Review
Silicic magma reservoirs in the Earth’s crustk
OlivieR Bachmann
1,
* and chRistian huBeR
2
1
Institute of Geochemistry and Petrology, ETH Zurich, 8092 Switzerland
2
School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia 30332, U.S.A.
aBstRact
Magma reservoirs play a key role in controlling numerous processes in
planetary evolution, including igneous differentiation and degassing, crustal
construction, and volcanism. For decades, scientists have tried to under-
stand what happens in these reservoirs, using an array of techniques such
as eld mapping/petrology/geochemistry/geochronology on plutonic and
volcanic lithologies, geophysical imaging of active magmatic provinces,
and numerical/experimental modeling. This review paper tries to follow
this multi-disciplinary framework while discussing past and present ideas. We specically focus on
recent claims that magma columns within the Earth’s crust are mostly kept at high crystallinity (“mush
zones”), and that the dynamics within those mush columns, albeit modulated by external factors (e.g.,
regional stress eld, rheology of the crust, pre-existing tectonic structure), play an important role in
controlling how magmas evolve, degas, and ultimately erupt. More specically, we consider how the
chemical and dynamical evolution of magma in dominantly mushy reservoirs provides a framework
to understand: (1) the origin of petrological gradients within deposits from large volcanic eruptions
(“ignimbrites”); (2) the link between volcanic and plutonic lithologies; (3) chemical fractionation of
magmas within the upper layers of our planet, including compositional gaps noticed a century ago
in volcanic series (4) volatile migration and storage within mush columns; and (5) the occurrence of
petrological cycles associated with caldera-forming events in long-lived magmatic provinces. The
recent advances in understanding the inner workings of silicic magmatism are paving the way to
exciting future discoveries, which, we argue, will come from interdisciplinary studies involving more
quantitative approaches to study the crust-reservoir thermo-mechanical coupling as well as the kinetics
that govern these open systems.
Keywords: Magma reservoir, mush, differentiation of the Earth, volcanism, volcanic degassing,
Invited Centennial article, Review article
intROductiOn
Determining the shape, size, depth, and state of magma bodies
in the Earth’s crust as well as how they evolve over time remain
key issues for several Earth science communities. With a better
determination of these variables, petrologists could construct more
accurate chemical models of differentiation processes; magma
dynamicists could postulate causes for magma migration, storage,
and interaction at different levels within the crust; volcanologists
could better predict the style and volume of upcoming volcanic
eruptions; and geochronologists could construct an evolutionary
trend with greater precision. Although our knowledge on magmatic
systems has expanded greatly over the past decades, much remains
to be done to constrain the multiphase, multiscale processes that
are at play in those reservoirs and the rates at which they occur.
Previous attempts to summarize the state of knowledge in this
field (e.g., Smith 1979; Lipman 1984; Sparks et al. 1984; Marsh
1989a; de Silva 1991) have greatly helped crystallizing ideas on
the contemporaneous state of knowledge and possible alleys for
future directions to explore. Since then, many more publications
have attempted to draw magma reservoirs geometries and internal
complexities [see Fig. 1 for a potpourri of such schematic diagrams
for several examples of large caldera systems in the U.S.A., as well
as the more recent reviews by Petford et al. (2000), Bachmann et
al. (2007b), Lipman (2007), Cashman and Sparks (2013), Cash-
man and Giordano (2014), and de Silva and Gregg (2014)]. At this
stage, it is clear that magmas are complex mixtures of multiple
phases (silicate melt, H
2
O-CO
2
-dominated fluid, and up to 10
different mineral phases in some systems) with very different
physical properties. For example, viscosity contrast between water
vapor and crystals can span more than 20 orders of magnitude,
and even vary significantly within one given phase (e.g., several
orders of magnitude for silicate melt), as composition-P-T-f
H
2
O
-f
O
2
vary (Mader et al. 2013). The way these magmas move and stall
in the crust will therefore reflect the complex interaction between
all these phases and the typically colder wall rocks.
Over the last decades, an impressive body of work has ac-
cumulated on the characterization of magma mainly through:
(1) high-precision, high-resolution geochemical work [with
ever more powerful instruments, including the atom-probe, e.g.,
Valley et al. (2014) and nanoSIMS in the recent past; e.g., Eiler
et al. (2011); Till et al. (2015)], and (2) experimental petrology
American Mineralogist, Volume 101, pages 2377–2404, 2016
0003-004X/16/0011–2377$05.00/DOI: http://dx.doi.org/10.2138/am-2016-5675 2377
* E-mail: olivier.bachmann@erdw.ethz.ch
k Open access: Article available to all readers online.
Downloaded from http://pubs.geoscienceworld.org/msa/ammin/article-pdf/101/11/2377/3597914/3_5675BachmannOApc.pdf
by guest
on 16 August 2022
BACHMANN AND HUBER: MAGMA RESERVOIRS
2378
(e.g., phase diagrams, trace element partitioning between phases,
isotopic fingerprinting), culminating in powerful codes for ther-
modynamic modeling (Ghiorso and Sack 1995a; Boudreau 1999;
Connolly 2009; Gualda et al. 2012). This work will continue to
greatly improve the constraints that we place on these magmatic
systems for years to come, but we argue that the field will have
to take into account kinetic, disequilibrium processes, which are
commonplace, and can be severe in the world of magmas.
Thermodynamics and kinetics are both valuable to understand
magmatic processes, and kinetics is bound to become even more
important in the following decades as we strive to better constrain
the timescale of dynamical processes associated with magma
transport and evolution in the crust. This Centennial Volume
of American Mineralogist appears as an appropriate landmark
to summarize the recent advances on the subject and bring the
multiple facets of this problem together to better frame the future
directions of research.
The most exciting scientific challenges are, we believe, those
that are driven by enigmas arising by observations of Nature. Be-
low are a few examples of key questions or long-standing observa-
tions that relate to magma reservoirs and magmatic differentiation:
(1) The fundamental dichotomy in supervolcanic deposits (also
called ignimbrites or ash-flow tuffs), which show, in most cases:
(a) compositionally and thermally zoned dominantly crystal-poor
units [sometimes grading to crystal-rich facies at the end of the
eruption (Lipman 1967; Hildreth 1979; Wolff and Storey 1984;
Worner and Schmincke 1984; de Silva and Wolff 1995)], and (b)
strikingly homogeneous crystal-rich dominantly dacitic units (the
“Monotonous Intermediates” of Hildreth 1981), known to be some
of the most viscous fluids on Earth (Whitney and Stormer 1985;
Francis et al. 1989; de Silva et al. 1994; Scaillet et al. 1998b;
Lindsay et al. 2001; Bachmann et al. 2002, 2007b; Huber et al.
2009; Folkes et al. 2011b; see also Bachmann and Bergantz 2008b;
Huber et al. 2012a for reviews).
(2) The striking resemblance in chemical composition (for ma-
jor and trace elements) between the interstitial melt in crystal-rich
silicic arc magmas and the melt-rich rhyolitic magmas (Bacon and
Druitt 1988b; Hildreth and Fierstein 2000; Bachmann et al. 2005).
(3) The ubiquitous observation of magma recharges from
depth prior to eruptions, and their putative role on the remobiliza-
tion of resident (often crystal-rich) magmas (“rejuvenation”), for
large and small eruptions (e.g., Sparks et al. 1977; Pallister et al.
1992; Murphy et al. 2000; Bachmann et al. 2002; Bachmann and
Bergantz 2003; Wark et al. 2007; Molloy et al. 2008; Bachmann
2010a; Cooper and Kent 2014; Klemetti and Clynne 2014; Wolff
and Ramos 2014; Wolff et al. 2015).
(4) The long-noticed compositional gap in volcanic series,
coined the Bunsen-Daly Gap after R. Bunsen and R. Daly’s early
observations (Bunsen 1851; Daly 1925, 1933), and confirmed
in many areas worldwide since (Chayes 1963; Thompson 1972;
Brophy 1991; Thompson et al. 2001; Deering et al. 2011a; Szy-
manowski et al. 2015) even where crystal fractionation, which
should lead to a continuous melt composition at the surface,
dominates (Bonnefoi et al. 1995; Geist et al. 1995; Peccerillo
et al. 2003; Macdonald et al. 2008; Dufek and Bachmann 2010;
Sliwinski et al. 2015).
(5) The complex relationship between silicic plutonic and
volcanic units with possible hypotheses (not mutually exclusive
in plutonic complexes) ranging from plutons being “failed erup-
tions” (crystallized melts; Tappa et al. 2011; Barboni et al. 2015;
Glazner et al. 2015; Keller et al. 2015) to plutons being “crystal
graveyards” (crystal cumulates; Bachl et al. 2001; Deering and
FiguRe 1. Recent schematic diagrams for the three big active caldera systems in the Western U.S.A. There are many other caldera systems around the
world (e.g., Hughes and Mahood 2008), and the focus on examples from the U.S.A. here is solely due to personal acquaintance. (a) Valles caldera (Wilcock
et al. 2010), (b) Long Valley caldera (Hildreth 2004), (c) Yellowstone, Ofcial web site and Lowenstern and Hurvitz (2008).
Downloaded from http://pubs.geoscienceworld.org/msa/ammin/article-pdf/101/11/2377/3597914/3_5675BachmannOApc.pdf
by guest
on 16 August 2022
BACHMANN AND HUBER: MAGMA RESERVOIRS
2379
Bachmann 2010; Gelman et al. 2014; Putirka et al. 2014; Lee
and Morton 2015). Related to this volcanic-plutonic connection
is the corollary that granites sensu stricto (evolved, high-SiO
2
,
low-Sr composition) appear to be less common than their volcanic
counterparts (e.g., Navdat database, Halliday et al. 1991; Hildreth
2007; Gelman et al. 2014).
(6) The observation that silicic plutons, when fully crystallized,
contain little volatile left (typically <1 wt% H
2
O that is bound up
in the minerals; e.g., Caricchi and Blundy 2015) despite the fact
they started with volatile concentrations above 6 wt%, implying
significant loss of volatile at different stages of the magma bod-
ies’ evolution [e.g., by first and second boiling, pre-, syn-, and
post-eruption degassing, see for example Anderson (1974); Wal-
lace et al. (1995); Candela (1997); Lowenstern (2003); Webster
(2004); Wallace (2005); Bachmann et al. (2010); Blundy et al.
(2010); Sillitoe (2010); Baker and Alletti (2012); Heinrich and
Candela (2012)].
(7) The well-known “Excess S,” highlighted by the much
higher S content released during eruptions (measured by remote
sensing or estimated from ice-core records) that can be accounted
for by the melt inclusion data [i.e., petrologic estimate, see, for
example, reviews by Wallace (2001); Costa et al. (2003); Shino-
hara (2008)].
(8) The remarkable cyclic activity observed large-scale vol-
canic systems, which culminates in caldera-forming eruptions
(“caldera cycles”) that can repeat itself several times, e.g., the
multi-cyclic caldera systems such as Yellowstone (Hildreth et al.
1991; Bindeman and Valley 2001; Christiansen 2001; Lowenstern
and Hurvitz 2008), Southern Rocky Mountain Volcanic field
(e.g., Lipman 1984, 2007; Lipman and Bachmann 2015), Taupo
Volcanic Zone (e.g., Wilson 1993; Wilson et al. 1995; Sutton et
al. 2000; Deering et al. 2011a; Barker et al. 2014), High Andes
(e.g., Pitcher et al. 1985; Petford et al. 1996; Lindsay et al. 2001;
Schmitt et al. 2003; de Silva et al. 2006; Klemetti and Grunder
2008), Campi Flegreii (e.g., Orsi et al. 1996; Civetta et al. 1997;
Gebauer et al. 2014), and Aegean arc (e.g., Bachmann et al. 2012;
Degruyter et al. 2015).
All those questions/observations require an overarching con-
cept of magma chamber growth and evolution that ties together
these seemingly disparate concepts. Before attempting such a
task, first this briefly outlines the methods that are typically used
to infer the state of these reservoirs, highlighting some of their
strengths and limitations.
hOw can we study magma ReseRvOiRs?
Sampling volcanic and plutonic lithologies
The foundation of volcanology is motivated by observations of
ongoing and past volcanic activity, with the purpose of understanding
the underlying processes that govern the evolution of magmatic systems
on Earth and other bodies of the Solar System (Wilson and Head 1994).
However, large-scale eruptions are rare (Simkin 1993; Mason et al. 2004)
and as volcanologists/petrologists we rely much on a forensic approach:
study magmatic rocks (both plutonic and volcanic) that formed/erupted
in the past, and try to reconstruct the conditions (P, T, f
O
2
, f
H
2
O
, f
SO
2
, …)
that prevailed in the reservoirs at the time of formation or eruption [see,
for example, a review of techniques by Putirka (2008) or Blundy and
Cashman (2008)]. Of course, volcanic and plutonic lithologies do not
provide the same type of information, and could/should be complementary.
Volcanic rocks carry limited spatial context for the reservoir, but provide
an instantaneous snapshot into the state of the magma body just before
the eruption. In contrast, plutonic rocks present a time-integrated image
of the magma accumulation zone, often with a history spanning multi-
million years (e.g., Greene et al. 2006; Walker et al. 2007; Schoene et al.
2012; Coint et al. 2013), for which we can tease out some information
about sizes and shapes of magmas bodies. Much of the geochemical data
acquired over the last century is now tabulated in large databases such as
Georoc (http://georoc.mpch-mainz.gwdg.de/georoc/), EarthChem (http://
www.earthchem.org), and Navdat (http://www.navdat.org), providing a
remarkable resource to analyze global geochemical problems (see recent
papers of Keller and Schoene 2012; Chiaradia 2014; Gelman et al. 2014;
Glazner et al. 2015; Keller et al. 2015).
Studying active volcanoes, gas, and geophysical
measurements
Measuring gases coming out of active volcanoes provides us with some
key information about the state of magmatic systems (e.g., Giggenbach
1996; Goff et al. 1998; Edmonds et al. 2003; Burton et al. 2007; Hum-
phreys et al. 2009). Due to their low density and viscosity, volcanic gases
can escape their magmatic traps and provide “a telegram from the Earth’s
interior,” as laid out by one of the pioneers of gas measurements, Sadao
Matsuo (e.g., Matsuo 1962). For example, the abundance of chemical
species like He, CO
2
, or Cl can help determine the type of magma that is
degassing (Shimizu et al. 2005), and the potential volume that is trapped
in the crust (Lowenstern and Hurvitz 2008). As mentioned above, the
amount of S released during eruptions is also key in estimating the pres-
ence of an exsolved gas phase in the magma reservoir prior to an eruption
(Scaillet et al. 1998a; Wallace 2001; Shinohara 2008). As discussed by
Giggenbach (1996), the composition of volcanic gases measured at the
surface integrates both deep, i.e., magma reservoir processes, and shallow
processes within the structure of the edifice, and possibly the interaction
with a hydrothermal system. As such it is often challenging to directly
relate gas composition to the conditions of shallow magma storage (e.g.,
Burgisser and Scaillet 2007) and a better monitoring strategy is generally
to look for relative changes in gas compositions and temperature rather
than focusing on absolute values.
With the exception of gas measurements, imaging active magma
reservoirs mainly involves geophysical methods. Several techniques are
now available, and provide different pictures of active magmas bodies.
Such techniques include: (1) seismic tomography [both active and passive
sources (see for example Dawson et al. 1990; Lees 2007; Waite and Moran
2009; Zandomeneghi et al. 2009; Paulatto et al. 2010), and ambient-noise
tomography (e.g., Brenguier et al. 2008; Fournier et al. 2010; Jay et al.
2012)]; (2) magneto-telluric surveys (e.g., Hill et al. 2009; Heise et al.
2010); (3) deformation studies using strain-meter, GPS, and satellite
data (e.g., Massonnet et al. 1995; Hooper et al. 2004; Lagios et al. 2005;
Newman et al. 2006; Hautmann et al. 2014); and (4) muon tomography
on steep volcanic edifices (Tanaka et al. 2007, 2014; Gibert et al. 2010;
Marteau et al. 2012).
Geophysical methods are based on the inversion of signals transmitted
through the crust and measured from or close to the surface. The choice of
methods and, in several cases, the frequency band of the signals studied de-
pends on the targeted resolution and depth of interest. Geophysical methods
probe variations in physical properties caused by the presence of partially
molten reservoirs in the crust and provide two- or three-dimensional images
of these systems. These inversions however do not provide an unequivocal
picture of the thermodynamical state of the magma storage region, because
several variables (melt fraction, temperature, exsolved gas content, com-
position, connectivity of the various phases) affect the elastic, magnetic,
and electric properties of magmas. Similarly, surface deformation signals
are not only function of the state, shape, orientation, and size of a magma
body, but also depend on the crustal response to stresses around the body
Downloaded from http://pubs.geoscienceworld.org/msa/ammin/article-pdf/101/11/2377/3597914/3_5675BachmannOApc.pdf
by guest
on 16 August 2022
BACHMANN AND HUBER: MAGMA RESERVOIRS
2380
(through anelastic deformation and movement along weakness planes such
as fractures and faults; Newman et al. 2006; Gregg et al. 2013).
In summary, geophysical surveys mainly provide constraints about the
depth, shape, and extent of a partially molten (or hot) region under an active
volcanic edifice. The many surveys using inSAR, GPS field campaignsm
or ambient-noise tomography where, as for gas monitoring, the focus lies
on relative changes rather than the effective state of the reservoir have
significantly increased our ability to monitor volcanic unrest (see recent
review by Acocella et al. 2015). They efficiently highlight rapid (decadal
or less) changes in the mechanical state of magmatic systems, which is
generally difficult to achieve using petrological data sets. However, the
spatial scales probed by geophysical imaging techniques are limited by
the resolution of the method considered. Under most circumstances, the
spatial resolution remains greater or equivalent to the kilometer scale (Waite
and Moran 2009; Farrell et al. 2014; Huang et al. 2015), which is in stark
contrast with the extremely high resolution of geochemical and petrologi-
cal data. As such, geophysical inversions can be thought of as upscaling
filters of the state of heterogeneous multiphase reservoirs. A fundamental
assumption underlying these inversions is that spatial resolution is com-
mensurable with a Representative Elementary Volume (REV) where the
physical properties of the heterogeneous medium can be (linearly) aver-
aged out into a single effective value (a more detailed discussion of the
limitations imposed by this assumption is provided below).
Reproducing the conditions prevailing in magma chambers
in the laboratory: Experimental petrology
(1) Phase-equilibrium experiments. Starting with N.L. Bowen at the
turn of the 20th century, many geoscientists have applied experimental
studies to study the thermodynamics that control the assemblage of phases
in magmas with different bulk compositions. For nearly a century, a large
amount of data has been collected on crystal-melt stability and composition
as a function of several variables (P, T, fugacities of volatile elements and
oxygen, compositions of magmas, cooling rate…). Much of these data are
now available in databases, some in web-based portals, as is the case for
the geochemical data (see for example Berman 1991; Holland and Powell
2011 or LEPR, http://lepr.ofm-research.org/YUI/access_user/login.php).
The data sets range from crystal-melt equilibria in the mantle melting
zones (e.g., see Poli and Schmidt 1995 to lower crustal (e.g., Muntener
and Ulmer 2006; Alonso-Perez et al. 2009) and upper crustal conditions
in many different settings (e.g., Johnson and Rutherford 1989; Johannes
and Holtz 1996; Scaillet and Evans 1999; Costa et al. 2004; Almeev et al.
2012; Gardner et al. 2014; Caricchi and Blundy 2015).
For shallow magma reservoirs in arcs, a third phase, volatile bubbles,
is also commonly present. This exsolved volatile phase does not only
affect the thermo-physical properties of magmas (see below) but can
also impact greatly the partitioning of some volatile trace elements and
therefore influence the chemical fractionation that takes place in magma
reservoirs. Laboratory experiments have played a major role in determin-
ing the solubility of volatile phases (H
2
O, CO
2
, S, CL, F…) as a function
of pressure, temperature, and melt composition [see the review by Baker
and Alletti (2012) and references therein]. Several species, such as Cl and
S compounds, have a strong influence on the fractionation and transport
of precious metals out of magma reservoirs to the site of ore deposits
and have therefore received a thorough attention (Scaillet et al. 1998a;
Williams-Jones and Heinrich 2005; Zajacz et al. 2008; Sillitoe 2010;
Fiege et al. 2014).
Chemical equilibrium between the melt and newly formed crystals
or crystal rims is often a decent assumption, but this assumption breaks
loose when changes taking place in the reservoir are more rapid than the
equilibration timescale (see Pichavant et al. 2007). Such rapid changes
in reservoirs’ conditions can include: (1) magma recharges and partial
(incomplete) mixing; (2) efficient physical separation of mineral and/or
bubble-melt; and (3) during rapid pressure drops associated with mass
withdrawal events (eruptions, dike propagation out of the chamber). In
these particular cases, kinetics controls the mass balance between the
different phases; one therefore needs to consider diffusion of elements
in silicate melts and minerals, as well as mineral and bubble growth or
dissolution. The diffusive transport of chemical species in silicate melts
is a strong function of: (1) viscosity (which is a proxy for the degree of
polymerization of the melt); (2) dissolved water content; and (3) to a lesser
extent, oxygen fugacity (redox conditions in the magma affect speciation).
We bring the attention of the interested readers to the review of Zhang and
Cherniak (2010) for additional information on this topic.
The presence of kinetics is readily observed in mineral zoning patterns
because solid-state diffusion is generally orders of magnitude slower than
diffusion in silicate melts. Over the last decade, several models have been
developed to use the diffusion of cations in silicate minerals and retrieve
timescales relevant to the dynamic evolution of magmas before and during
eruptions. These models rest heavily on experiments where the kinetics
of diffusion as a function of temperature and composition is measured
[see review of Zhang and Cherniak (2010) and reference therein]. Diffu-
sion modeling in minerals informs us on the interval of time between a
significant change in thermodynamic conditions of the magma (mixing
from recharges for example) and the time at which the minerals transition
across the closure temperature (e.g., eruption) where subsequent diffusion
can be ignored. To date, studies have explored the diffusion of a growing
quantity of cations in hosts such as quartz, plagioclase, pyroxene, biotite,
and olivines (for more mafic systems than the ones considered here; see
Fig. 2 for published examples).
Stable isotopes can also provide information regarding the thermo-
dynamic conditions (equilibrium fractionation between different phases)
and kinetics (kinetic fractionation) in the magma. Because of improved
accuracy and spatial resolution of analytical measurements, there is a
growing interest in using different stable isotope systems, such as O, S,
Ca, as well as other major elements, to fingerprint the oxidation state of
magmas (Metrich and Mandeville 2010; Fiege et al. 2015), crustal as-
similation (e.g., Friedman et al. 1974; Taylor 1980; Halliday et al. 1984;
Eiler et al. 2000; Bindeman and Valley 2001; Boroughs et al. 2012), or
mixing processes between magmas with different compositions (Richter
et al. 2003; Watkins et al. 2009a).
(2) Rheological experiments. The migration, convection, and defor-
mation of magmatic mixtures in the crust have direct implications on the
rate and efficiency of fractionation processes, on the cooling rate of magma
reservoirs, and eruption dynamics (e.g., Gonnermann and Manga 2007;
Lavallée et al. 2007; Cordonnier et al. 2012; Gonnermann and Manga 2013;
Parmigiani et al. 2014; Pistone et al. 2015). The response of magmas to
differential stresses is complex, owing to the multiphase nature of these
systems. Several research groups have performed deformation experiments
to better constrain rheological laws pertinent to magmas (Spera et al. 1988;
Lejeune and Richet 1995; Caricchi et al. 2007; Ishibashi and Sato 2007;
Champallier et al. 2008; Cimarelli et al. 2011; Mueller et al. 2011; Vona et
al. 2011; Pistone et al. 2012; Del Gaudio et al. 2013; Laumonier et al. 2014;
Moitra and Gonnermann 2015). These studies have focused on diverse
aspects of the rich non-linear dynamics of the rheology of suspensions,
such as: (1) the effect of the crystal volume fraction; (2) the effective shear
rate; and (3) the effect of polydisperse crystal size distributions and vari-
ous crystal shapes. These results clearly show that suspensions become
very stiff as crystal content approach or exceed 40 to 50% of the volume
of the magma, although they provide little information about the onset
and magnitude of a yield stress for crystal-rich suspensions. Even more
challenging is the development of physically consistent rheology laws for
three-phase magmas because of the additional complexity of the capillary
stresses (coupling bubbles, melt, and crystals), bubble deformation, three
phases lubrication effects and possible coalescence (Pistone et al. 2012;
Truby et al. 2014).
Magma rheology experiments have been synthesized to yield differ-
ent empirical (Lavallée et al. 2007; Costa et al. 2009) and semi-empirical
(Petford 2009; Mader et al. 2013; Faroughi and Huber 2015) models for
Downloaded from http://pubs.geoscienceworld.org/msa/ammin/article-pdf/101/11/2377/3597914/3_5675BachmannOApc.pdf
by guest
on 16 August 2022
BACHMANN AND HUBER: MAGMA RESERVOIRS
2381
the deformation of magmas under shear flow conditions. The mechanical
behavior of magmas remains a rich field for future investigations, and
several important challenges remain to constrain the rate at which magma
flows in reservoirs and up to the surface. They include:
• How to treat multiphase rheology when phase separation
takes place concurrently?
• How do bubbles interact with crystals suspended in viscous
melts, and how does this behavior depend on the strain rate
and volume fractions of each phases?
• Can we constrain the yield strength of crystal-rich magmas
under magma chamber conditions (imposed stresses rather
than imposed strain-rate)?
• By rheology, we often restrict ourselves to measure the shear
viscosity using experiments or models that relate stress and
deformation under very simple geometries. As viscosity is a
tensor, is it possible to measure its anisotropy for non-linear
multiphase systems with complex crystal shapes under vari-
ous strain rates and conditions pertinent to magma chambers?
The rheology of a magma body should therefore be considered as a
dynamical quantity, which can vary significantly (orders of magnitude)
because of changes in stress distributions, variations in crystal-melt-bubble
phase fractions and potentially many other factors.
Several other physical properties have a resounding impact on the
evolution of magmas in the crust. Magmas are far from thermal equilib-
rium in the mid to upper crust, and heat transfer in and out of these bodies
controls their chemical and dynamic evolution to a great extent (Bowen
1928; McBirney et al. 1985; Nilson et al. 1985; McBirney 1995; Rein-
ers et al. 1995; Thompson et al. 2002; Spera and Bohrson 2004). In that
context, the heat capacity, latent heat, and thermal conductivity of silicate
melts/minerals and CO
2
-H
2
O fluids are important to establish the timescale
over which these reservoirs crystallize in the crust and by extension the
vigor of convective motion in these systems. It is important to emphasize
that thermal convection in the classical sense is not directly applicable
to magma chambers under most conditions, because density contrasts
between the different phases far exceeds those by thermal expansion (e.g.,
Marsh and Maxey 1985; Bergantz and Ni 1999; Dufek and Bachmann
2010). Additionally, the rate of crystallization directly impacts the cooling
rate of magmas (due to non-linear release of latent heat during crystalliza-
tion, e.g., Huber et al. 2009; Morse 2011), and this latent heat buffering
becomes dominant as silicic magmas approach eutectic behavior near their
solidus (Gelman et al. 2013b; Caricchi and Blundy 2015). Some recent
studies have tested how the temperature dependence of thermal properties
influence the cooling and crystallization timescales of magmas in the crust
and showed that these effects are sometimes non-trivial (Whittington et
al. 2009; Gelman et al. 2013b; de Silva and Gregg 2014).
Tank experiments for magma chamber processes
An elegant approach to study the complex multiphase fluid dynam-
ics that takes place in magma reservoir is to resort to laboratory tank
experiments (see for example, McBirney et al. 1985; Nilson et al. 1985;
Jaupart and Brandeis 1986; Martin and Nokes 1988; Shibano et al. 2012).
Experiments provide the means to investigate some aspects of the complex
non-linear dynamics that prevail in these systems, however, in most cases, it
is difficult or impossible to satisfy dynamical similitude with real magmatic
systems. Nevertheless, experiments reveal interesting feedbacks that may
have been overlooked and can serve as a qualitative assessment of how
processes couple to each other. One of the main directions for laboratory
fluid dynamics experiments is to study how magmatic suspensions behave
at low Reynolds number during convection, phase segregation, and magma
recharges with mixing (e.g., McBirney 1980; Sparks et al. 1984; McBirney
et al. 1985; Davaille and Jaupart 1993; Koyaguchi et al. 1993). Here, the
term suspension is used loosely to describe both suspension with solid
particles (crystals) and bubbly emulsions.
As discussed throughout this review [and clearly alluded to by early
petrologists such as R. Daly as far back as the 1910s, e.g., Daly (1914)],
magma bodies are open reservoirs that exchange mass and energy with
their surroundings. Chemical heterogeneities and zonations then reflect
either phase separation by gravity (e.g., McBirney 1980, 1993; Hildreth
1981; Sparks et al. 1984) or incomplete mixing between magmas with
different compositions (Eichelberger 1975; Dungan et al. 1978; Blake
FiguRe 2. Examples of diffusional proles in quartz crystals from Taupo, New Zealand (Matthews et al. 2012), in pyroxene crystals from the
Bishop Tuff, California, U.S.A. (Chamberlain et al. 2014a), and plagioclase crystals from Cosigüina, Nicaragua (Longpré et al. 2014).
Downloaded from http://pubs.geoscienceworld.org/msa/ammin/article-pdf/101/11/2377/3597914/3_5675BachmannOApc.pdf
by guest
on 16 August 2022