1–69. relationships between magmatism and massive sulfide mineralization along
College Station, TX: Ocean Drilling Program. the Eastern Galapagos Spreading Center. Reviews in Economic Geology 8: 75–100.
Natland JH, Meyer PS, Dick HJB, and Bloomer SH (1991) Magmatic oxides and sulfides Perk N, Coogan LA, Karson JA, Klein EM, and Hanna H (2007) Primitive cumulates
in gabbroic rocks from Hole 735B and the later development of the liquid line of from the upper crust formed at the East Pacific Rise. Contributions to Mineralogy
descent. In: Von Herzen RP and Robinson PT (eds.) Proceedings of the Ocean and Petrology 154(5): 575–590.
Drilling Program, Scientific Results, vol. 118, pp. 75–111. College Station, TX: Pettigrew TL, Casey JF, and Miller DJ (1999) Proceedings of the Ocean Drilling
Ocean Drilling Program. Program, Initial Reports, vol. 179. College Station, TX: Ocean Drilling Program.
Navin DA, Peirce C, and Sinha MC (1998) The RAMESSES experiment – II. Evidence for Phipps Morgan J and Chen YJ (1993a) The Genesis of oceanic crust – Magma
accumulated melt beneath a slow spreading ridge from wide-angle refraction and injection, hydrothermal cooling, and crustal flow. Journal of Geophysical Research
multichannel reflection seismic profiles. Geophysical Journal International 98(B4): 6283–6297.
135: 746–772. Phipps Morgan J and Chen YJ (1993b) Dependence of ridge-axis morphology on
Nedimovic MR, Carbotte SM, Harding AJ, et al. (2005) Frozen magma lenses below the magma supply and spreading rate. Nature 364: 706–708.
oceanic crust. Nature 436: 1149–1152. Phipps Morgan J and Chen YJ (1994) Reply. Journal of Geophysical Research 99(B6):
Neo N, Yamazaki S, and Miyashita S (2009) Data report: Bulk rock compositions of 12031–12032.
samples from the IODP Expedition 309/312 sample pool, ODP Hole 1256D. Phipps Morgan J, Harding A, Orcutt J, Kent G, and Chen YJ (1994) An observational
In: Teagle DAH, Alt JC, and Umino S, et al. (eds.) Proceedings of IODP, 309/312. and theoretical synthesis of magma chamber geometry and crustal genesis along a
Washington, DC: Integrated Ocean Drilling Program Management International, Inc. mid-ocean ridge. In: Ryan MP (ed.) Magmatic Systems, pp. 139–178. San Diego,
Nicolas A (1992) Kinematics in magmatic rocks with special reference to gabbros. CA: Academic Press Inc.
Journal of Petrology 33(4): 891–915. Presnall DC and Hoover JD (1987) High pressure phase equilibrium constraints on the
Nicolas A, Boudier F, and France L (2009) Subsidence in magma chamber and the origin of mid-ocean ridge basalt. In: Mysen BO (ed.) Magmatic Processes:
development of magmatic foliation in Oman ophiolite gabbros. Earth and Planetary Physicochemical Principles, Geochemical Society Special Publication No. 1,
Science Letters 284: 76–87. pp. 75–90. University Park, Pa: The Geochemical Society.
Nicolas A, Freydier C, Godard M, and Vauchez A (1993) Magma chambers at mid-ocean Purdy GM, Kong LSL, Christeson GL, and Solomon SC (1992) Relationship between
ridges: How large? Geology 21: 53–56. spreading rate and the seismic structure of mid-ocean ridges. Nature 355: 815–817.
Nicolas A and Ildefonse B (1996) Flow mechanism and viscosity in basaltic magma Python M and Ceuleneer G (2003) Nature and distribution of dykes and related melt
chambers. Geophysical Research Letters 23(16): 2013–2016. migration structures in the mantle section of the Oman ophiolite. Geochemistry,
Nicolas A, Reuber I, and Benn K (1988) A new magma chamber model Geophysics, Geosystems 4: 8612.
based on structural studies in the Oman Ophiolite. Tectonophysics 151(1–4): Quick JE and Denlinger RP (1993) Ductile deformation and the origin of layered gabbro
87–105. in ophiolites. Journal of Geophysical Research 98(B8): 14015–14027.
Nielsen RL, Crum J, Bourgeois R, et al. (1995) Melt inclusions in high-An plagioclase Regelous M, Niu Y, Wendt JI, Batiza R, Greig A, and Collerson KD (1999) Variations in
from the Gorda Ridge: An example of the local diversity of MORB parental magmas. the geochemistry of magmatism on the East Pacific Rise at 10 300 N since 800 ka.
Contributions to Mineralogy and Petrology 122: 34–50. Earth and Planetary Science Letters 168: 45–63.
Nisbet EG and Fowler CMR (1978) The Mid-Atlantic Ridge at 37 and 45 N: Some Reid I and Jackson HR (1981) Oceanic spreading rate and crustal thickness. Marine
geophysical and petrological constraints. Geophysical Journal of the Royal Geophysical Research 5: 165–172.
Astronomical Society 54: 631–660. Reuber I (1990) Diapiric magma intrusions in the plutonic sequence of the Oman
Niu Y, Gilmore T, MacKie S, Greig A, and Bach W (2002) Mineral chemistry, whole-rock ophiolite traced by the geometry and flow patterns of the cumulates. In: Symposium
compositions and petrogenesis of Leg 176 gabbros: Data and discussion. on Diapirism with Special Reference to Iran, Teheran, pp. 315–338.
In: Natland JH, Dick HJB, Miller DJ, and Von Herzen RP (eds.) Proceedings of the Reynolds JR and Langmuir CH (1997) Petrological systematics of the Mid-Atlantic
Ocean Drilling Program, Scientific Results, vol. 176, pp. 1–60. Ridge south of Kane: Implications for ocean crust formation. Journal of Geophysical
O’Hara MJ (1965) Primary magmas and the origin of basalts. Scottish Journal of Research 102(B7): 14915–14946.
Geology 1(1): 19–40. Rhodes JM, Dungan MA, Blanchard DP, and Long PE (1979) Magma mixing at
O’Hara MJ (1968a) Are ocean floor basalts primary magma? Nature 220: 683–685. mid-ocean ridges: Evidence from basalts drilled near 22 N on the Mid-Atlantic
O’Hara MJ (1968b) The bearing of phase equilibria studies in synthetic and natural Ridge. Tectonophysics 55: 35–61.
systems on the origin and evolution of basic and ultrabasic rocks. Earth-Science Ridley I, Perfit MR, Smith MC, and Fornari DJ (2006) Magmatic processes in
Reviews 4: 69–133. developing oceanic crust revealed in a cumulate xenolith collected at the East Pacific
O’Hara MJ (1977) Geochemical evolution during fractional crystallization of a Rise, 9 500 N. Geochemistry, Geophysics, Geosystems 7(12): Q12O04.
periodically refilled magma chamber. Nature 266: 503–507. Robinson PT and Von Herzen R (1989) Proceedings of the Ocean Drilling Project, Initial
O’Hara MJ and Herzberg C (2002) Field relations, petrology, major element data and Reports, vol. 118. College Station, TX: Ocean Drilling Program.
phase equilibria in basalt petrogenesis. Geochimica et Cosmochimica Acta Robinson CJ, White RS, Bickle MJ, and Minshull TA (1996) Restricted melting under
66: 2167–2191. the very slow-spreading Southwest Indian Ridge. In: MacLeod CJ, Tyler PA, and
Onuma K and Tohara T (1983) Effect of chromium on phase relations in the join Walker CL (eds.) Tectonic, Magmatic and Hydrothermal Segmentation of
forsterite-anorthite-diopside in air at 1 atm. Contributions to Mineralogy and Mid-Ocean Ridges, Geological Society Special Publication, vol. 118, pp. 131–141.
Petrology 84: 174–181. London: Geological Society of London.
Ozawa K (1984) Olivine-spinel geospeedometry: Analysis of diffusion-controlled Roeder PL and Emslie RF (1970) Olivine-liquid equilibria. Contributions to
Mg-Fe2þ exchange. Geochimica et Cosmochimica Acta 48: 2597–2611. Mineralalogy and Petrology 29: 275–289.
Ozawa K, Meyer PS, and Bloomer SH (1991) Mineralogy and textures of iron-titanium Ross K and Elthon D (1993) Cumulates from strongly depleted mid-ocean-ridge basalt.
oxide gabbros and associated olivine gabbros from Hole 735B. In: Von Herzen RP Nature 365: 826–828.
and Robinson PT (eds.) Proceedings of the Ocean Drilling Program, Scientific Ross K and Elthon D (1997) Cumulus and postcumulus crystallisation in the oceanic
Results, vol. 118, pp. 41–74. College Station, TX: Ocean Drilling Program. crust: Major and trace-element geochemistry of Leg 153 Gabbroic rocks.
Pallister JS and Hopson CA (1981) Samail Ophiolite plutonic suite: Field relations, In: Karson JA, Cannat M, and Miller DJ (eds.) Proceedings of Ocean Drilling
phase variation, cryptic variation and layering, and a model of a spreading ridge Program, Scientific Results, vol. 153, pp. 333–353. College Station, TX: Ocean
magma chamber. Journal of Geophysical Research 86(B4): 2593–2644. Drilling Program.
Pan Y and Batiza R (2002) Mid-ocean ridge magma chamber processes: Constraints Rubin KH and Sinton JM (2007) Inferences on mid-ocean ridge thermal and magmatic
from olivine zonation in lavas from the East Pacific Rise at 9 300 N and 10 300 N. structure from MORB compositions. Earth and Planetary Science Letters
Journal of Geophysical Research 107(B1): ECV9-1–ECV 9-13. 260: 257–276.
540 The Lower Oceanic Crust

Ryan MP (1993) Neutral buoyancy and the structure of mid-ocean ridge magma Sun S-s and McDonough WF (1989) Chemical and isotopic systematics of oceanic
reservoirs. Journal of Geophysical Research 98(B12): 22321–22338. basalts: Implications for mantle composition and processes. In: Saunders AD and
Sato H, Aoki K-i, Okamoto K, and Fujita B (1978) Petrology and chemistry of basaltic Norry MJ (eds.) Magmatism in the Ocean Basins, Geological Society Special
rocks from Hole 396B, IPOD/DSDP Leg 46. In: Dmitriev L and Heirtzler J (eds.) Publication 42, pp. 313–345. London: Geological Society of London.
Initial Reports of the Deep Sea Drilling Project, vol. 46, pp. 115–142. Washington, Tait SR, Huppert HE, and Sparks RSJ (1984) The role of compositional convection in
DC: US Government Printing Office. the formation of adcumulate rocks. Lithos 17: 139–146.
Schilling J-G, Kingsley RH, and Devine JD (1982) Galapagos hot spot spreading center Tartarotti P, Cannat M, and Mével C (1995) Gabbroic dikelets in serpentinised
system I. Spatial petrological and geochemical variations (83oW-101oW). Journal of peridotites from the Mid-Atlantic Ridge at 23 200 N. In: Vissers RLM and Nicolas A
Geophysical Research 87(B7): 5593–5610. (eds.) Mantle and Lower Crustal Exposures in Ocean Ridges and in Ophiolites,
Schwartz JJ, John BE, Cheadle MJ, Reiners PW, and Baines AG (2009) Cooling history pp. 35–69. Dordrecht: Kluwer.
of Atlantis Bank oceanic core complex: Evidence for hydrothermal activity 2.6 Ma off Tiezzi LJ and Scott RB (1980) Crystal fractionation in a cumulate gabbro,
axis. Geochemistry, Geophysics, Geosystems 10(8): Q08020. Mid-Atlantic-Ridge 26 N. Journal of Geophysical Research 85: 5438–5454.
Schwartz JJ, John BE, Cheadle MJ, et al. (2005) Dating the growth of oceanic crust at a Tolstoy M, Harrding AJ, and Orcutt JA (1997) Deepening of the axial magma chamber
slow-spreading ridge. Science 310: 654–657. on the southern East Pacific Rise towards the Garrett Fracture Zone. Journal of
Searle M and Cox J (1999) Tectonic setting, origin, and obduction of the Oman Geophysical Research 102(B2): 3097–3108.
ophiolite. Geological Society of America Bulletin 111(1): 104–122. Tolstoy M, Waldhauser F, Bohnenstiehl DR, Weekly RT, and Kim WY (2008) Seismic
Seher T, Crawford WC, Singh SC, Cannat M, Combier V, and Dusunur D (2010) identification of along-axis hydrothermal flow on the East Pacific Rise. Nature
Crustal velocity structure of the Lucky segment of the Mid-Atlantic Ridge at 37 N 451: 181–185.
from seismic refraction measurements. Journal of Geophysical Research Toomey DR, Solomon SC, and Purdy GM (1988) Microearthquakes beneath median
115: B03103. valley of the Mid-Atlantic ridge near 23 N: Tomography and tectonics. Journal of
Shah AK and Buck WR (2001) Causes for axial high topography at mid-ocean ridges Geophysical Research 93(B8): 9093–9112.
and the role of crustal thermal structure. Journal of Geophysical Research 106(B12): Toomey DR, Solomon SC, Purdy GM, and Murray MH (1985) Microearthquakes
30865–30879. beneath the median valley of the Mid-Atlantic ridge near 23 N: Hypocenters and
Singh SC, Crawford WC, Carton H, et al. (2006a) Discovery of a magma chamber and focal mechanisms. Journal of Geophysical Research 90(B7): 5443–5458.
faults beneath a Mid-Atlantic Ridge hydrothermal field. Nature 442: 1029–1032. Van Ark E, Detrick RS, Canales JP, et al. (2007) Seismic structure of the Endeavour
Singh SC, Harding AJ, Kent GM, et al. (2006b) Seismic reflection images of the Moho Segment, Juan de Fuca Ridge: Correlations with seismicity and hydrothermal
underlying melt sills at the East Pacific Rise. Nature 442: 287–290. activity. Journal of Geophysical Research 112: B02401.
Singh SC, Kent GM, Collier JS, Harding AJ, and Orcutt JA (1998) Melt to mush Van Orman JA, Grove TL, and Shimizu N (2001) Rare earth element diffusion in
variations in crustal magma chamber properties along the ridge crest at the southern diopside: Influence of temperature, pressure, and ionic radius and an elastic model
East Pacific Rise. Nature 394: 874–878. for diffusion in silicates. Contributions to Mineralogy and Petrology 141: 687–703.
Sinha MC, Navin DA, MacGregor LM, et al. (1997) Evidence for accumulated melt Van Orman JA, Saal AE, Bourdon B, and Hauri EH (2006) Diffusive fractionation of
beneath the slow-spreading Mid-Atlantic Ridge. Philosophical Transactions of the U-series radionuclides during mantle melting and shallow-level melt–cumulate
Royal Society. Series A 355: 233–253. interaction. Geochimica et Cosmochimica Acta 70: 4797–4812.
Sinton JM and Detrick RS (1992) Mid-ocean ridge magma chambers. Journal of Vera EE, Mutter JC, Buhl P, et al. (1990) The structure of 0- to 0.2-m.y. old oceanic
Geophysical Research 97(B1): 197–216. crust at 9 N on the East Pacific Rise from expanded spread profiles. Journal of
Sleep NH (1975) Formation of oceanic crust: Some thermal constraints. Journal of Geophysical Research 95: 15529–15556.
Geophysical Research 80: 4037–4042. Villiger S, Muntener O, and Ulmer P (2007) Crystallization pressures of mid-ocean
Sleep NH and Barth GA (1997) The nature of the lower crust and shallow mantle ridge from major element variations of glasses from and fractional crystallization
emplaced at low spreading rates. Tectonophysics 279: 181–191. experiments. Journal of Geophysical Research 112: B01202.
Smewing JD (1981) Mixing characteristics and compositional differences in Watson EB, Wark DA, and Thomas JB (2006) Crystallisation thermometers for zircon
mantle-derived melts beneath spreading axes: Evidence from cyclically layered rocks and rutile. Contributions to Mineralogy and Petrology 151: 413–433.
in the ophiolite of north Oman. Journal of Geophysical Research 86(B4): 2645–2659. Webb SC (2008) Is there a deep hydrothermal circulation at the EPR? EOS, Transactions
Smith WHF and Sandwell DT (1997) Global sea floor topography from satellite altimetry American Geophysical Union. Abstract B21A-0327. Fall Meeting Supplement.
and ship depth soundings. Science 277: 1956–1962. Werner CD (1997) Data report: Geochemistry of rocks and minerals of the gabbro
Snow JE (1993) The Isotope Geochemistry of Abyssal Peridotites and Related Rocks. complex from the MARK area. In: Karson JA, Cannat M, Miller DJ, and Elthon D
PhD Thesis, Woods Hole Oceanographic Institute/MIT Joint Program. (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, vol. 153,
Snyder D, Gier E, and Carmichael I (1994) Experimental determination of the thermal pp. 491–504. College Station, TX: Ocean Drilling Program.
conductivity of molten CaMgSi2O6 and the transport of heat through magmas. West M, Menke W, Tolstoy M, Webb S, and Sohn R (2001) Magma storage beneath
Journal of Geophysical Research 99(B8): 15,503–15,516. Axial volcano on the Juan de Fuca mid-ocean ridge. Nature 413: 833–835.
Sobolev AV and Shimizu N (1993) Ultra-depleted primary melt inclusions in an olivine White RS, McKenzie D, and O’Nions RK (1992) Oceanic crustal thickness from seismic
from the Mid-Atlantic Ridge. Nature 363: 151–154. measurements and rare earth element inversions. Journal of Geophysical Research
Sohn RA, Fornari DJ, Von Damm KL, Hildebrand JA, and Webb SC (1998) Seismic and 97(B13): 19683–19715.
hydrothermal evidence for a cracking event on the East Pacific Rise crest at 9 500 N. Wilcock WSD, Archer SD, and Purdy GM (2002) Microearthquakes on the Endeavour
Nature 396: 159–161. segment of the Juan de Fuca Ridge. Journal of Geophysical Research 107(B12):
Sparks RSJ, Meyer P, and Sigurdsson H (1980) Density variations amongst mid-ocean 2336.
ridge basalts: Implications for magma mixing and the scarcity of primitive lavas. Wilcock WSD and Delaney JR (1996) Mid-ocean ridge sulfide deposits: Evidence for
Earth and Planetary Science Letters 46: 419–430. heat extraction from magma chambers or cracking fronts? Earth and Planetary
Stakes D, Mével C, Cannat M, and Chaput T (1991) Metamorphic stratigraphy of Hole Science Letters 145: 49–64.
735B. In: Von Herzen RP and Robinson PT (eds.) Proceedings of the Ocean Drilling Wilcock WSD, Hooft EEE, Toomey DR, et al. (2009) The role of magma injection in
Program, Scientific Results, vol. 118, pp. 153–180. College Station, TX: Ocean localizing black-smoker activity. Nature Geoscience 2: 509–513.
Drilling Program. Wilson DS, Clague DA, Sleep NH, and Morton JL (1988) Implications of magma
Stewart MA, Klein EM, and Karson JA (2002) Geochemistry of dikes and lavas from the convection for the size and temperature of magma chambers at fast spreading
north wall of the Hess Deep Rift: Insights into the four-dimensional character of ridges. Journal of Geophysical Research 93: 11974–11984.
crustal construction at fast spreading mid-ocean ridges. Journal of Geophysical Wilson DS, Teagle DAH, Alt JC, et al. (2006) Drilling into gabbro in intact ocean crust.
Research 107(B10): 2238. Science 312: 1016–1020.
Stolper E (1980) A phase diagram for mid-ocean ridge basalts: Preliminary results and Wood BJ and Blundy J (2003) Trace element partitioning under crustal and uppermost
implications for petrogenesis. Contributions to Mineralogy and Petrology mantle conditions: The influences of ionic radius, cation charge, pressure, and
74: 13–27. temperature. In: Carlson RW (ed.) Treatise on Geochemistry, Vol. 2: The Mantle and
Stolper E and Walker D (1980) Melt density and the average composition of basalt. Core, pp. 395–424. Oxford: Elsevier Pergamon.
Contributions to Mineralogy and Petrology 74: 7–12. Wolfe CJ, Purdy GM, Toomey DR, and Solomon SC (1995) Microearthquake
Suhr G, Hellebrand E, Johnson K, and Brunelli E (2008) Stacked gabbro units and characteristics and crustal velocity structure at 29 N on the Mid-Atlantic Ridge:
intervening mantle: A detailed look at a section of IODP Leg 305, Hole U1309D. The architecture of a slow-spreading segment. Journal of Geophysical Research
Geochemistry, Geophysics, Geosystems 9(10): Q10007. 100(B12): 24449–24472.
The Lower Oceanic Crust 541

Worster MG, Huppert HE, and Sparks RSJ (1990) Convection and crystallisation in IODP, 309/312. Washington, DC: Integrated Ocean Drilling Program Management
magma cooled from above. Earth and Planetary Science Letters 101: 78–89. International, Inc.
Yamazaki S, Neo N, and Miyashita S (2009) Data report: Whole-rock major and trace Yang H-J, Kinzler RJ, and Grove TL (1996) Experiments and models of anhydrous,
elements and mineral compositions of the sheeted dike–gabbro transition in ODP basaltic olivine-plagioclase-augite saturated melts from 0.001 to 10 kbar.
Hole 1256D. In: Teagle DAH, Alt JC, and Umino S, et al. (eds.) Proceedings of Contributions to Mineralogy and Petrology 124: 1–18.

You might also like