See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/247179925

Physical Properties of Magmas

Article · January 2000

CITATIONS READS
167 10,420

1 author:

Frank J. Spera
University of California, Santa Barbara
204 PUBLICATIONS 8,612 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Evolution of an open-system magma body at Chaos Crags: Nd-Sr isotope variations within mafic recharge magmas View project

Magma Chamber Simulator Research Group View project

All content following this page was uploaded by Frank J. Spera on 13 December 2017.

The user has requested enhancement of the downloaded file.

 Author's personal copy

Provided for non-commercial research and educational use only.

Not for reproduction, distribution or commercial use.

This chapter was originally published in the book The Encyclopedia of Volcanoes.
The copy attached is provided by Elsevier for the author's benefit and for the benefit of
the author's institution, for non-commercial research, and educational use. This includes
without limitation use in instruction at your institution, distribution to specific
colleagues, and providing a copy to your institution's administrator.

All other uses, reproduction and distribution, including

without limitation commercial reprints, selling or
licensing copies or access, or posting on open
internet sites, your personal or institution’s website or
repository, are prohibited. For exceptions, permission
may be sought for such use through Elsevier’s
permissions site at:
http://www.elsevier.com/locate/permissionusematerial

From Lesher, C.E., Spera, F.J., 2015. Thermodynamic and Transport Properties of Silicate
Melts and Magma. In: Sigurdsson, H., Houghton, B., Rymer, H., Stix, J., McNutt, S.
(Eds.), The Encyclopedia of Volcanoes, pp. 113–141.
ISBN: 9780123859389
Copyright © 2015 Elsevier Inc. All rights reserved.
Academic Press
 Author's personal copy

Chapter 5

Thermodynamic and Transport Properties

of Silicate Melts and Magma
Charles E. Lesher
Department of Earth and Planetary Sciences, University of California, Davis, CA, USA,
Department of Geoscience, Aarhus University, Aarhus, Denmark

Frank J. Spera
Department of Earth Science, University of California, Santa Barbara, CA, USA

Chapter Outline 4.3. Thermal Conductivity 125

1. Introduction 114 4.4. Diffusion 126
2. Magmatic Systems: Time and Length Scales 116 5. Conclusions 130
3. Magma Thermodynamic Properties 117 Supplementary Material for Thermodynamic and
3.1. Density and EOS 118 Transport Properties of Silicate Melts and Magma 131
3.2. Enthalpy, Entropy, and Heat Capacity 120 Acknowledgments 139
4. Magma Transport Properties 121 Further Reading 139
4.1. Melt Viscosity 121
4.2. Magma Viscosity 123

needed to completely fuse (melt) gabbro, the plutonic (crystalline)

Still, however, the highest value is set on glass that is nearly
equivalent of basaltic magma, is about 500 kJ/kg.
colorless or transparent, as nearly as possible resembling crystal.
magma High-temperature multiphase mixture of solids (cognate
For drinking vessels glass has quite superseded the use of silver crystals and exotic lithic fragments), silicate or carbonatitic
and gold
liquid, and HeOeCeSeCl-rich gas or supercritical fluid formed
Pliny the Elder (c. AD 23e79) by partial or total melting of parental source material.
magma rheology The rheological properties of magma depend on
temperature, bulk composition, pressure, phase assemblage
(melt  crystals  vapor), particle size and shape distribution,
GLOSSARY spatial arrangement of particles (structure), and shear rate in a
complex and intertwined manner characterized by multiple
basaltic magma Most typical type of magma erupted on Earth and feedback loops.
other terrestrial planets. Basaltic magma is erupted at tempera- magma transport phenomena This term encompasses all the
tures 1000e1300
C, and is made up mainly of SiO2 (50 wt%), dynamical processes responsible for the generation, ascent,
Al2O3 (15%), CaO (12%), and roughly equal amounts of FeO and eruption or emplacement of magma, its subsequent quenching or
MgO (10%). The alkalis (Na2O þ K2O) make up most of the solidification, and interaction in hydrothermal magmatic systems.
difference although trace amounts of every naturally occurring These processes involve simultaneous consideration of heat,
element are invariably present. About 25 km3 of basaltic magma mass, and momentum transport between relevant magmatic sub-
is generated and erupted or emplaced each year on Earth, pri- systems. Rates of momentum, heat, and mass transfer are widely
marily along the 75,000 km of diverging plate boundaries. The varied ranging from creeping (percolative) flow of small degree
density, specific heat, viscosity, and thermal conductivity of a partial melts through an otherwise solid parental source rock in
typical basaltic magma at 1200
C at low pressure are 2600 kg/ response to pressure, buoyancy, and viscous forces to the rapid
m3, 1450 J/kg K, 100 Pa s, and 0.6 W/m K, respectively. The heat (several hundred meters per second) eruption of low density

The Encyclopedia of Volcanoes. http://dx.doi.org/10.1016/B978-0-12-385938-9.00005-5 113

Copyright Ó 2015 Elsevier Inc. All rights reserved.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
114 PART | I Origin and Transport of Magma

highly expanded magmatic mixtures characteristic of high-silica, transport properties In typical dynamic (nonequilibrium) magmatic
volatile-rich rhyolitic magma that erupts to form pyroclastic flows systems, gradients in pressure, stress or velocity, temperature, and
of great natural hazard. The flow of magma in fractures, buoyant chemical potential give rise to transport of momentum, heat, and
rise of magma diapirs and evolution of magma within crustal mass, respectively. Momentum, heat, and mass can be transported
magma bodies that undergo simultaneous assimilation of wall by advective or diffusive flow. The transport properties that
rock, recharge of fresh parental magma and fractional crystalli- govern diffusive flow of momentum, heat, and mass are the vis-
zation and convective mixing (or unmixing) are all examples of cosity, thermal conductivity and self, tracer and chemical diffu-
magma transport phenomena. sivities, respectively.
metasomatism Metasomatism refers to the process whereby a pre-
existing igneous, sedimentary, or metamorphic rock undergoes
compositional and mineralogical transformations associated with NOMENCLATURE
chemical reactions triggered by the reaction of fluids (so-called
T Temperature (K)
metasomatic agents), which invade the protolith. A large-scale
P Pressure (Pa)
example is provided in subduction environments. Dehydration
r Density (kg/m3)
of hydrothermally altered oceanic crust and attached upper mantle
Cp Molar isobaric heat capacity (J/mol K)
generates fluids that migrate upwards and metasomatize the ul-
cp Specific isobaric heat capacity (J/kg K)
tramafic mantle wedge lying above the subducting slab. Because
Cpi, cpi Partial molar, specific isobaric heat capacity (J/mol K),
the peridotite solidus temperature is lowered by the presence of
(J/kg K)
water, metasomatism of the mantle wedge can trigger partial
S Entropy (J/K mol)
melting there. A smaller scale environment in which meta-
H Enthalpy (J/mol)
somatism is important occurs in contact metamorphic aureoles
s Interfacial surface energy (N/m)
surrounding granitic plutons when emplaced into cooler preex-
bT Isothermal compressibility,(vlnr/vP)T (Pa1)
isting crust. The magma body acts as a heat engine and drives
aP Isobaric expansivity, (vlnr/vT)P (Ke1)
hydrothermal circulation in the surrounding country rock.
Xi Mole fraction of ith component
Economically important mineral deposits form in contact meta-
Vi Partial molar volume of ith component (m3/mol)
morphic environments.
KT Isothermal bulk modulus (Pa)
rhyolitic magma Rhyolitic magma is erupted at temperatures
K0 Pressure derivative of bulk modulus
750e1000
C, and is made up mainly of SiO2 (75 wt%), Al2O3
h Viscosity (kg/m s h Pa s)
(13%), and the Na2O and K2O in roughly equal amounts (3e5%).
EA Activation energy for viscous flow (J/mol)
Only a small fraction of a cubic kilometer of rhyolitic magma is
VA Activation volume for viscous flow (m3/mol)
erupted each year mainly in continental environments although a
hr Relative viscosity, ratio of mixture viscosity to melt viscosity
significant fraction of rhyolitic and intermediate-composition
f Volume fraction dispersed phase (solid or vapor)
magmas may stagnate at depth within the crust crystallizing
kR Radiative (photon) conductivity (J/m K s)
there to form granitic plutons. Granitic batholiths are generally
k Thermal conductivity (J/m K s)
associated with subduction zone magmatism and can extend for
k Thermal diffusivity, k/rcp (m2/s)
hundreds to thousands of kilometers roughly parallel to the strike
D Self, tracer or chemical diffusivity (m2/s)
of oceanic trenches and along active continental margins. The
Ea Activation energy for diffusion (J/mol)
density, specific heat, viscosity, and thermal conductivity of
Va Activation volume for diffusion (J/mol)
typical rhyolitic magma bearing 2 wt% dissolved H2O at 900
C
at low pressure, are 2260 kg/m3, 1450 J/kg K, 500,000 Pa s, and
1 W/m K, respectively. A water-free rhyolitic melt of otherwise Subscripts
identical composition has a density and viscosity of 2350 kg/m3
and 1.2  1010 Pa s, respectively. The heat needed to completely fus fusion
fuse (melt) granite, the plutonic (crystalline) equivalent of rhyo- form formation
litic magma, is w300 kJ/kg. Pristine volatile contents of typical conf configuration
rhyolitic magmas lie in the range 1e6 wt% and are mainly H2O mix mixing
although finite amounts of CO2 and sulfur-rich gasses (H2S, SO2, r relative
etc.) are also present. Volcanic gasses are often corrosive and
contain HCl and HF as minor constituents.
thermodynamic properties Magma thermodynamic properties may 1. INTRODUCTION
be separated into two families: thermal functions and the equation
A central goal of studies in igneous petrology and volca-
of state (EOS). Thermal functions relate the temperature of a sub-
stance to its internal energy. These properties include the enthalpy,
nology is to understand the factors that lead to the compo-
entropy, and heat capacity. The EOS relates the density of a sub- sitional diversity of magmatic rocks and the related issue of
stance to its composition, pressure, and temperature, and important the origin of the Earth’s crust and mantle. In the crust are
properties include the isothermal compressibility and isobaric found essentially all of the material and energy resources
expansivity. The reactivity of a material depends on its chemical accessible to the world population of 7 billion. This under-
potential that involves both thermal functions and EOS data. standing is most powerful when framed in terms of

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 115

petrogenesisdthe origin of magma and the rocks formed or basaltic). Carbonatitic melts containing >50 wt% car-
from cooling and solidification of magmadwithin the bonate are also generated within the mantle. Although the
context of the coupling between the thermal evolution and volumetric rate of eruption of carbonatitic magma is very
chemical differentiation of Earth. The growth of oceanic and small compared to the roughly 25 km3/year of mid-ocean
continental crust throughout the w4560 million years of ridge basaltic magma produced, rare carbonatitic magmas
earth history, the extent of recycling of crust and lithosphere may be important agents for mantle metasomatism. Car-
by subduction, the relationship of mantle plumes to the bonatites also have an affinity with economically important
compositional differentiation of Earth, and the role of sub- diamond-bearing kimberlitic magmas, which are rapidly
duction in island arc magmatism and growth of continental erupted from depths of several hundred kilometers, if not
crust are problems that studies of magmas and magmatic more.
processes shed light upon. In addition, a close connection Common magmas vary nearly continuously in compo-
exists between magmatic diversity and practical problems, sition from basaltic (50 wt% SiO2) to rhyolitic (75 wt%
such as volcano forecasting, the mitigation of volcanic SiO2). These magmas usually contain small amounts (on the
hazards, and the discovery of material and energy resources order of a few weight percent) of dissolved H2O, CO2, and
such as ore deposits and geothermal heat. The thermody- other volatile species such as H2S, N2, HCl, HF, COS, and
namic and transport properties of magma are central to all SO2. H2O dissolved in melts occurs in two forms: molecular
of these considerations. H2O and as the hydroxyl polyanion OH. The ratio of mo-
There has been a sustained and accelerating effort in the lecular water to hydroxyl depends on the composition
last century to determine the properties of magma both in (devolatilized) of the melt. When sufficient volatile com-
the laboratory and by application of models based on the ponents are present, melt becomes supersaturated and a
chemistry and physics of materials. It is no exaggeration to discrete vapor or supercritical fluid forms. This fluid has a
claim that without these foundational measurements and low viscosity and density compared to silicate melts and is
theoretical models, petrology could not have evolved far particularly rich in the components OeHeCeSeNeCleF.
beyond the purely descriptive stage. In this chapter, we The speciation of this fluid depends on bulk composition,
provide an overview of current knowledge of the subject temperature, and pressure. At crustal pressures and tem-
and future directions. peratures, supercritical fluids, especially those rich in mo-
The composition of lava emitted from a volcanic center lecular H2O, are quite corrosive and can dissolve a few
reflects both the composition of its source as well as myriad percent or more by mass of other oxide components. In part,
dynamical phenomena operating during its generation, this is due to H2O’s dipolar nature and large dielectric con-
segregation, ascent, residence time in crust (storage) and stant making it a good solvent. At very high pressure and
eruption. For the purposes of this chapter, magma is defined temperature even larger quantities of solid may dissolve in
as a high temperature (generally >900 K), multiphase these fluids, while at conditions of subduction zone mag-
mixture of crystals, liquid, and vapor. The vapor can be matism, i.e., depths of w100 km (3 GPa) and temperatures
either a gas or a supercritical fluid. Herein, the term liquid of 700e1400
C, silicate melt and hydrous fluids can be
is used interchangeably with melt (i.e., they are synonyms) completely miscible. The concentration of dissolved vola-
and fluid generally restricted to a phase that is rich in H, O, tiles in a melt is strongly dependent on pressure because the
C, N, Cl, and S. The solid fraction of magma is primarily partial molar volume of a volatile species in the dissolved
made up of oxide and silicate crystals, the relative abun- state is much smaller than its molar volume in the gaseous or
dance of which depends strongly on composition, temper- supercritical fluid state. It is this difference in volume that
ature, pressure, and additional nonequilibrium or kinetic relates the Gibbs free energy, and hence solubility, to pres-
factors, such as cooling rate, rate of decompression, and sure at fixed temperature. The huge volumetric expansion of
rates of mass transfer by molecular diffusion, mechanical a magmatic mixture that accompanies exsolution of vola-
dispersion, and advective transport. Bits of local wall rock tiles is one of the primary causes of explosive volcanism as
(lithic inclusions), cognate crystal cumulates, or exotic pressureevolume (PV) expansion work is converted to ki-
xenoliths are also commonly found in plutonic and volcanic netic energy. Although at depths of a few kilometers magma
rocks and provide clues to understanding magma genesis ascent rates may be only a fraction of a meter per second,
and dynamics as well as a providing constraints on the bulk once volatile exsolution commences and PV work is con-
composition of the source. The liquid or melt portion of verted into kinetic energy, magma eruption speeds of order
magma is generally a multicomponent (OeSieAleCae 100e300 m/s can easily develop. Other factors, such as
MgeFeeKeNaeHeC) silicate liquid; crustal melts (si- magma viscosity and the rates of volatile component diffu-
licic or rhyolitic to intermediate or andesitic) are rich in sion, and their variations with temperature and pressure are
OeSieAleNaeKeH whereas melts generated within the also important in assessing the dynamics of explosive
Earth’s mantle are richer in OeSieAleCaeMgeFe (mafic volcanism.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
116 PART | I Origin and Transport of Magma

TABLE 5.1 Estimated Properties of Common Natural Melts at 1 bar (104 GPa) at Their Respective Liquidus
Temperatures. All Compositions Are Anhydrous Except Where Indicated. See Tables S5.1eS5.7 for More Exhaustive
Compilations of These Properties.

Liquidus Specific Heat Specific Isobaric Melt

Temperature of Fusion Density Heat Capacity Viscosity
Composition (
C) (kJ/kg) Dhfus (kg/m3) r (J/kg K) cp (Pa s) h

Granite (dry) 900 220 2349 1375 1.2  1010

Granite (2 wt% H2O) 900 250 2262 1604 5  105
Granodiorite 1100 354 2344 1388 1.3  106
Gabbro 1200 396 2591 1484 30.0
Eclogite 1200 570 2591 1484 e
Komatiite 1500 540 2748 1658 0.15

Peridotite 1600 580 2689 1793 0.25

Volcanic rocks, in particular, play a unique role in ef- compilations are not exhaustive and particular research
forts to understand magmatic transport phenomena applications may require a return to the original sources for
because, unlike plutonic rocks, they are quenched relatively additional details. Some of these sources are listed in
rapidly and provide more or less direct information Further Reading and many more are provided in the Sup-
regarding the composition of natural melts. In order to plementary Material section. The Nomenclature includes
understand the significance of the chemical composition of symbols used in the text.
volcanic rocks, the dynamical and physical aspects of its
parental magma evolution must first be unraveled. This is
not an easy task. The range of transport phenomena of 2. MAGMATIC SYSTEMS:
potential relevance to petrogenesis is quite varied and
covers large spatiotemporal scales. Although the dynamics
TIME AND LENGTH SCALES
can be complex, it is clear that the thermodynamic and Lifetimes of magmatic systems and processes vary
transport properties of magma are absolutely pivotal to the widelydfrom the rapid radiative quenching of a
success of any quantitative dynamical theory of magma millimeter-sized melt droplet (pyroclast) during its high-
genesis and transport. If twentieth-century research in speed ejection from a vent (several seconds), to the slow
petrology can be summarized in a few words, it was the cooling of a single lava flow (weeks or months), to the
century of quantificationdquantification of the energetics hundred thousand to million-year timescale of a large
of magma. Although far more needs to be learned, it is pluton cooling mainly by phonon (heat) conduction and
reasonable to claim that geologists now have a first-order hydrothermal convection. Individual magmatic hydrother-
understanding of the basic properties of the common mal systems, such as at Yellowstone National Park, USA,
magma types and are beginning to understand how such can remain active for a few millions of years because heat
properties relate to igneous petrogenesisdthe origin of conduction is intrinsically a slow process (rocks are good
igneous rocks. In the twenty-first century, new instrumental insulators) and because many magmatic systems are open
techniques are providing a treasure trove of information systems replenished by entry of magma rising from deeper
regarding the composition of crystals, glass inclusions, and in the crust or mantle below. This replenishment or
grain boundaries at the micron- to nanoscale. This infor- recharge magma is often hotter than resident magma
mation must ultimately be connected to dynamics of because it comes from greater depths where higher tem-
magmatic systems through the macroscopic petrogenetic peratures are usually found. In terms of dynamical process,
prism. This will require an even better understanding of the creeping percolation flow in a partial melt region is of the
properties of melts and magmas. In this chapter, the most order of several meters or less per year and may be con-
critical properties will be briefly reviewed in the context of trasted with the explosive eruption of a volatile-rich magma
petrogenesis. Representative data are presented in the fig- at several hundred meters per second. Although rates of
ures, Tables 5.1 and 5.2, and 11 data compilations archived heat, mass, and momentum transfer are wildly different in
as Supplementary Material Tables S5.1e S5.11. These these systems or subsystems, knowledge regarding the

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 117

hundreds to thousands of kilometers deep may have exis-

TABLE 5.2 Representative Phonon Thermal Conductivity ted. Modeling this system demands knowledge of the
of Geosilicate Liquids and Glasses, and Olivine. Data properties of magma and phase equilibria for temperatures
Sources and More Exhaustive Data Compilations Are in the range 1000e5000 K and pressures at the Earth’s
Provided in Tables S5.8 and S5.9. Experimental surface (104 GPa ¼ 1 bar) to the coreemantle boundary
Measurements Are Given in Roman Type. Values in Italic (135 GPa). Thus, knowledge of the physical properties of
Type Are Determined From Recent MD Simulations magmas is essential to virtually all facets of magmatism
(Tikunoff and Spera, 2014). throughout Earth history at a wide variety of scales in space
and time, including the environmental impacts and hazards
Temperature Thermal
of volcanism on virtually all inhabitants of our planet. This
(
C), Conductivity
is the realm of Earth System Petrology. Because the
Composition Pressure (GPa)1 (W/m K)
terrestrial planets and minor bodies (asteroids) of the solar
Olivine (Fo90) 800 2.69 system have a similar origin, study of magmatism is also a
1400 2.18 key element in understanding solar system origin and early
evolution. It has been estimated that tens of billions of
Olivine (Fo100) 1400 1.59
terrestrial planets, that is, planets composed of metal and
Basalt (glass) 300 1.48 rock, exist in the Milky Way galaxy, one of about 300
Obsidian (rhyolite glass) 300 1.67 billion galaxies in the observable Universe. The properties
of high-temperature silicate melts and magmas are there-
Rhyolite (liquid) 800e1100 1.5
fore relevant to phenomena at cosmological scales
NaAlSi3O8 (glass) 800 1.56 throughout the past 13.8 billion years.
NaAlSi3O8 (liquid) 1200 1.59
1800 1.45
3. MAGMA THERMODYNAMIC
1800, 10 2.16
PROPERTIES
CaMgSi2O6 (glass) 800 1.46
Magma properties can be separated into two groupsd
CaMgSi2O6 (liquid) 1100 1.21
equilibrium thermodynamic quantities and transport prop-
1800 1.14 erties. Important equilibrium thermodynamic quantities
1800, 10 2.02 include density (r), heat capacity (CP), third law and
configurational entropies (S), enthalpies of formation
Mg2SiO4 (liquid) 3000 1.06
(DHform), fusion (DHfus), mixing (Hmix), and interfacial
3300 0.90 surface energy (s). Application of equilibrium thermody-
3300, 10 2.11 namic data are most useful in addressing the influence of
source bulk composition, pressure, and temperature on the
1 4
Pressure is 1 bar (10 GPa) unless noted. composition and amount of melt generated during partial
fusion, as well as the composition of solids and residual
melts during solidification of magma. Although chemical
thermodynamic and transport properties of magma is crit- and thermal equilibrium are not perfectly attained in nature,
ical in order to meaningfully analyze and ultimately predict the concept of local equilibrium is useful because equilib-
the relevant dynamics regardless of the particular scale. rium is often closely approximated at some scale and
Length scales relevant to magma transport and genesis because it represents a reference state from which de-
also vary over many orders of magnitude. At the smallest viations from equilibrium can be assessed. When chemical
scale, submicrons to millimeters, nucleation and growth of equilibrium is assumed, all the classical theory of chemical
crystals, or vapor bubbles depend on local fluctuations in thermodynamics can be applied and many times this
melt structure and the rates of mass diffusion. At larger greatly simplifies a problem and allows one to test and
scale, differentiation involves the physical separation of discriminate between competing hypotheses. For example,
crystals, melt, and vapor in conduits and chambers over at constant temperature and pressure, minimization of the
meters to kilometers, while the generation, transport, Gibbs free energy for a fixed bulk composition allows one
and ponding of magmas can extend over hundreds of to compute the composition and abundance of each phase
kilometers. Batholithic terrains, the product of multiple in the equilibrium assemblage. To compute the energetics
emplacement of many individual plutons have aerial ex- of melting, heat capacities and enthalpies of fusion of
tents on the order of thousands of square kilometers. In the appropriate phases are essential. Although not strictly an
early part of Earth history, a globe-encircling magma ocean equilibrium process, calculation of magma heat transport

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
118 PART | I Origin and Transport of Magma

requires heat capacity and fusion enthalpy data, in addition the melt. Although Vi depends weakly on composition, to a
to thermal diffusivities. The variation of melt density with good approximation, it may be taken independent of
temperature, pressure, and composition is needed for composition and as a function of temperature and pressure
analysis of momentum transport since buoyancy is a sig- for most petrological calculations. Partial molar isothermal
nificant factor driving the segregation, ascent, and eruption compressibilities for the major oxide components have
of magma. The differentiation of an emplaced magma body been determined from ultrasonic sound speed laboratory
by gravity-driven crystal fractionation critically depends on experiments. These parameters can be used to compute
the density difference between melt and newly formed melt density as a function of temperature, composition,
crystals. Like an iceberg in the ocean, there are conditions and pressure up to several GPa using a simple empirical
in temperature and pressure space where crystals grown in a EOS
melt float rather than sink. The final solidified state of such " !
a system is obviously quite dependent upon the equation of X vVi
VðT; P; XÞ ¼ Xi Vi;Tr þ ðT  Tr Þ
state (EOS)dthe relationship between density, tempera- vT
P
ture, and pressuredfor all relevant phases. For example, ! #
the compressibility of magma, defined according to vVi
þ ðP  Pr Þ (5.2)
bT ¼ (vlnr/vP)T, informs us about the variation of magma vP
T
density with pressure and is therefore important for con-
straining buoyancy force(s) driving magma ascent and the where Tr and Pr are reference conditions (generally
depths of ponding where the density contrast between 1400
C and 104 GPa (1 bar), respectively) and Vi is the
magma and host rock vanishes. Magma compressibility partial molar volume of the ith component. Parameters for
also is an important factor governing the explosivity of computing melt density as a function of pressure (up to
magma at or near the earth’s surface, especially for several GPa) and temperature is presented in Table S5.1.
degassing, volatile-rich magmas. These examples show Densities of some common natural melts at their respective
that there is a deep connection between the equilibrium approximate liquid temperatures are given in Table 5.1 (and
thermodynamic properties of magma and its petrogenetic Table S5.2) and shown graphically as a function of tem-
transport history. Without fundamental property data, it is perature and volatile content in Figures 5.1 and 5.2. As a
impossible to exploit the equations governing conservation rule of thumb, adding FeO, Fe2O3, MgO, TiO2, and CaO to
of energy, mass, species, and momentum to solve quanti- a melt increases its density, whereas adding alkalis (Li2O,
tatively magma transport problems. The equilibrium Na2O, and K2O) and volatiles (H2O, CO2) have the oppo-
chemical thermodynamic model provides a logical starting site effect. The temperature derivative of the partial molar
point for the analysis of magma production. volumes of SiO2 and Al2O3, are both effectively zero. As a
Natural silicate melts contain SiO2, Al2O3, MgO, CaO,
iron oxides, the alkalis (Na2O and K2O), P2O5, transition
metals, and rare earth elements, and minor but important
amounts of volatile compounds made up of HeOeCe
SeCleFeN. A complete thermodynamic description of
such a complex multicomponent system over the relevant
range in temperatureepressureecomposition space is
beyond the scope of this chapter (see Chapter 6dChemical
Thermodynamics and the Study of Magmas by M. Ghiorso
and G. Gualdadfor details). Many excellent reviews have
also been published on the thermodynamic properties of
silicate liquids including the relationship between structure
and properties listed. A few of these are listed in Further
Reading. We summarize the salient features here.

3.1. Density and EOS

The density of a silicate melt may be determined by eval-
uating the quotient
r ¼ SXi Mi =SXi Vi (5.1)
FIGURE 5.1 Melt density as a function of temperature for natural melts
where Xi is the mole fraction, Mi is the molar mass, and Vi spanning the compositional range rhyolite to komatiite at 1 bar (104 GPa)
is the partial molar volume of the ith oxide component in and 1 GPa pressure. All compositions are volatile-free.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 119

abundant anion in the silicate Earth, the major incorpo-

ration mechanism of H2O into nominally anhydrous sil-
icates and oxides, including melts, is as the hydroxyl
anion, OH. Almost all of the nominally anhydrous
minerals that compose the earth’s crust and mantle can
incorporate measurable amounts of “water.” The amount
of water (generally as OH) incorporated into nominally
anhydrous minerals generally increases with pressure and
sometimes with temperature, and is typically in the range
50e300 ppm in the dominant minerals of the earth’s
upper most mantle (olivine, pyroxenes, and garnet). The
solubility of water in the (Mg,Fe)2SiO4 polymorphs
wadsleyite and ringwoodite stable in the transition zone
(w410 to w660-km-depth) is markedly higher, and in
total it has been estimated that several ocean’s worth of
water may be present in Earth’s mantle. The recent discov-
ery of upward of 1 wt% water in ringwoodite encapsulated
in a natural diamond from a Brazilian kimberlite bolsters
FIGURE 5.2 Melt density as a function of dissolved water content for the claim that much of the water recycled into the deep
natural melts spanning the range rhyolite to komatiite. Temperatures for mantle is trapped by the transition zone. This has important
each composition are characteristic eruption temperatures. Densities are implications for the phase relations, dynamics, and magmas
calculated at a pressure of 1 bar (104 GPa).
formed by partial melting of the mantle.
The change in volume of a crystalline silicate upon
consequence, the density of a melt rich in silica and fusion under isobaric conditions is important because it
alumina is quite insensitive to temperature. Densities of influences the slope of the melting curve in pressuree
melts vary from 2500 to 2900 kg/m3 at magmatic temper- temperature space and therefore the generation of magma
atures (w1100
C) for silicic through intermediate to mafic by partial fusion. For small degrees of partial melting under
and ultramafic compositions at low pressure. Density in- nearly isochoric conditions, melt pressure can rise above
creases as pressure rises, although increasing the concen- lithostatic values and the small melt overpressure can drive
tration of the alkaline earth metals, Ca, and Mg, serves to magma fracture and vertical transport. Additionally, melt
decrease the pressure dependence of melt density. The al- density influences the separation rate of melt from its
kali metals and water are relatively compressible and source. Volume changes accompanying melting of some
contribute significantly to the pressure dependence of silicates are given in Table S5.2. Typically, the molar vol-
density. It is important to note that small differences in ume of a crystalline phase increases by about 10% upon
composition can compensate for relatively large differ- fusion although silica (like ice) may shrink (i.e., become
ences in temperature because of the strong dependence of more dense) when fused at low pressure. Less is known
melt density on composition. Whereas a typical value for regarding densityetemperatureepressure relations or EOS
the isobaric expansivity is aP ¼ (vlnr/vT)P z 5  105/K, for natural carbonatite melts. Some data pertaining to the
the analogous quantity describing the variation of density density and viscosity of simple carbonate liquids as well as
with composition is of order 0.1 to 3 depending on the supercritical H2O are given in Table S5.3. Relative to
component. The volatile components H2O and CO2 have common silicate melts, carbonatite liquids are both less
an especially large effect on melt density. The partial dense (by about 30%) and less viscous by several orders of
specific density (Mi/Vi) for CO2 and H2O in natural melts magnitude compared to mafic melts and up to 10 orders of
are w1400 and 900 kg/m3, respectively, for typical magnitude compared to silicic ones.
crustal magmas. For comparison, Mi/Vi for K2O, MgO, At pressures corresponding to depths greater than about
FeO, and SiO2 are w2000, 3500, 5300, and 2200 kg/m3, 100 km (>3 GPa), the density of melts at superliquidus
respectively. Thus, a small amount of dissolved water temperatures can be determined from high-pressure static
dramatically lowers melt density and significantly affects (sinkefloat) and shock wave experiments. More recently, a
transport and thermodynamic melt properties such as vast array of spectroscopic tools, including X-ray absorp-
viscosity, mass diffusivities, thermal conductivity, and tion, tomography and diffraction/scattering, and ultrasonic
isobaric heat capacity. The temperature at which rocks measurements in conjunction with static high-pressure
begin to melt (the solidus temperature) is substantially conditions are being employed to augment data obtained
lowered when volatile constituents such as H2O are by more traditional approachesdin some cases accessing
present. Because oxygen is volumetrically by far the most subliquidus conditions more directly applicable to magma

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
120 PART | I Origin and Transport of Magma

genesis and differentiation. First-principles or classical MgO-rich komatiitic melts, probably the most common
molecular dynamics (MD) simulations are increasingly magma type erupted in the first billion or so years of earth
being utilized as a complement to experimental measure- history, become denser than olivine crystals at pressure
ments, and enable predictions beyond experimental reach. corresponding to depths of roughly 250e400 km. This has
A useful quasi-theoretical relationship is the important implications for differentiation of the silicate
BircheMurnaghan EOS that relates liquid (of fixed portion of Earth including internal mineralogical layering
composition) density to pressure and temperature accord- as well as possible chemical stratification. Moreover, the
ing to: compressibility of silicate liquids also bears on the dynamic
  stability of melts in the mantle, and the interpretation of low
3   7=3   5=3
P ¼ KT r rT;0  r rT;0 seismic velocity regions there. Models calling on the
2 presence of (neutrally buoyant) melt immediately above the
  (5.3)
3    coreemantle boundary (ultra low-velocity zone), at the top
1  ð4  K0 Þ r rT;0
2=3
1 of the transition zone (w410 km), and at a depth of about
4
220  30 km (Lehmann discontinuity) depend critically on
where rT,0 is the density of melt at 104 GPa and temper- the details of how silicate melts densify as pressure in-
ature T, KT is the isothermal bulk modulus (KT ¼ 1/bT) at creases. This is an area of very active research.
1 bar (104 GPa), K0 is the pressure derivative of KT, and r
is the melt density at pressure P and temperature T. Many
3.2. Enthalpy, Entropy, and Heat Capacity
other equations of state have been proposed and find
application in geoliquids. Compressibility data for silicate In addition to volumetric or EOS data, the calorimetric
melts are presented in Table S5.4. Normalized densitye properties of high temperature and molten silicates are
pressure relations for silicate liquids ranging from felsic to needed to analyze magma transport problems. These
ultramafic compositions are shown in Figure 5.3. Poly- properties inform us regarding the internal energy of melts
merized rhyolite liquid possessing an open network struc- and crystals and how internal energy and other closely
ture is vastly more compressibility than basalt or andesite, related thermodynamic functions change with temperature.
while highly depolymerized and compact melts such as These properties include the enthalpy, entropy, and heat
komatiite and peridotite are the least compressible. An capacity and are inextricably bound to the thermal evolu-
interesting application of high-pressure EOS studies is that tion of magma and relevant to processes such as partial
melting, solidification, and the advective transport of heat.
Melting temperatures, enthalpies of fusion, entropies of
fusion, and specific heats of fusion for some common
phases are listed in Table S5.5. The specific enthalpy of
fusion is the heat per unit mass needed at a reference
pressure (generally 104 GPa or 1 bar) to transform a
crystal or crystalline assemblage to the liquid state. There is
a wide variation in the specific enthalpy of fusion from 100
to 300 kJ/kg for silica polymorphs, albite, and sanidine, all
important components in crustal-derived melts, to
w1000 kJ/kg for refractory phases, such as forsterite, py-
rope, enstatite, and transition metal oxides relevant to mafic
and ultramafic compositions. Table 5.1 provides estimates
of fusion enthalpies for common compositions computed
using data from Table S5.5 and typical modal abundances.
There is a factor of three difference in the specific heat of
fusion for a model granite (w200 kJ/kg) compared to an
ultramafic composition (w600 kJ/kg). The more refractory
nature of the ultramafic composition is a reflection of the
higher fusion enthalpy of olivine and the pyroxenes relative
to quartz, alkali feldspar, and plagioclase. Garnet also has a
high specific fusion enthalpy. Melting eclogite (pyroxene
plus garnet) at high pressure requires more heat (570 kJ/kg)
FIGURE 5.3 Normalized density (r/ro) at 2000 K as a function of
pressure for natural silicate liquids ranging in composition from rhyolite to than a corresponding low-pressure plagioclase-pyroxene
peridotite computed using Eqn (5.3) and parameters from Table S5.4, gabbro (w400 kJ/kg). The relatively low fusion enthalpy
assuming an average dKT/dT ¼ 0.022. for mineral phases comprising continental crust implies

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 121

that anatexis of crust by heat exchange between mafic (OH)1 is w13,000 J/kg K, while water dissolved in its
magma and its surroundings is thermally efficient. It is molecular form (H2O) has a partial isobaric heat capacity of
worth noting that the heat required to completely melt w4700 J/kg K. For comparison, cp for ambient tap water is
Earth’s mantle w3  1030 J, is less than 10% of the kinetic 4184 J/kg K. Likewise, the isobaric heat capacity of silica
energy delivered to Earth by impact of a Mars-sized body glass increases by w30% with the addition of 1200 ppm
(15% mass of Earth) with an impact velocity equal to the OH, while adding 2 wt% H2O to an otherwise anhydrous
Earth’s escape velocity of 11.2 km/s. If, in fact, the earth granitic melt will increase cp by 20%. Thus, the addition or
did possess a magma ocean in the Hadean, then its lifetime loss of magmatic volatiles (namely water) can have sig-
would have been controlled by the rate at which w1030 J of nificant enthalpic effects that, in turn, will influence ascent
heat could be dissipated by conduction, radiation, and rates, eruptibility, solidification timescales, and countless
convection to space. other magmatic processes.
In addition to heat effects associated with solid to liquid
phase changes, it is important to consider sensible heat
effects or the variation of the enthalpy of the melt with 4. MAGMA TRANSPORT PROPERTIES
temperature. This is measured by the molar isobaric heat Transport properties of magma play a crucial role in
capacity, Cp. For most aluminosilicate glasses and melts of development of petrogenetic theory. Simple consideration
petrologic significance, Cp can be approximated as an ad- of the various stages of magma transport from source to
ditive function of oxide component partial molar isobaric surface makes this abundantly clear. Some examples
heat capacities. As temperature increases for glasses, an include the generation of magma by partial melting, melt
increased number of vibrational modes become excited and segregation by porous or channel flow, and magma ascent
so the isochoric heat capacity CV hðvE vTÞV increases toward by viscous flow in magma-filled propagating cracks. Other
the high temperature DulongePetit limit as vibrational examples include differentiation of magma by crystal
modes become saturated. The isobaric heat capacity is fractionation and the processes of magma mixing and
related to the isochoric one by the expression crustal anatexis. In all these phenomena, significant
Ta2
CP ¼ CV þ rbPT . Parameters for the common oxide com- amounts of momentum, heat, and chemical component
ponents to estimate the isobaric heat capacity of a glass as a transport occur. Transport of momentum, heat, and chem-
function of composition and temperature at low pressure ical species by magmas involves the material properties of
are listed in Table S5.6. At the glass transition temperature viscosity, thermal conductivity, and mass diffusivity,
(Tg), there is generally a discontinuous jump in isobaric respectively. Although data and models for the composition
heat capacity. The magnitude of this jump depends on the and temperature dependence of melt viscosity are avail-
atomic structure of the liquid. Highly polymerized able, less is known regarding the quantitative rheological
meltsdso-called, strong liquidsdexhibit little or no properties of magmatic mixtures. There are relatively few
discontinuity in heat capacity at Tg, while depolymerized, experimental data bearing on the thermal conductivity of
i.e., fragile, melts do. For petrologic purposes, it is suffi- melts at high temperatures; this is perhaps the most un-
cient to note that the jump in Cp at the glass transition certain of all magma transport properties but one with
temperature is small for silica-rich liquids (e.g., rhyolitic significance with respect to the thermal history of the Earth.
melts) and it becomes larger for more mafic liquids Chemical, self, and tracer diffusivity data exist and corre-
(andesitic and basaltic melts). lations have been developed for estimating diffusivities as a
For silicate melts (unlike silicate glasses), the temper- function of species charge and size for melts of various
ature dependence of Cp can generally be neglected unless temperature and composition, and treating more compli-
great accuracy is required. Parameters for estimating Cp for cated multicomponent diffusion and associated isotopic
melts as a function of composition at magmatic tempera- effects. The effects of pressure are not well understood, but
tures are presented in Table S5.7. Table 5.1 gives specific systematics are emerging for network former and network-
isobaric heat capacities (cp) for some common naturally modifying cations exhibiting different pressure de-
occurring anhydrous silicate melts. Silicic anhydrous melts pendencies. A brief review of these properties is presented
have cp around 1300e1400 J/kg K, whereas melts with below. The original references should be consulted for
higher MgO contents have specific heats around more detail.
1600e1700 J/kg K. Because the advective transport of heat
per unit volume is given by rcpDT magmas are good
transporters of mantle heat to crustal levels since DT is of
4.1. Melt Viscosity
order 100e1000 K. Magmatic water content can also have The dynamic viscosity (h) of magma is the material
a large effect on the heat capacity of magmas. The partial property that connects the shear stress in the fluid to the rate
isobaric heat capacity of water dissolved as hydroxyl of strain. The viscosity determines the rate of diffusion of

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
122 PART | I Origin and Transport of Magma

momentum. More precisely, the kinematic viscosity, The relationship between liquid structure at the atomic
defined v ¼ hr, measures the tendency of a melt or magma level and macroscopic properties (both transport and ther-
to diffuse velocity gradients. Critical issues include the modynamic) is an essential keystone of magma transport
effect of composition, pressure, and temperature on melt phenomena. The activation volume, a measure of the
viscosity, as well as the rheological properties of magmatic pressure dependence of viscosity, is usually a small fraction
suspensions. Although to a good approximation silicate of the molar volume of the melt and changes sign, from
liquids well above the glass transition temperature behave negative to positive, as the fraction of NBO increases and in
as Newtonian fluids (that is, they show a linear relationship the case of polymerized melts pressure increases. A typical
between shear rate and shear stress at fixed temperature and value for VA for a network tetrahedral melt such as NaAl-
pressure), this is not the case for magmatic multiphase Si3O8 at low pressure is around 6  106 m3/mol
suspensions that generally exhibit complex rheological (6 cm3/mol), whereas for more depolymerized melt
properties or for natural glasses that are described as (e.g., CaMgSi2O8) VA is around 3  106 m3/mol (þ3 cm3/
viscoelastic materials. mol). Melts with a negative value of VA show a decrease in
The temperature dependence of melt viscosity can be viscosity as pressure increasesdthe opposite is true for
described by several models. The simplest is the Arrhenian melts with positive VA.
model for a melt of fixed composition Fully polymerized network melts, for which essentially
all the oxygen is BO typically possess intermediate range
hðp; TÞ ¼ ho exp½ðEA þ PVA Þ=RT (5.4)
order defined by the formation of n-membered rings
In expression (5.4), ho is the asymptotic viscosity as (n ¼ 4e10) of tetrahedra at low pressure. As pressure in-
T / N, EA is the activation energy for viscous flow, VA is creases, the ring structure collapses and the “anomalous”
the activation volume for viscous flow, and R is the uni- effect of viscosity (VA) diminishes to a point that free-
versal gas constant. The activation energy and volume are volume effects dominate and viscosity increases with
constants for melt of fixed composition. The preexponential pressure (þVA). Elevated pressure also drives oxygen into
term is approximately a “universal” constant that varies higher (five and sixfold) coordination with Si and Al and
quite weakly with composition and is in the range 104 to such changes in the network structure also have been linked
105 Pa s. An optimal value of h z 104.6 Pa is found the change from anomalous to normal pressure dependence
empirically for melts in the range of natural compositions. of viscosity.
Activation energies vary systematically with composition Not all silicate melts follow the Arrhenian temper-
from values around 200 kJ/mol for mafic and ultramafic atureeviscosity relationship. This is especially true for
melts to 400 kJ/mol for more silicic compositions. fragile liquids, e.g., melts containing a high proportion
Increasing the concentration of nonframework components of NBO. An empirical relationship for predicting the
such as the alkalis, the alkaline earths, and especially H2O temperature dependence of viscosity data in fragile non-
results in creation of oxygens with only a single nearest Arrhenian melts at 1-bar pressure is the so-called
neighbor of Si or Al. The presence of these nonbridging TammanneVogeleFulcher (TVF) expression
oxygens (NBOs) destroys the network structure created by
h ¼ hO exp½B=ðT  TK Þ (5.5)
the linkage of tetrahedra to form 3 to 6þ membered rings at
intermediate atomic scales of order 0.5e1 nm. For the ideal where B and TK are functions of composition but not tem-
network tetrahedral melt, each oxygen is bound to two Si, perature. TK is a constant closely related to both the glass
two Al or one Si, and one Al simultaneously. The arche- transition temperature and the Kauzmann catastrophe tem-
typal example is molten silica in which each oxygen has perature. At T < TK, the entropy of supercooled liquid
two nearest neighbors of silicon and each silicon is sur- computed by extrapolation of high-temperature data is less
rounded by four nearest neighbors of oxygen. The increase than that of its corresponding crystal (the Kauzmann
in the concentration of NBO by addition of network mod- Paradox). The computed TVF viscosity (Eqn (5.5)) explodes
ifiers (e.g., H2O) lowers EA. Small (several percent by toward an infinite singularity as T / TK. The relationship
mass) concentrations of water have a dramatic effect in between the TVF and the Arrhenian model is found by
lowering the viscosity of silicate liquids because water is examining the temperature derivative of the viscosity. In the
89% oxygen (by mass) unlike any other common oxide Arrhenian model, vlnhArr =vð1=TÞ ¼ EA =R whereas for a
component. This is illustrated in Table 5.1 for the granitic TVF fluid vlnhTVF =vð1=TÞ ¼ BT2 =ðT  TK Þ2 . Thus, in
composition where the addition of 2 wt% H2O to a granitic the limit TK/ 0, the TVF and the Arrhenian models are
melt lowers melt viscosity by a factor of 105. H2O also identical with B ¼ EA/R. In general, TK is close to but
tends to lower the viscosity of more mafic compositions, usually somewhat less than Tg, the glass transition temper-
although the effect is less dramatic because fewer bridging ature. Typically, Tg for geosilicate compositions is
oxygen (BO) exist in anhydrous mafic melts. w1000 K. Tg for pure silica is somewhat higher (1500 K).

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 123

The TVF model therefore shows that the variation in melt

viscosity with reciprocal temperature decreases as temper-
ature increases and is not a constant (EA) as in the Arrhenian
model. The TVF empirical model draws some theoretical
foundation from the AdamseGibbseDiMarzio (AGD)
configurational entropy model. In AGD, viscosity depends
on temperature according to
   
h ¼ ho exp D TSconf T (5.6a)
where Sconf is the configurational entropy of the melt. Sconf
includes a contribution computed from liquid and glass
isobaric heat capacity data, as well as the residual entropy
of glass (frozen liquid) at the glass transition temperature,
Tg. The value of Sconf(T) is computed from

ZT  
DCp T
Sconf ðTÞ ¼ dT0 (5.6b)
T0 FIGURE 5.4 Viscosity as a function of temperature at 1 bar (104 GPa)
TK
for natural melts spanning the compositional range rhyolite to komatiite.
All compositions are volatile-free. The temperature range is illustrative of
where the heat capacity term in Eqn (5.6b) is the experi-
typical eruption temperatures for each composition.
mentally observed difference in liquid and solid heat
capacities. The parameter D is temperature independent but
varies with composition. An important aspect of the AGD
theory is that it connects thermodynamic and transport
properties. It is based on a theory of cooperative behavior
during dynamic rearrangement of atomic configurations in
the liquid state. The AGD theory collapses to the empirical
TVF expression provided the heat capacity difference is
given by the hyperbolic expression DCP ðTÞ ¼ CT T in Eqn
K

(5.6b). Although this identification is one of several pos-

sibilities, the collapse of the AGD expression to the TVF
provides some theoretical support for the validity of the
otherwise empirical TVF formulation.
An empirical model that gives melt viscosity as a function
of composition and temperature based on the TVF formu-
lation has been calibrated and in wide use (Giordano et al.,
2008). This empirical model gives the added benefit of
providing estimates of the glass transition temperature and
melt fragility as a function of melt composition at low
pressure. Pressure effects are not considered in the present FIGURE 5.5 Melt viscosity as a function of dissolved water content for
form of this model, however. The Arrhenian pressure natural melts spanning the range rhyolite to komatiite. Temperatures for
correction can be used to roughly account for pressure by each composition are characteristic eruption temperatures. Viscosities are
calculated for pressure equal to 1 bar (104 GPa). For the rhyolite
simple extension of the TVF. For less precise work, the
composition, two different models are shown. Note the very dramatic ef-
Arrhenian model still finds application because it is so simple fect of dissolved water on the viscosity of natural melts. Viscosity models
to implement, although extrapolating beyond the bounds of are from Shaw (1972) and Giordano et al. (2008).
the experimental measurements is risky at best. The viscosity
of naturally occurring silicate melts can span over 10 orders
of magnitude due to variations in composition and temper-
4.2. Magma Viscosity
ature (Table 5.1 and Figure 5.4). Dissolved water is espe- Far fewer data exist on the rheological properties of
cially effective in lowering the viscosity of polymerized magmatic suspensions compared to silicate melts. Magma,
melts (e.g., three orders of magnitude with addition of 3 wt a mixture of suspended crystals and vapor bubbles in a melt
H2O for rhyolite melt), while it has comparative small effects matrix, may be expected to exhibit complex rheological
on the viscosity of depolymerized (mafic and ultramafic) behavior when significant amounts of suspended crystals
liquids (see Figure 5.5). and bubbles (“particles”) are present. Silicate foams,

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
124 PART | I Origin and Transport of Magma

defined as concentrated emulsions with porosity greater tends to reduce the viscosity of a magma at a fixed f
than about 60 volume percent (f > 0.6) are found in small because smaller particles can fit in between the larger ones.
quantities at virtually all volcanic centers because magma The viscosity of a unimodal crystal-bearing magma with
typically contains volatiles (0.1e10 wt%) and the common f ¼ 0.3 is about 10 times higher than its value for the melt
volatiles (H2O and CO2) are practically insoluble in melts phase alone based on Eqn (5.7). In more detail, it is both the
at low pressure. Magma decompression during magma size and shape variation of individual crystals, as well as the
ascent commonly leads to volatile saturation. The impor- overall structure of the mixture, that strongly controls
tant variables governing the viscosity of multiphase rheological properties. An alternative model that purport-
mixtures include the shear rate, the volume fraction of solid edly accounts for the presence of melt entrapped between
and bubbles, the size and shape distributions of these particles gives smaller relative viscosities compared to Eqn
“particles” (solids and bubbles), temperature, pressure, and (5.7). The expression is
melt viscosity. There are no experiments exploring the  
rheological properties that account for these variables h f C1 fo
hr ¼ mix ¼ 1 (5.8)
collectively and comprehensively although some effects, hmelt fo
such as the effects of variable solid loading fractions, have
been explored. A complication is that in natural systems the where the C1 equals 2.5 and fo represents the random close
flow itself influences the distribution of particles via me- packing limit. The parameters C1 and fo vary depending on
chanical (sorting) and thermal effects. For example, viscous the particle size and shape distributions. For nonspherical
forces acting on particles can cause their movement relative particles, C1 is larger than the Einstein value of 5/2, used in
to melt and each other, e.g., flow differentiation. A het- Eqn (5.7). For high shear rates, some particle arrangements
erogeneous distribution of particles can evolve from an can form where spherical particles tend to form clusters and
initially homogeneous distribution under the influence of the mixture is capable of flow at volume fraction that ex-
particle-melt viscous drag and pressure forces. Similarly, ceeds the zero shear rate limit fo z 0.64. Presently, the
thermal gradients can effect the local concentration of most accurate model for a unimodal, solid þ melt sus-
phenocrysts thereby modifying velocity gradients. The pension at zero shear rate is
rheology of multiphase fluids is complex and currently the !C1 fo
subject of intensive study. Fortunately, there has been some h 1f
1fo

progress on two-phase mixtures (both crystal-melt sus- hr ¼ mix ¼ (5.9)

hmelt 1  ff
pensions and melt-vapor emulsions) applicable to o

magmatic systems. These studies enable one to approxi-

mate the viscosity of magmatic suspensions when Values for C1 and fo suitable for magma rheology cal-
suspensions are composed of monosized “particles” in the culations accounting for nonspherical crystals and the close
creeping flow regime at very low rates of shear. These pack limit are 3 and 0.64, respectively. Some typical rela-
models are valid from the dilute limit to the random tive suspension viscosities as a function of solid volume
packing volume fraction limit of f z 0.64. fraction are shown in Figure 5.6 based on Eqns (5.7)e(5.9)
To first order, the effect of increasing the volume frac- for creeping flow (low shear rate) of a suspension of rigid
tion of crystals, f, in a melt is to increase the relative spherical monodisperse particles with the packing limit
viscosity, hr, according to fo ¼ 0.64. Also shown is an approximation to the more
realistic case of inequant particles based on Eqn (5.9) with
hr ¼ hmix =hmelt ¼ ð1  f=fo Þ5=2 (5.7)
C1 ¼ 3. At high-volume fraction solids, these various
where hmix is the viscosity of the crystaleliquid suspension models exhibit order of magnitude differences in the
or mixture, hmelt is the viscosity of aphyric melt, and fo is relative viscositydsmall variations in C1 and fo amplify
the maximum packing fraction of the solid. Equation (5.7) these differences.
shows that the loading level of crystals is limited by fo, the In noncreeping flows, a nonlinear relation exists between
geometric limit for contact of solids in the most efficient the shear stress sustained by the mixture and the shear rate.
packing arrangement. For rhombohedral packing of Such suspensions exhibit power-law behavior. On a plot of
spherical particles, fo ¼ 0.74 whereas for simple cubic shear stress versus shear rate, the tangent at a given shear
packing and body-centered cubic packing fo ¼ 0.52 and rate is the effective viscosity of the suspension at that shear
0.60, respectively. In practice, random packing of spheres rate. Because the effective viscosity is a monotonically
generally gives fo in the range 0.60e0.65. Equation (5.7) decreasing function of shear rate at fixed temperature and
shows that the volume fraction and the maximum packing solid fraction, a value for the yield stress can sometimes be
fraction fo of solids are the two important parameters approximated by extrapolation of the tangent to the locus of
governing the viscous response of magma. Polydispersity, points in the shear stresseshear rate plane to the stress axis at
the presence of a nonunimodal distribution of particles, zero shear rate. More experimental work is required to

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 125

increasing bubble content. Natural systems span the range

from Ca < 1 to Ca[1 so it is important to consider the
dynamic regime when considering the effect of bubbles on
the viscosity of magma. Relative viscosity can drop by a
factor of 10 as the bubble content increases from 0 to w50
volume percent in the high Ca regime.
Self-organization of dispersed particles (solids or bub-
bles) to form structured mixtures can give rise to large
spontaneous changes in viscosity during flow. The forma-
tion of structure has been observed in laboratory models
and probably quite relevant during flow of crystal slurries
or vesiculated magma in chambers and conduits. Further
experimental, analytical, and numerical work is needed to
quantify the rheology of magmatic crystal-melt-vapor
suspensions.

FIGURE 5.6 Relative viscosity of magmatic suspensions containing

solids as a function of volume fraction of solids. The models refer to Eqns
4.3. Thermal Conductivity
(5.7)e(5.9). Note the congruity of models for low crystal loading volume Transport of magmatic heat occurs by several mechanisms
fraction and the divergence as the close packing limit of w0.64 is
approached. Relative viscosity models from Brouwers (2010). The coef-
including convection (heat transported by bulk flow),
ficient C1 ¼ 3 allows for particle nonsphericity, approximately based on phonon conduction (heat transported by atomic vibration),
Eqn (5.9). and radiation (an electromagnetic phenomenon involving
photon transfer). Radiation travels through a vacuum at the
speed of light; most gases are transparent to radiation.
develop quantitative expressions that relate volume fraction Radiative transport of heat may be important in transition
solid to the apparent yield stress. In the meantime, Eqn (5.9) metal poor melts due to their relative transparency. Since
may be used as an approximation keeping in mind its most gases are transparent, radiative heat transfer may also
inherent limitation to low shear rate flows. be important across bubbles in magmatic emulsions and
In bubble-melt mixtures, l viscosity can be either an foams. In contrast, radiative heat transport can generally be
increasing or decreasing function of the volume fraction of ignored in mafic melts containing significant amounts of
the low-viscosity phase (f) depending on the rate at which transition metals (e.g., Fe, Ti, Ni) because such melts are
_ bubbles act
the mixture is sheared. At low rates of shear (g), relatively opaque to thermal radiation, i.e., the mean free
as nondeformable inclusions and hr increases with photon path is relatively shortdon the order of millimeters.
increasing f similar to when solids (e.g., crystals) are The photon (radiative) conductivity (kR) can be approxi-
added to melt. In distinction, at high shear rate, low- mated in this case as
viscosity bubbles readily deform and the mixture
16 2 3
viscosity decreases with increasing bubble fraction. A krad ¼ sn T L (5.10)
dimensionless parameter termed the capillary number (Ca) 3
is useful in determining the appropriate rheodynamic where s is the StefaneBoltzmann constant (5.67  108 J/
_ b
regime. The capillary number is defined Ca ¼ hgr s where m2 K4 s), n is the index of refraction, and L is the mean free
_ rb, and s represent the melt viscosity, shear rate,
h, g, path for photons. The thermal transfer by photons depends
bubble radius, and melt-vapor interfacial tension, respec- on L as well as the emissivities of the surfaces across which
tively. The capillary number can be thought of as the ratio heat is being transferred. When L w 0 (as in an opaque
of viscous tractions acting on the boundaries of a bubble melt) or when one is concerned with transport of heat
that distort it from a spherical shape relative to interfacial across distances much larger than L, radiative transfer is
surface forces that tends to preserve its spherical shape. negligible. It is only when L is relatively large, as in
Small values of Ca correspond to conditions where surface transparent high-silica melts or in transparent vapor-melt
tension dominates and bubbles retain their spherical shapes. emulsions that photon conduction may become appre-
This is favored in finely dispersed (small bubble size), low- ciable. Because geometric factors are often relevant, the
viscosity emulsions subjected to low rates of deformation. importance of radiative heat transfer in geological pro-
One expects a relative viscosity relation of the form cesses (as opposed to laboratory small-scale experiments)
hr ¼ f(f,Ca) such that for small capillary number (roughly should be carefully considered.
Ca < 1), hr is an increasing function of f. In contrast, for The phonon thermal conductivity (k) provides a quan-
regimes with Ca > 1, the relative viscosity decreases with titative measure of the importance of phonon heat

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
126 PART | I Origin and Transport of Magma

conduction in solids and melts. The thermal diffusivity (k) relationships have important implications for the efficiency
defined k ¼ k/rCp is the relevant parameter in transient heat of heat transfer associated with molten regions of the
conduction problems. The thermal diffusivity involves a mantle.
combination of thermodynamic properties and the phonon
thermal conductivity. Quantized thermal waves called
phonons carry heat in silicate crystals, glasses, and melts.
4.4. Diffusion
Thermal resistivity (W), which is inversely proportional to Mass is transported in magmatic systems by bulk flow
thermal conductivity (i.e., k f W1), arises due to both (advection) and by diffusion. Their operation separately
phononephonon interaction and structural disorder. and collectively is responsible for convection in magma
Because disorder is an outstanding characteristic of the bodies. There is an extensive body of information per-
glassy and molten state, one may expect the mean free path taining to diffusion in silicate melts and glasses as a
for the dominant thermal phonons to approximately equal function of temperature, composition, and, to a lesser
the scale of structural disorder, roughly 0.3e0.6 nm in a extent, pressure. This is due in large measure to myriad
typical silicate glass or melt. The thermal conductivity of industrial and materials processing applications of diffu-
a solid or melt may be estimated according to the relation sion as well as the importance of diffusion in petrologic
processes such as crystal nucleation and growth, the growth
1 of vapor bubbles, the homogenization of melt inclusions,
k ¼ rCV cL (5.11) the resetting of geochronological clocks, and the mixing of
3
magmas. There is a deep connection between the atomic
where L is the phonon mean free path length, c is the sonic structure of a melt or glass and the mobility of its constit-
velocity (several kilometers per second), r is the melt uents. Hence study of the systematics of diffusion bridges
density and CV is the isochoric specific heat capacity. the gap between the microscopic and macroscopic realms
Because the structure of a melt or glass is sensitive to relevant to magma transport phenomena. Several excellent
pressure, one might anticipate the mean free phonon length reviews of diffusion for both industrial and geological
and hence k (other factors constant) to increase with materials are available and listed in Further Reading. In this
pressure. In general, for a silicate liquid k decreases as section, the types of diffusion important for magmatic
temperature increases but increases as pressure increases, systems are reviewed and typical values are given to indi-
consistent with expectations (see Table 5.2). cate the main trends and orders of magnitude. We review
Because radiative transfer tends to dominate at high theory and highlight applications of these data. Since we
temperatures, it can be difficult to cleanly separate the ef- are mainly concerned with magmatic systems discussion is
fects of phonon conduction from radiative transfer in lab- limited to properties for the liquid state, i.e., temperatures
oratory experiments. There are surprisingly few reliable above the glass transition (h < 1012 Pa s).
measurements of the thermal conductivity of liquid Diffusion involves the motion of different constituents
silicates, fewer for geochemically important compositions, at the molecular or atomic scale leading to local (inter)
and virtually no laboratory data regarding the effects of mixing or net transport of mass. From an atomistic stand-
pressure on phonon conduction. Fortunately, within the past point, diffusion is the consequence of the random walk of
few years, there have been some important advances. particles, often thermally activated, that depends on tem-
Values for the phonon thermal conductivity of some geo- perature, the size and charge density of the diffusing
silicate liquids, glasses, and crystals for a range of tem- species, and the viscosity of the surrounding medium.
peratures and pressures are given in Table 5.2 (Tables S5.8 Motion by random walk determines the intrinsic mobility
and S5.9 provide more exhaustive compilations). These of the diffusing species and is described by the self-
data show that as temperature increases, the thermal con- diffusion coefficient (D*). The pathways for random walk
ductivity typically decreases for both solids and liquids. (self-diffusion) cannot be measured directly in the labora-
Values if k for silicate liquids lie in the range 0.8e1.6 W/ tory; however, approximate values can be obtained by
m K with k generally higher from more polymerized liq- tracking the rate of transfer of readily identifiable compo-
uids. Finally, very recently, phonon conductivities for nents that are either chemically indistinguishable, i.e.,
molten NaAlSi3O8, CaMgSi2O6, and Mg2SiO4 have been different isotopes of a given element, or elements that are in
determined from MD simulations in the range such dilute concentration that their presence has no
2000e5000 K and 0e30 GPa (Tables 5.2 and S5.9). Values resolvable effect on the concentration of other components.
for molten albite and diopside at 0 GPa agree well with The term intradiffusion is used to describe the bidirectional
laboratory measurements at 1 bar, and confirm a weak exchange of isotopes of major, minor, or trace elements in
negative temperature dependence of k at constant pressure. an otherwise homogeneous system, while the term tracer
They further show that at constant temperature, k diffusion is reserved for cases where a trace element (or one
increases rather significantly as pressure increases. These of its isotopes) diffuses solely down its own concentration

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 127

gradient. Both are commonly assumed to be equal mea- it is not likely that this approach will have much practical
sures of self-diffusion provided (1) a constant activity utility for modeling natural magmatic systems ðn[4Þ
coefficientdwhich in the case of trace diffusion means the anytime soon.
tracer obeys Henry’s law (the concentration limits for An alternative and much simplified approach is to treat
Henrian behavior must be determined empirically, but is chemical diffusion in magmas as an effective binary pro-
commonly <1000 part per million), (2) pathways for cess. This is possible when chiefly the component of interest
diffusion are the same as for random walk, and (3) there are varies in concentration relative to all other components in
no significant mass-dependent (isotope) effects. There are the system. In the limit of Henry’s law this is tracer diffu-
instances where these essential requirements are not met sion. However, once the concentration exceeds the Henrian
and here diffusion coefficients provide at best semi- limit, the flux is no longer controlled solely by the magni-
quantitative constraints on intrinsic mobility in natural tude of the concentration gradient, but in how the activity of
systems only in so far as they mimic laboratory conditions. the species changes along that gradientdthis is the domain
More commonly in natural systems diffusion arises for chemical diffusion. The effective binary diffusion
from and occurs in the presence of chemical (potential) approach has also been used to quantify interdiffusion be-
gradients due to non-Henrian behavior of the tracer or tween vastly different magmas (such as basalt and rhyolite)
concentration gradients of other components at some where often the diffusion of SiO2 (Si and O being the
location and time. This is referred to as chemical diffusion slowest diffusing elements) is considered relative to all the
and is distinguished by simultaneous fluxes of components. other components or where all the network formers (Si, Al,
From a phenomenological perspective, the chemical flux of Feþ3, etc.) are lumped together as one component and all
a given component can be described by Fick’s first law the network modifiers (Feþ2, Mg, Ca, Na, and K in excess of
relating the net flux of that component to a negative that needed to charge balance tetrahedral Al, Feþ3, etc.) are
gradient in its chemical potential or activity. Where combined as the other. While again the diffusivities deter-
chemical diffusion involves just two components of a mined by this approach are only applicable to natural
multicomponent system a simple binary solution to Fick’s systems in so far as the laboratory conditions simulate na-
law is sufficient; however, if there are net displacements of ture, effective binary diffusivities especially for SiO2 or
more than two components a full description of chemical network formers do place limits on the length and time-
diffusion requires a (n e 1) by (n e 1) diffusion matrix with scales for chemical homogenization during magma com-
elements Dij. That is, for an n-component system, the flux mingling. This enables us to evaluate how species with high
of the ith component depends on the chemical potential intrinsic mobility are locally redistributed by chemical
gradient of any independent set of n e 1 components and diffusion within the more sustained major element gradient
not simply the ith component. Because the chemical due to sluggish diffusion of network-forming cations and
diffusion rate of the ith component depends on n e 1 oxygen (Figure 5.7). Consider the diffusive length during
chemical potential gradients whereas isotopic equilibration the 105 year solidification time of a 103 km3 gabbroic pluton
of the ith component depends on the self-diffusivity of that (characteristic lengthw10 km). For O and Si, diffusion can
species, decoupling between elemental and isotopic con- effect changes in concentration over w3 m, while in the
centrations is possible. Furthermore, due to nonideality case of Li and H2O having diffusivities two to three orders
and/or diffusive coupling effects, even for ideal solutions, of magnitude larger diffusive exchange can occur over
chemical diffusion can involve a net flux of elements up, w60 m. Possible ramifications for isotopic composition of
rather than down, their concentration gradients. This is fast and slow diffusion species such as Sr and Nd, respec-
referred to as uphill diffusion and is well documented in tively, are illustrated in Figure 5.8.
experiments. In nature, enrichment in K2O along the Nearly 65 years ago L.S. Darken proposed that the
granite/gabbro interface and biotite-rich rinds often sur- chemical diffusivity Dci can be expressed as
rounding mafic enclaves in granitic or granodioritic magma  
has been attributed to uphill diffusion. d ln gi
Dci ¼ kTMi 1 þ (5.12)
In recent years, there have been a number of experi- d ln xi
mental studies of chemical diffusion involving three and
four component systems (e.g., CaOeMgOeAl2O3eSiO2) where k is the Boltzmann constant, T is temperature, gi and
that have succeeded in constraining the full diffusion xi are the activity coefficient and mole fraction of species i,
coefficient matrix. These studies have shown that the on- respectively, and Mi is the intrinsic mobility (velocity of
diagonal Dii correspond closely to self-diffusivities, while species i per unit force). Equation (5.12) provides an
the off-diagonal terms are typically smaller, although not explicit expression relating the chemical diffusivity to both
always the case, and commonly negative, which can lead to species i intrinsic mobility (random walk velocity) and its
uphill diffusion. Given the challenges constraining the activity gradient causing a directional flux. Consider first
diffusion coefficient matrix for even up to four components the case of uniform composition. Since there is no chemical

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
128 PART | I Origin and Transport of Magma

FIGURE 5.8 Effects of temperature, time, and diffusion length scale for
Sr and Nd isotopic homogenization and bulk chemical homogenization of
basalt with entrained blobs of rhyolite. Diffusive impedance is the time (t)
in years over the length scale of the compositional heterogeneity (l) in
meters squared, t/l2, is the diffusive impedance. Contours for 50 and 95%
of the way to complete homogenization of isotopic composition and bulk
chemical composition are represented by the dotted and solid curves,
respectively. Approximate liquidus and solidus temperatures indicated by
stippled horizontal bars delimiting the temperature interval where hy-
bridization of basalt is likely to occur. Modified after Figure 12 in Lesher,
FIGURE 5.7 Schematic diagram showing progressive hybridization of C.E., 1994. Kinetics of Sr and Nd exchange in silicate liquids e theory,
mafic (A, red) and felsic (B, yellow) magmas when mixing is limited by a experiments, and applications to uphill diffusion, isotopic equilibration,
large viscosity contrast and diffusion. (A) Solid curves with arrows show and irreversible mixing of magmas. Journal of Geophysical Research 99,
the paths for the mafic magma (upper trajectories) and complementary 9585e9604 with permission from John Wiley & Sons Publications.
felsic magma (lower trajectories) as a function of time leading eventually
to bulk homogenization in time t3. These trends deviate markedly from
those predicted for simple binary mixing shown by the dashed curves. Herein lies the common assumption that tracer diffusion
Uphill diffusion of Sr is depicted when the gradient in activity is opposite
and self-diffusion are equivalent measures of intrinsic
to the gradient in concentration, negative Darken diffusivity (see text). In
(B) the covariation of 87Sr/86Sr and 143Nd/144Nd ratios reflects the fact that mobilitydbut again only in the limit of Henry’s law.
the intrinsic mobility of Sr is greater than of Nd (see Tables S5.10 and The equivalency of the self-diffusion and intradiffusion
S5.11). Isotopic equilibrium is practically achieved in time t, which may be as might be derived from an isotope diffusion couple
less than that required for the system to achieve complete chemical ho- experiment is not so straightforward, since isotopes of the
mogenization (t3). Modified after Figure 2 in Lesher, C.E., 1990.
same element have mass differences and thus their self-
Decoupling of chemical and isotopic exchange during magma mixing.
Nature 344, 235e237 with permission Nature Publishing Group. diffusivities may differ. Drawing again on Darken’s work,
the diffusivity for exchange of two isotopes of the same
gi
element (isotope 1 with mass m1 and isotope 2 with mass
flux, dd ln
ln xi ¼ 0 and Di ¼ kTMi . This is tantamount to our
c
m2) in an otherwise homogeneous solution depends on their
definition of self-diffusivity, D*, so that relative concentrations and self-diffusivities
 
 d ln gi D ¼ x1 D2 þ x2 D1
Di ¼ Di 1 þ
c
(5.13) (5.14a)
d ln xi
where
Now consider species i present in dilute concentration and  b
obeying Henry’s law, i.e., gi ¼ constant. In this case, while m2
gi D1 ¼ D2 (5.14b)
dlnxi is nonzero, dlngi will be zero and thus dd lnln xi ¼ 0. m1

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 129

b is typically 0.5 for gases but markedly smaller for from Eqn (5.4), the Arrhenius relation can also describe the
condensed phases. In the case of silicate melts, 6Li has been dependence of diffusivity on pressure and temperature
shown to diffusive w12% faster than 7Li giving b ¼ 0.2,
while the differences in diffusivities of isotopes of Fe, Mg, D ¼ D0 exp½ðEa þ PVa Þ=RT (5.15)
and Ca are well below 2%, constraining b < 0.06 for these
elements. b for Si is expected to be markedly smaller where D0 is constant for a fixed composition, and Ea and Va
(0.025) meaning that differences in self-diffusivities for represent the activation energy and activation volume for
Si isotopes are likely far too smaller to be resolved by diffusive transport, respectively. Equations (5.15) is a use-
current mass spectrometry methods. Furthermore, from ful starting point for any analysis of the temperature and
 
Eqn (5.14b) it can be appreciated that as mm1 /1, D1 /D2 so pressure dependence of diffusivities; however, certainly in
2

that for most petrologic applications involving traditional the case of silicate melts the assumption of constant Ea and
radiogenic systems (Sr, Nd, Hf, and Pb) the differences in Va may only be valid for restricted temperature and pres-
isotope self-diffusivities will be so small that diffusive sure intervalsdfor many of the same reasons that Eqn (5.4)
fractionation effects will be negligible. On the other hand, fails for fragile liquids. However, for diffusion the pre-
this also means that, in practice, the self-diffusivity can be exponential term (D0) depends strongly on melt composi-
determined directly by measuring the flux of one isotope tion and the size and charge of the species. As such, Eqn
relative to others of that element in an otherwise homoge- (5.15) often does not provide sensible results if the diffu-
neous mediumdin other words, from an intradiffusion sivities being considered are chemical diffusivities that
couple experiment. depend on the details of the compositional gradients used to
Returning to Eqn (5.13), in most situations encoun- determine Dij. It is best to limit consideration to experi-
gi
tered in magmatic systems the ratio dd ln
ln xi will have a value mental measurements that serve as good proxies for self-
other than zero, while its magnitude is expected to vary in diffusivities, i.e., tracer and intradiffusivities (see Tables
space and time. Thus, the Darken diffusivity has the S5.10 and S5.11). Ea is found to be of order 300e400 kJ/
peculiarly feature of being either positive, negative, or mol for network-forming atoms like Si and Al and in the
zero depending on whether the activity of species i in- range 250e180 kJ/mol for transition metals and the alka-
creases, decreases, or is constant with concentration, line earth metals. Alkali metals have activation energies in
respectively. While the former case is usually encountered the range 120e220 kJ/mol.
and certainly prevails in the absence of gradients in other Less is known about Va for silicate melts. Va is typi-
components, the latter situations arise due to strong non- cally positive for network modifiers such as Ca and Mg
ideality accompanying gradients in other components, (þ2 to þ12 cm3/mol) and can be positive or negative for
particularly silica content. The occurrence of uphill network-forming cations (Si) and oxygen, depending on
diffusion in this context can then be appreciated as a direct composition and pressure. Positive Va is consistent with
consequence of a negative Darken diffusivity for that more conventional free-volume models of ionic diffusion,
element, while for rapidly diffusing species concentration while to explain negative Va one often appeals to coop-
differences can be established essentially under quasi- erative modes of diffusion. For basalt to rhyolite compo-
equilibrium conditions where the activity gradient tends sitions at low to moderate pressures, Va for Si and O range
to zero. These concentration differences can persist as from 2 to 15 cm3/mol with more negative values
long as there remain differences in, for example, silica associated with more silicic (polymerized) melts. Both
content. The main point regarding magma mixing is that experiments and MD simulations show that Va for Si and
ultimately mass transfer depends on chemical diffusion O can change from negative to positive at high pressure.
operating at the molecular scale. The process is generally The maximum in the diffusivity has been attributed to
not explicable by linear mixing, and due to differences in progressive stabilization of high coordinated network
self-diffusivities and the complexities of the activitye species with pressure, but there is also mounting evidence
composition relations significant decoupling among that the anomalous pressure dependence of network for-
chemical components and isotopic systems is possible mers up to 5e6 GPa is intimately connected to the
(Figures 5.7 and 5.8). Armed with this appreciation can compressible nature of polymerized silicate melts over
help in deciphering the often complex compositional this pressure range. In contrast, Va for Si and O is positive
relationship among hybrid rocks formed by magma even at low pressure in depolymerized nature liquids such
mingling. as diopsidic melt suggesting that given their more
A great deal is known about the effects of temperature compact structure diffusion is restricted by available free-
and composition on self/tracer and effective binary chem- volume and more readily accomplished by motion of in-
ical diffusivities at atmospheric pressure, while our un- dividual ions rather than through cooperative motion.
derstanding of the effects pressure is meager. Following Granted, these expectations are based on limited data for

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
130 PART | I Origin and Transport of Magma

DLi DNa > DK > DRb > DCs. The diffusivity of the car-
bonate group (CO3) is w1.2  1010 m2/s whereas for
H2O, D ¼ w1  109 m2/s, and depends on the concen-
tration of dissolved water. At fixed temperature, D*’s for
oxygen and most high field strength cations, including Si,
are larger in basalt melt compared to rhyolite, by a factor of
5e500; but interestingly the alkalis shown little to no
dependence on magma composition. These relationships
are illustrated in Figure 5.9 using an empirical model based
on the assumption that Df h1 and fit to experimental data
for self-diffusion and independent constraints on viscosity.
While such empirical models do have limitations, notably
missing is a comprehensive picture of the effects of volatile
FIGURE 5.9 Tracer diffusivities for the alkalis, intermediate field species (H2O, CO2, etc.) and pressure, they do permit order
strength elements (IFSE bounded by Z2/r ¼ 3 and 8) and high field strength of magnitude estimates of self-diffusivities for ions of
elements (HFSE; Z2/r > 10) from 1300
C (upper bound) to 1000
C different size and charge over a broad range of natural
(lower bound) and spanning composition of basalt (red, left hand bound) to magma compositions and temperature. This information is
rhyolite (yellow, right hand bound) computed from equations presented by
Mungall (2002). Diffusivities are plotted against log viscosity from
essential for modeling the dynamics and timescales of
Giordano et al. (2008). magmatic processes fundamentally controlled by diffusion
at the molecular to atomic length scales.

high-pressure conditions and further laboratory measure-

ments and first-principle calculations are sorely needed to
5. CONCLUSIONS
reliably constraint processes involving diffusion in the The important thermodynamic and transport properties of
deep interior of our own and other planetary bodies. magma have been reviewed in this chapter. The complete
Studies of diffusion in conjugation with viscous flow are description of multicomponent and multiphase magma at
also critical for understanding the connections between temperatures from 800 to 5000 K and pressures from
these transport processes on a fundamental physical/ 104 GPa (surface) to 135 GPa (base of mantle) is an
structural level. ambitious program that will take many years to accom-
Roughly, Ea and Va for oxygen self-diffusion in a melt plish. Although an enormous task, there are reasons to be
are equivalent to EA and VA for viscous flow because optimistic. Within the last quarter century, knowledge of
viscous flow in a silicate melt is ultimately related to the the thermodynamic properties of high-temperature crys-
mobility of oxygen, the predominant constituent. This talline, molten, and glassy silicates including melts of
expectation derives from the classic expressions proposed natural composition has grown dramatically. Although still
by Einstein, Glasstone, Eyring, among others, that assume far from complete, thermodynamic properties for many
Df h1 . This simple proportionality has been very useful in crystalline phases relevant to magmatic systems are known
understanding relative magnitudes of self-diffusivities and at the high temperature and pressure conditions extending
their dependence on magma composition (Figure 5.9). As a at least to the base of the lithosphere and within the
rule, oxygen together with Si and other high field strength asthenosphere where primary liquids are formed beneath
cations tend to have the smallest self-diffusivities for a spreading ridges, within upwelling mantle plumes and at
given temperature and melt composition, while diffusivities subduction zones. An approximate model for the Gibbs
for the transition, alkaline earth, and alkali metals are uni- free energy of natural silicate liquids has also been devel-
formly larger. In general, network-modifying ions of high oped in the past 15 years and is undergoing continuous
charge and large size have smaller self-diffusivities than revision and expansion. These advances have improved
small ions of low charge, other factors remaining the same. efforts to describe and predict equilibrium phase relations
For example, in a basaltic melt at 1300
C, oxygen and the involved in the melting and solidification of magma.
network cations Si and Al have D* w3  1012 m2/s. Similarly, studies of the temperature, pressure, and
This compares with rare earth and transition metals composition dependence of melt viscosity, the rheology of
(e.g., Fe, Ti), and alkaline earths (e.g., Mg, Ca, Sr, Ba) for magmatic (solidemeltevapor) mixtures, chemical, tracer,
which D* w 1  1011 m2/s and w2  1011 m2/s, and self-diffusion and thermal conductivity have been
respectively. A typical ranking for basaltic melts is made for a variety of natural and synthetic melts and
DBa DSr > DCa > DMg > DFe > DREE > DTi. The alkalis glasses. Correlations have been developed for some prop-
are significantly more mobile with D* w3  1010 m2/s erties valid in specific regions of temperatureepressuree
and exhibit a strong dependence on ionic radius with composition space that enable one to obtain order of

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 131

magnitude or better estimates of many critical properties. computed are estimates, they do provide a means for un-
These have been greatly aided by significant advances in derstanding, at the atomic level, the effect of temperature,
experimental techniques, especially for high pressures and pressure, and composition on physical properties and for
utilizing a variety of radiation sources and spectroscopic identifying critical laboratory experiments that should be
tools, and analytical methods that are probing at ever performed.
smaller length scales with hitherto unfathomable precision Study of the properties of molten, glassy, and high-
and accuracy. Development of models based on condensed temperature crystalline silicates and oxides of geolog-
matter physics has come to supersede earlier purely ical relevance is a burgeoning field for last research. The
empirical models and provide for rational interpolations rate of acquisition of new information is accelerating as
and a basis for careful extrapolation. In the computational new techniques are applied to old problems. There are
realm, MD simulations are increasingly being applied to hundreds of volcanic eruptions each year and many of
investigate the structure and properties of molten silicates these present dangers involving the loss of life and
at very high temperatures and pressures. This method, property. The first step in taming magma is to understand
which involves use of a potential energy expression its nature from a fundamental and broad perspective. At
describing the electrostatic interactions between the the same time a deeper understanding of the evolution of
various atoms in a fluid or melt, enables one to compute planet Earth and, by analogy, other terrestrial planets
thermodynamic, spectroscopic, and transport properties both within our solar system and beyond depends on
under a variety of conditions. Although properties so knowledge regarding the physical properties of magma.

SUPPLEMENTARY MATERIAL FOR THERMODYNAMIC AND TRANSPORT PROPERTIES

OF SILICATE MELTS AND MAGMA

TABLE S5.1 Partial Molar Volumes, Thermal Expansions and Compressibilities

of Oxide Components. Note That (vVi/vP)T h bTVi, Where bT Is the
Isothermal Partial Molar Compressibility of the ith Component and
(vVi/vT)p h aP Vi, Where aP Is the Isobaric Expansivity.

Vliq ðT; P; XÞ ¼ S Xi ½Vi;1673K þ ðvVi =vTÞp ðT  1673Þ þ ðvVi =vPÞT P

Vi,1673K (vVi/vT)P (vVi/vP)T

(106 m3/mol) (109 m3/mol$K) (106 m3/mol$GPa)

SiO2 26.86  0.03 0.0 1.89  0.02

TiO2 23.16  0.26 7.24  0.46 2.31  0.06

Al2O3 37.42  0.09 0.0 2.26  0.09
Fe2O3 42.13  0.28 9.09  3.49 2.53  0.09
FeO 13.65  0.15 2.92  1.62 0.45  0.03
MgO 11.69  .08 3.27  0.17 0.27  0.07
CaO 16.53  0.06 3.74  0.12 0.34  0.05

Na2O 28.88  0.06 7.68  0.10 2.40  0.05

K2O 45.07  0.09 12.08  0.20 6.75  0.14
Li2O 16.85  0.15 5.25  0.81 1.02  0.06
H2O 26.27  0.5 9.46  0.83 3.15  0.61
CO2 25.4  1.0 10.86  1 3.82  0.8

Modified after Lange and Carmichael (1990), Lange (1997), Ochs and Lange (1997), and Ghiorso and Gualda
(2014, this volume). Uncertainties are 1s.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
132 PART | I Origin and Transport of Magma

TABLE S5.2 Molar Volume of Typical Silicate Crystals and Melts at High Temperature and 104 GPa. Depolymerized
Compositions Typically Decrease in Density 10e20% upon Fusion. The Volume Change Is Less for Tetrahedral Fluids
Such as SiO2. Molar Volumes at 1400
C, Except for H2O at 273 K.

Melting Vmelt Vcrystal DVfusion

Mineral Temperature (K) (106 m3/mol) (106 m3/mol) Vm/Vc (106 m3/mol)

SiO2 (cristobalite) 1999 26.91 27.44 0.98 0.53

NaAlSi3O8 1393 112.83 104.13 1.08 8.64
KAlSi3O8 1473 121.24 111.72 1.09 9.75
CaAl2Si2O8 1830 108.41 103.42 1.05 5.48
CaSiO3 1817 43.89 41.87 1.05 2.31

MgSiO3 1830 38.85 33.07 1.18 5.99

CaMgSi2O6 1665 81.80 69.74 1.18 12.62
Ca2MgSi2O7 1727 99.70 97.10 1.03 2.60
Ca3MgSi2O8 1848 118.09 105.92 1.12 13.07
Fe2SiO4 1490 53.15 48.29 1.11 5.49
Mg2SiO4 2174 52.37 47.41 1.12 5.53

CaTiSiO5 1670 67.61 57.57 1.17 10.04

H2O (ice) 273 18.01 19.64 0.92 1.63

Modified from Lange and Carmichael (1990).

TABLE S5.3 Density and Viscosity of Molten Carbonate (Jones et al., 1995 and Sykes et al., 1992)
and Supercritical H2O (Haar et al., 1984).

Composition Pressure (GPa) Temperature (
C) Density (kg/m3) Viscosity (Pa s)

K2Ca(CO3)2 104 975 2014 d

2.5 950 2750 0.032

2.5 1150 2580 0.018

4.0 1050 2800 0.023
K2CO3 4 1500 3100 0.023
K8Ca3Mg(CO3)8 2 1250 e 0.065
K2Mg(CO3)2 3 800 e 0.036
3 900 e 0.022

5.5 1200 e 0.006

H2O 0.3 500 e 104
800 e 7105

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 133

TABLE S5.4 Approximate Values of the Isothermal Bulk Modulus (KT) and Its Pressure derivative (K0 ¼ dKT/dP)
for Use in the BircheMurnaghan Equation of State for Some Simple and Naturally Occurring Silicate Melts at
High Temperature (To). The Temperature Derivative, dKT/dT, has Been Constrained Only for a Few
Compositions.

KT Data
Composition (GPa) K0 dKT/dT To (K) Source*

Simple Systems
CaMgSi2O6 22.4 6.9 d 1773 (1)
CaAl2Si2O8 17.9 5.3 d 1923 (1)
Fe2SiO4 19.4 5.3 d 1773 (2)
Naturally Occurring
Rhyolite (0e8 wt% H2O) 11.5 6.5 0.0016 1273 (3)

Andesite (0e 9 wt% H2O) 17.2 6.3 d 1273 (4)

Basalt (anhydrous) 19.3 4.4 d 1673 (5)
Basalt (anhydrous) 20.8 4.6 d 1673 (6)
Basalt (2 wt% H2O) 18.2 4 d 2573 (6)
Basalt (8 wt% H2O) 14.5 3.2 2473 (6)
Komatiite (anhydrous) 26.0 4.3 d 2073 (5)

Komatiite (anhydrous) 23.1 4.9 d 2073 (6)

Peridotite (anhydrous) 23.4 6.2 d 2273 (6)
Peridotite (anhydrous) 24 7.3 0.0027 2100 (7)
Peridotite (5 wt% H2O) 8.8 9.9 0.0022 1773 (8)
Peridotite (w2 wt% CO2) 23 8.5 0.01 1800 (9)

*(1) Rigden et al., (1989); (2) Thomas et al., 2012; (3) Malfait et al., 2014a; (4) Malfait et al., 2014b; (5) Agee, 1998; (6) refits by Jing and Karato, 2008, of
published sink/float constraints, see additional references therein; (7) Sakamaki et al., 2010; (8) Sakamaki et al., 2009; (9) Sakamaki et al., 2011.

TABLE S5.5 Melting Temperature, Molar and Specific Enthalpy of Fusion and Entropy of
Fusion at 104 GPa for Some Silicate and Oxide Phases.

Melting DHfus DSfus Dhfus

Formula Temperature (K) (kJ/mole) (J/mole-K) (kJ/kg)

H2O 273 7.401 27.11 116

NH3 195 5.657 29.01 333
CH4 90.6 0.937 10.34 58
ZrO2 (baddeleyite) 3123 87.03 27.87  0.02 706
Fe0.947O 1652 31.3  0.2 19.0  0.1 454
FeTiO3 1640 21.7  0.1 13.2  0.1 143

TiO2 (rutile) 1870 67.0  0.1 35.8  0.1 838

Fe2O3 (hematite) 1895 114.5  0.2 60.4  0.2 717
PbO (Massicot) 1170 25.52  0.01 21.81 114

(Continued)

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
134 PART | I Origin and Transport of Magma

TABLE S5.5 Melting Temperature, Molar and Specific Enthalpy of Fusion and Entropy of
Fusion at 104 GPa for Some Silicate and Oxide Phases.dCont’d
Melting DHfus DSfus Dhfus
Formula Temperature (K) (kJ/mole) (J/mole-K) (kJ/kg)

Fe3O4 (Magnetite) 1870 138.07  0.05 73.83 596

LiAlO2 1883 87.9  0.3 46.7  0.5 1333
MgAlO2 2408 107  11 44  4 752

Ca2Fe2O5 1750 151.0  0.5 86.3  0.1 556

Al2O3 2323 107.5  54 46.3  23 1054
SiO2 (quartz)a 1700 9.40  1.0 5.53  0.56 157
SiO2 (cristobalite) 1999 8.92  1.0 4.46  0.50 149
MgSiO3a 1834 73.2  6.0 39.9  3.3 729
CaSiO3 (wollastonite)a 1770 61.7  4.0 34.9  2.3 531

CaSiO3 (pseudowoll) 1817 57.3  2.9 31.5  1.6 493

CaMgSi2O6 1665b 137.7  2.4 82.7  1.4 636
Ca2MgSi2O7 1727 123.9  3.2 71.7  1.9 454
Ca3MgSi3O8b 1848 125  15 67.5  8.1 350
Fe2SiO4 1490 a
89.3  1.1 59.9  0.7 438
Mn2SiO4 1620 89.0  0.5 55.2  0.3 633

Mg2SiO4 2174 142  14 65.3  6 1010

CaTiSiO5 1670 123.8  0.4 74.1  0.2 755
NaAlSi3O8 1393 64.5  3.0 46.3  2.2 246
NaAlSi2O6 a
1100 59.3  3.0 53.9  2.7 293
NaAlSiO4a 1750 49.0  2.1 28.0  1.2 345
Na2Si2O5 1147 35.6  4.1 31.0  3.6 196

KAlSi3O8a 1473 57.7  4.2 39.2  2.8 207

CaAl2SiO8 1830 133.0  4.0 72.7  2.2 478
a
Mg3Al2Si3O12 1500 243  8 162  5 603
Mg2Al4Si5O18 a
1740 346  10 199  6 591
KMg3AlSi3O10F2 1670 308.8  1.3 185  1 733
a
Metastable congruent melting.
b
Incongruent melting.
Data compiled from Kelly (1936), Robie and Hemingway (1995), Ghiorso (2004) and references therein.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 135

TABLE S5.6 Coefficients for Estimating the Isobaric Heat Capacity of Silicate Glasses Valid for
Temperatures in the Range 400 K < T < 1000 K at 104 GPa. Xi Is the Mole Fraction of the ith
P
Oxide Component. The gfw Is Defined According to gfw ¼ XiMi, Where Mi is the Molar Mass of
the ith Component.
X X X  
Cp ðTÞ ¼ ai X i þ bi X i T þ ci Xi T2 J K1 gfw1

ai bi  102 ci  105

SiO2 66.354 0.7797 28.003

TiO2 33.851 6.4572 4.470
Al2O3 91.404 4.4940 21.465
Fe2O3 58.714 11.3841 19.915
FeO 40.949 2.9349 7.6986
MgO 32.244 2.7288 1.7549

CaO 46.677 0.3565 1.9322

Na2O 69.067 1.8603 2.9101
K2O 107.194 3.2194 28.929

Modified after Stebbins et al. (1984).

TABLE S5.7 Partial Molar Isobaric Heat Capacity for Molten Oxide Components Applicable to
Silicate Melts at 104GPa. Cp is Approximately Independent of Temperature at T 1400 K.
X  
Cp ¼ Cpi Xi J K1 gfw1

Molar Specific
Cpi (J/gfw K) cpi (J/kg K)

SiO2 80.0  0.9 1331

TiO2 111.8  5.1 1392
Al2O3 157.6  3.4 1545

Fe2O3 229.0  18.4 1434

FeO 78.9  4.9 1100
MgO 99.7  7.3 2424
CaO 99.9  7.2 1781
SrO 88.7  8 855
BaO 83.4  6.0 544

H2O molecular 41  14.0 2278


(OH) hydroxyl 153  18 8500
Li2O 104.8  3.2 3507
Na2O 102.3  1.9 1651
K2O 97.0  5.1 1030
Rb2O 97.9  3.6 524

Modified from Stebbins et al. (1984). Water values from Bouhifd et al. (2006).

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
136 PART | I Origin and Transport of Magma

TABLE S5.8 Thermal Conductivity of Some Geosilicate Crystals, Glasses and Melts at 1 bar.

Composition and state Temperature (
C) Thermal Conductivity (W/m K) Data Source*

Olivine (Fo90) 800 2.69 (1)

1400 2.18 (1)
Forsterite 400 2.47 (1)

800 1.84 (1)

1400 1.59 (1)
Orthopyroxene (bronzite) 0 4.62 (1)
300 3.05 (1)
Diopside 20 4.27 (1)
Diabase 300 2.09 (1)

Obsidian (rhyolite glass) 0 1.34 (1)

300 1.67 (1)
500 1.89 (1)
Basalt glass 0 1.15 (1)
300 1.48 (1)
SiO2 (glass) 600 1.76 (1)

1230 1.87 (1)

NaAlSi3O8 (glass) 25 1.37 (2)
800 1.56 (2)
CaAl2Si2O8 (glass) 25 1.13 (2)
800 1.43 (2)
CaMgSi2O6 (glass) 25 1.24 (2)

800 1.46 (2)

KAlSi3O8 (glass) 25 1.12 (2)
800 1.44 (2)
SiO2 (liquid) 1400 1.32 (2)
1700 1.41 (2)
CaMgSi2O6 (liquid) 1100 1.21 (3)

CaAl2SiO8 (liquid) 1200 1.45 (3)

NaAlSi3O8 (liquid) 1200 1.59 (3)
KAlSi3O8 (liquid) 1200 1.45 (3)
Rhyolite (liquid) 800e1100 w1.5 (3)

*(1) Clark (1966); (2) Hofmeister et al. (2009); (3) Pertermann et al. (2008).

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 137

TABLE S5.9 Thermal Conductivity for Geosilicate Liquids at Elevated Pressure Determined
by Molecular Dynamics Simulation.

Temperature (
C) Pressure (GPa) Thermal Conductivity (W/m K)

CaMgSi2O6 liquid
1800 0 1.14
1800 10 2.02

2800 0 1.02
2800 10 1.94
4000 0 0.77
4000 10 1.85
NaAlSi3O8 liquid
1800 0 1.45

1800 10 2.16
2800 0 0.99
2800 10 2.55
3750 0 0.89
3750 10 1.89
Mg2SiO4 liquid

3000 0 1.06
3300 0 0.90
3300 6.3 1.83
3300 10 2.11
3800 0 0.89
4300 0 0.68

Data from Tikunoff and Spera (2014).

TABLE S5.10 Activation Energies (Ea) and Preexponential Factors (Do) for Self/Tracer
Diffusivities in Anhydrous Basaltic (46e49.7 wt% SiO2) Melts. Extreme Caution Should Be
Exercised Extrapolating Beyond the Experimental T and P Range.

lnDo Data
Species P (GPa)* T Range (K) Ea (kJ/mol) (Do in m2/s) Source#

Li 104 (a) 1569e1673 116 11.8 (1)

4
Na 10 (a) 1577e1692 164 9.3 (2)
4
Cs 10 (a) 1567e1675 274 4.4 (2)
4
Ca 10 (a) 1538e1723 184 9.8 (3)
4
Sr 10 (a) 1538e1723 182 10.5 (3)
4
Sr 10 (a) 1567e1675 191 9.7 (2)
Sr 1 (gr) 1528e1738 136 14.4 (4)

(Continued)

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
138 PART | I Origin and Transport of Magma

TABLE S5.10 Activation Energies (Ea) and Preexponential Factors (Do) for Self/Tracer
Diffusivities in Anhydrous Basaltic (46e49.7 wt% SiO2) Melts. Extreme Caution Should Be
Exercised Extrapolating Beyond the Experimental T and P Range.dCont’d
lnDo Data
Species P (GPa)* T Range (K) Ea (kJ/mol) (Do in m2/s) Source#

Ba 104 (a) 1538e1723 165 12.0 (3)

4
Ba 10 (a) 1572e1692 170 11.7 (2)
4
Sc 10 (a) 1572e1675 198 10.3 (2)
4
Mn 10 (a) 1567e1671 166 11.2 (2)
Feþ3 104 (a) 1570e1675 265 4.8 (2)
Co 104 (a) 1567e1671 197 9.0 (2)
Co 104 (a) 1538e1723 152 12.1 (3)
Si 1 (gr) 1593e1873 167 12.8 (6)
O 104 (v) 1593e1773 251 6.6 (7)

O 1 (gr) 1593e1873 172 12.4 (6)

Euþ3, Gd 104 (a) 1593e1713 170 12.1 (5)
þ3 4
Eu 10 (a) 1567e1673 268 5.0 (8)
þ3 4
Eu 10 (a) 1577e1672 248 9.3 (8)

*(a), run in air; (v), run at variable log fO2 between 0 and 3.3; (gr), enclosed in graphite capsule.
#
(1) Lowry et al., 1981; (2) Lowry et al., 1982; (3) Hofmann and Magaritz, 1977; (4) Lesher, 1994; (5) Magaritz and Hofmann,
1978; (6) Lesher et al., 1996; (7) Canil, 1990; (8) Hendersen et al., 1985.

TABLE S5.11 Activation Energies (Ea) and Preexponential Factors (Do) for Self/Tracer
diffusivities in Anhydrous Rhyolitic (69e76 wt% SiO2) Melts. Extreme Caution Should Be
Exercised Extrapolating Beyond the Experimental T and P Range.

lnDo Data
Species P (GPa)* T Range (K) Ea (kJ/mol) (Do in m2/s) Source*

Li 104 (a) 1573e1673 84 13.8 (1)

Na 104 (a) 1573e1673 152 9.6 (2)

Cs 104 (a) 1573e1673 219 10.7 (2)

Cs 104 (a) 1063e1573 201 13.8 (3)
Ca 104 (a) 903e1203 284 1.6 (3)
Ba 104 (a) 1573e1673 130 16.6 (2)
4
Ba 10 (a) 1433e1588 199 11.2 (2)
Sr 1 (gr) 1528e1738 136 16.3 (4)
4
Mn 10 (a) 1573e1673 210 10.6 (2)
þ3 4
Fe 10 (a) 1573e1673 177 14.4 (2)
4
Co 10 (a) 1573e1673 167 13.9 (2)
B 1 (gr) 1573e1773 400 1.7 (5)
Ga 1 (gr) 1573e1773 347 4.3 (5)
Si 1 (gr) 1573e1773 139 23.0

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 139

TABLE S5.11 Activation Energies (Ea) and Preexponential Factors (Do) for Self/Tracer
diffusivities in Anhydrous Rhyolitic (69e76 wt% SiO2) Melts. Extreme Caution Should Be
Exercised Extrapolating Beyond the Experimental T and P Range.dCont’d
lnDo Data
Species P (GPa)* T Range (K) Ea (kJ/mol) (Do in m2/s) Source*

Si 1 (gr) 1733e1935 380 3.8 (6)

Si 2 (gr) 1733e1935 305 8.3 (6)
Si 4 (gr) 1733e1935 163 15.9 (6)

O 1 (gr) 1733e1935 293 8.9 (6)

O 2 (gr) 1733e1935 264 10.2 (6)
O 4 (gr) 1733e1935 155 15.9 (6)
Euþ3 104 (a) 1473e1673 253 9.0 (2)
Euþ3 104 (a) 973e1323 289 7.2 (3)
Ce 104 (a) 1148e1373 490 6.3 (3)

Nd 1 (gr) 1528e1738 198 13.7 (4)

Uþ6 104 (a) 1250e1923 364 6.7 (7)
4
Th 10 (a) 1250e1923 369 7.0 (7)

*(a), run in air; (gr), enclosed in graphite capsule.

*(1) Cunningham et al., 1983; (2) Hendersen et al., 1985; (3) Jambon, 1982; (4) Lesher, 1994; (5) Baker, 1992; (6) Tinker et al.,
2001; (7) Mungall, 1997.

ACKNOWLEDGMENTS
Clark Jr, S.P., 1966. Thermal conductivity. GSA Memoirs 97, 459e482.
This material is based upon work supported by the National Science Cunningham, G.J., Henderson, P., Lowry, R.K., Nolan, J., Reed, S.J.B.,
Foundation under grants EAR-1019887 and EAR-1215714, and the Long, J.V.P., 1983. Lithium diffusion in silicate melts. Earth Planet
U.S. Department of Energy under contract DE-FG-03-91ER-14211. Sci Lett 65, 203e205.
This research also used resources of the National Energy Research Dingwell, D.B., Webb, S.L., 1989. Structural relaxation in silicate melts
Scientific Computing Center (NERSC) supported by U.S. Depart- and non-Newtonian melt rheology in geologic processes. Physics and
ment of Energy contract DE-AC02-05CH11231. CEL acknowledges Chemistry of Minerals 16, 508e516.
support from the Danish National Research Foundation for the Niels Dobson, D.P., Jones, A.P., Rabe, R., Sekine, T., Kurita, K., Taniguchi, T.,
Bohr Professorship at Aarhus University during production of this Kondo, T., Kato, T., Shimomura, O., Urakawa, S., 1996. In-situ
chapter. Any opinions, findings, and conclusions or recommenda- measurement of viscosity and density of carbonate melts at high
tions expressed in this chapter are those of the authors and do not pressure. Earth Planet. Sci. Lett 143, 207e215.
necessarily reflect the views of agencies funding this work. Ghiorso, M.S., 2004. An equation of state for silicate melts. III. Analysis
of shock compression data and mineral fusion curves. Am. J. Sci 304,
752e810.
FURTHER READING Ghiorso, M., Gualda, G., 2015. Chemical Thermodynamics and the Study
Agee, C.B., 1998. Crystal-liquid density inversions in terrestrial and lunar of Magmas. In: Sigurdsson, H., Houghton, B.F., McNutt, S.R.,
magmas. Phys. Earth Planet. Int 107, 63e74. Rymer, H., Stix, J. (Eds.), Encyclopedia of Volcanoes. Elsevier,
Baker, D.R., 1992. Tracer diffusion of network formers and multicom- pp. 143e161.
ponent diffusion in dacitic and rhyolitic melts. Geochim Cosmochim Giordano, D., Russell, J.K., Dingwell, D.B., 2008. Viscosity of magmatic
Acta 56 (2), 617e631. liquids: a model. Earth and Planetary Science Letters 271, 123e134.
Bouhifd, M., Whittington, A., Roux, J., Richet, P., 2006. Effect of water Haar, L., Gallagher, J., Kell, G., 1984. NBS/NRC Steam Tables: Ther-
on the heat capacity of polymerized aluminosilicate glasses and modynamic and Transport Properties and Computer Programs for
melts. Geochimica et Cosmochimica Acta 70, 711e722. Vapor and Liquid States of Water in SI Units. Hemisphere Publishing
Brouwers, H.J.H., 2010. Viscosity of a concentrated suspension of rigid Corporation, Washington, 320 pp.
monosized particles. Physical Review E 81, 051402. Henderson, P., Nolan, J., Cunningham, G.C., Lowry, R.K., 1985.
Canil, D., Muehlenbachs, K., 1990. Oxygen diffusion in a Fe-rich basalt Structural controls and mechanisms of diffusion in natural silicate
melt. Geochim Cosmochim Acta 54, 2947e2951. melts. Contrib Mineral Petrol 89, 263e272.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
140 PART | I Origin and Transport of Magma

Hofmann, A.W., Magaritz, M., 1977. Diffusion of Ca, Sr, Ba, and Co in a Nevins, D., Spera, F.J., 1998. Molecular dynamics simulations of molten
basalt melt - implications for geochemistry of mantle. J Geophy Res CaAl2Si2O8: Dependence of structure and properties on pressure. Am
82, 5432e5440. Mineral 83, 1220e1230.
Hofmeister, A.M., Whittington, A.G., Pertermann, M., 2009. Transport Ochs III, F.A., Lange, R.A., 1997. The partial molar volume, thermal
properties of high albite crystals and near-endmember feldspar and expansivity, and compressibility of H2O in NaAlSi3O8 liquid: new
pyroxene glasses and melts to high temperature. Contrib. Mineral. measurements and an internally consistent model. Contrib Mineral
Petrol 158, 381e400. Petrol 129, 155e165.
Jambon, A., 1982. Tracer diffusion in granitic melts - Experimental Pertermann, M., Whittington, A.G., Hofmeister, A.M., Spera, F., Zayak, J.,
results for Na, K, Rb, Cs, Ca, Sr, Ba, Ce, Eu to 1300 Degrees C and a 2008. Transport properties of low-sanidine single-crystals, glasses and
model of calculation. J Geophy Res 87, 797e810. melts at high temperature. Contrib Mineral Petrol 155, 689e702.
Jing, Z., Karato, S., 2008. Compositional effect on the pressure derivatives Richet, P., 1983. Viscosity and configurational entropy of silicate melts.
of bulk modulus of silicate melts. Earth Planet Sci Lett 272, 429e436. Geochim Cosmochim Acta 48, 471e483.
Jones, A.P., Dobson, D., Genge, M., 1995. Comment on physical prop- Richet, P., Bottinga, Y., 1995. Rheology and configurational entropy of
erties of carbonatite magmas inferred from molten salt data, and silicate melts. Rev.. in Mineral 32, 67e89.
application to extraction patterns from carbonatite-silicate magma Rigden, S.M., Ahrens, T.J., Stolper, E.M., 1989. High-pressure equation
chambers. Geol. Mag 132, 121e121. of state of molten anorthite and diopside. J. Geophys. Res 94,
Kelley, K.K., 1936. Contributions to data on theoretical metallurgy: V. Heats 9508e9522.
of fusion of inorganic compounds. U.S. Bur. Mines Bull 393, 166. Rivers, M.L., Carmichael, L.S.E., 1987. Ultrasonic studies of silicate
Lange, R.A., Carmichael, I.S.E., 1990. Thermodynamic properties of melts. Jour. Geophys. Res 92, 9247e9270.
silicate liquids with emphasis on density, thermal expansion and Robie, R.A., Hemingway, B.S., 1995. Thermodynamic properties of
compressibility. Reviews in Mineral 24, 25e59. minerals and related substances at 298.15 K and 1 bar (105 pascals)
Lange, R.A., 1997. A revised model for the density and thermal expan- pressure and at higher temperatures. USGS Bulletin 2131.
sivity of K2O-Na2O-CaO-MgO-Al2O3-SiO2 liquids from 700 to 1900 Sakamaki, T., Ohtani, E., Urakawa, S., Hidenori Terasaki, H.,
K: extension to crustal magmatic temperatures. Contrib. Mineral. Katayama, Y., 2011. Density of carbonated peridotite magma at high
Petrol 130, 1e11. pressure using an X-ray absorption method. Am. Mineral 96,
Lesher, C.E., 2010. Self-diffusion in silicate melts: theory, observations 553e557.
and applications to magmatic systems. Reviews in Mineralogy and Sakamaki, T., Ohtani, E., Urakawa, S., Suzuki, A., Katayama, Y., 2009.
Geochemistry 72, 269e309. Measurement of hydrous peridotite magma density at high pressure
Lesher, C.E., Hervig, R.L., Tinker, D., 1996. Self diffusion of network using the X-ray absorption method. Earth Planet Sci Lett 287,
formers (silicon and oxygen) in naturally occurring basaltic liquid. 293e297.
Geochim Cosmochim Acta 60, 405e413. Sakamaki, T., Ohtani, E., Urakawa, S., Suzuki, A., Katayama, Y., 2010.
Lowry, R.K., Henderson, P., Nolan, J., 1982. Tracer diffusion of some Density of dry peridotite magma at high pressure using an X-ray
alkali, alkaline-earth and transition element ions in a basaltic and an absorption method. Am. Mineral 95, 144e147.
andesitic melt, and the implications concerning melt structure. Scarfe, C.M., Mysen, B.O., Virgo, D., 1987. Pressure dependence of the
Contrib Mineral Petrol 80, 254e261. viscosity of silicate melts. In: Mysen, B.O. (Ed.), Magmatic Pro-
Lowry, R.K., Reed, S.J.B., Nolan, J., Henderson, P., Long, J.V.P., 1981. cesses: Physicochemical Principles, Geochem. Sec, 1, pp. 59e67.
Lithium tracer-diffusion in an alkali-basaltic melt - an ion-microprobe Special Publication.
determination. Earth Planet Sci Lett 53, 36e40. Shaw, H.R., 1972. Viscosities of magmatic silicate liquids: an
Magaritz, M., Hofmann, A.W., 1978. Diffusion of Sr, Ba and Na in empirical method of prediction. American Journal of Science 272,
obsidian. Geochim. Cosmochim. Acta 42, 595e605. 870e893.
Malfait, W.J., Seifert, R., Petitgirard, S., Mezouar, M., Sanchez-Valle, C., Spera, F.J., Ghiorso, M., Nevins, D., 2011. Structure, thermodynamic and
2014b. The density of andesitic melts and the compressibility of transport properties of liquid MgSiO3: comparison of molecular
dissolved water in silicate melts at crustal and upper mantle condi- models and laboratory results. Contributions to Mineralogy and
tions. Earth Planet Sci Lett 393, 31e38. Petrology 75, 1272e1296.
Malfait, W.J., Seifert, R., Petitgirard, S., Perrillat, J.-P., Mezouar, M., Stebbins, J.F., Carmichael, I.S.E., Moret, L.K., 1984. Heat capacities and
Ota, T., Nakamura, E., Lerch, P., Sanchez-Valle, C., 2014a. Super- entropies of silicate liquids and glasses. Contrib Mineral Petrol 86,
volcano eruptions driven by melt buoyancy in large silicic magma 131e148.
chambers. Nature Geoscience 7, 122e125. Stixrude, L., Lithgow-Bertelloni, C., 2010. Thermodynamics of the
Mungall, J.E., Dingwell, D.B., 1997. Actinide diffusion in a haplogranitic earth’s mantle. In: Wentzcovitch, R., Stixrude, L. (Eds.), Theoretical
melt: Effects of temperature, water content and pressure. Geochim and Computational Methods in Mineral Physics, Reviews in Mineral
Cosmochim Acta 61, 2237e2246. and Geochemistry, 71, pp. 465e484.
Mungall, J.E., 2002. Empirical models relating viscosity and tracer Sykes, D., Baker, M.B., Wyllie, P.J., 1992. Viscous properties of car-
diffusion in magmatic silicate melts. Geochimica et Cosmochimica bonate melts at high pressure, EOS. Trans. Am. Geophy. Union 73,
Acta 66, 125e143. 372.
Mysen, B.O., Richet, P., 2005. Silicate Glasses and Melts. Elsevier, Thomas, C.W., Liu, Q., Agee, C.B., Asimow, P.D., Lange, R.A., 2012.
New York. Multi-technique equation of state for Fe2SiO4 melt and the density of
Navrotsky, A., 1995. Energetics of silicate melts. Reviews in Mineral 32, Fe-bearing silicate melts from 0 to 161 GPa. J. Geophys. Res 117,
121e149. B10206.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

 Author's personal copy
Chapter | 5 Thermodynamic and Transport Properties of Silicate Melts and Magma 141

Tikunoff, D., Spera, F.J. Thermal conductivity of molten and glassy Urbain, G., Bottinga, Y., Richet, P., 1982. Viscosity of liquid silica, sili-
NaAlSi3O8, CaMgSi2O6 and Mg2SiO4 by non equilibrium molec- cates and alumino-silicates. Geochimica et Cosmochimica Acta 46,
ular dynamics at elevated temperature and pressure: part 1 e 1061e1072.
methods and results. American Mineralogist, 99(11e12), pp. Zarzycki, J., 1991. In: Chan, R.W. (Ed.), Glasses and the Vitreous State.
2328e2336. Cambridge University Press, Cambridge, pp. 505.
Tinker, D., Lesher, C.E., 2001. Self diffusion of Si and O in dacitic liquid Zhang, Y., Ni, H., Chen, Y., 2010. Diffusion data in silicate melts.
at high pressures. Am Mineral 86, 1e13. Reviews in Mineralogy and Geochemistry 72, 311e408.

The Encyclopedia of Volcanoes, Second Edition, 2015, 113e141

View publication stats