A Reappraisal of Redox Melting in the Earth`s Mantle as a Function

JOURNAL OF PETROLOGY
VOLUME 52
NUMBERS 7 & 8
PAGES 1363^1391
2011
doi:10.1093/petrology/egq061
A Reappraisal of Redox Melting in the
Earth’s Mantle as a Function of Tectonic
Setting and Time
STEPHEN F. FOLEY*
GEOCYCLES RESEARCH CENTRE AND INSTITUTE OF GEOSCIENCES, UNIVERSITY OF MAINZ, BECHERWEG 21, 55099
MAINZ, GERMANY
RECEIVED FEBRUARY 2, 2010; ACCEPTED SEPTEMBER 14, 2010
ADVANCE ACCESS PUBLICATION NOVEMBER 3, 2010
Redox melting refers to any process by which melt is generated by the
contact of a rock with a fluid or melt with a contrasting oxidation
state. It was originally applied to melting owing to the oxidation of
reduced CH4 - and H2-bearing fluids in contact with more oxidized
blocks in the mantle, particularly recycled crustal blocks.This oxidation mechanism causes an increase in the activity of H2O by the reaction of CH4 with O2, and the increased aH2O causes a rapid drop
in the solidus temperature, and is here termed hydrous redox melting
(HRM). Recently, a second redox melting mechanism (carbonate
redox melting; CRM) has been discovered that operates in more oxidized conditions, and may post-date the first mechanism in the same
geographical area, explaining the sequence of igneous rock types
from lamproites to ultramafic lamprophyres that occurs during the
development of rifts through cratons. The CRM mechanism relies
on the oxidation of solid carbon as graphite or diamond that has
accumulated in the lithosphere over time. The solidus temperature
for rocks with both CO2 and H2O is lower than in conditions with
H2O alone; it does not occur at depths less than 65 km, but has
recently been confirmed experimentally to depths of at least 200 km.
Melts produced by HRM are not SiO2-undersaturated, even at
depths of 200 km, and may often resemble lamproites or SiO2-rich
picrites, whereas melts produced by CRM are always SiO2-undersaturated and range from carbonatitic to ultramafic lamprophyric or
melilititic with increasing degree of melting. The operation of redox
melting may be more common than has been recognized because the
oxidation state of the upper mantle is not uniform as a function of
depth, geodynamic setting or geological time. The general decrease of
oxygen fugacity (fO2) of c. 0·7 log units per 1 GPa pressure increase
dictates that rapidly subducted oceanic lithosphere will be considerably more oxidized than ambient mantle peridotite at depths of
*Corresponding author. Telephone: þ49-6131-392-2845.
Fax: þ49-6131-392-3070. E-mail: [email protected]
200^300 km. Hydrothermal alteration (serpentinization), addition
of continental or carbonate sediments, and dehydration reactions
during subduction all contribute to the heterogeneity of oxidation
states in the subducted slab, which may vary over 6 log units; this
raises the potential for redox reactions on local and regional scales.
The oceanic lithosphere has a lower average fO2 than either continental or cratonic mantle lithosphere at a given depth, so that the
HRM mechanism dominates in recycled blocks and at the base of
the continental lithosphere. The higher thermal gradients dictate
that HRM is more common in the modern Earth beneath ocean islands and in upwelling mantle currents than in subduction zones.
The oxidation state of the mantle is often described as having been
constant since 3·5 Ga, but this overlooks the bias towards continental
samples. Redox melting of oxidized recycled blocks (at approximately
the fayalite^magnetite^quartz buffer) in the mantle was not important in the Hadean and Archaean, as it had to await the gradual
oxidation of the mantle and the establishment of the subduction process, as well as the stabilization of the continents. The lack of
CRM explains the lack of carbonatites before 2·7 Ga. However, the
lower fO2 of the Archaean asthenosphere and higher volatile contents
caused more prevalent HRM in the Hadean and Archaean mantle.
Degassing is controlled by solubility of volatile species in melts,
which are H2O-rich but C-poor in reducing conditions. Silicate
melts under reduced conditions contain much less carbon but more nitrogen than melts in the modern mantle, arguing for a nitrogen-rich,
CO2-poor early atmosphere.
KEY WORDS:
redox melting; craton; recycled crust; Archean; mantle
degassing; fO2
ß The Author 2010. Published by Oxford University Press. All
rights reserved. For Permissions, please e-mail: journals.permissions@
oup.com
JOURNAL OF PETROLOGY
VOLUME 52
I N T RO D U C T I O N
The oxidation state of the Earth’s mantle has remained
controversial for decades. Both the chemistry of volcanic
rocks and their constituent minerals and minerals in
mantle peridotites have been used as oxygen sensors; however, the information they provide often does not match
up. Whereas volcanic rocks are usually found to lie in the
oxygen fugacity (fO2) range of the nickel^nickel oxide
(NNO) or fayalite^magnetite^quartz (FMQ) oxygen buffers (Carmichael & Nicholls, 1967; Haggerty & Tompkins,
1983; Foley, 1985; Carmichael, 1991; Ballhaus, 1993; Canil,
1997; Be¤zos & Humler, 2005; Lee et al., 2005; Mallmann
& O’Neill, 2009), the mantle peridotite samples, whether
from xenoliths in volcanic rocks or from alpine peridotite
bodies, show a greater range at lower oxygen fugacities between the FMQ and iron^wu«stite (IW) buffers (Fig. 1;
Eggler, 1983; O’Neill & Wall, 1987; Wood & Virgo, 1989;
Wood, 1991; Woodland & Koch, 2003; Woodland et al.,
2006; Frost & McCammon, 2008). An increasing body of
evidence indicates that the deeper mantle is even more
reduced and may be metal-saturated (Ballhaus & Frost,
1994; Rohrbach et al., 2007; Frost & McCammon, 2008).
Figure 1 indicates the position of the oxygen buffers
referred to above and shows the range of fO2 typical for
most volcanic and mantle rocks. The wide range of
oxygen fugacities exhibited by mantle peridotite samples
indicates that, provided that oxidation states vary over
small distances, there is a large potential for
redox-controlled reactions that will include melting reactions. This would be in keeping with modern conceptions
of the mantle as containing a complex intermingling of
pyroxenites, eclogites and other mineral assemblages with
peridotite (Alle'gre & Turcotte, 1986; Foley et al., 2001;
Jacob, 2004; Sobolev et al., 2007), thus resembling a migmatite more than the uniform peridotite as in older notions.
The melting points of peridotite and other ultramafic
rocks likely to be present in the mantle have been shown
by experimental studies to be extremely susceptible to the
presence of even small amounts of volatile components
such as H2O, CO2 and CH4 (Green, 1973; Eggler, 1976;
Wyllie, 1978; Taylor & Green, 1988; Dasgupta &
Hirschmann, 2006; Foley et al., 2009). The term ‘redox melting’ was coined by Taylor (1985) for melting caused by the
increase in water activity as a result of the oxidation of
reduced H2O þ CH4 fluids, accompanied by the precipitation of solid carbon. It was, however, anticipated by
Wyllie (1980), who wrote ‘if reduced oxygen fugacity raises
the solidus for the system peridotite^C^H^O. . . partial
melting would then only occur where temperature were
raised or oxygen fugacity was increased’.
Here, the concept of ‘redox melting’ is extended to include a second mechanism of melting in more oxidized
conditions in which a further drop in the melting point is
achieved by the introduction of CO2 into the system by
NUMBERS 7 & 8
JULY & AUGUST 2011
the oxidation of carbon to carbonate. Recent experimental
studies at 4^6 GPa (Foley et al., 2009) confirmed the findings of Wallace & Green (1988) at lower pressures
(53·1GPa) that the melting temperature of peridotite in
the presence of both H2O and CO2 is lower than with
H2O or CO2 alone. The change in melting temperature
under oxidizing conditions is not as large as that under
reducing conditions, but the melting points are still lower,
making the appearance of the effects of this second mechanism in the rock record more likely, particularly in cratonic areas. The melting curves of mantle peridotite at
various oxidation states were surprisingly poorly constrained until recently, and there is still considerable room
for improvement under reduced conditions. Furthermore,
the involvement of ultramafic assemblages other than peridotite is now thought to be important during melting of
the mantle (Foley, 1992a, 1992b; Hirschmann & Stolper,
1996; Pertermann & Hirschmann, 2003; Sobolev et al.,
2005, 2007); however, the effect of oxygen fugacity on the
melting of these components is yet to be studied.
In the context of this study, I argue that melting of the
mantle as a result of redox reactions is more important
than has been realized, and in some geodynamic situations, such as cratonic mantle lithosphere and the deep
recycling of lithospheric blocks, it may be more important
than temperature changes. The rejuvenation of cratonic
blocks by episodic infiltration of small-degree partial melt
is likely to result from the first, reduced, redox melting
mechanism, here termed hydrous redox melting (HRM),
and later by the second, more oxidized, mechanism (carbonate redox melting; CRM). The consecutive action of
both redox melting mechanisms is a logical consequence
of the erosion of cratonic lithosphere mantle (Foley, 2008).
The HRM mechanism may have been more important in
the first half of Earth history before oxidation of the uppermost parts of the mantle and the surface. The operation of
redox melting during the deep recycling of lithospheric
mantle blocks and in melting processes in the early Earth
and in the evolution of the Earth’s mantle is discussed
below.
T H E OX I DAT I O N S TAT E O F T H E
MA NTLE LITHOSPH ERE A ND
A ST H E NO S P H E R E
It is now recognized that the oxygen fugacity in the Earth’s
upper mantle varies by several orders of magnitude both
vertically and laterally (Ballhaus & Frost, 1994; Woodland
et al., 2006; Foley, 2008; Frost & McCammon, 2008;
Mackwell, 2008). However, available data on the upper
mantle oxidation state are often generalized to give an
average value that lies around FMQ 1 to FMQ (Fig. 1).
Average values tend to be emphasized [e.g. FMQ 0·41
for mid-ocean ridge basalt (MORB) glasses, Be¤zos &
1364
FOLEY
REDOX MELTING IN MANTLE
Fig. 1. A comparison of commonly used oxygen buffers plotted in terms of oxygen fugacity against temperature (a) and pressure (b). OIE,
olivine^iron^enstatite corrected for Mg-number ¼ 90 indicates the approximate fO2 of metal saturation in mantle peridotite, although a
Ni-rich metal appears at around IW (O’Neill & Wall, 1987). EMOG ¼ enstatite-magnesite-olivine-graphite. CW is not a buffer, but indicates
the position of the maximum H2O content in COH fluids. The plot against pressure (b) is less commonly used, but shows the divergence of
FMQ from NNO and the higher relative fO2 of CW at high pressures, which promotes the importance of the reduced realm in Fig. 9. Oxygen
fugacities throughout this study are given relative to FMQ. Volcanic rocks range mostly around and slightly above FMQ^NNO (grey box;
Carmichael, 1991) whereas xenolith samples of mantle rocks range from slightly above FMQ down to IW (very few samples in the light grey
part of box). These ranges are not temperature-specific, but are intended to show the generally higher oxidation state of volcanic rocks relative
to their mantle sources.
1365
JOURNAL OF PETROLOGY
VOLUME 52
Humler, 2005; FMQ þ 0·24 for continental spinel lherzolites, Bryndzia & Wood, 1990], whereas exceptions, deviations and regional variations that are potentially
important for the operation of redox melting are not.
Therefore, the information on the redox state of the uppermost mantle is summarized here with a view to assessing
the heterogeneities resulting from juxtaposition of blocks
because of geodynamic movements, or from possible
changes in the oxidation state through the evolution of the
Earth.
Information on the oxidation state of the upper mantle
can potentially be gleaned from several sources, including
the Fe3þ/Fe2þ ratios and redox-sensitive trace element
ratios of volcanic rocks, oxybarometry on mantle-derived
minerals and xenoliths, on diamonds and mineral inclusions within them, and from evidence for the presence of
minerals such as carbonates whose stability is restricted to
particular fO2 conditions.
The most studied, and consequently best understood,
region of the mantle in the context of oxygen fugacity is
the continental lithosphere, based on extensive calculations with oxybarometers that depend on the exchange of
iron as Fe3þ and Fe2þ between component minerals of
peridotites (Mattioli & Wood, 1986; O’Neill & Wall, 1987;
Wood & Virgo, 1989; Ballhaus et al., 1991). There are several
calibrations, each with an uncertainty of around 0·5 log
units fO2, so that even a mantle with uniform fO2 would
show a scatter of calculated fO2 values. However, I
show here a compilation of the available data as a series
of histograms (Figs 2^4), which shows that the variations
are much larger than the uncertainties, such that blocks
with contrasting fO2 are likely to become juxtaposed
in the mantle. A complicating factor is that the fO2 of
peridotites should decrease by c. 0·7 FMQ units per
1GPa increase in pressure (Ballhaus & Frost, 1994).
For this reason, the information contained in the histograms is summarized in Fig. 5, which shows a series of parallel lines representing the average fO2 for peridotites
in various tectonic settings. The orthogonal distances
between the parallel lines represent real, pressureindependent differences in the fO2 of peridotites in these
tectonic settings.
Garnet peridotites sampled by kimberlites in cratonic regions have consistently been shown to have lower fO2 and
to show a greater range of fO2 between FMQ and
FMQ 5, with most values falling between FMQ 2 and
FMQ 4 (Frost & McCammon, 2008). These are shown
here to have an average of FMQ 2·83 (Fig. 2, top left
panel), although there appear to be craton-specific regional differences as shown particularly by the lower average
fO2 of FMQ 4·05 for the Baltic craton (Fig. 3, left
panels; Woodland & Peltonen, 1999). The large range for
cratonic peridotites is shown in Fig. 5 (after Frost &
McCammon, 2008) together with the slope expected
NUMBERS 7 & 8
JULY & AUGUST 2011
from the pressure effect (Ballhaus & Frost, 1994;
thick blue line). This allows a pressure correction for comparison with peridotites from other tectonic settings. The
effect of pressure is illustrated by the difference between
non-cratonic lithosphere and abyssal peridotites, which
show a similar range in the histograms (Fig. 2), but are
41 log unit fO2 apart in Fig. 5. This is because of the
lower equilibration pressure of the abyssal spinel lherzolites (assumed average of 1GPa or 33 km depth, corresponding to the approximate depth of separation of melts
from their mantle source) in comparison with the
deeper-derived continental xenoliths, many of which contain garnet.
For the oceanic mantle, much less information is available, and the age information is much more restricted because of the ephemeral nature of oceanic crust and
lithosphere (the ocean crust is restricted to the last 54%
of Earth history). An indirect method of estimation of the
oxidation state at mid-ocean ridges is the Fe3þ/Fe2þ ratio
in mid-ocean ridge basalts. The first compilation by
Christie et al. (1986) found fO2 values of FMQ 1·35 for
Atlantic MORB glass rinds and FMQ 1·5 for Pacific
MORB glass rinds (‘MORB-1’ in Fig. 5), whereas the
more slowly crystallized basalt cores of pillows showed the
more oxidized values of FMQ þ 0·5 for the Atlantic and
FMQ þ 0·3 for the Pacific. This information was emphasized in a previous contribution on redox melting (Foley,
1988), not only for the more reduced state of oceanic relative to continental basalts, but also as evidence for
late-stage oxidation of the melts between rapid quenching
of the glassy rims and slower cooling of the basalt pillow
cores. The first line of evidence has been called into question by a more recent compilation of glass analyses: Be¤zos
& Humler (2005) found a more oxidized average value of
FMQ 0·41 (‘MORB-2’ in Fig. 5), and this has been used
as confirmation of the homogeneity of mantle redox states
(Frost & McCammon, 2008). However, the discrepancy of
41 log unit fO2 in pressure-corrected values between abyssal peridotites (Bryndzia et al., 1989; Bryndzia & Wood,
1990) and continental peridotites remains (Fig. 5), and a
more reduced state of the oceanic mantle is also favoured
by the expectation that melts will oxidize during emplacement (Ballhaus, 1993), as was also indicated by the difference in Fe3þ/Fe2þ ratios between pillow centres and their
rapidly chilled rims (Christie et al., 1986). Assuming the
same pressure dependence of 0·7 FMQ units per GPa
pressure (Ballhaus & Frost, 1994), the average value for
abyssal peridotites of FMQ 1·05 can be extrapolated to
higher pressures to form an array for the convecting
sub-oceanic asthenosphere (dashed line in Fig. 5) that is
about 0·7 log units below that of the cratonic lithosphere
and 41 log unit below the non-cratonic continental lithosphere. This difference means that the average fO2 of abyssal peridotites at 35^45 km beneath mid-ocean ridges
1366
FOLEY
REDOX MELTING IN MANTLE
Fig. 2. Variation of oxidation states in mantle rocks as a function of tectonic setting on a global scale, expressed as the deviation (FMQ) from
the FMQ buffer (see Fig. 1 for position relative to other buffers). The average for the oceanic mantle is lower than for the non-cratonic continental lithosphere, which itself shows considerable variation. Higher values are typical of supra-subduction zone and ocean island settings. The limited data for rift-influenced continental mantle include modifed Archean mantle, and may therefore manifest a considerable oxidation effect
relative to the generally uniformly lower oxidation state of cratonic lithosphere. Data sources: cratonic mantleçLuth et al. (1990), Woodland &
Peltonen (1999), McCammon et al. (2001), Woodland & Koch (2003), McCammon & Kopylova (2004), Creighton et al. (2009, 2010), Lazarov et al.
(2009); oceanic and abyssal peridotitesçBryndzia & Wood (1990), Nasir (1996), Canil et al. (2006); continental lithosphere (non-cratonic)ç
Wood & Virgo (1989), Luth et al. (1990), Ionov & Wood (1992), Qi et al. (1995), Luhr & Aranda-Gomez (1997), Nasir et al. (2006, 2010);
supra-subduction zone xenolithsçWood & Virgo (1989), Canil et al. (1990), Brandon & Draper (1996), Parkinson & Arculus (1999), McInnes et al.
(2001), Parkinson et al. (2003), Bryant et al. (2007),Wang et al. (2008, 2009); oceanic within-plate xenolithsçBallhaus (1993); rift-influenced continental mantleçRudnick et al. (1994), Foley et al. (2006); orogenic massifsçWoodland et al. (1992,1996, 2006), Song et al. (2009).
corresponds to about 85^95 km in the continental
lithosphere.
Further variation is visible when one considers
rift-related xenoliths and those from supra-subduction
zone environments. The rift-influenced peridotites are
from cratonic (Tanzania) or close to cratonic (eastern
Antarctica) regions, and show higher pressure-corrected
fO2 than the rest of the continental mantle. This is barely
visible from the histogram (Fig. 2), but variations within
the sample set from both areas show that the lowest fO2
values are from samples least affected by overprinting
(Rudnick et al., 1994; Foley et al., 2006). This is interpreted
to indicate the oxidizing effect of melt infiltration in the
plume-related uplift of the Tanzanian craton and beneath
the developing Lambert^Amery rift in Antarctica (Foley
et al., 2006).
1367
JOURNAL OF PETROLOGY
VOLUME 52
NUMBERS 7 & 8
JULY & AUGUST 2011
Fig. 3. Variation in the oxidation state of cratonic lithosphere of Archaean age (left panels) and in supra-subduction zone peridotites from continental arc (upper right) and island arc (lower right) settings. Data sources as in Fig. 2. Whereas the variation in cratonic mantle lithosphere
may be largely owing to the effect of pressure (see also Fig. 5), the arc peridotites show a larger range in oxidation states at a more restricted
range of pressures.
It has frequently been shown that later metasomatism of
peridotites caused by the infiltration of melts leads to an increase in the fO2 of the peridotites (McGuire et al., 1991;
Frost & McCammon, 2008; Creighton et al., 2009), so that
the position of the continuous blue line for cratonic
peridotites in Fig. 5 would be lower by c. 1 log unit if it is
to represent the original pre-metasomatic redox state.
Information from orogenic peridotite massifs (average
FMQ 0·21; Fig. 2) documents this oxidizing effect of
melt infiltration (Woodland et al., 1996). Histograms for
orogenic peridotites with textural and chemical evidence
for melt infiltration are compared with those for probably
unaffected (or less affected) peridotites in Fig. 4. The average for the metasomatically overprinted peridotites is
FMQ þ 0·38, 0·86 log units higher than for the unaffected
peridotites (FMQ 0·48). It is increasingly recognized
from textural relationships and chemical zonation that
clinopyroxenes and garnets in continental peridotites are
often introduced later (Glaser et al., 1999; Simon et al.,
2007; Rehfeldt et al., 2008). This is important for the reconstruction of mantle conditions during Archaean times; as
melt infiltration tends to correlate with increasing oxidation state, the implication is that the cratonic mantle originally had lower fO2 than it now displays.
The fO2 of peridotites from the mantle wedge above
subducting slabs is often summarized as being more oxidized than in stable continental areas (McInnes et al.,
2001; Canil, 2002; Parkinson et al., 2003; Malaspina et al.,
2009). The average of available data is FMQ þ 0·51 (Fig. 2,
top right panel), including a substantial proportion of
Fig. 4. Oxidation state of peridotites from orogenic massifs showing
the oxidizing effect of metasomatic melt infiltration averaging c. 0·86
log units fO2. Data sources as in Fig. 2. Melt infiltration by similar
processes is thought to be widespread in the mantle lithosphere, particularly the lower continental lithosphere.
rocks with fO24FMQ þ1. There is also a distinct difference between peridotite xenoliths from island arcs (average FMQ þ1·3) and those from continental arcs (average
FMQ 0·16; Fig. 3, right panels). The pressure-corrected
average (Fig. 5) is 1·6 log units above the cratonic mantle,
and 2·3 log units above the asthenosphere value. These
more oxidized values are usually attributed to the infiltration of melts or water-rich fluids associated with the subducting slab. However, evidence for the sense of redox
1368
FOLEY
REDOX MELTING IN MANTLE
Fig. 5. A summary of oxygen fugacity data from mantle samples as a function of their pressure of origin (modified after Frost & McCammon,
2008). The continuous blue line has a slope of 0·7 FMQ units per GPa (Ballhaus & Frost, 1994) indicating the pressure effect on fO2 that explains most of the cratonic array with a scatter of 1 log units. The dashed line is drawn parallel, but originates from the average value for abyssal peridotites (yellow square at FMQ 1·05; see Fig. 2), showing that the asthenospheric mantle should be more reduced than even the
cratonic peridotites in their present state. MORB-1 is the average fO2 of MORB after Christie et al. (1986), and MORB-2 the corrected value
after Be¤zos & Humler (2005). The more oxidized position than average abyssal peridotites reinforces the conclusion of Christie et al. (1986)
that MORB melts oxidize on their way to the surface. Further lines for the non-cratonic continental lithosphere (green), rift-influenced continental lithosphere (red) and supra-subduction zone peridotites (orange) are plotted parallel, but through the average values shown in the histograms of Fig. 2 plotted at the depths indicated by thermobarometry. The grey symbols show the positions at 1GPa, a reasonable depth for
extensive melting beneath mid-ocean ridges, to facilitate direct comparison of the relative oxidation state in different tectonic settings. The
array of lines indicates a variation of at least 2^3 log units fO2 at any depth.
change during subduction processes is complex and partly
contradictory; there are examples of subduction metasomatism with fO2 between 1 and 5 log units below FMQ
(Song et al., 2009; Wang et al., 2009), and metal-bearing
peridotites are known from the sub-Kamchatka mantle
(Ishimaru et al., 2009).
A more recent development in the estimation of redox
states in the mantle source regions of magmas uses the
V/Sc ratio (Li & Lee, 2004; Lee et al., 2005). This is based
on the similar incompatibility of the two elements during
melting, so that their fractionation at various degrees of
melting should be minimal, coupled with the fact that vanadium exists in several oxidation states, whereas Sc occurs
only in the 3þ state in geological conditions. There are
now enough results to compare melts from different geodynamic environments, but these are not consistent with
those from oxygen barometry: current estimates see no
consistent difference between MORB, ocean island basalts
and island arc basalts (Lee et al., 2005; Mallmann &
O’Neill, 2009) despite the differences delineated above
from oxygen barometry of peridotites.
The V/Sc ratio of basalts has also been used to suggest
that the oxidation state of basalt sources has remained essentially constant through geological time, with a V/Sc
average for modern MORB of 6·74 compared with an
average for Archaean basalts of 6·34, indicating a change
of less than 0·3 FMQ units since 3·5 Ga (Li & Lee,
2004). A similar result was obtained from Cr abundances
in volcanic rocks by Delano (2001). However, the oxybarometry results for cratonic peridotites indicate a great
range of redox states, with some appreciably more reduced
than the extrapolation of abyssal peridotite values (Fig. 5;
Ballhaus, 1993; Woodland & Peltonen, 1999; Kadik, 2003),
leading to an apparent contradiction. Rare occurrences of
1369
JOURNAL OF PETROLOGY
VOLUME 52
coexisting cohenite (Fe3C) and Fe-metal in cratonic diamonds (Jacob et al., 2004) would appear to correspond to
the lowest possible fO2 conditions for diamond formation
(Wood, 1993). A possible solution to the paradox is that
the data on Archaean basalts considered by Delano (2001),
Li & Lee (2004), and Berry et al. (2008) are restricted to
continental settings, which show only the more oxidized
part of the upper mantle range on the modern Earth. The
same is true for late Archaean komatiites, which are also
no more reduced than modern continental basalts (Canil,
1997; Berry et al., 2008). The best estimates for the composition of Archaean oceanic crust are provided by cratonic
eclogite xenoliths in kimberlites; a compilation of the limited available V/Sc data gives an average of 7·38, encompassing eclogites from West and South Africa, Siberia and
Canada (Jacob & Foley, 1999; Barth et al., 2001; Jacob
et al., 2005; Smart et al., 2009), with regional averages ranging from 4·7 to 7·57. Care must be taken here to eliminate
fractionated rock compositions, which tend to have higher
V/Sc, because many eclogites are thought to represent
former gabbros and not volcanic rocks (Jacob, 2004).
Li & Lee (2004) introduced a selection window of
8^12% MgO to restrict their attention to unfractionated
volcanic rocks. If we restrict our attention to the very few
data for eclogites inferred from their compositions and
high d18O values to be picritic volcanic rocks and not gabbros (Jacob & Foley, 1999), then the V/Sc range of 4·9^5·9
may indicate slightly more reducing conditions for oceanic
crust production at 2·6 Ga than the current global
MORB average (6·74; Li & Lee, 2004). The exact difference in terms of fO2 depends on the degree of melting (Li
& Lee, 2004), but is between 0·5 and 0·7 log units, bearing
in mind the picritic composition of the ‘volcanic’ late
Archaean eclogites (Jacob & Foley, 1999). If the oxidation
of the mantle was gradual and continuous, then the early
Archaean mantle was probably more than one log unit
more reduced than in modern MORB magma sources.
This contrasts with the conclusions from Cr and V/Sc in
continental Archaean volcanic rocks, and may indicate
that redox contrasts between tectonic settings in the
Archaean were at least as large as those seen in Figs 2^5
for the modern Earth. The continental signal from the
Archaean is not typical, it is just preferentially preserved.
The picture emerging from these various lines of evidence provides abundant potential for the geodynamic
juxtaposition of blocks with contrasting redox state, both
in terms of average values and the variation of redox
states in all geodynamic environments (Figs 2^5). The
pressure effect dictates that more reduced conditions will
prevail in the lower cratonic lithosphere and also in the asthenospheric mantle in close proximity to it. In these regions the HRM mechanism is likely to be widespread,
whereas the CRM mechanism will be commoner closer to
the surface, or where more oxidized blocks are found at
NUMBERS 7 & 8
JULY & AUGUST 2011
great depth. Another implication of the pressure effect is
that the hypothesized initiation of melting beneath
mid-ocean ridges at c. 300 km in the presence of CO2
(Dasgupta et al., 2007) must be considered very unlikely
because the redox state will generally be far too low to
allow the stability of CO2. Instead, methane and/or diamond would dominate the carbon species and the mantle
may even be metal-saturated at these depths (Ballhaus &
Frost, 1994; Rohrbach et al., 2007; Frost & McCammon,
2008). The most important conclusions for the following
discussion are that the boundary between the lithosphere
and convecting asthenosphere is likely to correspond to a
redox front, and that the range of redox states at any
given depth is 42 log units even without the presence of
recycled blocks, leaving ample potential for redox melting
reactions.
R E D OX M E LT I N G B Y
H Y D R AT I O N ( H R M ) : OX I DAT I O N
OF R E DUC E D F LU I D
COMPONENTS
The redox melting mechanism originally proposed by
Taylor (1985) is referred to here as redox melting by hydration (HRM) and operates in relatively reduced conditions
slightly above the IW buffer at depths corresponding to
the lower cratonic lithosphere (Fig. 1). Given the evidence
summarized above for the oxidation state of mantle rocks,
it can be expected to apply to melt production in deeper
levels of the mantle (150^250 km), except directly below
mid-ocean ridges where it may operate at shallower levels.
Melting is a indirect product of the oxidation of methane
to form water and solid carbon:
CH4 þO2 ¼ 2H2 O þ C:
The increase in aH2O results in depression of the melting
temperature by hundreds of degrees in the presence of
water (Green, 1973). The effect is shown in Fig. 6a on a
plot of the C/(C þ H2) ratio against fO2. All mixtures of
C þ H þ O fluid components falling within the grey
shaded region are carbon-saturated, so that solid carbon
in the form of either graphite or diamond coexists with a
fluid composition lying on the continuous curved line
(Frost, 1979). The C, H and O components of a fluid under
reducing conditions around the IW buffer consist principally of CH4, H2O and H2, and will release water whilst
precipitating carbon on encountering more oxidized
rocks. The increase in water activity may result in melting
if the solidus is depressed sufficiently to meet the ambient
geotherm. Whether melting occurs will depend on the concentration of volatile components and thus the amount of
water released by the redox mechanism and the variation
in geothermal gradient with geodynamic setting. The
rapid change in the H2O/CH4 ratio of fluids in the fO2
1370
FOLEY
REDOX MELTING IN MANTLE
Fig. 6. The mechanism of redox melting by hydration (HRM). (a) Carbon saturation curve for COH fluids at realistic mantle melting conditions of 3 GPa and 1300 K (10278C). Compositions in the pale grey area consist of solid carbon plus a fluid with a composition on the curved
line (carbon saturation curve). Fluids at intermediate fO2, about halfway between IW and FMQ, consist of 496% H2O at this pressure and
temperature, effectively separating a region of reduced CH4 þ H2O fluids from oxidized CO2 þ H2O fluids. When reduced fluids come into contact with oxidized material they move vertically in the diagram until carbon saturation is reached, and then increase in H2O content, which
can lead directly to melting by the depression of melting point by increased water activity (see Fig. 8). The CH4/H2O ratio is strongly dependent
on fO2 at about IW þ1, as is the CO2/H2O ratio at FMQ 0·5. (b) Species abundance as a function of fO2 between FMQ and IW ^ 1, showing
the rapid transition from CO2- to H2O-rich fluids at FMQ 0·5, and from H2O- to CH4-rich fluids at IW þ1 (FMQ 3 to FMQ 4). The
line ‘water-maximum’ denotes the fO2 at which H2O content of the fluid is highest.
region 0·5^1·5 log units above IW (at the temperature^
pressure conditions of Fig. 6) restricts the operation of
HRM to this narrow range of fO2: this is clearer in Fig.
6b, in which the molar amounts of fluid species are shown
plotted against fO2 for the same pressure^temperature
conditions as in Fig. 6a. The carbon saturation curve indicates that fluid compositions pass through a maximum
H2O content (496% in Fig. 6a) between 1·5 and 2·0 log
units above IW, and then experience a rapid increase in
CO2/H2O over the next 1 log unit, reaching extremely
CO2-rich and H2O-poor compositions whilst still below
FMQ. This will be important for the second, CRM, mechanism at a later stage.
The possibility of melting as a result of HRM depends
on the geothermal gradient and on the magnitude of the
increase in aH2O. The effect of increasing aH2O on the solidus of peridotite is estimated in Fig. 7 from the limited experimental data on the reduced solidus by Taylor & Green
(1988). This diagram only approximates conditions in the
Earth’s mantle because it does not account for dissolved
silicates in the fluid, which will increase in importance towards higher pressures, possibly resulting in termination
of the solidus at the second critical end point (Kessel
et al., 2005). Above this pressure, supercritical liquids that
compositionally mimic melts will occur at temperatures
below the extrapolation of the solidus (white dotted line;
1371
JOURNAL OF PETROLOGY
VOLUME 52
NUMBERS 7 & 8
JULY & AUGUST 2011
Fig. 7. Dependence of the melting curve (solidus) of peridotite on water activity (after Taylor & Green, 1988). The intermediate curves for
aH2O ¼ 0·35, 0·70 and 0·85 are most relevant for redox melting (HRM), whereby the increasing solubility of silicate material in fluids towards
the higher pressures shown here decreases the aH2O in fluids. The higher temperatures relative to water-saturated conditions dictate that
HRM is a realistic mechanism for the continental lithosphere as indicated by the SE Australia geotherm, and may also occur in the lower
reaches of cratons where the deposition of diamond will result. The white circle indicates the approximate position of the second critical
end-point in peridotitic systems: the solidus does not exist at higher pressures than this (dotted line), but is replaced by a continuum between
fluid- and liquid-like behaviour termed ‘supercritical liquid’ to emphasize its liquid-like element partitioning behaviour (Kessel et al., 2005).
Kessel et al., 2005). The position of the second critical end
point, shown in Fig. 7 as a white circle, is very poorly
known, with experimental estimations varying from
3·8 to 10 GPa (Stalder et al., 2001; Mibe et al., 2007). The
position shown at a depth of c. 220 km is probably a minimum for peridotite, as this is only slightly above the pressure determined for eclogite (Kessel et al., 2005), which
has a much lower MgO concentration. Because of its high
content of MgO and low contents of SiO2, Na2O and
Al2O3, the correct position of the second critical end
point for natural mantle peridotite may be deeper than
the lower end of Fig. 7. As the second critical end point is
approached, the amount of silicate material dissolved in
the fluid increases greatly and this will result in a drop in
aH2O, leading to an increase in the solidus temperature.
A third uncertainty concerns the solidus of peridotite in
H2O-rich conditions. Grove et al. (2006) determined the
solidus temperature to be as low as 8508C at pressures between 2·5 and 3 GPa with 14·5 wt % H2O, which is 150^
2008C lower than that shown in Fig. 7. However, Green
et al. (2010) interpreted the glasses in Grove et al.’s
experiments as being quenched out of a fluid phase: their
results indicate a solidus temperature of 13508C at 6 GPa,
which is close to the aH2O ¼ 0·85 line in Fig. 7, possibly because of high silicate solute concentrations.
Despite these uncertainties, Fig. 7 illustrates the effect of
increasing aH2O in decreasing the melting temperature,
and thus the principle of the HRM mechanism. Referring
to the example of Fig. 6, the water activity at fO2 ¼ IW
before the HRM mechanism begins is likely to be well
below 0·35 as a result of the abundance of CH4 at this
fO2. The solidus for these redox conditions (aH2O ¼ 0·35)
is far above the geotherms for continental areas (Fig. 7).
Following oxidation of methane accompanied by the deposition of carbon, the aH2O will have increased to
around the 0·85 line, and the peridotite solidus will be
very close to the geothermal gradient. It should be noted
here that CH4 þ H2O fluids are strongly non-ideal
(Taylor, 1985; Matveev et al., 1997), so that a small molar
CH4 content results in a higher aCH4, and correspondingly lower aH2O, thus accentuating the role of changing
CH4/H2O in the fluid. At higher pressures than those
1372
FOLEY
REDOX MELTING IN MANTLE
higher pressures where fluids with 499% H2O are stable
over several log units fO2 (Woermann & Rosenhauer,
1985). The HRM mechanism remains constrained to a
small fO2 range at the lower fO2 end of this water maximum, so that the fO2 range applicable to HRM moves towards lower fO2 at higher pressures as shown in Fig. 9.
This diagram divides oxygen fugacity^pressure space into
reduced and oxidized realms (Foley, 1994), separated by
the lightly coloured central region in which all fluids are
extremely water-rich, so that no important change in fluid
composition can occur to cause melting in this region. In
terms of the potential for redox melting, this area of intermediate fO2 is relatively inert or ‘barren’. The conditions
of HRM are indicated by the diagonally lined region as
being limited to a very restricted range of fO2 (Fig. 9).
The rapid change from H2O-rich to CO2-rich fluids in
less than one log unit fO2 above the water maximum also
constrains melting to be restricted to a very small fO2
range. Thus, redox melting is active only in the diagonally
lined regions in Fig. 9 despite the existence of a wide
range of oxidation states in peridotite samples (Arculus,
1985; Frost & McCammon, 2008).
Fig. 8. Pressure effect on the width of the water maximum. Curved
lines show the carbon saturation curves (as in Fig. 6) for 3, 5, 7 and
10 GPa. Fluids below the curve are CH4 þ H2O mixtures, whereas
those above the curve are CO2 þ H2O mixtures. These regions are
separated by a point of extreme H2O content in the fluid (circles).
Squares indicate the positions of intersection with the IW buffer. The
pale shaded area emphasizes the existence of fluids with 490% H2O
over 6 log units fO2 at depths of 300 km, in contrast to the darker
shaded area for 3 GPa, which spreads over only 2·5^3 log units. This
pressure effect leads to the polarization of redox melting conditions
shown by the width of the pale shaded area in Fig. 9.
corresponding to Fig. 6, the water maximum widens and
the influence of CH4 is suppressed still further
(H2O499%; Fig. 8). Here, the relevant peridotite solidus
will be essentially indistinguishable from the watersaturated solidus in Fig. 7 and melting will occur where
the solidus cuts the geothermal gradient, which should
apply in parts of the lower cratonic lithosphere. The water
maximum is less marked at lower pressures, but HRM
will nevertheless apply away from cratons because the
form of the reduced solidus (Taylor & Green, 1988) is conducive to cutting the higher geothermal gradients at
depths of 60^100 km.
The exact form of the carbon saturation surface and its
position relative to fO2 buffers varies with pressure and
temperature, whereby particularly the pressure effect is
important for mantle melting by HRM. Figure 8 shows
that the water maximum is more marked and wider at
R E D OX M E LT I N G B Y
C A R B O N AT I O N ( C R M ) :
OX I DAT I O N T O C A R B O N AT E S TA B L E C O N D I T I O N S
A second mechanism by which melting of peridotite can
occur as a result of a change in fO2 without transport of
heat has become apparent from recent experiments on the
melting of peridotite in the presence of both H2O and
CO2 (Foley et al., 2009). A further increase in fO2 from
conditions corresponding to the water maximum will lead
to a rapid increase in the ratio of CO2 to H2O in the fluid
as long as the fluid remains saturated in carbon (Fig. 6).
The oxidation may occur by infiltration of fluids or melts
along cracks in the lithosphere or along grain boundaries
in the convecting mantle, or by simple juxtaposition of
blocks with contrasting oxygen fugacities by geodynamic
movements. If little or no carbon is present in the rock or
in the infiltrating fluid, then the carbon will soon be exhausted and the system cannot proceed to CO2-rich conditions. However, saturation in carbon will be ensured
wherever solid carbon is abundant in the rock in the form
of either graphite or diamond, so that the previous operation of HRM in the same rock would provide the ideal
pretreatment of the mantle for this second mechanism.
The abundance of diamonds in the lower cratonic lithosphere, documented by peridotite and eclogite xenoliths as
well as mineral inclusions in diamonds (Boyd & Gurney,
1986; Stachel & Harris, 2008) thus means that the lower
cratonic lithosphere may be especially susceptible to CRM.
1373
JOURNAL OF PETROLOGY
VOLUME 52
NUMBERS 7 & 8
JULY & AUGUST 2011
Fig. 9. Regions of operation of the two redox melting mechanisms in pressure^fO2 space. The dark shaded areas indicate reduced, CH4bearing and oxidized CO2-bearing conditions (as in Fig. 6). In the palest shaded area, H2O accounts for 490% of the C^O^H components
in the fluid, and this widens towards higher pressures, thus effectively polarizing the melting areas (diagonal line shading), so that the fO2
values of HRM and CRM diverge towards higher pressures. Melting conditions are limited to small areas on this graph, but the mantle can
be quickly oxidized across the H2O-rich area. The lines for asthenosphere, cratonic mantle, continental lithosphere and supra-subduction zone
mantle correspond to the average oxygen fugacity lines in Fig. 5. These indicate that CRM is unlikely at lower lithosphere levels unless an unusual oxidation event triggered by the oxidation of solid carbon is realized. CRM may be important in subduction zones. In contrast, much of
the mantle at depths of 100^300 km will be close to the fO2 of operation of the HRM mechanism, which may thus be a prevalent cause of incipient melting.
The rapid production of CO2 dictates that the melting
curve now corresponds to peridotite melting in the presence of CO2 and H2O. Experiments on peridotite with
small amounts (0·4^2 wt %) of both H2O and CO2 by
Wallace & Green (1988) indicated that the melting point
lies at 960^9808C between 65 and 100 km and that a field
for carbonatitic melt lies directly above the solidus. This
corrected earlier experiments (see Olafsson & Eggler,
1983, and references therein) in which the solidus temperature may have been overestimated as a result of alkali^carbonate melt pockets being dissolved during preparation of
polished mounts. However, only recently have experiments
determining the position of the solidus at higher pressures
(43^10 GPa) become available. First results came from
melting of peridotite in the presence of CO2 only (i.e. without H2O), which indicate the presence of carbonate-rich
melt at temperatures close to the water-bearing solidus
(Canil & Scarfe, 1990; Dasgupta & Hirschmann, 2006;
Brey et al., 2008). Experiments at 4^6 GPa with similar
small amounts of both H2O and CO2 to those of Wallace
& Green (1988) have recently confirmed that the solidus is
lower for the mixed volatile phase with respect to either
CO2 or H2O alone (Fig. 10; Foley et al., 2009). These experiments are from a K2O-enriched peridotite composition,
but the position of the solidus with similar amounts of
H2O and CO2 is confirmed in unenriched Hawaiian and
1374
FOLEY
REDOX MELTING IN MANTLE
depending on pressure; Figs 8 and 9), the oxidation of methane causes an increase in aH2O, which in turn depresses
the melting point of the rocks. This is the HRM mechanism. In more oxidized conditions (FMQ 1·5 to
FMQ 0·5 depending on pressure), redox melting is
caused by the oxidation of solid carbon to carbonate. A
very minor amount of water is needed to minimize the solidus temperature, but this CRM is essentially a ‘dry’ mechanism steered principally by carbonation; it does not
produce H2O but depends on reduction of aH2O by the resulting carbonate. The fO2 of operation of the two redox
melting mechanisms diverges to greater depths (Fig. 9).
M E LT C O M P O S I T I O N S
P RO D U C E D B Y T H E H R M A N D
CRM MECH A NISMS
Fig. 10. Comparison of melting curves for reduced (CH4 þ H2O;
after Taylor & Green, 1988; Green & Falloon, 1998) and oxidized
(CO2 þ H2O; Foley et al., 2009) conditions. The reduced solidus follows H2O-undersaturated, CO2-free melting to 3 GPa and then a
low-aH2O solidus at higher pressures. During hydrous redox melting
(HRM), the solidus moves to lower temperatures owing to the increase in aH2O as indicated by the arrows. Melts are not SiO2-undersaturated and resemble olivine lamproites. The oxidized solidus for
H2O þ CO2-bearing conditions is that determined at 4^6 GPa by
Foley et al. (2009), and extrapolated as a dashed line to higher pressures parallel to the solidus for CO2 alone (Dasgupta &
Hirschmann, 2006). Melts just above the solidus are carbonate-rich;
the transition to SiO2-undersaturated silicate melts (25^40 wt %
SiO2) needs experimental clarification at pressures 46 GPa.
However, these solidi may cease to exist if the second critical
end-point is reached (white-filled circles show possible approximate
positions). This possibility is indicated by the pale grey areas in
which supercritical liquids will resemble melts more than fluids in
terms of their element compositions (Kessel et al., 2005). (See text for
further discussion.) UML ¼ ultramafic lamprophyre.
MORB-pyrolite (S. F. Foley et al., unpublished data). The
same uncertainties in the position of the second critical
end point apply for Fig. 10 as for Fig. 7. This melting curve
means that melting is likely to occur in the lower reaches
of cratonic mantle lithosphere as long as the appropriate
mixture of volatiles is available. Thus, both HRM and
CRM mechanisms have the potential to cause melting in
the lower cratonic lithosphere and in areas of the mantle
in other geodynamic settings at depths of 120^300 km
(Figs 7, 8 and 10). A comparison of the P^T positions of
cratonic and asthenospheric geotherms with the
volatile-free peridotite solidus indicates that melting at
this depth is otherwise unlikely (McKenzie & Bickle,
1988; Wyllie, 1988).
In summary, redox melting at two distinct and
well-defined fO2 conditions in the mantle is caused by different petrological mechanisms. In reduced conditions
(IW þ 0·5 to IW þ1·5 or FMQ 3·5 to FMQ 4·5,
The question of whether redox melting as opposed to decompression melting or melting as a result of an increase
in temperature is most important in the natural conditions
corresponding to diverse geodynamic situations will eventually be answered by comparing the melt compositions
found in high-pressure experiments with those of natural
volcanic rocks or interstitial phases in mantle samples.
Redox melting processes may often be concentrated in
areas enriched in ultramafic rock types other than peridotite. However, experimental determinations of melt compositions in mantle assemblages such as garnet pyroxenite
(Irving, 1974; Adam et al., 1992; Pertermann &
Hirschmann, 2003) or those containing abundant hydrous
minerals (Lloyd et al., 1985; Thibault et al., 1992; Foley
et al., 1999) are much rarer and less comprehensive than
studies of peridotite melting, and the effects of redox state
on melt composition are still only poorly known. Here,
melt compositions expected for CRM and HRM are compared with those likely to be produced by other mechanisms under various conditions to form the basis for an
assessment of the relative importance of redox melting
that can be improved by future experimental
investigations.
If the convecting mantle contains only trace amounts of
H2O and CO2, as is generally thought (Wyllie, 1980;
Dasgupta et al., 2007), then the solidus of peridotite will
generally lie considerably above the geothermal gradient
expected for the mantle defined by a conductive lid overlying an adiabatic gradient within the convecting mantle
(Wyllie, 1988). The position of this geotherm depends
largely on the temperature in deeper parts of the mantle,
and thus on the convection patterns, but in none of the
cases illustrated in Fig. 7 does the geothermal gradient
cross the volatile-free solidus. This means that the mantle
is mostly in the solid state; it is only in exceptional circumstances that it melts.
1375
JOURNAL OF PETROLOGY
VOLUME 52
The most geodynamically reasonable scenarios for the
partial melting of volatile-free peridotite in the upper
mantle are by decompression melting of upwelling mantle
beneath mid-ocean ridges at depths of 20^50 km and by
melting in the laterally spreading heads of mantle plumes
beneath either oceanic or continental lithosphere (Wyllie,
1988; McKenzie & Bickle, 1988). This conclusion is in
agreement with geophysical observations of the upper
mantle such as seismic-wave velocity (White et al., 1992)
and electrical conductivity profiles (Shankland & Waff,
1977), which discount widespread extensive melting except
in these regions of the uppermost 120 km of the mantle.
Melting in plume heads beneath oceanic lithosphere will
be concentrated at depths of 70^100 km, depending on the
distance from the nearest mid-ocean spreading centre and
thus on the thickness the conductive lid of lithosphere has
attained, whereas the source regions of continental flood
basalts are likely to be deeper (Neal et al., 1997).
Extensive experimental data on the melt compositions
generated by melting of dry peridotite show that there is a
general increase in MgO and alkali content, and a decrease in silica activity with increase in pressure. In addition, at a given pressure, an increase in the temperature,
and thus the degree of melting, leads to a decrease in
alkali content and an increase in MgO (Green, 1970;
Jaques & Green, 1980; Takahashi, 1986; Falloon & Green,
1987, 1988; Kinzler & Grove, 1992; Herzberg & O’Hara,
1998). These trends would result in the production of tholeiitic melts at depths of 15^40 km beneath mid-ocean
ridges (Falloon et al., 1988; Fujii, 1989), whereas plume-head
melts at higher pressures would be olivine basalt to picrite.
Low-degree melting at 460 km depth may produce alkaline, silica-undersaturated compositions that explain the
alkaline magma series observed in ocean islands (Jaques
& Green, 1980). Integrated studies of experimental petrology and mineral chemistry (Green et al., 2001) indicate
that mantle temperatures and potential temperatures are
probably higher than commonly modelled (e.g.
McKenzie & Bickle, 1988), so that picritic melts may be
commoner beneath both mid-ocean ridges and ocean islands than is often thought. In Archaean times, higher
mantle temperatures may have led to the production of komatiitic melts (18^30% MgO; Arndt, 1977; Nisbet et al.,
1993), although it is debatable if these melts were restricted
to plumes (Abbott et al., 1994; Be¤dard, 2006), or were typical of oceanic crust in the Archaean (Arndt, 1983; Nisbet
& Fowler, 1983).
Experimental studies of melts relevant to hydrous redox
melting are extremely rare: Taylor & Green (1988) melted
pyrolite in the presence of mixed H2O þ CH4 fluids and
delineated the melting curve between 1·5 and 3·5 GPa (corresponding to 50^120 km depth), but did not report analyses of melt compositions. Indeed, there are no published
melt compositions from peridotite with this volatile
NUMBERS 7 & 8
JULY & AUGUST 2011
mixture available to date, so that melt compositions can
only be estimated from indirect information.
The geodynamically reasonable scenarios for HRM are
(1) close to the base of continental, particularly cratonic,
lithosphere owing to the proximity of the solidus to the ambient geotherm, and (2) interaction of reduced mantle
with more oxidized, recycled blocks within the asthenosphere (Green et al., 1987). In the second of these, a higher
geothermal gradient applies, and melting is achieved by
suppression of the solidus curve owing to the increase in
water activity by the HRM mechanism (Fig. 7).
Melt compositions can be estimated from the solubilities
of available volatile species and their effects on the structure of silicate melts taken from studies of melt compositions that do not correspond directly to melts of garnet or
spinel peridotite. C^O^H fluid compositions at upper
mantle pressures and fO2 conditions corresponding to the
HRM mechanism are dominated by CH4 and H2O
(Fig. 6). Water is known to be very soluble at high pressures
and to depolymerize the aluminosilicate network of silicate
melts by breaking bridging oxygen bonds, causing a reduction in the average size of aluminosilicate species
(Burnham, 1979; Stolper, 1982; Mysen et al., 1982). The
petrological corollary of this effect is the expansion of
the stability fields of minerals with higher ratios of
network-modifying cations to network-forming cations,
meaning that olivine is stabilized relative to pyroxenes
during melting (Kushiro, 1975; Gupta & Green, 1988).
Partial melts of peridotite with H2O alone may be alkaline
(nephelinite or basanite; Green, 1973; Millhollen et al.,
1974) because of the presence of low-degree melts over a
considerable temperature range before major melting
occurs (Green & Falloon, 1998).
The few studies of the solubility mechanisms of methane
in silicate melts indicate that only 0·2 wt % carbon dissolves in reduced form (Taylor & Green, 1987) and this
may increase to a maximum of 0·5% by dissolution of
CH4 groups in less polymerized melts such as nephelinites
with NBO/Tof 0·9 (Mysen, 1987; Mysen et al., 2009). This
major difference in solubilities of H2O and CH4 means
that the position of the solidus and the melt compositions
are determined principally by H2O, with CH4 having
little more than a dilution effect on the water activity
(Taylor & Green, 1988; Fig. 7). Melt compositions produced
by HRM will, therefore, be similar to melts of peridotite
with small amounts of H2O; these may include alkaline
compositions, but the lack of carbonate ions in the melt
will prevent a strong degree of undersaturation in silica.
A similar effect has been shown in lamproitic systems
with mixed H2O and CH4 volatiles present, for which no
silica-undersaturated melts are produced at all at depths
up to at least 200 km (Foley, 1993).
Melt compositions produced at higher oxygen fugacities
by the CRM mechanism will differ greatly from those
1376
FOLEY
REDOX MELTING IN MANTLE
produced by HRM. This is due to the presence of abundant CO2 and the contrasting effect it has on the structure
of silicate melts, which is to cause polymerization of the
aluminosilicate network by forming complexes with
network-modifying cations (Mysen et al., 1982). This expands the stability field of pyroxenes and garnet relative
to olivine in the melting peridotite, meaning that the contribution of olivine components to the melt is greater so
that melts have lower SiO2 contents than their dry or hydrous counterparts at the same pressure. The solubility of
carbon in the form of carbonate at pressures 420 kbar is
high, of the order of 20 wt % compared with 50·5 wt %
at 51kbar (Brey, 1976; Wyllie & Huang, 1976), so that its
effects on melt compositions are strong.
Experiments on peridotite with CO2 alone at upper
mantle pressures have shown initial melt fractions to be
carbonatitic with less than 10 wt % SiO2 (Sweeney, 1994;
Dasgupta et al., 2007; Brey et al., 2008). At higher melt fractions, the SiO2 contents of melts increase, but the resulting
carbonate-bearing silicate melts are still notably
SiO2-poor compared with anhydrous melts, corresponding
to melilitites or ultramafic lamprophyres. Current results
differ as to whether the transition from carbonatitic to silicate melts is abrupt (Moore & Wood, 1998; Dasgupta
et al., 2007) or continuous (Brey et al., 2008; Foley et al.,
2009; Litasov & Ohtani, 2009a). Experiments with both
H2O and CO2 are most relevant to the CRM mechanism,
as these correspond to the stable volatile mixture (Fig. 6)
and together suppress the melting point more than CO2
alone (Fig. 10). First results by Wallace & Green (1988) at
3 GPa emphasized the high Na2O contents of initial carbonatitic melts, whereas later experiments have shown
that K2O can also be enriched in carbonatitic melts
(Thibault et al., 1992; Sweeney, 1994; Foley et al., 2009;
Ghosh et al., 2009). A potentially important effect of
increasing pressure is that initial melts may become less
carbonatitic towards 50^60 kbar (Foley et al., 2009).
The difference in melt types produced by the two redox
melting mechanisms in comparison with melts of dry peridotite is best illustrated by considering the melt compositions expected at similar pressure conditions to those
commonly resulting from decompression melting of dry
peridotite. For example, beneath ocean islands, melting
will be concentrated beneath the oceanic lithosphere at
depths of 70^100 km, where initial melts will be slightly
nepheline-normative and higher degree of melts will be picritic (Jaques & Green, 1980). Melting in reduced,
H2O þ CH4-bearing conditions at similar depths will
occur at lower temperatures owing to the influence of
H2O, and melt compositions will be mildly alkaline.
Here, the influence of water on depressing the melting
temperature is more important than its depolymerizing
effect on the melt structure. This is because the drop in solidus results in a wide temperature interval of low-degree
melting referred to by Green (1990) as the ‘incipient melting regime’. Within this temperature interval, melts
remain alkaline, and would become SiO2-richer than dry
melts only at higher temperatures within the ‘major melting regime’. At the same depths, melting by CRM will
occur at even lower temperatures than HRM, and melts
will be markedly lower in SiO2, resembling melilitites or
much lower SiO2 melts resembling ultramafic lamprophyres that are probably not seen at the surface in an
unreacted state.
D I S C U S S I O N A N D A P P L I C AT I O N S
Partial melting in the upper mantle as a result of one of the
redox melting mechanisms can occur wherever redox
state varies greatly over relatively small distances, and
may operate in a greater variety of conditions and geodynamic settings than the original definition of redox melting
intended (Taylor, 1985; Taylor & Green, 1987). This was restricted to HRM at low oxygen fugacities close to the IW
buffer (Figs 3 and 5), whereas the CRM mechanism
may operate in different geodynamic situations, and
partly in the same settings at a later stage of development
(Foley, 2008). Figures 2^5 demonstrate that variation of
fO2 in mantle rocks is characteristic of all tectonic settings,
and that the redox contrast of blocks derived from differing
tectonic settings (e.g. oxidized subducted blocks in a
reduced deep mantle environment) will in many cases
be strong.
Here, three situations are considered in which redox
melting may be or have been most important through the
evolution of the Earth; (1) the rejuvenation of cratonic
lithosphere by thinning and erosion as a precursor to continental rifting; (2) the interaction of recycled lithospheric
blocks from subduction or delamination processes with ambient mantle at deeper levels; (3) redox melting in reduced
conditions in the upper mantle of the Hadean to
Archaean Earth. This is not an exhaustive list of possibilities, as the juxtaposition of rock types with contrasting
lithologies and oxidation states may be common in the
convecting upper mantle (Alle'gre & Turcotte, 1986; Foley
et al., 2001; Sobolev et al., 2007).
Rejuvenation of cratons and rifts through
cratons
Cratons are typified by the long-term stability of the crust
and lithospheric mantle beneath it; however, evidence is
mounting that there may be more magmatic activity at
the base of cratons than has generally been assumed. This
evidence comes from the proven removal of the lithosphere
beneath the North China craton (Xu, 2001; Gao et al.,
2004), from young melt infiltration events in peridotite
xenoliths (Konzett et al., 2000; Simon et al., 2007; Rehfeldt
et al., 2008), geochemical investigations of inclusions in
diamonds (Richardson et al., 1993; Shimizu & Sobolev,
1377
JOURNAL OF PETROLOGY
VOLUME 52
NUMBERS 7 & 8
JULY & AUGUST 2011
Fig. 11. Redox melting during the rejuvenation and breakup of cratons may be due to operation of both HRM and CRM mechanisms consecutively. During the initial stages of breakup of the North Atlantic craton to produce the Labrador Sea, melting in reducing conditions produced
lamproites at 1400^1200 Ma. Following the erosion of the base of the craton root, impingement of the more oxidized upwelling asthenosphere
caused redox melting owing to depression of the solidus from reduced conditions (CH4 þ H2O; dashed line, right panel) to oxidized conditions
(CO2 þ H2O; continuous line, right panel). Incipient melting following the initial oxidation event first causes veining of the overlying mantle
wedge that is still reduced, and with further development of the rift base, the solidus depression moves as a wave upwards and re-melts the recently enriched wedge. The resulting melts derived by CRM are ultramafic lamprophyres, emplaced mostly during two episodes at around
610 and 55 Ma. Further development of the rift resulted in melililtic to nephelinitic magmatism during the Mesozoic, derived from depths of
100^120 km, with nearby reactivation of ultramafic lamprophyre melting as a result of the steeply sloping sides of the Archaean cratonic lithosphere, which still allows CRM at high pressures (left panel). G = graphite, D ¼ diamond. Diagrams combined and modified after Tappe et al.
(2006, 2007).
1995; Jacob et al., 2000), and from rift magmatism. The
most extreme form of reactivation of cratonic lithosphere
is manifested in successful rifts through cratons that proceed to the production of oceanic lithosphere, as in the
Labrador Sea rift between Canada and Greenland. In
other areas, similar processes can be seen in peridotite
xenoliths and volcanic rocks of ‘unsuccessful’ and current
rifts at the margins of cratonic blocks, as in Antarctica
(Foley et al., 2006) and Congo^Tanzania (Link et al., 2010).
The Labrador Sea area provides evidence for the series
of igneous rocks produced during the development of cratonic rifts and has been described in detail by Tappe et al.
(2006, 2007, 2008). It results from several episodes prior to
and including the rifting event that eventually produced
new oceanic crust: (1) a 1400^1200 Ma event produced
lamproitic magmas from great depths; (2) a 610^570 Ma
event produced mostly carbonate-rich ultramafic lamprophyres; (3) a Mesozoic event produced nephelinitic and
similar rocks (Tappe et al., 2007, 2008). Lamproites originate at the base of the lithosphere at craton margins and
their chemistry can be explained only by melting in
reduced conditions, probably triggered by HRM (Foley,
1989a, 1989b). The lack of CO2 or carbonate under
these conditions results in melts that are not silicaundersaturated, despite being silica-poor, and thus
lamproitic rather than belonging to the melilitite^
carbonatite series (Foley, 1993). Lamproitic melts produced
by HRM may be common at the base of the lithosphere
and may act as regular agents of enrichment in potassium
and other incompatible elements; however, they are unlikely to reach the surface and form igneous rocks in most
cases. This explains the tendency of lamproites to occur
around the margins of cratons and not in their centres
(Janse & Sheahan, 1995).
The second event characterized by ultramafic lamprophyres can be ascribed to the effects of the CRM mechanism (Fig. 11; Tappe et al., 2006). Thinning of the lower
cratonic lithosphere reactivates and oxidizes diamond that
was deposited during earlier, largely local, HRM events
that produced the lamproitic melts noted above by the release of H2O from the oxidation of CH4. The carbon from
the methane was left in the residue as diamond formed,
particularly in the time period 1·4^1·2 Ga. In the later episode around 600 Ma, this diamond was oxidized by the
juxtaposition of upwelling asthenospheric mantle and the
thinning continental lithosphere; the resulting oxidized
carbon caused a drop in the solidus temperature (Fig. 10).
This CRM mechanism is common in continental
rifts around the world, resulting in the frequent association of continental rifting with carbonate-rich melts.
1378
FOLEY
REDOX MELTING IN MANTLE
Ultramafic lamprophyres or melilitite^carbonatite series
rocks are also found in the Lambert^Amery rift in eastern
Antarctica (Foley et al., 2002a), and in off-craton continental rifts such the Oslo and Rhine rifts (Keller, 1981;
Andersen, 1988; Schleicher et al., 1990; Dahlgren, 1994).
In the Labrador Sea area, isotope studies have indicated
that melting involved phlogopite-bearing vein assemblages
originating from earlier magmatic episodes beneath the
later rift, but that a major component was derived from
melting of peridotite at 120^160 km depth in the presence
of both H2O and CO2 (Tappe et al., 2008). The age of the
mica-bearing vein assemblages can be constrained by
Rb^Sr isotopes, and varies from several hundred million
years in the Labrador Sea and eastern Antarctica (Foley
et al., 2002a, 2006) to just 3 Ma in the western branch of
the east African Rift (Link et al., 2010).
The third, Mesozoic, melting event beneath the
Labrador Sea occurs at much shallower depths
(90^100 km), producing nephelinites, although the steep
lithosphere^asthenosphere boundary dictates that melting
at deep levels occurs quasi-contemporaneously at geographically close locations (Tappe et al., 2008). In the western branch of the East African Rift, the kamafugite series
(potassic melilitites and kalsilitites) are the equivalent of
the ultramafic lamprophyres (Rosenthal et al., 2009);
alkali basalts and nephelinites occur further south in the
Virunga and South Kivu provinces (Rogers et al., 1998;
Furman & Graham, 1999; Platz et al., 2004; Chakrabarti
et al., 2009) away from the junction with the craton (Link
et al., 2010).
The extent of applicability of the HRM mechanism to
the lower cratonic lithosphere will depend on the mechanism of formation and oxidation state of cratonic mantle,
plus the extent to which it has been infiltrated by later
melts. The original oxidation state is now largely overprinted by the pressure effect, which gives rise to the general slope in Fig. 5; however, the uncertainty on the
vertical axis in Fig. 5 reflects a mixture of the memory of
the original oxidation state and the metasomatic overprinting. The combined effect of multiple melt infiltration
episodes (Simon et al., 2007; Rehfeldt et al., 2008) has probably raised the original fO2 by more than one full log
unit for many parts of the cratonic lithosphere. The formation mechanism of cratonic mantle lithosphere is still
debated, with low-pressure near-surface subduction competing with a high-pressure plume origin to explain the
geochemical depletion in terms of high degrees of melting
(Lee, 2006). Both are likely to result in low oxygen fugacities; the plume hypothesis invokes material from greater
depths where the oxygen fugacity may even be low
enough to result in metal saturation (Rohrbach et al.,
2007), whereas the assembly of accreted arc-related oceanic
lithosphere implies formation in an Archaean oceanic
environment that was probably more reduced than today.
In both cases, an important point is that melting close to
the IW buffer causes reduction and drying of the residue
by melting, because water is transported away in the melt.
The fO2 of the melt will be close to the water maximum
(Fig. 6), whereas the residue will be located at the underside of the carbon saturation curve in Figs 6 and 8. This
will be 2^5 log units below that of the melt, with greater
fO2 fractionation at higher pressures (Fig. 8). Partial melting is thus an important mechanism leading to diversification of fO2 in mantle rocks (Kadik, 1988, 1997).
Metasomatic overprinting will have occured periodically throughout the age of the lithosphere, often resulting
in redox reactions during crystallization of the metasomatic melts and fluids, as shown by diamonds of different
ages (Richardson et al., 1993; Shimizu & Sobolev, 1995;
Jacob et al., 2000). X-ray tomographic studies of
diamond-bearing mantle rocks have shown that the diamonds are often spatially restricted to zones of fluid or
melt infiltration (Keller et al., 1999), and that the diamonds
are associated with mineral assemblages such as cohenite,
moissanite and carbonates that reflect several log units difference in fO2 (Eggler & Lorand, 1984; Ulmer et al., 1998;
Jacob et al., 2004; Pal’yanov et al., 2005). The large variety
of ages of the infiltration assemblages increases the likelihood of redox contrasts and thus of redox melting as time
progresses.
In summary, the probable formation of cratonic lithosphere by heterogeneous lateral accretion followed by numerous melt infiltration events over its long history has
probably resulted in a region of the lithosphere with
mixed redox states susceptible to later redox melting reactions. The HRM mechanism at more reducing conditions
will tend to be more important in the lower parts of the
cratonic lithosphere, and results in silica-saturated, potassic melts. In contrast, the more oxidized CRM mechanism
will be concentrated in areas of cratonic lithospheric thinning, at the edges of cratons, or in continental regions
away from cratons; the resulting melts will belong to the
carbonatite^melilitite series, with aillikitic ultramafic
lamprophyres as the deepest-derived expression.
Melting in recycled blocks and their input
into magmas
Subduction causes juxtaposition of blocks of varying redox
state and is also a major cause of the polarization of redox
states in the mantle. The interaction of oxidized subducted
blocks with reduced CH4 þ H2O fluids emanating from
the deep mantle was considered as one of the principal
mechanisms for redox melting (HRM) by Green et al.
(1987, 1990). We can now take a closer look at redox processes during subduction, both as a function of spatial
scale and rock type, and as a function of time through
Earth history. The change in oxidation state of subducting
material is probably the major controlling factor in the
gradual oxidation of internal processes that pre-empted
1379
JOURNAL OF PETROLOGY
VOLUME 52
NUMBERS 7 & 8
JULY & AUGUST 2011
Fig. 12. Schematic illustration showing various stages in the formation, alteration and subduction of oceanic crust and lithosphere relevant to
the fixation of redox conditions and later redox melting. Colours refer to oxidation state as shown in the inset and not to the origin of the
rocks. (1) MORB oxidation state is approximately FMQ ^ 1, and hydrothermal alteration results in serpentine and magnetite formation.
(2) The lower lithosphere is slightly more reduced than basalts at the surface, and is impregnated by more water-rich, low-degree melts that
form pyroxenite veins. (3) Part of the water is released during subduction, whereas carbonates introduced at the surface may remain in the
solid residue (yellow blocks). (4) Subducted sedimentary material provides Na and K that may help to depress the solidus temperature at a
later stage. (5) The subducted lithosphere at depths of 150^350 km contains rocks with a mixture of redox states, but is generally more oxidized
than the surrounding mantle (IW to IW þ1) owing to insufficient time for adjustment to the lower fO2 typical of high-pressure conditions
(Ballhaus & Frost, 1994). Redox melting is probably rare in the subduction zone environment because of the low thermal gradients. (6)
Recycled blocks in the mantle at 250^400 km, derived either by peeling off from the lower oceanic lithosphere or from deep recycling of ocean
crust, are mostly at fO2 2^3 log units above the ambient mantle peridotite in the region of HRM, but generally not oxidized enough for
widespread CRM. Water release by oxidation of methane in the lower blocks leads to melting by HRM at higher levels (7). Many ‘plumes’
may be upward movement of small-degree melts and fluids that cause major melting beneath the lithosphere under ocean islands (Green
et al., 1987, 2001), and are also a contributor to basaltic melts at mid-ocean ridges. Redox melting may be concentrated in pyroxenitic material
in the recycled blocks.
the great oxidation event at the Earth’s surface in the early
Proterozoic.
Figure 12 summarizes some of the important processes
related to the formation and destruction of oceanic crust
that have an effect on possible later redox melting events.
It has been hypothesized that redox melting beneath
ocean islands as a result of the interaction of oxidized
blocks of subducted lithosphere with reduced CH4-rich
fluids from deeper in the mantle results in melting caused
by an increase in water activity (the HRM mechanism)
and that such melts contribute to ocean island magmatism
(Green et al., 1987; Green, 1990). This mechanism may be
more important than thermal plumes in generating melts
beneath oceanic islands, as implied by the comparable
liquidus temperatures of Hawaiian and mid-ocean ridge
picritic melts and the higher water contents of oceanic
island melts, so that ‘hotspots’ may more correctly be
termed ‘wetspots’ (Green et al., 2001). We can now consider
the subduction process in more detail, as our understanding both of subduction and of the redox profile of the
deeper mantle has developed since these earlier contributions (Green et al., 1987, 1990; Green, 1990). Stages of ocean
1380
FOLEY
REDOX MELTING IN MANTLE
crust formation, alteration, subduction and recycling that
contribute to the heterogeneity of redox states are illustrated in Fig. 12.
The oxidation state of the melts forming the basaltic
crust at mid-ocean ridges is constrained by oxybarometry
of peridotites (Bryndzia et al., 1989) and the Fe3þ/Fe2þ
ratios of basalts (Be¤zos & Humler, 2005) to be close to
FMQ 1, bearing in mind that rapid late-stage oxidation
occurs after chilling of pillow rims (Christie et al., 1986).
This is changed locally before entry into the subduction
zone by hydrothermal alteration and sedimentation. The
major effect of hydrothermal alteration is serpentinization
of basalts and ultramafic rocks and the growth of other hydrous minerals such as chlorite and hydroxides in the basalts (Fig. 12, point 1). The serpentinization of olivine helps
to polarize oxidation states on a very local scale within
the crust, as it leads to the formation of magnetite coexisting with serpentine, which can release hydrogen at a later
stage (Sleep et al., 2004). At the crust formation stage, the
newly formed lower lithosphere is slightly more reduced
than the FMQ 1 typical of the crust, and this is characteristic of the lithosphere that later becomes involved in
subduction (Fig. 12, point 2). Furthermore, incomplete extraction of low-degree melts leads to impregnation of the
lithosphere with solidified melts, which take the form of
pyroxenite veins (Foley et al., 2001). Water is an incompatible component, so these melts are enriched in water and
therefore slightly more oxidized than the residue.
Following formation and associated serpentinization of
the ocean crust at mid-ocean ridges, a fractionation of
fO2 of at least 2 log units between parts of the crust has already probably taken place.
During subduction, the metamorphism of the variably
altered basaltic crust and its veneer of sediment leads to release of water from a series of hydrous phases (Fig. 12,
point 3), whereas a significant proportion of the H2O can
be taken to depths of 200 km (Ulmer & Trommsdorff,
1995; Schmidt & Poli, 1998). At these depths, material leaving the subducting slab will be either a water-rich fluid
with a high silicate solute content (Kessel et al., 2005) or a
melt, particularly of clastic sediments (Parkinson &
Arculus, 1999; Hermann & Spandler, 2008; Plank et al.,
2009). The residue that continues its way into the deeper
mantle will in most cases be more reduced. In contrast,
carbonates have been shown to be stable above the melting
point of basaltic material in the form of carbonated eclogites (Yaxley & Green, 1994); thus CO2 may persist in
recycled subducted blocks where it coexists on a geodynamically small spatial scale with the more reduced residues
of fluid and melt loss.
Degassing of deeper parts of the serpentinized slab will
further polarize oxidation states in the residue, as the C^
O^H components of the fluids are dominated by H2O
and the width of the H2O-dominated fluid region widens
to higher pressures (Woermann & Rosenhauer, 1985;
Fig. 8). This leads to stronger reduction of reduced rocks
and oxidation of those more oxidized than the ‘water-maximum’ (Fig. 8) than occurs at lower pressures. Evidence
for this polarization of oxidation states in subduction
zones is found in the form of metals and methane-bearing
fluid inclusions in xenoliths (Ishimaru et al., 2009; Song
et al., 2009), whereas the general oxidation state of
supra-subduction zone mantle appears to be more oxidized
(Figs 2 and 4; Parkinson et al., 2003; Bryant et al., 2007).
The upper portions of the crust may include subducted
sedimentary material in places (Fig. 12, point 4); this is important as Na and K further reduce the melting point of
peridotite under oxidized conditions (Foley et al., 2009;
Ghosh et al., 2009; Litasov & Ohtani, 2009a) where
mixing is intimate (Prelevic et al., 2008).
The aggregate effect of these processes is that subducted
slabs can be expected to be very heterogeneous in oxidation state (5^6 log units fO2) when they reach depths between 150 and 200 km, where the ambient mantle should
be around 3 log units below FMQ (Fig. 5). The oxidation
state of large parts of the subducted slab will be higher
than that of the ambient mantle because of the persistence
of carbonates, and the adjustment of the oxygen fugacity
of these near-surface materials to higher pressures will be
too slow to match the rate of subduction, which will take
oceanic lithosphere to depths of 100^300 km in less than
10 Myr at normal spreading rates (Fig. 12, point 5).
Despite the resulting mixture of oxidation states in the
deeper regions of subduction zones, redox melting is unlikely to be important except in subducted sedimentary
material (Hermann & Spandler, 2008; Plank et al., 2009),
as low thermal gradients will not allow the melting point
of eclogite or peridotite to be reached, even in volatile-rich
conditions. Instead, much of the mobility of elements may
be controlled by fluids in modern subduction zones
(Ishikawa & Nakamura, 1994; Poli & Schmidt, 1995;
Stalder et al., 1998) without the widespread melting of
mafic rocks in the form of garnet amphibolite or eclogite
that may have been prevalent during the late Archaean
(Rapp et al., 1991; Martin, 1999; Foley et al., 2002b). There
will, nevertheless, be many fluid redox reactions within
and above the subducting slab, such as the formation of
diamonds from fluids that come into contact with eclogite
blocks of differing oxidation states (Bulanova, 1995;
Stachel & Harris, 2008).
The main location for the redox melting of recycled material will be in larger blocks at depths of 300^600 km
(Green et al., 1987). These may be partly assimilated
blocks of the oceanic mantle lithosphere whose compositional difference from the convecting mantle is relatively
small, so that they become detached from the underside of
the subducting lithosphere when the thermal contrast lessens at depths of 300^500 km (Ringwood, 1982), thus
1381
JOURNAL OF PETROLOGY
VOLUME 52
gradually thinning the subducting slab from below (Fig. 12,
point 5). These blocks will contain strands of garnet pyroxenite derived from nephelinitic to basaltic melts that
failed to reach the surface beneath the mid-ocean ridge
(Foley et al., 2001; Fig. 12, point 2). Crustal blocks consisting
of mostly ultramafic rocks and eclogites may be transported laterally between 300 and 500 km depth or be returned to these depths after deeper recycling and
transformation through higher pressure hydrous (or nominally anhydrous) phases (Thompson, 1992; Kohlstedt
et al., 1996). However deeply the blocks are recycled, the
ambient conditions of the convecting mantle will be at
fO2 of FMQ 4 to FMQ 6 at depths of 200^300 km
(6^8 GPa pressure; Fig. 5). The original lithosphere oxidation state will have varied from FMQ 1 to FMQ 2 at
its formation, although the recycled blocks will exhibit a
larger variation owing to hydrothermal and subduction
processes discussed above. These will vary from FMQ 0
in carbonate-bearing assemblages to fO2 as low as
FMQ^6 in ultramafic blocks that lost water-rich fluids or
melts at high pressures. This means that most redox melting events related to these blocks will involve the HRM
mechanism (Fig. 12, point 6; see also Figs 6 and 9); carbonates, although present in eclogites, will be rare and the
CRM mechanism will not be important. This conclusion
contrasts with the suggestion that incipient melting below
mid-ocean ridges will be largely CO2-controlled
(Dasgupta et al., 2007). Reduced carbon, however, will be
common in deeply subducted material, partly as a result
of the polarization of oxidation states noted above that
concentrates H and C in the more reduced products,
whereas oxide minerals and carbonates are concentrated
in the oxidized products. This dictates that recycled blocks
are a significant source of reduced carbon and hydrogen
(Poli et al., 2009), resulting in the release of CH4 and H2
that lead to H2O release and thus potential melt production by HRM. The vertical arrangement of several blocks
may result in H2O release from the lower ones (Fig. 12,
point 6) and redox melting in the upper ones (Fig. 12,
point 7).
A principal unknown factor in the assessment of redox
melting in recycled blocks is the melting points of pyroxenite under volatile-present conditions. There are far more
experimental data available on the melting point of peridotite with CO2 and H2O þ CO2 mixtures (Wallace &
Green, 1988; Canil & Scarfe, 1990; Dasgupta et al., 2007;
Brey et al., 2008; Foley et al., 2009; Ghosh et al., 2009;
Litasov & Ohtani, 2009b) than on eclogite and pyroxenite
(Yaxley & Green, 1994; Hammouda, 2003; Yaxley & Brey,
2004; Dasgupta et al., 2004), for which a mixed
H2O þ CO2 volatile melting point has not yet been
determined.
The melting points of both peridotite and eclogite^
pyroxenite in reducing conditions are very poorly known.
NUMBERS 7 & 8
JULY & AUGUST 2011
The incompatibility of Na and K, possibly resulting from
sediment input into subduction zones, and their incorporation into carbonate-rich melts cause further reduction of
the solidus temperature in comparison with simple system
experimental determinations (Foley et al., 2009; Ghosh
et al., 2009; Litasov & Ohtani, 2009a).
The conclusions reached in this section presume subduction as it occurs on the modern Earth. However, before
the advent of an oxygen-rich atmosphere, which followed
the gradual oxidation of near-surface geological reservoirs,
the thickness and composition of the oceanic crust, the
subduction process itself (if present at all), and the oxidation state of various crust and lithosphere components
may have differed greatly.
Redox melting in the Archaean and
Hadean mantle
There is considerable debate about whether the oxidation
state of the Earth’s upper mantle has changed through
time; some scenarios invoke either rapid or gradual
changes during the Hadean and Archaean based on interpretations of mantle rocks, core formation processes, or of
the development of atmospheric and surface conditions
(Kasting et al., 1993; Catling et al., 2001; Kump et al., 2001;
Kadik, 2003; Galimov, 2005). In contrast, recent summaries derived from indicators of the oxidation state in the
source of volcanic rocks (V contents in olivine, Cr contents
and V/Sc ratios of bulk-rocks) claim that there has been
no appreciable change in the oxidation state of the upper
mantle since 3·5 Ga (Canil, 1997; Delano, 2001; Li & Lee,
2004; Berry et al., 2008). However, these conclusions are
based on continental rocks and so only probe the continental lithosphere. It was argued above that the oceanic
mantle today is more reduced than the continental mantle
and that there is a general trend towards more oxidized
continental mantle lithosphere later in the Earth’s history
(Fig. 5). Kasting et al. (1993) noted that a constant mantle
oxidation state implies an oxygen-bearing Archaean atmosphere, which contrasts with evidence for a major increase in atmospheric oxygen levels in the early
Proterozoic (Farquhar et al., 2000; Pavlov & Kasting,
2002; Bekker et al., 2004). Furthermore, the bias from samples of continental Archaean volcanic rocks is strong, considering that there was a much smaller volume of
continental crust and lithosphere before the major period
of crustal growth in the late Archaean and Proterozoic
(Taylor & McLennan, 1995; Hawkesworth & Kemp,
2006). The Archaean mantle sample comes mostly from
garnet peridotites that have been part of the continental
lithosphere since the mid- to late Archaean (Pearson,
1999), and range in fO2 mostly between 1·5 and 4·5 log
units below the FMQ buffer (Figs 2 and 5). Cratonic peridotites show, on average, lower FMQ values than
non-cratonic ones (Fig. 5), although their oxidation states
may have been raised by about 1 log unit by widespread
1382
FOLEY
REDOX MELTING IN MANTLE
later metasomatism and thus represent only maximum
values. Melt infiltration in orogenic massifs has produced
an average oxidation of 0·86 FMQ units (Fig. 4); in
these the textural evidence for metasomatism (Woodland
et al., 1996) is more subtle than in cratonic peridotites,
which appear to have been largely dunite and harzburgite
in their original pre-metasomatic state (Simon et al., 2007;
Rehfeldt et al., 2008).
Correcting for oxidation through more than 1 log unit
fO2 as a result of these strong, multiple metasomatic
events, it appears acceptable to conclude that the
Archaean asthenospheric mantle was more reduced than
the current sub-ridge mantle, equivalent to approximately
FMQ 1·5 to FMQ 2·0 at 35^40 km depth, and that the
record of Archaean basalts is biased towards continental
regions. Furthermore, the ages of cratonic peridotites and
eclogites that give the evidence for more reducing
Archaean conditions are almost entirely from the late
Archaean (2·9^2·5 Ga), raising the possibility that the
mantle had progressively oxidized from even more reduced
conditions during the Hadean and early Archaean. I
argue here that the upper mantle was more reduced than
on the modern Earth and that it became oxidized at a
later stage by recycling of surface material that could
occur only after the gradual oxidation of near-surface
layers. A more reduced upper mantle for the Earth is in
keeping with best estimates of the oxidation states of
FMQ 6 for the Moon, and FMQ 3 to FMQ 5 for
Mars (Wadhwa, 2008), both considerably lower than continental mantle samples from the Earth. These were
‘frozen in’ by the early cessation of tectonic movements on
both the Moon and Mars (Stevenson, 2001; Boyet &
Carlson, 2007). The slightly higher value for the Earth
may be real, owing to its larger size, which caused a
greater concentration of Fe2O3 in the mantle following dissociation of FeO into Fe8 and Fe2O3 at the higher pressures
of core formation (Frost et al., 2004). The gradual oxidation
of near-surface layers through the Archaean and early
Proterozoic is well documented in the appearance of oxidized sediments and decrease in banded iron formations
and uraninite deposits (Schidlowski, 1981; Canfield, 2005).
A corollary of this reduced early mantle model is that the
subduction process itself, if it happened in a similar
manner to today, was not an oxidized process until at
least the late Archaean, meaning that processes within the
oxidized realm of Fig. 9 were volumetrically insignificant.
The implications for possible volatile-controlled melting
are as follows. The rarity of continental lithosphere and its
generally lower oxidation state, coupled with the reduced
state of subduction, make the occurrence of melting in the
presence of carbonates, and thus the operation of the
CRM mechanism, unlikely during the Archaean
and Hadean. This is backed up by the fact that no carbonatite magmatism earlier than 2·7^2·6 Ga is known
(Cavell et al., 1992; Tilton & Bell, 1994; Villeneuve & Relf,
1998). In contrast, the more reduced upper mantle and
also its probable higher volatile contents, particularly in
the Hadean when degassing was still more vigorous,
allowed the widespread action of the HRM mechanism in
the convecting mantle.
Melting that occurs at near-surface conditions in a
reduced mantle with fO2 around FMQ 4 to FMQ 5
will cause a change of the oxidation state because of the
differential solubilities of various volatile components.
Water behaves as an incompatible component during melting, and at depths of 40^60 km (assuming a mid-ocean
ridge scenario as in the Phanerozoic) several weight per
cent of H2O can be dissolved in melts. In the case of
carbon, more than 20 wt % CO2 would be soluble in silicate melts at460 km depth (Wyllie & Huang,1976), whereas the solubility of reduced carbon is limited to 0·2 wt %,
with a maximum of 0·5 wt %, which applies only if the
melts are strongly depolymerized (Taylor et al., 1987;
Mysen et al., 2009). This has a reducing effect on the residue, forcing carbon saturation, after which water-rich
melts at approximately FMQ 2 will coexist with
carbon-bearing residues at around FMQ 4 (Fig. 6). This
fractionation effect between melt and residue results in a
basaltic crust that is more oxidized than the residue and
thus represents the first stage of the gradual oxidation of
near-surface layers. The fractionation is greater at higher
pressures because of the widening of the water maximum
with increasing pressure (Fig. 8). CO2 does not exist at
these fO2 conditions (Fig. 6) and water activities are generally lower than at intermediate fO2 as a result of being
diluted by methane. This means that water-bearing solidi
with lower water activities will be more relevant for the
Hadean and Archaean, and these lie at higher temperatures than for more oxidized conditions (Figs 7 and 10). A
further implication of this is that higher temperatures in
the Archaean mantle do not necessarily mean much more
extensive melting than later in Earth history, as is usually
concluded from assessments that compare the melting behaviour of dry peridotite at contrasting temperatures
(McKenzie & Bickle, 1988; Bickle et al., 1994). However,
once the melting point is attained, there will be a more
limited temperature interval of incipient melting before
major melting occurs.
The ocean crust resulting from this melting behaviour
was thicker and more MgO-rich than later in Earth history, although the estimates of 30^45 km that are sometimes suggested (McKenzie & Bickle, 1988; Kent et al.,
1996) are probably excessive because of overestimation of
mantle temperatures and underestimation of the effects of
changing volatile concentrations and speciation through
time. Following internal magmatic differentiation of picritic parental melts (Jacob & Foley, 1999; Foley et al., 2003), a
large proportion of the lower crust would consist of
1383
JOURNAL OF PETROLOGY
VOLUME 52
pyroxenites and wehrlites, and these would be preferentially returned to the mantle (Foley et al., 2003).
The return flow of lithospheric blocks to the mantle by
subduction or delamination would have been dominated
by olivine-bearing pyroxenites with reduced oxidation
states, rather than by oxidized and more silica-rich eclogites as in later stages of Earth history. The involvement of
oxidized blocks in recycling had to await the oxidation of
the Earth’s surface in the late Archaean.
The more reduced fO2 of the upper mantle during the
Hadean and Archaean has important implications for the
compositional development of the atmosphere. High atmospheric CO2 concentrations have been invoked in
many scenarios (e.g. Owen et al., 1979), but are difficult to
reconcile with palaeosol data, which discount CO2 levels
of several per cent (Rye et al., 1995; Sheldon, 2006). A
more recent suggestion involves a higher pressure atmosphere dominated by nitrogen (Goldblatt et al., 2009). The
key to the degassing of volcanic gas species is the content
of the various volatile species in the melts during melting,
which depends on their presence in the source and their
solubilities at high pressures where the melts form. The
solubilities of all volatile components increase with increasing pressure, meaning that degassing occurs by release of
dissolved volatiles as melts move towards the surface.
Given the evidence for progressive oxidation of the mantle
throughout the Hadean and Archaean, early rapid degassing of CO2 is not realistic, because melts would have contained little carbon as a result of the low solubility of CH4
(Taylor et al., 1987; Mysen et al., 2009). The solubility of
CO2 in melts increases sharply at pressures corresponding
to about 60 km (Eggler, 1976; Wyllie & Huang, 1976), meaning that abundant volcanic degassing of CO2 could occur
only after oxidation of the mantle to at least FMQ 1 progressed to this depth. This may explain the higher calculated pCO2 in equilibrium with palaeosols in the early
Proterozoic (Sheldon, 2006). Until this time, degassing species were C-poor, but rich in H and N. Available data indicate that the solubility of nitrogen is very low at relatively
high and intermediate oxygen fugacities where N2 is
stable, but increases by at least a factor of five at low
oxygen fugacities that are now typical of the deep mantle
and may have prevailed in the upper mantle during the
Hadean and Archaean. The reason lies in the different solution mechanisms; molecular N2 dissolves only in spaces
between silicate network components, whereas reduced nitrogen dissolves as NH2 units in the silicate structure itself
(Libourel et al., 2003; Mysen et al., 2008).
NUMBERS 7 & 8
JULY & AUGUST 2011
particularly in the modern mantle at depths of
200^300 km, where more oxidized recycled material
meets peridotite where ambient fO2 conditions are
around FMQ 3 to FMQ 5. Recycled blocks may contain material with extremely heterogeneous oxidation
states owing to hydrothermal alteration at mid-ocean
ridges, sediment input at subduction zones and dehydration reactions during subduction that result in a mixture
of reduced and oxidized carbon. The involvement of
recycled materials is especially important in ocean island
magmatism (Sobolev et al., 2007) because of the interplay
of redox melting (HRM) and the presence of pyroxenitic
former ocean crust. This results in the release of H2O at
200^300 km depth and the promotion of melting, contributing considerably to melting in rising mantle plumes.
In Hadean and Archaean times, the uppermost mantle
was in a more reduced state than today, and the less
degassed state means that HRM probably operated in the
uppermost mantle more commonly than today without
the need for recycled blocks; subduction would not have
brought more oxidized material to great depths in the
mantle because of the lack of oxidized surface conditions.
In globally more reduced conditions, the solubilities
(Taylor & Green, 1987; Mysen et al., 2008) and thus the
degassing behaviour of carbon and nitrogen were opposite
to those in modern volcanic gases. Carbon solubility was
low and so little CO2 was degassed, whereas the solubility
of nitrogen was higher, resulting in a nitrogen-rich early atmosphere. Partial melting under reduced conditions concentrates H2O in the melt, thus causing variations in fO2
between more oxidized melts and more reduced residues.
This is the principal driving mechanism for the oxidation
of the upper mantle through time.
Carbonate redox melting (CRM) was essentially inoperative until the late Archaean when a large proportion of
the continental crust was formed together with the stabilization of the continental mantle lithosphere. The CRM
mechanism involves the oxidation of solid carbon accumulated in the lithosphere over time, and is therefore commonest during the reactivation of cratons. This explains
the absence of carbonatites until 2·7 Ga, and the association of carbonatites with intraplate continental areas and
continental rifts. Furthermore, it implies that CRM is a logical consequence of earlier HRM in the same area,
which caused the enrichment in carbon as graphite or
diamond.
AC K N O W L E D G E M E N T S
CONC LUSIONS
The importance of each of the two redox melting mechanisms varies as a function of geodynamic setting and geological time. Hydrous redox melting (HRM) involves the
oxidation of CH4- and H2-bearing fluids and operates
I am grateful to Yaoling Niu for inviting me to present an
earlier version of this work at the symposium in honour of
Peter Wyllie and the 50th anniversary of the Journal of
Petrology at the Goldschmidt Conference 2009 in Davos.
My interest in redox melting processes was stimulated at
an early stage by discussions with Wayne Taylor, David
1384
FOLEY
REDOX MELTING IN MANTLE
Green, Matthias Rosenhauer and Eduard Woermann. The
current reconsideration is prompted by interdisciplinary
discussions within the Geocycles Research Centre in
Mainz. I am grateful to Bob Luth, Arno Rohrbach, Guil
Mallmann and Jo«rg Hermann for comments that improved the manuscript.
FUNDING
Interdisciplinary research on solid Earth^atmosphere
interactions is part of the Programme of the Geocycles
Research centre of the University of Mainz.
R EF ER ENC ES
Abbott, D., Burgess, L., Longhi, J. & Smith, W. H. F. (1994). An empirical thermal history of the Earth’s upper mantle. Journal of
Geophysical ResearchçSolid Earth 99, 13835^13850.
Adam, J., Green, T. H. & Day, R. A. (1992). An experimental study of
two garnet pyroxenite xenoliths from the Bullenmerri and Gnotuk
Maars of western Victoria, Australia. Contributions to Mineralogy and
Petrology 111, 505^514.
Alle'gre, C. J. & Turcotte, D. L. (1986). Implications of a twocomponent marble-cake mantle. Nature 323, 123^127.
Andersen, T. (1988). Evolution of peralkaline calcite carbonatite
magma in the Fen complex, Southeast Norway. Lithos 22, 99^112.
Arculus, R. J. (1985). Oxidation status of the mantle: past and present.
Annual Review of Earth and Planetary Sciences 13, 75^95.
Arndt, N. T. (1977). Ultrabasic magmas and high-degree melting of
the mantle. Contributions to Mineralogy and Petrology 64, 205^221.
Arndt, N. T. (1983). Role of a thin, komatiite-rich oceanic crust in the
Archean plate-tectonic process. Geology 11, 372^375.
Ballhaus, C. (1993). Redox states of lithospheric and asthenospheric
upper mantle. Contributions to Mineralogy and Petrology 114, 331^348.
Ballhaus, C. & Frost, B. R. (1994). The generation of oxidized CO2bearing basaltic melts from reduced CH4-bearing upper-mantle
sources. Geochimica et Cosmochimica Acta 58, 4931^4940.
Ballhaus, C., Berry, R. F. & Green, D. H. (1991). High pressure experimental calibration of the olivine^orthopyroxene^spinel oxygen
geobarometer: implications for the oxidation state of the upper
mantle. Contributions to Mineralogy and Petrology 107, 27^40.
Barth, M. G., Rudnick, R. L., Horn, I., McDonough, W. F.,
Spicuzza, M. J., Valley, J. W. & Haggerty, S. E. (2001).
Geochemistry of xenolithic eclogites from West Africa, Part I: A
link between low MgO eclogites and Archean crust formation.
Geochimica et Cosmochimica Acta 65, 1499^1527.
Be¤dard, J. H. (2006). A catalytic delamination-driven model for
coupled genesis of Archaean crust and sub-continental lithospheric
mantle. Geochimica et Cosmochimica Acta 70, 1188^1214.
Bekker, A., Holland, H. D., Wang, P. L., Rumble, D., Stein, H. J.,
Hannah, J. L., Coetzee, L. L. & Beukes, N. J. (2004). Dating the
rise of atmospheric oxygen. Nature 427, 117^120.
Berry, A. J., Danyushevsky, L. V., O’Neill, H. S. C., Newville, M. &
Sutton, S. R. (2008). Oxidation state of iron in komatiitic melt inclusions indicates hot Archaean mantle. Nature 455, 960^963.
P
Be¤zos, A. & Humler, E. (2005). The Fe3þ/ Fe ratios of MORB glasses
and their implications for mantle melting. Geochimica et
Cosmochimica Acta 69, 711^725.
Bickle, M., Nisbet, E. & Martin, A. (1994). Archean greenstone belts
are not oceanic crust. Journal of Geology 102, 121^138.
Boyd, F. R. & Gurney, J. J. (1986). Diamonds and the African lithosphere. Science 272, 472^477.
Boyet, M. & Carlson, R. W. (2007). A highly depleted moon or a nonmagma ocean origin for the lunar crust? Earth and Planetary Science
Letters 262, 505^516.
Brandon, A. D. & Draper, D. S. (1996). Constraints on the origin of
the oxidation state of mantle overlying subduction zones: An example from Simcoe, Washington, USA. Geochimica et Cosmochimica
Acta 60, 1739^1749.
Brey, G. (1976). CO2 solubility and solubility mechanisms in silicate
melts at high pressures. Contributions to Mineralogy and Petrology 57,
215^221.
Brey, G. P., Bulatov, V. K., Girnis, A. V. & Lahaye, Y. (2008).
Experimental melting of carbonated peridotite at 6^10 GPa.
Journal of Petrology 49, 797^821.
Bryant, J. A., Yogodzinski, G. M. & Churikova, T. G. (2007). Melt^
mantle interactions beneath the Kamchatka arc: Evidence from
ultramafic xenoliths from Shiveluch volcano. Geochemistry,
Geophysics, Geosystems 8, doi:10.1029/2006GC001443.
Bryndzia, L. T. & Wood, B. J. (1990). Oxygen thermobarometry of
abyssal spinel peridotitesçthe redox state and C^O^H volatile
composition of the Earth’s sub-oceanic upper mantle. American
Journal of Science 290, 1093^1116.
Bryndzia, L. T., Wood, B. J. & Dick, H. J. B. (1989). The oxidation state
of the Earth’s sub-oceanic mantle from oxygen thermobarometry
of abyssal spinel peridotites. Nature 341, 526^527.
Bulanova, G. P. (1995). The formation of diamond. Journal of
Geochemical Exploration 53, 1^23.
Burnham, C. W. (1979). The importance of volatile constituents. In:
Yoder, H. S. (ed.) The Evolution of the Igneous Rocks. Washington,
DC: Princeton University Press, pp. 439^483.
Canfield, D. E. (2005). The early history of atmospheric oxygen:
Homage to Robert A. Garrels. Annual Review of Earth and Planetary
Sciences 33, 1^36.
Canil, D. (1997). Vanadium partitioning and the oxidation state of
Archaean komatiite magmas. Nature 389, 842^845.
Canil, D. (2002). Vanadium in peridotites, mantle redox and tectonic
environments: Archean to present. Earth and Planetary Science Letters
195, 75^90.
Canil, D. & Scarfe, C. M. (1990). Phase relations in peridotite þ CO2
systems to 12 GPa: Implications for the origin of kimberlite and carbonate stability in the Earth’s upper mantle. Journal of Geophysical
ResearchçSolid Earth 95, 15805^15816.
Canil, D., Virgo, D. & Scarfe, C. M. (1990). Oxidation state of mantle
xenoliths from British Columbia, Canada. Contributions to
Mineralogy and Petrology 104, 453^462.
Canil, D., Johnston, S. T. & Mihalynuk, M. (2006). Mantle redox in
Cordilleran ophiolites as a record of oxygen fugacity during partial
melting and the lifetime of mantle lithosphere. Earth and Planetary
Science Letters 248, 106^117.
Carmichael, I. S. E. (1991). The redox state of basic and silicic
magmas: a reflection of their source regions? Contributions to
Mineralogy and Petrology 106, 129^141.
Carmichael, I. S. E. & Nicholls, J. (1967). Iron^titanium oxides and
oxygen fugacities in volcanic rocks. Journal of Geophysical Researchç
Solid Earth 72, 4665^4687.
Catling, D. C., Zahnle, K. J. & McKay, C. P. (2001). Biogenic methane, hydrogen escape, and the irreversible oxidation of early
Earth. Science 293, 839^843.
Cavell, P. A., Wijbrans, J. R. & Baadsgaard, H. (1992). Archean magmatism in the Kaminak Lake area, District of Keewatin,
Northwest territoriesçages of the carbonatite-bearing alkaline
1385
JOURNAL OF PETROLOGY
VOLUME 52
complex and some host granitoid rocks. Canadian Journal of Earth
Sciences 29, 896^908.
Christie, D. M., Carmichael, I. S. E. & Langmuir, C. H. (1986).
Oxidation states of mid-ocean ridge basalt glasses. Earth and
Planetary Science Letters 79, 397^411.
Creighton, S., Stachel, T., Matveev, S., Hofer, H., McCammon, C. &
Luth, R. W. (2009). Oxidation of the Kaapvaal lithospheric
mantle driven by metasomatism. Contributions to Mineralogy and
Petrology 157, 491^504.
Creighton, S., Stachel, T., Eichenberg, D. & Luth, R. W. (2010).
Oxidation state of the lithospheric mantle beneath Diavik diamond
mine, central Slave craton, NWT, Canada. Contributions to
Mineralogy and Petrology 159, 645^657.
Dahlgren, S. (1994). Late Proterozoic and Carboniferous ultramafic
magmatism of carbonatitic affinity in southern Norway. Lithos 31,
141^154.
Dasgupta, R. & Hirschmann, M. M. (2006). Melting in the Earth’s
deep upper mantle caused by carbon dioxide. Nature 440, 659^662.
Dasgupta, R., Hirschmann, M. M. & Withers, A. C. (2004). Deep
global cycling of carbon constrained by the solidus of anhydrous,
carbonated eclogite under upper mantle conditions. Earth and
Planetary Science Letters 227, 73^85.
Dasgupta, R., Hirschmann, M. M. & Smith, N. D. (2007). Water follows carbon: CO2 incites deep silicate melting and dehydration beneath mid-ocean ridges. Geology 35, 135^138.
Delano, J. W. (2001). Redox history of the Earth’s interior since similar
to 3900 Ma: Implications for prebiotic molecules. Origins of Life
and Evolution of the Biosphere 31, 311^341.
Eggler, D. H. (1976). Does CO2 cause partial melting in the
low-velocity layer of the mantle? Geology 4, 69^72.
Eggler, D. H. (1983). Upper mantle oxidation state: evidence from olivine^orthopyroxene^ilmenite assemblages. Geophysical Research
Letters 10, 365^368.
Eggler, D. H. & Lorand, J. L. (1994). Sulfides, diamonds, and mantle
fO2. In: Meyer, H. O. A. & Leonardos, O. H. (eds) Diamonds: Characterization, Genesis and Exploration. Brasilia: CPRM, pp. 160^169.
Falloon, T. J. & Green, D. H. (1987). Anhydrous partial melting of
MORB pyrolite and other peridotite compositions at 10 kbar: implications for the origin of primitive MORB glasses. Mineralogy and
Petrology 37, 181^219.
Falloon, T. J. & Green, D. H. (1988). Anhydrous partial melting of
peridotite from 8 to 35 kb and the petrogenesis of MORB. Journal
of Petrology , Special Lithosphere Issue, 379^414.
Falloon, T. J., Green, D. H., Hatton, C. J. & Harris, K. L. (1988).
Anhydrous partial melting of a fertile and depleted peridotite
from 2kb to 30kb and application to basalt petrogenesis. Journal of
Petrology 29, 1257^1282.
Farquhar, J., Bao, H. M. & Thiemens, M. (2000). Atmospheric influence of Earth’s earliest sulfur cycle. Science 289, 756^758.
Foley, S. F. (1985). The oxidation state of lamproitic magmas.
Tschermaks Mineralogische und Petrographische Mitteilungen 34, 217^238.
Foley, S. F. (1988). The genesis of continental basic alkaline magmas:
an interpretation in terms of redox melting. Journal of Petrology ,
Special Lithophere Issue, 139^161.
Foley, S. F. (1989a). Experimental constraints on phlogopite chemistry
in lamproites. 1. The effect of water activity and oxygen fugacity.
EuropeanJournal of Mineralogy 1, 411^426.
Foley, S. F. (1989b). The genesis of lamproitic magmas in a reduced,
flourine-rich mantle. In: Ross, J. (ed.) Kimberlites and Related Rocks.
Melbourne: Blackwell, pp. 616^631.
Foley, S. (1992a). Petrological characterization of the source components of potassic magmasçgeochemical and experimental constraints. Lithos 28, 187^204.
NUMBERS 7 & 8
JULY & AUGUST 2011
Foley, S. (1992b). Vein-plus-wall-rock melting mechanisms in the lithosphere and the origin of potassic alkaline magmas. Lithos 28,
435^453.
Foley, S. F. (1993). An experimental study of olivine lamproite: first results from the diamond stability field. Geochimica et Cosmochimica
Acta 57, 483^489.
Foley, S. F. (1994). Geochemical and experimental studies of the origin
of ultrapotassic igneous rocks. Neues Jahrbuch fu«r Mineralogie,
Abhandlungen 167, 1^55.
Foley, S. F. (2008). Rejuvenation and erosion of the cratonic lithosphere. Nature Geoscience 1, 503^510.
Foley, S. F. (2009). The renaissance of redox melting. Geochimica et
Cosmochimica Acta 73, A388.
Foley, S. F., Musselwhite, D. S. & Van der Laan, S. R. (1999). Melt
compositions from ultramafic vein assemblages in the lithospheric
mantle: a comparison of cratonic and non-cratonic settings. In:
Gurney, J. J., Gurney, J. L., Pascoe, M. D. & Richardson, S. H.
(eds) J. B. Dawson Volume. Cape Town: Red Roof Design,
pp. 238^246.
Foley, S. F., Petibon, C. M., Jenner, G. A. & Kjarsgaard, B. A. (2001).
High U/Th partitioning by clinopyroxene from alkali silicate and
carbonatite metasomatism: an origin for Th/U disequilibrium in
mantle melts? Terra Nova 13, 104^109.
Foley, S. F., Andronikov, A. V. & Melzer, S. (2002a). Petrology of
ultramafic lamprophyres from the Beaver Lake area of Eastern
Antarctica and their relation to the breakup of Gondwanaland.
Mineralogy and Petrology 74, 361^384.
Foley, S., Tiepolo, M. & Vannucci, R. (2002b). Growth of early continental crust controlled by melting of amphibolite in subduction
zones. Nature 417, 837^840.
Foley, S. F., Buhre, S. & Jacob, D. E. (2003). Evolution of the Archaean
crust by delamination and shallow subduction. Nature 421, 249^252.
Foley, S. F., Andronikov, A. V., Jacob, D. E. & Melzer, S. (2006).
Evidence from Antarctic mantle peridotite xenoliths for changes
in mineralogy, geochemistry and geothermal gradients beneath a
developing rift. Geochimica et Cosmochimica Acta 70, 3096^3120.
Foley, S. F., Yaxley, G. M., Rosenthal, A., Buhre, S., Rapp, R. P. &
Jacob, D. E. (2009). The composition of near-solidus melts of peridotite in the presence of CO2 and H2O at 40 and 60 kbar. Lithos,
Proceedings of the 9th International Kimberlite Conference 112S, 274^283.
Frost, B. R. (1979). Mineral equilibria involving mixed volatiles in a
C^O^H fluid phase: the stabilities of graphite and siderite.
AmericanJournal of Science 279, 1035^1059.
Frost, D. J. & McCammon, C. A. (2008). The redox state of Earth’s
mantle. Annual Review of Earth and Planetary Sciences 36, 389^420.
Frost, D. J., Liebske, C., Langenhorst, F., McCammon, C. A.,
Tronnes, R. G. & Rubie, D. C. (2004). Experimental evidence for
the existence of iron-rich metal in the Earth’s lower mantle. Nature
428, 409^412.
Fujii, T. (1989). Genesis of mid-ocean ridge basalts. In: Saunders, A. D.
& Norry, M. J. (eds) Magmatism in the Ocean Basins. Geological
Society, London, Special Publications 42, 137^146.
Furman, T. & Graham, D. (1999). Erosion of lithospheric mantle beneath the East African Rift system: geochemical evidence from
the Kivu volcanic province. Lithos 48, 237^262.
Galimov, E. M. (2005). Redox evolution of the Earth caused by a
multi-stage formation of its core. Earth and Planetary Science Letters
233, 263^276.
Gao, S., Rudnick, R. L., Yuan, H. L., Lui,Y. S., Xu, W. L., Ling, W. L.,
Ayers, J., Wang, X. C. & Wang, Q. H. (2004). Recycling lower continental crust in the North China craton. Nature 432, 892^897.
Ghosh, S., Ohtani, E., Litasov, K. D. & Terasaki, H. (2009). Solidus of
carbonated peridotite from 10 to 20 GPa and origin of
1386
FOLEY
REDOX MELTING IN MANTLE
magnesiocarbonatite melt in the Earth’s deep mantle. Chemical
Geology 262, 17^28.
Glaser, S. M., Foley, S. F. & Gunther, D. (1999). Trace element compositions of minerals in garnet and spinel peridotite xenoliths
from the Vitim volcanic field, Transbaikalia, eastern Siberia. Lithos
48, 263^285.
Goldblatt, C., Claire, M. W., Lenton, T. M., Matthews, A. J.,
Watson, A. J. & Zahnle, K. J. (2009). Nitrogen-enhanced greenhouse warming on early Earth. Nature Geoscience 2, 891^896.
Green, D. H. (1970). The origin of basaltic and nephelinitic magmas.
Transactions of the Leicester Philosophical and Literary Society 64, 28^54.
Green, D. H. (1973). Experimental melting studies on a model upper
mantle composition at high pressures under water-saturated and
water-undersaturated conditions. Earth and Planetary Science Letters
19, 37^53.
Green, D. H. (1990). The role of oxidation^reduction and C^H^O
fluids in determining melting conditions and magma compositions
in the upper mantle. Proceedings of the Indian Academy of Sciences 99,
153^165.
Green, D. H. & Falloon, T. J. (1998). Pyrolite: a Ringwood concept
and its current expression. In: Jackson, I. (ed.) The Earth’s Mantle.
Cambridge: Cambridge University Press, pp. 311^378.
Green, D. H., Falloon, T. J. & Taylor, W. R. (1987). Mantle-derived
magmasçroles of variable source peridotite and variable C^H^O
fluid compositions. In: Mysen, B. O. (ed.) Magmatic Processes:
Physicochemical Principles. Washington, DC: Geochemical Society,
pp. 139^154.
Green, D. H., Taylor, W. R. & Foley, S. F. (1990). The Earth’s upper
mantle as a source for volatiles. In: Herbert, H. K. & Ho, S. E.
(eds) Stable Isotopes and Fluid Processes in Mineralization. Perth:
University of Western Australia, pp. 17^34.
Green, D. H., Falloon, T. J., Eggins, S. M. & Yaxley, G. M. (2001).
Primary magmas and mantle temperatures. European Journal of
Mineralogy 13, 437^451.
Green, D. H., Hibberson, W. O., Kovaks, I. & Rosenthal, A. (2010).
Water and its influence on the lithosphere^asthenosphere boundary. Nature (in press).
Grove, T. L., Chatterjee, N., Parman, S. W. & Medard, E. (2006). The
influence of H2O on mantle wedge melting. Earth and Planetary
Science Letters 249, 74^89.
Gupta, A. K. & Green, D. H. (1988). The liquidus surface of the
system forsterite^kalsilite^quartz at 28 kb under dry conditions, in
presence of H2O, and of CO2. Mineralogy and Petrology 39, 163^174.
Haggerty, S. E. & Tompkins, L. A. (1983). Redox state of Earth’s upper
mantle from kimberlitic ilmenites. Nature 303, 295^300.
Hammouda, T. (2003). High-pressure melting of carbonated eclogite
and experimental constraints on carbon recycling and storage in
the mantle. Earth and Planetary Science Letters 214, 357^368.
Hawkesworth, C. J. & Kemp, A. I. S. (2006). Evolution of the continental crust. Nature 443, 811^817.
Hermann, J. & Spandler, C. J. (2008). Sediment melts at sub-arc
depths: An experimental study. Journal of Petrology 49, 717^740.
Herzberg, C. & O’Hara, M. J. (1998). Phase equilibrium constraints
on the origin of basalts, picrites, and komatiites. Earth-Science
Reviews 44, 39^79.
Hirschmann, M. M. & Stolper, E. M. (1996). A possible role for garnet
pyroxenite in the origin of the ‘garnet signature’ in MORB.
Contributions to Mineralogy and Petrology 124, 185^208.
Ionov, D. A. & Wood, B. J. (1992). The oxidation state of
subcontinental mantle: oxygen thermobarometry of mantle xenoliths from central Asia. Contributions to Mineralogy and Petrology 111,
179^193.
Irving, A. J. (1974). Geochemical and high pressure experimental studies of garnet pyroxenite and pyroxene granulite xenoliths from
the Delegate basaltic pipes, Australia. Journal of Petrology 15, 1^40.
Ishikawa, T. & Nakamura, E. (1994). Origin of the slab component in
arc lavas from across-arc variation of B and Pb isotopes. Nature
370, 205^208.
Ishimaru, S., Arai, S. & Shukuno, H. (2009). Metal-saturated peridotite in the mantle wedge inferred from metal-bearing peridotite
xenoliths from Avacha volcano, Kamchatka. Earth and Planetary
Science Letters 284, 352^360.
Jacob, D. E. (2004). Nature and origin of eclogite xenoliths from kimberlites. Lithos 77, 295^316.
Jacob, D. E. & Foley, S. F. (1999). Evidence for Archean ocean crust
with low high field strength element signature from diamondiferous eclogite xenoliths. Lithos 48, 317^336.
Jacob, D. E., Viljoen, K. S., Grassineau, N. & Jagoutz, E. (2000).
Remobilization in the cratonic lithosphere recorded in polycrystalline diamond. Science 289, 1182^1185.
Jacob, D. E., Kronz, A. & Viljoen, K. S. (2004). Cohenite, native iron
and troilite inclusions in garnets from polycrystalline diamond aggregates. Contributions to Mineralogy and Petrology 146, 566^576.
Jacob, D. E., Bizimis, M. & Salters, V. J. M. (2005). Lu^Hf and geochemical systematics of recycled ancient oceanic crust: evidence
from Roberts Victor eclogites. Contributions to Mineralogy and
Petrology 148, 707^720.
Janse, A. J. A. & Sheahan, P. A. (1995). Catalog of world wide diamond and kimberlite occurrencesça selective and annotative
approach. Journal of Geochemical Exploration 53, 73^111.
Jaques, A. L. & Green, D. H. (1980). Anhydrous melting of peridotite
at 0^15 kb pressure and the genesis of tholeiitic basalts.
Contributions to Mineralogy and Petrology 73, 287^310.
Kadik, A. (1988). Effects of melting on the evolution of fluid and redox
conditions in the upper mantle. Geochemistry International 25, 76^84.
Kadik, A. (1997). Evolution of Earth’s redox state during upwelling of
carbon-bearing mantle. Physics of the Earth and Planetary Interiors
100, 157^166.
Kadik, A. A. (2003). Mantle-derived reduced fluids: Relationship to
the chemical differentiation of planetary matter. Geochemistry
International 41, 844^855.
Kasting, J. F., Eggler, D. H. & Raeburn, S. P. (1993). Mantle redox
evolution and the oxidation-state of the Archean atmosphere.
Journal of Geology 101, 245^257.
Keller, J. (1981). Carbonatitic volcanism in the Kaisertstuhl alkaline
complexçevidence for highly fluid carbonatitic melts at the
Earth’s surface. Journal of Volcanology and Geothermal Research 9,
423^431.
Keller, R. A., Taylor, L. A., Snyder, G. A., Sobolev, V. N., Carlson, W.
D., Bezborodov, S. M. & Sobolev, N. V. (1999). Detailed pull-apart
of a diamondiferous eclogite xenolith: implications for mantle processes during diamond genesis. In: Gurney, J. J., Gurney, J. L.,
Pascoe, M. D. & Richardson, S. H. (eds) J. B. Dawson Volume.
Cape Town: Red Roof Design, pp. 397^402.
Kent, R. W., Hardarson, B. S., Saunders, A. D. & Storey, M. (1996).
Plateaux ancient and modern: Geochemical and sedimentological perspectives on Archaean oceanic magmatism. Lithos 37,
129^142.
Kessel, R., Schmidt, M. W., Ulmer, P. & Pettke, T. (2005). Trace element signature of subduction-zone fluids, melts and supercritical liquids at 120^180 km depth. Nature 437, 724^727.
Kinzler, R. J. & Grove, T. L. (1992). Primary magmas of mid-ocean
ridge basalts 2. Applications. Journal of Geophysical ResearchçSolid
Earth 97, 6907^6926.
1387
JOURNAL OF PETROLOGY
VOLUME 52
Kohlstedt, D. L., Keppler, H. & Rubie, D. C. (1996). Solubility of
water in the alpha, beta and gamma phases of (Mg, Fe)(2)SiO4.
Contributions to Mineralogy and Petrology 123, 345^357.
Konzett, J., Armstrong, R. A. & Gunther, D. (2000). Modal metasomatism in the Kaapvaal craton lithosphere: constraints on
timing and genesis from U^Pb zircon dating of metasomatized
peridotites and MARID-type xenoliths. Contributions to Mineralogy
and Petrology 139, 704^719.
Kump, L. R., Kasting, J. F. & Barley, M. E. (2001). Rise of atmospheric oxygen and the ‘upside-down’ Archean mantle. Geochemistry,
Geophysics, Geosystems 2, doi:10.1029/2000GC000114.
Kushiro, I. (1975). On the nature of silicate melt and its significance in
magma genesis; regularities in the shift of the liquidus boundaries
involving olivine, pyroxene, and silica minerals. American Journal of
Science 275, 411^431.
Lazarov, M., Woodland, A. B. & Brey, G. P. (2009). Thermal state and
redox conditions of the Kaapvaal mantle: A study of xenoliths
from the Finsch mine, South Africa. Lithos 112, 913^923.
Lee, C. T. A. (2006). Geochemical/petrologic constraints on the origin
of cratonic mantle. In: Benn, K., Mareschal, J.-C. & Condie, K.
C. (eds) Archean Geodynamics and Environments. Washington, DC:
American Geophysical Union, pp. 89^114.
Lee, C. T. A., Leeman, W. P., Canil, D. & Li, Z. X. A. (2005). Similar
V/Sc systematics in MORB and arc basalts: Implications for the
oxygen fugacities of their mantle source regions. Journal of Petrology
46, 2313^2336.
Li, Z. X. A. & Lee, C. T. A. (2004). The constancy of upper mantle
fO2 through time inferred from V/Sc ratios in basalts. Earth and
Planetary Science Letters 228, 483^493.
Libourel, G., Marty, B. & Humbert, F. (2003). Nitrogen solubility in
basaltic melt. Part I. Effect of oxygen fugacity. Geochimica et
Cosmochimica Acta 67, 4123^4135.
Link, K., Koehn, D., Barth, M. G., Tiberindwa, J., Barifaijo, E.,
Aanju, K. & Foley, S. F. (2010). Continuous cratonic crust between
the Congo and Tanzania Archaean terranes of western Uganda.
International Journal of Earth Sciences 99, doi:10.1007/
s00531-010-0548-8.
Litasov, K. D. & Ohtani, E. (2009a). Phase relations in the
peridotite-carbonate-chloride system at 7·0^16·5 GPa and the role
of chlorides in the origin of kimberlite and diamond. Chemical
Geology 262, 29^41.
Litasov, K. D. & Ohtani, E. (2009b). Solidus and phase relations of
carbonated peridotite in the system CaO^Al2O3^MgO^SiO2^
Na2O^CO2 to the lower mantle depths. Physics of the Earth and
Planetary Interiors 177, 46^58.
Lloyd, F. E., Arima, M. & Edgar, A. D. (1985). Partial melting of a
phlogopite-clinopyroxenite nodule from south-west Uganda: an experimental study bearing on the origin of highly potassic continental rift volcanics. Contributions to Mineralogy and Petrology 91, 321^329.
Luhr, J. F. & Aranda-Gomez, J. J. (1997). Mexican peridotite xenoliths
and tectonic terranes: Correlations among vent location, texture,
temperature, pressure, and oxygen fugacity. Journal of Petrology 38,
1075^1112.
Luth, R., Virgo, D., Boyd, F. & Wood, B. (1990). Ferric iron in mantle
derived garnets. Contributions to Mineralogy and Petrology 104, 56^72.
Mackwell, S. (2008). Rheological consequences of redox state. In:
MacPherson, G. J. (ed.) Oxygen in the Solar System. Mineralogical
Society of America and Geochemical Society, Reviews in Mineralogy and
Geochemistry 68, 555^569.
Malaspina, N., Poli, S. & Fumagalli, P. (2009). The oxidation state of
metasomatized mantle wedge: insights from C^O^H-bearing
garnet peridotite. Journal of Petrology 50, 1533^1552.
NUMBERS 7 & 8
JULY & AUGUST 2011
Mallmann, G. & O’Neill, H. S. C. (2009). The crystal/melt partitioning of V during mantle melting as a function of oxygen fugacity
compared with some other elements (Al, P, Ca, Sc, Ti, Cr, Fe, Ga,
Y, Zr and Nb). Journal of Petrology 50, 1765^1794.
Martin, H. (1999). Adakitic magmas: modern analogues of Archaean
granitoids. Lithos 46, 411^429.
Mattioli, G. S. & Wood, B. J. (1986). Upper mantle oxygen fugacity recorded by spinel lherzolites. Nature 322, 626^628.
Matveev, S., Ballhaus, C., Fricke, K., Truckenbrodt, J. &
Ziegenbein, D. (1997). Volatiles in the Earth’s mantle: I. Synthesis
of CHO fluids at 1273 K and 2·4 GPa. Geochimica et Cosmochimica
Acta 61, 3081^3088.
McCammon, C. & Kopylova, M. G. (2004). A redox profile of the
Slave mantle and oxygen fugacity control in the cratonic mantle.
Contributions to Mineralogy and Petrology 148, 55^68.
McCammon, C. A., Griffin, W. L., Shee, S. R. & O’Neill, H. S. C.
(2001). Oxidation during metasomatism in ultramafic xenoliths
from the Wesselton kimberlite, South Africa: implications for the
survival of diamond. Contributions to Mineralogy and Petrology 141,
287^296.
McGuire, A. V., Dyar, M. D. & Nielson, J. E. (1991). Metasomatic oxidation of upper mantle peridotite. Contributions to Mineralogy and
Petrology 109, 252^264.
McInnes, B. I. A., Gregoire, M., Binns, R. A., Herzig, P. M. &
Hannington, M. D. (2001). Hydrous metasomatism of oceanic
sub-arc mantle, Lihir, Papua New Guinea: petrology and geochemistry of fluid-metasomatised mantle wedge xenoliths. Earth and
Planetary Science Letters 188, 169^183.
McKenzie, D. & Bickle, M. J. (1988). The volume and composition of
melt generated by extension of the lithosphere. Journal of Petrology
29, 625^679.
Mibe, K., Kanzaki, M., Kawamoto, T., Matsukage, K. N., Fei, Y. W.
& Ono, S. (2007). Second critical endpoint in the peridotite^H2O
system. Journal of Geophysical ResearchçSolid Earth 112, doi:10.1029/
2005JB004125.
Millhollen, G. L., Irving, A. J. & Wyllie, P. J. (1974). Melting interval
of peridotite with 5·7 per cent water to 30 kilobars. Journal of
Geology 82, 575^587.
Moore, K. R. & Wood, B. J. (1998). The transition from carbonate to
silicate melts in the CaO^MgO^SiO2^CO2 system. Journal of
Petrology 39, 1943^1951.
Mysen, B. O. (1987). Magmatic silicate melts: relations between bulk
composition, structure and properties. In: Mysen, B. O. (ed.)
Magmatic Processes: Physicochemical Principles. Washington, DC:
Geochemical Society, pp. 375^399.
Mysen, B. O., Virgo, D. & Seifert, F. A. (1982). The structure of
silicate melts; implications for chemical and physical properties
of natural magma. Reviews of Geophysics and Space Physics 20,
353^383.
Mysen, B. O., Yamashita, S. & Chertkova, N. (2008). Solubility and solution mechanisms of NOH volatiles in silicate melts at high pressure and temperatureçamine groups and hydrogen fugacity.
American Mineralogist 93, 1770.
Mysen, B. O., Fogel, M. L., Morrill, P. L. & Cody, G. D. (2009).
Solution behavior of reduced C^O^H volatiles in silicate melts at
high pressure and temperature. Geochimica et Cosmochimica Acta 73,
1696^1710.
Nasir, S. (1996). Oxygen thermobarometry of the Semail harzburgite
massif, Oman and United Arab Emirates. European Journal of
Mineralogy 8, 153^163.
Nasir, S., Al-Sayigh, A., Alharthy, A. & Al-Lazki, A. (2006).
Geochemistry and petrology of Tertiary volcanic rocks and related
1388
FOLEY
REDOX MELTING IN MANTLE
ultramafic xenoliths from the central and eastern Oman
Mountains. Lithos 90, 249^270.
Nasir, S., Everard, J. L., McClenaghan, M. P., Bombardieri, D. &
Worthing, M. A. (2010). The petrology of high pressure xenoliths
and associated Cenozoic basalts from northeastern Tasmania.
Lithos 118, 35^49.
Neal, C. R., Mahoney, J. J., Kroenke, L. W., Duncan, R. A. &
Petterson, M. G. (1997). The Ontong-Java Plateau. In:
Mahoney, J. J. & Coffin, M. F. (eds) Large Igneous Provinces.
continental, oceanic, and planetary flood volcanism. AGU, Washington:
Geophysical Monograph, pp. 183^216.
Nisbet, E. G. & Fowler, C. M. R. (1983). Model for Archean plate tectonics. Geology 11, 376^379.
Nisbet, E. G., Cheadle, M. J., Arndt, N. T. & Bickle, M. J. (1993).
Constraining the potential temperature of the Archean mantle: a
review of the evidence from komatiites. Lithos 30, 291^307.
Olafsson, M. & Eggler, D. H. (1983). Phase relations of amphibole,
amphibole^carbonate, and phlogopite^carbonate peridotite: petrologic constraints on the asthenosphere. Earth and Planetary Science
Letters 64, 305^315.
O’Neill, H. S. C. & Wall, V. J. (1987). The olivine^orthopyroxene^
spinel oxygen geobarometer, the nickel precipitation curve, and
the oxygen fugacity of the Earth’s upper mantle. Journal of Petrology
28, 1169^1191.
Owen, T., Cess, R. D. & Ramanathan, V. (1979). Enhanced CO2
greenhouse to compensate for reduced solar luminosity on early
Earth. Nature 277, 640^642.
Pal’yanov, Y. N., Sokol, A. G., Tomilenko, A. A. & Sobolev, N. V.
(2005). Conditions of diamond formation through carbonate^
silicate interaction. EuropeanJournal of Mineralogy 17, 207^214.
Parkinson, I. J. & Arculus, R. J. (1999). The redox state of subduction
zones: insights from arc-peridotites. Chemical Geology 160, 409^423.
Parkinson, I. J., Arculus, R. J. & Eggins, S. M. (2003). Peridotite xenoliths from Grenada, Lesser Antilles Island Arc. Contributions to
Mineralogy and Petrology 146, 241^262.
Pavlov, A. A. & Kasting, J. F. (2002). Mass-independent fractionation
of sulfur isotopes in Archean sediments: Strong evidence for an
anoxic Archean atmosphere. Astrobiology 2, 27^41.
Pearson, D. G. (1999). The age of continental roots. Lithos 48, 171^194.
Pertermann, M. & Hirschmann, M. M. (2003). Partial melting experiments on a MORB-like pyroxenite between 2 and 3 GPa:
Constraints on the presence of pyroxenite in basalt source regions
from solidus location and melting rate. Journal of Geophysical
ResearchçSolid Earth 108, doi:10.1029/2000JB000118.
Plank, T., Cooper, L. B. & Manning, C. E. (2009). Emerging
geothermometers for estimating slab surface temperatures. Nature
Geoscience 2, 611^615.
Platz, T., Foley, S. F. & Andre¤, L. (2004). Complex evolution of the
Nyiragongo volcanics, Virunga Province. Journal of Volcanology and
Geothermal Research 136, 269^295.
Poli, S. & Schmidt, M. W. (1995). H2O transport and release in subduction zonesçexperimental constraints on basaltic and andesitic
systems. Journal of Geophysical ResearchçSolid Earth 100,
22299^22314.
Poli, S., Franzolin, E., Fumagalli, P. & Crottini, A. (2009). The transport of carbon and hydrogen in subducted oceanic crust: An experimental study to 5 GPa. Earth and Planetary Science Letters 278,
350^360.
Prelevic, D., Foley, S. F., Romer, R. & Conticelli, S. (2008).
Mediterranean Tertiary lamproites derived from multiple source
components in postcollisional geodynamics. Geochimica et
Cosmochimica Acta 72, 2125^2156.
Qi, Q., Taylor, L. A. & Zhou, X. M. (1995). Petrology and geochemistry of mantle peridotite xenoliths from SE China. Journal of
Petrology 36, 55^79.
Rapp, R. P., Watson, E. B. & Miller, C. F. (1991). Partial melting of
amphbiolite and eclogite and the origin of Archean trondhjemites
and tonalites. Precambrian Research 51, 1^25.
Rehfeldt, T., Foley, S. F., Jacob, D. E., Carlson, R. W. & Lowry, D.
(2008). Contrasting types of metasomatism in dunite, wehrlite and
websterite xenoliths from Kimberley, South Africa. Geochimica et
Cosmochimica Acta 72, 5722^5756.
Richardson, S. H., Harris, J. W. & Gurney, J. J. (1993). 3 generations
of diamonds from old continental mantle. Nature 366, 256^258.
Ringwood, A. E. (1982). Phase transformations and differentiation in
subducted lithosphereçimplications for mantle dynamics, basalt
petrogenesis, and crustal evolution. Journal of Geology 90, 611^643.
Rogers, N. W., James, D., Kelley, S. P. & De Mulder, M. (1998). The
generation of potassic lavas from the eastern Virunga province,
Rwanda. Journal of Petrology 39, 1223^1247.
Rohrbach, A., Ballhaus, C., Golla-Schindler, U., Ulmer, P.,
Kamenetsky, V. S. & Kuzmin, D. V. (2007). Metal saturation in the
upper mantle. Nature 449, 456^458.
Rosenthal, A., Foley, S. F., Pearson, D. G., Nowell, G. M. & Tappe, S.
(2009). Magmatic evolution at the propagating tip of a continental
riftça geochemical study of primitive alkaline volcanic rocks of
the western branch of the East African Rift. Earth and Planetary
Science Letters 284, 236^248.
Rudnick, R., McDonough, W. & Orpin, A. (1994). Northern
Tanzanian peridotite xenoliths: a comparison with Kaapvaal peridotites and inferences on metasomatic interactions. In: Meyer, H.
O. A. & Leonardos, O. H. (eds) Kimberlites, Related Rocks and
Mantle Xenoliths. Brasilia: CPRM, pp. 336^353.
Rye, R., Kuo, P. H. & Holland, H. D. (1995). Atmospheric carbondioxide concentrations before 2.2 billion years ago. Nature 378,
603^605.
Schidlowski, M. (1981). Uraniferous constituents of the Witwatersrand
Conglomerates: ore-microscopic observations and implications for
the Witwatersrand metallogeny. In: US Geological Survey, Professional
Papers 1161, N1^N29.
Schleicher, H., Keller, J. & Kramm, U. (1990). Isotope studies on alkaline volcanics and carbonatites from the Kaiserstuhl, Federal
Republic of Germany. Lithos 26, 21^35.
Schmidt, M. W. & Poli, S. (1998). Experimentally based water budgets
for dehydrating slabs and consequences for arc magma generation.
Earth and Planetary Science Letters 163, 361^379.
Shankland, T. J. & Waff, H. S. (1977). Partial melting and electrical
conductivity anomalies in the upper mantle. Journal of Geophysical
ResearchçSolid Earth 82, 5409^5417.
Sheldon, N. D. (2006). Precambrian paleosols and atmospheric CO2
levels. Precambrian Research 147, 148^155.
Shimizu, N. & Sobolev, N. (1995). Young peridotitic diamonds from
the Mir kimberlite pipe. Nature 375, 394^397.
Simon, N. S. C., Carlson, R. W., Pearson, D. G. & Davies, G. R.
(2007). The origin and evolution of the Kaapvaal cratonic lithospheric mantle. Journal of Petrology 48, 589^625.
Sleep, N. H., Meibom, A., Fridriksson, T., Coleman, R. G. & Bird, D.
K. (2004). H2-rich fluids from serpentinization: Geochemical and
biotic implications. Proceedings of the National Academy of Sciences of
the USA 101, 12818^12823.
Smart, K. A., Heaman, L. M., Chacko, T., Simonetti, A.,
Kopylova, M., Mah, D. & Daniels, D. (2009). The origin of
high-MgO diamond eclogites from the Jericho Kimberlite,
Canada. Earth and Planetary Science Letters 284, 527^537.
1389
JOURNAL OF PETROLOGY
VOLUME 52
Sobolev, A. V., Hofmann, A. W., Sobolev, S. V. & Nikogosian, I. K.
(2005). An olivine-free mantle source of Hawaiian shield basalts.
Nature 434, 590^597.
Sobolev, A. V., Hofmann, A. W., Kuzmin, D. V., Yaxley, G. M.,
Arndt, N. T., Chung, S. L., Danyushevsky, L. V., Elliott, T.,
Frey, F. A., Garcia, M. O., Gurenko, A. A., Kamenetsky, V. S.,
Kerr, A. C., Krivolutskaya, N. A., Matvienkov, V. V.,
Nikogosian, I. K., Rocholl, A., Sigurdsson, I. A.,
Sushchevskaya, N. M. & Teklay, M. (2007). The amount of recycled
crust in sources of mantle-derived melts. Science 316, 412^417.
Song, S. G., Su, L., Niu, Y. L., Lai, Y. & Zhang, L. F. (2009). CH4 inclusions in orogenic harzburgite: Evidence for reduced slab fluids
and implication for redox melting in mantle wedge. Geochimica et
Cosmochimica Acta 73, 1737^1754.
Stachel, T. & Harris, J. W. (2008). The origin of cratonic diamondsç
Constraints from mineral inclusions. Ore Geology Reviews 34, 5^32.
Stalder, R., Foley, S. F., Brey, G. P. & Horn, I. (1998). Mineral aqueous
fluid partitioning of trace elements at 900^12008C and
3·0^5·7 GPa: New experimental data for garnet, clinopyroxene,
and rutile, and implications for mantle metasomatism. Geochimica
et Cosmochimica Acta 62, 1781^1801.
Stalder, R., Ulmer, P., Thompson, A. B. & Gunther, D. (2001). High
pressure fluids in the system MgO^SiO2^H2O under upper
mantle conditions. Contributions to Mineralogy and Petrology 140,
607^618.
Stevenson, D. J. (2001). Mars’ core and magnetism. Nature 412, 214^219.
Stolper, E. (1982). Water in silicate glasses; an infrared spectroscopic
study. Contributions to Mineralogy and Petrology 81, 1^17.
Sweeney, R. J. (1994). Carbonatite melt compositions in the Earth’s
mantle. Earth and Planetary Science Letters 128, 259^270.
Takahashi, E. (1986). Melting of a dry peridotite KLB-1 up to 14 GPa:
Implications on the origin of peridotitic upper mantle. Journal of
Geophysical ResearchçSolid Earth 91, 9367^9382.
Tappe, S., Foley, S. F., Jenner, G. A., Heaman, L. M., Kjarsgaard, B.
A., Romer, R. L., Stracke, A., Joyce, N. & Hoefs, J. (2006).
Genesis of ultramafic lamprophyres and carbonatites at Aillik Bay,
Labrador: A consequence of incipient lithospheric thinning beneath the North Atlantic craton. Journal of Petrology 47, 1261^1315.
Tappe, S., Foley, S. F., Stracke, A., Romer, R. L., Kjarsgaard, B. A.,
Heaman, L. M. & Joyce, N. (2007). Craton reactivation on the
Labrador Sea margins: 40Ar/39Ar age and Sr^Nd^Hf^Pb isotope
constraints from alkaline and carbonatite intrusives. Earth and
Planetary Science Letters 256, 433^454.
Tappe, S., Foley, S. F., Kjarsgaard, B. A., Romer, R. L., Heaman, L.
M., Stracke, A. & Jenner, G. A. (2008). Between carbonatite and
lamproiteçdiamondiferous Torngat ultramafic lamprophyres
fromed by fluid-fluxed melting of cratonic MARID-type metasomes. Geochimica et Cosmochimica Acta 72, 3258^3286.
Taylor, S. R. & McLennan, S. M. (1995). The geochemical evolution of
the continental crust. Reviews of Geophysics 33, 241^265.
Taylor, W. R. (1985). The role of C^O^H fluids in upper mantle
processesça theroetical, experimental and spectroscopic study,
PhD thesis, University of Tasmania, Hobart.
Taylor, W. R. & Green, D. H. (1987). The petrogenetic role of methane:
effect on liquidus phase relations and the solubility mechanism of
reduced C^H volatiles. In: Mysen, B. O. (ed.) Magmatic Processes:
Physicochemical Principles. Washington, DC: Geochemical Society,
pp. 121^138.
Taylor, W. R. & Green, D. H. (1988). Measurement of reduced peridotite^C^O^H solidus and implications for redox melting of the
mantle. Nature 332, 349^352.
NUMBERS 7 & 8
JULY & AUGUST 2011
Thibault, Y., Edgar, A. D. & Lloyd, F. E. (1992). Experimental investigation of melts from a carbonated phlogopite lherzolite: implications for metasomatism in the continental lithosphere. American
Mineralogist 77, 784^794.
Thompson, A. B. (1992). Water in the Earth’s upper mantle. Nature 358,
295^302.
Tilton, G. R. & Bell, K. (1994). Sr^Nd^Pb isotope relationships in late
Archean carbonatites and alkaline complexesçapplications to the
geochemical evolution of Archean mantle. Geochimica et
Cosmochimica Acta 58, 3145^3154.
Ulmer, G. C., Grandstaff, D. E., Woermann, E., Gobbels, M.,
Schonitz, M. & Woodland, A. B. (1998). The redox stability of moissanite (SiC) compared with metal^metal oxide buffers at 1773 K
and at pressures up to 90 kbar. Neues Jahrbuch fu«r Mineralogie,
Abhandlungen 172, 279^307.
Ulmer, P. & Trommsdorff, V. (1995). Serpentine stability to mantle
depths and subduction-related magmatism. Science 268, 858^861.
Villeneuve, M. E. & Relf, C. (1998). Tectonic setting of 2·6 Ga carbonatites in the Slave Province, NW Canada. Journal of Petrology
39, 1975^1986.
Wadhwa, M. (2008). Redox conditions on small bodies, the Moon and
Mars. In: MacPherson, G. J. (ed.) Oxygen in the Solar System.
Mineralogical Society of America and Geochemical Society, Reviews of
Mineralogy and Geochemistry 68, 493^510.
Wallace, M. E. & Green, D. H. (1988). An experimental determination of primary carbonatite magma composition. Nature 335,
343^346.
Wang, J., Hattori, K. H., Li, J. P. & Stern, C. R. (2008). Oxidation
state of Paleozoic subcontinental lithospheric mantle below the
Pali Aike volcanic field in southernmost Patagonia. Lithos 105,
98^110.
Wang, J., Hattori, K. H., Kilian, R. & Stern, C. R. (2009).
Metasomatism of sub-arc mantle peridotites below southern South
America: reduction of fO2 by slab-melt. Contributions to Mineralogy
and Petrology 153, 607^624.
White, R. S., McKenzie, D. & O’Nions, R. K. (1992). Oceanic crustal
thickness from seismic measurements and rare-earth element
inversions. Journal of Geophysical Research 97, 19683^19715.
Woermann, E. & Rosenhauer, M. (1985). Fluid phases and the redox
state of the Earth’s mantleçextrapolations based on experimental,
phase-theoretical and petrological data. Fortschritte der Mineralogie
63, 263^349.
Wood, B. J. (1991). Oxygen barometry of spinel peridotites. In:
Lindsley, D. H. (ed.) Oxide Minerals. Mineralogical Society of America,
Reviews in Mineralogy 25, 417^431.
Wood, B. J. (1993). Carbon in the core. Earth and Planetary Science Letters
117, 593^607.
Wood, B. J. & Virgo, D. (1989). Upper mantle oxidation-stateçferric
iron contents of lherzolite spinels by 57Fe Mo«ssbauer spectroscopy
and resultant oxygen fugacities. Geochimica et Cosmochimica Acta 53,
1277^1291.
Woodland, A. B. & Koch, M. (2003). Variation in oxygen fugacity
with depth in the upper mantle beneath the Kaapvaal craton,
Southern Africa. Earth and Planetary Science Letters 214, 295^310.
Woodland, A. B. & Peltonen, P. (1999). Ferric iron contents of garnet
and clinopyroxene and estimated oxygen fugacities of peridotie
xenoliths from the eastern Finland kimberlite province. In:
Gurney, J. J., Gurney, J. L., Pascoe, M. D. & Richardson, S. H.
(eds) P. H. Nixon Volume. Cape Town: Red Roof Design, pp. 904^911.
Woodland, A. B., Kornprobst, J. & Wood, B. J. (1992). Oxygen thermobarometry of orogenic lherzolite massifs. Journal of Petrology 33,
203^230.
1390
FOLEY
REDOX MELTING IN MANTLE
Woodland, A. B., Kornprobst, J., McPherson, E., Bodinier, J. L. &
Menzies, M. A. (1996). Metasomatic interactions in the lithospheric
mantle: Petrologic evidence from the Lherz massif, French
Pyrenees. Chemical Geology 134, 83^112.
Woodland, A. B., Kornprobst, J. & Tabit, A. (2006). Ferric iron in orogenic lherzolite massifs and controls of oxygen fugacity in the
upper mantle. Lithos 89, 222^241.
Wyllie, P. J. (1978). Mantle fluid compositions buffered in peridotite^
CO2^H2O by carbonates, amphibole, and phlogopite. Journal of
Geology 86, 687^713.
Wyllie, P. J. (1980). The origin of kimberlite. Journal of Geophysical
Research 85, 6902^6910.
Wyllie, P. J. (1988). Magma genesis, plate tectonics, and chemical differentiation of the Earth. Reviews of Geophysics 26, 370^404.
Wyllie, W. J. & Huang, W. L. (1976). High CO2 solubilities in mantle
magmas. Geology 4, 21^24.
Xu, Y. G. (2001). Thermo-tectonic destruction of the Archaean lithospheric keel beneath the Sino-Korean Craton in China: Evidence,
timing and mechanism. Physics and Chemistry of the Earth Part Aç
Solid Earth and Geodesy 26, 747^757.
Yaxley, G. M. & Brey, G. P. (2004). Phase relations of
carbonate-bearing eclogite assemblages from 2·5 to 5·5 GPa: implications for petrogenesis of carbonatites. Contributions to Mineralogy
and Petrology 146, 606^619.
Yaxley, G. M. & Green, D. H. (1994). Experimental demonstration of
refractory carbonate-bearing eclogite and siliceous melt in the subduction regime. Earth and Planetary Science Letters 128, 313^325.
1391