Earth and Planetary Science Letters 482 (2018) 33–43

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Fast intraslab fluid-flow events linked to pulses of high pore fluid

pressure at the subducted plate interface
Stephan Taetz a,∗ , Timm John b , Michael Bröcker a , Carl Spandler c , Andreas Stracke a
a
Institut für Mineralogie, Universität Münster, Corrensstraße 24, 48149 Münster, Germany
b
Institut für Geologische Wissenschaften, Freie Universität Berlin, Malteserstr. 74-100, 12449 Berlin, Germany
c
Economic Geology Research Centre, James Cook University, Townsville, 4811, Australia

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

Article history: A better understanding of the subduction zone fluid cycle and its chemical–mechanical feedback requires
Received 22 February 2017 in-depth knowledge about how fluids flow within and out of descending slabs. Relicts of fluid-flow
Received in revised form 11 October 2017 systems in exhumed rocks of fossil subduction zones allow for identification of the general relationships
Accepted 23 October 2017
between dehydration reactions, fluid pathway formation, the dimensions and timescales of distinct
Available online 9 November 2017
Editor: M. Bickle
fluid flow events; all of which are required for quantitative models for fluid-induced subduction zone
processes. Two types of garnet–quartz–phengite veins can be distinguished in an eclogite-facies mélange
Keywords: block from the Pouébo Eclogite Mélange, New Caledonia. These veins record synmetamorphic internal
Li chronometry fluid release by mineral breakdown reactions (type I veins), and infiltration of an external fluid (type
subduction zones II veins) with the associated formation of a reaction selvage. The dehydration and fluid migration
fluid flow system documented by the type I veins likely occurred on a timescale of 105 –106 years, based on average
diffusion modelling subduction rates and metamorphic conditions required for mineral dehydration and fluid flow. The
plate interface
timeframe of fluid–rock interaction between the external fluid and the wall-rock of the type II veins is
seismic slip
quantified using a continuous bulk-rock Li-diffusion profile perpendicular to a vein and its metasomatic
selvage. Differences in Li concentration between the internal and external fluid reservoirs resulted in a
distinct diffusion profile (decreasing Li concentration and increasing δ 7 Li) as the reaction front propagated
into the host rock. Li-chronometric constraints indicate that the timescales of fluid–rock interaction
associated with type II vein formation are on the order of 1 to 4 months (0.150+ 0.14
−0.08 years). The short-
lived, pulse-like character of this process is consistent with the notion that fluid flow caused by oceanic
crust dehydration at the blueschist-to-eclogite transition contributes to or even dominates episodic pore
fluid pressure increases at the plate interface, which in turn, may trigger slip events reported from many
subduction zones.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction John et al., 2012; Peacock, 1993). Expelled fluids migrate via chan-
nelized flow through the subducted slab and mantle wedge (e.g.
Element cycling at subduction zones controls the global geo- Ague, 2011; Angiboust et al., 2014; John et al., 2008, 2012;
chemical budget of many elements and is a crucial process for Plümper et al., 2017; Scambelluri and Philippot, 2001; Taetz et
the evolution of Earth’s mantle–crust–atmosphere system. As sub- al., 2016). Field observations document that these fluids may react
ducting plates heat up during their passage into the deeper with their wall rocks, thereby mobilising and transporting solu-
mantle, hydrous minerals within the different rocks of the sub- ble elements (e.g. LILE, LREE) (Bebout, 2007; Gao et al., 2007;
ducted slab become progressively unstable, eventually break down, Herms et al., 2012; John et al., 2004, 2012), and thereby induce
and release water. The most prominent dehydration reactions the formation of melts in the mantle wedge with distinct chemical
are related to the transformation of hydrous amphiboles and signatures (e.g. John et al., 2004; Vrijmoed et al., 2013).
chlorite to anhydrous omphacite and garnet during eclogitisa- Various seismic to nonseismic slip phenomena (e.g. megath-
tion of the subducting oceanic crust (e.g. Gao and Klemd, 2001; rust earthquakes, episodic tremor and slip events, silent earth-
quakes, creep) occurring at the plate interface have been linked
to high pore fluid pressure or even sudden increase of the lo-
* Corresponding author. cal pore fluid pressure regime (e.g. Audet et al., 2009; Beroza
E-mail address: stephan.taetz@uni-muenster.de (S. Taetz). and Ide, 2011; Fagereng and Diener, 2011; Moreno et al., 2014;

https://doi.org/10.1016/j.epsl.2017.10.044
0012-821X/© 2017 Elsevier B.V. All rights reserved.
34 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43

Schurr et al., 2014). The required rise of fluid pressure is proba-

bly caused by dehydration reactions within the descending slab
(e.g. Dragovic et al., 2015; Fagereng and Diener, 2011; Moreno
et al., 2014; Schurr et al., 2014). Several studies suggest that
the periodicity of some of these events is controlled by cyclic
changes in pore fluid pressure, with slip occurring when a suf-
ficient amount of fluid had been accumulated (e.g. Frank et al.,
2015; Gomberg et al., 2010). The increased permeability induced
by the slip event reduces the elevated pore pressure (Skarbek and
Rempel, 2016), but owing to continuing fluid production within
the descending slab, the system recovers (Plümper et al., 2017;
Taetz et al., 2016) until again a critical fluid pressure is reached.
Key towards a better understanding of the subduction zone
fluid cycle and its chemical and mechanical feedback along the
plate interface is obtaining in-depth understanding of how flu-
ids flow within and out of the descending slab. Developing re-
liable quantitative models of fluid-mediated mass transfer in a
subducting slab requires quantifying the general relationships be-
tween mineral reactions, fluid flow and element mobilization dur-
ing fluid–rock interaction. (e.g. Angiboust et al., 2014; Bebout,
2007; Gao et al., 2007; John et al., 2004; Spandler et al., 2011;
Taetz et al., 2016). The dimensions and timescales of distinct fluid
flow events and the physico-chemical processes triggered by fluid
flow are crucial parameters of such models. However, direct in-
vestigation of dehydration-related fluid flow in subducting slabs
is impossible and, owing to the dm- to m-sized dimensions of
the flow structures (e.g. Gao et al., 2007; Gao and Klemd, 2001;
Herms et al., 2012; John et al., 2008, 2012; Spandler and Her-
mann, 2006), are not detectable in-situ by geophysical methods.
Consequently, samples exhumed long after (i.e. millions of years)
fluid–rock interaction occurred provide the only direct observa-
tional evidence. But determining the duration of fast processes
that have happened in the geologic past is a difficult task. Isotopic
chronometers commonly used to date the high-pressure rocks (e.g.
Lu–Hf, Sm–Nd, Ar–Ar, Rb–Sr for absolute ages) are too imprecise Fig. 1. a) Simplified geological map of New Caledonia (after Aitchison et al., 1995);
to resolve short-duration events of several thousand years or less b) close-up of the Pam Peninsula (after Clarke et al., 1997) with mineral distribution
(e.g. Ague and Baxter, 2007; Bons, 2001; Camacho et al., 2005; patterns. The position of the sampling site is indicated by the red square. (Gln =
John et al., 2012). Determining the relative timescales of diffusive glaucophane; Lws = lawsonite; Ab = albite; Ep = epidote; Alm = almandine).

processes that affect fast diffusing elements, however, is one pos-

sible means to quantify the timing of such processes (e.g. John et The high-pressure/low-temperature (HP/LT) metamorphic rocks
al., 2012; Watson and Baxter, 2007). In fluid flow related struc- of the Pouébo Eclogite Mélange in New Caledonia provide an ex-
tures diffusion may occur on the cm- to outcrop-scale between cellent opportunity to study the fluid flux in a subduction zone
pristine wall rock and the fluid conduit, as documented in the for- setting. The focus of this study is on the fluid dynamics recorded
mation of reaction selvages (e.g. Ague, 2003; Beinlich et al., 2010; by two types of garnet–quartz–phengite veins that cross-cut eclog-
Penniston-Dorland and Ferry, 2008). In such reaction zones, min- ite facies mélange blocks from this occurrence (Spandler and Her-
eral reaction at the fluid–rock interface is considered to proceed mann, 2006; Taetz et al., 2016). These veins document both inter-
faster than the rate of fluid transport (e.g. Jamtveit et al., 2000; nal fluid release by mineral breakdown reactions and infiltration of
John et al., 2012). Consequently, in these fluid dominated systems, external fluids (Spandler and Hermann, 2006; Taetz et al., 2016).
element transport within the reaction selvage is primarily con- Taetz et al. (2016) investigated the main petrological, geochemical
trolled by bulk diffusion (e.g. Watson and Baxter, 2007). Lithium- and geochronological characteristics of the rocks cut by the veins
chronometry based on bulk-diffusion modelling is a tool that al- and developed a conceptual model for vein formation. Here, we
lows the determination of the timescales of such fast and rela- present results of Li chronometry based on bulk diffusion mod-
tively cool (<700 ◦ C) geological processes (e.g. John et al., 2012; elling that provide quantitative constraints on the duration of a
Teng et al., 2006; Watson and Baxter, 2007). Lithium is a trace distinct vein-forming fluid flow event.
element in both fluid and solid phases, and is not strongly par-
titioned into any metamorphic phases. Nevertheless, Li can un- 2. Geological setting
dergo significant kinetic fractionation between the two stable iso-
topes 6 Li and 7 Li, resulting in faster diffusion of 6 Li compared The metamorphic HP/LT ophiolite complex located in the NE
to 7 Li (e.g. Richter et al., 2006). This process leads to isotopic part of the main island of New Caledonia (Fig. 1) is character-
variations that are easily resolvable using modern analytical tech- ized by a regular pattern of mineral isograds. The prograde re-
niques (e.g. John et al., 2012; Penniston-Dorland et al., 2010; gional P –T gradient ranges from 0.3 GPa, 250 ◦ C in the SW to ca.
Teng et al., 2006). Therefore, diffusion modelling based on Li has 1.9–2.4 GPa, 500–650 ◦ C in the NE (Carson et al., 1999; Clarke et
the advantage that both concentration profiles and isotope ratio al., 1997; e.g. Vitale Brovarone and Agard, 2013). The highest grade
profiles can be used, which provides additional measures of con- eclogite-facies rocks occur in the Pouébo Eclogite Mélange (PEM)
sistency and robustness to the models. (e.g. Carson et al., 1999; Clarke et al., 1997; Spandler et al., 2005),
 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43 35

which consists of blocks and tectonic slices that are enclosed in a

meta-ultramafic (serpentinite and talc-schist) or metasedimentary
matrix (e.g. Cluzel et al., 2012; Spandler et al., 2005). The mafic
rocks of the PEM comprise garnet amphibolites and eclogites orig-
inating from diverse protoliths from a late Cretaceous to Paleocene
back-arc basin. HP/LT rocks indicate peak-metamorphic conditions
of 1.9–2.4 GPa and 500–650 ◦ C, reached at ca. 44–38 Ma (Baldwin
et al., 2007; Pirard and Spandler, 2017; Spandler et al., 2005;
Taetz et al., 2016), followed by rapid exhumation and cooling to
about 125 ◦ C at ca. 34 Ma (Baldwin et al., 2007). For details on the
tectonic and metamorphic history see Cluzel et al. (2012), Spandler
et al. (2005), Baldwin et al. (2007), Vitale Brovarone and Agard
(2013) and Taetz et al. (2016).

3. Samples and vein system formation

All samples used for Li-isotope analysis were collected from

a single metabasic block of the PEM that records the blueschist-
eclogite transition. The sample is located ca. 2.5 km east of the
mouth of Amoss Creek (S 20 16.748, E 164 26.829) and has been
described in detail by Spandler and Hermann (2006) and Taetz et
al. (2016) (Fig. 1). The studied block (ca. 2 × 4 m in size) shows two
types of well-preserved HP veins in an amphibole-rich eclogitic
host rock (Spandler and Hermann, 2006; Taetz et al., 2016). The
mineral assemblage of the host rock includes garnet porphyrob-
lasts and a matrix that mainly consists of glaucophane, barroisitic
hornblende, omphacite and epidote, as well as phengite and ru-
tile as minor phases. Apatite, zircon and titanite are present as
accessory mineral phases. Two types of garnet–phengite–quartz
veins cross-cut the host rock. Type I veins have no obvious re-
action zone and are up to 1 cm thick and can reach 0.5 m in
length. These veins are interpreted to form from fluids released
by metamorphic dehydration of chlorite, amphibole, and epidote,
which then filled dynamically-connecting pore spaces, before be-
ing channelized into veins of variable size (Plümper et al., 2017;
Taetz et al., 2016). Type II veins occur more rarely and are asso-
ciated with a distinct omphacite-rich metasomatic selvage (Figs. 2,
3). The studied type II vein is up to 5 cm thick and can be traced
over ca. 1 m (Figs. 2, 3c). Type II veins are attributed to fluid
influx from an external source (Taetz et al., 2016). The metaso-
matic selvage is formed by fluid–rock interaction along a reaction
front due to chemical disequilibrium between internal (from in situ
fluid production) and external (from fluid production elsewhere in
the same slab) fluids (Taetz et al., 2016). Vein and selvage forma-
tion occurred shortly prior to peak metamorphic conditions, which
are constrained to 540 ± 15 ◦ C using Zr-in-rutile thermometry
(host rock, selvage, type II vein) and 500–560 ◦ C and 1.9–2.2 GPa
using Grt–Cpx–Ph thermobarometry (host rock and type II vein)
(Taetz et al., 2016). These values are in agreement with P –T es-
timates for other rocks from the PEM (e.g. Carson et al., 1999;
Clarke et al., 1997; Vitale Brovarone and Agard, 2013). Internal Rb–
Sr isochrons for host rock, selvage and type II vein all yielded an
age of ca. 38 Ma, which dates vein formation and associated fluid–
rock interaction (Taetz et al., 2016).
Fig. 4 schematically shows the evolution of the vein system
from small scale initial fluid release by individual mineral grains
towards channelized flow on the outcrop-scale. The initial break-
down of fluid-bearing phases (chlorite, epidote, amphibole) oc-
curred at around 400 ◦ C and 1.6 GPa and continued under increas-
Fig. 2. Field images. (a) Outcrop overview displaying both vein types. A continuous
ing prograde metamorphic conditions towards peak conditions of sampling profile is shown by white rectangles. Numbers denote distance from the
ca. 540 ◦ C and 2.2 GPa (Taetz et al., 2016). The release of a free type II vein in cm. Individual samples used for Li chronometry are shown by orange
fluid phase is associated with the onset of chlorite breakdown at rectangles. Dashed lines outline the transition zone between selvage and host rock.
ca. 460–480 ◦ C and 1.9 GPa (Figs. 4a, b; 5) which liberates H2 O. (b) Close-up of a type II vein. (c) Close-up of a network of type I veins.

Due to the lower molar volume (i.e. greater density) of the newly
36 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43

Fig. 3. (a) Thin section photographs of the profile used for Li isotope analysis illustrating the type II vein with a sharp contact to the selvage and the gradual transition from
selvage to host rock which is partially cross-cut by type I veins. Mineral abbreviations after Whitney and Evans (2010). (b) Mineral modal abundance along the profile (Taetz
et al., 2016). (c, d) Whole rock Li concentration (blue) and δ 7 Li (orange). Symbol size is in accordance with error bars. Symbol dimensions on the x-axis indicate the sample
size along which the data was averaged. Error bars on the y-axis indicate an assumed uncertainty of ±1 μg/g for the Li concentration and the uncertainty on measured
δ 7 Li respectively. Coloured curves show the best fitting modelling results for different amounts of reaction-induced porosity φR /φ0 . Grey curves show the best fit for deviant
values at 5 and 10 cm distance from the vein with φR /φ0 = 5. (hst = host-selvage transition; v = vein).

formed garnet-omphacite assemblage these processes locally in- 4. Methods

creased the porosity in the host rock by ca. 2.3% and allowed
accumulation of the liberated fluid phase (Figs. 4a; 5; see also 4.1. Analytical techniques
Plümper et al., 2017). With further breakdown of epidote and am-
A continuous profile (ca. 35 cm) was sampled perpendicular
phibole and the associated higher density of newly formed solid to the main type II vein including material from the vein, reac-
phases (garnet, omphacite) the porosity increased, eventually lead- tion halo and host rock. Additional drill cores were taken at the
ing to an interconnected network that allowed fluid propagation maximum distance from the type II vein (44 and 53 cm away, re-
through the host rock in vein structures (Fig. 4c) that can develop, spectively) and the marginal part of the outcrop to obtain material
for example, by hydrofracturing or porosity waves (e.g. Bons, 2001; that most likely remained unaffected by fluid infiltration-induced
alteration.
Connolly and Podladchikov, 2007; Plümper et al., 2017). It is likely
All analytical work, except for the porosity measurements, was
that the studied rock volume was infiltrated by such fluids along carried out at the Institut für Mineralogie, Universität Münster,
a structurally weak zone, for example a pre-existing dehydration Germany. Isotope analysis were conducted on whole rock pow-
vein (Taetz et al., 2016) (Fig. 4d). ders and mineral separates prepared from 14 rock slabs (each ca.
 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43 37

Fig. 4. Schematic evolution of the vein system from (a) initial mineral breakdown, (b) small scale fluid flow in vein-like porosity, (c) dehydration vein formation and (d)
infiltration of an external fluid by a short-lived fluid pulse event. The external fluid reacts with the wall rock and the internal fluid. Advecting fluids can eventually reach
the plate interface shear zone, where they increase fluid pressure by adding to the already fluid filled porosity. Elevated fluid pressure at the plate interface may induce slip
events indicated by red stars. Multiple fluid pulses of varying intensity (p 1 to p 4 ) travel along the plate interface and flatten out with increasing distance.

4.5 × 2.5 × 1 cm with a weight of ca. 40 g) and two drill cores (ca. ing ca. 50–200 ng Li were taken up in 0.67 M HNO3 /methanol 30%
100 g each) (Figs. 2; 3). Thin-sections of the individual slabs allow v/v for ion exchange chromatography. For geological reference ma-
a correlation of petrographic, geochemical and isotopic character- terials (BHVO-2, JR-2, JB-2) ca. 100 mg of powder was digested and
istics. aliquots equivalent of 12.5 mg rock were used for column chem-
All sample material was cleaned with deionized water, crushed istry.
in a steel mortar and ground in an agate ring mill. Whole rock Lithium was separated by ion exchange chromatography using
powders (ca. 50–100 mg per sample) were digested in HF–HNO3 the method described in Magna et al. (2004, 2006). For the main
(5:1) in a tabletop setup. After evaporation to dryness the residue separation of Li from other cations, columns filled with 2.1 ml of
was re-dissolved in 6 M HCl and homogenized on a hot plate BioRad® AG 50W-X8 (200–400 mesh) cation exchange resin were
overnight followed by a second evaporation step. Aliquots contain- used. The samples were loaded in 0.67 M HNO3 /methanol 30%
38 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43

Lab Petroleum Services, Houston, Texas, USA. Sample plugs with

a diameter of 2.54 cm were loaded into hydrostatic core holders.
The baseline pore volume for the sample is determined by calcu-
lating the bulk volume from callipered sample dimensions minus
the grain volume measured in a helium porosimeter (using Boyle’s
Law). Multiple in-situ measurements were conducted under net
confined stress (NCS) from 0 to 0.062 GPa. The porosity is re-
calculated based on apparent pore volume reduction at each NCS.
The technique typically yields values within ±0.2% uncertainty.
The resulting porosity has been extrapolated to the relevant pres-
sure of 2.0 GPa using an exponential function, similar to Katayama
et al. (2012) (supplementary Fig. S1).

4.2. Lithium partition coefficients

The underlying rationale for a plausible application of Li

chronometry is the selection of realistic partition coefficients oper-
ating at equilibrium conditions. Partition coefficients of Li between
min/fluid
individual minerals and fluid (K d(Li) ) are difficult to deter-
mine and experimental results at 700–1100 ◦ C cover a wide range
min/fluid
(ca. 0.05–2.1; e.g. Tomascak et al., 2016). Additionally, K d(Li) is
strongly temperature sensitive, and no experimental data is cur-
rently available for the relevant temperatures. Therefore, we have
= 3.59 for 540 ◦ C from the experimentally
cpx/fluid
Fig. 5. P –T pseudosection (Perple_X) in the system TNMnCaKFMASCHO for the av- extrapolated K d(Li)
erage bulk composition of the reference samples 40-A, 41-A, 18-II. Details on the derived relationship of Caciagli et al. (2011):
calculation are given in Taetz et al. (2016). Stable phases in numbered fields: (1)
cpx/fluid
Chl, Ep, Amp, Ph, Gt, Omp, Rbk, Qz, Rt; (2) Amp, Ph, Gt, Omp, Qz, Rt, H2 O; (3) Ph,
LnD Li = −7.3(±0.5) + 7.0(±0.7) × 1000/ T . (2)
Gt, Omp, Qz, Rt, H2 O. Field (4) resembles subjacent fields plus carphorite. Grey lines
show the combined density of solid phases, which increases with higher P –T con- min/fluid
ditions due to breakdown of fluid bearing phases and formation of denser eclogite. To achieve a consistent dataset of K d(Li) for all relevant min-
Blue dashed lines show the combined volume fraction of fluid (OH) bearing phases. eral phases (Grt, Brs, Ep, Gln, Omp, Ph, Qz, Rt; mineral abbrevi-
P –T path after Taetz et al. (2016). (For interpretation of the references to colour in ations after Whitney and Evans, 2010) this value was combined
this figure legend, the reader is referred to the web version of this article.) min/cpx
with K d(Li) data obtained from a series of natural rock sam-
ples according to the model of Marschall et al. (2007a) (Table 1).
v/v; cations were eluted in 1 M HNO3 /methanol 80% v/v. A sec- WR/fluid
ond cation exchange column (0.6 ml of BioRad® AG 50W-X12, Clinozoisite data was used for epidote. K d(Li) has been subse-
200–400 mesh) was used for removal of Cr, Al and small amounts quently derived from the sum of this set of partition coefficients
of Na with 0.5 M HNO3 . These cations may remain after the multiplied with the mineral modal abundance along the sampling
first step owing to possible resin degradation, and the treatment profile (Marschall et al., 2007a; Taetz et al., 2016).
with HNO3 /methanol mixtures (Jeffcoate et al., 2004; Magna et al.,
2006). All acids and methanol were cleaned by sub-boiling distil- 5. Results
lation (three times).
Lithium isotopic analysis were conducted on a Thermo Scientific Based on mineralogical, textural and geochemical characteris-
Neptune Plus MC-ICP-MS. Samples were introduced with a CETAC tics, the studied rock profile is divided into four parts: vein, sel-
Aridus II de-solvating sample introduction system equipped with vage, selvage-host rock transition zone and host rock (Taetz et al.,
a teflon spray chamber and a Savillex Nebulizer (certified 48 μl 2016) (Fig. 3a). The Li concentrations and δ 7 Li values of samples
uptake rate). All samples were measured 3 times in a standard- representing the different segments of the profile are summarized
sample bracketing setup with L-SVEC as standard Li solution. δ 7 Li in Table 1.
values (h) of the samples were calculated relative to the average WR/fluid
The low K d(Li) of the vein (0.22; Table 1; Fig. 3b) reflects the
of the bracketing L-SVEC measurements (Eq. (1)). mineral mode with only phengite as a significant Li carrier. The
  WR/fluid
(7 Li/6 Li)sample omphacite-rich selvage shows the highest K d(Li) value (= 3.10)
7
δ Li (h) = 7 6 − 1 × 1000 (1) that decreases towards the host rock where it reaches an average
( Li/ Li)L-SVEC
value of ca. 2.3. Based on this set of partition coefficients, the Li
To evaluate accuracy and precision of the method, various rock concentration in the transport vein fluid is estimated at ca. 14 μg/g,
standards (BHVO-2, JB-2, JR-2) and Li standard solutions (IRMM- whereas the Li concentration of fluid in pore spaces of the wall
016) have been measured during each analytical session. Results rock is estimated at ca. 6 μg/g (Table 1).
and additional details on Li isotope analysis are given in the sup- The type II vein is characterized by low Li concentrations of
plementary material (Text and Table S1). Lithium concentrations 2–3 μg/g with δ 7 Li = −1.63 ± 0.07h (Fig. 3b; Table 1). At the sharp
have been determined by relative beam intensity measurements contact to the omphacite-rich metasomatic selvage zone a signifi-
against pre-diluted standard solutions with similar concentrations cant compositional step is developed that is characterized by the
(within 5% for 50 ppb Li solutions). Analytical uncertainty during highest Li concentrations (up to 45 μg/g) and corresponding low-
analysis was <0.2 μg/g. An uncertainty of ±1 μg/g was assigned est δ 7 Li (−2.31 ± 0.24h) of the studied profile. With the selvage
for modelling to accommodate for any potential Li loss during sam- gradually adjusting to the host-rock mineralogy over a distance of
ple digestion. ca. 3–5 cm, the Li concentration decreases towards ca. 16–18 μg/g
The porosity of the most pristine host rock sample 40A, a drill in the host rock where it remains constant. The opposite trend is
core taken 44 cm from the type-II vein, was measured by Core shown by the δ 7 Li values that increase towards 0.83 ± 0.04h from
 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43 39

Table 1
Mineral modal abundances* , Li concentrations, Li diffusion coefficients and δ 7 Li values of individual samples along the profile.

Sample 41-A 40-A 18-II 17-II 16-II 16-A 15-C 15-II 15-A 15-I 14-C 14-B 14-A 14-IV(2)
type host host host host host host host host host selvage selvage selvage selvage vein
min/fluid dist. to −53 −44 −33 −26 −18 −11.5 −10 −5 −3.5 −2.8 −1.5 −0.5 0
K d(Li)
vein [cm]
0.108 Grt 23.2 23.9 n.d. 22.7 20.9 n.d. 20.7 22.8 28.9 20.2 11.6 6.5 5.3 62.9
4.313 Brs 9.5 9.3 n.d. 7.3 7.9 n.d. 9.5 10.2 11.1 7.1 3.8 1.2 0.1 0.0
0.040 Ep 12.1 11.9 n.d. 11.2 13.1 n.d. 13.1 13.1 12.4 9.7 7.2 6.7 6.0 0.0
4.313 Gln 16.5 22.0 n.d. 15.4 11.6 n.d. 12.5 12.6 14.7 12.4 6.5 2.3 0.6 0.0
3.594 Omp 28.6 25.8 n.d. 33.6 39.6 n.d. 38.7 34.7 24.8 44.8 67.2 80.5 85.3 0.0
1.186 Ph 1.0 0.1 n.d. 0.9 0.8 n.d. 0.0 0.4 0.1 0.0 0.0 0.0 0.0 13.2
0.004 Qz 6.0 3.1 n.d. 5.8 3.1 n.d. 2.4 3.3 5.1 3.0 1.6 1.1 0.8 21.8
0.003 Rt 3.1 4.1 n.d. 3.1 3.0 n.d. 3.1 3.0 3.0 2.9 2.1 1.7 1.8 2.1
total 100.0 100.1 100.0 100.0 100.0 100.1 100.0 100.1 100.0 100.0 99.9 100.0
WR/fluid
K d(Li) 2.19 2.31 n.d. 2.23 2.30 n.d. 2.37 2.26 2.04 2.48 2.87 3.05 3.10 0.22

LiOmp μg/g 49.6 48.6 48.7 50.4 49.7 n.d. n.d. 51.1 n.d. n.d. n.d. 47.2 n.d. 49.9
LiWR μg/g 17.6 17.7 18.6 18.1 18.4 18.8 18.9 16.0 15.7 24.8 30.4 37.8 39.6 2.4
Lifluid μg/g 8.2 6.9 n.d. 7.2 7.0 n.d. 6.8 6.2 7.8 8.9 9.7 11.1 11.9 10.8

δ 7 Li h n.d. 0.83 0.89 n.d. n.d. 0.79 0.80 1.24 1.32 −0.29 −0.84 −1.95 −2.31 −1.63
±0.08 ±0.12 ±0.14 ±0.12 ±0.17 ±0.07 ±0.15 ±0.06 ±0.12 ±0.24 ±0.07
*
Modal abundances are based on Rietveld analysis. An uncertainty of ±5% is assumed. Higher uncertainties are expected for Ph due to possible method specific texture
effects (Taetz et al., 2016); n.d. = not determined. Mineral abbreviations after Whitney and Evans (2010).

selvage to host rock. Two host rock samples collected at ca. 5 and spaces of the host rock (ca. 6 μg/g) drives Li diffusion from the
10 cm distance to the vein show slightly lower Li concentrations vein into the host rock creating the observed compositional gra-
(13–14 μg/g) and elevated δ 7 Li values (+1.28 ± 0.04h) compared dients (Fig. 3). Lithium is then incorporated into individual phases
to the samples collected at greater distance to the vein. This de- of the mineral assemblage by dissolution–precipitation reactions
viation from the general trend can be attributed to the influence (Putnis and John, 2010), as recorded by elevated Li concentrations
of a larger type I vein that could not be avoided during sample in the selvage relative to the host rock (Fig. 3).
preparation (Fig. 2a; highlighted by white arrow).
6.2. Lithium as a monitor of fluid–rock interaction
6. Discussion
The Li concentration in the host rock (ca. 16–18 μg/g) is to a
6.1. Mechanisms of element exchange between type II vein and host large extent internally buffered (Taetz et al., 2016), in good agree-
rock ment with the notion that the studied eclogite represents former
seawater-altered oceanic crust (Spandler and Hermann, 2006). Al-
Textural and mineral chemical zoning patterns suggest that an tered oceanic crust can incorporate up to ca. 35 μg/g Li (Chan et
external fluid infiltrated the host rock along a pre-existing weak al., 2002), whereas fresh oceanic crust is characterized by lower Li
zone (e.g. a former type I vein) and that fluid–rock interaction af- concentrations of ca. 4–9 μg/g with an average of 6.5 μg/g (Gale
fected the host rock in the immediate vicinity of the newly formed et al., 2013). The positive δ 7 Li value of the host rock (ca. +1h)
type II vein (Taetz et al., 2016). While the external fluid, possibly is also consistent with that of altered oceanic crust, although the
buoyancy driven, may flow in open fractures along pressure gradi- latter exhibits a considerable range in δ 7 Li, from ca. −3 to +14h
ents (e.g. Connolly and Podladchikov, 2007; Oliver and Bons, 2001), (e.g., Chan et al., 2002). Nevertheless, δ 7 Li values in the selvage are
the fluid–rock interaction is caused by chemical gradients between distinctly lower (up to −2.31h) and can clearly be distinguished
the external fluid and the internal fluid that is stored in the dy- from the host rock.
namically forming porosity of the host rock (Taetz et al., 2016). Diffusion occurs in two principal regimes in fluid flow-related
Thus, the reaction front propagation is driven by the related dif- structures: (1) on the intragrain scale between different growth
fusive element transport leading to the formation of the reaction zones of single crystals, which result from continuous growth on
selvage (Figs. 2; 3a). In such reaction zones the mineral reaction pre-existing grains or from partial recrystallization during fluid-
along the fluid–solid interface is faster than the transport of the mediated dissolution–precipitation processes (e.g. Ague and Baxter,
components involved in the reactions (e.g. Jamtveit et al., 2000; 2007; Jonas et al., 2014), and; (2) on the cm- to outcrop-scale be-
John et al., 2012; Putnis and John, 2010). The advective compo- tween the pristine wall rock and the fluid conduit, as documented
nent of the external fluid is focused into its fracture-determined by the formation of reaction selvages (e.g. John et al., 2012;
flow path, and so is insignificant compared to diffusion in terms of Jonas et al., 2014; Taetz et al., 2016).
fluid–rock interaction within the immediate wall-rock (Ague, 2011; The Li concentration and isotope composition can therefore be
Connolly and Podladchikov, 2007; John et al., 2012; Oliver and used to decipher metasomatic processes in subduction zones (John
Bons, 2001). Consequently, for such fluid-dominated systems, el- et al., 2012; Penniston-Dorland et al., 2010). The potential changes
ement transport within the reaction selvage is mainly controlled in the Li isotope ratios during dehydration processes from min-
by bulk diffusion (e.g. John et al., 2012; Penniston-Dorland et eral breakdown to channelized fluid flow are limited (Marschall
al., 2010; Watson and Baxter, 2007). For systems such as the et al., 2007b). Experimental studies (e.g. Wunder et al., 2006)
one described here, kinetic effects on the duration of the diffu- show that Li isotope fractionation between mineral (Cpx, Amp)
sive processes can be neglected, because these proceed signifi- and fluid at 550 ◦ C is about −4h, whereas mineral–mineral iso-
cantly faster than the fluid–rock interaction (Jamtveit et al., 2000; tope fractionation, for example between amphibole and omphacite,
John et al., 2012). The inferred difference in Li concentration be- is negligible. Additionally, as host rock and selvage minerals both
tween the vein-forming fluid (ca. 14 μg/g) and the fluid-filled pore equilibrated with a high-pressure fluid, the fractionation effect
40 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43

should be equal along the entire profile. Therefore, δ 7 Li variations time of Ω ≈ 2.60 × 10−4 . The model was additionally run with its
along the sampling profile are most likely caused by fractionation parameters set to reflect the lower and upper bounds of the an-
during diffusion (e.g. John et al., 2012; Marschall et al., 2007b; alytical uncertainties on δ 7 Li and Li concentration. The resulting
Penniston-Dorland et al., 2010). uncertainty in Ω is less than 5% in both cases. To account for the
additional uncertainty introduced by the choice of β , we assume a
6.3. Lithium chronometry total uncertainty of ±11% for the resulting Ω . This value for Ω can
be inserted into Eq. (3) (John et al., 2012) to estimate the duration
Fluids released from hydrous phases by breakdown reactions of fluid–rock interaction in years:
in the subduction zone may facilitate seismic and non-seismic
WR/fluid
slip events along the plate interface by creating a significant in- Ω ρs K d(Li) L2
crease of pore fluid pressure (e.g. Audet et al., 2009; Beroza τtotal, years ≈ × × × (3)
3 × 10 7 ρf ∅WR D Li
and Ide, 2011; Fagereng and Diener, 2011; Moreno et al., 2014;
Schurr et al., 2014). If correct, fluid release events should corre- where ρs is the average density of the host rock which is set to
spond in space and time with slip events, but such relationships 3.24 g/cm2 and ρf is the density of the fluid calculated from the
are poorly constrained. Direct investigations of intraslab fluid flow IAPWS-95 pure water equation of state (Wagner and Pruß, 2002)
events are currently not possible, but vein systems in exhumed HP set to 1.21 g/cm3 . The parameter L is the length of the profile
metamorphic rocks can provide insights into fluid flow processes (44 cm). The uncertainties on these values can be neglected as
operating at depth. these are considered to be very accurate and only have a very
In the present case, formation of the type I veins is linked to minimal effect on the resulting duration. The calculated duration is
internal dehydration processes, whereas formation of the type II most sensitive to changes in the extrapolated values for φ WR , D Li
veins and associated selvages is related to infiltration of external WR/fluid
and K d(Li) . φ WR is the diffusion relevant background porosity of
fluids (Taetz et al., 2016). Both Li concentration and δ 7 Li record a
the host rock. The value of 4.87 × 10−3 at 2.0 GPa was extrapolated
distinct diffusion pattern across the studied profile (Fig. 3), which
from measurements on the reference sample 40A (Table S2) and
provides an ideal opportunity to determine the duration of the
lies in the same range as calculated porosity data from greenschist-
fluid–rock interaction by Li chronometry, using a quantitative dif-
facies rocks by Skelton (2011). We estimate an uncertainty of
fusion model developed previously (see John et al., 2012 for a
±10% resulting from the extrapolation. D Li (= 1.4 × 10−8 m2 s−1
detailed description). This model estimates the non-dimensional
at 2.0 GPa and 540 ◦ C) is the diffusion coefficient of Li in aqueous
time Ω by minimum misfit between the measured and the mod-
fluid calculated by extrapolation of experimentally determined val-
elled data. The non-dimensional time can then be converted into
ues for 0.1–0.5 GPa and 500–600 ◦ C (Oelkers and Helgeson, 1988)
dimensional time, as outlined below.
via an Arrhenius equation. The uncertainty on D Li is estimated
The length of the studied Li isotope profile (44 cm), the Li con-
to be about ±14% based on the 5–10% estimate for the data by
centrations and δ 7 Li values of the most distant host rock sample
Oelkers and Helgeson (1988) with an additional 10% for the data
and the type II vein have been used as the starting points for WR/fluid
modelling in non-dimensional time. Additional parameters are the extrapolation to the relevant P –T conditions. K d(Li) is the Li
WR/fluid
reaction-induced variation in K d(Li) between the maximum val- distribution coefficient in the host rock at 540 ◦ C and 2.0 GPa. The
WR/fluid
uncertainty on K d(Li) is the least constrained. It is based on an
ues in selvage and the average host rock (−45%), which define the
min/cpx
parameter Kd = −0.45, the reaction-induced change of the poros- internally consistent dataset of K d(Li) by Marschall et al. (2007a)
ity φR /φ0 (with φR = transient porosity during reaction and φ0 = and their analytical uncertainty as well as the modal abundance
starting background porosity), and β , which describes how much estimate (Taetz et al., 2016) that together result in an overall un-
faster 6 Li diffuses than 7 Li by kinetic fractionation. The parameter certainty of ±10%. A larger uncertainty is, however, introduced by
β is set to 0.018 according to the best fit of the measured δ 7 Li pro- the combination of this dataset with the extrapolated values for
file, which is in accordance with experimentally determined values cpx/fluid
K d(Li) based on an experimentally derived relationship (Eq. (2);
for aqueous fluids (β = 0.015) by Richter et al. (2006). Note that cpx/fluid
Caciagli et al., 2011). K d(Li) has been calculated within the up-
the model is rather insensitive to changes in β in terms of the re-
sulting non-dimensional time (Ω ). A change in β by 50% (±0.009) per and lower limits of Eq. (2) (Caciagli et al., 2011) which cor-
WR/fluid
results in a shift of Ω by <10%. responds to 30% uncertainty and hence the resulting K d(Li) can
Fig. 3 (c, d – coloured curves) shows the resulting diffusion pro- be reasonably well constrained with a 32% uncertainty. All values
files for different amounts of reaction-induced porosity (φR /φ0 = inserted into Eq. (3) yield the total duration of the fluid rock inter-
1–100). The modelled diffusion profiles are generally in accordance action as follows, with the best fit with φR /φ0 = 5:
with the measured values. Differences originating from varying the
2.60 × 10−4 3.24 2.28 0.442
reaction-induced porosity between 1 and 1000 are small, due to τtotal, years ≈ × × ×
the short length of the diffusion profile. However, only the mod- 3 × 107 1.21 4.87 × 10−3 1.4 × 10−8
elled curve for φR /φ0 = 5 plots within the error bars of all mea- ≈ 0.150+ 0.14
−0.08 yrs (4)
sured data points (δ 7 Li and Li concentration), with the best fit (red
curve) representing a non-dimensional time of Ω ≈ 2.60 × 10−4 . The best estimate for the maximum and minimum duration of
To validate our modelling of the measured data, a X 2 parameter fluid–rock interaction of 1 to 4 months (0.150+ 0.14
−0.08 years) results
has been calculated. The modelled diffusion profile (Kd = −0.45, from the modification of all values by their uncertainties to maxi-
φR /φ0 = 5), applied to the analytical results for Li at the given mum and minimum values, which shows the robustness of the Li
spatial distribution results in reduced X 2 = 0.9. Thus, the model chronometry method to determine the duration of fluid rock inter-
reproduces the analytical data. action processes.
The deviant values at 5 and 10 cm distance in the profile are
related to an overlapping type I vein and could therefore not be re- 6.4. Duration of fluid flow events in subducting slabs
produced by the model. To evaluate the introduced uncertainty, the
starting composition of the model was changed to represent these The applicability of Li chronometry on vein–wall rock systems
data points (model 2). Similar to model 1, the result for φR /φ0 = 5 is restricted to simple cases that only record a single, clearly de-
(Fig. 3b; grey curve) produced the best fit with a non-dimensional fined fluid–rock interaction event. A prerequisite is the formation
 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43 41

of a resolvable Li gradient in a reaction halo and, additionally, the it very likely that the type II vein-forming fluids were generated
rocks should lack any subsequent retrograde mineral reactions at in a deeper part of the subducting slab, simply because higher
lower metamorphic grades. The mélange block in this case study fluid volumes are needed to form such veins (Connolly and Podlad-
is a rare example of where these conditions are met and are well chikov, 2007; John et al., 2012; Plümper et al., 2017). The dynam-
preserved; almost all other exposures of HP rocks in New Caledo- ically forming porosity within dehydrating slabs ultimately leads
nia (and indeed across the globe) lack prograde veins that record to highly channelized fluid escape systems (Plümper et al., 2017;
single fluid-flow events and/or are affected by retrograde meta- Taetz et al., 2016). Once the fluid volume exceeds a critical thresh-
morphism. Nevertheless, Taetz et al. (2016) and Plümper et al. old, rapid and channelized fluid ascent through the subducting
(2017) showed that the developed patterns of fluid flow structures slab will be dominated by hydraulic decompaction–compaction
match those expected to form during dynamic rock dehydration phenomena, such as buoyancy-driven porosity waves (Connolly
in a subduction zone environment (Type I veins). Of significance and Podladchikov, 2007) or mobile hydrofractures (Bons, 2001).
is the finding that short-duration processes can cause the fea- The overall dehydration, fluid accumulation and fluid migration
tures described in this study. In fact, the calculated duration for occurs on timescales of 105 –106 years, estimated mainly from
fluid–rock interaction in the New Caledonia vein system of 1–4 the geometry and convergence rate of the subduction system
months is comparable to other inferred timescales of fluid–rock (Dragovic et al., 2015; John et al., 2012; Plümper et al., 2017;
interaction (e.g. Ague and Baxter, 2007; Camacho et al., 2005; Taetz et al., 2016). Individual fluid release steps, however, are
John et al., 2012). The distance an element diffuses into a medium much faster and highly localized in space and time, with pulses
(e.g., the host rock) is generally proportional to the square root as short as hundreds of years to months (Camacho et al., 2005;
of time available for the process to complete. This allows a rough John et al., 2012, this study).
comparison with similar diffusion profiles. For example, the time The currently available relatively small data set for such dis-
span of ca. 0.150 years needed for the evolution of a ca. 4 cm tinct pulses of slab fluid release indicates the short-lived, pulse-
broad selvage corresponds to a reaction front propagation of about like character of these processes. This is in accordance with a
26 cm2 /yr. We have recalculated the timescale needed to de- model suggesting that channelized fluid flow related to synmeta-
velop the Li diffusion profile of a fluid flow-induced reaction halo morphic dehydration reactions, for example at the blueschist-to-
in HP rocks from the Chinese Tianshan reported by John et al. eclogite transition, contributes to – or even dominates – fluid
(2012) using the updated partition coefficients and porosity data accumulation that eventually will cause slip phenomena by ris-
of this study (Fig. S1). This results in an estimated time of ca. ing pore fluid pressures at the plate interface. If these slip events
760 years for fluid–rock interaction for about 1 m, i.e. a reaction are sufficiently large and rapid, they may even trigger megathrust
front propagation of 0.13 cm2 /yr. Considering the exponentially in- earthquakes (cf. Fagereng and Diener, 2011; Moreno et al., 2014;
creasing time needed for diffusion with increasing distance from Schurr et al., 2014).
the vein, this corresponds to 3.3 cm2 /yr on the first 4 cm. The
initial velocities are broadly within the same range and suggest
7. Summary and conclusions
diffusion-driven reaction progress at a velocity of cm/yr with the
difference predominantly controlled by the porosity of the host
Two types of HP/LT veins in an eclogitic host rock from the PEM
rock. Camacho et al. (2005) suggested a duration in the order of
record two distinct dehydration processes related to the blueschist-
ca. 10 years for fluid pulses associated with shear zone formation
to-eclogite transition. As outlined by Spandler and Hermann (2006)
at eclogite-facies conditions. In this context, it is also worth men-
and Taetz et al. (2016), the formation of these veins can be at-
tioning that the large-scale pulsed flow-induced heating associated
with mountain building processes is envisaged to occur within a tributed to internal and external fluid flow processes, which oc-
maximum duration of 105 years (Ague and Baxter, 2007). curred on different time scales. The results of this study can be
To form veins and to achieve extensive interaction with the summarized as follows:
wall rock in such a short time, highly channelized fluid flow
with a high fluid volume or a highly reactive fluid is necessary 1. The overall dehydration, fluid accumulation and fluid migra-
(e.g. John et al., 2004, 2012). Plümper et al. (2017) showed that tion documented by the type I veins is a continuous process
dehydration ultimately leads to channelized flow, already at the and occurred on a timescale of 105 –106 years that is mainly
onset of reaction, and that pervasive porous flow does not ap- given by the geometry and convergence rate of the subduction
ply to dehydrating slabs. The petrogenetic model suggested by system.
Taetz et al. (2016) for the development of the New Caledonian 2. Different Li concentrations in both fluid reservoirs of the type
type I veins further supports this interpretation. The timescales II veins and the host rock led to the development of a char-
of fluid accumulation related to dehydration can be broadly con- acteristic diffusion profile (decreasing Li concentration and in-
strained by combining P –T data from pseudosection calculations creasing δ 7 Li) that allows for the application of Li chronom-
(e.g. Taetz et al., 2016) with age constraints for the PEM. As etry. In contrast to the type I veins, individual fluid release
schematically shown in Fig. 5, dehydration mainly took place be- steps documented by type II veins are much faster and highly
tween 460–540 ◦ C at 1.8–2.1 GPa. This corresponds to a normal localized in space and time. Diffusion modelling indicates a
subduction zone geotherm of 8 ◦ C/km and an associated ca. 80 ◦ C duration of 0.150+ 0.14
−0.08 years (1 to 4 months) for the selvage-
temperature increase as the slab descended by about 10 km depths forming fluid-flow event.
(vertically). Assuming an average convergence rate of 5–10 cm/a 3. The channelized fluid pulses associated with metamorphic slab
and a slab dip of about 45◦ this leads to a time of ca. 2–5 × dehydration can result in accumulation of high fluid volumes
105 years for the slab to pass through the relevant P –T window, within short time intervals to allow for a fast increase of pore
and the associated formation of dehydration-related type I veins. fluid pressures.
Dragovic et al. (2015) concluded that pulsed dehydration events 4. The short-lived character of this process (<1 year), suggests
with water loss of up to 0.5 wt% can occur in less than 8 × 105 that fluid release related to oceanic plate dehydration is likely
years in a subducting slab, which is in good agreement with our responsible for episodic pore fluid pressure increases at the
estimate for type I vein formation. The type II veins, however, plate interface. These occur localized in space and time and
are linked to an external fluid that is related to dehydration pro- may trigger various seismic to non-seismic slip phenomena at
cesses nearby or elsewhere in the subducting slab. We consider the plate interface.
42 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43

Acknowledgements Gao, J., Klemd, R., 2001. Primary fluids entrapped at blueschist to eclogite transition:
evidence from the Tianshan meta-subduction complex in northwestern China.
Contrib. Mineral. Petrol. 142, 1–14.
The Deutsche Forschungsgemeinschaft financially supported
Gomberg, J., et al., 2010. Slow-slip phenomena in Cascadia from 2007 and beyond:
this research through grant JO 349/5-1. We thank H. Baier for lab-
a review. Geol. Soc. Am. Bull. 122, 963–978.
oratory assistance and J. Berndt for his support on the electron Herms, P., John, T., Bakker, R.J., Schenk, V., 2012. Evidence for channelized external
microprobe and the LA-ICP-MS. Editorial handling by Mike Bickle fluid flow and element transfer in subducting slabs (Raspas Complex, Ecuador).
and constructive comments by Besim Dragovic and an anonymous Chem. Geol. 310–311, 79–96.
reviewer are greatly appreciated. Jamtveit, B., Austrheim, H., Malthe-Sorenssen, A., 2000. Accelerated hydration of the
Earth’s deep crust induced by stress perturbations. Nature 408, 75–78.
Appendix A. Supplementary material Jeffcoate, A.B., Elliott, T., Thomas, A., Bouman, C., 2004. Precise/small sample size de-
terminations of lithium isotopic compositions of geological reference materials
and modern seawater by MC-ICP-MS. Geostand. Geoanal. Res. 28, 161–172.
Supplementary material related to this article can be found on- John, T., Gussone, N., Podladchikov, Y.Y., Bebout, G.E., Dohmen, R., Halama, R., Klemd,
line at https://doi.org/10.1016/j.epsl.2017.10.044. R., Magna, T., Seitz, H.-M., 2012. Volcanic arcs fed by rapid pulsed fluid flow
through subducting slabs. Nat. Geosci. 5, 489–492.
References John, T., Klemd, R., Gao, J., Garbe-Schonberg, C.D., 2008. Trace-element mobiliza-
tion in slabs due to non steady-state fluid–rock interaction: constraints from an
Ague, J.J., 2003. Fluid infiltration and transport of major, minor, and trace elements eclogite-facies transport vein in blueschist (Tianshan, China). Lithos 103, 1–24.
during regional metamorphism of carbonate rocks, Wepawaug Schist, Connecti- John, T., Scherer, E.E., Haase, K., Schenk, V., 2004. Trace element fractionation during
cut, USA. Am. J. Sci. 303, 753–816. fluid-induced eclogitization in a subducting slab: trace element and Lu–Hf–Sm–
Ague, J.J., 2011. Extreme channelization of fluid and the problem of element mobil- Nd isotope systematics. Earth Planet. Sci. Lett. 227, 441–456.
ity during Barrovian metamorphism. Am. Mineral. 96, 333–352. Jonas, L., John, T., King, H.E., Geisler, T., Putnis, A., 2014. The role of grain boundaries
Ague, J.J., Baxter, E.F., 2007. Brief thermal pulses during mountain building recorded and transient porosity in rocks as fluid pathways for reaction front propagation.
by Sr diffusion in apatite and multicomponent diffusion in garnet. Earth Planet. Earth Planet. Sci. Lett. 386, 64–74.
Sci. Lett. 261, 500–516. Katayama, I., Terada, T., Okazaki, K., Tanikawa, W., 2012. Episodic tremor and slow
Aitchison, J.C., Clarke, G.L., Meffre, S., Cluzel, D., 1995. Eocene arc-continent colli- slip potentially linked to permeability contrasts at the Moho. Nat. Geosci. 5,
sion in New-Caledonia and implications for regional southwest pacific tectonic 731–734.
evolution. Geology 23, 161–164. Magna, T., Wiechert, U.H., Halliday, A.N., 2004. Low-blank isotope ratio measure-
Angiboust, S., Pettke, T., Hoog, J.C.M.D., Caron, B., Oncken, O., 2014. Channelized
ment of small samples of lithium using multiple-collector ICPMS. Int. J. Mass
fluid flow and eclogite-facies metasomatism along the subduction shear zone.
Spectrom. 239, 67–76.
J. Petrol. 55, 883–916.
Magna, T., Wiechert, U., Halliday, A.N., 2006. New constraints on the lithium isotope
Audet, P., Bostock, M.G., Christensen, N.I., Peacock, S.M., 2009. Seismic evidence for
compositions of the Moon and terrestrial planets. Earth Planet. Sci. Lett. 243,
overpressured subducted oceanic crust and megathrust fault sealing. Nature 457,
336–353.
76–78.
Baldwin, S.L., Rawling, T., Fitzgerald, P.G., 2007. Thermochronology of the New Cale- Marschall, H.R., Altherr, R., Rupke, L., 2007a. Squeezing out the slab – modelling the
donian high-pressure terrane: implications for middle Tertiary plate boundary release of Li, Be and B during progressive high-pressure metamorphism. Chem.
processes in the southwest Pacific. Spec. Pap., Geol. Soc. Am. 419, 117–134. Geol. 239, 323–335.
Bebout, G.E., 2007. Metamorphic chemical geodynamics of subduction zones. Earth Marschall, H.R., Pogge von Strandmann, P.A.E., Seitz, H.M., Elliott, T., Niu, Y.L., 2007b.
Planet. Sci. Lett. 260, 373–393. The lithium isotopic composition of orogenic eclogites and deep subducted
Beinlich, A., Klemd, R., John, T., Gao, J., 2010. Trace-element mobilization during Ca- slabs. Earth Planet. Sci. Lett. 262, 563–580.
metasomatism along a major fluid conduit: eclogitization of blueschist as a con- Moreno, M., Haberland, C., Oncken, O., Rietbrock, A., Angiboust, S., Heidbach, O.,
sequence of fluid–rock interaction. Geochim. Cosmochim. Acta 74, 1892–1922. 2014. Locking of the Chile subduction zone controlled by fluid pressure before
Beroza, G.C., Ide, S., 2011. Slow earthquakes and nonvolcanic tremor. Annu. Rev. the 2010 earthquake. Nat. Geosci. 7, 292–296.
Earth Planet. Sci. 39, 271–296. Oelkers, E.H., Helgeson, H.C., 1988. Calculation of the thermodynamic and trans-
Bons, P.D., 2001. The formation of large quartz veins by rapid ascent of fluids in port properties of aqueous species at high pressures and temperatures: aqueous
mobile hydrofractures. Tectonophysics 336, 1–17. tracer diffusion coefficients of ions to 1000 ◦ C and 5 kb. Geochim. Cosmochim.
Caciagli, N., Brenan, J.M., McDonough, W.F., Phinney, D., 2011. Mineral–fluid parti- Acta 52, 63–85.
tioning of lithium and implications for slab–mantle interaction. Chem. Geol. 280, Oliver, N.H.S., Bons, P.D., 2001. Mechanisms of fluid flow and fluid–rock interaction
384–398. in fossil metamorphic hydrothermal systems inferred from vein–wallrock pat-
Camacho, A., Lee, J.K.W., Hensen, B.J., Braun, J., 2005. Short-lived orogenic cycles and terns, geometry and microstructure. Geofluids 1, 137–162.
the eclogitization of cold crust by spasmodic hot fluids. Nature 435, 1191–1196.
Peacock, S.M., 1993. The importance of blueschist-eclogite dehydration reactions in
Carson, C.J., Powell, R., Clarke, G.L., 1999. Calculated mineral equilibria for eclogites
subducting oceanic-crust. Geol. Soc. Am. Bull. 105, 684–694.
in CaO–Na2 O–FeO–MgO–Al2 O3 –SiO2 –H2 O: application to the Pouebo Terrane,
Penniston-Dorland, S.C., Ferry, J.M., 2008. Element mobility and scale of mass trans-
Pam Peninsula, New Caledonia. J. Metamorph. Geol. 17, 9–24.
Chan, L.-H., Alt, J.C., Teagle, D.A.H., 2002. Lithium and lithium isotope profiles port in the formation of quartz veins during regional metamorphism of the
through the upper oceanic crust: a study of seawater–basalt exchange at ODP Waits River Formation, east-central Vermont. Am. Mineral. 93, 7–21.
Sites 504B and 896A. Earth Planet. Sci. Lett. 201, 187–201. Penniston-Dorland, S.C., Sorensen, S.S., Ash, R.D., Khadke, S.V., 2010. Lithium iso-
Clarke, G.L., Aitchison, J.C., Cluzel, D., 1997. Eclogites and blueschists of the Pam topes as a tracer of fluids in a subduction zone melange: Franciscan Complex,
Peninsula, NE New Caledonia: a reappraisal. J. Petrol. 38, 843–876. CA. Earth Planet. Sci. Lett. 292, 181–190.
Cluzel, D., Maurizot, P., Collot, J., Sevin, B., 2012. An outline of the geology of New Pirard, C., Spandler, C., 2017. The zircon record of high-pressure metasedimentary
Caledonia; from Permian-Mesozoic Southeast Gondwanaland active margin to rocks of New Caledonia: implications for regional tectonics of the south–west
Cenozoic obduction and supergene evolution. Episodes 35, 72–86. Pacific. Gondwana Res. 46, 79–94.
Connolly, J.A.D., Podladchikov, Y.Y., 2007. Decompaction weakening and channeling Plümper, O., John, T., Podladchikov, Y.Y., Vrijmoed, J.C., Scambelluri, M., 2017. Fluid
instability in ductile porous media: implications for asthenospheric melt segre- escape from subduction zones controlled by channel-forming reactive porosity.
gation. J. Geophys. Res., Solid Earth 112. Nat. Geosci. 10, 150–156.
Dragovic, B., Baxter, E.F., Caddick, M.J., 2015. Pulsed dehydration and garnet growth Putnis, A., John, T., 2010. Replacement processes in the Earth’s crust. Elements 6,
during subduction revealed by zoned garnet geochronology and thermodynamic 159–164.
modeling, Sifnos, Greece. Earth Planet. Sci. Lett. 413, 111–122.
Richter, F.M., Mendybaev, R.A., Christensen, J.N., Hutcheon, I.D., Williams, R.W., Stur-
Fagereng, A., Diener, J.F.A., 2011. Non-volcanic tremor and discontinuous slab dehy-
chio, N.C., Beloso, A.D., 2006. Kinetic isotopic fractionation during diffusion of
dration. Geophys. Res. Lett. 38.
ionic species in water. Geochim. Cosmochim. Acta 70, 277–289.
Frank, W.B., Shapiro, N.M., Husker, A.L., Kostoglodov, V., Bhat, H.S., Campillo, M.,
Scambelluri, M., Philippot, P., 2001. Deep fluids in subduction zones. Lithos 55,
2015. Along-fault pore-pressure evolution during a slow-slip event in Guerrero,
213–227.
Mexico. Earth Planet. Sci. Lett. 413, 135–143.
Gale, A., Dalton, C.A., Langmuir, C.H., Su, Y., Schilling, J.-G., 2013. The mean compo- Schurr, B., et al., 2014. Gradual unlocking of plate boundary controlled initiation of
sition of ocean ridge basalts. Geochem. Geophys. Geosyst. 14, 489–518. the 2014 Iquique earthquake. Nature 512, 299–302.
Gao, J., John, T., Klemd, R., Xiong, X.M., 2007. Mobilization of Ti–Nb–Ta dur- Skarbek, R.M., Rempel, A.W., 2016. Dehydration-induced porosity waves and
ing subduction: evidence from rutile-bearing dehydration segregations and episodic tremor and slip. Geochem. Geophys. Geosyst. 17, 442–469.
veins hosted in eclogite, Tianshan, NW China. Geochim. Cosmochim. Acta 71, Skelton, A., 2011. Flux rates for water and carbon during greenschist facies meta-
4974–4996. morphism. Geology 39, 43–46.
 S. Taetz et al. / Earth and Planetary Science Letters 482 (2018) 33–43 43

Spandler, C., Hermann, J., 2006. High-pressure veins in eclogite from New Caledo- Vitale Brovarone, A., Agard, P., 2013. True metamorphic isograds or tectonically
nia and their significance for fluid migration in subduction zones. Lithos 89, sliced metamorphic sequence? New high-spatial resolution petrological data for
135–153. the New Caledonia case study. Contrib. Mineral. Petrol. 166, 451–469.
Spandler, C., Pettke, T., Rubatto, D., 2011. Internal and external fluid sources for Vrijmoed, J.C., Austrheim, H., John, T., Hin, R.C., Corfu, F., Davies, G.R., 2013. Metaso-
eclogite-facies veins in the monviso meta-ophiolite, Western Alps: implications matism in the ultrahigh-pressure Svartberget garnet-peridotite (Western Gneiss
for fluid flow in subduction zones. J. Petrol. 52, 1207–1236. Region, Norway): implications for the transport of crust-derived fluids within
Spandler, C., Rubatto, D., Hermann, J., 2005. Late Cretaceous–Tertiary tectonics of the the mantle. J. Petrol. 54, 1815–1848.
southwest Pacific: insights from U–Pb sensitive, high-resolution ion microprobe Wagner, W., Pruß, A., 2002. The IAPWS formulation 1995 for the thermodynamic
(SHRIMP) dating of eclogite facies rocks from New Caledonia. Tectonics 24. properties of ordinary water substance for general and scientific use. J. Phys.
Chem. Ref. Data 31, 387–535.
Taetz, S., John, T., Bröcker, M., Spandler, C., 2016. Fluid–rock interaction and evolu-
Watson, E.B., Baxter, E.F., 2007. Diffusion in solid-Earth systems. Earth Planet. Sci.
tion of a high-pressure/low-temperature vein system in eclogite from New Cale-
Lett. 253, 307–327.
donia: insights into intraslab fluid flow processes. Contrib. Mineral. Petrol. 171,
Whitney, D.L., Evans, B.W., 2010. Abbreviations for names of rock-forming minerals.
90.
Am. Mineral. 95, 185–187.
Teng, F.Z., McDonough, W.F., Rudnick, R.L., Walker, R.J., 2006. Diffusion-driven ex- Wunder, B., Meixner, A., Romer, R.L., Heinrich, W., 2006. Temperature-dependent
treme lithium isotopic fractionation in country rocks of the Tin Mountain peg- isotopic fractionation of lithium between clinopyroxene and high-pressure hy-
matite. Earth Planet. Sci. Lett. 243, 701–710. drous fluids. Contrib. Mineral. Petrol. 151, 112–120.
Tomascak, P.B., Magna, T., Dohmen, R., 2016. Advances in Lithium Isotope Geochem-
istry. Springer International Publishing, Cham.