Magmas Under Pressure: Advances in High-Pressure Experiments on Structure and Properties of Melts

Magmas under Pressure: Advances in High-Pressure Experiments on Structure and Properties of Melts summarizes recent advances in experimental technologies for studying magmas at high pressures. In the past decade, new developments in high-pressure experiments, particularly with synchrotron X-ray techniques, have advanced the study of magmas under pressure. These new experiments have revealed significant changes of structure and physical properties of magmas under pressure, which significantly improves our understanding of the behavior of magmas in the earth’s interior.

This book is an important reference, not only in the earth and planetary sciences, but also in other scientific fields, such as physics, chemistry, material sciences, engineering and in industrial applications, such as glass formation and metallurgical processing.

• Includes research and examples of high-pressure technologies for studying the structure and properties of magma
• Summarizes the current knowledge on the structure and properties of high-pressure magma
• Highlights the importance of magma in understanding the evolution of the earth’s interior

• Language: English
• Publisher: Elsevier Science
• Release date: Apr 6, 2018
• ISBN: 9780128112748

Book preview

Magmas Under Pressure

Advances in High-Pressure Experiments on Structure and Properties of Melts

Editors

Yoshio Kono

Chrystèle Sanloup

Table of Contents

Cover image

Title page

Copyright

List of Contributors

Preface

Part 1. Magmas in the Earth's Interior

Chapter 1. Primary Melt Compositions in the Earth's Mantle

1. Introduction

2. Melting of Mantle Peridotite

3. Storage and Availability of Volatile Components in the Mantle

4. Melting of Rocks Other Than Peridotite in the Mantle

5. Geodynamic Variations in the Sources and Compositions of Melts

Chapter 2. Carbon-Bearing Magmas in the Earth's Deep Interior

1. Introduction

2. Problems of Experimental Techniques to Study Systems With Volatiles

3. Mantle Temperature and Silicate Solidi

4. Redox State in the Deep Upper and Lower Mantle

5. Melting and Phase Relations in the Carbon-Bearing Systems

6. Deep Carbon Cycle, Melting, and Material Transport in the Earth's Mantle

7. Concluding Remarks

Chapter 3. The Influence of Pressure on the Properties and Origins of Hydrous Silicate Liquids in Earth's Interior

1. Introduction

2. H2O Solubility and Speciation in Silicate Liquids

3. Effect of Pressure on Hydrous Melting

4. Physical Properties of Hydrous Silicate Melts

5. Concluding Remarks

Chapter 4. Melting in the Earth's Deep Interior

1. Introduction

2. Do We Have Geophysical Evidence for Melting in the Deep Earth?

3. Melting Diagrams for the Earth's Lower Mantle

4. Partial Melt Compositions and Melting/Crystallization Sequences

5. Melting of MORB at the Core–Mantle Boundary

6. Influence of Volatile Incompatible Elements

7. Melting in the Mantle and the Thermal Profile of the Earth

8. Conclusions

Part 2. Advances in Experimental Studies of Melts at High Pressures

Chapter 5. X-Ray Diffraction Structure Measurements

1. Introduction

2. Experimental Techniques

3. Examples of the Results by the Technique

4. Challenge and Future Perspectives

Chapter 6. X-Ray Absorption Spectroscopy Measurements

1. Introduction

2. Principles of X-Ray Absorption Spectroscopy

3. Analysis of EXAFS Spectra in Highly Disordered and Liquid Systems

4. Analysis of XANES in Highly Disordered and Liquid Systems

5. High-Pressure Apparatuses for X-Ray Absorption Spectroscopy

6. Examples

7. X-Ray Absorption Spectroscopy by X-Ray Raman Scattering

Chapter 7. Synchrotron Mössbauer Spectroscopy Measurement

1. Introduction

2. Mössbauer Spectroscopy for 57Fe Isotope

3. High-Pressure 57Fe Mössbauer Spectroscopy Using a Diamond-Anvil Cell

4. Nuclear Resonant Scattering of Synchrotron Radiation

5. Synchrotron Mössbauer Spectroscopy With Synchrotron Mössbauer Source

6. Beam Properties and Data Analysis of Synchrotron Mössbauer Source Radiation

7. Beam Focusing System for High-Pressure Synchrotron Mössbauer Source Spectroscopy

8. Applications

9. Summary

Chapter 8. Vibrational Properties of Glasses and Melts

1. Introduction

2. Infrared and Raman Spectroscopy: A Brief Introduction

3. In Situ Measurements: Instrumentation and Challenges

4. Silicate Speciation and Network Structure

5. Volatiles in Silicate Melts

6. Challenges and Future Perspectives

Chapter 9. Density and Elasticity Measurements for Liquid Materials

1. Introduction

2. Density Measurements

3. Sound Velocity and Elasticity Measurements

4. Equations of State

5. Applicability of Each Method and Concluding Remarks

Chapter 10. Viscosity Measurement

1. Introduction

2. Falling Sphere Viscosity Measurement at the Beamline 16-BM-B in the APS

3. Viscosity of Melts at High Pressures

4. Current Challenges in Viscosity Measurement at High Pressures

Chapter 11. Electrical Conductivity Measurement

1. Introduction

2. Electrical Conduction in Melts

3. Electrical Conductivity Measurement of Melts

4. Electrical Conductivity Data on Silicate and Carbonate Melts

5. Electrical Conductivity of Partial Molten Rocks

6. Concluding Remarks and Future Directions

Part 3. Current Knowledge on Structure and Properties of Magmas Under Pressure

Chapter 12. Structure and Properties of Silicate Magmas

1. Structure of Silicate Magmas

2. Density of Silicate Magmas

3. Viscosity of Silicate Magmas

4. Elastic Wave Velocity of Silicate Glasses

5. Conclusion

Chapter 13. Densification Mechanisms of Oxide Glasses and Melts

1. Introduction

2. X-Ray and Neutron Diffraction Methods

3. Densification of SiO2 Glass

4. Densification of GeO2 Glass

5. Role of Fivefold AO5 Units in the Densification of SiO2, GeO2, and Silicates

6. Ring Closure and the Zipper Mechanism

7. Densification of B2O3 Glass

8. Oxygen-Packing Fraction

9. First Sharp Diffraction Peak as a Marker for Structural Change

10. First Sharp Diffraction Peak as an Indicator of the System Density

11. Summary and Future Perspectives

Chapter 14. Silicate Glasses Under Ultrahigh Pressure Conditions

1. Introduction

2. High-Pressure Brillouin Scattering Spectroscopy

3. Acoustic Wave Velocity Measurements for Silicate Glasses Under Ultrahigh Pressure

4. Recent Experimental and Theoretical Progress

5. Analogy Between Silicate Glasses and Silicate Melts in Si–O Coordination

6. Geophysical Implications

7. Concluding Remarks and Outlook

Chapter 15. Melts Under Extreme Conditions From Shock Experiments

1. Introduction: Shock Experiments and Magmas

2. Theory of Shock Compression

3. Experimental Methods

4. Shock Studies of Silicate Melts

5. Conclusion and Future Prospects

Chapter 16. Simulation of Silicate Melts Under Pressure

1. Introduction

2. Computational Methodology

3. Calculated Results and Analysis

4. Summary

Author Index

Subject Index

Copyright

Elsevier

Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands

The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom

50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

Copyright © 2018 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

ISBN: 978-0-12-811301-1

For information on all Elsevier publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Candice Janco

Acquisition Editor: Amy Shapiro

Editorial Project Manager: Tasha Frank

Production Project Manager: Prem Kumar Kaliamoorthi

Designer: Mark Rogers

Typeset by TNQ Books and Journals

List of Contributors

Paul D. Asimow,     California Institute of Technology, Pasadena, CA, United States

Suraj K. Bajgain

Louisiana State University, Baton Rouge, LA, United States

Florida State University, Tallahassee, FL, United States

Charlotte J.L. de Grouchy,     University of Edinburgh, Edinburgh, United Kingdom

Guillaume Fiquet,     Sorbonne Université, UMR CNRS 7590, Paris, France

Stephen F. Foley,     Macquarie University, Sydney, NSW, Australia

Dipta B. Ghosh,     Louisiana State University, Baton Rouge, LA, United States

Bijaya B. Karki,     Louisiana State University, Baton Rouge, LA, United States

Yoshio Kono,     Carnegie Institution of Washington, Argonne, IL, United States

Konstantin D. Litasov,     Sobolev Institute of Geology and Mineralogy SB RAS, Novosibirsk, Russia

Wim J. Malfait,     EMPA, Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland

Craig E. Manning,     University of California, Los Angeles, CA, United States

Takaya Mitsui,     National Institutes for Quantum and Radiological Science and Technology, Sayo, Hyogo, Japan

Motohiko Murakami,     ETH Zürich, Zürich, Switzerland

Keisuke Nishida,     The University of Tokyo, Tokyo, Japan

Zsanett Pintér,     Macquarie University, Sydney, NSW, Australia

Tatsuya Sakamaki,     Tohoku University, Sendai, Miyagi, Japan

Philip S. Salmon,     University of Bath, Bath, United Kingdom

Chrystèle Sanloup,     Sorbonne Université, Paris, France

Anton Shatskiy,     Sobolev Institute of Geology and Mineralogy SB RAS, Novosibirsk, Russia

Hidenori Terasaki,     Osaka University, Osaka, Japan

Max Wilke,     Universität Potsdam, Potsdam, Germany

Takashi Yoshino,     Okayama University, Misasa, Tottori, Japan

Preface

Magmas are essentially high-pressure objects, that are formed at depth, either in the present Earth beneath volcanoes, mid-ocean ridges, subduction zones, and hotspots, or in the deep magma ocean in the early Earth. Knowledge of magmas under pressure is therefore a prerequisite to understand the current and past evolution of the Earth. Efforts have been made to understand the structure and properties of magmas from natural observations and laboratory experiments. However, knowledge on the structure and properties of magmas under pressure has been limited, due to experimental difficulties, for example, the necessity of an in situ analytical probe under high-pressure and high-temperature conditions. In the past decade, new developments in high-pressure experiments, particularly with synchrotron X-ray techniques, have advanced the study of magmas under pressure. These new experiments have revealed significant changes to the structure and physical properties of magmas under pressure, which has significantly improved our understanding of the behavior of magmas in the Earth's interior. In addition, the high-pressure technologies developed in user facilities have expanded the community for the study of magmas under pressure.

This book aims to summarize recent advances in experimental technologies for studying magmas at high pressures, and what is known on the structure and properties of magmas under pressure, as a reference of current knowledge, present issues, and future perspectives for the study of magmas under pressure. This book consists of three sections: magmas in the Earth's interior (Chapters 1–4), advances in experimental studies of melts at high pressures (Chapters 5–11), and current knowledge of the structure and properties of magmas under pressure (Chapters 12–16). The first section (Chapters 1–4) reviews the melting of silicate and carbonate compositions at the pressure and temperature conditions in the Earth's mantle based on natural melt compositions and experimental petrology inputs. The second section (Chapters 5–11) introduces the latest experimental technologies for studying melts at high-pressure and high-temperature conditions. The third section (Chapters 12–16) reviews current knowledge on the structure and properties of magmas at high pressures based on experimental and theoretical approaches.

The knowledge of the structure and properties of magmas at high pressures is of benefit to a wide range of Earth and planetary scientists for discussing and modeling magmatism in the various tectonic settings in the Earth's interior. In addition, the information of the recent high-pressure experimental technologies will be valuable for various scientists, engineers, and students not only for studying magmas in Earth science but also for studying liquids, glasses, and other amorphous solids in physics, chemistry, and material sciences as well as engineering.

We would like to thank the chapter authors for their valuable contributions. We acknowledge Amy Shapiro, Tasha Frank, and Anita Mercy Vethakkan for their editorial support.

Yoshio Kono

Chrystèle Sanloup

Part 1

Magmas in the Earth's Interior

Outline

Chapter 1. Primary Melt Compositions in the Earth's Mantle

Chapter 2. Carbon-Bearing Magmas in the Earth's Deep Interior

Chapter 3. The Influence of Pressure on the Properties and Origins of Hydrous Silicate Liquids in Earth's Interior

Chapter 4. Melting in the Earth's Deep Interior

Chapter 1

Primary Melt Compositions in the Earth's Mantle

Stephen F. Foley, and Zsanett Pintér     Macquarie University, Sydney, NSW, Australia

Abstract

Primary melts are the melt compositions in equilibrium with their source at the time of extraction from that source. Many mantle-derived melts are derived chiefly from peridotite, but have additional components in the source that originate as recycled crust and sediments, or as ultramafic rocks formed by solidification of migrating mantle melts. Most melts originate between 40 and 120  km depth, with exceptions up to 250  km, particularly around and beneath cratons, where fluxed by volatile components. Melts of dry peridotite are varieties of basalt with increasing MgO and alkalies toward high pressures: they range from picrite and komatiite at high degrees of melting to basanite and nephelinite at small melt fractions. More alkaline compositions, including melilitite, kimberlite, lamprophyre, and carbonatite, require H2O and CO2 in the source. Water promotes melts richer in silica than basalt, whereas CO2 causes SiO2-poor melts. At high pressures, primary melts may be transitional between silicate and carbonate melt compositions. Volatile components depress the melting point of peridotite by 150–400°C relative to dry melting conditions, and melting in oxidized conditions occurs at lower temperatures than in reduced conditions or with H2O alone. Incipient melts are widespread, universal precursors to more voluminous melt production and cause modification and evolution of the lithosphere.

Keywords

CO2; H2O; Mantle; Primary melts; Redox; Volatile components

Chapter Outline

1. Introduction

2. Melting of Mantle Peridotite

2.1 Melting of Peridotite Without Volatile Components

2.2 Melting of Peridotite in the Presence of Water

2.3 Melting in the Presence of Carbon Dioxide

2.4 Melting of Peridotite With H2O and CO2

2.5 Melting in Reducing Conditions

3. Storage and Availability of Volatile Components in the Mantle

3.1 Nominally Anhydrous Minerals

3.2 Hydrous Phases

3.3 Carbon-Bearing Phases

4. Melting of Rocks Other Than Peridotite in the Mantle

4.1 Recycling of Ocean Crust Basalt/Gabbro

4.2 Ultramafic Cumulates From the Ocean Crust and Arcs

4.3 Serpentinites: Hydrothermally Altered Ultramafic Rocks

4.4 Clastic Sedimentary Material

4.5 Carbonate Sedimentary Material

4.6 Ultramafic Rocks Originating From Solidification of Transient Incipient Melts

5. Geodynamic Variations in the Sources and Compositions of Melts

5.1 Oceanic Lithosphere

5.2 Continental Lithospheric Mantle

5.3 Cratonic Lithosphere

5.4 Subduction Zones

5.5 Shallow Convergent Zones

Acknowledgments

References

1. Introduction

The term primary magma has a long history that predates the advent of high-pressure equipment to study the melting of mantle rocks. At the time of the introduction of the end-loaded piston–cylinder apparatus (Boyd and England, 1958), which initiated abundant experimental studies of mantle melting, Turner and Verhoogen (1960) lamented that the term primary magma was frequently used but seldom defined. This haziness remained through the 1960s, with imprecise definitions such as partial melting of the mantle, yielding primary magmas (O'Hara, 1965), and liquids formed by processes of partial melting or complete melting of mantle rock (Green and Ringwood, 1967), which moved to the surface without further modification (Thompson, 1974). Presnall's (1979) simple formulation would be acceptable for almost all petrologists: primary magma will refer to a magma as it exists immediately after separation from its source region. Here, we refine this by explicitly including a requirement expedient for experimental petrologists, and which they have followed implicitly for years (Wyllie, 1987)—namely, equilibrium: a primary melt is the melt composition in equilibrium with its source at the time of extraction from that source. This is distinct from a parental melt, which is the starting point for an observed crystal fractionation series, but need not be primary, and from a primitive melt, which may have been modified since extraction from its source. Note that the emphasis of these definitions is on the melts' unmodified, pristine state rather than on the identity of the mantle rock that melts: most experimental studies assume the mantle to be a simple, four-phase peridotite consisting of olivine, orthopyroxene, clinopyroxene, and an aluminous phase: garnet, spinel, or plagioclase depending on pressure.

Partial melting on the modern Earth is almost entirely restricted to a magmatic zone between 40 and about 250  km depth. Most melting occurs in the uppermost 140  km, ranging from decompression melting beneath midocean ridges to picrites in large igneous provinces. Melting deeper than 140  km occurs only where promoted by volatile components. The scope of this chapter is restricted to melting of the mantle with and without volatile components: we do not consider melting deeper than 250  km (see Chapters 2 and 4), major melting on the hot early Earth, or details of melting in subduction zones (Chapter 3).

First H2O, and later CO2, were added to peridotite melting studies, whereas reduced conditions in which H2O  +  CH4 fluids are stable remain little studied. Volatiles have two important effects: firstly a strong depression of melting temperatures, which results in a broad incipient melting regime (Green and Falloon, 1998), and secondly, a large expansion in the compositional range of melts. These melt compositions are only now in the process of being staked out.

Heterogeneity in the sources of mantle-derived melts was a major topic in the last quarter of the 20th century, and was initially implicitly understood to refer to trace element and isotope variations in peridotite, or the presence of additional minor phases such as amphibole or carbonate in peridotite (Oxburgh, 1964; McGetchin and Besancon, 1973). It is increasingly recognized that peridotite is not the only relevant rock, but that others, probably present as dikes or blocks in the source, play a role in many melts that may nevertheless fall under the above definition of primary melts: the melt is in equilibrium with the source at the time it is extracted, but the source is not a single, homogeneous rock type. Peridotite is usually one of the rocks, but the other may result from recycling of oceanic crust or continental material or from the migration and solidification of melts within the mantle to form other, usually ultramafic, rocks. Eclogites, representing recycled ocean crust, are the most studied of this suite of alternatives, whereas the number of melting studies of ultramafic rocks (e.g., Foley et al., 1999; Kogiso et al., 2004) and sedimentary rocks (e.g., Nichols et al., 1994; Hermann and Spandler, 2008; Thomsen and Schmidt, 2008) is increasing. This has led to a recent crescendo of experiments that mix two rock types in single experiments to characterize the reaction between melt and adjacent rocks (e.g., Yaxley and Green, 1998; Mallik et al., 2015; Wang et al., 2017). These reaction processes need to be understood, as they may take place entirely at temperatures below the melting point of peridotite (Foley, 1992a), resulting in a large range of melts that pre-empt major melting in specific geodynamic environments. The composition and action of volatiles and recycled materials depends strongly on the past and present geodynamic environment and on the mechanisms of storage of the volatiles in the mantle. Hydrous and carbon-bearing minerals have stability fields that make them more stable in some environments than in others.

2. Melting of Mantle Peridotite

The concept that basaltic melts originate by partial fusion of peridotite stems from Bowen (1928), who made this suggestion based on an estimate of mantle peridotite composition that now appears very crude (MgO 18.2  wt%, Mg# 67.1) and even fails to meet modern criteria for the definition of peridotite at all (containing >42% Opx, only 13% olivine and 20% feldspar; Washington, 1925). Thirty-five years later, peridotite compositions were much better known; natural peridotite compositions with Mg# (100  Mg/(Mg  +  Fe)) around 90 (O'Hara and Mercy, 1963) were very close to model mantle compositions such as pyrolite (Green and Ringwood, 1963).

2.1. Melting of Peridotite Without Volatile Components

The melt compositions produced by dry melting of peridotites are summarized and projected onto the base of the basalt tetrahedron and as a function of pressure and temperature in Fig. 1.1. These compositions were charted by early studies at pressures up to 3.5  GPa (Green, 1970, 1971) and have been more precisely determined with improved techniques since then. Most melt compositions are silica-saturated, falling in the triangle olivine–orthopyroxene–feldspar (Ol-Opx-Fs), with the exception of melts at very shallow depths (<25  km), which are quartz-normative (i.e., they lie to the quartz side of the Fs-Opx line in Fig. 1.1A), and of initial melts at high pressures (Falloon and Green, 1988), which are nepheline-normative. With increasing degree of melting, the melt compositions move toward the Ol-Opx line, corresponding to increasing MgO content (Jaques and Green, 1980; Falloon and Green, 1988), but vectors trend at acute angles to the Fs-Opx line, meaning that the melts change their degree of SiO2-saturation only slightly (Fig. 1.1A).

General trends with increasing pressure are apparent, with increasing MgO and decreasing SiO2, expressed as the degree of SiO2-saturation, being the most significant (Fig. 1.1). Initial melts are nepheline (Ne)-normative at >1  GPa, but move into the Ol-Opx-Fs field as melting progresses (Figs. 1.1 and 1.2). There are still very few melt compositions determined for pressures above 3.5  GPa: first indications suggested extremely MgO-rich melts with >30  wt% MgO at all pressures from 5 to 14  GPa (Takahashi, 1986), whereas more recent determinations indicate 18%–20% MgO for 10% melting at 4  GPa, increasing to 23%–25% MgO at 15%–20% melting and 6–7  GPa (Walter, 1998). Melts are picritic to komatiitic at high degrees of melting at all mantle depths greater than about 70  km (Fig. 1.2; Takahashi, 1986; Falloon and Green, 1987; Walter, 1998). The reason for these changes is that olivine is a peritectic phase at low pressures, meaning that the proportion of this low-SiO2, high-MgO phase increases in the residue, so that melts show the opposite trend (Walter, 1998). With increasing pressure, the ratio of olivine/Opx in the residue decreases steadily, allowing more MgO to enter the melt. The stability of garnet in the residue also increases, so that the Al2O3 content of melts drops from above 12  wt% at 3  GPa to <6  wt% at 7  GPa (Falloon and Green, 1988; Walter, 1998).

Figure 1.1  Compositions of primary melts produced by dry melting of peridotite (pyrolite). (A) Ovals indicate initial melts; numbers in them are pressures in GPa. Blue arrows (black in print version) show progression of melt compositions away from alkaline compositions with increasing degree of melting. Colored backgrounds indicate SiO 2 -undersaturated ( pink ) (dark gray in print version), SiO 2 -saturated ( green ) (light gray in print version), and SiO 2 -oversaturated ( yellow ) (white in print version) melt compositions. (B) Melt compositions shown as rock names as a function of pressure and temperature. Again, pink (dark gray in print version) shows alkaline, SiO 2 -undersaturated compositions close to the solidus ( red line ) (black line in print version) at higher pressures. After (A) Falloon, T.J., Green, D.H., 1987. Anhydrous partial melting of MORB pyrolite and other peridotite compositions at 10 kbar: implications for the origin of primitive MORB glasses. Mineral. Petrol. 37, 181–219; (B) Jaques, A., Green, D., 1980. Anhydrous melting of peridotite at 0–15 kb pressure and the genesis of tholeiitic basalts. Contrib. Mineral. Petrol. 73, 287–310.

Figure 1.2  Melt types produced as a function of degree of melting of peridotitic mantle at three different pressures. Melt compositions at very low melt fractions are uncertain; more alkaline compositions are controlled by volatile components. Modified after Green, D.H., 1970. The origin of basaltic and nephelinitic magmas. Trans. Leic. Philos. Lit. Soc. 64, 28–54.

Melt compositions at low degrees of melting are controversial due to the difficulty of analyzing small glass areas whose composition may be substantially falsified by modification by crystallization during the quenching phase at the end of experiments (Green, 1976). There is thus a poorly defined zone of initial melts (Fig. 1.1A), which also partly reflects variations in the fertility of peridotite compositions in different experimental studies. Estimates of melt compositions in this zone were initially addressed by synthetically increasing the amount of fusible components by subtracting 40% olivine from the peridotite compositions, which also facilitated microscopic identification of the phases coexisting with melt (Green, 1973a,b). More exact knowledge of melt compositions was later improved by two reversal techniques. Sandwich experiments, in which a melt composition measured in, or estimated from, a first melting experiment, is equilibrated with peridotite at the same pressure–temperature conditions in a second experiment (Stolper, 1980; Fujii and Scarfe, 1985). An alternative reversal technique uses melt traps of diamond or carbon spheres (Hirose and Kushiro, 1993; Baker and Stolper, 1994) to provide larger, more measurable melt areas, following a technique developed for experiments with fluids (Ryabchikov et al., 1989). Together, these reversal experiments indicated that the first melts are rich in Na2O (Falloon et al., 1997, 2008; Robinson et al., 1998), and that those from more fertile sources will be alkaline (nepheline-normative) at pressures at which more depleted peridotites produce melts with no normative nepheline (Falloon et al., 1997, 2001; Robinson et al., 1998). Near solidus partial melts become rich in Na2O and K2O, whereby sodium has a stronger effect of causing SiO2-undersaturated melt compositions than K2O (Kushiro, 1975; Gupta and Green, 1988; Conceição and Green, 2000). The position of the beginning of melting line in Fig. 1.1 is thus fuzzy because of both uncertainties in melt composition and of bulk source composition.

These results appear to confirm early suggestions that nephelinites and perhaps melilitites are the most alkaline melts that can be produced by melting of peridotite (Fig. 1.2; Green, 1971). This leaves many magma compositions such as kimberlites and some lamprophyres completely unexplained. The limit at nephelinite is consistent with trace element studies (Frey et al., 1978), which suggested that mantle peridotite could not produce the enrichment in incompatible trace elements seen in alkaline melts unless the source had experienced enrichment of trace elements by metasomatism.

2.2. Melting of Peridotite in the Presence of Water

Although successful for the explanation of voluminous basalt melts, melting of dry peridotite cannot account for rarer melt types that are richer in alkalies and show appreciable contents of the volatile components H2O and CO2. The first experimental studies of the effect of H2O on melting of peridotites (Green, 1973b; Millhollen et al., 1974; Mysen and Boettcher, 1975) showed that the melting temperature was depressed through hundreds of degrees and that amphibole is stable at the solidus. However, because of the worsened quenching behavior, melt compositions were difficult to ascertain and results were controversial: the position of the solidus and the composition of melts remains a contentious issue (Grove et al., 2006; Till et al., 2012; Green et al., 2014). The current status is summarized in Fig. 1.3, in which the solidus for peridotite + H2O is compared to that for dry and other volatile-bearing conditions.

Figure 1.3  Comparison of the melting curves of peridotite for dry conditions (CMAS, dry TQ, MPY, HPY after Green and Falloon, 1998 ; Presnall et al., 2002 ) and with volatiles in the system. Green line (dark gray thin line in print version)   =   peridotite   +   CO 2 (after Falloon and Green, 1989 ; dashed line extension from Dasgupta and Hirschmann, 2006 ); blue line (light gray in print version)   =   peridotite   +   H 2 O (HZ1 with ∼200   ppm H 2 O; Green et al., 2010 , 2014 ; Kovács et al., 2012 ); heavy red line (heavy black in print version)   =   oxidized solidus (CO 2 + H 2 O; Foley et al. (2009) . purple line (mid gray in print version)   =   reduced solidus (CH 4 + H 2 O), which follows the H 2 O-undersaturated curve to 3   GPa and then moves to higher temperatures as H 2 O activity decreases (after Green and Falloon, 1998 ). Thin gray lines are estimated solidi for different H 2 O activities with reduced C–O–H ( Taylor and Green, 1988 ). Increasing redox conditions can depress the relevant melting curve as shown by the black arrow from the reduced to the new oxidized solidus.

The water-undersaturated melting curve at <3  GPa is defined by the stability of pargasitic amphibole. The position of this curve in Fig. 1.3 is approximate, because amphibole may be considerably more stable in bulk compositions with lower Mg# and higher alkali contents (Kushiro, 1970; Millhollen et al., 1974; Wallace and Green, 1991). At pressures higher than the stability limit of amphibole, the melting temperature drops sharply, as in many conditions the rock is now saturated in water, because the nominally anhydrous minerals (NAMs) accommodate a maximum of only 200  ppm H2O (Kovács et al., 2012). In rocks with higher K2O, phlogopite will become stable at the solidus, and this defines the slope of the peridotite + H2O curve at higher pressures (Wyllie, 1977).

and affects melt compositions strongly due to its depolymerizing action (Wasserburg, 1957), resulting in SiO2-undersaturated melts rich in normative olivine that may contain ∼25  wt% H2O. Solidus curves for dry peridotites vary with the degree of fertility (gray–black lines in Fig. 1.3): for example, at 1  GPa, Hawaiian pyrolite (∼1225°C) lies at the lowest temperature, ∼50°C below midocean ridge basalt (MORB) pyrolite (∼1275°C), with the depleted Tinaquillo peridotite slightly higher still (∼1300°C) (Fig. 1.3). These compositional differences are small compared to the effect of H2O.

2.3. Melting in the Presence of Carbon Dioxide

Carbon speciation in the upper mantle is controlled by the oxygen fugacity (Frost and Wood, 1997; Luth, 1999). Under oxidized conditions, CO2-bearing fluids, melts, and carbonate minerals are stable, whereas at reduced conditions CH4 and solid carbon as graphite or diamond are stable. Initial experiments showed carbon dioxide–rich melts to be unquenchable, and so CO2 was initially investigated by determining its solubility in simple systems and in volcanic melts (Eggler, 1976; Wyllie and Huang, 1976), followed by thorough theoretical predictions of its effect on melting of the mantle (Eggler and Holloway, 1977; Wyllie, 1978).

Carbon dioxide has a similar effect on the mantle solidus to H2O, depressing the solidus of carbonated peridotite by several hundred degrees (Falloon and Green, 1989; Dalton and Presnall, 1998a; Dasgupta and Hirschmann, 2006; Dasgupta et al., 2007). The solidus curve sinks to much lower temperatures above ∼2  GPa due to the stabilization of solid carbonate in place of CO2 vapor, and describes a steeper slope than dry melting conditions (Dasgupta et al., 2013). The effect of CO2 on melt compositions is, however, very different from that of H2O. Carbon dioxide increases the stability of Opx relative to olivine by increasing the degree of polymerization of the silicate network in the melt (Mysen, 1983; Ni and Keppler, 2013). Both CO2 and high Na2O and K2O contents drive melt compositions to lower SiO2 contents (Fig. 1.4; Kushiro, 1975; Gupta and Green, 1988), resulting in melts that are strongly SiO2-undersaturated. This may explain melilititic melts that could not be derived from peridotite either dry or with H2O alone (Brey and Green, 1977), and K-rich undersaturated rocks such as kamafugites, ultramafic lamprophyres (Wendlandt and Eggler, 1980), and carbonatites. Some studies found a progression from carbonatitic melts with 5–6  wt% SiO2 to carbonated silicate melts with 30–35  wt% SiO2 (Dalton and Presnall, 1998b; Dasgupta and Hirschmann, 2007; Ghosh et al., 2014), but the transition from carbonatitic to carbonate-bearing silicate melt remains controversial (Dalton and Presnall, 1998b; Moore and Wood, 1998; Lee and Wyllie, 2000; Brey et al., 2008; Litasov et al., 2008).

2.4. Melting of Peridotite With H2O and CO2

Many more experiments have been done on the melting of peridotite with either H2O or CO2 alone than with both H2O and CO2, despite these being less realistic for the real world. Following experiments on the single-volatile systems in the 1970s, elaborate phase diagrams relevant to H2O  +  CO2 were produced, which saw melting curves for mixed volatiles intermediate between the pure H2O and CO2 curves (e.g., Eggler and Holloway, 1977; Wyllie, 1978). However, these proved incorrect once experiments with both H2O and CO2 became available, which showed that the melting point was depressed even more than with H2O alone, and that a field for carbonate melt occurs between 2 and 3  GPa (Fig. 1.3; Wallace and Green, 1988). The carbonatite melt field is wide where it coexists with amphibole in peridotite with small amounts of both CO2 (2.3  wt%) and H2O (0.3  wt%). Melts in the carbonatite melt field between 2 and 3  GPa (Fig. 1.3) are strongly enriched in Na2O and many highly incompatible elements (Wallace and Green, 1988; Sweeney et al., 1992; Dasgupta and Hirschmann, 2006). At >3  GPa, above the stability limit of pargasitic amphibole, the solidus moves to lower temperatures, but a thin zone with carbonate-rich melt persists to at least 6  GPa (Fig. 1.3; Foley et al., 2009) and can be extrapolated to higher pressures as shown in Fig. 1.3, assuming that the melting curve remains parallel to the better-determined position of peridotite  +  CO2 without H2O (Dasgupta and Hirschmann, 2006; Ghosh et al., 2014).

Figure 1.4  Effect of various volatile species on the olivine/orthopyroxene cotectic curve in the simple system nepheline–quartz–forsterite. H 2 O moves the curve toward the quartz corner, favoring SiO 2 -richer melt compositions, whereas CO 2 has the opposite effect. The positions for reduced fluid mixtures are due to the reduced activity and presence of minor dissolved CO 2 for the CH 4 + H 2 O curve. Pressure   =   2.8   GPa. Compiled from Taylor, W., Green, D., 1987. The petrogenetic role of methane: effect on liquidus phase relations and the solubility mechanism of reduced C–H volatiles, In: Mysen, B.O. (Ed.), Magmatic Processes: Physicochemical Principles. Geochemical Society, University Park, PA, pp. 121–138; Gupta, A.K., Green, D.H., 1988. The liquidus surface of the system forsterite-kalsilite-quartz at 28 kb under dry conditions, in presence of H2O, and of CO2. Mineral. Petrol. 39, 163–174; Green, D.H., Falloon, T.J., 1998. Pyrolite: a ringwood concept and its current expression. In: Jackson, I. (Ed.), The Earth's Mantle: Composition, Structure, and Evolution, pp. 311–378.

At low melt fractions, melt compositions progress gradually from carbonatitic at <10% melting to carbonate-bearing silicate melts akin to ultramafic lamprophyres at 20%–30% melting (Fig. 1.5; Foley et al., 2009), agreeing with some of peridotite  +  CO2 (Dalton and Presnall, 1998b; Dasgupta et al., 2007). In water-rich mixtures, this may encounter immiscibility between carbonate and silicate melts at intermediate degrees of melting (Kiseeva et al., 2012), but this has not been confirmed for peridotitic compositions.

Fig. 1.6 summarizes primary melt compositions that are predicted from experiments with various CO2  +  H2O mixtures, and compares them to the products of dry peridotite melting. Although melts of dry peridotite become more alkaline with increasing pressure at low degrees of melting (Fig. 1.2), nephelinites appear to represent the limit of extractable melts (3%–5% melting accounting for crystal growth effects; Faul, 1997), and even melilitites require both CO2 and H2O in their sources (Brey and Green, 1977). Differing ratios of CO2/H2O cause melts resembling alkaline magma types that only rarely reach the surface as low-volume dikes or diatremes. CO2-rich conditions may give rise to carbonatites and aillikites (carbonate-rich ultramafic lamprophyres; Rock, 1986), whereas intermediate CO2/H2O produces kimberlite, alnöite, and melilitite (Fig. 1.6). Minettes and lamproites cannot be produced in CO2-rich conditions. Their high contents of potassium and incompatible elements and abundant hydrous phases indicate that they are produced from mixed-rock sources and cannot be generated from peridotite alone (Foley, 1990, 1992b).

Figure 1.5  Changing melt compositions produced from peridotite in oxidizing conditions (with H 2 O and CO 2 ) with increasing temperature above the solidus at a pressure of 5   GPa. Initial melts are dolomitic carbonatite with <5   wt%

SiO2, whereas at 30% melting, melts resemble ultramafic lamprophyres (aillikites) with around 30  wt% SiO2. The main changes are an increase in SiO2 and decrease in CaO. After Foley, S., Yaxley, G., Rosenthal, A., Buhre, S., Kiseeva, E., Rapp, R., Jacob, D., 2009. The composition of near-solidus melts of peridotite in the presence of CO2 and H2O between 40 and 60 kbar. Lithos 112, 274–283.

Figure 1.6  Schematic representation of the melt types produced at very low degrees of melting from peridotite with mixed volatile components. Carbonatite and aillikite can originate in oxidized melting conditions, whereas minette (alkaline lamprophyres) and lamproite require H 2 O-rich conditions that may be more reducing. Most of the melts in the volatile-present area will also require mixed source rocks containing nonperidotitic rock types.

The most important effects of H2O and CO2 are that they define the extent of the incipient melting region (Green and Falloon, 1998) and exert the main control on melt compositions within it. Melts from the incipient melting region are rarely seen at the Earth's surface but nevertheless play an important role in the evolution of the mantle, because they may exist and be mobile over a temperature interval of several hundred degrees before major melting begins. At pressures of 4–6  GPa, melts with H2O and CO2 resemble aillikites, which have SiO2 contents intermediate between carbonatites and melilitites, which cannot be formed at lower pressures because they fall in a miscibility gap (Hamilton et al., 1979; Kjarsgaard and Peterson, 1991).

2.5. Melting in Reducing Conditions

Under reducing conditions, C–O–H fluids do not contain CO2, but consist of CH4  +  H2O  +  H2 (±C2H6) in proportions that vary widely over a relatively small range in oxygen fugacity close to and above the iron–wüstite buffer (IW; Fig. 1.7). Under these conditions, carbon is rendered essentially immobile during melting, existing as graphite or diamond with only small amounts (≤0.2  wt%) dissolved in the melt (Taylor and Green, 1987; Mysen et al., 2009). The active role of H2O is thus much greater, so that CH4 can be essentially viewed as a dilutant that reduces the activity of H2O. This expresses itself in the phase relationships, with the Ol-Opx cotectic lying between the positions for H2O-saturated and dry conditions (Fig. 1.4). Because of the variable CH4/H2O ratios, the position of the melting curve varies over several hundred degrees at pressures of 3  GPa and greater (Fig. 1.3; Taylor and Green, 1988). Higher melting temperatures correspond to more reducing conditions with higher CH4/H2O ratios, because aH2O is lower. The curve in Fig. 1.3 assumes more reducing conditions with increasing depth, indicating that the melting temperature is higher than in more oxidizing mantle conditions (Taylor and Green, 1988; Litasov et al., 2014).

The compositions of melts in the presence of mixed CH4  +  H2O volatiles are still virtually unknown (dissolved in the melt in addition to OH− (Taylor and Green, 1988; Odling et al., 1997; Stagno and Frost, 2010). The curve favored by Green and Falloon (1998) is probably realistic for conditions within the Earth because the fO2, and thus H2O/CH4, are expected to decrease toward greater depths (Frost and McCammon, 2008). At depths of 150  km, the fO2 is estimated to be ∼IW + 1, where carbonate will not be stable, and graphite/diamond will coexist with H2O  +  CH4 fluids.

The rapid variation in aH2O with pressure gives rise to an important mechanism for the generation and exhaustion of melts, redox melting (Taylor and Green, 1987), and its opposite, redox freezing (Rohrbach and Schmidt, 2011). With increasing fO2, the H2O/CH4 ratio increases, rapidly depressing the melting point due to the increase in aH2O, which may cause melting (lower box in Fig. 1.7A). Later, a second redox melting mechanism was identified for more oxidized conditions in which stored diamond or graphite is released into melts as carbonate (upper box in Fig. 1.7A): this operates because the solidus for H2O  +  CO2 is below that of H2O alone (Foley, 2011). These melting and freezing mechanisms will be important in geodynamic situations in which rocks with contrasting oxidation states become juxtaposed, such as during the rejuvenation and rifting of ancient cratonic lithosphere (Tappe et al., 2006; Foley, 2008, 2011) and in the collision of numerous small continental and oceanic plates (Prelević et al., 2013). Similar effects will occur more gradually in upwelling asthenosphere, where oxidation of reduced carbon species will be followed by melting of carbonate accompanying the reduction of the Fe³+ in residual phases (Stagno and Frost, 2010; Stagno et al., 2013). This will lead to SiO2-undersaturated basaltic melt production as the dissolved carbonate components are released in the form of CO2 at pressures of ∼2  GPa (Stagno et al., 2013).

Figure 1.7  Volatile speciation (C–O–H system) as a function of oxygen fugacity (fO 2 ). (A) shows fluid compositions on a scale from H-rich (left) to C-rich (right); (B) shows the same information in terms of percentages of species. For compositions in the light gray area in (A), fluids lie on the carbon saturation curve thick blue line (gray in print version) and are CO 2   +   H 2 O mixtures ( green area ) in oxidized conditions, and CO 2 -free, dominated by CH 4   +   H 2 O in reduced conditions ( light violet area ). These are separated by a zone in which fluids consist almost exclusively of H 2 O ( horizontal line labeled water-maximum). Redox melting occurs at the fO 2 of the boxes in (A): the lower box is the oxidation of CH 4 to H 2 O, and the upper box is the oxidation of C to CO 2 . Modified after Foley, S.F., 2011. A reappraisal of redox melting in the Earth's mantle as a function of tectonic setting and time. J. Petrol. 52, 1363–1391.

3. Storage and Availability of Volatile Components in the Mantle

The distribution of volatiles in the Earth's mantle has been a hot topic because of their great impact on partial melting (Green, 1973a; Gaetani and Grove, 1998; Hirose and Kawamoto, 1995; Hirschmann, 2006; Green et al., 2010, 2014), viscosity and seismic properties of the mantle (Dixon et al., 2004; Jung and Karato, 2001), electrical conductivity (Karato, 1990; Wang et al., 2008), and deformation characteristics (Hirth and Kohlstedt, 1996; Mizukami et al., 2004).

Upper mantle–derived magmas carry indications of the volatile signatures of their source region. Usually, unaltered mantle-derived magmas and primary melt inclusions contain up to several weight percent volatiles: these and analyses of mantle rocks imply complex volatile source region distributions ranging from 30 to 7000  ppm (Fig. 1.8). Fertile lherzolite may store ∼100–200  ppm volatiles (mainly H2O; Green et al., 2010), whereas ocean island basalts indicate higher volatile contents. There are many candidates for the storage sites of these volatiles in the upper mantle. Besides NAMs from the mantle, accessory hydrous and carbon-bearing phases include mica, amphiboles, and carbon in reduced (graphite, diamond, SiC) or oxidized (carbonate) forms.

3.1. Nominally Anhydrous Minerals

The most established techniques for determination of water (H2O, H+, and OH− in the structure) in the NAMs from the mantle are Fourier transform infrared spectrometry (FTIR) and secondary ion mass spectrometry (SIMS) (Rossman, 2006), now supplemented by new methods such as μ-elastic recoil detection analysis (Raepsaet et al., 2008) and Raman spectroscopy (Thomas et al., 2009).

Based on a compilation of natural and experimental studies (the Pannonian Uniform Lithospheric Database, puli.mfgi.hu), nominally anhydrous minerals could store several hundreds to thousands ppm water and up to 3–5  ppm CO2 in their structure. Olivine is the most abundant constituent in the lithospheric mantle (∼60% modal) and the most studied mantle mineral for water contents. Natural and experimental studies (Ingrin et al., 2013; Kovács et al., 2012; Peslier et al., 2010), however, have revealed that olivine does not preserve the water signature of the source region but loses (or gains) water during sampling by the host magma, retaining a maximum of ∼100  ppm water (Kovács et al., 2012, and references therein). Experiments have demonstrated that the storage capacity could be ∼200  ppm wt% H2O (Hauri et al., 2006; Tenner et al., 2009; Green et al., 2010; Demouchy and Bolfan-Casanova, 2016) in the lithosphere and asthenosphere, and possibly ∼4000–8900  ppm in the deeper mantle (12–14  GPa; Smyth et al., 2006; Ferot and Bolfan-Casanova, 2012). The H2O uptake in olivine is affected by the presence of mixed volatiles, with H2O contents in olivine dropping by half as CO2/H2O increases from 1.5 to 4 (Sokol et al., 2013).

The H2O content of pyroxenes (ortho- and clinopyroxene) is considered to be unmodified by sampling (values ranging from ∼1.9 to 2.4 (i.e., Hirth and Kohlstedt, 1996; Ingrin and Skogby, 2000; Peslier et al., 2002), but as natural and experimental datasets expand, scattering between ∼1 and 5 (Fig. 1.9) has been found.

Figure 1.8  Maximum bulk volatile (H 2 O   +   CO 2 ) storage capacity of upper-mantle peridotite, assuming no melt present. Sources estimates for mantle-derived magmas from [1] Dixon et al. (1995) , [2] Danyushevsky et al. (2000) , and [3] Dasgupta and Hirschmann (2010) . [4] Storage capacity estimates of representative ultramafic assemblies based on the water storage capacity of NAMs after Kovács et al. (2012) . Calculated H 2 O storage capacities for upper-mantle rocks based on experimental studies: [5] pyrolite ( Ringwood, 1962 ), [6] HZ1 harzburgite ( Green et al., 2010 ), [7] Hirschmann et al. (2009) , [8] Tenner et al. (2012) , [9] Ardia et al. (2012) . Deep mantle rocks at 12–14   GPa: [10] Smyth et al. (2006) and [11] Ferot and Bolfan-Casanova (2012) .

(∼4.5–5.0) than the main trend (Bonadiman et al., 2009; Yu et al., 2011; Hao et al., 2012; Warren and Hauri, 2014; Pintér et al., 2015). Natural mantle NAMs show lower concentrations than experimental studies (Kovács et al., 2012), which may indicate that volatile storage in the upper mantle is below the maximum (Kovács et al., 2012; Demouchy and Bolfan-Casanova, 2016). Furthermore, the impact of partial melting, metasomatism, melt–rock interaction, and refertilization processes on the storage capacity have not yet been investigated in detail. Therefore variations in composition, mafic-ultramafic character, mineralogy, major and trace element chemistry, fO2, fH2O, etc., could potentially cause strong variation in the substitution mechanisms of H2O in NAMs, strongly influencing the storage capacity.

Figure 1.9  Partitioning of water between coexisting orthopyroxene and clinopyroxene. Black lines indicate 1:1, 1:2, 1:3, 1:4, and 1:5 ratios for DH 2 O (Cpx/Opx) for reference. Data sources for natural mantle xenoliths from different geodynamic settings: sub-arc ( Peslier et al., 2002 ; Peslier and Luhr, 2006 ; Falus et al., 2008 ); rift regions ( Li et al., 2008 ; Denis et al., 2013 ); abyssal ( Warren and Hauri, 2014 ); continental ( Bell and Rossman, 1992 ; Yang et al., 2008 ; Bonadiman et al., 2009 ; Hao et al., 2014 ; Pei et al., 2015 ; cratonic settings: Xia et al., 2010 , 2013 ; Baptiste et al., 2012 ; Kolesnichenko et al., 2017 ); and xenoliths from regions of young and/or active extension ( Bonadiman et al., 2009 ; Hao et al., 2012 ; Yu et al., 2011 ; Warren and Hauri, 2014 ; Pintér et al., 2015 ). Experimental partitioning trends are shown as thick lines with ranges ( Hauri et al., 2006 ; Tenner et al., 2009 ; Kovács et al., 2012 ; Novella et al., 2014 ).

3.2. Hydrous Phases

The volatile storage capacity in the mantle lithosphere and up to at least 8  GPa pressure are mostly controlled by the stability of hydrous and carbon-bearing species. Amphibole and mica are the most common hydrous phases in upper mantle rocks. Pargasitic amphibole is the dominant hydrous phase in peridotite at <3  GPa, increasing the volatile storage capacity toward ∼0.6  wt% (Green et al., 2010, 2014) and locally even ∼1–2  wt%. Pargasitic amphibole stability is compromised by jadeitic clinopyroxene, and is thus limited to 90–100  km in Ca-rich, fertile mantle assemblages (Green et al., 2010, 2014; Pirard and Hermann, 2015; Mandler and Grove, 2016). However, in clinopyroxene-free rocks such as dunite and harzburgite, amphibole is richer in Na and K and poorer in Ca (magnesiokatophorite), which increases its stability toward higher pressure (∼4.5  GPa; Pirard and Hermann, 2015). The stability of amphibole is also increased by fluorine and influenced by bulk H2O, Na, and K activities, Na/Ca ratio, and fO2 and Cl contents (Foley, 1991; Konzett and Ulmer, 1999; Mandler and Grove, 2016). As a result, amphiboles may break down over a large pressure and temperature range: calcic amphiboles from ∼2.5  GPa (Fumagalli and Poli, 2005) to ∼4  GPa (Mandler and Grove, 2016), whereas alkali amphiboles (K-richterite) may be stable up to ∼14  GPa (Sudo and Tatsumi, 1990; Konzett et al., 1997; Trønnes, 2002). Amphiboles are most likely crystallization products from infiltrating volatile-rich asthenospheric melts (Menzies et al., 1985; Foley, 1992a), although some may precipitate from H2O-rich fluids (Lloyd et al., 1987; Erlank et al., 1987). Melts cause the formation of pargasitic amphiboles by refertilization of peridotites (Wallace and Green, 1991; Niida and Green, 1999), whereas a wider compositional range may crystallize in distinct assemblages in veins (Pirard and Hermann, 2015).

The breakdown of pargasitic amphibole may be associated with several geophysical discontinuities, including the lithosphere–asthenosphere boundary beneath the oceanic crust (Green and Liebermann, 1976; Sifré et al., 2014), in young continental lithosphere (Posgay et al., 1995; Tašárová et al., 2009), and the mid-lithospheric discontinuity in thicker continental lithosphere (Thybo, 2006; Selway et al., 2015; Hansen et al., 2015).

Phlogopitic mica is a common minor phase in upper-mantle rocks, occurring in peridotites (Dawson and Powell, 1969; Delaney et al., 1980; Canil and Scarfe, 1989) and more commonly as distinct assemblages such as mica–amphibole–rutile–ilmenite–diopside (MARID) and mica–pyroxenites (Dawson and Smith, 1977; Lloyd et al., 1987; Gamble et al., 1988). Experiments show phlogopite to be the main hydrous phase in peridotite between 3  GPa, where it replaces amphibole, and 6–8  GPa, where it reacts with pyroxene components to form potassic richterite and garnet (Sudo and Tatsumi, 1990; Konzett and Ulmer, 1999; Trønnes, 2002). It may be stable to 9–12  GPa in some assemblages (Trønnes, 2002).

It is important to note that incipient melts may be dominated by the melting of the hydrous phases, particularly amphibole, just above the solidus, producing basanitic/nephelinitic (from Ca-amphibole) and lamproitic (from K-richterite) melts with compositions that broadly resemble the amphiboles (Foley et al., 1999). However, residual NAMs nevertheless retain some water until the temperature where anhydrous melting begins.

In parts of the mantle that have experienced recent subduction, additional phases may introduce large amounts of water into the mantle. These include serpentine (Ulmer and Trommsdorff, 1995) and, at depths up to 300  km, phengite and lawsonite (Scambelluri and Philippot, 2001; Schmidt and Poli, 1998), which may contain 8–15  wt% H2O.

3.3. Carbon-Bearing Phases

The amount of carbon dissolved in the major phases of peridotite (NAMs) is very low—usually below detection limits of FTIR and SIMS—with maxima increasing from 2 to 12  ppm at 11  GPa (Rosenthal et al., 2015; Shcheka et al., 2006). Carbon is stored in distinct carbon-bearing phases that are dependent on the redox state: carbonate at high fO2, graphite or diamond at low fO2, and carbides under reducing conditions in the presence of an Fe metallic phase.

Experiments and theoretical considerations indicate that carbonates (12–14  wt% C) should be stable as dolomite in peridotites in oxidizing conditions at pressures above 2  GPa (Eggler and Baker, 1982). Dolomite is replaced by magnesite at higher pressures (Wyllie, 1978; Falloon and Green, 1989), whereas calcite should exist only in nonperidotitic veins. Evidence in natural rocks for carbonate stability in the mantle has been controversial, with rare occurrences described from xenoliths, some of which are interpreted as evidence for metasomatic reactions involving carbonate melts (Ionov et al., 1993; Ionov, 1998; Laurora et al., 2001). In many cases, carbonate melts infiltrating the lithosphere are exhausted by the formation of clinopyroxene and the loss of CO2 (Yaxley et al., 1991; Rudnick et al., 1993).

Carbonated melts may be stable at depths of ∼150  km (Stagno and Frost, 2010), but the pressure limit in the mantle is more likely limited by the decrease in oxidation state with increasing depth than by the maximum pressure limit of carbonate. Carbonates may occur at much deeper levels in subduction zones, where mixed and diverse oxidation states permit the transport of carbonated sediments and ocean crust, the latter even beyond the solidus (Yaxley and Brey, 2004; Grassi and Schmidt, 2011). Melts produced in this environment will be K-rich Mg–Ca–Fe carbonatites that are effective metasomatic agents in the overlying mantle wedge.

At low fO2 around IW, corresponding to much of the lower cratonic lithosphere (Woodland and Peltonen, 1999; Frost and McCammon, 2008), diamond will be the main repository for carbon. The release of carbon from diamond will be principally by oxidation by infiltration of melts or juxtaposition of oxidized blocks (Foley, 2008, 2011).

The occurrence of carbides in mantle rocks has been controversial because of the prohibitively low oxygen fugacity required to stabilize them in mantle peridotite (Mathez, 1995; Ulmer et al., 1998). Silicon carbide (moissanite) is known as inclusions in diamond (Moore and Gurney, 1989; Otter and Gurney, 1989; Klein-BenDavid et al., 2004) or associated with the crystallization of polycrystalline diamond (Jacob et al., 2011). Many of these occurrences may be related to redox melting or freezing processes. More recently, evidence for extremely low fO2 has been accumulating, with moissanite found associated with chromites in ophiolites (Trumbull et al., 2009; Griffin et al., 2016a) together with nitrides and native metals. These may be caused by extremely local reduction by CH4  +  H2 fluids (Griffin et al., 2016b), possibly related to immiscibility between H2O and H2 (Bali et al., 2013).

4. Melting of Rocks Other Than Peridotite in the Mantle

Following the definition of primary magma as the melt composition in equilibrium with its source at the time of extraction, the composition and mineral assemblages in the source are not specified. Mantle sources have historically been implicitly assumed to be peridotite, but a more modern and nontraditional view includes mixed rocks in the mantle source, and fits both this definition and mounting geochemical evidence in igneous rocks and their minerals.

The Earth's mantle is a complex and heterogeneous mixture of mostly ultramafic and mafic rock types and is the time-integrated product of partial melting and recycling of crust over more than 4000  million  years. The rocks other than peridotite are so diverse that no compositional label effectively covers them, and so we refer to them simply as nonperidotitic mantle rocks (NPMR) in the following text. The NPMR probably mostly occur surrounded by peridotite, and many have lower melting points than peridotite due to silicic and aluminous compositions or to high concentrations of volatile components. Volatile components are concentrated in melts during partial melting, so that rocks that result from refreezing of melts in the mantle commonly contain hydrous or carbonate minerals, and as a consequence have melting points that are almost universally lower than that of peridotite. Many of the melts produced directly in the mantle from NPMR may never be seen at the surface, and yet they contribute in diluted form—often swamped by later major melting of peridotite—to magmas that are. These components are the source of much of the variation in isotope and trace element signatures in igneous rocks.

Also important for the effectiveness of metasomatism and mass transfer by later melts is the scale of blocks and dikes of different composition within the mantle, which ranges from centimeter-scale veins in xenoliths (Irving, 1980; Ancochea and Nixon, 1987) to blocks tens of meters wide in massif peridotites (Ceuleneer and Le Sueur, 2008; Tilhac et al., 2016). Partial melting of large blocks has the potential to produce large volumes of mobile melt, and this may be more common in subduction environments than in the lithospheric blocks usually sampled in massif peridotites because of the tectonic juxtaposition of crustal and mantle rocks during near-surface subduction processes.

It is important to acknowledge and emphasize that improved understanding of the role of mixed rock regions in the mantle is imperative not only for our comprehension of the mechanisms of mantle melting, but it will also have important knock-on effects for all areas of research involved with the mantle, many of which are currently locked into the paradigm of uniform peridotite mantle. Examples are geophysics, deep mantle mineral physics, deformation studies, and the deep carbon cycle (Biellmann et al., 1993; Karato and Jung, 1998; Tommasi et al., 2000). Regional studies will also need to make allowance for the influence of NPMR in the mantle.

Here, we give a brief overview of the effects of melting or dehydration of the most likely NPMR, and then consider variations in the sources and melt compositions with geodynamic setting.

4.1. Recycling of Ocean Crust Basalt/Gabbro

Basalts and gabbros of the ocean crust transform to eclogite at depths of 30–50  km, which are the best studied of all subducted rocks due to evidence for their deep recycling and involvement in oceanic magmatism (Hofmann and White, 1982; Hirschmann and Stolper, 1996; Sobolev et al., 2007). Slabs often reach transition zone depths (Irifune and Ringwood, 1987; van der Hilst et al., 1998; van Mierlo et al., 2013), and some cases may be recycled through the lower mantle (Fukao and Obayashi, 2013).

Partial melts of eclogites are SiO2-oversaturated, with near-solidus melts ranging from andesitic to dacitic (Yaxley and Green, 1998; Pertermann and Hirschmann, 2003). Alkali contents of melts are very sensitive to K and Na contents in the rock and to melting of accessory phases (Spandler et al., 2008; Rosenthal et al., 2014):