You are on page 1of 26

Lithos 334–335 (2019) 205–230

Contents lists available at ScienceDirect

Lithos

journal homepage: www.elsevier.com/locate/lithos

Seismicity and mineral destabilizations in the subducting mantle up to


6 GPa, 200 km depth
Thomas P. Ferrand ⁎
Earthquake Research Institute, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan

a r t i c l e i n f o a b s t r a c t

Article history: In the subducting oceanic lithosphere, a significant part of the seismicity is triggered in the mantle, especially
Received 28 June 2018 along the upper and lower Wadati-Benioff planes. Several studies have investigated the potential involvement
25 February 2019 of dehydration reactions in the triggering mechanism of mantle earthquakes. Recent experimental results reveal
Accepted 11 March 2019
that, under subduction conditions, mechanical instabilities nucleate in strong stable mineral aggregates during
Available online 15 March 2019
the destabilization of minor amounts of antigorite, i.e. the high-temperature serpentine, through a stress transfer,
Keywords:
without any fluid overpressure. Here I confront these laboratory results to seismological observations. On one
Earthquakes hand, most of the natural hydrous magnesium silicates seem to be known, with experimentally deduced stability
Intraslab limits up to 7 GPa, at least, available as relatively accurate estimates. On the other hand, recent achievements in
DDST thermal structure of subduction zones combined with precise hypocentre relocation give access to pressure and
Dehydration temperature conditions at earthquakes hypocentres. A series of P-T diagrams summarizes the stability limit of
Transformation minerals that may be part of natural peridotites or serpentinites with variable compositions at pressures from
Mantle 0.5 to 6 GPa and temperatures from 200 to 950 °C, and compares it with seismicity. Both hydrous and anhydrous
Stress transfer
phases are considered. A myriad of minor metamorphic reactions could participate in a transformation-driven
Lower Wadati-Benioff plane
stress transfer, even if the stability limits of serpentine minerals seem to correlate with most of the observed
Double seismic structure
Peridotite seismicity.
Hydration © 2019 Elsevier B.V. All rights reserved.
Hydrous phases
Water transfer
Serpentine
Antigorite
Chlorite
Clinochlore
Talc
DHMS

1. Introduction Kirby, 1987; Schubnel et al., 2013) and/or enstatite (Akashi et al.,
2009; Xu et al., 2018). In this study, I focus on subduction mantle earth-
Mantle earthquakes occur in different geological contexts, such as quakes (30–200 km), a large amount of which is triggered 10–40 km be-
shearing oceanic transform faults (e.g. Abercrombie and Ekström, neath the subduction interface in the lower Wadati-Benioff plane
2001; Antolik et al., 2006; Delescluse and Chamot-Rooke, 2007; (Brudzinski et al., 2007; Burbach and Frohlich, 1986; Hacker et al.,
Deplus et al., 1998; Henry, 2000), continental strike-slip faults (e.g. 2003; Hasegawa et al., 1978; Peacock, 2001; Yamaoka et al., 1986). A
Ferrand et al., 2018; Ueda et al., 2008) and, for most, subduction zones recent study (Abers et al., 2013) calculated the pressure and tempera-
(e.g. Yamaoka et al., 1986; Burbach and Frohlich, 1986; Abers and ture conditions at earthquakes hypocentres using both thermal models
Sarker, 1996; Peacock, 2001; Abers et al., 2006, 2009, 2013; Andersen of subduction zones (Syracuse et al., 2010; van Keken et al., 2011) and
et al., 2014; Scambelluri et al., 2017). Among the latter, geophysicists precise hypocentre relocation techniques (e.g. Kita et al., 2010a,
distinguish between intermediate-depth (≈35–400 km) and deep- 2010b). They concluded that in warm subduction zones (Cascadia,
focus (N400–700 km; Frohlich, 2006) earthquakes, respectively related Tokai and Kii) most of the earthquakes are triggered in the mantle,
to dehydration reactions (e.g. Jung and Green, 2004; Ferrand et al., near the Moho or some kilometres deeper, whilst earthquakes clearly
2017) and phase transitions of olivine (e.g. Green and Houston, 1995; form two distant planes, i.e. the upper and lower Wadati-Benioff planes,
in cold subduction zones (e.g. Tohoku and Hokkaido).
⁎ Corresponding author. The mechanism of mantle earthquakes is still a matter of debate
E-mail address: ferrand@eri.u-tokyo.ac.jp (T.P. Ferrand). (Section 2.4), but recent studies have demonstrated mantle hydration

https://doi.org/10.1016/j.lithos.2019.03.014
0024-4937/© 2019 Elsevier B.V. All rights reserved.
206 T.P. Ferrand / Lithos 334–335 (2019) 205–230

tens of kilometres below the Moho (e.g. Cai et al., 2018; Shillington et al., (Pawley and Wood, 1995), but compositional variations, e.g. Tschermak
2015) and the presence of transient aqueous fluid within the lower substitution, are observed in HP hydrated talc-like phases (e.g. Chinnery
Wadati-Benioff plane (Bloch et al., 2018). High-pressure dehydration et al., 1999; Fumagalli and Poli, 2005). Considering the complexity of
experiments have evidenced a stress transfer upon antigorite destabili- serpentinized faults (Fig. 3), this study aims to show to what extent dehy-
zation (DDST; Ferrand et al., 2017; Fig. 1), which would explain recent dration reactions could suffice to explain all lower-plane earthquakes in
seismological observations (Kita and Ferrand, 2018). Up to now, the the subducting mantle at the Nankai, Japan, Kuril and Cascadia trenches
lower Wadati-Benioff plane has been proposed to correspond to the de- (Abers et al., 2013).
hydration “isotherm” of antigorite (e.g. Abers et al., 2013; Hacker et al., Furthermore, using recent relocation data (Sippl et al., 2018) and
2003; Peacock, 2001) or to the hydration limit of the oceanic lithosphere slab thermal structure (Wada and Wang, 2009) for northern Chile, I es-
(Ferrand et al., 2017; Kita and Ferrand, 2018). Other studies consider timated the P-T conditions at hypocenters in the subducting mantle as
that antigorite has nothing to do with lower-plane seismicity was previously done for other subduction zones (Abers et al., 2013).
(Reynard et al., 2010) and that it could be due to self-localizing thermal This intermediate-age lithosphere (≈ 45 Ma) shows two distinct seis-
runaway within peridotite above 600 °C (e.g. Kelemen and Hirth, 2007; micity planes within the subducting mantle, separated by ≈ 25 km
Ohuchi et al., 2017). and merging into a highly seismic body between 90 and 150 km.
For the five above-mentioned populations of mantle earthquakes Thus, this study also discusses how such puzzling seismicity could be re-
(Fig. 2; Abers et al., 2013), constraints on earthquake location and ther- lated to mineral destabilizations within the subducting mantle of the
mal structure are independent, and hypocentre depth error is lower Nazca plate.
than 5 km for Cascadia (Abers et al., 2013; Parsons et al., 1998) and
only 2 km beneath northern Japan (e.g. Kita et al., 2010b; Zhao et al., 2. Geological and experimental contexts
1997) and Nankai (Tokai and Kii; e.g. Hirose et al., 2008). As shown on
Fig. 2, a large variability of P-T conditions at hypocentres exist, especially 2.1. Hydration and carbonation of oceanic peridotites
in the cold mantle slab northeastern Japan (Tohoku and Hokkaido).
Hypocentre P-T conditions presented in Fig. 2 show a partial correlation At least 25% of the oceanic floor, forming either at magma-poor
with major dehydration reactions, e.g. chrysotile and antigorite, i.e. ser- (slow spreading) or magma-rich (fast spreading) ridges could be only
pentine minerals. composed by serpentinized mantle (e.g. Cannat et al., 2010). At slow-
Natural serpentinites, mostly made of antigorite and other serpen- spreading oceanic trenches, the “seismic Moho” indicates the depth at
tine minerals, present significant chemical variations from place to which pervasive alteration of the peridotites stops, i.e. limit of high
place (e.g. John et al., 2011; Kendrick et al., 2011; Scambelluri et al., reaction-induced fracturation (e.g. Dunn et al., 2017). The seismic
2001, 2004; Trommsdorff and Evans, 1980), which have a significant Moho is usually not very deep, only 2–4 km below the seafloor near
impact on the structural properties of antigorite (Uehara and Shirozu, the ridge axis (Dunn et al., 2017; Escartín et al., 1997). But it varies
1985). Notably, the Al3+ content may significantly impact its stability and the impact of alteration on seismic velocities can reach about
by MgSi-AlAl Tschermak substitution (Bromiley and Pawley, 2003; 10 km below the seafloor, especially along normal faults, detachments
Padrón-Navarta et al., 2013). Chemical variations are also widely ob- and fracture zones (e.g. Canales et al., 2000). Along seismic faults this
served in other hydrous phases, such as humites (e.g. Trommsdorff and depth may significantly increase as fluid flows are favored by brittle de-
Evans, 1980; Evans & Trommsdorff, 1983). Talc composition does not formation. Thus maximum alternation, i.e. localized serpentinization, is
usually vary much from the end-member formula Mg3Si4O10(OH)2 deeper than the “seismic Moho” (Andreani et al., 2007).

Fig. 1. Dehydration-driven stress transfer from the lab to the slab. Geometrical relationships between serpentinites and peridotites both in the lab and in natural systems and implications
on earthquakes triggering: (a) Situation within experimental samples (adapted from Ferrand et al., 2017) and near the tip of deep serpentinized faults; (b) Situation within a subducting
slab (modified after Kita and Ferrand, 2018); (b) Recap of the up-to-date understanding of mantle earthquakes mechanisms within a subducting slab undergoing dehydration.
T.P. Ferrand / Lithos 334–335 (2019) 205–230 207

Fig. 2. Seismicity in the mantle of subducting slabs and stability of major hydrous phases. Diagram showing P-T conditions at hypocentres in the subducting mantle beneath Hokkaido,
Tohoku, Kii, Tokai and Cascadia as calculated by Abers et al. (2013). Curves show the stability limits of chrysotile (Chr), antigorite (Atg), talc (Ta) and clinochlore (Chl). For each of
these phases, colour shades indicate the destabilization P-T window depending on local variations of peridotite bulk composition in the MSH and/or MASH systems. For composition
and abbreviation of mineral phases, see Tables 1 and 2. For details see Figs. 4–7. For additional phases see Figs. 8–10. References: B.1986 = Berman et al. (1986); P.2003 = Pawley
(2003); P.2005 = Perrillat et al. (2005); S&S.1977 = Staudigel and Schreyer (1977); U&T.1999 = Ulmer and Trommsdorff (1999); W&S.1997 = Wunder and Schreyer (1997).

Carbon oxidizing conditions at depth in the oceanic lithosphere is Moho, respectively, through serpentinization of deep faults, which can
unknown, but carbonation could occur along with serpentinization, be extended to 35 km based on an estimate of the brittle-ductile transi-
for instance in deep hydrothermal systems within ophicarbonates of tion of the lithosphere (Ranero et al., 2003). Evidence of the direct link
mid-oceanic ridges (Escartín et al., 1997; Trommsdorff and Evans, between mantle hydration and the generation of dehydration-
1977). Experiments reveal that serpentinization triggers carbonation induced intermediate-depth seismicity is also demonstrated offshore
in CO2-rich systems (e.g. Klein and McCollom, 2013). In addition, field Alaska (Shillington et al., 2015). Such hydration up to the lower plane
observations reveal that the back-reaction of serpentinite with the is also consistent with full waveform modeling performed both for Hok-
COH-fluid promotes deep carbonation during subduction (Scambelluri kaido (Garth and Rietbrock, 2014) and Chile (Garth and Rietbrock,
et al., 2016). Major uncertainties exist regarding deep carbon cycles, 2017).
but recent studies about deep life reveal that serpentinites can host mi- Slower velocities in the lower-plane (Nakajima et al., 2009) have
crobial activity (Ménez et al., 2012), especially within hydrothermal been confirmed by high-resolution data (Bloch et al., 2018), which indi-
systems (Kelley et al., 2005; McCollom, 2007), which could lead to effi- cate a higher VP/VS ratio (≥2.0 ± 0.1) typical of free fluid. Bloch et al.
cient carbon storage within the hydrated lithospheric mantle. (2018) demonstrated a deep hydration of the lithospheric mantle of
Reaction-driven cracking is expected to play a critical role in the sus- the subducting Nazca plate up to the lower-plane and suggested that
tainability of fluid flow during either hydration or carbonation dehydration-induced fluids escape through a transient crack network
(e.g. Jamtveit et al., 2008; Jamtveit et al., 2009; Kelemen and Hirth, (porosity ~0.1 vol%). The higher VP/VS values at the lower plane most
2012; Rudge et al., 2010). Fluid migration is usually favored by likely indicate the presence of minor fluids that cannot immediately es-
preexisting heterogeneities within the uppermost mantle (e.g. Debret cape, whereas higher connectivity of/within serpentinized zones
et al., 2013). Moreover, neutron scattering techniques reveal nano- (Fig. 1) between the planes would allow permanent permeable path-
scale pore spaces within the structure of serpentine minerals that can ways, especially thanks to the structure of serpentine minerals
serve as fluid pathways (Tutolo et al., 2016), allowing further hydration (Schwarzenbach, 2016; Tutolo et al., 2016). Indeed, the abnormally
of ultramafic rocks (Schwarzenbach, 2016). high VP/VS ratio is only observed in the vicinity of the lower-plane and
not in the inter-plane aseismic region.
2.2. Extent of mantle hydration and carbonation in subduction context Carbonates are soluble in aqueous fluids at relatively low tempera-
tures (e.g. Caciagli and Manning, 2003; Facq et al., 2014; Frezzotti
Observational facts show evidence of deep faults at numerous sub- et al., 2011; Milesi et al., 2015) suggesting that, if hydrous fluids are in-
duction trenches (Cai et al., 2018; Ivandic et al., 2010; Naif et al., 2015, volved, efficient slab decarbonation should occur at relatively shallow
2016; Ranero et al., 2003; Shillington et al., 2015). At the Mariana sub- depths (Debret et al., 2018). But numerous studies prove that carbon
duction zone, ocean-bottom seismic data recently revealed that signifi- reaches great depth through subduction, in the form of graphite
cant serpentinization reaches at least 24 km below the Moho (Cai et al., (Pasteris, 1981) or aragonite inclusions found within olivine (N100 km
2018), consistently with the flexure model that predicts a neutral plane depth; Humphreys et al., 2010). Notably, dehydration fluids are charac-
at around 30 km below the seafloor (Emry et al., 2014). For other sub- terized by high oxygen fugacity (e.g. Debret and Sverjensky, 2017),
duction zones, despite the absence of substantially reduced seismic which slightly delays further dehydration reactions (Bretscher et al.,
velocities at 10 km and deeper (Reynard et al., 2010), seismic reflection 2018), especially via the change of iron oxidation state due to
surveys have revealed deep reflections that can be interpreted as serpentinizaion (e.g. Debret et al., 2015; Merkulova et al., 2016). Such
bending-related serpentinized faults at the Middle America Trench dehydration-induced oxidizing conditions would allow further carbon-
(Ivandic et al., 2010; Ranero et al., 2003) and offshore Alaska ation reaction, i.e. transfer of CO2 to greater depths.
(Shillington et al., 2015). These observations indicate that the litho- Most importantly, with increasing pressure and temperature, H2O is
spheric mantle is partially hydrated at least 8 km and 15km below the a supercritical fluid in case of H2O saturation (e.g. Ferrando et al., 2005).
208 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Fig. 3. An incomplete correlation between dehydration and lower-plane seismicity? (a) Sketch of a representative unit of fault surface and gouge in the MSH system considering only
antigorite and brucite, and associated P-T diagram questioning the “correlation” between dehydration reactions and lower-plane seismicity; (b) Situation in the MASH system
considering antigorite and clinochlore; (c) Suggested situation for a natural mantle vs natural seismicity. For references, see Fig. 2.

In such conditions, H2O is a solvent and CO2 may be a solute. But, as 2.3. Multiple destabilizations of hydrous phases within serpentinites
most of the hydrated mantle only contains few H2O, it is mainly
adsorbed at grain boundaries or integrated as defects within crystals Antigorite (Schweizer, 1840) is the high-temperature serpentine va-
(e.g. Bai and Kohlstedt, 1992; Martin and Donnay, 1972). In a riety and carries in its structure about 12–13 wt% of bound H2O on aver-
subducting slab, as long as H2O is preferentially integrated into hydrous age (e.g. Kunze, 1961; Mellini et al., 1987; Uehara, 1998; Uehara and
phases (b600 °C), slab decarbonation would occur only in case of H2O Kamata, 1994). It is a rock-forming mineral and the most abundant hy-
saturation, i.e. depending on hydrous phases connectivity and aqueous drous phase in the mantle undergoing subduction. The stability field of
fluids pathways. antigorite is thus commonly used to predict the depth where fluid is
T.P. Ferrand / Lithos 334–335 (2019) 205–230 209

released in subduction zones (e.g. Hacker et al., 2003; Peacock, 2001; 2.4. Decades of debate about (de)serpentinization and earthquakes
Schwartz et al., 2013). A first-order model of serpentinized fault in the mechanisms
MSH system would be an antigorite gouge with variable thickness, but
it would only correlate with part of the seismological observations The link between the dehydration of serpentinites and mantle seis-
(Fig. 3a). micity has been a matter of debate for decades (e.g. Abers et al., 2013;
Actually, a large number of hydrous phases may coexist with serpen- Ferrand et al., 2017; Frohlich, 2006; Hacker et al., 2003; Incel et al.,
tine minerals within serpentinites (Section 4.1), such as talc (4.7 wt% 2017; Kirby et al., 1996; Raleigh and Paterson, 1965). Potential mecha-
H2O, e.g. Berman et al., 1986; Pawley and Wood, 1995) or chlorite nisms are recapped in Fig. 1c. In the followings, I distinguish between
(12.97 wt% H2O, e.g. Staudigel and Schreyer, 1977; Pawley, 2003). triggering and nucleation mechanisms.
Table 1 lists the hydrous phases (Section 4.1) that are likely within Embrittlement upon dehydration of deforming antigorite occurs in
serpentinites, at least in small amounts and for limited P-T conditions. the laboratory at low pressure (0.5 GPa, i.e. 16 km) and is the root of
The natural complexity of serpentinized faults is questioned in this the “dehydration embrittlement” model (Raleigh and Paterson, 1965).
study, such as phases diversity and fluid pathways (Fig. 3b,c). In this model, fluid overpressure due to dehydration drives brittle yield-
One of the most important differences between hot and cold ing (e.g. Frohlich, 2006; Hacker et al., 2003). Fluid overpressure is also
subducting slabs is that major dehydration reactions, e.g. serpentine proposed to trigger high-pressure hydrofracturing in the cold top-slab
minerals, are respectively associated with positive and negative volume mantle (Padrón-Navarta et al., 2010). The limitations of such fluid-
change (solid + fluid; e.g. Hacker et al., 2003; Abers et al., 2013; Gasc induced processes have been discussed in several studies (e.g. Ferrand
et al., 2011; Incel et al., 2017; Ferrand et al., 2017). Nevertheless, nega- et al., 2017; Thielmann et al., 2015).
tive volume change also occurs within the mantle of hot subducting Consequently, the weakening of dehydrating serpentinites, i.e.
slabs, for instance during the dehydration of talc or chlorite (Fig. 2). mostly rock-forming antigorite and other hydrous phases, has been

Table 1
Hydrous phases and associated formula considered in this study.

CaO-MgO-Al2O3-SiO2-H2O system (CMASH) for pressures between 0.5 and 6.5 GPa. DHMS refers to “Dense Hydrous Magnesium Silicates”. For carbonates, see Table 3. Except for goethite,
iron is not indicated in the table. The substitution of Mg2+ by Fe2+ in magnesium silicates is common, but the iron content does not exceed 1%, which, for most phases, has minor impact on
stability. Besides, Fe3+ and Ti3+ may exist in hydrous silicates, especially replacing Al3+ in Al-bearing minerals. As titanium and fluorine have a major impact on humites stability (see Fig. 9),
Ti3+ and F− are indicated for chondrodite and clinohumite. In most mantle peridotites, hydrous phases are minor phases. P-T diagrams that direclty deal with the stability of a phase are
indicated with a bold number. When numbers are not bold, it means that the phase is involved in a reaction in a P-T diagram that does not focus on this phase.
210 T.P. Ferrand / Lithos 334–335 (2019) 205–230

considered as the cause for dynamic rupture nucleation in the mantle the amount and the distribution of antigorite in slightly/locally
(e.g. Hacker et al., 2003). However, at mantle pressures (N1 GPa), serpentinized peridotite bodies (Ferrand et al., 2017).
dehydrating antigorite is not seismic (Chernak and Hirth, 2010, 2011; To understand the complex phenomenon of transformation-driven
Gasc et al., 2011, 2017; Okazaki and Hirth, 2016). Nevertheless, earthquakes, two models could be integrated: the thermal runaway
dehydration-driven embrittlement is observed in serpentinized perido- model (Kelemen and Hirth, 2007) and the stress transfer model
tites at subducting-mantle P-T conditions (e.g. Ferrand et al., 2017; Jung (DDST, Ferrand et al., 2017). In other words, dynamic weakening, re-
et al., 2004; Jung and Green, 2004; Xia, 2013). quired for rupture at depth, could be operated by thermal runaway,
These results are confirmed by field observations, where pseu- which would be triggered by local antigorite dehydration in the vicinity
dotachylytes, i.e. fossilized earthquakes, are observed in large dry peri- of the future fault (Fig. 1).
dotite bodies (Ferrand et al., 2018; Scambelluri et al., 2017) or
partially/locally serpentinized peridotites (Andersen and Austrheim, 2.5. About the upper Wadati-Benioff plane
2006; Andersen et al., 2014; Deseta et al., 2014a, 2014b; Dunkel et al.,
2017). Concurrently with these observations of brittle features in peri- Salty brines have been found in eclogitized hydrous peridotites from
dotite, only ductile deformation is reported in dehydrating serpentinites the Erro-Tobbio massif, i.e. former plate interface, which suggests that
(Plümper et al., 2017), consistently with experiments (e.g. Chernak and significant amounts of Cl, Na and K could be incorporated into peridotites
Hirth, 2010, 2011). The Erro Tobbio serpentinite of the Voltri massif, during subduction (Scambelluri et al., 1997). Then, many other hydrous
which reached 100 km depth (i.e. 3 GPa) 30 million years ago (e.g. minerals may exist within metasomatized peridotites, such as Na- and
Scambelluri et al., 1995; Scambelluri et al., 2004), shows dehydration K-bearing minerals, e.g. pargasite NaCa2Mg4Al3Si6O22(OH)2 (e.g. Lykins
features organized as anastomosed channels (Plümper et al., 2017). and Jenkins, 1992) and phlogopite KMg3AlSi3O10(OH)2 (e.g. Fumagalli
The dehydration reaction in serpentinites seems to drive ductile defor- et al., 2009). These minerals may form within serpentinites but only in
mation, and no connexion between dehydrating serpentinite bodies near-surface conditions, when peridotite reaches the seafloor. Similarily,
and faults could be observed on the field. as lawsonite CaAl2Si2O7(OH)2·H2O and zoizite Ca2Al3[Si2O7][SiO4]O(OH)
Another mechanism, which would not require the presence of hy- (Schmidt and Poli, 1994) do not contain Mg and are generally not found
drous phases, has been proposed to explain the double Wadati-Benioff within peridotites, these minerals are not considered in this study.
structure, especially concerning the lower plane (Reynard et al., 2010). In- From a petrological point of view, the lower Wadati-Benioff plane is
deed, until recently (Bloch et al., 2018; Cai et al., 2018), seismic imaging the simpliest part of the double-seismic zone, as it consists at first order
has generally documented the absence of substantially reduced velocities of locally and/or slightly hydrated peridotite. Considering this relatively
deep below the oceanic Moho, thus demonstrating little hydration if any simple composition, it is likely that this seismicity has only one trigger-
(Reynard et al., 2010). Earthquakes would be triggered by the nucleation ing mechanism. As long as we do not know whether there is oceanic
of self-sustained shear-heating instabilities (John et al., 2009; Kelemen crust or not and/or how many triggering mechanisms exist near the
and Hirth, 2007; Prieto et al., 2013), as supported by recent high- plate interface, I do not focus on the upper plane. Experiments have
pressure deformation experiments on dunite (Ohuchi et al., 2017). shown that mineral destabilizations within the oceanic crust, such as
Shear heating in dry ultramafic mylonitic layers would lead to viscous lawsonite and glaucophane dehydration, trigger seismicity (e.g. Incel
rupture between 600 and 800 °C (Kelemen and Hirth, 2007). It is however et al., 2017; Okazaki and Hirth, 2016). The numerous dehydration and
important to note that, such thermal runaway (John et al., 2009; Kelemen other metamorphic reactions that may occur within the oceanic crust
and Hirth, 2007) must be preceded by an initial grain size reduction should be recapped, but it is not the point of this study.
(Thielmann, 2018a), easily obtained during localized mineral transforma-
tions (Green II et al., 2015; Thielmann, 2018b). In the laboratory (Ohuchi 2.6. Experimental stability curves vs pseudosections
et al., 2017) the shear strength of the dunite sample is reached at sample
scale; for natural intraslab earthquakes, a trigger is needed, and mineral It should be recalled here that the experimental stability curves de-
transformations are good candidate for it. A thermal runaway process tailed in this review fundamentally differ from pseudosections. Experi-
that would be associated with mineral transformations, e.g. antigorite de- mental P-T curves are deduced from direct observations of samples
hydration, could participate in paving the way for reconciliation of these after experiments mimicking the geological processes; pseudosections,
apparently contradictory studies. Nonetheless, the endothermique nature constrained from experimental results, are based on thermodynamic
of dehydration reactions would constitute a negative feedback in models calculations in equilibrated systems. Used in many petrological and
considering thermal runaway activated by frictional heating (e.g. Brantut mineralogical studies (e.g. Bretscher et al., 2018; Fumagalli and Poli,
et al., 2011; Yamashita and Schubnel, 2016). 2005; Padrón-Navarta et al., 2013), pseudosections are a powerful tool
The impact of sliding on a fault segment on the stress state along the to understand metamorphic parageneses.
rest of the fault and along neighbouring faults varies depending on their Nonetheless, pseudosections only consider phases that are defined as an
geometries, the lithologies they go through and the fluids that circulate input in the calculation. Moreover, considering relatively low temperatures
in their gouges (e.g. Guéguen and Palciauskas, 1994). Mineral reactions (b600 °C) in subduction context and variable amount of minor hydrous/
necessarily induce grain size reduction, which could drive nanometric carbonated phases within fault gouges (Fig. 3), local equilibria within
flow and cause extreme weakening during fault slip (Green II et al., serpentinized faults should be considered. In other words, rocks are not
2015). Local transformation and associated grain size reduction may expected to be equilibrated at the fault scale. In sake of clarity, this paper
also drive stress transfers and trigger earthquakes (Ferrand et al., contains only experimental P-T curves and no pseudosection. It leaves
2017). It has also been noted that slow seismic events may lead to stress also the door open to potential undiscovered or misunderstood phases,
transfers triggering large earthquakes in Japan (Obara and Kato, 2016). such as high-pressure hydrated talc phases (Section 4.1.1), with various
Recent experimental results (Ferrand et al., 2017) revealed that H2O contents, which could consist of a polysomatic series (Section 4.2.4.2).
antigorite destabilization is the trigger for dynamic failure in slightly
serpentinized peridotites (5 vol% antigorite), leading to the radiation 3. Methods and results
of elastic waves and pseudotachylytes formation (Ferrand et al.,
2017). These laboratory observations lead to the establishment of a 3.1. Review of all experimental studies on the stability of mantle minerals
new phenomenological model, the DDST (Dehydration-Driven Stress
Transfer), which may describe how intraslab mantle earthquakes are lo- Stability limits of a wide spectrum of major and minor mantle phases
cated along the LWBP. The latter seismicity zone is neither a tectonic are presented in a series of P-T diagrams (Figs. 4–11). These diagrams
contact nor a thermal limit. Its position seems mainly controlled by summarize prograde stability limits and take into account potential
T.P. Ferrand / Lithos 334–335 (2019) 205–230 211

large chemical variations due to fluids. Based on geophysical imaging and Trommsdorff, 1999; Wunder, 1998), field observations (e.g.
studies (Ivandic et al., 2010; Ranero et al., 2003; Reynard et al., 2010; Trommsdorff, 1983; Chopin, 1984; Schwartz et al., 2013) and associated
Shillington et al., 2015), experimental datasets (Bromiley and Pawley, thermodynamic considerations (e.g. Berman et al., 1986; Wunder and
2003; Chinnery et al., 1999; Evans, 1977; Evans, 2004; Evans et al., Schreyer, 1997; Holland and Powell, 1998; Hilairet et al., 2006), I con-
1976; Fockenberg, 1995; Fransolet and Schreyer, 1984; Fumagalli sider a large spectrum of phases in the MSH and MASH systems
et al., 2014; Guggenbuehl, 1994; Johannes, 1968; Massonne et al., (Figs. 4–9), with a wide range of variations both in their chemical com-
1981; Merkulova et al., 2016; Pawley, 2003; Pawley and Wood, 1995, positions and possible paragenesis. Transformations in anhydrous peri-
1996; Pawley and Wood, 1996; Perrillat et al., 2005; Schreyer, 1988; dotite are also considered (Fig. 10), as well as decarbonation reactions
Staudigel and Schreyer, 1977; Ulmer and Trommsdorff, 1995; Ulmer and other carbonate mineral stability limits (Fig. 11).

Fig. 4. Stability of antigorite on a prograde metamorphic path and sketch of antigorite lattice. (a) P-T diagram presenting up-to-date knowledge on the stability limit of antigorite up to
6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry and H2O saturation as expected in a subduction context. Background coloured dots
correspond to P-T conditions at hypocentres of intermediate-depth mantle earthquakes for well-studied subducting slabs beneath Hokkaido, Tohoku, Kii, Tokai and Cascadia (see
Fig. 2). For composition and abbreviation of mineral phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations and thermodynamic
equilibrium numerical studies, see the text. References: B.1986 = Berman et al. (1986); B&N.1998 = Bose and Navrotsky (1998); B&P.2003 = Bromiley and Pawley (2003); E.1976 =
Evans et al. (1976); E.2004 = Evans (2004); H.2006 = Hilairet et al. (2006); P.2005 = Perrillat et al. (2005); U&T.1999 = Ulmer and Trommsdorff (1999); W&S.1997 = Wunder and
Schreyer (1997). (b) Antigorite crystal lattice 3D structure (m = 17; adapted from Uehara, 1998). Silica tetrahedra SiO4− 4 (lilac) share three of their four oxygen atoms (red) with each
other, i.e. “silica layer”, and share one with the octahedral sites, i.e. “brucite layer”. Ideally, the latter are occupied by bivalent cations (orange spheres, mostly Mg2+ and Fe2+). Protons
+ −
(H ) of the hydroxyles (OH ), represented with small blue dots, connect the silica and brucite layers (hydrogen bond). For sake of clarity, the geometry of the brucite layer is
represented separately, showing the six oxygen atoms at the octahedron edges.
212 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Fig. 5. Stability of chrysotile, lizardites and amesite on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limit of serpentine minerals chrysotile,
lizardites and amesite up to 6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry and H2O saturation as expected in a subduction context. See
Fig. 2 for the background coloured dots. The stability of amesite is not certain as it and clinochlore could get mixed up (Ackermand et al., 1975). For composition and abbreviation of mineral
phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations and thermodynamic equilibrium numerical studies, see the text. References: A.1975 =
Ackermand et al. (1975); B.1986 = Berman et al. (1986); C&C.1979 = Caruso and Chernosky (1979); C.1973a = Chernosky (1973a); C.1973b = Chernosky (1973b); E.1976 = Evans et al.
(1976); E.1977 = Evans (1977); E.2004 = Evans (2004); J.1968 = Johannes (1968); P; S.2013 = Schwartz et al. (2013); T.1983 = Trommsdorff (1983); U&T.1999 = Ulmer and
Trommsdorff (1999).

Antigorite is the most studied hydrous magnesium silicate and the was previously done for other subduction zones (Abers et al., 2013).
most common hydrous phase in the subducting mantle. Its stability The thermal structure calculated by Wada and Wang (2009), at latitude
limits are shown in Fig. 4a. Antigorite lattice structure is detailed in 23.5°S, is the most accurate up to date for this region. It gives similar re-
Fig. 4b, with description of “silica” and “brucite” layers, often referred sults as the model of van Keken et al. (2011) for a given slab geometry
to in descriptions of phyllosilicates such as serpentine, talc and chlorite and boundary conditions (Ikuko Wada, personal communication). The
minerals. Documented reactions involving other serpentine minerals thermal model of Wada and Wang (2009) is also consistent with atten-
(chrysotile, lizardites and amesite) are summarized in Fig. 5. The stabil- uation tomography results (Schurr et al., 2006).
ity limits of chlorite and Mg-chloritoid are summarized in Fig. 6. Most Down to 120 km depth, a constant dip of the subduction interface of
reactions involving talc and HP hydrous talc-like minerals are presented 21° is considered and well consistent with both the slab surface imaging
in Fig. 7, except reactions involving chlorite or Mg-chloritoid (Fig. 6). and the relocated seismicity (Sippl et al., 2018). At great depths, how-
The stability of Phase A and amphiboles are presented in Fig. 8. Reac- ever, a change occurs in the slab geometry between latitudes 22°S and
tions involving other Mg- and/or Al-rich hydrous silicates are presented 22.5°S (Sippl et al., 2018), which constrains to consider an additional
in Fig. 9. Complementally, Fig. 10 summarizes reactions of anhydrous bend of 14°, leading to a total dip of 35° from 120 to 200 km depth.
phases, some of which typical of ultramafic rocks (e.g. forsterite, In Fig. 13, I present the cross-section at latitude 21.5°S with results of
enstatite, diopside), some other possibly locally present due to fluid cir- P and T estimates for the case of an extrapolation of the thermal struc-
culation at faults and metasomatism (e.g. Bonnemains et al., 2017). Fi- ture with constant dip of 21° (Fig. 13a and b), and considering the addi-
nally, as the peridotite is likely to be partially carbonated, at least in tional bend at 120 km depth (Fig. 13c and d). I consider no change in the
the first kilometres below the Moho, decarbonation reactions and thermal structure of the slab between these two scenarios. For compar-
other carbonate mineral stability limits are summarized in Fig. 11. ison, the same exercise at latitude 23°S, close to the determined thermal
Additional works on antigorite polysomatism and on the impact of structure (23.5°S), is displayed in Fig. 13e and f.
H2O saturation on the stability of antigorite are recapped in Fig. 12. Errors for hypocentre relocation do not exceed 2 km (Sippl et al.,
All the references used are specified in captions and discussed in the 2018), which means that pressure estimates are accurate with uncer-
following sections. Mineral compositions and abbreviations considered tainties lower than 60 MPa. In contrast, the temperature estimates are
in this study are listed in Table 1, 2 and 3 respectively for hydrous strongly affected by uncertainties about the thermal structure of the
phases except carbonates, anhydrous phases except carbonates and car- subducting slab. For a given thermal structure, the T estimates are ob-
bonate minerals. tained considering a linear T increase between the displayed isotherms.

3.2. Estimate of the P-T conditions at hypocentres beneath northern Chile 4. Discussion

In this study, I use high-quality relocation data recently published for Major advances have been made in the understanding of mineralogical
northern Chile (Sippl et al., 2018), at latitude 21.5°S and 23°S, to esti- diversity and stability limits for various bulk compositions thanks to ex-
mate the P-T conditions at hypocentres in the subducting mantle as tended experimental studies and associated thermodynamic calculations
T.P. Ferrand / Lithos 334–335 (2019) 205–230 213

Fig. 6. Stability of chlorite and Mg-Chloritoid on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limit of chlorite (clinochlore and sudoite) and
Mg-chloritoid up to 6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry and H2O saturation as expected in a subduction context. See Fig. 2 for
the background coloured dots. For composition and abbreviation of mineral phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations
and thermodynamic equilibrium numerical studies, see the text. When not specified, the curves are adapted from Chopin and Schreyer (1983). References: A.1975 = Ackermand et al.
(1975); C.1984 = Chopin (1984); E.1976 = Evans et al. (1976); E.1977 = Evans (1977); E.2004 = Evans (2004); F&S.1984 = Fransolet and Schreyer (1984); F.1995 = Fockenberg
(1995); F.1998a = Fockenberg (1998a); F.1998b = Fockenberg (1998b); F.2014 = Fumagalli et al. (2014); G.1994 = Guggenbuehl (1994); J.1968 = Johannes (1968); M.1981 =
Massonne et al. (1981); P&W.1995 = Pawley and Wood (1995); P.2003 = Pawley (2003); S.1988 = Schreyer (1988); S.2013 = Schwartz et al. (2013); S&S.1977 = Staudigel and
Schreyer (1977); T.1983 = Trommsdorff (1983); U&T.1999 = Ulmer and Trommsdorff (1999).

in the MSH system (e.g. Pawley and Wood, 1995; Wunder and Schreyer, the importance of these stability fluctuations on the relevance and ap-
1997; Chinnery et al., 1999; Pawley, 2003). An important parameter con- plicability of transformation-driven stress transfers such as the one evi-
trolling the stability of major hydrous phases appears to be the bulk com- denced by recent experimental results (DDST; Ferrand et al., 2017). To
position of the peridotite, e.g. the presence or absence of brucite shifting do so, this paper deliberately puts the emphasis on well-studied and
the stability of serpentine minerals by 100–150 °C and 2–3 GPa (Figs. 4 widely-observed phases as much as on rarer or less observed minerals.
and 5). The fluids also significantly impact serpentinization, especially re-
action kinetics (e.g. Pens et al., 2016). Additional work in the MASH system 4.1. A diversity of phases with abrupt stability limits
(e.g. Chollet et al., 2009; Bromiley and Pawley, 2002, 2003) and comple-
mentary studies considering additional minor amounts of cations (e.g. 4.1.1. A continuous discovery of new phases in transforming serpentinites
Fe2+, Fe3+, Cr3+ and Ti3+; e.g. Ulmer and Trommsdorff, 1995; Fumagalli Many studies have tried to refine thermodynamic calculations to
et al., 2014) or anions (e.g. F−, Cl−; e.g. Weiss, 1997) allow to better con- fit experimental data in internally consistent datasets (e.g. Berman
strain the stability of natural phases. Both deep H2O and CO2 cycles have et al., 1986; Berman, 1988; Holland and Powell, 1998; Wunder and
been reappraised in light of experiments, geological observations and ther- Schreyer, 1997; Wunder, 1998) and this work is still progressing. Fur-
modynamic calculations (e.g. Irifune et al., 1998; Comodi et al., 2007; ther updates of reaction curves positions are likely. This review shows
Scambelluri et al., 2016; Dunn et al., 2017; Cai et al., 2018; Sverjensky a first-order general agreement between all the experimental works
et al., 2014). Complementary studies explored deep fluid processes and el- which have been conducted since the 70s, as illustrated by the series
ements cycling especially through antigorite burying and deep destabiliza- of P-T diagrams (Figs. 4–11).
tion (Scambelluri et al., 1997, 2001, 2004; John et al., 2011). Section 4.1 Below 6 GPa, antigorite is the most abundant hydrous phase in the
summarizes the state of the art in the MSH and MASH systems, and ad- mantle between 300 and 700 °C (Fig. 4). It is the rock-forming mineral
dresses the question of “metastability”. of serpentinites, which may include various other phases. Serpentinite
In Section 4.2, I highlight several parameters that undoubtedly influ- volumes transport large amounts of water into the mantle (Irifune et al.,
ence the position of reaction curves in the P-T diagrams presented above 1998). During its transportation at depth, serpentine minerals transfer
(Figs. 4–11). The parameters are (4.2.1) bulk composition, (4.2.2) min- water to other phases of the MSH system such as talc (e.g. Pawley and
eral chemistry, (4.2.3) fluid composition and saturation level, (4.2.4) Wood, 1995; Bose and Navrotsky, 1998; Bromiley and Pawley, 2003;
mineral structure and (4.2.5) structural interactions between phases. Figs. 4a and 5), HP hydrated talc (i.e. 10-Å and talc-like phases, e.g.
Taken all together, these factors are likely to allow substantial variations Chinnery et al., 1999; Fumagalli and Poli, 2005; Comodi et al., 2007) or
in the stability of either major or minor mantle phases. Finally, I discuss phase A (e.g. Pawley and Wood, 1996; Chinnery et al., 1999; Chollet
214 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Fig. 7. Stability of talc and 10-Å phase on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limit of talc and 10-Å phase up to 6 GPa, i.e. 200 km.
Reactions are indicated with coloured curves referring to different chemistry and H2O saturation as expected in a subduction context. See Fig. 2 for the background coloured dots. For
composition and abbreviation of mineral phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations and experiments-based numerical
studies, see the text. For sake of clarity, the stability of talc in the presence of chlorite or chloritoid is only drawn on Fig. 6. When not obvious, small black arrows refer to the reaction
sense as indicated in the reactions list. References: B.2000 = Bailey and Holloway (2000); B.1986 = Berman et al. (1986); C.1999 = Chinnery et al. (1999); C.2009 = Chollet et al.
(2009); C.1984 = Chopin (1984); E.1977 = Evans (1977); G.1984 = Guggenbuehl (1994); K.1966 = Kithara et al. (1966); P.1998 = Pawley (1998); P&W.1995 = Pawley and Wood
(1995); U&T.1999 = Ulmer and Trommsdorff (1999); W&S.1997 = Wunder and Schreyer (1997); Y&A.1977 = Yamamoto and Akimoto (1977).

Fig. 8. Stability of phase A and Mg-amphiboles on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limit of phase A, anthophyllite,
cummingtonite, tremolite and tschermakite up to 6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry. See Fig. 2 for the background
coloured dots. For composition and abbreviation of mineral phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations and experiments-based
numerical studies, see the text. When not obvious, small black arrows refer to the reaction sense as indicated in the reaction list. References: B.1986 = Berman et al. (1986);
B&N.1998 = Bose and Navrotsky (1998); H.2006 = Hilairet et al. (2006); J.1983 = Jenkins (1983); P&W.1995 = Pawley and Wood (1995); P&W.1996 = Pawley and Wood (1996);
T.1983 = Trommsdorff (1983); U&T.1999 = Ulmer and Trommsdorff (1999); W&S.1997 = Wunder and Schreyer (1997); W.1998 = Wunder (1998).
T.P. Ferrand / Lithos 334–335 (2019) 205–230 215

Fig. 9. Stability of other relevant hydrous minerals on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limits of brucite, diaspore, clinohumites,
chondrodites, staurolite, carpholite and Sursassites up to 6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry. See Fig. 2 for the background
coloured dots. For composition and abbreviation of mineral phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations and experiments-based
numerical studies, see the text. When not obvious, small black arrows refer to the reaction sense as indicated in the reaction list. When not specified, the curves are adapted from
Chopin and Schreyer (1983). The gray shaded region corresponds to the reaction 2 Gth → Hmt + H2O (FO system; Majzlan et al. (2003)). References: B.1986 = Berman et al. (1986);
F.1998a = Fockenberg (1998a); F.1998b = Fockenberg (1998b); P&W.1996 = Pawley and Wood (1996); P.2000 =Pawley (2000); U&T.1999 = Ulmer and Trommsdorff (1999);
W.1997 = Weiss (1997); W&S.1997 = Wunder and Schreyer (1997); W.1998 = Wunder (1998).

Fig. 10. Stability of minerals in anhydrous peridotites on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limits of anorthite, spinel, garnets,
quartz, enstatite, corundum, kyanite and sapphirine in unhydrous peridotites up to 6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry. See
Fig. 2 for the background coloured dots. For composition and abbreviation of mineral phases, see Tables 1 and 2. For details about the catalogue of published experimental investigations
and experiments-based numerical studies, see the text. When not obvious, small black arrows refer to the reaction sense as indicated in the reaction list. References: A.1975 = Ackermand
et al. (1975); B.1986 = Berman et al. (1986); B.1991 = Bohlen et al. (1991); H&P1998 = Holland and Powell (1998); H&E1971 = Hensen and Essene (1971); M&K.1998 = Morishita and
Kodera (1998); M.2001 = Morishita et al. (2001); O.1976 = Obata et al. (1976); O'H.1971 = O'Hara et al. (1971); U&S.2001 = Ulmer and Stalder (2001).
216 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Fig. 11. Stability of carbonate minerals in peridotites on a prograde metamorphic path. P-T diagram presenting up-to-date knowledge on the stability limits of carbonated peridotites up to
6 GPa, i.e. 200 km. Reactions are indicated with coloured curves referring to different chemistry. See Fig. 2 for the background coloured dots. For composition and abbreviation of mineral
phases, see Tables 2 and 3. For details about the catalogue of published experimental investigations and experiments-based numerical studies, see the text. When not obvious, small black
arrows refer to the reaction sense as indicated in the reaction list. The impact of CO2 saturation within the aqueous fluid is indicated for some decarbonation reactions in the CMSH system
(light and dark green; Tenthorey and Cox, 2003; Trommsdorff and Evans, 1977) and in the CMS system (light and dark gray lines after Zhou and Hsu, 1992). The shaded gray area refers to
all plausible positions of the stability curves in wet natural systems. Purple lines are contours of log10mCaCO 3 (calcite concentration in aqueous fluid) as a function of P and T, without
considering the calcite-aragonite transition (Caciagli and Manning, 2003). References: J&P.1971 = Johannes and Puhan (1971); K.1975 = Kushiro et al. (1975); M.1996 = Martinez
et al. (1996); W&H.1976 = Wyllie and Huang (1976); Z&H.1992 = Zhou and Hsu (1992).

et al., 2009; Fig. 8). Especially, 10-Å phase, i.e. a HP hydrated talc bearing stability field (Comodi et al., 2007). Other hydrated talc phases (e.g.
13.6 wt% H2O (Yamamoto and Akimoto, 1977), has been identified as a Chinnery et al., 1999; Perrillat et al., 2005) have been discovered experi-
good candidate for water transfer into the mantle beyond the antigorite mentally. Such HP phyllosilicates are inexistent or extremely rare on the

Fig. 12. Antigorite metasomatism and influence of H2O saturation on its stability. P-T diagram presenting the evolution of antigorite a-axis wavelength m (green numbers and lines;
Wunder et al., 2001) and stability limit depending on H2O saturation (blue numbers and lines; Perrillat et al., 2005). See Fig. 2 for seismicity (background coloured dots). See Fig. 4 for
references of reaction curves (black lines).
T.P. Ferrand / Lithos 334–335 (2019) 205–230 217

Table 2
Anhydrous phases and associated formula considered in this study.

CaO-MgO-Al2O3-SiO2 system (CMAS), except carbonates (see Table 3), for pressures between 0.5 and 6.5 GPa. Here the Fe2+ content is only indicated for hematite, olivine and ilmenite, but
minor substitution of Mg2+ by Fe2+ is common in all magnesium silicates. The last column indicates in which figures the phases are involved. P-T diagrams that direclty deal with the
stability of a phase are indicated with a bold number. When numbers are not bold, it means that the phase is involved in a reaction in a P-T diagram that does not focus on this phase.

Earth surface, as they are prone to localize deformation, which prevents Å(0.65)” in Table 1) could actually be one of the so-called “talc-like” phases.
preservation throughout exhumation. For instance, a “talc-like” phase The relative stabilities of HP hydrated talc phases should depend, as for
was initially observed thanks to in-situ X-ray diffraction during HP-HT ex- serpentines and talc, on H2O saturation (e.g. Perrillat et al., 2005;
periments investigating the stability limit of antigorite (Perrillat et al., Hilairet et al., 2006) and SiO2 activity (e.g. Worden et al., 1991;
2005; Fig. 4), and a similar phase was recently observed on the field Manning, 1995).
(Plümper et al., 2017). The 10-Å phase has an intermediate composition In the MSH system, it seems that a large diversity of high-grade
and structure between talc and vermiculite (Pawley and Wood, 1995). phases exists, some of which having close composition or structure.
The “talc-like” phase has an intermediate composition between talc and Most of the dense hydrous phases would be stable in a P-T range of a
the 10-Å phase (Perrillat et al., 2005) and should have intermediate struc- few GPa and some hundreds of degrees. In addition, hot slabs may en-
ture. The water-poor 10-Å phase evidenced by Chinnery et al. (1999; “10- dure seismicity due to Mg-amphiboles destabilization (e.g.

Table 3
Carbonate phases and associated formula considered in this study.

Carbonate phases in the CaO-MgO-Al2O3-SiO2-CO2-H2O system for pressures between 0.5 and 6.5 GPa. The mentioned hydrated carbonates are stable in serpentinite, but their stability
limits have not been determined yet. P-T diagrams that direclty deal with the stability of a phase are indicated with a bold number. When numbers are not bold, it means that the
phase is involved in a reaction in a P-T diagram that does not focus on this phase.
218 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Fig. 13. Seismicity in the subducting mantle beneath northern Chile. (a) Relocated events from 2007 to 2014 on a W-E profile 21.5°S after Sippl et al. (2018), with thermal model from
Wada and Wang (2009) with extrapolation from 150 to 200 km depth considering a constant slab dip of 21°; (b) P-T diagram showing conditions at hypocentres within the subducting
mantle according to section (a); (c) Same as (a) considering an additional bend of 14° from the centre of the highly seismic body, consistently with the slab geometry at that latitude (Sippl
et al., 2018) as confirmed by attenuation tomography (Schurr et al., 2006); (d) P-T diagram showing conditions at hypocentres within the subducting mantle according to section (c);
(e) W-E cross-section at latitude 23°S; (f) P-T diagram for section (e). For the cross-sections, the horizontal axis indicates the distance from the trench. On the P-T diagrams, the curves
show the stability limits of chrysotile (Chr), antigorite (Atg), talc (Ta) and clinochlore (Chl). For each of these phases, colour shades indicate the destabilization P-T window depending
on local variations of peridotite bulk composition in the MSH and/or MASH systems. For composition and abbreviation of mineral phases, see Tables 1 and 2. For details see Figs. 4–7.
For additional phases see Figs. 8–10. References: B.1986 = Berman et al. (1986); P.2003 = Pawley (2003); P.2005 = Perrillat et al. (2005); S&S.1977 = Staudigel and Schreyer
(1977); U&T.1999 = Ulmer and Trommsdorff (1999); W&S.1997 = Wunder and Schreyer (1997).

Greenwood, 1963; Berman et al., 1986; Evans and Ghiorso, 1995; Fig. 8); 4.1.2. Diversity of phases within serpentinites
in cold slabs, humites could be involved (Akimoto et al., 1977; Serpentinite bodies contain additional phases, such as serpentine-
Trommsdorff and Evans, 1980; Evans and Trommsdorff, 1983; Dymek like hydrated inosilicates, i.e. carlosturanite (Compagnoni et al., 1985;
et al., 1988; Weiss, 1997; Fig. 9). 17.14 wt% H2O), and balangeroite (Compagnoni et al., 1983;
In nature, mostly due to chemical variability (see Section 4.2), even Bonaccorsi et al., 2012; 11.95 wt% H2O). Depending on bulk composition
more phases exist, a significant number of which may have not been and H2O saturation, carlosturanite forms in the stability field of
discovered yet. In this study I especially emphasize the MASH system, antigorite through the reaction 6 Chr + 3 Br + 3 H2O → Crl and
in which many phases have stability limits fitting seismological balangeroite through the reaction 8 Chr + 18 Br → Bal + 14 H2O
observations, e.g. chlorite (e.g. Staudigel and Schreyer, 1977; Pawley, (Evans, 2004). The stability limits of these minerals are unknown but
2003; Fumagalli et al., 2014), Mg-Chloritoid (Vidal et al., 1994), Mg- they have been evidenced as stable together with antigorite and chrys-
Carpholite (Viswanathan, 1981; Ferraris et al., 1992; Vidal et al., otile (Ferraris et al., 1987; Groppo and Compagnoni, 2007). Even if they
1992), Mg-Staurolite (Fockenberg, 1998a), Yoderite (Schreyer, 1988), are rarely observed in natural samples, we cannot exclude that such
and Mg-Sursassites (also known as MgMgAl-pumpellyite; Schreyer phases could reach higher ratios at depth within H2O-saturated
et al., 1991; Fockenberg, 1998b; Bromiley and Pawley, 2002). Except serpentinized faults.
chlorite, these phases of the MASH system have not been reported in Other serpentine minerals may also share part of the stability field of
natural HP chlorite-bearing harzburgites that formed via serpentinite chrysotile, lizardite and antigorite, such as polygonal serpentine (e.g.
dehydration (Evans and Trommsdorff, 1978; Trommsdorff et al., Mellini, 1986; Mitchell and Putnis, 1988; Viti, 2010; Baronnet and
1998). It is important to recall here that, as most phases are involved Devouard, 1996), conical serpentine and polyhedral serpentine
in retrograde reactions during exhumation processes (e.g. Figs. 6 and (Andreani et al., 2008; Cressey et al., 2008). These serpentine minerals
9), they could be rare on outcrops and still have an important role at were found stable within antigorite-rich rocks (e.g. Banfield et al.,
depth. 1995) but their stability limits are unknown.
T.P. Ferrand / Lithos 334–335 (2019) 205–230 219

Bulk composition (Section 4.2.1), mineral chemistry (Section 4.2.2) 4.2.1.1. Stability of each phase depending on bulk composition. In the MSH
and fluids (Section 4.2.3) are known as the key parameters controlling system, antigorite dehydration occurs at pressures and temperatures
relative stabilities of these phases within serpentinites, along with respectively ≈2 GPa and ≈150 °C lower in brucite-saturated condi-
antigorite polysomatism (Section 4.2.3). Especially, structural reorgani- tions (Fig. 4). Similar observations are made for chrysotile (Fig. 5),
zation and H2O redistribution in serpentinites or serpentinized perido- even if the reaction Chr + Br → A + 10Å + H2O is deduced
tites is likely to occur. from the overall topology of the system and was not directly tested
with dedicated experiments. Similar observations are made in the
MASH and CMASH systems or for any other chemistry, e.g. includ-
4.1.3. The question of “metastability” ing minor amounts of Fe, Ni or Cr, which do not change the topol-
Chrysotile is said to be “metastable” (Evans, 2004) as it may trans- ogy of the overall system.
form to other serpentine minerals with same composition and different Within peridotites the stability of clinochlore, the Mg-chlorite,
structure through recrystallization under hydrothermal conditions. depends on the presence of enstatite in the system, but it only shifts
Lizardite has also been noticed as “metastable” (e.g. O'Hanley et al., the stability curves by ≈50 °C down (Fig. 6). Other minerals, such as ky-
1989; Mellini and Zanazzi, 1989). A significant number of studies refer anite, diapore and corundum, which would reduce clinochlore stability
to “metastability”, “metastable dehydration reactions” or “metastable by tens of degrees, are not expected in the deep mantle, except if local
extension” of the antigorite stability field (e.g. Trommsdorff and high-Al, low-Mg metasomatism exists in near Moho-conditions. Espe-
Evans, 1980; Wunder and Schreyer, 1997). Many studies questioned cially, between 2 and 5 GPa, Mg-chloritoid, poor in Mg compared to
the potential link between subduction seismicity and the dehydration clinochlore, may form from chlorite between 400 and 550 °C and desta-
of serpentine minerals (e.g. Peacock, 2001; Hacker et al., 2003; Abers bilize between 600 and 750 °C depending on the presence of Al-rich
et al., 2013). minerals. Furthermore, in a talc-rich system, Mg-chloritoid is
Antigorite “metastability” is especially invoked to explain why its destabilized at ≈3.5 GPa, leading to the formation of Mg-sursassite
dehydration curve does not fully fit the lower Wadati-Benioff plane of (Fockenberg, 1998b).
double seismic structures (e.g. Peacock, 2001). Indeed, as shown in Quartz or aqueous SiO2 may also be present within peridotites (e.g.
Fig. 4, a significant part of the seismicity does not fit antigorite destabi- Bonnemains et al., 2017), which, at 2 GPa, lowers the stability of
lization, even when bulk composition and phases chemical variability clinochlore by 300 °C and 4 GPa. On the contrary, moderate SiO2 activity
are considered (Fig. 4; Sections 4.2.1 and 4.2.2). However, experimental transforms chrysotile and lizardite into antigorite and tends to stabilize
work evidenced that, for pressures between 1 and 6 GPa, antigorite de- antigorite; furthermore, high SiO2 activity transforms serpentine min-
stabilizes quickly in the stability field of Fo + En + H2O (above 600–700 erals into talc (Schwartz et al., 2013). Chrysotile also transforms to
°C), which suggests that “metastable” antigorite does not exist in a antigorite in the presence of talc or tremolite (Fig. 5). The extent of
subducted slab (Inoue et al., 2009; Ferrand et al., 2017). talc occurrence depends, as the latter, on the bulk Mg-Si ratio (Pawley
The kinetics of antigorite dehydration is fast enough to trigger brittle and Wood, 1995). Especially, the presence of forsterite decreases the
failure in subduction conditions (Perrillat et al., 2005; Ferrand et al., stability of talc by ≈100 °C at 1 GPa to ≈250 °C at 2.5 GPa (Berman
2017; Liu et al., 2019), it is not metastable (Inoue et al., 2009; Ferrand, et al., 1986; Pawley, 1998). Hence considering near-Moho forsterite-
2019). Hence, other phases (Fig. 4), are likely not to be metastable, poor bulk composition, talc may stay stable or heterogeneously trans-
which was demonstrated for talc and 10-Å phase (Chollet et al., 2009). form, especially depending on fluid circulation and ion charge of these
By definition, metastability is the ability of a mineral to persist for an fluids.
unbounded time in an unstable equilibrium or metastable state. The ki-
netics of metamorphic reactions depends on the presence of a fluid 4.2.1.2. Coexisting hydrous phases. In nature, various minor phases may
phase, which dramatically increases the mobility of atoms within a coexist within serpentinite bodies or along serpentinized faults. Nota-
rock volume (e.g. Fyfe, 1958; Rubie, 1990, and reference therein). In bly, coexisting phases may be less stable than each phase of the mixture.
other words, the absence of fluids is known to favor metastability. But For example, at 3 GPa, antigorite (Fig. 4) and talc (Fig. 7) are stable up to
for dehydration reactions, this kinetic effect is in competition with ≈650 °C and ≈750 °C respectively, and the association of those phases
H2O saturation. Indeed, as a reaction product, H2O stabilizes antigorite destabilizes around 570 °C (Figs. 4 and 7). Note that similar behavior is
(Figs. 4 and 12; Perrillat et al., 2005). “Metastability” in transforming observed with brucite, as the latter dehydrates above 1000 °C at those
serpentinites at subduction conditions could actually be due to fluid cir- pressures (e.g. Berman et al., 1986; Kanzaki, 1991; Fig. 9). Between 2
culations within serpentinized faults, allowing ions mobility between and 4 GPa, the association talc-chlorite-chloritoid is also less stable
different regions of the serpentinized peridotite. Fluid percolation and than any of the three phases involved (Figs. 6 and 7).
repeated circulation events may allow reaction products to migrate, as Within peridotites, talc is generally considered stable up to 1.5 GPa
a supercritical fluid, up to the subduction channel. In other words, igno- and 700 °C only and most studies show pseudosections in which talc
rance on local variations of mantle composition and connectivity of is absent at higher pressures (e.g. Fumagalli and Poli, 2005; Padrón-
potential fluid pathways seems to be called “metastability”. The latter Navarta et al., 2013). Nevertheless, talc itself can be stable to much
may also refer to the difference in the position of stability limit in the higher P and T, depending on its chemistry and on the mineralogical en-
P-T diagram between theoretical pure phases and natural minerals vironment (e.g. Yamamoto and Akimoto, 1977; Pawley and Wood,
with variable ions diversity (Section 4.2.2). 1995). Experimental data (Fig. 7) show that talc within peridotites can
only be stable up to 2 GPa at 500 °C and 1.5 GPa at 600 °C; but talc within
antigorite-rich serpentinites is, depending on chemistry, stable to
4.2. Parameters controlling phases stability 3–4 GPa at 500 °C and 2.5–4 GPa at 600 °C. As a consequence,
considerding potential high hydration in near-Moho conditions or
4.2.1. Impact of bulk composition on the stability of mineral assemblages along deep bending faults, talc may locally exist within deep
As summarized in Fig. 2. Bulk composition seriously impacts the sta- serpentinites.
bility limit of major mantle hydrous phases, especially depending on the Potential structural associations of hydrous phases are discussed in
presence of brucite, enstatite and forsterite in the vicinity of the latter. Section 4.2.5.
Figs. 4–7 detail the various reactions that could affect antigorite
(Fig. 4), chrysotile, lizardite (Fig. 5), chlorite, Mg-chloritoid (Fig. 6) 4.2.2. Impact of chemical composition on mineral stability
and talc (Fig. 7) in the MSH system and other systems that are relevant In dry mantle peridotite, Al is mostly present in spinel and garnets,
for serpentinized mantle. but some Al also integrates clinopyroxenes. In contrast, in serpentinized
220 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Table 4
Composition of antigorite samples from various studies.

Composition of oxides. Contributions below 0.05 for all samples are not indicated.
a
Corrected for Carbonate impurity and absorbed water; *This study.
Depending on the study, bulk analysis (Bulk an.) was performed by electron microprobe (EMP), X-ray fluorescence (XRF) or XRD (X-ray diffraction; this study). When investigated, H2O
m M mFe3þ
and iron ions determination (det.) is specified. X Mg ¼ 1=ð1 þ mFe Mg
Mg MFe
Þ; X Fe3þ ¼ ðm þmFe2þ Þ. References: E.1976 = Evans et al. (1976); J&S.1981 = Johannes and Schreyer (1981); W&S.1997
Fe3þ

= Wunder and Schreyer (1997); U&T.1995 = Ulmer and Trommsdorff (1995); B&P.2003 = Bromiley and Pawley (2003); P.2005 = Perrillat et al. (2005); I.2009 = Inoue et al. (2009);
G.2011 = Gasc et al. (2011); G.2017 = Gasc et al. (2017); F.2017 = Ferrand et al. (2017).

peridotites substantial Al is preferentially integrated into various hy- often confused with chlorite, is not well known but is likely to be
drous phases, such as chlorite or other Al-rich phases (Fig. 6), but also much greater (Ackermand et al., 1975).
in smaller amounts in MSH minerals, e.g. antigorite (Fig. 4), lizardite Clinochlore is the hydrous silicate that stays stable the longest in pe-
(Fig. 5) or talc (Fig. 7). ridotites below 5 GPa (Staudigel and Schreyer, 1977; Pawley, 2003;
Fumagalli et al., 2014). Its stable structure may include various cations,
such as Fe2+ or Cr3+ in significant quantities (e.g. Ulmer and
4.2.2.1. Antigorite composition and stability. Natural antigorite has sys-
Trommsdorff, 1999; Fumagalli et al., 2014). Notably, such compositional
tematic compositional variations depending on its origin due to differ-
variations may delay the reaction Chl + 2 En → 2 Fo + Py + H2O by
ent formation contexts (Uehara and Shirozu, 1985), as confirmed by
30–80 °C.
various analyses using electron microprobe or X-ray fluorescence
Talc composition does not usually vary much from the end-member
(Table 4). Comparing the MSH and MASH compositions (Bromiley and
formula Mg3Si4O10(OH)2 (Pawley and Wood, 1995), with as most com-
Pawley, 2003), the reaction Atg → 14 Fo + 20 En + 31 H2O is shifted
mon substitutions up to ≈0.15 Al replacing Si and ≈ 0.1 Fe replacing Mg
by 40–100 °C at 2 and 5 GPa, respectively (Fig. 4). This stabilizing effect
(Deer and Zussman, 1962). Nonetheless, such substitutions may shift
of Al3+ on antigorite seems even greater with increasing cations diver-
talc stability by 80–100 °C (Fig. 7; Ulmer and Trommsdorff, 1999). Com-
sity, e.g. adding Fe2+ or Ni2+ in antigorite, or adding Ca2+ or Cr3+ in the
paring talc stability in the CMSH and CMASH systems, Al3+ appears to
serpentinite (e.g. Ulmer and Trommsdorff, 1995), which is likely for
stretch the reaction Ta + Fo → 5 En + H2O by N100 °C at 2 GPa. Compar-
most of the unpublished natural compositions (e.g. Bose and
ing the MSH and CMASH systems, the reaction Ta → 3 En + Coe + H2O
Navrotsky, 1998). Such inclusive compositions would stretch the stabil-
is shifted by N120 °C at 4 GPa (Fig. 7).
ity limit by 60–150 °C compared to the MSH composition (Fig. 4). In ad-
The stability fields of the 10-Å phase (Fig. 7) and phase A (Fig. 8) are
dition, it is important to note that the “Al-bearing” sample of Bromiley
also shifted to higher pressures due to ions diversity, so that their ther-
and Pawley (2003) is also Fe-rich and that Fe determination was not
mal limit decreases by some tens of degrees (e.g. Chollet et al., 2009;
performed. When Fe3+, as a trivalent cation, could stabilize antigorite
Ulmer and Trommsdorff, 1999).
to higher T and P (Bromiley and Pawley, 2003; Padrón-Navarta et al.,
According to the literature, chemistry has limited impact on the sta-
2008, 2013; Debret et al., 2014; Debret et al., 2015) via Tschermak sub-
bility of Mg-amphiboles (Fig. 8). However, substantial amounts of triva-
stitutions (Evans et al., 2012; Padrón-Navarta et al., 2013), Fe2+ is
lent cations, especially Al3+ and Fe3+, are preferentially incorporated in
known to decrease its stability by ≈ 15 °C (Fig. 4, CMASH system, 0.8
amphiboles by Tschermak substitution (e.g. Ulmer and Trommsdorff,
b XFe b 1%; Ulmer and Trommsdorff, 1999). Considering these two an-
1999), which leads to various structures and stabilities. For instance,
tagonist effects, 10 wt% FeO reduces antigorite stability by only 25 °C
tschermakite, i.e. Al-tremolite (Jenkins, 1983), is stable to ≈ 880 °C
(Merkulova et al., 2016).
(i.e. limit shifted by 70 °C) at low pressure. Also, and most importantly,
it should be noticed that cummingtonite, the monoclinic polymorph of
4.2.2.2. Composition and stability of other hydrous phases. Chrysotile and anthophyllite, is stable to 2 GPa at ≈740 °C (MSH system), which can
polygonal serpentine may bear some Al (e.g. Andreani et al., 2007, 0.5 be extended to ≈780 °C if the iron content reaches 25% (Evans and
and 4 wt% of Al2O3). On the contrary, lizardite naturally bears substan- Ghiorso, 1995). Thus, as for serpentine minerals, the breakdown of am-
tial Al. The impact of Al on the stability of lizardite has been studied at phiboles could explain a large part of mantle earthquakes within hot
low pressure (Chernosky, 1973a, 1973b; Caruso and Chernosky, subducting slabs (Fig. 8).
1979), i.e. a shift by ≈ 60 °C between Lz(0.2) and Lz(0.5) (Fig. 5; see Finally, clinohumite, which could be seen as a HP hydrated olivine, is
Table 1 for compositions). The stability of amesite, the Al-serpentine, widely observed in antigorite rocks, stabilizing by adapting its
T.P. Ferrand / Lithos 334–335 (2019) 205–230 221

composition to P-T conditions (e.g. Trommsdorff and Evans, 1980; with increasing H2O activity, with a maximum in H2O-saturated con-
Evans and Trommsdorff, 1983; Scambelluri et al., 1991; Weiss, 1997). ditions (Fig. 12; Perrillat et al., 2005). Chrysotile dehydration (Fig. 5)
Indeed, clinohumite can bear substantial Fe2+, Fe3+, Ti3+ and F−, may contribute to H2 O saturation and delay antigorite instability.
stretching its stability limit by hundreds of degrees (Fig. 9; e.g. López Similarly, if brucite is present in the system in limited amounts,
Sánchez-Vizcaíno et al., 2005). part of antigorite could react at about 500 °C, stabilizing the system
via H2 O saturation, which would allow stress to rise again in the
4.2.2.3. Anhydrous phases. Anhydrous phases (Fig. 9; Table 2) also have fresh peridotite surrounding serpentinized zones (i.e. serpentinized
varying compositions. Various proportions of Al3+ or Ca2+ in pyroxens faults or ductile zones, e.g. Scambelluri et al., 2017).
and garnets significantly change their stability fields (e.g. Morishita It should be recalled that H2O saturation at high P most likely refers
et al., 2001; O'Hara et al., 1971). Nevertheless, as the latter form solid so- to defects diffusing within crystals. In other words, neither sustainable
lutions, they are not likely to destabilize abruptly, but rather gradually porosity nor fluid pressure is required.
adjust their composition to the P-T evolution. For this reason, only a
few compositions are considered in Fig. 9. 4.2.3.2. Impact of H2O on the stability of other phases. Presumably, for any
phase, variations in H2O saturation could explain most of the remaining
4.2.2.4. Carbonated peridotites. Below 1 GPa, mechanical instabilities in discrepancies that cations variability does not account for. Furthermore,
carbonates are involved in seismicity (e.g. Kurzawski et al., 2016). At supposing H2O-saturated conditions for most of the experimental
higher pressures, considering carbonated peridotites and CO2 in the results, a reduction of the saturation in near-instability conditions for
aqueous fluid (e.g. Trommsdorff and Evans, 1977; Zhou and Hsu, any of the listed phases might shift the curves 50–100 °C toward
1992), additional reactions should be considered (Fig. 11). Carbonate lower temperatures.
phases that are expected in carbonated peridotites are listed in The H2O content may vary in the 10-Å phase (e.g. Pawley and Wood,
Table 3. Some serpentinite decarbonation reactions are considered 1995; Chinnery et al., 1999; Chollet et al., 2009), which has a large impact
(Tenthorey and Cox, 2003; Trommsdorff and Evans, 1977), but most on its stability (Fig. 7). Its ideal composition in the MASH system has been
of the knowledge focuses on anhydrous systems up to date (Kushiro established to be Mg3-y/2Aly/3Si4O10(OH)2·2H2O (Yamamoto and
et al., 1975; Martinez et al., 1996; Wyllie and Huang, 1976; Zhou and Akimoto, 1977), with possible substitution of Mg2+ by Fe2+. As discussed
Hsu, 1992). As suggested by Fig. 11, some of these reactions fit mantle above (Section 4.1.1), Chinnery et al. (1999) actually studied the stability
seismicity. Near the Moho, peridotite carbonation is expected (e.g. of a near-10-Å phase, that could actually be one of the "talc-like" phases.
Wyllie and Huang, 1976) and associated decarbonation reactions A possible explanation for the uncertainty in H2O-saturation during
seem to fit part of the mantle seismicity. Hydrous carbonates (Table 4) talc stability tests (Fig. 7) is directly related to the existence of "talc-like"
are most likely in hydrated peridotites but, unfortunately, their stability phases that form during loading and heating, progressively integrating
fields have not been studied yet. H2O in their structure. In other words, to get H2O-saturated conditions
near the talc stability limit, a significant oversaturation is needed. As
4.2.2.5. Phase transitions. Enstatite, one of the most abundant phases in argued by Chinnery et al. (1999), the H2O content of HP hydrated talc
peridotites, endures two transitions in the P-T window considered in should depend on the composition of the starting material, especially
this study. Both of them fit some mantle seismicity (Fig. 10). Especially, trivalent cations. And, as the stability of all hydrated phases itself de-
the transition lcEn → oEn surprisingly fits seismicity beneath Hokkaido pends on chemistry, little chemical variations are likely to significantly
at 5–5.5 GPa. Furthermore, recent experimental work (Xu et al., 2018) change the stability limit of such partially hydrated talc minerals.
suggests that the high-pressure transition in orthoenstatite above Ulmer and Trommsdorff (1999) reported large differences in the
12 GPa (≈400 km) triggers part of the deep earthquakes. experimentally deduced reaction Ta + Fo → 5 En + H2O (Kithara
The quartz-coesite transition correlates particularly well with the et al., 1966; Yamamoto and Akimoto, 1977; Guggenbuehl, 1994;
seismicity beneath Hokkaido and Tohoku at ≈2.5 GPa. But quartz is Pawley, 1998) and argued that the only likely explanation they could
not expected in deep peridotites, except if quartz is locally present as offer is possible differences in the activity of H2O in the different studies.
a minor phase. These earthquakes also correlate with the destabilization In other words, the results of Yamamoto and Akimoto (1977) would
of serpentine + brucite, which is more likely. correspond to water-saturated conditions whilst the experiments of
Several transitions in carbonates also correlate with seismological Pawley (1998) would represent water-free conditions.
observations (Fig. 10). For near-Moho mantle earthquakes (e.g.
Cascadia), the transition Ca → Ara could be involved in seismicity. For 4.2.3.3. Impact of CO2 on decarbonation reactions and antigorite stability.
the lower Wadati-Benioff plane of cold slabs (e.g. northern Japan), the Aqueous fluid circulating in faulted peridotites are likely to contain up
presence of carbonates is less probable, but the question of fluid circula- to 0.1 wt% CO2 (Trommsdorff and Evans, 1977; Zhou and Hsu, 1992).
tion in deep bending faults at trenches is still debated. If so, dolomite If so, CO2 saturation influences (de)carbonation reactions:
could locally form in deep faults down to the brittle-ductile transition,
i.e. 20–40 km depth, and its decomposition during subduction could Di þ Ca → Ak þ CO2
play a role in seismicity.
Hence, surprisingly, this review highlights potential involvement of Wo þ Mc þ CO2 → Di þ Ca
phase transitions within serpentinites, peridotites or serpentinized pe-
ridotites between 1 and 6 GPa. These reactions are shifted to higher pressures by ≈1 GPa by de-
creasing CO2 saturation from 0.1 wt% down to 0.02 wt% (Fig. 11). As it
4.2.3. Impact of fluid composition and saturation level stabilizes carbonates, increasing CO2 saturation stabilizes antigorite,
but this effect is very limited (Tenthorey and Cox, 2003; Trommsdorff
4.2.3.1. Impact of H2O on antigorite stability. Antigorite stability is and Evans, 1977).
known to depend on H2 O saturation (e.g. Perrillat et al., 2005;
Hilairet et al., 2006; Fig. 4). At 2.5 GPa, H2O saturation shifts the sta- 4.2.4. Polysomatism, water content and mineral stability
bility limit from 570 °C (Perrillat et al., 2005) to 670 °C at least
(Bromiley and Pawley, 2003; Ulmer and Trommsdorff, 1995). 4.2.4.1. Antigorite self-stabilized by polysomatism. Antigorite is an atypical
Perrillat et al. (2005) showed that a “talc-like” phase forms in H2O- serpentine variety for several reasons that are detailed below. In the
unsaturated conditions, whereas antigorite remains stable in H2O- MSH system, antigorite exhibits 12 ideal discrete compositions depend-
saturated conditions. The stability limit gradually shifts to higher T ing on P-T conditions (Wunder et al., 2001). According to Kunze (1961),
222 T.P. Ferrand / Lithos 334–335 (2019) 205–230

antigorite compositional variations can be expressed by the formula This means that H2O saturation should not favor the transformation
M3m-3T2mO5m(OH)4m-6 where M are octahedral cationic sites (Mg2+, Chr → Atg or its reverse reaction (Wunder et al., 2001). Hence, as it is
Fe2+, Ni2+, Al3+), T are tetrahedral cationic sites (Si4+, Al3+, Fe3+) not a dehydration reaction, antigorite could contribute to almost 100%
and m is the number of tetrahedra in a single chain along the a-axis, of serpentinites in brucite-poor environments, as suggested by field
i.e. wavelength. Notably, antigorite H2O content depends on m. For nat- observations.
ural antigorite, m = ⟦13;24⟧, which corresponds to [12.09;12.52] wt% of Finally, even if most of the studies do not consider it as a key param-
H2O (Mellini et al., 1987), and m decreases with increasing metamor- eter, the stability limit of antigorite varies with P and m, as follows:
phic grade of serpentinites (Kunze, 1961; Uehara and Kamata, 1994).
In subduction P-T conditions, increasing T or decreasing P reduces m 5 Atgð14Þ → 72 Fo þ 17 Ta þ 108 H2 O ðP ¼ ½0:5; 1:5 GPaÞ
(Fig. 12), which corresponds to a volume reduction of the antigorite
crystal lattice (Wunder et al., 2001). For 18 N m N 14 and depending AtgðmÞ → ðm−3Þ Fo þ ðm þ 3Þ En þ ð2m−3Þ H2 O ð14 bm b16Þ
on relative amounts of brucite (stable to 400–500 °C in subduction
zones, Fig. 4a, 5 and 9) and H2O, successive polysomatism reactions
5 AtgðmÞ → ðm−3Þ A þ ð2m þ 6ðm þ 1ÞÞ En
can be written as follows (Wunder et al., 2001; Fig. 12):
þ ðm þ 6ðm−1ÞÞ H2 O ð16 bm b18Þ

ð3 þ mÞ Atgðmþ1Þ → ð4 þ mÞ AtgðmÞ þ 6 Fo þ 9 H2 O
As explained in Section 4.2.3 and 4.2.1, these dehydration reactions
ðmþ1Þ ðmÞ are likely to be delayed depending H2O saturation and potential compo-
m Atg → ð1 þ mÞ Atg þ 3 Br
sitional variations.

This means that antigorite polysomatism reactions are minor 4.2.4.2 H2O content of high-pressure hydrated talc phases: A polysomatic
dehydration reactions, which may, to a certain extent, maintain H2O- series? As shown in previous sections, investigations on HP hydrated
saturated conditions, i.e. self-stabilization (Perrillat et al., 2005; talc are still very active, especially questioning the composition, struc-
Fig. 12). Depending on bulk composition and potential fluids, these ture and stability of “talc-like” phases (e.g. Perrillat et al., 2005; Chollet
structural adjustments of antigorite may occur or not, which is likely et al., 2009). The 10-Å phase (e.g. Yamamoto and Akimoto, 1977),
to shift its stability limit. which can be seen as one of the so-called “talc-like” phases, can be
In addition, when chrysotile destabilizes in the stability field of written in the MASH system as (Mg,Al)3(Si,Al)4O10(OH)2·2H2O.
Atg(m), brucite is formed regardless of m as follows: Complementary experiments evidenced a similar phase with lower
H2O-content, which is written (Mg,Al)3(Si,Al)4O10(OH)2·0.65H2O
m Chr → AtgðmÞ þ 3 Br ð16bmb18Þ (Chinnery et al., 1999), that I suggest as one of the “talc-like” phases.
Such variation in H2O content significantly changes the stability field

Fig. 14. Examples of interlayering and other close associations of hydrous minerals. (a) Scanning electron microscopy image showing an association between sudoite, clinochlore,
muscovite (Ms) and pyrophyllite in metasediments from Verrucano, Italy (adapted from Giorgetti et al., 1997). (b) High-resolution Transmission Electron Microscopy (TEM) image
(adapted from Dódony and Buseck, 2004) showing an intimate intergrowth interlayer of clinochlore, lizardite and chrysotile in the Baltimore mafic complex. Each of these structures
shows coherent, continuous boundaries with the others. (c-d) TEM images showing various associations between antigorite, clinochlore, dozyite (D), chrysotile and polygonal serpentine
(PS) in a serpentinite from Wood's Mine, Pennsylvania (adapted from Banfield et al., 1995). Polygonal serpentine develops as connexions between larger grains, mostly clinochlore. Ab-
breviations that are not specified here are detailed in Table 1.
T.P. Ferrand / Lithos 334–335 (2019) 205–230 223

of HP hydrated talc (Fig. 7) and is considered to be due to H2O unsatu- 4.3. A joint effect of successive minor transformation-driven stress
rated conditions during its formation (Chinnery et al., 1999). transfers?
Most probably, “talc-like” phases are a population of phases with
slightly variable lattice structures, close to talc and 10-Å phase, i.e. 4.3.1. A full correlation between seismicity and mineral transformations
with different water contents (e.g. Fumagalli and Stixrude, 2007), and Considering all the factors that control the stability of mineral assem-
only formed in narrow P-T conditions. However, even if some enstatite blages in natural peridotite below 6 GPa (Section 4.2), I demonstrate that
crystals are observed after antigorite dehydration, a “talc-like” phase is a large majority of the stability limits drawn in Figs. 4–11 correlate with
formed and preserved above 800 °C (Ferrand et al., 2017), which indi- mantle earthquakes beneath Japan, Cascadia and northern Chile. Thus, it
cates that “talc-like” phases are stable at higher T than (low-pressure) is reasonable to think that, if present, any of these reactions could partic-
talc. Once these phases are formed, their stability could extend to high ipate in peridotite embrittlement. Moreover, as suggested by geophysical
pressure and temperature as for talc and 10-Å phase (e.g. Pawley and imaging (Ranero et al., 2003; Ivandic et al., 2010; Shillington et al., 2015)
Wood, 1995; Chollet et al., 2009). and confirmed by most recent studies (Cai et al., 2018; Bloch et al., 2018),
In light of this review, and by comparison with antigorite, I propose the H2O transfer into the mantle is likely to occur thanks to deep
that talc and HP hydrated talc phases could be regarded as a serpentinized faults, although almost all the peridotite remains fresh
polysomatic series. Further investigations are required to understand within the downgoing slab (Reynard et al., 2010).
whether “talc-like” phases have discrete or continuous compositions.
4.3.2. Transformation-driven grain size reduction and stress transfers
4.2.5. Interlayering and other close mineral associations Intermediate-depth earthquakes are thought to initiate by thermal
Minerals of the serpentine, talc and chlorite groups are phyllosilicates. runaway processes (e.g. John et al., 2009; Kelemen and Hirth, 2007).
As mentioned in Section 4.2.1, the stability fields of many hydrous silicates Such strain localization is grain size sensitive and is likely to propagate
discussed in this study overlap. Interactions between coexisting phases and accelerate by grain size reduction up to the unstable propagation,
may significantly reduce destabilization temperatures. However, close as- i.e. seismic rupture (Thielmann et al., 2015; Thielmann, 2018a, 2018b).
sociation of phases may drive additional stability. In particular, associated Dehydration reactions of single phases are fast phenomena, as dem-
phases may stabilize each other by forming fully-fledged minerals, such onstrated for antigorite (Perrillat et al., 2005; Inoue et al., 2009; Ferrand
as kulkeite (e.g. Schreyer et al., 1982) or dozyite (Bailey et al., 1995). et al., 2017) and talc (Chollet et al., 2009). Such transformations sud-
Interlayering has been documented for various assemblages, such as denly lead to extremely fine mineral aggregates (Ferrand, 2017), and
muscovite + chlorite (e.g. Giorgetti et al., 1997), lizardite + chlorite any other reaction is likely to induce grain size reduction. Grain size is
(Dódony and Buseck, 2004) and chlorite + berthierine (Fe-Al-serpen- the key parameter driving strain localization by thermal runaway
tine; Xu and Veblen, 1996). Various associations between antigorite, (Thielmann et al., 2015; Thielmann, 2018a, 2018b). If hydrous phases
clinochlore, dozyite, chrysotile and polygonal serpentine were evi- are present in a future fault zone, it is reasonable to include its dehydra-
denced by Banfield et al. (1995). The latter studies presented SEM and tion in the strain localization process. For instance, antigorite dehydra-
TEM images showing interlayered minerals and other close mineral as- tion may be either a cause or a consequence of strain localization (e.g.
sociations forming coherent boundaries (Fig. 14). Brantut et al., 2010, 2011; Brantut and Sulem, 2012; Platt et al., 2015).
In the MASH system, Al-free chrysotile may form depending on stoi- Furthermore, any new HP phase or reaction discovered in the past
chiometry within synthetic lizardite samples (Liz(0.2); Chernosky, years allows imagining successive grain size reduction events, which
1973a). Lizardite and chlorite coexist and could stabilize each other could actually participate in keeping relatively small grain size at some
(Caruso and Chernosky, 1979). Intimate relations have been imaged be- places. For example, the discovery of yoderite (Schreyer, 1988), stable
tween chrysotile, lizardite and clinochlore (Dódony and Buseck, 2004). from ≈750 to ≈870 °C between 1.5 and 2 GPa, adds a reaction curve
Dozyite (Bailey et al., 1995; Banfield et al., 1995) is a natural 1:1 regular to Fig. 6, which correlates with earthquakes around 770 °C beneath Kii
interstratification of trioctahedral serpentine and chlorite, with an inter- or Cascadia. Such localized grain size reduction events could be accom-
mediate formula between clinochlore and amesite (Al-serpentine). panied by slow slab-scale continuous stress transfers, equivalent of
Kulkeite is a natural interlayered talc-chlorite mineral (Schreyer what was observed in the laboratory during the dehydration of slightly
et al., 1982). On the field, it may be confused with mica because of sim- serpentinized peridotites (Ferrand et al., 2017). These stress transfers
ilar shape and aspect. It is important to notice that the talc layer of would largely be accommodated by “self-localizing” thermal runaway
kulkeite is hydrated, and that observations suggest intermediate stabil- (e.g. John et al., 2009), i.e. viscous flow in the periphery of transforming
ity between talc and chlorite (Schreyer et al., 1982), relevant for hot mineral assemblies, i.e. planar patches. In other words, mineral transfor-
subducted slabs (Figs. 4 and 5). Similarily, a mixed layered structure mations induce grain size reduction, which initiates strain localization
formed of clinochlore and 10-Å phase was synthetized (Fumagalli and (e.g. Thielmann, 2018a, 2018b), driving local viscous flow propagating
Poli, 2005) and could thus be called HP (hydrated) kulkeite. by thermal runaway up to a regional mechanical instability, i.e. perido-
As stated by Fumagalli and Poli (2005), kulkeite would destabilize in tite embrittlement.
the presence of clinopyroxene (Ca,Mg)2Si2O6. For example, diopside According to the DDST model (Ferrand et al., 2017), at uppermost
and kulkeite would transform into enstatite and garnet as follows: mantle pressures, i.e. between 1 and 2 GPa, the dehydration reaction
is likely to drive a mechanical instability in peridotites only if the
dehydrating mineral constitutes a minor phase, e.g. 5 vol% in the exper-
Ku þ Di → 5 En þ Fo þ GtCaMg2 þ 5 H2 O iments (Ferrand et al., 2017). This volume ratio cannot be directly ex-
trapolated, as the phase network connectivity mostly depends on the
Kulkeite is qualified as “metastable” in chlorite-rich rocks (Schreyer shape of the inclusions, which should be closer to surfaces than volumes
et al., 1982) and dozyite would also be stable up to 800 °C at least in nature, i.e. deep faults. Thus, some minor phases discussed in this
(Banfield et al., 1995). However, as closely associated phases, they could study may have major importance.
endure much faster transformation once the assembly becomes unstable. On one hand, increasing the number of phases should dispatch po-
Finally, interlayering is an alternative explanation for the most ex- tential stress transfers along a fault network, which would favor
treme observed stability of talc (Fig. 7; e.g. Yamamoto and Akimoto, aseismic deformation. On the other hand, regional variability in both
1977) compared to theoretical reactions (Pawley, 1998). Possible fault network geometry and phases diversity may drive significant
interlayered talc-like phases, i.e. with various H2O contents, could also stress accumulations where reactions are rare or inexistent. It is then
participate in the high stability of “talc-like” phases observed in the lab- important to keep in mind that antigorite (Fig. 4) and chlorite (Fig. 6),
oratory (e.g. Ferrand et al., 2017). in spite of significant influence of above-mentioned parameters on
224 T.P. Ferrand / Lithos 334–335 (2019) 205–230

their stability, are the most widely stable rock-forming hydrous phases Paterson, 1965), it could contribute to the triggering mechanism of
below 5 GPa and that their stability limits vary much less than the ones earthquakes in the upper Wadati-Benioff plane, especially during the
of talc (Fig. 7) for example. Furthermore, considering local variations of dehydration of lawsonite (Incel et al., 2017; Okazaki and Hirth, 2016).
all above-mentioned parameters, N90% of the mantle earthquakes cor- As discussed by several papers (e.g. Reynard et al., 2010; Thielmann,
relate with antigorite destabilization; chlorite remains a good candidate 2018a), overpressure requires available fluids, which are not likely in
for earthquakes at higher temperatures. the subducting mantle, especially at depth N5 km below the Moho.
Seismic fault nucleation in peridotites could originate in a chain Moreover, most of the dehydration reactions (Figs. 4–8) induce a nega-
mechanism with grain size as central parameter. Sliding would become tive volume change (solid + fluid) between 1.5 and 5 GPa, which is in-
unstable after a critical displacement at a certain scale, as suggested by compatible with fluid overpressure (e.g. Jung et al., 2004; Gasc et al.,
the DDST model (Ferrand et al., 2017). Considering potential fluid per- 2011).
colation and repeated circulation events, which would allow reaction Using both a rupture criterion (Ohnaka, 2003) and the percola-
products to migrate up to the subduction channel, understanding the tion theory (Stauffer and Aharony, 1992), the DDST model (Dehy-
triggering mechanisms of earthquakes in the slab mantle remains dration-Driven Stress Transfer, Section 4.3.4) predicts that
challenging. mechanical instabilities could nucleate due to high stress intensity
at the tip of antigorite clusters in olivine-antigorite aggregates
4.3.3. Dehydration embrittlement vs dehydration-driven stress transfer (Ferrand et al., 2017). If so, depending on the regional stress
The “dehydration-embrittlement” model (Raleigh and Paterson, state, any unstable minor phase could locally induce a critical stress
1965) hypothesizes fluid-induced embrittlement. According to this transfer. The most important parameters controlling the earth-
model, brittle yielding of the rocks would be due to fluid overpressure quake triggering should be the relative dimensions of clusters of
due to the dehydration of phases (e.g. Hacker et al., 2003; Frohlich, unstable phase and volumes of stable peridotite, as well as the spa-
2006). Firstly evidenced at low pressure in the laboratory (Raleigh and tial distribution of the clusters.

Fig. 15. Comparison between mantle earthquakes beneath Japan, Cascadia and Chile. (a) P-T diagram showing P-T conditions at hypocentres in the subducting mantle beneath Hokkaido,
Tohoku, Kii, Tokai and Cascadia (Abers et al., 2013), with additional data for northern Chile (latitude 23°S); (b) Same as (a) with additional data for northem Chile (21.5°S), i.e. where un-
certainties remain regarding the thermal structure due to both the change in subduction obliquity and the additional bend of the slab. Curves show the stability limits of chrysotile (Chr),
antigorite (Atg), talc (Ta) and clinochlore (Chl). For each of these phases, colour shades indicate the destabilization P-T window depending on local variations of peridotite bulk compo-
sition in the MSH and/or MASH systems. For composition and abbreviation of mineral phases, see Tables 1 and 2. For details see Figs. 4–7. For additional phases see Figs. 8–11. References:
B.1986 = Berman et al. (1986); P.2003 = Pawley (2003); P.2005 = Perrillat et al. (2005); S&S.1977 = Staudigel and Schreyer (1977); U&T.1999 = Ulmer and Trommsdorff (1999);
W&S.1997 = Wunder and Schreyer (1997).
T.P. Ferrand / Lithos 334–335 (2019) 205–230 225

Stress state description in a polyphasic mineral aggregate depends Two distinct seismicity planes exist within the subducting mantle of
on the observation scale. Indeed, as experimentally demonstrated by the Nazca Plate, separated by ≈25 km and merging into a highly seismic
Ferrand et al. (2017), low sample-scale stress state can coexist with body between 90 and 150 km depth (Sippl et al., 2018). The upper-
local high stress intensities, especially due to local mineral transforma- plane mantle seismicity could partly originate in somehow large back-
tions. Both experimental and seismological observations are consistent ground stresses in the vicinity of the subducting interface (b10 km
with the DDST model, which appears more accurate than “dehydration away), but could also be triggered by the destabilization of chrysotile
embrittlement” to explain mantle earthquakes. Recent field work and lizardite, i.e. transition to antigorite (Fig. 5; Evans et al., 1976;
(Dunkel et al., 2017; Scambelluri et al., 2017; Ferrand et al., 2018) and Evans, 2004; Schwartz et al., 2013).
seismological data (Cai et al., 2018; Bloch et al., 2018; Kita and The earthquakes of the lower Wadati-Benioff plane fully correlate
Ferrand, 2018) together support DDST rather than dehydration embrit- with the destabilization of serpentine minerals, i.e. dehydration reac-
tlement to explain the earthquakes of the lower Wadati-Benioff plane. tions (Fig. 13f and 15a). At latitude 21.5°S, however, it seems that the
lower-plane seismicity extends up to the limit of stability of talc
4.3.4. Heterogeneity, phases connectivity and transformation-driven stress (Fig. 13b). This shift in temperature estimations (40 °C between
transfers Fig. 13d and f), is likely to be due to a slight change in subduction obliq-
In natural materials, homogeneity can be defined at a certain scale. uity (Plunder et al., 2018), which happens around latitude 22°S (Sippl
Nonetheless, rocks are usually assemblages of various minerals, con- et al., 2018).
taining impurities at the grain scale, e.g. fluid inclusions, or at crystal lat- The increase of seismicity at 90–100 km depth (Fig. 13), also ob-
tice scale, e.g. dislocations or ions substitution. At larger scales, lithology served in Japan (Fig. 15), well correlates with pressure-induced
contrasts have various origins and faults constitute additional heteroge- antigorite dehydration, i.e. transition to phase A in brucite-saturated
neity. In other words, homogeneity remains a concept. During deforma- conditions (Atg + 71 Br → 17 A + 51 H2O). At higher pressures and
tion, heterogeneities may interact. The evolution of this interaction has temperatures, the highly seismic body could be explained by a series
been studied using percolation models (e.g. Lockner and Madden, 1991; of transformations releasing minor amounts of H2O and forming HP
Hansen et al., 1991; Reuschle, 1998). hydrated talc and phase A (Figs. 4 and 5). Taking into account the addi-
The percolation theory (Broadbent and Hammersley, 1957) allows tional bend of the slab near 120 km depth (Fig. 13c), most of the seis-
describing clusters connectivity in a system close to its critical state, micity between 3 and 6 GPa (90–200 km) correlates quite well with
i.e. “percolation threshold”. Near the threshold, power laws describe serpentine and chlorite destabilizations. Notably, the highly seismic
fractal evolution of clusters geometry (Stauffer and Aharony, 1992). body between 90 and 150 km (3–4.5 GPa) correlates with a series of de-
Widely used to study fluid circulations, the percolation theory has re- hydration reactions affecting chrysotile, antigorite and chlorite with
cently been used to study mechanical stress (Tordesillas et al., 2009, highly negative volume changes, which facilitates rupture nucleations.
2011; Burnley, 2013). Mechanical interaction between clusters can be Such observations are well consistent with experimental results, show-
numerically studied like interactions between fault segments (e.g. ing that dehydration-induced earthquakes are triggered at much lower
Baud and Reuschlé, 1997). Moreover, statistical models of cracks and temperatures above 3 GPa (Ferrand et al., 2017).
crack length distribution give access to relative dimensions and specific Because of its intermediate age (≈45 Ma), the Nazca plate
surfaces of the different phases in the system (e.g. Dienes, 1982; subducting beneath northern Chile and the Altiplano completes the
Guéguen and Dienes, 1989). dataset of P-T conditions corresponding to the nucleation of mantle
A rock volume may deform and/or transform faster depending on earthquakes. Altogether, the relocated hypocenters show a wide range
confinement and drainage conditions (e.g. Wong et al., 1997). of nucleation P-T conditions, confirming that, if mineral transformations
Dehydrating antigorite is first localized in veins containing the dehydra- are involved in earthquakes triggering mechanisms, several reactions
tion products (Hermann et al., 2000; López Sánchez-Vizcaíno et al., are involved. Nonetheless, the destabilization of serpentine minerals
2005) and prone to form anastomosed channels within serpentinites in various conditions (Figs. 4 and 5) remains sufficient to explain the
(Plümper et al., 2017). As discussed in Section 4.2.3, dehydration prod- majority of the seismicity (Fig. 15), along with chlorite minerals for
ucts mobility may strongly depend on the connectivity of hydrous the highest temperatures.
phases networks, leading to potential apparent metastability (Section
4.1.3). Finally, recent seismic attenuation studies (e.g. Shiina et al.,
2018) suggest that the lower Wadati-Benioff plane correlates with 5. Conclusions
slightly lower seismic velocities, which would be consistent with re-
peated strong rheological contrast on the waves pathway, e.g. between This study participates in checking the validity of the DDST model
serpentinized faults and fresh peridotite volumes, as argued by Ferrand (Dehydration-Driven Stress Transfer) compared to other models such
et al. (2017) and Kita and Ferrand (2018). Finally, recent high- as dehydration embrittlement and self-localizing thermal runaway in
resolution data show that the lower plane of the subducting Nazca light of most recent experimental and seismological achievements. Min-
plate contains a minor fluid phase (Bloch et al., 2018), which transiently erals stability limits, which significantly vary with bulk composition,
escapes through the slab toward the subduction interface. This is quite phases chemistry (especially Al3+), fluid saturation, fluid composition
consistent with theoretical calculations on generation and maintenance and other factors, show correlations with seismicity. Thus, if present,
of pore pressure excess in a dehydrating system (Wong et al., 1997). But any unstable mineral, listed in Tables 1, 2 and 3 or not, is likely to con-
pore fluids in such conditions are transient, as the connectivity of fluid tribute in the initiation of strain localization by significant grain size re-
pathways would be favored by preexisting heterogeneities (Debret duction, which drives further propagation by thermal runaway. As a
et al., 2013) and by the structure of serpentine minerals itself (Tutolo major part of the seismicity questioned here corresponds to P-T condi-
et al., 2016; Schwarzenbach, 2016). tions leading to significant negative volume changes upon dehydration,
fluid overpressure is not expected, especially for the lower Wadati-
4.4. About northern Chile Benioff plane of cold subducting slabs.
Previous experiments have shown that hydrous phases are not
Using relocated hypocenters (Sippl et al., 2018) and an accurate metastable in the P-T window considered in this study (Perrillat et al.,
thermal model (Wada and Wang, 2009), I show that the mantle earth- 2005; Inoue et al., 2009; Chollet et al., 2009; Chollet et al., 2011). More-
quakes within the subducting Nazca plate beneath northern Chile could over, studies on dehydrating antigorite show stable sliding and aseismic
be all triggered by mineral transformations (Fig. 13), consistently with deformation in deep serpentinites (Chernak and Hirth, 2010, 2011; Gasc
previous results for other subduction zones (Fig. 15; Abers et al., 2013). et al., 2011, 2017; Okazaki and Hirth, 2016) when serpentinized
226 T.P. Ferrand / Lithos 334–335 (2019) 205–230

peridotite show embrittlement in the same conditions (Jung et al., Akimoto, S., Yamamoto, K., Aoki, K., 1977. Hydroxyl-clinohumite and Hydroxyl-
chondrodite: possible H2O-bearing minerals in the upper mantle. High Pressure Re-
2004; Jung and Green, 2004; Xia, 2013; Ferrand, 2017; Ferrand et al., search 163–172.
2017). Andersen, T.B., Austrheim, H., 2006. Fossil earthquakes recorded by pseudotachylytes in
Antigorite destabilization correlates with most of the mantle seis- mantle peridotite from the Alpine subduction complex of Corsica. Earth and Plane-
tary Science Letters 242 (1–2), 58–72.
micity studied. However, the series of P-T diagrams presented here Andersen, T.B., Austrheim, H., Deseta, N., Silkoset, P., Ashwal, L.D., 2014. Large subduction
show that antigorite stability alone does not explain all observed mantle earthquakes along the fossil Moho in Alpine Corsica. Geology 42 (5), 395–398.
seismicity (Fig. 4), and that other serpentine minerals (Fig. 5), along Andreani, M., Mével, C., Boullier, A.-M., Escartín, J., 2007. Dynamic control on serpentine
crystallization in veins: constraints on hydration processes in oceanic peridotites.
with other hydrous phases should be involved. For instance, as sug- Geochemistry, Geophysics, Geosystems 8 (2).
gested by previous studies (e.g. Hacker et al., 2003; Abers et al., 2013), Andreani, M., Grauby, O., Baronnet, A., Muñoz, M., 2008. Occurrence, composition and
chlorite remains a good candidate to explain HT seismicity in hot mantle growth of polyhedral serpentine. European Journal of Mineralogy 20 (2), 159–171.
Antolik, M., Abercrombie, R.E., Pan, J., Ekström, G., 2006. Rupture characteristics of the
slabs (e.g. Kii, Tokai and Cascadia; Fig. 6) as well as beneath northern
2003 Mw 7.6 mid-Indian Ocean earthquake: implications for seismic properties of
Chile (Figs. 13 and 15). Notably, the transitions from serpentine min- young oceanic lithosphere. Journal of Geophysical Research - Solid Earth 111 (B4).
erals to phases A and 10-Å correlate with higher mantle seismicity be- Bai, Q., Kohlstedt, D.L., 1992. Substantial hydrogen solubility in olivine and implications
tween 90 and 150 km depth. for water storage in the mantle. Nature 357 (6380), 672.
Bailey, E., Holloway, J.R., 2000. Experimental determination of elastic properties of talc to
Mantle earthquakes between 0.5 and 6 GPa, at least, are very likely 800 C, 0.5 GPa; calculations of the effect on hydrated peridotite, and implications for
to be triggered by dehydration reactions. This could also play a role in cold subduction zones. Earth and Planetary Science Letters 183 (3–4), 487–498.
the triggering process of other earthquakes, such as upper-plane earth- Bailey, S.W., Banfield, J.F., Barker, W.W., Katchan, G., 1995. Dozyite, a 1:1 regular inter-
stratification of serpentine and chlorite. American Mineralogist 80 (1–2), 65–77.
quakes. Disqualified in the laboratory (Gasc et al., 2011, 2017; Okazaki Banfield, J.F., Bailey, S.W., Barker, W.W., Smith, R.C., 1995. Complex polytypism: Relation-
and Hirth, 2016; Ferrand et al., 2017), “dehydration embrittlement” is ships between serpentine structural characteristics and deformation. American Min-
not likely in the slab mantle at intermediate depths. Up to date, DDST eralogist 80 (11−12), 1116–1131.
Baronnet, A., Devouard, B., 1996. Topology and crystal growth of natural chrysotile and
(Ferrand et al., 2017) remains the most accurate model to explain the polygonal serpentine. Journal of Crystal Growth 166 (1–4), 952–960.
earthquakes of the lower Wadati-Benioff plane. Furthermore, as Baud, P., Reuschlé, T., 1997. A theoretical approach to the propagation of interacting
discussed in this paper, DDST would not be contradictory with thermal cracks. Geophysical Journal International 130 (2), 460–468.
Berman, R.G., 1988. Internally-consistent thermodynamic data for minerals in the system
runaway models, but complementary. Na2O-K2O-CaO-MgO-FeO-Fe2O3-Al2O3-SiO2-TiO2-H2O-CO2. Journal of Petrology 29
Water-free transformations (e.g. antigorite formation; Fig. 5) and (2), 445–522.
anhydrous reactions (Fig. 10), especially the enstatite transitions, Berman, R.G., Engi, M., Greenwood, H.J., Brown, T.H., 1986. Derivation of internally-
consistent thermodynamic data by the technique of mathematical programming: a
might be involved in the triggering mechanism of earthquakes within
review with application the system MgO-SiO2-H2O. Journal of Petrology 27 (6),
slabs at cold subduction zones. The paragenetic evolution of Al reparti- 1331–1364.
tion in peridotite (Fig. 10) and decarbonation reactions (Fig. 11) also Bloch, W., John, T., Kummerow, J., Salazar, P., Krüger, O.S., Shapiro, S.A., 2018. Watching
show meaningful correlations, which require further investigations. In Dehydration: Seismic Indication for Transient Fluid Pathways in the Oceanic Mantle
of the Subducting Nazca Slab. Geochemistry, Geophysics, Geosystems 19 (9),
any case, the “Transformation-Driven Stress Transfer” (TDST) involves 3189–3207.
minor phases that are prone to induce local grain size reduction. These Bohlen, S.R., Montana, A., Kerrick, D.M., 1991. Precise determinations of the equilibria
local metamorphic transformations are likely to drive stress transfers kyaniteb—N sillimanite and kyaniteb—N andalusite and a revised triple point for
Al2SiO5 polymorphs. American Mineralogist 76 (3–4), 677–680.
that initiate catastrophic strain localization in the mantle. Bonaccorsi, E., Ferraris, G., Merlino, S., 2012. Crystal structure of 2M and 1A
polytypes of balangeroite. Zeitschrift für Kristallographie Crystalline Materials
227 (7), 460–467.
Bonnemains, D., Escartin, J., Mével, C., Andreani, M., Verlaguet, A., 2017. Pervasive silicifi-
Acknowledgments cation and hanging wall overplating along the 13° 20′ N oceanic detachment fault (M
id-A tlantic R idge). Geochemistry, Geophysics, Geosystems 18 (6), 2028–2053.
I thank Christian Sippl for his explanations about recent relocation Bose, K., Navrotsky, A., 1998. Thermochemistry and phase equilibria of hydrous phases in
the system MgO-SiO2-H2O: Implications for volatile transport to the mantle. Journal
data beneath Chile, Ikuko Wada for giving me access to the up-to-date of Geophysical Research - Solid Earth 103 (B5), 9713–9719.
thermal structure at latitude 23°S, and Nadège Hilairet and Javier Brantut, N., Sulem, J., 2012. Strain localization and slip instability in a strain-rate harden-
Escartín for the thoughtful discussions. Many thanks to Natsumi ing, chemically weakening material. Journal of Applied Mechanics 79 (3), 031004.
Brantut, N., Schubnel, A., Corvisier, J., Sarout, J., 2010. Thermochemical pressurization of
Hokanishi for her help with XRF analysis and to Hatsuki Yamauchi and
faults during coseismic slip. Journal of Geophysical Research - Solid Earth 115,
Takehiko Hiraga for great support. I thank the International Research Pro- B05314.
motion Office of the Earthquake Research Institute for supporting this Brantut, N., Han, R., Shimamoto, T., Findling, N., Schubnel, A., 2011. Fast slip with inhibited
study. I also thank Philippe Agard, associate editor of the special issue, temperature rise due to mineral dehydration: evidence from experiments on gyp-
sum. Geology 39 (1), 59–62.
and two anonymous reviewers, whose comments strongly improved Bretscher, A., Hermann, J., Pettke, T., 2018. The influence of oceanic oxidation on
the paper. serpentinite dehydration during subduction. Earth and Planetary Science Letters
499, 173–184.
Broadbent, S.R., Hammersley, J.M., 1957. Percolation processes. Mathematical Proceedings
References of the Cambridge Philosophical Society 53, 629–641.
Bromiley, G.D., Pawley, A.R., 2002. The high-pressure stability of Mg-sursassite in a
Abercrombie, R.E., Ekström, G., 2001. Earthquake slip on oceanic transform faults. Nature model hydrous peridotite: a possible mechanism for the deep subduction of signif-
410 (6824), 74–77. icant volumes of H2O. Contributions to Mineralogy and Petrology 142 (6),
Abers, G.A., Sarker, G., 1996. Dispersion of regional body waves at 100–150 km depth be- 714–723.
neath Alaska: in situ constraints on metamorphism of subducted crust. Geophysical Bromiley, G.D., Pawley, A.R., 2003. The stability of antigorite in the systems MgO-SiO2-
Research Letters 23 (10), 1171–1174. H2O (MSH) and MgO-Al2O3-SiO2-H2O (MASH): the effects of Al3+ substitution on
Abers, G.A., van Keken, P.E., Kneller, E.A., Ferris, A., Stachnik, J.C., 2006. The thermal struc- high-pressure stability. American Mineralogist 88 (1), 99–108.
ture of subduction zones constrained by seismic imaging: Implications for slab dehy- Brudzinski, M., Thurber, C., Hacker, B., Engdahl, E., 2007. Global prevalence of double
dration and wedge flow. Earth and Planetary Science Letters 241 (3), 387–397. Benioff zones. Science 316 (5830), 1472–1474.
Abers, G.A., MacKenzie, L.S., Rondenay, S., Zhang, Z., Wech, A.G., Creager, K.C., 2009. Imag- Burbach, G.V., Frohlich, C., 1986. Intermediate and deep seismicity and lateral structure of
ing the source region of Cascadia tremor and intermediate-depth earthquakes. Geol- subducted lithosphere in the Circum-Pacific Region. Reviews of Geophysics 24 (4),
ogy 37 (12), 1119–1122. 833–874.
Abers, G.A., Nakajima, J., van Keken, P.E., Kita, S., Hacker, B.R., 2013. Thermal-petrological Burnley, P.C., 2013. The importance of stress percolation patterns in rocks and other poly-
controls on the location of earthquakes within subducting plates. Earth and Planetary crystalline materials. Nature Communications 4.
Science Letters 369, 178–187. Caciagli, N.C., Manning, C.E., 2003. The solubility of calcite in water at 6-16 kbar and 500-
Ackermand, D., Seifert, F., Schreyer, W., 1975. Instability of sapphirine at high pressures. 800°C. Contributions to Mineralogy and Petrology 146 (3), 275–285.
Contributions to Mineralogy and Petrology 50 (2), 79–92. Cai, C., Wiens, D.A., Shen, W., Eimer, M., 2018. Water input into the Mariana subduction
Akashi, A., Nishihara, Y., Takahashi, E., Nakajima, Y., Tange, Y., Funakoshi, K.I., 2009. zone estimated from ocean-bottom seismic data. Nature 563, 389–392.
Orthoenstatite/clinoenstatite phase transformation in MgSiO3 at high-pressure and Canales, J.P., Detrick, R.S., Lin, J., Collins, J.A., Toomey, D.R., 2000. Crustal and upper mantle
high-temperature determined by in situ X-ray diffraction: Implications for nature seismic structure beneath the rift mountains and across a nontransform offset at the
of the X discontinuity. Journal of Geophysical Research - Solid Earth 114 (B4).
T.P. Ferrand / Lithos 334–335 (2019) 205–230 227

Mid-Atlantic Ridge (35 N). Journal of Geophysical Research - Solid Earth 105 (B2), Escartín, J., Hirth, G., Evans, B., 1997. Effects of serpentinization on the lithospheric
2699–2719. strength and the style of normal faulting at slow-spreading ridges. Earth and Plane-
Cannat, M., Fontaine, F., Escartín, J., 2010. Serpentinization and associated hydrogen and tary Science Letters 151 (3–4), 181–189.
methane fluxes at slow spreading ridges. Diversity of Hydrothermal Systems on Evans, B.W., 1977. Metamorphism of alpine peridotite and serpentinite. Annual Review of
Slow Spreading Ocean Ridges. vol. 188. AGU, Washington, D. C, pp. 241–264. Earth and Planetary Sciences 5 (1), 397–447.
Caruso, L.J., Chernosky, J.V., 1979. The stability of lizardite. The Canadian Mineralogist 17 Evans, B.W., 2004. The serpentinite multisystem revisited: chrysotile is metastable. Inter-
(4), 757–769. national Geology Review 46 (6), 479–506.
Chernak, L., Hirth, G., 2010. Deformation of antigorite serpentinite at high temperature Evans, B.W., Ghiorso, M.S., 1995. Thermodynamics and petrology of cummingtonite.
and pressure. Earth and Planetary Science Letters 296 (1–2), 23–33. American Mineralogist 80 (7–8), 649–663.
Chernak, L., Hirth, G., 2011. Syndeformational antigorite dehydration produces stable Evans, B.W., Trommsdorff, V., 1978. Petrogenesis of garnet lherzolite, Cima di Gagnone,
fault slip. Geology 39 (9), 847–850. Lepontine Alps. Earth and Planetary Science Letters 40 (3), 333–348.
Chernosky, J.V.J., 1973a. An Experimental Investigation of the Serpentine and Evans, B.W., Trommsdorff, V., 1983. Fluorine hydroxyl titanian clinohumite in Alpine re-
Chlorite Group Minerals in the System MgO-Al2O3-SiO2-H2O. Unpublished PhD the- crystallized garnet peridotite: compositional controls and petrologic significance.
sis. MIT. American Journal of Science 283, 355–369.
Chernosky, J.V., 1973b. The stability of chrysotile, Mg3Si2O5(OH)4, and the free energy of Evans, B.W., Johannes, W., Oterdoom, H., Trommsdorff, V., 1976. A multisystem approach
formation of talc, Mg3Si4O10(OH)2. Geological Society of America Annual Meeting to serpentine equilibria. SMPM 56, 79–93.
Program and Abstracts. Evans, B.W., Dyar, M.D., Kuehner, S.M., 2012. Implications of ferrous and ferric iron in
Chinnery, N.J., Pawley, A.R., Clark, S.M., 1999. In situ observation of the formation of 10 Å antigorite. American Mineralogist 97 (1), 184–196.
phase from talc + H2O at mantle pressures and temperatures. Science 286 (5441), Facq, S., Daniel, I., Montagnac, G., Cardon, H., Sverjensky, D.A., 2014. In situ Raman study
940–942. and thermodynamic model of aqueous carbonate speciation in equilibrium with ara-
Chollet, M., Daniel, I., Koga, K.T., Petitgirard, S., Morard, G., 2009. Dehydration kinetics of gonite under subduction zone conditions. Geochimica et Cosmochimica Acta 132,
talc and 10 Å phase: consequences for subduction zone seismicity. Earth and Plane- 375–390.
tary Science Letters 284 (1–2), 57–64. Ferrand, T.P., 2017. Reproduction expérimentale d'analogues de séismes mantelliques par
Chollet, M., Daniel, I., Koga, K.T., Morard, G., van de Moortèle, B., 2011. Kinetics and mech- déshydratation de l'antigorite & Comparaison à des pseudotachylites naturelles (Doc-
anism of antigorite dehydration: Implications for subduction zone seismicity. Journal toral dissertation, Paris Sciences et Lettres).
of Geophysical Research - Solid Earth 116, 1–9. Ferrand, T.P., 2019. Neither Antigorite nor its Dehydration Is “Metastable” (American
Chopin, C., Schreyer, W., 1983. Magnesiocarpholite and magnesiochloritoid: two Mineralogist – Highlight & Breaktrhough).
index minerals of pelitic blueschists and their preliminary phase relations in Ferrand, T.P., Hilairet, N., Incel, S., Deldicque, D., Labrousse, L., Gasc, J., ... Schubnel, A., 2017.
the model system MgOAl2O3-SiO2- H2O. American Journal of Science Orville Dehydration-driven stress transfer triggers intermediate-depth earthquakes. Nature
283-A, 72–96. Communications 8.
Chopin, C., 1984. Coesite and pure pyrope in high-grade blueschists of the Western Alps: a Ferrand, T.P., Labrousse, L., Eloy, G., Fabbri, O., Hilairet, N., Schubnel, A., 2018. Energy bal-
first record and some consequences. Contributions to Mineralogy and Petrology 86 ance from a mantle pseudotachylyte, Balmuccia, Italy. Journal of Geophysical Research:
(2), 107–118. Solid Earth 123 (5), 3943–3967.
Comodi, P., Cera, F., Nazzareni, S., Dubrovinsky, L., 2007. Raman spectroscopy of the 10-A Ferrando, S., Frezzotti, M.L., Dallai, L., Compagnoni, R., 2005. Multiphase solid inclusions in
phase at simultaneously HP-HT. European Journal of Mineralogy 19 (5), 623–629. UHP rocks (Su-Lu, China): remnants of supercritical silicate-rich aqueous fluids re-
Compagnoni, R., Ferraris, G., Flora, L., 1983. Balangeroite, a new fibrous silicate related to leased during continental subduction. Chemical Geology 223 (1–3), 68–81.
gageite from Balangero, Italy. American Mineralogist 68 (1-2), 214–219. Ferraris, G., Mellini, M., Merlino, S., 1987. Electron-diffraction and electron-microscopy
Compagnoni, R., Ferraris, G., Mellini, M., 1985. Carlosturanite, a new asbestiform rock- study of balangeroite and gageite; crystal structures, polytypism, and fiber texture.
forming silicate from Val Varaita, Italy. American Mineralogist 70, 767–772. American Mineralogist 72 (3–4), 382–391.
Cressey, G., Cressey, B.A., Wicks, F.J., 2008. Polyhedral serpentine: a spherical analogue of Ferraris, G., Ivaldi, G., Goffé, B., 1992. Structural study of a magnesian ferrocarpholite:
polygonal serpentine? Mineralogical Magazine 72 (6), 1229–1242. are carpholites monoclinic? Neues Jahrbuch für Mineralogie – Monatshefte
Debret, B., Sverjensky, D.A., 2017. Highly oxidising fluids generated during serpentinite 337–347.
breakdown in subduction zones. Scientific Reports 7 (1), 10351. Fockenberg, T., 1995. New experimental results up to 100 kbar in the system MgO-Al2O3-
Debret, B., Nicollet, C., Andreani, M., Schwartz, S., Godard, M., 2013. Three steps of SiO2-H2O (MASH): preliminary stability fields of chlorite, chloritoid,
serpentinization in an eclogitized oceanic serpentinization front (Lanzo Massif- staurolite, MgMgAl-pumpellyite, and pyrope. Bochumer Geologische und
Western Alps). Journal of Metamorphic Geology 31 (2), 165–186. Geotechnische Arbeit 44, 39–44.
Debret, B., Andreani, M., Muñoz, M., Bolfan-Casanova, N., Carlut, J., Nicollet, C., ... Trcera, N., Fockenberg, T., 1998a. An experimental investigation on the P-T stability of Mg-staurolite
2014. Evolution of Fe redox state in serpentine during subduction. Earth and Plane- in the system MgO-Al2O3-SiO2-H2O. Contributions to Mineralogy and Petrology 130
tary Science Letters 400, 206–218. (2), 187–198.
Debret, B., Bolfan-Casanova, N., Padrón-Navarta, J.A., Martin-Hernandez, F., Andreani, M., Fockenberg, T., 1998b. An experimental study of the pressure-temperature stability of
Garrido, C.J., López Sánchez-Vizcaíno, V., Gómez-Pugnaire, M.T., Muñoz, M., Trcera, N., MgMgAl-pumpellyite in the system MgO-Al2O3-SiO2-H2O. American Mineralogist
2015. Redox state of iron during high-pressure serpentinite dehydration. Contribu- 83 (3–4), 220–227.
tions to Mineralogy and Petrology 169 (4), 36. Fransolet, A.M., Schreyer, W., 1984. Sudoite, di/trioctahedral chlorite: a stable low-
Debret, B., Bouilhol, P., Pons, M.L., Williams, H., 2018. Carbonate transfer during the onset temperature phase in the system MgO-Al2O3-SiO2-H2O. Contributions to Mineralogy
of slab devolatilization: new insights from Fe and Zn stable isotopes. Journal of Petrol- and Petrology 86 (4), 409–417.
ogy 59 (6), 1145–1166. Frezzotti, M.L., Selverstone, J., Sharp, Z.D., Compagnoni, R., 2011. Carbonate dissolution
Deer, W.A., Zussman, J., 1962. Rock-Forming Minerals: Sheet Silicates. vol. 3. Wiley. during subduction revealed by diamond-bearing rocks from the Alps. Nature Geosci-
Delescluse, M., Chamot-Rooke, N., 2007. Instantaneous deformation and kinematics of the ence 4 (10), 703.
India-Australia Plate. Geophysical Journal International 168 (2), 818–842. Frohlich, C., 2006. Deep Earthquakes. Cambridge university Press.
Deplus, C., Diament, M., Hébert, H., Bertrand, G., Dominguez, S., Dubois, J., Malod, J., Fumagalli, P., Poli, S., 2005. Experimentally determined phase relations in hydrous perido-
Patriat, P., Pontoise, B., Sibilla, J.-J., 1998. Direct evidence of active deformation in tites to 6.5 GPa and their consequences on the dynamics of subduction zones. Journal
the eastern Indian oceanic plate. Geology 26 (2), 131–134. of Petrology 46 (3), 555–578.
Deseta, N., Andersen, T.B., Ashwal, L., 2014a. A weakening mechanism for intermediate- Fumagalli, P., Stixrude, L., 2007. The 10 Å phase at high pressure by first principles calcu-
depth seismicity? Detailed petrographic and microtextural observations from lations and implications for the petrology of subduction zones. Earth and Planetary
blueschist facies pseudotachylytes, Cape Corse, Corsica. Tectonophysics 610, Science Letters 260 (1–2), 212–226.
138–149. Fumagalli, P., Zanchetta, S., Poli, S., 2009. Alkali in phlogopite and amphibole and their ef-
Deseta, N., Ashwal, L., Andersen, T.B., 2014b. Initiating intermediate-depth earthquakes: fects on phase relations in metasomatized peridotites: a high-pressure study. Contri-
insights from a HP-LT ophiolite from Corsica. Lithos 206, 127–146. butions to Mineralogy and Petrology 158 (6), 723.
Dienes, J.K., 1982. Permeability, percolation and statistical crack mechanics. The 23rd US Fumagalli, P., Poli, S., Fischer, J., Merlini, M., Gemmi, M., 2014. The high-pressure stability
Symposium on Rock Mechanics (USRMS). American Rock Mechanics Association. of chlorite and other hydrates in subduction melanges: experiments in the
Dódony, I., Buseck, P.R., 2004. Serpentines close-up and intimate: an HRTEM view. Inter- system Cr2O3-MgO-Al2O3-SiO2-H2O. Contributions to Mineralogy and Petrology 167
national Geology Review 46 (6), 507–527. (2), 979.
Dunkel, K.G., Austrheim, H., Renard, F., Cordonnier, B., Jamtveit, B., 2017. Localized Fyfe, W.S., 1958. Metamorphic reactions and metamorphic facies. Geological Society of
slip controlled by dehydration embrittlement of partly serpentinized dunites, America 73.
Leka Ophiolite complex, Norway. Earth and Planetary Science Letters 463, Garth, T., Rietbrock, A., 2014. Order of magnitude increase in subducted H2O due to hy-
277–285. drated normal faults within the Wadati-Benioff zone. Geology 42 (3), 207–210.
Dunn, R.A., Arai, R., Eason, D.E., Canales, J.P., Sohn, R.A., 2017. Three-dimensional seismic Garth, T., Rietbrock, A., 2017. Constraining the hydration of the subducting Nazca plate
structure of the Mid-Atlantic Ridge: an investigation of tectonic, magmatic, and hy- beneath Northern Chile using subduction zone guided waves. Earth and Planetary
drothermal processes in the rainbow area. Journal of Geophysical Research - Solid Science Letters 474, 237–247.
Earth 122 (12), 9580–9602. Gasc, J., Schubnel, A., Brunet, F., Guillon, S., Mueller, H., Lathe, C., 2011. Simultaneous
Dymek, R.F., Boak, J.L., Brothers, S.C., 1988. Titanian chondrodite-and titanian acoustic emissions monitoring and synchrotron X-ray diffraction at high pressure
clinohumite-bearing metadunite from the 3800 Ma Isua supracrustal belt, West and temperature: Calibration and application to serpentinite dehydration. Physics
Greenland: Chemistry, petrology, and origin. American Mineralogist 73. of the Earth and Planetary Interiors 189, 121–133.
Emry, E.L., Wiens, D.A., Garcia-Castellanos, D., 2014. Faulting within the pacific plate at the Gasc, J., Hilairet, N., Yu, T., Ferrand, T., Schubnel, A., Wang, Y., 2017. Faulting of natural
mariana trench: implications for plate interface coupling and subduction of hydrous serpentinite: Implications for intermediate-depth seismicity. Earth and Planetary Sci-
minerals. Journal of Geophysical Research - Solid Earth 119 (4), 3076–3095. ence Letters 474, 138–147.
228 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Giorgetti, G., Memmi, I., Nieto, F., 1997. Microstructures of intergrown phyllosilicate Kanzaki, M., 1991. Dehydration of brucite (Mg(OH)2) at high pressures detected by differ-
grains from Verrucano metasediments (northern Apennines, Italy). Contributions to ential thermal analysis. Geophysical Research Letters 18 (12), 2189–2192.
Mineralogy and Petrology 128 (2–3), 127–138. Kelemen, P.B., Hirth, G.A., 2007. Periodic shear-heating mechanism for intermediate-
Green, H.W., Houston, H., 1995. The mechanics of deep earthquakes. Annual Review of depth earthquakes in the mantle. Nature 446, 787–790.
Earth and Planetary Sciences 23 (1), 169–213. Kelemen, P.B., Hirth, G., 2012. Reaction-driven cracking during retrograde metamor-
Green II, H.W., Shi, F., Bozhilov, K., Xia, G., Reches, Z., 2015. Phase transformation and phism: olivine hydration and carbonation. Earth and Planetary Science Letters 345,
nanometric flow cause extreme weakening during fault slip. Nature Geoscience 8 81–89.
(6), 484–489. Kelley, D.S., Karson, J.A., Früh-Green, G.L., Yoerger, D.R., Shank, T.M., Butterfield, D.A., ...
Greenwood, H.J., 1963. The synthesis and stability of anthophyllite. Journal of Petrology 4 Jakuba, M., 2005. A serpentinite-hosted ecosystem: the lost city hydrothermal field.
(3), 317–351. Science 307 (5714), 1428–1434.
Groppo, C., Compagnoni, R., 2007. Metamorphic veins from the serpentinites of the Pie- Kendrick, M.A., Scambelluri, M., Honda, M., Phillips, D., 2011. High abundances of noble
monte Zone, western Alps, Italy: a review. Periodico di Mineralogia 76, 127–153. gas and chlorine delivered to the mantle by serpentinite subduction. Nature Geosci-
Guéguen, Y., Dienes, J., 1989. Transport properties of rocks from statistics and percolation. ence 4 (11), 807.
Mathematical Geology 21 (1), 1–13. Kirby, S.H., 1987. Localized polymorphic phase transformations in high-pressure faults
Guéguen, Y., Palciauskas, V., 1994. Introduction to the Physics of Rocks. Princeton Univer- and applications to the physical mechanism of deep earthquakes. Journal of Geophys-
sity Press. ical Research - Solid Earth 92 (B13), 13789–13800.
Guggenbuehl, E., 1994. Stabilität von Talk, Forsterit und Enstatit im Feld und Experiment. Kirby, S., Engdahl, R.E., Denlinger, R., 1996. Intermediate-depth intraslab earthquakes and
Diploma thesis. University of Zurich, Switzerland. arc volcanism as physical expressions of crustal and uppermost mantle metamor-
Hacker, B.R., Peacock, S.M., Abers, G.A., 2003. Subduction factory: 2. Intermediate-depth phism in subducting slabs. Subduction Top to Bottom 195–214.
earthquakes in subducting slabs are linked to metamorphic dehydration reactions. Kita, S., Ferrand, T.P., 2018. Physical Mechanisms of Oceanic Mantle Earthquakes: Com-
Journal of Geophysical Research - Solid Earth 108. parison of Natural and Experimental Events (Scientific Reports).
Hansen, A., Hinrichsen, E.L., Roux, S., 1991. Scale-invariant disorder in fracture and related Kita, S., Okada, T., Hasegawa, A., Nakajima, J., Matsuzawa, T., 2010a. Anomalous deepening
breakdown phenomena. Physical Review B 43 (1), 665. of a seismic belt in the upper-plane of the double seismic zone in the Pacific slab be-
Hasegawa, A., Umino, N., Takagi, A., 1978. Double-planed structure of the deep seismic neath the Hokkaido corner: possible evidence for thermal shielding caused by
zone in the northeastern Japan arc. Tectonophysics 47 (1–2), 43–58. subducted forearc crust materials. Earth and Planetary Science Letters 290 (3),
Henry, P., 2000. Fluid flow at the toe of the Barbados accretionary wedge constrained by 415–426.
thermal, chemical, and hydrogeologic observations and models. Journal of Geophys- Kita, S., Okada, T., Hasegawa, A., Nakajima, J., Matsuzawa, T., 2010b. Existence of
ical Research - Solid Earth 105 (B11), 25,855–25,872. interplane earthquakes and neutral stress boundary between the upper and lower
Hensen, B.J., Essene, E.J., 1971. Stability of pyrope-quartz in the system MgO-Al2O3-SiO2. planes of the double seismic zone beneath Tohoku and Hokkaido, northeastern
Contributions to Mineralogy and Petrology 30 (1), 72–83. Japan. Tectonophysics 496 (1), 68–82.
Hermann, J., Müntener, O., Scambelluri, M., 2000. The importance of serpentinite Kithara, S., Takenouchi, S., Kennedy, G.C., 1966. Phase relations in the system MgO-
mylonites for subduction and exhumation of oceanic crust. Tectonophysics 327 SiO2-H2O at high-temperatures and pressures. American Journal of Science 264,
(3–4), 225–238. 223–233.
Hilairet, N., Daniel, I., Reynard, B., 2006. Equation of state of antigorite, stability field Klein, F., McCollom, T.M., 2013. From serpentinization to carbonation: new insights
of serpentines, and seismicity in subduction zones. Geophysical Research Letters from a CO2 injection experiment. Earth and Planetary Science Letters 379,
33 (2). 137–145.
Hirose, F., Nakajima, J., Hasegawa, A., 2008. Three-dimensional seismic velocity structure Kunze, G., 1961. Antigorit. Strukturtheoretische Grundlagen und ihre praktische
and configuration of the Philippine Sea slab in southwestern Japan estimated by Bedeutung für die weitere Serpentin-Forschung. Fortschritte in Mineralogie 9,
double-difference tomography. Journal of Geophysical Research - Solid Earth 113 206–324.
(B9). Kurzawski, R.M., Stipp, M., Niemeijer, A.R., Spiers, C.J., Behrmann, J.H., 2016. Earthquake
Holland, T.J.B., Powell, R.T.J.B., 1998. An internally consistent thermodynamic data nucleation in weak subducted carbonates. Nature Geoscience 9 (9), 717.
set for phases of petrological interest. Journal of Metamorphic Geology 16 (3), Kushiro, I., Satake, H., Akimoto, S., 1975. Carbonate-silicate reactions at high presures and
309–343. possible presence of dolomite and magnesite in the upper mantle. Earth and Plane-
Humphreys, E.R., Bailey, K., Hawkesworth, C.J., Wall, F., Najorka, J., Rankin, A.H., 2010. Ara- tary Science Letters 28, 116–120.
gonite in olivine from Calatrava, Spain - evidence for mantle carbonatite melts from Liu, T., Wang, D., Shen, K., Liu, C., Yi, L., 2019. Kinetics of antigorite dehydration: Rapid de-
N100 km depth. Geology 38 (10), 911–914. hydration as a trigger for lower-plane seismicity in subduction zones. American Min-
Incel, S., Hilairet, N., Labrousse, L., John, T., Deldicque, D., Ferrand, T., Wang, Y., Renner, J., eralogist: Journal of Earth and Planetary Materials 104 (2), 282–290.
Morales, L., Schubnel, A., 2017. Laboratory earthquakes triggered during Lockner, D.A., Madden, T.R., 1991. A multiple-crack model of brittle fracture: 2. Time-
eclogitization of lawsonite-bearing blueschist. Earth and Planetary Science Letters dependent simulations. Journal of Geophysical Research - Solid Earth 96 (B12),
459, 320–331. 19643–19654.
Inoue, T., Yoshimi, I., Yamada, A., Kikegawa, T., 2009. Time-resolved X-ray diffraction anal- López Sánchez-Vizcaíno, V., Trommsdorff, V., Gómez-Pugnaire, M.T., Garrido, C.J.,
ysis of the experimental dehydration of serpentine at high pressure. Journal of Min- Müntener, O., Connolly, J.A.D., 2005. Petrology of titanian clinohumite and olivine at
eralogical and Petrological Sciences 104 (2), 105–109. the high-pressure breakdown of antigorite serpentinite to chlorite harzburgite
Irifune, T., Kubo, N., Isshiki, M., Yamasaki, Y., 1998. Phase transformations in serpentine (Almirez Massif, S. Spain). Contributions to Mineralogy and Petrology 149 (6),
and transportation of water into the lower mantle. Geophysical Research Letters 25 627–646.
(2), 203–206. Lykins, R.W., Jenkins, D.M., 1992. Experimental determination of pargasite stability rela-
Ivandic, M., Grevemeyer, I., Bialas, J., Petersen, C.J., 2010. Serpentinization in the trench- tions in the presence of orthopyroxene. Contributions to Mineralogy and Petrology
outer rise region offshore of Nicaragua: constraints from seismic refraction and 112 (2–3), 405–413.
wide-angle data. Geophysical Journal International 180 (3), 1253–1264. Majzlan, J., Grevel, K.D., Navrotsky, A., 2003. Thermodynamics of Fe oxides: Part II.
Jamtveit, B., Malthe-Sorenssen, A., Kostenko, O., 2008. Reaction enhanced permeability Enthalpies of formation and relative stability of goethite (α-FeOOH),
during retrogressive metamorphism. Earth and Planetary Science Letters 267, lepidocrocite (γ-FeOOH), and maghemite (γ-Fe2O3). American Mineralogist 88 (5–
620–627. 6), 855–859.
Jamtveit, B., Putnis, C., Malthe-Sorenssen, A., 2009. Reaction induced fracturing Manning, C.E., 1995. Phase-equilibrium controls on SiO2 metasomatism by aqueous fluid
during replacement processes. Contributions to Mineralogy and Petrology 157, in subduction zones: reaction at constant pressure and temperature. International
127–133. Geology Review 37 (12), 1074–1093.
Jenkins, D.M., 1983. Stability and composition relations of calcic amphiboles in ultramafic Martin, R.F., Donnay, G., 1972. Hydroxyl in the mantle. American mineralogist. Journal of
rocks. Contributions to Mineralogy and Petrology 83 (3–4), 375–384. Earth and Planetary Materials 57, 554–570.
Johannes, W., 1968. Experimental investigation of the reaction forsterite + H2O = ser- Martinez, I., Zhang, J., Reeder, R.J., 1996. In situ X-ray diffraction of aragonite and dolomite
pentine + brucite. Contributions to Mineralogy and Petrology 19, 309–315. at high pressure and high temperature: evidence for dolomite breakdown to arago-
Johannes, W., Puhan, D., 1971. The calcite-aragonite transition, reinvestigated. Contribu- nite and magnesite. American Mineralogist 81 (5–6), 611–624.
tions to Mineralogy and Petrology 31 (1), 28–38. Massonne, H.J., Mirwald, P.W., Schreyer, W., 1981. Experimentelle Uberprtifung der
Johannes, W., Schreyer, W., 1981. Experimental introduction of CO2 and H2O into Mg- Reaktionskurve Chlorit + Quarz = Talk + Disthen im System MgO-Al2O3-SiO2-
cordierite. American Journal of Science 281 (3), 299–317. H2O. Fortschritte in Mineralogie 59, 12–123.
John, T., Medvedev, S., Rüpke, L.H., Andersen, T.B., Podladchikov, Y.Y., Austrheim, H., 2009. McCollom, T.M., 2007. Geochemical constraints on sources of metabolic energy for
Generation of intermediate-depth earthquakes by self-localizing thermal runaway. chemolithoautotrophy in ultramafic-hosted deep-sea hydrothermal systems. Astro-
Nature Geoscience 2 (2), 137. biology 7, 933–950.
John, T., Scambelluri, M., Frische, M., Barnes, J.D., Bach, W., 2011. Dehydration of Mellini, M., 1986. Chrysotile and polygonal serpentine from the Balangero serpentinite.
subducting serpentinite: implications for halogen mobility in subduction zones and Mineralogical Magazine 50, 301–306.
the deep halogen cycle. Earth and Planetary Science Letters 308 (1–2), 65–76. Mellini, M., Zanazzi, P.F., 1989. Effects of pressure on the structure of lizardite-1T.
Jung, H., Green, H.W., 2004. Experimental faulting of serpentinite during dehydration: Im- European Journal of Mineralogy 1 (1), 13–19.
plications for earthquakes, seismic low-velocity zones, and anomalous hypocenter Mellini, M., Trommsdorff, V., Compagnoni, R., 1987. Antigorite polysomatism: behaviour
distributions in subduction zones. International Geology Review 46 (12), 1089–1102. during progressive metamorphism. Contributions to Mineralogy and Petrology 97,
Jung, H., Green II, H.W., Dobrzhinetskaya, L.F., 2004. Intermediate-depth earthquake 147–155.
faulting by dehydration embrittlement with negative volume change. Nature 428 Ménez, B., Pasini, V., Brunelli, D., 2012. Life in the hydrated suboceanic mantle. Nature
(6982), 545–549. Geoscience 5 (2), 133.
T.P. Ferrand / Lithos 334–335 (2019) 205–230 229

Merkulova, M., Muñoz, M., Vidal, O., Brunet, F., 2016. Role of iron content on serpentinite Ranero, C., Morgan, J., McIntosh, K., Reichert, C., 2003. Bending-related faulting and man-
dehydration depth in subduction zones: experiments and thermodynamic modeling. tle serpentinization at the Middle America trench. Nature 425, 367–373.
Lithos 264, 441–452. Reuschle, T., 1998. A network approach to fracture: the effect of heterogeneity and load-
Milesi, V., Guyot, F., Brunet, F., Richard, L., Recham, N., Benedetti, M., Dairou, J., Prinzhofer, ing conditions. Pure and Applied Geophysics 152 (4), 641–665.
A., 2015. Formation of CO2, H2 and condensed carbon from siderite dissolution in the Reynard, B., Nakajima, J., Kawakatsu, H., 2010. Earthquakes and plastic deformation of an-
200-300°C range and at 50 MPa. Geochimica et Cosmochimica Acta 154, 201–211. hydrous slab mantle in double Wadati-Benioff zones. Geophysical Research Letters
Mitchell, R.H., Putnis, A., 1988. Polygonal serpentine in segregation-textured kimberlite. 37, L24309.
The Canadian Mineralogist 26 (4), 991–997. Rubie, D.C., 1990. Role of kinetics in the formation and preservation of eclogites. Eclogite
Morishita, T., Kodera, T., 1998. Finding of corundum-bearing gabbro boulder possibly de- Facies Rocks, pp. 111–140.
rived from the Horoman Peridotite Complex, Hokkaido, northern Japan. 岩鉱. vol. 93 Rudge, J.F., Kelemen, P.B., Spiegelman, M., 2010. A simple model of reaction-induced
(2) pp. 52–63. cracking applied to serpentinization and carbonation of peridotite. Earth and Plane-
Morishita, T., Arai, S., Gervilla, F., 2001. High-pressure aluminous mafic rocks from the tary Science Letters 291 (1–4), 215–227.
Ronda peridotite massif, southern Spain: significance of sapphirine-and corundum- Scambelluri, M., Strating, E.H., Piccardo, G.B., Vissers, R.L.M., Rampone, E., 1991. Alpine
bearing mineral assemblages. Lithos 57 (2–3), 143–161. olivine-and titanian clinohumite-bearing assemblages in the Erro-Tobbio peridotite
Naif, S., Key, K., Constable, S., Evans, R.L., 2015. Water-rich bending faults at the Middle (Voltri Massif, NW Italy). Journal of Metamorphic Geology 9 (1), 79–91.
America Trench. Geochemistry, Geophysics, Geosystems 16 (8), 2582–2597. Scambelluri, M., Piccardo, G.B., Philippot, P., Robbiano, A., Negretti, L., 1997. High salinity
Naif, S., Key, K., Constable, S., Evans, R.L., 2016. Porosity and fluid budget of a water-rich fluid inclusions formed from recycled seawater in deeply subducted alpine
megathrust revealed with electromagnetic data at the Middle America Trench. Geo- serpentinite. Earth and Planetary Science Letters 148 (3–4), 485–499.
chemistry, Geophysics, Geosystems 17, 4495–4516. Scambelluri, M., Bottazzi, P., Trommsdorff, V., Vannucci, R., Hermann, J., Gòmez-Pugnaire,
Nakajima, J., Tsuji, Y., Hasegawa, A., Kita, S., Okada, T., Matsuzawa, T., 2009. Tomographic M.T., López Sánchez-Vizcaíno, V., 2001. Incompatible element-rich fluids released by
imaging of hydrated crust and mantle in the subducting Pacific slab beneath Hok- antigorite breakdown in deeply subducted mantle. Earth and Planetary Science Let-
kaido, Japan: evidence for dehydration embrittlement as a cause of intraslab earth- ters 192 (3), 457–470.
quakes. Gondwana Research 16 (3–4), 470–481. Scambelluri, M., Fiebig, J., Malaspina, N., Müntener, O., Pettke, T., 2004. Serpentinite sub-
Obara, K., Kato, A., 2016. Connecting slow earthquakes to huge earthquakes. Science 353 duction: implications for fluid processes and trace-element recycling. International
(6296), 253–257. Geology Review 46 (7), 595–613.
Obata, M., 1976. The solubility of Al2O3 in orthopyroxenes in spinel and plagioclase peri- Scambelluri, M., Bebout, G.E., Belmonte, D., Gilio, M., Campomenosi, N., Collins, N.,
dotites and spinel pyroxenite. American Mineralogist 61 (7-8), 804–816. Crispini, L., 2016. Carbonation of subduction-zone serpentinite (high-pressure
O'Hanley, D.S., Chernosky, J.V., Wicks, F.J., 1989. The stability of lizardite and chrysotile. ophicarbonate; Ligurian Western Alps) and implications for the deep carbon cycling.
The Canadian Mineralogist 27 (3), 483–493. Earth and Planetary Science Letters 441, 155–166.
O'Hara, M.J., Richardson, S.W., Wilson, G., 1971. Garnet-peridotite stability and occurrence Scambelluri, M., Pennacchioni, G., Gilio, M., Bestmann, M., Plümper, O., Nestola, F., 2017.
in crust and mantle. Contributions to Mineralogy and Petrology 32 (1), 48–68. Fossil intermediate-depth earthquakes in subducting slabs linked to differential
Ohnaka, M., 2003. Constitutive scaling law and a unified comprehension for frictional slip stress release. Nature Geoscience 10, 960–996.
failure, shear fracture of intact rock and earthquake rupture. Journal of Geophysical Scambelluri, M., Pettke, T., Cannaò, E., 2015. Fluid-related inclusions in Alpine high-pres-
Research - Solid Earth 108, B2. sure peridotite reveal trace element recycling during subduction-zone dehydration of
Ohuchi, T., Lei, X., Ohfuji, H., Higo, Y., Tange, Y., Sakai, T., ... Irifune, T., 2017. Intermediate- serpentinized mantle (Cima di Gagnone, Swiss Alps). Earth and Planetary Science
depth earthquakes linked to localized heating in dunite and harzburgite. Nature Geo- Letters 429, 45–59.
science 10 (10), 771. Schmidt, M.W., Poli, S., 1994. The stability of lawsonite and zoisite at high pressures: ex-
Okazaki, K., Hirth, G., 2016. Dehydration of lawsonite could directly trigger earthquakes in periments in CASH to 92 kbar and implications for the presence of hydrous phases in
subducting oceanic crust. Nature 530 (7588), 81. subducted lithosphere. Earth and Planetary Science Letters 124 (1–4), 105–118.
Padrón-Navarta, J.A., Sánchez-Vizcaíno, V.L., Garrido, C.J., Gómez-Pugnaire, M.T., Jabaloy, A., Schreyer, W., 1988. Experimental studies on metamorphism of crustal rocks under man-
Capitani, G.C., Mellini, M., 2008. Highly ordered antigorite from Cerro del Almirez HP- tle pressures. Mineralogical Magazine 52, 1), 1–26.
HT serpentinites, SE Spain. Contributions to Mineralogy and Petrology 156 (5), 679. Schreyer, W., Medenbach, O., Abraham, K., Gebert, W., Müller, W.F., 1982. Kulkeite, a new
Padrón-Navarta, J.A., Tommasi, A., Garrido, C.J., López Sánchez-Vizcaíno, V., Gómez- metamorphic phyllosilicate mineral: Ordered 1∶ 1 chlorite/talc mixed-layer. Contri-
Pugnaire, M.T., Jabaloy, A., Vauchez, A., 2010. Fluid transfer into the wedge controlled butions to Mineralogy and Petrology 80 (2), 103–109.
by high-pressure hydrofracturing in the cold top-slab mantle. Earth and Planetary Schreyer, W., Maresch, W.V., Baller, T., 1991. A new hydrous, high-pressure phase with a
Science Letters 297 (1–2), 271–286. pumpellyite structure in the system MgO-Al2O3-SiO2-H2O. Progress in Metamorphic
Padrón-Navarta, J.A., López Sánchez-Vizcaíno, V., Hermann, J., Connolly, J.A., Garrido, C.J., and Magmatic Petrology. Cambridge University Press, Cambridge, pp. 47–64.
Gómez-Pugnaire, M.T., Marchesi, C., 2013. Tschermak's substitution in antigorite Schubnel, A., Brunet, F., Hilairet, N., Gasc, J., Wang, Y., Green II, H.W., 2013. Deep focus
and consequences for phase relations and water liberation in high-grade earthquake analogs recorded at high pressure and temperature in the laboratory. Sci-
serpentinites. Lithos 178, 186–196. ence 341, 1377–1380.
Parsons, T., Trehu, A.M., Luetgert, J.H., Miller, K., Kilbride, F., Wells, R.E., ... Christensen, N.I., Schurr, B., Rietbrock, A., Asch, G., Kind, R., Oncken, O., 2006. Evidence for lithospheric de-
1998. A new view into the Cascadia subduction zone and volcanic arc: Implications tachment in the Central Andes from local earthquake tomography. Tectonophysics
for earthquake hazards along the Washington margin. Geology 26 (3), 199–202. 415 (1–4), 203–223.
Pasteris, J.D., 1981. Occurrence of graphite in serpentinized olivines in kimberlite. Geology Schwartz, S., Guillot, S., Reynard, B., Lafay, R., Debret, B., Nicollet, C., Lanari, P., Auzende,
9 (8), 356–359. A.L., 2013. Pressure-temperature estimates of the lizardite/antigorite transition in
Pawley, A., 1998. The reaction talc + forsterite = enstatite + H2O: New experimental re- high pressure serpentinites. Lithos 178, 197–210.
sults and petrological implications. American Mineralogist 83 (1), 51–57. Schwarzenbach, E.M., 2016. Research focus: serpentinization and the formation of fluid
Pawley, A., 2000. Stability of clinohumite in the system MgO-SiO2-H2O. Contributions to pathways. Geology 44 (2), 175–176.
Mineralogy and Petrology 138, 284–291. Schweizer, E., 1840. Ueber den Antigorit, ein neues Mineral. Advanced Synthesis & Catal-
Pawley, A., 2003. Chlorite stability in mantle peridotite: the reaction clinochlore + ysis 21 (1), 105–109.
enstatite = forsterite + pyrope + H2O. Contributions to Mineralogy and Petrology Shiina, T., Nakajima, J., Matsuzawa, T., 2018. P-wave attenuation in the Pacific slab be-
144 (4), 449–456. neath northeastern Japan revealed by the spectral ratio of intraslab earthquakes.
Pawley, A.R., Wood, B.J., 1995. The high-pressure stability of talc and 10Å -phase: poten- Earth and Planetary Science Letters 489, 37–48.
tial storage for H2O in subduction zones. American Mineralogist 80, 998–1003. Shillington, D.J., Bécel, A., Nedimović, M.R., Kuehn, H., Webb, S.C., Abers, G.A., ... Mattei-
Pawley, A.R., Wood, B.J., 1996. The low-pressure stability of phase A, Mg7Si2O8(OH)6. Con- Salicrup, G.A., 2015. Link between plate fabric, hydration and subduction zone seis-
tributions to Mineralogy and Petrology 124, 90–97. micity in Alaska. Nature Geoscience 8 (12), 961–964.
Peacock, S., 2001. Are the lower planes of double seismic zones caused by serpentine de- Sippl, C., Schurr, B., Asch, G., Kummerow, J., 2018. Seismicity structure of the Northern
hydration in subducting oceanic mantle? Geology 29 (4), 299–302. Chile forearc fromN 100,000 double-difference relocated hypocenters. Journal of Geo-
Pens, M., Andreani, M., Daniel, I., Perrillat, J.P., Cardon, H., 2016. Contrasted effect of alumi- physical Research: Solid Earth 123 (5), 4063–4087.
num on the serpentinization rate of olivine and orthopyroxene under hydrothermal Staudigel, H., Schreyer, W., 1977. The upper thermal stability of clinochlore,
conditions. Chemical Geology 441, 256–264. Mg5Al[AlSi3O10](OH)8, at 10-35 kb PH2O. Contributions to Mineralogy and Petrology
Perrillat, J.-P., Daniel, I., Koga, K.T., Reynard, B., Cardon, H., Crichton, W.A., 2005. Kinetics of 61 (2), 187–198.
antigorite dehydration; a real-time X-ray diffraction study. Earth and Planetary Sci- Stauffer, D., Aharony, A., 1992. Introduction to Percolation Theory. 2nd ed. Taylor &
ence Letters 236, 899–913. Francis, London.
Platt, J.D., Brantut, N., Rice, J.R., 2015. Strain localization driven by thermal decomposition Sverjensky, D.A., Stagno, V., Huang, F., 2014. Important role for organic carbon
during seismic shear. Journal of Geophysical Research - Solid Earth 120 (6), 4405–4433. in subduction-zone fluids in the deep carbon cycle. Nature Geoscience 7 (12),
Plümper, O., John, T., Podladchikov, Y.Y., Vrijmoed, J.C., Scambelluri, M., 2017. Fluid escape 909.
from subduction zones controlled by channel-forming reactive porosity. Nature Geo- Syracuse, E.M., van Keken, P.E., Abers, G.A., 2010. The global range of subduction zone
science 10 (2), 150. thermal models. Physics of the Earth and Planetary Interiors 183 (1–2), 73–90.
Plunder, A., Thieulot, C., van Hinsbergen, D.J., 2018. The effect of obliquity on temperature Tenthorey, E., Cox, S.F., 2003. Reaction-enhanced permeability during serpentinite dehy-
in subduction zones: insights from 3-D numerical modeling. Solid Earth 9 (3), 759. dration. Geology 31 (10), 921–924.
Prieto, G.A., Florez, M., Barrett, S.A., Beroza, G.C., Pedraza, P., Blanco, J.F., Poveda, E., 2013. Thielmann, M., 2018a. Grain size assisted thermal runaway as a nucleation mechanism for
Seismic evidence for thermal runaway during intermediate-depth earthquake rup- continental mantle earthquakes: Impact of complex rheologies. Tectonophysics. 746,
ture. Geophysical Research Letters 40 (23), 6064–6068. 611–623.
Raleigh, C.B., Paterson, M.S., 1965. Experimental deformation of serpentinite and its tec- Thielmann, M., 2018b. Intermediate depth earthquakes due to grain size assisted thermal
tonic implications. Journal of Geophysical Research 70 (16), 3965–3985. runaway: what are the odds? JpGU Meeting 2018.
230 T.P. Ferrand / Lithos 334–335 (2019) 205–230

Thielmann, M., Rozel, A., Kaus, B.J.P., Ricard, Y., 2015. Intermediate-depth earthquake gen- Viti, C., 2010. Serpentine minerals discrimination by thermal analysis. American Mineral-
eration and shear zone formation caused by grain size reduction and shear heating. ogist 95 (4), 631–638.
Geology 43 (9), 791–794. Wada, I., Wang, K., 2009. Common depth of slab-mantle decoupling: Reconciling diver-
Tordesillas, A., Zhang, J., Behringer, R., 2009. Buckling force chains in dense granular as- sity and uniformity of subduction zones. Geochemistry, Geophysics, Geosystems 10
semblies: physical and numerical experiments. Geomechanics and Geoengineering: (10).
An International Journal 4 (1), 3–16. Weiss, M., 1997. Clinohumites: A Field and Experimental Study. Doctoral dissertation.
Tordesillas, A., Shi, J., Tshaikiwsky, T., 2011. Stress-dilatancy and force chain evolution. In- Swiss Federal Institute of Technology Zurich.
ternational Journal for Numerical and Analytical Methods in Geomechanics 35 (2), Wong, T.F., Ko, S.C., Olgaard, D.L., 1997. Generation and maintenance of pore pressure ex-
264–292. cess in a dehydrating system 2. Theoretical analysis. Journal of Geophysical Research -
Trommsdorff, V., 1983. Metamorphose magnesiumreicher Gesteine: Kritischer Vergleich Solid Earth 102 (B1), 841–852.
von Natur, experiment und thermodynamischer datenbasis. Fortschritte in Worden, R.H., Droop, G.T.R., Champness, P.E., 1991. The reaction antigorite → olivine + talc
Mineralogie 61, 283–308. + H2O in the Bergell aureole, N. Italy. Mineralogical Magazine 55 (380), 367–377.
Trommsdorff, V., Evans, B.W., 1977. Antigorite-ophicarbonates: phase relations in a Wunder, B., 1998. Equilibrium experiments in the system MgO–SiO2–H2O (MSH): stabil-
portion of the system CaO-MgO-SiO2-H2O-CO2. Contributions to Mineralogy and ity fields of clinohumite-OH[Mg9Si4O16(OH)2], chondrodite-OH[Mg5Si2O8(OH)2] and
Petrology 60 (1), 39–56. phase A(Mg7Si2O8(OH)6). Contributions to Mineralogy and Petrology 132 (2),
Trommsdorff, V., Evans, B.W., 1980. Titanian hydroxyl-clinohumite: formation and break- 111–120.
down in antigorite rocks (Malenco, Italy). Contributions to Mineralogy and Petrology Wunder, B., Schreyer, W., 1997. Antigorite: High pressure stability in the system MgO-
72 (3), 229–242. SiO2-H2O (MSH). Lithos 41, 213–227.
Trommsdorff, V., Sánchez-Vizcaíno, V.L., Gomez-Pugnaire, M.T., Müntener, O., 1998. High Wunder, B., Wirth, R., Gottschalk, M., 2001. Antigorite pressure and temperature depen-
pressure breakdown of antigorite to spinifex-textured olivine and orthopyroxene, SE dence of polysomatism and water content. European Journal of Mineralogy 13 (3),
Spain. Contributions to Mineralogy and Petrology 132 (2), 139–148. 485–496.
Tutolo, B.M., Mildner, D.F.R., Gagnon, C.V.L., Saar, M.O., Seyfried, W.E., 2016. Nanoscale Wyllie, P.J., Huang, W.L., 1976. Carbonation and melting reactions in the system CaO-
constraints on porosity generation and fluid flow during serpentinization. Geology MgO-SiO2-CO2 at mantle pressures with geophysical and petrological applications.
44, 103–106. Contributions to Mineralogy and Petrology 54 (2), 79–107.
Ueda, T., Obata, M., Di Toro, G., Kanagawa, K., Ozawa, K., 2008. Mantle earthquakes frozen Xia, G., 2013. Experimental Studies on Dehydration Embrittlement of Serpentinized Peri-
in mylonitized ultramafic pseudotachylytes of spinel-lherzolite facies. Geology 36, dotite and Effect of Pressure on Creep of Olivine. PhD dissertation. University of Cal-
607–610. ifornia, Riverside, USA.
Uehara, S., 1998. TEM and XRD study of antigorite superstructures. The Canadian Miner- Xu, H., Veblen, D.R., 1996. Interstratification and other reaction microstructures in the
alogist 36 (6), 1595–1605. chlorite-berthierine series. Contributions to Mineralogy and Petrology 124 (3–4),
Uehara, S., Kamata, K., 1994. Antigorite with a large supercell from Saganoseki, Oita Pre- 291–301.
fecture, Japan. The Canadian Mineralogist 32, 93–103. Xu, J., Zhang, D., Fan, D., Zhang, J.S., Hu, Y., Guo, X., ... Zhou, W., 2018. Phase transitions in
Uehara, S., Shirozu, H., 1985. Variations in chemical composition and structural properties orthoenstatite and subduction zone dynamics: effects of water and transition metal
of antigorites. Mineralogical Journal 12, 299–318. ions. Journal of Geophysical Research - Solid Earth 123 (4), 2723–2737.
Ulmer, P., Stalder, R., 2001. The Mg (Fe) SiO3 orthoenstatite-clinoenstatite transitions at Yamamoto, K., Akimoto, S.I., 1977. The system MgO-SiO2-H2O at high pressures and tem-
high pressures and temperatures determined by Raman-spectroscopy on quenched peratures; stability field for hydroxyl-chondrodite, hydroxyl-clinohumite and 10Å-
samples. American Mineralogist 86 (10), 1267–1274. phase. American Journal of Science 277 (3), 288–312.
Ulmer, P., Trommsdorff, V., 1995. Serpentine stability to mantle depths and subduction Yamaoka, K., Fukao, Y., Kumazawa, M., 1986. Spherical shell tectonics: Effects of sphericity
related magmatism. Science 268, 858–861. and inextensibility on the geometry of the descending lithosphere. Reviews of Geo-
Ulmer, P., Trommsdorff, V., 1999. Phase relations of hydrous mantle subducting to 300 physics 24 (1), 27–53.
km. Conference Paper, Geochemical Society, Houston, pp. 259–282. Yamashita, T., Schubnel, A., 2016. Slow slip generated by dehydration reaction coupled
van Keken, P.E., Hacker, B.R., Syracuse, E.M., Abers, G.A., 2011. Subduction factory: 4. with slip-induced dilatancy and thermal pressurization. Journal of Seismology 20
Depth-dependent flux of H2O from subducting slabs worldwide. Journal of Geophys- (4), 1217–1234.
ical Research - Solid Earth 116 (B1). Zhao, D., Matsuzawa, T., Hasegawa, A., 1997. Morphology of the subducting slab boundary
Vidal, O., Goffé, B., Theye, T., 1992. Experimental study of the stability of sudoite and in the northeastern Japan arc. Physics of the Earth and Planetary Interiors 102 (1–2),
magnesiocarpholite and calculation of a new petrogenetic grid for the system FeO- 89–104.
MgO-Al2O3-SiO2-H2O. Journal of Metamorphic Geology 10 (5), 603–614. Zhou, J., Hsu, L.C., 1992. The stability of merwinite in the system CaO-MgO-SiO2-H2O-
Vidal, O., Theye, T., Chopin, C., 1994. Experimental study of chloritoid stability at high CO2 with CO2-poor fluids. Contributions to Mineralogy and Petrology 112 (2–3),
pressure and various fO2 conditions. Contributions to Mineralogy and Petrology 385–392.
118 (3), 256–270.
Viswanathan, K., 1981. The crystal structure of a Mg-rich carpholite. American Mineralo-
gist 66 (9–10), 1080–1085.

You might also like