ABSTRACT
Title of Dissertation: Earth?s Radiogenic Heat Production,
and Models for the Composition of
the Deep Continental Crust
Laura Sammon
Doctor of Philosophy, 2022
Dissertation Directed by: Professor William F. McDonough
Department of Geology
Much of the continental crust, the 40+ km thick plates of rock that make up the
outer shell of our planet, is inaccessible to us living on its surface. Thus its composition
is a mystery. We lack the technology to sample it directly at depths past 5 km, aside
from a few deep (expensive) drill holes, so we must come up with a clever alternative for
establishing its composition.
The deep crust, the lower two-thirds of the continent, serves as a supporting root.
When continents collide, they make mountain ranges, or when pulled apart they make rift
valleys and basins. The composition of the deep crust, and specifically its silica (SiO2),
molecular water, and heat producing element (HPE: K, Th, U) contents, directly influ-
ence the crust?s rheology during tectonic events and its potential for deadly earthquakes.
Its chemical makeup is the sum of 4.5 billion years of crustal evolutionary processes that
continuously shape and reshape the platform upon which society sits. An accurate de-
scription of the deep crust, however, requires careful integration of many different data
sources.
My research combines geochemistry with thermodynamics, geophysics, mineral-
physics, seismology, and even particle physics to produce self-consistent models for the
crust?s composition. Using thermodynamic calculations, I generate densities and seismic
sound wave speeds from a range of chemical compositions. Matching these forecasted
models to Earth?s seismic and gravity data allows me to translate the deep crust?s physical
properties into chemical compositions on both the regional and the global scale. Impor-
tantly, by quantifying not only the compositions, but also the uncertainties and the misfit
in these results, I can better define the differences between competing models for crust
deformation and evolution.
Charting the distribution of Earth?s geochemical resources has led to our collab-
orations with particle physicists, who need our expertise to determine the frequency of
radioactive decay and therefore the amount of HPE decay emissions (known as geoneu-
trinos) in the crust; this geoneutrino flux is the background signal in their nuclear physics
experiments. Their global flux measurements constrain our models for heat production
and the amount of radiogenic energy that heats the Earth ? which provides power to man-
tle convection, plate tectonics, and the destruction and creation of more continental crust.
Our main sources of data are threefold. First, we have critically compiled geochemical
analyses of >10,000 rock samples from pre-existing literature (Earthchem.org and affil-
iates). Second, we use geophysical data provided by sources such as the United States
Geological Survey, the Earthscope USArray, and others to determine which of our geo-
chemical samples could produce Earth?s observed seismic and density signals. Third, we
partner with particle physicists in the United States, Canada, Italy, Japan, and China to
jointly interpret data from three international geoneutrino detectors.
By focusing on Earth as a whole system we seek a comprehensive understanding of
its natural hazards and resources. Using multidisciplinary constraints, my goal is to build
compositional models of the continental crust, with quantifiable uncertainties, that can be
applied regionally and at larger scales. These findings will provide predictive insights on
the strength and response of the continents when subjected to the dynamic processes of
plate tectonics.
Earth?s Radiogenic Heat Production,
and the Composition of
the Deep Continental Crust
by
Laura Sammon
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
2022
Advisory Committee:
Professor William F. McDonough, Chair/Advisor
Associate Professor Nicholas C. Schmerr
Assistant Professor Megan Newcombe
Dr. Ondr?ej S?ra?mek
Professor Gregory Sullivan, Dean?s Representative
? Copyright by
Laura Sammon
2022
Dedication
To my friends, my family at Maryland - every success you made sure was celebrated, and
every failure you made sure was not in vain.
ii
Acknowledgments
I was told at the beginning of graduate school (I cannot remember the source) that
you can only get two out of the three options: a good advisor, good research, or good
coworkers. Luckily, I never really had to choose.
Though Bill started out as my academic advisor, he accumulated many titles over
the years. He has been a graduate school mentor and colleague, yes, but also someone
to confide in, to share my hopes and fears with. He was, at times, a PR representative, a
sous chef, a taxi driver, an emergency contact, a travel agent, and a supplier of absolutely
terrible ?dad jokes?.
Many other collaborators helped push this research forward as well. I would like to
acknowledge especially Walter Mooney at the USGS for his insight, data, and financial
support for a trip to California that acted as a turning point in my research projects and
my morale. I also want to thank many others: Ingrida Semenec from the SNO+ team,
an incredible physicist and persistent friend; Wiesen Shen and Wolfgang Szwillus, who
taught me enough so that I could pretend to be a geophysicist; Scott Wipperfurth, who was
an invaluable mentor in the beginning of my program, especially when my advisor was
halfway across the globe; and everyone who has read and reviewed my papers over the
years, especially Roberta Rudnick, Derek Schutt, Derrick Hasterok, and Jamie Connolly,
all of whom I bothered on multiple occasions.
iii
I cannot even begin to name all of the graduate students and professors in the Geol-
ogy Department at Maryland who have had such an incredible impact on my life. Never
did I expect such congeniality from professors, though it is not surprising to find among
geologists. The geology graduate student body at UMD was more of a dream team. Ev-
ery new year, I made more friends, and it became harder and harder to watch old friends
graduate and move on. I would not trade you all for the shiniest rock or biggest pot of
coffee in the world.
And thank you so much to my loving family and support team. Throughout the
PhD process you have stayed by my side, on my side when I needed it most. You believed
when I doubted, you held me while I cried, and you listened (or tried very, very hard to)
when I rambled about neutrinos, codes, presentations, and rocks in general.
My research was financially supported by NSF grants, the Ann G. Wylie Fellow-
ship, a GSA graduate student research grant, ESSIC travel grants, the UMD Geology
Department and Graduate School.
iv
Table of Contents
Dedication ii
Acknowledgements iii
Table of Contents v
List of Tables viii
List of Figures x
Chapter 1: Introduction 1
1.1 Neutrino Geoscience and Crust Composition . . . . . . . . . . . . . . . . 2
1.1.1 Geoneutrino Detection . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Deep Crust Composition Models . . . . . . . . . . . . . . . . . . . . . . 6
1.2.1 Relating Chemical Composition to Seismic Observations . . . . . 7
Chapter 2: Quantifying Earth?s the radiogenic heat budget 11
2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Lithospheric Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5.1 Near-Field and Far-Field Lithosphere . . . . . . . . . . . . . . . 20
2.5.2 Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Importance of the Near-Field Lithosphere Model . . . . . . . . . . . . . 25
2.7 Heat Flux Constraints on Lithospheric Models . . . . . . . . . . . . . . . 28
2.8 The Future of Neutrino Geoscience . . . . . . . . . . . . . . . . . . . . . 31
2.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.10 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Chapter 3: A Geochemical Review of Amphibolite, Granulite, and Eclogite Fa-
cies Lithologies: Perspectives on the Deep Continental Crust 36
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.4 The Art and Science of Deep Crustal Modeling . . . . . . . . . . . . . . 40
3.5 Datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
v
3.5.1 Amphibolite, Granulite, and Eclogite . . . . . . . . . . . . . . . . 43
3.5.2 Potential Biases . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.6 Major Element Compositions . . . . . . . . . . . . . . . . . . . . . . . . 50
3.6.1 SiO2, MgO, FeO, and the Daly Gap . . . . . . . . . . . . . . . . 50
3.6.2 The Constancy of Al and Ga . . . . . . . . . . . . . . . . . . . . 58
3.7 Minor and Trace Element Composition . . . . . . . . . . . . . . . . . . . 62
3.7.1 Rare Earth Elements . . . . . . . . . . . . . . . . . . . . . . . . 63
3.7.2 Heat Producing Elements . . . . . . . . . . . . . . . . . . . . . . 66
3.8 Distributions that Trend, Periodically . . . . . . . . . . . . . . . . . . . . 69
3.9 A Basalt, by Any Other Name . . . . . . . . . . . . . . . . . . . . . . . 73
3.10 Constructing the Continental Crust . . . . . . . . . . . . . . . . . . . . . 76
3.11 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Chapter 4: Localized Lower Crustal Composition from Joint-Inversion of Geo-
chemical and Geophysical Datasets 88
4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.2 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.4 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.4.1 Compositional Modeling of the Lower Crust . . . . . . . . . . . . 93
4.4.2 Geologic Setting . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.5 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.7.1 Lower Crust Composition . . . . . . . . . . . . . . . . . . . . . 104
4.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Chapter 5: The composition of the deep continental crust inferred from geochem-
ical and geophysical data 115
5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.2 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.4.1 Model Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.4.2 Model Uncertainties . . . . . . . . . . . . . . . . . . . . . . . . 126
5.4.3 Quality, Expense, and Time: Global vs. Local Models . . . . . . . 128
5.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.5.1 Empirical Composition-Velocity Trends . . . . . . . . . . . . . . 129
5.5.2 Deep Crustal Density . . . . . . . . . . . . . . . . . . . . . . . . 133
5.5.3 Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.6.1 SiO2 and Overall Deep Crustal Composition . . . . . . . . . . . . 139
5.6.2 CaO and Sr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.6.3 Heat Production and Moho Heat Flux . . . . . . . . . . . . . . . 148
5.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
vi
Chapter 6: Conclusions and Future Work 160
6.1 Key conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
6.2 Future geoneutrino project locations . . . . . . . . . . . . . . . . . . . . 163
Appendix A:Supporting Information for Chapter 2 166
A.1 Full electron antineutrino flux equation . . . . . . . . . . . . . . . . . . . 166
A.2 Heat production from K, Th, and U decay . . . . . . . . . . . . . . . . . 167
Appendix B:Supporting Information for Chapter 3 168
B.1 Eu Anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
B.2 Fluid Mobile Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
B.2.1 K, Rb, and Cs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
B.2.2 La/Th and Th/U . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
B.2.3 Alkalis and Alkaline Earth Metals, in General . . . . . . . . . . . 173
B.3 High Field-Strength Elements . . . . . . . . . . . . . . . . . . . . . . . . 174
B.4 Transition Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
B.5 Halogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
B.6 Supplementary Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Appendix C:Supporting Information for Chapter 4 208
C.1 Compositional Maps and Uncertainties . . . . . . . . . . . . . . . . . . . 208
vii
List of Tables
2.1 Borexino Models for the upper crust in the NFL, bulk calculated Signal,
and Radiogenic Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.1 Major element compositions for amphibolite, granulite, and eclogite . . . 55
3.2 Major oxide compositions from amphibolites, granulites, and eclogites
compared to other basalts . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.3 Continental crust elemental abundances . . . . . . . . . . . . . . . . . . 77
3.4 Comparison of expected deep crustal physical properties . . . . . . . . . 85
4.1 Uncertainties associated with deep crustal modeling parameters . . . . . . 99
4.2 Colorado Plateau SiO2 content . . . . . . . . . . . . . . . . . . . . . . . 106
4.3 Northern Basin and Range SiO2 content . . . . . . . . . . . . . . . . . . 106
4.4 Southern Basin and Range SiO2 content . . . . . . . . . . . . . . . . . . 107
4.5 Major element concentrations in the lower crust of the southwestern US . 107
4.6 Summary of lower crustal seismic properties and mineralogies . . . . . . 108
5.1 Tectonic regimes by surface area . . . . . . . . . . . . . . . . . . . . . . 125
5.2 SiO2 concentration for different tectonic regimes . . . . . . . . . . . . . 137
5.3 Calculated middle and lower crust bulk composition . . . . . . . . . . . . 139
5.4 Comparison of continental crust composition estimates . . . . . . . . . . 147
5.5 Heat production calculation parameters . . . . . . . . . . . . . . . . . . . 153
5.6 Heat production in the continental crust . . . . . . . . . . . . . . . . . . 154
A.1 Heat production and geoneutrino flux results . . . . . . . . . . . . . . . . 166
A.2 Radionuclide heat production . . . . . . . . . . . . . . . . . . . . . . . . 167
B.1 Amphibolite Facies Lithologies . . . . . . . . . . . . . . . . . . . . . . . 180
B.2 Granulite Facies Xenoliths . . . . . . . . . . . . . . . . . . . . . . . . . 183
B.3 Post-Archean Granulite Terrains . . . . . . . . . . . . . . . . . . . . . . 186
B.4 Archean Granulite Terrains . . . . . . . . . . . . . . . . . . . . . . . . . 189
B.5 Eclogite Facies Xenoliths . . . . . . . . . . . . . . . . . . . . . . . . . . 192
B.6 Eclogite Facies Terrains . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
B.7 Amphibolite Facies Lithologies . . . . . . . . . . . . . . . . . . . . . . . 198
B.8 Granulite Facies Xenoliths . . . . . . . . . . . . . . . . . . . . . . . . . 199
B.9 Post Archean Granulite Facies Terrains . . . . . . . . . . . . . . . . . . . 200
B.10 Archean Granulite Facies Terrains . . . . . . . . . . . . . . . . . . . . . 201
viii
B.11 Eclogite Facies Xenoliths . . . . . . . . . . . . . . . . . . . . . . . . . . 202
B.12 Eclogite Facies Terrains . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
ix
List of Figures
2.1 Geoneutrino signal strength vs. distance from Italian detector . . . . . . . 16
2.2 U concentration in common crustal rocks . . . . . . . . . . . . . . . . . . 21
2.3 Schematic drawing of the location of the Borexino experiment and its
Near-Field lithosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Near-Field Lithosphere vs. Mantle geoneutrino signal trade-off . . . . . . 23
2.5 Two different Near-Field Lithosphere models . . . . . . . . . . . . . . . 24
2.6 Total geoneutrino signal vs. lithospheric signal . . . . . . . . . . . . . . . 26
2.7 Geological and surface heat flow maps of the central Italian peninsula . . 27
2.8 Map of geoneutrino detectors . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1 Mg# vs. SiO2 for medium- to high- grade metamorphics . . . . . . . . . 52
3.2 Al Saturation Index for metamorphic lithologies . . . . . . . . . . . . . . 61
3.3 Rare earth element median values . . . . . . . . . . . . . . . . . . . . . . 64
3.4 Rare Earth element abundances in amphibolite, granulite, and eclogite . . 65
3.5 Elemental abundances in amphibolites . . . . . . . . . . . . . . . . . . . 71
3.6 Elemental abundances in granulites . . . . . . . . . . . . . . . . . . . . . 72
3.7 Major element abundances compared to basalt . . . . . . . . . . . . . . . 74
3.8 MORB normalized rare earth element abundances in amphibolite, gran-
ulite, and eclogite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.9 Bulk crustal elemental abundances from different models . . . . . . . . . 83
4.1 Sample collection and seismometer locations for this study . . . . . . . . 91
4.2 Perple X predicted velocities and densities vs. SiO2 . . . . . . . . . . . . 95
4.3 Modeling process flow chart . . . . . . . . . . . . . . . . . . . . . . . . 97
4.4 Geophysical properties of the southwestern US . . . . . . . . . . . . . . 102
4.5 Composition probability histograms for three southwestern US geological
provinces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.6 SiO2 abundance in the southwestern US . . . . . . . . . . . . . . . . . . 103
4.7 Seismic velocities vs. isochemical velocity prediction . . . . . . . . . . . 105
4.8 Schematic of geological structure and composition in the southwestern US 113
5.1 Weighted area proportion of tectonic regimes . . . . . . . . . . . . . . . 119
5.2 Global tectonic regimes and seismic velocity profile locations . . . . . . . 121
5.3 Conceptual illustration of how velocity distributions are used to predict
crust compositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.4 Vp vs. SiO2 for amphibolite and granulite facies lithologies . . . . . . . . 130
x
5.5 Vs vs. SiO2 for amphibolite and granulite facies lithologies . . . . . . . . 131
5.6 Vp/Vs vs. SiO2 for amphibolite and granulite facies lithologies . . . . . . 132
5.7 Density vs. depth for different types of continental crust . . . . . . . . . . 134
5.8 Global SiO2 concentration in the lower crust . . . . . . . . . . . . . . . . 136
5.9 Global SiO2 concentration at various depths . . . . . . . . . . . . . . . . 138
5.10 Global CaO concentration and uncertainty in the lower crust . . . . . . . 140
5.11 Global Sr concentration and uncertainty in the lower crust . . . . . . . . . 141
5.12 Global U abundances and uncertainties in the lower crust . . . . . . . . . 149
5.13 Heat production in the layers of the continental crust . . . . . . . . . . . 151
5.14 Moho heat fluxes for different tectonic provinces . . . . . . . . . . . . . 155
5.15 Global Moho heat flux . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.16 Crustal U concentration vs. Moho heat flux . . . . . . . . . . . . . . . . 157
B.1 Eu anomalies in amphibolite, granulite, and eclogite . . . . . . . . . . . . 169
B.2 Fluid mobile element concentrations, amphibolite and granulite . . . . . . 171
B.3 La/Th vs. Th/U . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
B.4 Fluid mobile elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
B.5 Location map and SiO2 content of the metamorphic sample datasets . . . 204
B.6 Location map and Mg# of the metamorphic sample datasets . . . . . . . . 205
B.7 Nb vs. Ta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
B.8 Ln(Th) vs. Ln(U) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
C.1 Vs as a function of temperature . . . . . . . . . . . . . . . . . . . . . . . 208
C.2 Uncertainty from SiO2 calculations . . . . . . . . . . . . . . . . . . . . . 209
C.3 Distribution of mafic oxides MgO + FeO . . . . . . . . . . . . . . . . . . 210
C.4 Uncertaintise in MgO + FeO . . . . . . . . . . . . . . . . . . . . . . . . 211
C.5 Relative uncertainty in deep crustal composition . . . . . . . . . . . . . . 212
C.6 Median Vp from the joint model . . . . . . . . . . . . . . . . . . . . . . 213
C.7 Median Vs from the joint model . . . . . . . . . . . . . . . . . . . . . . 214
C.8 Median Vs from seismic data . . . . . . . . . . . . . . . . . . . . . . . . 215
C.9 Example seismic inversions . . . . . . . . . . . . . . . . . . . . . . . . . 215
xi
Chapter 1: Introduction
The processes that make Earth a unique and habitable planet are tied to its compo-
sition, past and present. The starting materials Earth inherited during its formation were
crucial to early planetary differentiation (McDonough et al., 2020), and the resources that
are left today still fuel its mantle convection, plate tectonics, and the geodynamo. This
fuel is primarily expressed as heat. Earth?s total heat budget of 46 ? 3 terrawatts (TW,
1012 watts) (Jaupart et al., 2016), is a combination of primordial (kinetic) heat from plan-
etary accretion and of radiogenic heat. With two sources of fuel, Earth?s internal engine
is like that of a hybrid car. Unfortunately, we do not have a fuel gauge to tell us what
proportion of heat is primordial or radiogenic. While humans are unlikely to be around
when both of Earth?s fuel tanks reach ?empty?, knowing Earth?s heat evolution and future
trajectory tells us about the abundance of valuable elements, and helps us understand the
cycles of heating and cooling in other rocky planets ? and their potential for hosting life.
The radioactive decay of heat producing elements (HPEs: U, Th, K) contributes
anywhere between 10 and 30 TW of energy to Earth (Javoy, 1995; McDonough and Sun,
1995; Turcotte and Schubert, 2014), and is the only process that has added significant
heat to the Earth?s interior since roughly 4.4 billion years ago. Th and U in particular
occupy unique positions in the study of Earth?s heat production and composition. First,
1
their radioactive isotopes, 232Th, 238U, and trivial amounts of 235U, account for >80%
of Earth?s radiogenic heat generation (McDonough et al., 2020). Second, Th and U are
refractory elements which condensed out of the solar nebula in conserved ratios to 26
other refractory elements (Bellini et al., 2013). Accurately and precisely determining
the concentration of Th and/or U in the bulk Earth therefore allows us to calculate the
abundances of those other refractory elements as well, which include Ca, Al, Nb, and the
economically valuable rare earth elements.
Because of the chemical differentiation of Earth?s layers, the highest concentration
of HPEs is locked in Earth?s continental crust. Unfortunately, we who live on the surface
can only directly access the uppermost parts of the crust (the deepest drill hole, Kola
Superdeep Borehole, was abandoned at 12 km depth). The deep continental crust and
the mantle are inaccessible to traditional geological sampling methods. This dissertation
advances the study of Earth?s composition by reconciling geoneutrino fluxes measured
by particle physicists with geochemical, geophysical, and seismological observations, to
produce an internally consistent compositional model of Earth. The overarching goal
of this research is to build an Earth compositional model with quantifiable uncertainties
using an interdisciplinary approach, and to highlight the need for further collaboration
among the geosciences.
1.1 Neutrino Geoscience and Crust Composition
Geoneutrinos are a direct probe into Earth?s interior. Geoneutrinos, abbreviated ??e,
are electron antineutrinos emitted during the radioactive decay of Earth?s natural isotope
2
systems. These antineutrinos carry with them unique energy tags, allowing us to identify
each ??e?s radioactive parent isotope. Measurements of geoneutrino fluxes can therefore
directly, quantitatively constrain the number of Th and U decays occurring inside Earth.
Geoneutrinos are nearly massless, chargeless particles. Once generated, these ghost-
like particles leave Earth to traverse the universe. Due to their elusive nature (interaction
cross-section of < 10?47m?2), we only detect about one ??e per 1019 that pass through a
one kiloton detector, or roughly 10 ??e per year. Modern geoneutrino detectors have the
capability to measure ??e from Th and U isotopes, which carry enough energy to pass the
instruments? 1.8 MeV reaction threshold. Because the decay rates of 232Th and 238U are
well-known (McDonough et al., 2020), and the ratio of Th to U is (relatively) constant
(Wipperfurth et al., 2018), we can calculate the absolute abundances of these HPEs from
geoneutrino measurements.
Geoneutrino detectors, however, cannot measure directional information, meaning
that the ??e from HPE decays in Earth?s deep mantle look the same as ??e?s from the crust.
Earth?s total geoneutrino emission signal is a combination of geoneutrinos produced from
all of Earth?s silicate layers. Negligible geoneutrinos are emitted from Earth?s core (Wip-
perfurth et al., 2018). The number of ??e observed by physicists is therefore:
Stotal = Scrust +Smantle (1.1)
The total signal, Stotal , is reported in Terrestrial Neutrino Units (TNU) to normal-
ize measurements among different detectors. One TNU equals one antineutrino per one
kiloton of scintillation fluid (1018 free protons) per year. Stotal is proportional to the con-
3
centrations of Th and U divided by the square of their distance (r?2) from the detector.
Although the mantle is Earth?s largest silicate reservoir, variations in the local crust?s con-
centrations of Th and U have the strongest effect on the ??e signal simply because the local
crust is closer to the detector (smaller r).
To understand how many geoneutrinos are coming from the mantle, and therefore
how much radioactive heat is left to power mantle convection, we must first determine
Scrust , a crustal correction for the abundance of Th and U. Chapter 2 of this dissertation
shows that slight differences in the modeled upper crustal correction vastly impact the
final calculation of radiogenic heat in the mantle. Chapter 2?s discussion highlights the
importance of understanding the geology local to each geoneutrino detector and demon-
strates why geoscientists need to take an active role in the calculation of the geoneutrino
signal. Chapters 3 ? 5 quantify the composition and uncertainties of the deeper portions
of the continental crust through a combination of geochemical data analyses and geo-
physical observations. Due to its inaccessibility and poorly constrained major and trace
element abundances, the composition of the deep continental crust has been a point of
debate among geoscientists for decades (see Dumond et al. (2018); Hacker et al. (2015)
for summaries of these debates). However, we combine independent information from
geochemical studies (Chapter 3) with petrological, geophysical, and seismological infor-
mation (Chapters 4, 5) to construct a more precise model for deep crustal composition.
Our joint geochemical-geophysical approach to determining the major element and HPE
content of the deep crust is effective on a local (Chapter 4) and a global (Chapter 5) scale,
both of which are necessary for calculating the local (Near-Field) and global (Far-Field)
contribution of Th and U to the total geoneutrino signal.
4
1.1.1 Geoneutrino Detection
Geoneutrinos are emitted during the decay of four primary nuclides, along with
energy, reported as heat (Q) in MeV:
40K?40 Ca+??+ ??e +Q(1.31MeV )
232T h??208 Pb+6? +4??+4??e +Q(42.6MeV )
235U ??207 Pb+7? +4??+4??e +Q(46.4MeV )
238U ??206 Pb+8? +6??+6??e +Q(51.7MeV )
As stated earlier, though, only ??e from the Th and U decay chains are energetic
enough to accomplish the inverse beta decay reaction (IBD, equation 1.2). In kiloton-
sized detectors of scintillating fluid, ?? +e collides with a free proton (p , hydrogen nucleus)
transforming it into a positron (e+) and a neutron (n). This reaction results in two flashes
of light that are correlated in time, space, and energy.
??e + p+? e++n (1.2)
The emitted positron promptly annihilates with an electron, producing two photons
summing to 1.022 MeV as the first detectable flash of light. About ?200 microseconds
later, the neutron thermalizes and is captured by a nearby proton to form a 2H+ deuterium
nucleus. This reaction produces another flash of light with a discrete energy of 2.2 MeV,
(the nuclear binding energy for deuterium). The two flashes of light and the separation
timer of 200 ?s allow for the unique identification of ??e with low background noise.
5
1.2 Deep Crust Composition Models
The deep continental crust?s chemical composition is central to the debate of crustal
formation, evolution, strength, temperature, and bulk composition. The inaccessible na-
ture of the deep crust forces reliance on analogue samples and modeling results to de-
termine its bulk composition. Medium- to high-grade continental metamorphic litholo-
gies serve as geochemical markers for the deep crust. These amphibolite, granulite, and
eclogite facies rocks are sampled through tectonically emplaced or eroded exposures of
metamorphic rock (terrains) or through deep crustal xenoliths that are rapidly carried to
the surface through volcanic eruptions. The difficulty remains in determining whether
these isolated occurrences of deep crustal material are representative of the deep crust as
a whole. On the other hand, geophysical surveying techniques, such as seismic veloc-
ity measurements, provide a means of measuring large portions of the deep crust in situ,
but only report on the physical properties of the deep crust rather than its chemical state.
However, a combination of both geochemical and geophysical analyses can place con-
straints on deep crustal composition. The physical properties of small-scale geochemical
samples, measured in laboratory experiments, can be compared to large scale observations
of the crust.
Christensen and Fountain (1975) were among the first to formally predict lower
crustal compositions from laboratory derived Vp velocities in granulite. By comparing
laboratory experiments to large seismic datasets, the authors identified average seismic
and compositional trends in the deep crust. Holbrook et al. (1992), Rudnick and Foun-
tain (1995), and successive publications offered interpretations of suites of seismic sur-
6
veys in the context of global crustal structure, incorporating even more petrological and
geochemical analyses of deep crustal samples. Following the work of Holbrook et al.,
Christensen and Mooney (1995) developed a model which addressed the global crustal
structure through synthesizing large seismic data sets. To combat sparse data coverage in
remote areas, such as Antarctica or central Africa, the models extrapolate seismic profiles
of various tectonic provinces to regions of similar crustal structure (e.g., island arcs, oro-
gens, Archean crust, etc.). The CRUST and LITHO family of models (Bassin et al., 2000;
Mooney et al., 1998; Pasyanos et al., 2014) has applied the same extrapolation method to
generate what are now the most prevalent of global seismic models for the lithosphere.
CRUST models define Vp, Vs, density, and thickness structures for different tectonic
settings by relating Vp field measurements to Vs and density through empirical seismic
observations. Recent iterations of similar techniques (e.g., Szwillus et al., 2019) more
explicitly outline their methods and quantify the statistical uncertainty in their models.
1.2.1 Relating Chemical Composition to Seismic Observations
Comparing geochemical measurements to geophysical observations requires trans-
lation between bulk rock composition and seismic velocities. A rock?s composition con-
trols its mineralogy, which in turn dictates its physical properties, such as its density,
compressibility, and shearibility. The bulk compressibility modulus (K), shear modulus
(?), and density of a rock define the velocities at which seismic waves travel through it:
?
K + 43 ?V p = (1.3)
?
7
?
?
V s = (1.4)
?
where Vp is compressional wave velocity and Vs is shear wave velocity.
Thermodynamic modeling softwares can use Voigt1-Reuss2-Hill3 (VRH) averag-
ing to build bulk rock compositions from individual mineral constituents. One such soft-
ware, Perple X (Connolly, 2005), predicts equilibrium mineral assemblages from bulk
rock compositions using Gibbs free energy minimizations. From these mineral assem-
blages, VRH averaging is used to generate bulk rock properties. Perple X derives mineral
phase relationships from experimental and empirical input databases, e.g. from the exper-
iments conducted by Holland and Powell (2004). Connolly (2005) offers an overview of
the software?s free energy minimization technique for calculating mineral assemblages.
Chapter 3 of this dissertation explains in detail how Vp and Vs are predicted for differ-
ent deep crustal lithologies, the temperature and pressure variations considered, and the
uncertainties associated with converting major oxide compositions to seismic velocities.
Chapters 4 and 5 then compare Perple X-derived seismic velocities to field observations
of Vp and Vs.
Trace elements, such as Th and U, cannot be directly derived from seismic proper-
ties. By definition, they are not major rock forming constituents and therefore cannot be
modelled as thermodynamic rock components. However, many trace elements have quan-
tifiable relationships to one or more major oxides. Chapter 3 analyzes the predictable
1W. Voigt, Lehrbuch der Krista Uphysik (B. B. Teubner, Leipzig, 1928), p. 739.
2A. Reuss, Z. Angew. Math. Mech. 9, 49 (1929)
3R. Hill, Proc. Phys. Soc. (London) 65,349 (1952)
8
behaviors between HPEs and SiO2. Chapter 5 then generates trace element maps as a
function of major oxide abundance using a simple bivariate probability analysis.
The goal of this project was to investigate Earth?s radiogenic heat production by placing
multiple, independent constraints on its composition. The research presented here focuses
specifically on creating joint geochemical-geophysical models for the deep continental
crust. The inaccessibility and heterogeneity of the deepest portions of the continental
crust compels us to reconcile datasets from different fields of geology and physics. By
incorporating geochemical, petrological, and geophysical datasets, and accounting for the
uncertainties within each, this dissertation lays one more brick in the foundation of a a
holistic Earth model.
In addition, our models of continental crust composition assess the concentration
of heat producing elements (K, Th, and U) in areas local to geoneutrino detectors and
throughout the Earth. The detection of geoneutrinos, ??e, emitted from within Earth is a
tool for directly measuring Earth?s heat production capacity and ultimately the abundance
of > 25 elements. These geoneutrino measurements, however, require accurate crustal
composition corrections. This document outlines a multidisciplinary approach to solving
for crustal composition, of the deep crust in particularly, first on the local, then on the
global scale.
Each new chapter?s specific undertakings depend on observations and techniques
developed in previous parts of the project. Chapter 2 outlines the importance of accurate
crustal composition models for geoneutrino and radiogenic heat production calculations.
9
In-depth geochemical characterization of amphibolite and granulite samples (Chapter 3),
including major element and HPE correlations, builds an extensive library of possible
deep crustal compositions. Thermodynamic modeling translates these compositions to
seismic velocities, which are then compared to both regional (Chapter 4) and global
(Chapter 5) seismic data. The conclusion to this dissertation (Chapter 6) discusses the
future of neutrino geoscience and crustal compositional modelling, including which av-
enues should be pursued next.
10
Chapter 2: Quantifying Earth?s the radiogenic heat budget
2.1 Overview
Earth?s internal heat drives mantle convection and plate tectonics. Measurements
of geoneutrino fluxes show what portion of that heat is radiogenic, as opposed to pri-
mordial. For a geoneutrino flux measurement to be interpreted into total radiogenic heat,
though, a lithospheric correction must be applied. An accurate geochemical understand-
ing of the rocks surrounding geoneutrino detectors is of the utmost importance. Here,
we show how two different local geological models around the Borexino detector in Italy
lead to two very different radiogenic heat results. We show that Borexino?s results agree
with other geoneutrino experiments when Italy?s regional geology is properly taken into
account.
[1] This chapter has been submitted to Earth and Planetary Science Letters as:
Sammon, L. G., & McDonough, W. F. (2022). Quantifying Earth?s radiogenic heat
budget.
[2] LGS and WFM contributed to the conceptualization of this project. LGS con-
structed the synthetic models. LGS and WFM wrote and edited this manuscript together.
11
2.2 Abstract
Earth?s internal heat drives its dynamic engine, causing mantle convection, plate
tectonics, and the geodynamo. These renewing and protective processes, which make
Earth habitable, are fueled by a primordial and radiogenic heat. For the past two decades,
particle physicists have measured the flux of geoneutrinos, electron antineutrinos emitted
during ?? decay. These ghost-like particles provide a direct measure of the amount of
heat producing elements (HPE: Th & U) in the Earth and in turn define the planet?s ab-
solute concentration of the refractory elements. The geoneutrino flux has contributions
from the lithosphere and mantle. Detector sensitivity follows a 1/r2 (source detector sep-
aration distance) dependence. Accordingly, an accurate geologic model of the Near-Field
Lithosphere (NFL, closest 500 km) surrounding each experiment is required to define the
mantle?s contribution. Because of its proximity to the detector and enrichment in HPEs,
the local lithosphere contributes ?50% of the signal and has the greatest effect on inter-
preting the mantle?s signal.
We re-analyzed the upper crustal compositional model used by Agostini et al. (2020)
for the Borexino experiment. We documented the geology of the western Near-Field re-
gion as rich in potassic volcanism, including some centers within 50 km of the detector.
In contrast, the Agostini study did not include these lithologies and used only a HPE-
poor, carbonate-rich, model for upper crustal rocks in the surrounding ?150 km of the
Borexino experiment. Consequently, we report 3? higher U content for the local upper
crust, which produces a 50% decrease in Earth?s radiogenic heat budget, when compared
to their study. Results from the KamLAND and Borexino geoneutrino experiments are at
12
odds with one another and predict mantle compositional heterogeneity that is untenable.
Combined analyses of the KamLAND and Borexino experiments using our revised local
models strongly favor an Earth with ?20 TW present-day total radiogenic power. The
next generation of geoneutrino detectors (SNO+, counting; and JUNO, under construc-
tion) will better constrain the HPE budget of the Earth.
2.3 Introduction
A combination of primordial and radiogenic energy drives Earth?s engine, with the
former coming from planetary accretion and the latter from decay of K, Th, and U. Our
planetary vehicle lacks a fuel gauge to define the amount of fuel left to power plate tec-
tonics, mantle convection, and the geodynamo. Defining the thermal evolution of the
planet gives insights into the cooling and crystallization history of the core, the temporal
variation in mantle viscosity, and the nature of the cosmic building blocks of the Earth.
With the dawn of geoneutrino detection (Araki et al., 2005), we now have the opportunity
to define the Earth?s radiogenic fuel budget, which in turn can specify the proportional
contribution of these heat producing elements (K, Th, U) in the crust and mantle.
Twenty years have passed since particle physicists began detecting the Earth?s emis-
sion of geoneutrinos (chargeless and near-massless particles emitted during ?? decay)
(Araki et al., 2005). The first generation of detectors (KamLAND in Japan and Borexino
in Italy) have reported their flux measurements and interpreted their data in the context
of an assumed geological model. The precision of the flux measurement (? ) continues
?
to improve with exposure time, as it follow counting statistics (? ? 1/ N, N=number
13
of observed events). The accuracy of the interpretation and its uncertainties depends on
the assumed geological model. To interpret the geoneutrino flux measurement, one uses
a detailed assessment of the Th and U abundances and distribution in the lithosphere sur-
rounding the detector (closest ?500 km, which typically contributes 40 to 50% of the
measured signal). A reference model is assumed for contributions from the remaining
global lithosphere and mantle, with the Earth?s core having negligible quantities of K,
Th, and U, and no significant contribution to the signal. Combined analyses of the results
from the KamLAND and Borexino experiments favor an Earth with?20 TW present-day
total radiogenic power (or a ?16 TW Earth for just Th and U power) (Bellini et al., 2021;
McDonough et al., 2020). This finding indicates that ?40% of the Earth?s estimated
power of 46 ? 3 TW (Jaupart et al., 2015) comes from radiogenic sources.
Controversy remains, however, regarding the assumed geological model used to
describe the local lithospheric contribution to the geoneutrino flux. For the lithosphere
surrounding the KamLAND detector the various geological models predicting the local
3D distribution of Th and U differ by a factor of 1.4, based on their reported geoneutrino
fluxes (Enomoto, 2006; Huang et al., 2013; Wipperfurth et al., 2020). In contrast, for the
Borexino detector the various predictions differ by a factor of 3 (Agostini et al., 2020;
Coltorti et al., 2011; Huang et al., 2013; Wipperfurth et al., 2020). The interpretation of
the regional geology is important for geoneutrino studies as it fundamentally influences
the final result, and the global abundances of Th and U.
The latest interpretation of geoneutrino data from the Borexino experiment (Agos-
tini et al., 2020) predicts a low contribution from their local crust to the overall geoneu-
trino signal. Consequently, their inferred mantle geoneutrino signal is high (?25 TW
14
from Th+U), as well as their calculation for the bulk Earth?s radiogenic power (?38 TW
from K+Th+U), with model uncertainties at ?34% (Agostini et al., 2020). This predic-
tion contrasts with other geoneutrino experiments (Agostini et al., 2015; Gando et al.,
2013) and numerous geochemical (Huang et al., 2013; Javoy et al., 2010; McDonough
and Sun, 1995, e.g.) and geophysical (Jaupart et al., 2016; Turcotte and Schubert, 2014,
e.g.) models for Earth. Agostini et al. (2020) places their upper limit of uncertainty at 51
TW of radiogenic heat production, which is outside of all geological observations.
Here we review the data for constructing a local geological model for the litho-
sphere immediately surrounding the Borexino detector. We evaluate the local geological
model used in Agostini et al. (2020) and compare it with competing models. We then
test whether such models are consistent with the known regional geology and heat flux
constraints. Using these findings, we identify the best local lithospheric models for the
Borexnio experiment. Relying on the same principles, we discuss the competing local
lithospheric models for the next generation of geoneutrino experiments.
2.4 Background
Neutrinos are weakly-interacting fundamental particles that stream freely through
matter, carrying information about their decay source. Detection of electron antineutrinos
(??e) is accomplished via the Inverse Beta Decay (IBD) reaction with a free protons (p):
??e + p? e++ n [n, neutron; e+, positron] with an energy threshold of Ethr?? = 1.8 MeV.e
This restriction allows detection of only the highest energy antineutrinos produced during
some of the ?? decays in the 238U and 232Th decay chains (Araki et al., 2005).
15
100
500
1000
5000
10000
Figure 2.1: The strength of a geoneutrino signal depends on the abundance of the emitter
(Th or U), and the 1/distance2 from the emitted to the detector, regardless of direction. A
detector in central Italy (Borexino) sees the strongest signal (yellow) from its immediate
surrounding geology and the weakest signal from the opposite side of Earth (pink). The
outer and inner core do not contribute to the geoneutrino signal and are grayed-out.
16
Distance from Detector  (km)
Earth?s total geoneutrino emission comes from the lithosphere and mantle, with the
number of ??e observed (i.e., S, signal) by physicists is therefore:
Stotal = Slithosphere +Smantle (2.1)
Stotal is reported in Terrestrial Neutrino Units (TNU) to normalize between detectors of
different sizes; 1 TNU equals 1 antineutrino detection per 1 kiloton of scintillation fluid
(1032 free protons) per year of exposure in a 100% efficient detector. Stotal is proportional
to the concentration of U times its sensitivity factor (?) and the same for Th, divided by
the square of their distance (r) from the detector:
?[U ]+? [T h]
Stotal ? 2 (2.2)r
Sensitivity factors account for a neutrino?s interaction cross-section, which scales with its
discrete energy, and accounts for the higher energy neutrino from U being more detectable
than those from Th, despite 4? lower concentration levels of the former.
Figure 2.1 shows the sensitivity of Stotal relative to distance from the detector in central
Italy. At a known decay rate, a relatively constant (232Th/238U)molar value (Wipperfurth
et al., 2018), and an assumed K/U value, we calculate the abundance of the heat producing
elements (K, Th, and U; HPEs). Please refer to Supplementary equation S1-Eq1 for the
full calculation of the total ??e signal.
Compositional variations in the local lithosphere have the strongest effect on the
geoneutrino signal because the lithosphere is closer to the detector (smaller r) and is 100-
fold enriched in HPE relative to the mantle. Although the Earth?s mantle is largest silicate
17
reservoir, its low U concentration (?10 ng/g) and distance (greater r) causes its signal to
be muted.
To determine the contribution of geoneutrinos from the mantle, and therefore how
much radioactive heat is left to power mantle convection, plate tectonics, or the geody-
namo, we must first determine the U and Th concentrations in the lithosphere surrounding
the detector. Subtracting the lithospheric signal from the total signal is done to establish
the mantle value and its Th and U content. The Slithosphere has Near-Field Lithospheric
(NFL) and Far-Field Lithospheric (FFL) contributions. Thus, the mantle geoneutrino sig-
nal is:
Smantle = Stotal? (SNFL +SFFL) (2.3)
The relative contributions of these components are: Near-Field lithosphere (40 to 50%),
Far-Field lithosphere (30 to 40%, i.e., global lithospheric signal), and mantle (?25%)
(Wipperfurth et al., 2020). The lithosphere includes the mechanically coupled, underlying
lithospheric mantle, which has limited compositional variation (McDonough, 1990) and
contributes little (order ?1 TNU, <10% of the signal) to the lithospheric signals (Huang
et al., 2013). Araki et al. (2005) observed that the first 50 km and 500 km from KamLAND
contributes ?25% and ?50%, respectively, of the total signal.
Modeling uncertainties: The relative uncertainties on the flux measurement at Kam-
LAND and Borexino experiments improve over time; KamLAND went from ?54% to
?15% uncertainty for its measured flux, while Borexino went from?42% to?19%. The
modern mantle with depleted and enrich domains is predicted to show only ?10% total
18
variation in its geoneutrino signal (S?ra?mek et al., 2013). Likewise, only ?10% relative
variation is observed in estimates of the Far-Field lithospheric signal. Typically, the up-
per crust (i.e., the top 1/3 of the crust) contributes ?70% of the geoneutrino signal from
the lithosphere. Hence, the greatest impact on interpreting the mantle signal comes from
accurately predicting the upper crustal composition, that is, the SNFL.
2.5 Lithospheric Modeling
Disentangling the mantle?s contribution to Stotal is a major goal of geoneutrino stud-
ies. Doing so requires accurate models for SLithosphere. Importantly, uncertainties (statisti-
cal and systematic) in the NFL model contribute most significantly to uncertainties in the
modern mantle and global results.
Given the limited (?10%) variation in the mantle?s signal, one expects its predicted
values from different geoneutrino experiments to agree at this level. However, the local
estimates of the modern mantle Smantle range from?30?13 TNU (power from K, Th, and
U) by the Borexino team (Agostini et al., 2020) to ?7?1.6 TNU by the KamLAND team
(Gando et al., 2013). Consequently, the disparate nature of these findings either means
(1) the mantle is grossly heterogeneous (i.e., beyond scales envisaged by geology), or (2)
there are substantial inaccuracies in lithospheric modeling.
The distribution, volume, composition (HPE content), and petrology of the forma-
tions surrounding a detector must be accurately determined for its contribution to SNFL.
Shales and granites are enriched in HPEs, whereas peridotites and carbonates normally
are not. However, the degree of HPE enrichment is variable even within a given rock type.
19
HPE concentrations differ among igneous, metamorphic, and sedimentary rocks, and be-
tween silicate and carbonate lithologies (Figure 2.2). It is therefore crucial to model ac-
curately the proportional contribution of each geological formation and its HPE content
near a detector.
The Borexino geoneutrino experiment at Gran Sasso National Laboratory was lo-
cated outside of L?Aquila, Italy, in the central Italian peninsula (Figure 2.3, 13.57?E,
42.45?N, with 1.4 km of rock overburden). The Apennines consist primarily of fore-
land basin sediments and siliciclastic foredeep basin sediments, covered by Middle Pleis-
tocene to Recent volcanics (on the western side) and continental shelf and marine deposits
(Cosentino et al., 2010; Vignaroli et al., 2019). The marine deposits are mainly dolomitic
(marble, where metamorphosed). Extensional forces from mantle spreading to the west
of the Apennines have led to a fault-block system of grabens filled with terreginous sedi-
ments in a region known as the Tyrrhenian Extensional Zone (Cosentino et al., 2010). As
a result, the uppermost crust near the detector could contain a mixture lithologies ranging
from < 1 ppm to > 4 ppm U (Figure 2.2).
2.5.1 Near-Field and Far-Field Lithosphere
The Near-Field Lithosphere (NFL) is oftentimes, for the sake of computational ease,
treated as the 4?latitude ? 6?longitude area centered on the detector (Huang et al., 2013),
rather than a circle with a 500 km diameter. The Far-Field Lithosphere (FFL) consists
of the rest of the Earth?s lithosphere (oceanic and continental). The crucial step, which
requires geoscientific expertise, is determining the concentration and distribution of HPEs
20
High TNU
Granites
Shales
Sandstones
Carbonates
Basalts
Peridotites
Low TNU
0 2 4 6 8
U (ppm)
Figure 2.2: The average and range of U (and Th) depends on rock lithology. Granites
tend to have higher average HPE content while carbonates and mafic rocks have lower
averages. Sandstones, on the other hand, can have a wide range of U content depending
on their formation and surrounding rocks. The white bar for each rock type shows the
interquartile range of U concentrations from the Earthchem.org Database https://www.
earthchem.org. See text for the definition of TNU.
21
10?E 12?E
16?E 45?N
40?N
t
Upper Cru
s
~20 - 30 km
t
Middle Cr
us
st
Lower Cru
ntle ~150 km
Moho Lithosphe
ric Ma
Figure 2.3: A schematic drawing of the location of the Borexino experiment and its Near-
Field lithosphere (NFL; highlighted colored map in the center). Though the global abun-
dance of U and Th contribute to the measured geoneutrino signal, the (continental) crust
immediately surrounding the detector has the strongest effect on the signal.
in the lithologies of the Near-Field Lithosphere.
SFFL is a global average of the continental and oceanic lithospheric contribution to a
detector?s farfield geoneutrino flux. Model predictions for the SFFL at existing and future
planned detector sites are consistent, with estimates agreeing at better than the ?20%
level. The competing predictions for BorexinoSFFL agree at 16?1 TNU (Agostini et al.,
2020; Coltorti et al., 2011; Huang et al., 2013; Wipperfurth et al., 2020).
Whether a signal is from a moderate source of heat producing elements in the litho-
sphere near the detector or from a more concentrated mantle source is where discrepancies
are introduced. To illustrate this point, and to highlight the need for accurate lithospheric
models for the area surrounding geoneutrino detectors, we walk through the impacts of
two different scenarios of upper crustal concentrations for Th and U near the Borexino
geoneutrino detector.
Figure 2.4 illustrates the signal trade-off between HPE content of the Near-Field
Lithosphere and mantle. Stotal depends on the total mass of HPEs and their distance from
22
16 TNU
NFL
NFL
FFL FFL
Mantle
Mantle
Agostini?20 Generic
Figure 2.4: The total geoneutrino signal, Stotal (length of the boxes in the figure) measured
at a given detector remains relatively constant over time; the uncertainty decreases as the
number of geoneutrino events detected increases. The amount of signal attributed to the
Near-Field Lithosphere (NFL, yellow) determines how much signal must come from the
mantle (blue). The average signal of the Far-Field Lithosphere (FFL, brown) generally
stays the same (i.e., 16?1 TNU for Borexino).
the detector. The non-uniqueness of the modeling drives us to construct more accurate
3D descriptions of the HPE contents of the Near-Field Lithosphere, to evaluate better the
mantle HPE concentrations.
2.5.2 Numerical Model
Figure 2.5 presents two NFL models used to analyze the effects of vastly different
abundances of Th and U in the upper crust surrounding the Borexino detector: (1) a low
Th+U content (e.g., dominantly carbonate) and (2) medium Th+U content (e.g.,shale-like,
or averaged carbonate + siliciclastics + volcanic). These idealized models are comparable
to those reported in (1) Agostini et al. (2020) and Coltorti et al. (2011), and (2) Huang
et al. (2013), Wipperfurth et al. (2020), and McDonough et al. (2020).
Using Monte Carlo numerical modeling (Wipperfurth et al., 2020), we determined
23
47 TNU
47 TNU
U = 0.80 ppm U = 2.7 ppm
Th = 2.1 ppm Th = 10 ppm 
3.0
Upper Crust Upper Crust
2.0
Middle Crust Middle Crust
1.0
Lower Crust Lower Crust
0
NFL Model 1 NFL Model 2
Figure 2.5: Two different Near-Field Lithosphere models illustrate low (Agostini?20) and
medium (Generic) U and Th scenarios in the uppermost crust near our geoneutrino de-
tector. The middle and lower crust are kept the same among the three models since we
are primarily interested in the effects of upper crustal compositional changes. See Huang
et al. (2013) for discussions on middle/lower/deep crustal geoneutrino contributions.
24
U (ppm)
the expected BorexinoSNFL assuming two different scenarios: low and medium HPE con-
tents for the upper crust. The HPE content of the middle, and lower crust and lithospheric
mantle are taken from Sammon and McDonough (2021); Sammon et al. (2022). For the
physical description of the local lithosphere, we use the LITHO1.0 model (Pasyanos et al.,
2014) (i.e. density, distance from detector) with 1?latitude x 1?longitude horizontal reso-
lution for the upper, middle, and lower crust and lithospheric mantle. Table 2.1 lists the
compositional model parameters for the NFL, its signal, and that for the total lithosphere
and mantle. This table also reveals the predicted power of the mantle and bulk Earth for
these two different upper crustal models and thus NFL models.
A factor of three difference in the HPE budget of the upper crust for these two NFL
models produces a factor of ?2 difference in both the estimated mantle and bulk Earth
radiogenic power (Figure 2.6). These gross differences in the predicted radiogenic power
demonstrate the significance of producing an accurate NFL model.
2.6 Importance of the Near-Field Lithosphere Model
The Apennines of the central Italian peninsula exposes a geological paradox across
its eastern and western divide. Its Adriatic eastern side is composed of a compressional
fold and thrust belt, whereas its Tyrrhenian western side is composed of extensional fault-
block mountains. The paradox of this mountain belt is the juxtaposition of both compres-
sional and extensional tectonic forces over a relatively narrowed (?150 km) east-west
traverse.
Figure 2.7 shows that carbonate sediments surround the Borexino detector, whereas
25
60
50
40
TW
30
40
20
30
Agostini?20
20
Generic 10
10
0
0 10 20 30 40
S  (TNU)
lithosphere
Figure 2.6: The lithospheric geoneutrino signal (predicted, Slithosphere) vs. the measured
geoneutrino signal (Stotal) for the Agostini at al. model and Generic Model introduced in
Table 2.1. The Agostini at al. Model has a smaller predicted lithospheric signal, attribut-
ing 21.3+7.3?10 TNU to lithosheric U and Th. The Generic model has a higher concentration
of U and Th in the upper crust of the NFL, and therefore a greater lithospheric flux,
32.3+7.9?6.4 TNU. The dotted lines with slopes = 1 show the y-intercept for each model. The
y-intercept is the Smantle. The blue-shaded area shows the measured S +10.8total of 47?9.6 TNU
26
Bulk Earth
S  (TNU)
total
A Surface Geology B Surface Heat Flux (mW/m2) 250
 44? N  ?Quaternary  volcano  44  N  Quaternary  volcano
volcanic deposit Borexino detector
Radicofani carbonate deposit Larderello Radicofani outer Apennine frontLarderello 200
Torre Alfina Borexino detector Torre Alfina
 43? N  ?Amiata San Venanzo outer Apennine front  43  N  Amiata San Venanzo
Vulsini Polino Vulsini Polino
Cimini Cimini 150
Vico Cupaello Vico Cupaello
 42? N  Sabatini  42? N  Sabatini
Colli Albani
T Roccamonfina
Colli Albani Roccamonfina
yrr Th ye rn rh
100
? ia
enia Campi Flegrei
 41  N  n S Campi Flegreiea  41
? N  n Se
Vulture a Vulture
Ischia Ischia150 km Vesuvius 150 km Vesuvius 50
 40? N   40? N  
 10? E  12? E  14? E  16? E  10? E  12? E  14? E  16? E 0
Figure 2.7: A simplified geological map (A) of the central Italian peninsula showing
extensive volcanism on the western portion and carbonate platforms to the east. The red
dashed line circles the Borexino detector (blue star) at a radius of 50 km. Quaternary
volcanic deposits in the west coincide with high surface heat flux (B).
the western half of the Near-Field region exposes extensive deposits of Neogene to Qua-
ternary igneous rocks (Lima et al., 2005; Xhixha et al., 2014). The Tuscan and Roman
magmatic provinces are exposed all throughout the Tyrrhenian side of the Apennines and
coastal plains. This western portion of the Italian peninsula is enriched in K, Th, and
U, with some rocks containing as much as 25 ?g/g U (Conticelli et al., 2007), which is
slightly less than 10 times enriched over average upper crustal rocks (Rudnick and Gao,
2014).
These western Tuscan and Roman magmatic rocks are HPE-enriched and make up
a significant portion of the upper crust of the NFL. Some of these rocks are within 50 km
of the Borexino detector and need to be incorporated into any NFL model, but unfortu-
nately these lithologies were not discussed by (Agostini et al., 2020; Coltorti et al., 2011).
Agostini et al. (2020) highlighted the central tile, which includes the area within?100 km
of the Borexino detector and noted ?Up to a distance of ?150 km from Borexino, 100%
of the geoneutrino signal is generated from the LOC [local lithosphere].? Nearly all of
27
the volcanoes identified in Figure 2.7, some of which are enormous volcanic centers, are
within 150 km of the Borexino detector. In addition, the CROP 11 seismic refraction line
that the Agostini et al. (2020) model cites as evidence for 13 km of carbonate sediments
shows thick layers of siliciclastic sediments as well (e.g., Di Luzio et al. (2009); Patacca
et al. (2008).
To develop our alternative model of the BorexinoNFL, we followed the practices of
Huang et al. (2013) and McDonough et al. (2020) and used a generic, average upper crust
composition (Rudnick and Gao, 2014). Using such a generic model for the upper crust of
the NFL results in a mantle and bulk Earth model that is consistent with studies that favor
a 20 TW radiogenic Earth (Bellini et al., 2021; McDonough et al., 2020).
Disparities between the predicted HPE concentrations in the upper crust for the
NFL cause the greatest systematic uncertainties in calculated radiogenic heat production.
Constructing a purely carbonate versus a generic upper crust around the detector changes
the expected mantle radiogenic heat budget from 30 TW to 13 TW, respectively. These
contrasting models illustrate the consequences of modeling different proportions of HPE
lithologies for the NFL. Consequently, inaccurate estimates of the subsurface composition
near a detector vastly change the implications of the observed geoneutrino signal Stotal .
2.7 Heat Flux Constraints on Lithospheric Models
To further assess the upper crustal model of the BorexinoNFL we turned to the avail-
able heat flux data for the central Apennines (Pauselli et al., 2019; Verdoya et al., 2021).
Given the regional tectonic setting discussed above, it is not surprising to observe a clear
28
distinction between the western, high heat flux (>150 mW/m2) and the eastern low heat
flux (<70 mW/m2) provinces (Pauselli et al., 2019) (Figure 2.7). Moreover, using ob-
servable crustal radiogenic heat production data, Verdoya et al. (2021) concluded that low
surface heat flux estimates (e.g., values <45 mW/m2) are unreliable in the Apennines.
This study also concluded that the central Apennines region has an average heat flux of
?70 mW/m2 (with eastern and western limbs being approximately 55 and 150 mW/m2,
respectively). On average, the BorexinoNFL has a relatively normal continental surface heat
flux value (e.g., ?63 mW/m2, (Lucazeau, 2019)).
Surface heat flux is the sum of contributions from heat production in the crust and
the heat flux across the Moho. The total surface heat flux (TotalHF ) can be expressed as
the sum of crustal and Moho heat fluxes:
TotalHF ?CrustHF +MohoHF (2.4)
Normally, a regionally averaged surface heat flux (e.g., ?63 mW/m2) is dominated
by an upper crustal fraction (i.e., 50 to 60%) and, less so, by a ?1/3 contribution from
the Moho heat flux (i.e, 21 ? 10 mW/m2) (Sammon et al., 2022). If we assume a generic
crustal compositional model (Table 2.1), the regional TotalHF for the Italian peninsula
appears normal in terms of its heat production and surface heat flux (i.e.,?70 mW/m2). In
contrast, assuming the compositional model for the NFL adopted by Agostini?20 (Coltorti
et al., 2011) puts the CrustHF contribution at 24 mW/m2 and a MohoHF of 46 mW/m2 ?
more than double the global average. While this level of Moho heat flux is possible, it
is only observed in areas of recent volcanism, which contradicts the low HPE carbonate
29
shelf model.
The Agostini et al. (2020) model for the mantle?s radiogenic heat (30 TW) is also
inconsistent with their choice of a 8.1 TW global lithosphere model. The Earth has 46? 3
TW of heat (Jaupart et al., 2016), which is both radiogenic and primordial in origin, with
other contributions including 3 TW from oceanic hot spots (Jaupart et al., 2007; Labrosse
et al., 2001), 0.4 TW from tidal heating, crust-mantle differentiation, and thermal con-
traction (Jaupart et al., 2007), and a minimum of 6 TW to 12 TW from secular cooling of
the mantle (Labrosse, 2007). Consequently, for Agostini et al. (2020)?s accounting to be
correct, it leaves anywhere from -2 to -8 TW for the core-mantle boundary (CMB) heat
flux, meaning that the mantle is radiating 2 to 8 TW of heat into Earth?s core as it heats up
over time. Our alternative model has 7.6+2.1?1.6 TW for the global lithosphere (Wipperfurth
et al., 2020) and 12.9 TW in the mantle. This model yields a CMB heat flux of 10 to 16
TW, in agreement with estimates from previous studies (Korenaga, 2008; Labrosse, 2002;
Labrosse et al., 2001; Nakagawa and Tackley, 2010; Roberts et al., 2003).
The first experiment to detect geoneutrinos, KamLAND, in Kamioka, Japan, pre-
dicts a low radiogenic power Earth, 11.2+7.9?5.1 TW for Th and U only, or 14 TW when
including the decay of other isotopes (Gando et al., 2013). This result is intermediate
between the low H (H = heat production) estimates for the Earth (Javoy et al., 2010) and
middle H estimates (McDonough and Sun, 1995; Palme and O?Neill, 2014). The NFL
model used by the KamLAND team (Enomoto et al., 2007) predicts an Earth with a low
radiogenic power, whereas that proposed by Wipperfurth et al. (2020) predicts an Earth
with 20 TW of total radiogenic power.
These KamLAND results challenge the Earth model of Agostini et al. (2020) that
30
predicts 38 TW of radiogenic power. Either (1) the geological compositions of the Kam-
LAND and/or the Borexino models need to be thoroughly re-investigated, or (2) one
would have to predict a hemispherical dichotomy in the mantle?s composition. The latter
hypothesis is, of course, unsupported by empirical data on the composition of mid-ocean
ridge basalts and ocean island basalts. The second hypothesis seems completely unten-
able.
In summary, we document the significance of geology?s input into interpreting the
particle physics flux data. The combined results for KamLAND and Borexino experi-
ments strongly favor a 20 TW radiogenic Earth model. Moreover, these results confirm
that the bulk Earth has a 1.9? enrichment in refractory elements over a CI chondritic
composition (Yoshizaki and McDonough, 2021).
2.8 The Future of Neutrino Geoscience
High resolution crustal models accounting for the specific types and proportions
of lithologies surrounding each geoneutrino detector must be constructed to interpret
geoneutrino flux measurement. The geology underlying active geoneutrino detectors
(Figure 2.8) in Gran Sasso, Italy, Kamioka, Japan, and Sudbury, Canada, reveal com-
plicated tectonic features (e.g., (paleo-)subduction and synorogenic extension, ocean-
continent subduction zone, large impact structure). Geoneutrino data already exists from
two of these locations, but these crustal models are either low resolution or in conflict with
one another. We must reconcile the geoneutrino signal at each location with improved lo-
cal and regional geology. We must use a wide range of independent geoscientific data to
31
 180? E
 150? W OBD  150? E
 120? W  45
? N K amLAND  120? E
JUNO
?
SNO+ CJPL
  90? W   90? E
  60? W   60? E
Borexino
  30? W   30? E
   0?
Figure 2.8: Borexino and KamLAND will be joined by the next generation of geoneu-
trino detectors, including SNO+, which is already counting, and JUNO, which is under
construction. The under-development CJPL detector next to the Himalayas marks the
fifth detector in the northern hemisphere, allowing for unprecedented mantle resolution.
The OBD (ocean bottom detector) experiment is a mobile device and its position can be
optimized as being 3000 km away from South America, Australia, and the core mantle
boundary.
constrain the composition of the NFL. Moreover, our compositional models needs to be
internally consistent with available heat flow, geochemistry/petrology, structural geology,
and seismology data to reduce the systematic uncertainties on Earth?s HPE content and
thermal budget.
There are three more geoneutrino projects under construction or development: Jiang-
men Underground Neutrino Observatory (JUNO, Figure 2.8 purple dot) in southeastern
China, which will be 20x larger than any existing detector (An et al., 2016); China Jin-
32
ping Underground Laboratory (CJPL, Figure 2.8 green dot) sited on the eastern slope of
the Tibetan plateau and Himalayan ramp and at 2.4 km depth (Beacom et al., 2017); and
OBD, a movable, Ocean Bottom Detector (Figure 2.8 teal dot) proposed by a team of sci-
entists and engineers working with JAMSTEC (Sakai et al., 2022). These projects each
represent massive feats of engineering and decades-long data collection experiments and
require substantial geoscientific input.
The decay of HPEs contribute substantially to Earth?s internal heat. By quantify-
ing Earth?s geoneutrino flux, we can precisely establish how much fuel from HPEs is left
to power mantle convection and the recycling processes of plate tectonics. Geoneutri-
nos studies use modern physics technology to measure directly and instantaneously the
current compositional properties of the inaccessible mantle. Th and U exist in Earth in
constant, chondritic ratios to 26 other elements (McDonough and Sun, 1995); if we con-
strain the abundance of HPEs, we can establish Earth?s concentrations of Ca, Al, Nb, and
the economically valuable rare earth elements. With the second generation of geoneutrino
detectors on the horizon, geoscientists and physicists are poised to unravel Earth?s heat
budget from the tallest mountains to the bottom of the oceans.
2.9 Conclusion
The power of geoneutrino studies lies in directly quantifying the amount of heat
producing elements in the bulk Earth. Deep reservoirs in Earth that before were un-
reachable are being sampled by particle physicists, but these studies have not reached a
consensus on what their results mean for mantle heat production. The geoneutrino sig-
33
Agostini?20 Generic Units
K 9,600 23,200 ?g/g
Th 3.0 10.5 ?g/g
U 0.8 2.7 ?g/g
HP? 0.19 0.62 nW/kg
SNFL 9.7 16.6 TNU
SLithosphere 25.7 32.3 TNU
SMantle 21.3 14.7 TNU
Mantle 30 13 TW
Total 38 20 TW
UC? local model for the Upper Continental Crust. NFL = Near-Field Lithosphere (i.e., closest
?500 km to a detector). Units: ?g/g (10?6 kg/kg); TNU (Terrestrial Neutrino Unit, see text for
details); TW (Terra Watts, 1012 watts). R* radiogenic power. HP? Heat Production.
Table 2.1: Borexino Models for the upper crust in the NFL, bulk calculated Signal, and
Radiogenic Power
nal at a given detector is a combination of crust-sourced and mantle-sourced Th and U
decays. Since geoneutrinos do not carry directional information, the lithospheric signal
must be constrained to quantify the mantle?s abundances of Th and U.
Approximately 50% of the geoneutrino signal is produced from the Near-Field
Lithosphere (NFL), with 25% of the signal coming from the HPEs within 50 km of the
detector. Conflicting Near-Field Lithospheric compositional models lead to profoundly
different consequences for the predicted HPE content in the mantle and Earth?s thermal
evolution.
The Borexino particle physics team (Agostini et al., 2020) modeled the NFL sur-
rounding their detector as predominantly carbonate, with low concentrations of Th and
U. Their model therefore requires most of the geoneutrino signal to come from the distant
mantle, implying a 30 TW of mantle radiogenic heat production. Consequently, >80% of
34
R* Heat Signal UC?
all of the Earth?s internal heat is radiogenic. This high heat production mantle is inconsis-
tent with measurements from the detector at KamLAND and with heat flux observations.
Alternatively, the inclusion of Neogene to Recent, HPE-rich volcanic deposits in the
Borexino NFL region results in a more normal average upper crustal composition for Th
and U. Using this upper crustal model (versus a low HPE model) can explain the Borexino
signal, resulting in 13 TW of radiogenic power in the mantle or a 20 TW radiogenic Earth.
It is therefore imperative to produce high-resolution NFL maps with accurate proportions
of each HPE lithology.
The direct measurement of geoneutrinos can provide crucial insights into the sources
and distribution of heat producing elements in the Earth. When paired with accurate ge-
ological knowledge, these high-energy antineutrinos emitted from HPE decays within
Earth helps establish the composition of the planet?s building blocks as well as the fuel
left to power Earth?s dynamic interior.
2.10 Acknowledgments
LGS and WFM gratefully acknowledge the NSF for support (grants EAR1650365
and EAR2050374) and University of Maryland for a Wiley fellowship. Both authors
would also like to acknowledge Jordan Goldstein, who worked with them to create the
figures and schematics in this paper.
35
Chapter 3: A Geochemical Review of Amphibolite, Granulite, and Eclog-
ite Facies Lithologies: Perspectives on the Deep Continental
Crust
3.1 Overview
The composition and origins of the bottom 23 of the continental crust has been a
topic of geologic debate for many years. Because of the inaccessible depths of these
middle and lower sections of the continents, we cannot sample them directly. We must
rely on rocks brought to the surface through mountain building and magma entrainment
processes. Deep crustal rocks delivered via these processes come from a wide variety of
depths and encompass many different chemical compositions. This chapter seeks to char-
acterize the average composition of the deep crust (typically from 15 to 40 km beneath
the surface) and identify the processes that produced the crust?s present-day, chemically
layered structure.
[1] This chapter has been published as:
Sammon, L. G., & McDonough, W. F. (2021). A Geochemical Review of Amphibo-
lite, Granulite, and Eclogite Facies Lithologies: Perspectives on the Deep Continen-
tal Crust.
36
[2] LGS wrote this text and curated the deep crustal dataset with significant insights
and text additions from WFM. LGS would like to acknowledge Caerwyn Hartten for her
assistance in curating the metamorphic datasets.
3.2 Abstract
Debate abounds regarding the composition of the deep (middle + lower) continental
crust. Exhumed medium- and high-grade metamorphic rocks, which range in composition
from mafic to felsic, provide information about the bulk composition of the deep crust.
This study presents a global compilation of geochemical data on amphibolite (n = 6500),
granulite (n = 4000), and eclogite (n = 200) facies lithologies and quantifies trends, uncer-
tainties, and sources of bias in the deep crust sampling. The continental crust?s Daly Gap
is well documented in amphibolite and most granulite facies lithologies. Igneous differ-
entiation processes likely control the compositional layering in the crust. Al2O3, Lu, and
Yb vary little from top to bottom of the crust. In contrast, SiO2, light rare earth elements,
Th, and U show a wider range of abundances throughout. Because of oversampling of
mafic lithologies, our predictions are a lower bound on middle crustal composition. Ad-
ditionally, the distinction between granulite facies terrains (intermediate SiO2, high heat
production, high incompatibles) or granulite facies xenoliths (low SiO2, low heat produc-
tion, low incompatibles) as being the best analogs of the deep crust remains disputable.
We have incorporated both rock types, along with amphibolite facies lithologies, to de-
fine a deep crustal composition that approaches 57.6 wt.% SiO2. This number, however,
represents a compositional middle ground, as seismological studies indicate a general in-
37
crease in density and Vp and Vs velocity with increasing depth. Future studies should
analyze more closely the depth dependent trends in deep crustal composition to develop
composition models that are not limited to a three-layer crust.
3.3 Introduction
The composition of the deep continental crust has been the subject of many stud-
ies for the past half century because of its importance in crustal evolution and the lack of
consensus on its composition. The combined middle and lower continental crust (referred
to here as the ?deep crust?) are the integrated chemical products of billions of years of
crust formation and deformation, yet their inaccessibility (deeper than 10 km) has led to
a poorly constrained compositional model for the lower two-thirds of the continent. The
deep continental crust can be sampled through tectonically emplaced or eroded exposures
of high-grade metamorphic rock (referred to here as ?terrains?) or deep crustal xeno-
liths that are rapidly carried to the surface through volcanic eruptions. The composition
of these deep crustal analogues ranges widely, encompassing lithologies from metamor-
phosed basalt to granite. Varied tectonic regimes and widespread crustal heterogeneity
have led to numerous geochemical and geophysical models that help to explain local phe-
nomena, but struggle to produce a coherent global picture. Attempts to resolve the debate
are limited by nonunique solutions and poorly quantified uncertainties. Defining the bulk
compositional properties of the deep continental crust and describing its depth dependent
changes endures as a long-standing challenge. Thus, the deep crustal composition puzzle
remains, troublingly, unsolved.
38
Rudnick (1995) posited the paradox of the continental crust: the continental crust
has an andesitic composition, however melts from the mantle are basaltic. In doing so, she
identified that the formation of continental crust, as compared to making oceanic crust,
must be an open system process involving, to different extents, weathering, intra-crustal
melting (leaving behind a dense residue), and delamination as some of the operating pro-
cesses. Consequently, the geochemical uncertainty associated with deep crust compo-
sition has led to competing models for crust formation (Bu?rgmann and Dresen, 2008;
Hacker et al., 2015; Rudnick and Gao, 2014).
In developing their model, Hacker et al. (2015) outlines two processes that they
envisage as shaping crustal evolution: delamination and relamination. Delamination oc-
curs when gravitationally unstable material in the deep crust, such as eclogite and other
garnet-rich lithologies, separates and flows into the less dense underlying mantle. This
process leads to a dense, mafic deep crust as eclogitization occurs but before the lower
crust delaminates. In contrast, the process of relamination thrusts subducting sediment
under the continental crust, resulting in a more felsic, less dense lower crust. While indi-
vidual examples can be found to support each of these processes, the difficulty remains in
determining the dominant pattern of crust evolution.
The continental crust is conventionally split into upper,?middle, and lower layers,
though distinct seismic or petrological/geochemical boundaries are not always evident
(Holbrook et al., 1992). Petrological and geochemical studies of the deep continental
crust have therefore sought to define its composition through analysis of various high
grade metamorphic lithologies. It is difficult to gauge, however, if isolated metamorphic
samples are representative of the entire deep crust. Temperature and pressure, and there-
39
fore metamorphic grade, increase with increasing depth in the crust, though the geother-
mal gradient varies by up to a factor of ? 3 depending on continental crust type and
tectonic regime (Christensen and Mooney, 1995). If a pressure of 1 GPa is reached at
35 km (assuming an average crustal density of 2,900 kg/m3 (Wipperfurth et al., 2020)),
the deep crust could plausibly be composed of greenschist, amphibolite, granulite, and/or
eclogite facies lithologies. However, amphibolite and granulite facies material dominate
what are interpreted as deep crustal cross-sections (such as metamorphosed terrains ex-
humed in the Ivrea-Verbano Zone, Italy), with minimal evidence for greenschist facies
lithologies (Rudnick and Gao, 2014). Eclogite facies lithologies likely contribute to oro-
genic regions with thickened deep crust (pressures up to 1.5-2 GPa following the same
density scheme as above)(Leech, 2001; Lombardo and Rolfo, 2000). For these reasons,
this study focuses on amphibolite, granulite, and eclogite facies lithologies as potential
major components of the deep crust. We report on an expanded database developed by
Rudnick and Presper (1990) and added to by Hacker et al. (2015), including data sourced
from Earthchem.org; we examine the chemical trends among various medium to high
grade metamorphic lithologies to understand and better characterize what is the average
composition of the deep continental crust and follow this with implications for crustal
differentiation and evolution processes.
3.4 The Art and Science of Deep Crustal Modeling
In many ways, predicting the composition of the deep continental crust is as much
an art as it is a science, because deep crustal models depend not only on the input data,
40
but also the approach each modeler takes to interpreting said data. The definition of the
deep crust depends on the question each researcher is trying to address, and is therefore
neither a static nor universal term. These differing can sometimes lead to confusion and
produce seemingly contradictory models of the crust when in fact, each model is simply
looking at the crust through a different lens.
How many layers should we split the crust into? What is the scale of lateral varia-
tions in the crust? The answers differ based on the model. This fundamental question is
the crux of the disagreement between popular composition models (Hacker et al., 2015;
Rudnick and Gao, 2014). While some split the crust into two layers (Hacker et al., 2015),
upper and lower, shallow and deep, others split it into three (upper, middle, and lower) or
more sections (Christensen and Mooney, 1995; Mooney et al., 1998). Thus, debates about
compositional models need to be clear about their specific crustal mass fractions. Much
geophysical effort has gone into determining the layering and seismic structure of the
continental crust. Such topics are beyond the scope of this study, but we want to bring the
concept of model resolution to the readers? attention so that they can appreciate the com-
plexity of the task of modeling deep crust composition and be mindful that we are taking
but one approach. Nevertheless, averaging geochemical data on large and diverse sample
sets has merits. An average deep crustal composition provides an integrated view of how
the crust has evolved through time, informing us of the dominant crust formation path-
ways, though unique tectonic regions may still require specialized compositional models.
Bulk crust and silicate Earth compositions are also useful when comparing Earth to other
planetary bodies, such as Mars or Super Earths. On a more concrete level, crustal correc-
tions are required for geophysical applications, such as tomographic models of the mantle
41
or geoneutrino studies, and an average crustal composition is detailed enough to provide
further resolution than PREM (Dziewonski and Anderson, 1981) but simple enough to
implement without significant detriment to calculation time. As such, we have decided to
press forward with a global-scale deep crustal model.
Classically, two approaches have been taken to assess deep crustal composition:
sample driven modeling and process driven modeling. Sample-driven models base their
conclusions on the premise that deep crustal analogue samples, such as mafic high grade
metamorphic xenoliths, are by and large representative of the composition of the deep
crust. Empirical analyses are the main source of data for this type of model. This geo-
chemical inverse model takes measured element concentrations from exhumed rocks and
derives the conditions under which they formed. A second approach considers physical
processes and constraints that build the deep crust and the effects of crust formation and
evolution. A variety of mafic and felsic compositions can satisfy the geophysical observ-
ables, such as Vp or viscosity (Hacker et al., 2015; Shinevar et al., 2018). These forward
process models consider all possible geochemical solutions, avoiding the potential bias
of xenoliths, which may be sampling a restricted portion of the deep crust, or whose
chemistry has been influenced by their limited eruption environments. Both approaches
have their strengths, by being based on petrological and geochemical observation or by
reducing sampling bias, and in the end, both methods are valid.
This study more closely resembles the first approach, using samples to infer deep
crustal composition. We are mindful of the potential biases this leads to (refer to Section
3.5.2). For the sake of comparison to other models, we operate under the assumption of a
three-layer crust, though we advocate for embracing the potential for vertical and lateral
42
compositional variation by analyzing the full spectrum of available data. Future studies
should move beyond bisecting or trisecting the crust, taking advantage of the quality and
resolution of both geochemical and geophysical data currently being produced.
3.5 Datasets
3.5.1 Amphibolite, Granulite, and Eclogite
For the rest of this study, ?amphibolite?, ?granulite?, and ?eclogite? will refer to
rocks in those metamorphic facies, without imposing constraint on composition. Both
amphibolite and granulite facies lithologies range from mafic (< 52 wt.% SiO2) to fel-
sic (>68 wt.% SiO ) in composition, and can have Mg#?s (molar Mg2 Mg+Fe ) that resemble
the mantle (Mg# ? 89), the upper continental crust (Mg# ? 30), or any number in be-
tween. Eclogite facies lithologies are less heterogeneous than amphibolite or granulite
facies. Eclogite facies mineral assemblages are dominated by (clino)pyroxene and gar-
net, leaving less room for variations in silica content. Please note that ?eclogite? as a
metamorphic facies is less restrictive in composition than the largely garnet-omphacite,
bi-mineralic rock, eclogite, and can include arclogite samples.
The medium pressure (e.g., 0.2-0.8 GPa) and temperature (e.g., 200-600?C) equi-
librium assemblages of amphibolite facies lithologies presumably reflects the conditions
of the middle continental crust. Granulite facies lithologies are widely held to comprise
the lower continental crust, with its base being defined seismically by the Moho. Am-
phibolite facies lithologies are generally sampled through exhumed terrains and are more
rarely sampled through xenoliths. Granulite facies lithologies can be sampled via terrains
43
or xenoliths. Granulite facies xenoliths have predominantly mafic to intermediate-mafic
(45 - 55 wt.% SiO2) silica content, while granulite facies terrains span the range of mafic
to felsic. Granulite facies rocks are distinguished from amphibolite facies rocks by the
dehydration of hydrous mineral phases (Rudnick and Fountain, 1995). The water-rich
minerals that can occur in amphibolite, such as amphiboles and micas, break down into
pyroxenes in the granulite stability field due to higher temperatures. Granulite facies
metamorphism can initiate at 600?C, meaning that any granulite facies rocks present in
areas where the crust is thin and/or the lower crust is at temperatures < 600?C are likely
in thermal disequilibrium. Granulite facies lithologies, however, are only expected to
undergo retrograde metamorphism under limited circumstances due to the kinetic bar-
rier of rehydration (Semprich and Simon, 2014). Thus, many studies still use metastable
granulite as a lower crustal analogue.
The eclogite facies is traditionally bounded by the pressures and temperatures re-
quired to transform basaltic mineral assemblages into high pressure clinopyroxene and
garnet ? rutile ? accessory minerals. Though it can be difficult to achieve the pressures
required to form eclogite in average continental crustal settings (crustal thicknesses <
40 km), eclogite facies materials may be a significant component of modern and paleo-
orogenic belts (Leech, 2001; Lombardo and Rolfo, 2000). Eclogite facies lithologies,
however, may not be preserved through time even in this thicker crust because, unlike
granulite, they are more susceptible to retrograde metamorphism (Abbott and Greenwood,
2001; Carswell and Cuthbert, 1986; Dokukina and Mints, 2019; Krogh et al., 1994). De-
spite their occurrence being potentially localized to orogenic regions in the crust, eclogite
facies lithologies are considered germane to the discussion of deep continental crust be-
44
cause eclogite is commonly found at subduction zones, one of the main tectonic settings
where continental crust is generated (Hawkesworth and Kemp, 2006; Rudnick, 1995).
3.5.2 Potential Biases
The first step in analyzing a dataset is to admit that it is potentially biased. Through-
out this paper, we scrutinize the statistical uncertainty of deep crust compositions. Sys-
tematic uncertainties, however, are not so easily quantified. This section offers what
limited insight we have on the potential for systematic bias in our deep crust sample set.
The analyses and conclusions in the rest of this paper are generally founded upon the
assumption that the following systematic biases have a limited effect on our dataset, and
if any datasets do fall prey to bias, they can be amended without significantly changing
the overall picture of deep crust composition.
Our compendium of deep crustal samples, available in the supplemental informa-
tion of this paper, consists of published data from various sources, most of which are
available on Earthchem.org (www.earthchem.org). We used a subset of the data avail-
able, limiting our calculations and analyses to samples whose major oxide content is
reported and totals to 100 ? 10%. Because of the numerous opportunities for bias in our
dataset, we only limited samples by metamorphic grade and major oxide totals. Remov-
ing the oxide totals filter does not substantially change the distributions of most elements,
but tends to increase the data scatter. The filtered and unfiltered data sheets are available
as supplemental information.
45
3.5.2.1 Location Bias
The global distribution of medium and high grade metamorphic samples shows little
correlation between composition and location (Figures S1 and S2). In fact, samples of
mafic and felsic compositions are often found within the same region. We are, of course,
limited to areas where terrains and/or xenoliths have been exposed at Earth?s surface,
but our data include samples from all seven continents. Amphibolite facies lithologies
have been extensively studied in crust of various ages. Granulite facies lithologies are
also widely sampled, though the xenoliths are relegated to areas that have experienced
uncommon eruptions of mafic, xenolith-bearing magmas. In addition, Archean granulite
facies terrains are generally restricted to cratonic regions, as active tectonic environments
tend not to preserve Archean-aged rocks. Eclogite facies xenoliths and terrains are our
most limited datasets, but >200 samples are still available for study. Many eclogite facies
samples are from the western United States, potentially biasing the dataset towards the
eclogites of the Franciscan Complex and eclogites formed from oceanic crust subduction
(Tsujimori et al., 2006). South America and Antarctica are not represented in the eclogite
facies dataset.
3.5.2.2 Buoyancy and Transport Mechanism Bias
The deep crust may not be fully represented by the analogue samples that have
reached Earth?s surface. Medium and high grade metamorphic lithologies that have sur-
vived surface transport do not necessarily reflect the full distribution, abundance, or com-
position of the deep crust. Felsic terrains could be over-represented at the surface due
46
to their lower densities. Buoyancy is a significant dynamical force that may play a criti-
cal role in determining what types of metamorphic terrains outcrop at the surface (Gerya
et al., 2002; Kelemen and Behn, 2016).
On the other hand, eruption type and location may likewise bias xenolith compo-
sitions (Jaupart and Mareschal, 2003), including contaminating them with the basaltic
lavas. (Rogers and Hawkesworth, 1982; Rudnick and Presper, 1990; Rudnick and Taylor,
1987). Studies have also found that felsic xenoliths often cannot withstand the frequently
hot, violent eruptions that transport samples to the surface and tend to be re-assimilated
(Halliday et al., 1993; Rudnick and Fountain, 1995). Granulite facies xenoliths in partic-
ular could be biased by location and/or eruption method: they tend to be co-located with
cratonic crust and they are often carried by kimberlite eruptions and fast-erupting alkali
basaltic volcanism (Rudnick and Presper, 1990; Russell et al., 2012). Cratonic xenoliths,
which presumably have not experienced recent tectonic activity aside from their transport
to Earth?s surface, may be sampling the composition of an older deep crust.
3.5.2.3 Preservation and Exposure Bias
We recognize also the potential of sample preservation bias. Recent studies outline
different weathering rates for different metamorphic rock compositions (Ohta and Arai,
2007; Price and Velbel, 2003). Age and weathering rate, along with protolith composition,
may affect the current metamorphic sample population.
Metastable conditions in the deep crust are another concern. Granulitic lithologies
would not be in thermal equilibrium under most projected geotherms (Kusznir and Park,
47
1987). On the other hand, it has been proposed that felsic granulite facies lithologies will
not transform into eclogite facies under high pressures but remain as metastable granulite
(Hacker et al., 2010). By contrast, the middle continental crust should be stable in the
greenschist facies and the lower crust in amphibolite facies. This lower grade combi-
nation is not often observed in exposed cross-sections, which, on average, display high
temperature/pressure metamorphism and have a mean pressure of 0.8 GPa (i.e., 25-30
km depth)(Brown and Johnson, 2019). The abundance of amphibolite and granulite fa-
cies material in (what we deem to be) deep crustal cross-sections suggests that the deep
crust reached peak metamorphic conditions some time in the past and has since cooled
off. Additionally, metamorphic terrains may undergo retrogression during the exhuma-
tion process. We classify these observations as a ?preservation bias? because we are
preferentially preserving metastable mineral assemblages.
3.5.2.4 Sample Collection and Naming Bias
Lastly we face the bias that we as scientists impose ourselves: collection and clas-
sification bias. Unique localities can be over-sampled for their novelty, and thus, overly
abundant in the dataset. Common andesitic rocks, with their lack of attractive phenocrysts
and dull grayish-pink hue, may unfortunately be glossed over in favor of more attractive
samples (apologies to Dr. J. Blundy and colleagues). Oversampling the same locations
seems to plague the amphibolite facies dataset most, with many nearly identical samples
in Japan, Alaska, the western United States, and the Appalachian region of the eastern
United States. We look more closely at the consequences of this redundant sampling in
48
Section 3.10. However, for the main analysis of amphibolite facies lithologies in this pa-
per, we keep all amphibolite facies samples in the datasets so that we can see the full span
of available data.
Metamorphic lithologies can be categorized by texture (e.g., schist, gneiss) or pressure-
temperature (P-T) grade (e.g., amphibolite, granulite), and incomplete changes in lithol-
ogy can lead to subjective naming. Unfortunately, >6000 gneisses, schists, and meta-
igneous samples were excluded from this study because they were not accompanied by
mineral assemblage information. Metamorphic texture on its own cannot be correlated to
precise P-T conditions.
To mitigate the oversampling of individual geologic formations, we averaged all
samples of the same facies collected within 0.2?x 0.2?latitude x longitude of each other.
This averaging did not change the median composition of granulite facies xenoliths. The
median composition of eclogite facies xenoliths and terrains, and granulite facies terrains
of all ages changed by < 4%. For amphibolite facies lithologies, however, the median
composition, especially SiO2, increased drastically by >10%.
While an unknown amount of bias plagues our dataset, over 10,000 samples con-
tribute to our understanding of deep crustal composition. Systematic differences among
the different metamorphic lithologies are discussed in the appropriate sections. These
differences, where quantifiable, serve as markers for different possible deep crustal com-
positions. This study focuses on a contextualized overview of compositional aspects; it
does not delve deeply into metamorphic processes. Should any systematic errors funda-
mentally shift our understanding of the deep crust, that in and of itself would be worthy
of future assays.
49
3.6 Major Element Compositions
3.6.1 SiO2, MgO, FeO, and the Daly Gap
The abundance of major oxides in deep crustal analogue samples is difficult to sum-
marize with a single value and uncertainty. Elemental distributions are not always well
defined by the convenient-to-describe Gaussian, normal, log normal, or gamma distri-
bution functions. Table 3.1 reports summary statistics for amphibolite, granulite, and
eclogite facies lithologies major oxide content, but are by no means the most compre-
hensive descriptions of these complex distributions. Unless indicated otherwise, we will
reference the median ? 12 the interquartile ranges because of their resistance to skewness
and outliers.
A discussion of the significance of median versus mean should also include the
practice of evaluating element ratios. Should one consider the representative element
ratio to be represented by a ratio of the means or a mean of the ratios? (Likewise, be rep-
resented by a ratio of the medians or a median of the ratios?) There is no simple answer
to this question; it has been debated extensively without reconciliation. The fundamen-
tal question asks ? how representative is one?s data set of the geological domain being
evaluated? For the deep crust, there are many unknowns including unknown unknowns.
Hence our preference is to use median values and a median of the ratio, as these resist the
influence of skewness and outliers.
Because the distributions of major oxides tend to be skewed and/or multi-modal,
our oxide totals are between 90 and 95% when summing the means or medians. We
50
delve further into assessing the modality of our distributions in Appendix B. Appendix
tables B1 - B6 list distribution parameters and summary statistics for all elements. We
acknowledge our hubris in attempting to parameterize non-parametric distributions, but
we are condemned to using bite-sized descriptions of data in the somewhat Sisyphean
task of quantifying the chemical composition of a crust we cannot easily access.
The most noticeable data trend is the bimodal distribution of primitive and evolved
samples, illustrated by Figure 3.1. The phenomenon published by Rudnick and Presper
(1990) persists in this dataset of over 4,000 granulite facies samples and is also present in
over 6,000 amphibolite facies samples. Granulite facies xenoliths are dominantly mafic,
having <55 wt.% SiO2 and ranging from mantle-like Mg#?s ? 89 to Mg#?s of 45-50.
Granulite facies terrains encompass both mafic and felsic compositions. The felsic sam-
ples follow a Fe-enrichment/Mg-depletion trend, leading to a double-peaked structure,
resembling a chair, when plotted in Mg# vs. SiO2 space. There is also an age-dependent
trend in composition within the granulite facies terrains dataset: older, Archean samples
are more evolved than Post-Archean samples. Amphibolite facies lithologies show the
same chair-like structure, but with a greater concentration of mafic samples. No distinc-
tion is made between amphibolite facies terrains and xenoliths in the dataset because of
the scarcity of amphibolite facies xenolith data.
The corollary to this bimodality is the ?missing? intermediate samples between 53
and 68 wt.% SiO2. The well-documented Daly Gap (Daly, 1914) describes an appar-
ent dearth of intermediate composition rocks, and is collectively observed in all 586,000
metamorphic and igneous samples in the Earthchem.org database. Thermodynamic insta-
bility of intermediate compositions (Daly, 1914; Dufek and Bachmann, 2010) and liquid
51
100 100
Amphibolite Facies A Granulite Facies B 0.30Lithologies 0.30 Xenoliths
80 80 0.25
0.25
0.20
60 0.20 60
0.15 0.15
40 40
0.10 0.10
20 20
N = 4595 0.05 N = 1215 0.05
0 0
0 20 40 60 80 100 0 20 40 60 80 100
SiO2 (wt.%) SiO2 (wt.%)
100 Post-Archean Granulite 100C Archean Granulite 0.15Facies Terrains 0.12 Facies Terrains D
80 80
0.10
60 0.08 60
0.10
40 0.06 40
0.04 0.05
20 20
N = 1348 0.02 N = 1268
0 0 0
0 20 40 60 80 100 0 20 40 60 80 100
SiO2 (wt.%) SiO2 (wt.%)
Figure 3.1: Mg# vs. SiO2 for A) amphibolite lithologies, and granulite facies B) xeno-
liths, C) Post-Archean terrains, and D) Archean terrains. Color indicates relative data
point density. Blue and red fields mark mafic and felsic SiO2 abundances. Mg# is calcu-
lated as molar [Mg][Mg]+[Fe] . All show high concentrations of mafic and/or felsic compositions
and comparatively few compositions of intermediate SiO2.
52
Mg# Mg#
% Data Density % Data Density
Mg# Mg#
% Data Density % Data Density
immiscibility (Charlier et al., 2011; Reubi and Blundy, 2009), among other hypotheses
(Jackson et al., 2018; Yamasaki, 2018), have been proposed to explain the gap. While
it is possible that these rocks are not representative of the crust, we conclude that this is
dubious given its coherence across multiple lithologies.
The systematically mafic composition of granulite facies xenoliths was noted by
Rudnick and Presper (1990) along with many other studies thereafter. Among the pro-
posed explanations for the relative abundance mafic xenoliths are that felsic xenoliths are
less likely to survive the eruption process (Halliday et al., 1993; Rudnick and Fountain,
1995) and that xenoliths might sample deeper regions of the crust than terrains (Bohlen
and Mezger, 1989; Rudnick and Fountain, 1995). Terrains, on the other hand might be
biased towards sampling shallower or more felsic regions because mafic terrains are less
buoyant and less likely to reach the surface (Gerya et al., 2008). Granulite facies ter-
rains also show aged-based compositional biases, with Archean samples (61.5? 8.5 wt.%
SiO2) being, in general, more evolved than Post-Archean samples (61.5? 8.5 wt.% SiO2)
despite having similar median values. Studies have suggested that the ages recorded in
these high grade metamorphic samples have been affected by open system behavior (Ash-
wal et al., 1999) or, as more traditionally argued, hotter temperatures in the Archean al-
lowed for greater amounts of delamination of mafic material, leaving the Archean crust
enriched in felsic components (Martin, 1986).
There is no discernible compositional difference between granulite facies terrains
and xenoliths of comparable SiO2. Most other compositional trends, such as elevated
median CaO in granulite facies xenoliths or rare earth element enrichment in terrains
(discussed later), correlate to the sample?s silica content. The composition of granulite
53
facies lithologies seems to have little dependence on location (other than the fact that
xenoliths are generally most accessible in regions that have experienced volcanism); if
surface transport mechanisms are affecting the composition of these granulite samples,
then they are not doing so beyond preferentially selecting for certain SiO2.
The strong preference for mafic compositions in amphibolite facies lithologies is
likely biased by mineralogy and geologic naming conventions. Amphibolite facies litholo-
gies unsurprisingly contain amphibole minerals, which generally form in mafic rock com-
positions. Felsic rocks of similar metamorphic grade seem to be categorized as schists,
gneisses, or even metapelites. It is likely that many amphibolite facies samples were
excluded from our study because they were given a textural metamorphic grade designa-
tion. Thousands of intermediate and felsic gneisses could not be assigned to amphibolite
or granulite facies because of insufficient metadata.
The eclogite facies xenoliths and terrains are limited to 46.2?1.2 and 47.2?2.2
wt.% SiO2, respectively. This is likely due to the stricter definition of ?eclogite?, which
can refer to a bi-mineralic rock or require basaltic mineral assemblages to reach high
pressure. Eclogite facies lithologies have Mg# of 30 to 90, with no correlation to location
or method of surface transport.
54
Table 3.1: Major Element Compositions
Mean Median Geo- ?- STD IQR Geo- ?- N (fil- N (origi-
Mean Mean STD STD tered) nal)
Amphibolite Facies Lithologies
SiO2 59.1 57.1 58.4 59.1 9.43 18.2 1.17 9.37 2240 5490
TiO2 0.830 0.730 0.684 0.830 0.494 0.653 1.92 0.50 2240 5490
Al2O3 14.8 14.8 14.7 14.8 1.57 2.15 1.11 1.58 2240 5490
FeOT 5.31 5.26 4.33 5.31 2.88 5.03 2.03 3.29 847 2510
MnO 0.130 0.138 0.111 0.130 0.0650 0.120 1.85 0.07 2220 5430
MgO 4.13 3.70 2.93 4.13 2.81 5.03 2.57 3.27 2240 5490
CaO 5.94 5.69 4.58 5.94 3.61 6.95 2.23 4.13 2230 5480
Na2O 2.96 2.97 2.78 2.96 0.985 1.46 1.46 1.06 2230 5480
K2O 1.72 1.32 1.18 1.72 1.33 2.05 2.60 1.42 2230 5470
P2O5 0.148 0.138 0.125 0.148 0.0830 0.110 1.84 0.084 2157 5284
Mg# 46.4 46.8 45.0 46.4 10.6 15.1 1.28 11.2 2220 5430
Granulite Facies Xenoliths
SiO2 51.6 50.2 51.3 51.6 5.05 4.96 1.10 4.88 147 1490
TiO2 0.975 0.923 0.879 0.975 0.415 0.581 1.62 0.436 144 1480
Al2O3 16.2 16.2 16.1 16.2 1.74 2.31 1.11 1.75 147 1490
FeOT 6.76 6.71 6.42 6.76 2.15 2.75 1.39 2.16 81 723
MnO 0.150 0.147 0.143 0.150 0.043 0.056 1.36 0.045 143 1440
55
MgO 7.05 7.19 6.60 7.05 2.29 3.14 1.48 2.55 145 1480
CaO 9.23 9.56 8.83 9.23 2.48 3.27 1.37 2.73 147 1490
Na2O 2.56 2.48 2.44 2.56 0.744 1.14 1.38 0.783 147 1490
K2O 0.779 0.658 0.604 0.779 0.543 0.716 2.10 0.535 147 1480
P2O5 0.163 0.153 0.130 0.163 0.101 0.132 2.08 0.106 143 1434
Mg# 56.9 57.1 56.2 56.9 9.05 12.9 1.18 9.16 145 1480
Post Archean Granulite Facies Terrains
SiO2 59.3 61.5 58.6 59.3 8.73 16.10 1.16 8.83 145 1660
TiO2 0.871 0.763 0.752 0.871 0.438 0.650 1.78 0.462 122 1630
Al2O3 15.8 15.5 15.6 15.8 2.05 2.88 1.14 2.04 122 1630
FeOT 5.75 5.59 5.33 5.75 2.10 3.33 1.50 2.21 81 758
MnO 0.129 0.130 0.120 0.129 0.0447 0.0711 1.50 0.049 120 1600
MgO 5.06 4.10 3.79 5.06 3.64 5.06 2.25 3.70 123 1630
CaO 6.11 5.41 4.83 6.11 3.76 7.17 2.08 4.05 123 1630
Na2O 2.51 2.50 2.33 2.51 0.860 1.32 1.51 0.954 122 1620
K2O 1.87 1.73 1.30 1.87 1.36 2.09 2.66 1.53 155 1700
P2O5 0.150 0.122 0.119 0.150 0.100 0.111 2.05 0.100 112 1454
Mg# 48.0 45.8 46.5 48.0 12.0 15.7 1.28 11.9 119 1610
Archean Granulite Facies Terrains
SiO2 60.1 61.5 59.5 60.1 8.44 16.9 1.15 8.52 123 1530
TiO2 0.655 0.609 0.575 0.655 0.338 0.422 1.68 0.327 122 1490
Al2O3 15.3 14.9 15.2 15.3 2.10 2.36 1.14 2.02 122 1510
56
FeOT 5.12 4.13 4.06 5.12 3.28 5.08 2.05 3.37 78 902
MnO 0.117 0.101 0.101 0.117 0.061 0.094 1.79 0.063 112 1440
MgO 4.36 3.72 3.40 4.36 2.88 4.53 2.10 2.96 123 1520
CaO 5.33 4.49 4.25 5.33 3.24 5.65 2.08 3.48 122 1510
Na2O 3.05 3.24 2.83 3.05 1.06 1.63 1.51 1.17 120 1500
K2O 1.50 1.20 1.14 1.50 1.06 1.67 2.18 1.07 125 1630
P2O5 0.163 0.156 0.144 0.163 0.082 0.087 1.67 0.080 105 1426
Mg# 48.3 47.0 47.3 48.3 9.81 12.2 1.22 9.56 122 1500
Eclogite Facies Xenoliths
SiO2 46.2 46.2 46.1 46.2 1.91 2.46 1.04 1.93 15 173
TiO2 0.617 0.607 0.490 0.617 0.344 0.621 2.15 0.406 15 173
Al2O3 16.1 15.7 15.9 16.1 2.79 2.76 1.17 2.61 15 173
FeOT 8.56 8.74 8.39 8.56 1.64 2.63 1.23 1.72 6 46
MnO 0.186 0.176 0.183 0.186 0.037 0.035 1.20 0.034 15 172
MgO 11.7 11.6 11.3 11.7 3.02 4.34 1.31 3.07 15 173
CaO 11.4 11.3 11.1 11.4 2.49 4.99 1.26 2.55 15 173
Na2O 2.06 1.58 1.61 2.06 1.49 1.31 2.06 1.39 15 173
K2O 0.375 0.200 0.211 0.375 0.404 0.380 3.09 0.374 14 131
P2O5 0.065 0.063 0.055 0.065 0.030 0.035 1.97 0.037 12 86
Mg# 51.8 47.9 50.8 51.8 9.90 15.2 1.21 9.71 13 123
Eclogite Facies Terrains
SiO2 47.5 47.2 47.4 47.5 2.64 5.33 1.06 2.65 14 60
57
TiO2 1.33 1.09 1.14 1.33 0.715 1.23 1.79 0.732 14 60
Al2O3 14.9 15.0 14.8 14.9 2.14 3.26 1.16 2.16 14 60
FeOT 9.06 8.59 8.48 9.06 3.27 6.30 1.44 3.27 11 31
MnO 0.227 0.192 0.212 0.227 0.089 0.115 1.41 0.080 14 60
MgO 8.11 7.98 7.80 8.11 2.42 2.80 1.31 2.25 14 60
CaO 11.1 11.1 10.9 11.1 2.13 3.85 1.22 2.18 14 60
Na2O 2.74 2.56 2.36 2.74 1.57 2.25 1.72 1.46 14 59
K2O 0.454 0.428 0.314 0.454 0.334 0.691 2.54 0.371 12 57
P2O5 0.134 0.108 0.097 0.134 0.106 0.117 2.27 0.102 14 57
Mg# 41.7 40.8 40.8 41.7 8.84 9.60 1.24 8.95 14 60
STD - standard deviation
IQR - interquartile range
Geo-Mean - geometric mean
Geo-STD - geometric standard deviation (reported in log units)
? - gamma function mean, ? ?? (scale parameter ? shape parameter)
?
??ST D - gamma function standard deviation, ? ??2
3.6.2 The Constancy of Al and Ga
Notably, Al2O3 content remains relatively constant (i.e., ?12% variation) through-
out all samples. Though eclogite facies lithologies have slightly elevated Al2O3 content
compared to the other samples (Table 3.1), estimates for Al2O3 only range from 14-17
58
wt.%. The Al2O3 values of granulite facies lithologies are roughly 5-15% lower than the
commonly accepted lower crustal Al2O3 values of Rudnick and Gao (2014) though still
within the study?s given error. Our estimated Al2O3 content in granulite facies lithologies
are more in line with Wedepohl (1995) and Gao et al. (1998) lower crustal values.
Elements of the same group in the periodic table tend to behave similarly. For
example, the abundance of Ga tracks with Al and Ge tracks with Si (De Argollo and
Schilling, 1978). Comparable to Al, Ga concentrations are nearly constant in amphibolite
and granulite facies lithologies: median abundance ranges from 17.3 to 19.5 ppm. Eclog-
ite facies samples again behave differently. Due to the significantly smaller sample sizes
of the eclogite lithologies. There is little or no data reported for Ge and so we predict its
concentration in the deep crust to be relatively invariable at about 1.3-1.4 ppm, based on
chemical trends for igneous rocks (De Argollo and Schilling, 1978).
3.6.2.1 Understanding Protolith Populations
A comparison of the molar abundances of Al to alkali metals and alkaline earths
provides a potential provenance indicator for the origin of deep crustal rocks. Sedimen-
tary rocks typically have Al2O3 contents of ? 20 wt.% (Taylor and McLennan, 1985),
whereas most igneous rocks vary from 12 to 19 wt.% Al2O3 (De La Roche et al., 1980).
Anorthosites and other plagioclase-rich cumulate rocks, however, can have much higher
Al2O3 contents.
Al content and Aluminum Saturation Index have in the past been used to help infer
the protolith of deep crustal samples. When a rock?s Aluminum Saturation Index (ASI;
59
molar Al2O3/(CaO + Na2O + K2O)) > 1, it is classified as peraluminous (Zen, 1988), but
with no characterization of its source of origin, so caution is needed. Though sedimentary
rocks tend to have higher Al contents (ASI = 1.12, excluding carbonates), Keller et al.
(2015) found that common crustal cumulates are also often Al2O3-rich and. Zen (1988)
reported granites having ASI values between 1 and 1.4 and noted that these rocks can
be derived from a variety of source lithologies, with the proviso that for large bodies
of strongly peraluminous granitic rocks peraluminous sources seem necessary. Chappell
et al. (2012) observed that many I(igneous)-type granites are peraluminous and owe their
origins to partial melting of more mafic source rocks. They also noted that gradations
from peraluminous felsic granites to metaluminous igneous compositions are seen for
rock suites that have a shared, closed isotopic system.
Unknowns remain significant regarding the abundance of sedimentary lithologies
the deep crust. Our samples have median ASI values ranging from 0.65 to 1.06, yet the
distribution of aluminous indices is wide and sometimes asymmetrical (Figure 3.2). Am-
phibolite facies lithologies and granulite facies xenoliths have ASI values comparable to
igneous lithologies (Earthchem.org data, median ASI = 0.76), but eclogite facies litholo-
gies have a median ASI values lower than either. The eclogite facies distributions lack
the long tail of higher ASI values which will skew their medians somewhat despite its
robustness to outliers. If eclogite facies lithologies were not sufficiently sampled to en-
counter high ASI samples, then it is possible that their actual median ASI value is higher.
The change in eclogite ASI distribution shape would lend the appearance of a log-normal
trend among the datasets.
It is difficult to distinguish between metaigneous and metasedimentary protoliths
60
1000 A Amphibolite Facies 400 B Granulite Facies
Lithologies Xenoliths
500 0.94 200 0.87
0 0
0 1 2 3 0 1 2 3
ASI ASI
400 C Post Archean Granulite 400 D Archean Granulite
Facies Terrains Facies Terrains
200 1.06 200 1.06
0 0
0 1 2 3 0 1 2 3
ASI ASI
30 30
E F
Eclogite Facies Eclogite Facies
20 Xenoliths 20 Terrains
10 0.67 10 0.65
0 0
0 1 2 3 0 1 2 3
ASI ASI
Figure 3.2: Natural log of the aluminum saturation index index (ASI), molar Al2O3/(CaO
+ Na2O + K2O). Red numbers indicate the median value for each sample type. The ASI
value is an ambiguous indicator of a rock?s protolith; rocks with ASI >1 can be from ig-
neous or metasedimentary protoliths can be an indicator of metasedimentary contributions
to sample populations since sediments are typically more Al rich than igneous sources.
The more mafic datasets have smaller ASI values while the granulite facies terrains can
more often have values around 1.
61
# of Samples # of Samples # of Samples
for deep crustal lithologies with certainty (Wilkinson et al., 2009). Hacker et al. (2015)
identified 44% Archean and Post-Archean granulite-facies rocks as peraluminous and
noted that they may be metasedimentary; going further, they suggested that amphibolite-
facies terrains have similar statistics and that 16% granulite-facies xenoliths that are per-
aluminous may be metasedimentary. We do not find compelling evidence that the ASI
value provides unambiguous indication of what is a metasediment. In fact, as cautioned
by Chappell et al. (2012), many peraluminous rocks are igneous, including those de-
rived from the remelting of igneous rocks. Given that the term peraluminous does not
effectively identify what might be a metasediment, we turned to a machine learning algo-
rithm to predict a metamorphic protolith from major element chemistry (Hasterok et al.,
2019b). This method, however, also produced unclear results. The algorithm showed low
confidence in whether the protoliths were igneous or sedimentary, with ratings close to
0 instead of -1 (confidently igneous) or 1 (confidently sedimentary). A broader view of
factors must be considered to determine the formation and evolution processes of the deep
crust.
3.7 Minor and Trace Element Composition
Here we discuss key geochemical trends seen in incompatible elements; other ob-
servations are not covered here for the sake of brevity. The rest of the data is addressed
in more detail in Appendix B, which reviews our findings for fluid mobile elements, high
field strength elements (HFSE), transition metals, and other important groups of elements.
Regardless of surface transport mechanism (eruption as xenoliths or tectonic emplace-
62
ment as terrains), there are no differences in trace element content between granulites
of similar SiO2 content. Therefore, granulite facies xenoliths and terrains can be treated
as one lithology when discussing silica-correlated compositional trends. Eclogite facies
xenoliths and terrains have fewer data points, so it remains unclear whether or not they
should be given the same treatment.
3.7.1 Rare Earth Elements
Figures 3.3 and 3.4 illustrate that the rare earth element patterns of all of the sam-
ples are congruent, having greater variation in the light rare earths (LREE) and than in the
heavy rare earth elements (HREE). The amphibolite and granulite data show LREE en-
richments and their variability is comparable to that shown by Rudnick and Gao (2014),
with granulite facies terrains having the highest median concentrations of La through Nd.
Igneous processes ? rather than metamorphic changes or chemical weathering ? control
the relative enrichment in LREEs seen in granulite facies terrains compared to granulite
facies xenoliths or amphibolite lithologies. The greatest abundance of La and Ce is not
seen in the most hydrated samples (amphibolites) but in the most evolved samples (gran-
ulite facies terrains). Eclogite facies lithologies are relatively depleted in LREE compared
to amphibolites, yet they are more enriched than granulite facies xenoliths. The standard
deviation of the REE distributions narrows from La to Lu. Eclogite facies samples surpris-
ingly show no relative enrichments in HREEs, which would be typical of rocks with more
abundant garnet. The HREEs concentrations, especially Tm, Yb, and Lu are identical for
all of the metamorphic facies in question.
63
64
A B
101 101
Amphibolite Facies Lithologies Granulite Facies Xenoliths
100 100
La Ce Pr Nd  Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd  Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
C D
101 101
Post Archean Granulite Facies Terrains Archean Granulite Facies Terrains
100 100
La Ce Pr Nd  Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd  Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Eclogite Facies Xenoliths E Eclogite Facies Terrains F
101 101
100 100
La Ce Pr Nd  Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd  Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Figure 3.3: Rare earth element median values. The gray shaded bands represent the 1st and 3rd quartile values for each REE.
P . M .  N o r m a l i z e d P . M .  N o r m a l i z e d P . M .  N o r m a l i z e d
A b u n d a n c e A b u n d a n c e A b u n d a n c e
P . M .  N o r m a l i z e d P . M .  N o r m a l i z e d P . M .  N o r m a l i z e d
A b u n d a n c e A b u n d a n c e A b u n d a n c e
35
Rare Earth Elements - Medians
Amphibolite Facies Lithologies
30 Granulite Facies Xenoliths
Post Archean Granulite Facies Terrains
25 Archean Granulite Facies Terrains
Eclogite Facies Xenoliths
20 Eclogite Facies Terrains
15
10
5
0
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Figure 3.4: While the most variation is seen in the light rare earth elements, the heavy
rare earth elements are tightly knit around 0.25 ppm (5x primitive mantle). Note the
linear scale. The black line and gray shaded region surrounding it is the Rudnick and Gao
(2014) lower crustal composition ? 15%.
Amphibolite facies lithologies and granulite facies xenoliths span similar ranges of
La/Yb, 8.16 ? 6.5 and 7.52 ? 3.2, respectively. Archean granulite facies terrains have a
much higher median value (La/Yb = 16.0 ? 10.2) than Post-Archean terrains (La/Yb =
10.1? 4.50) despite having similar SiO2 content, yet La/Yb and is correlated to SiO2 and
forms the same chair-like structure when in natural log La/Yb space as SiO2 vs. Mg#. The
bimodal structure suggests that the Daly Gap affects La/Yb, and that the ratio reflects the
original igneous processes that formed the rock rather than metamorphism or weathering.
Flatter rare earth patterns (lower La/Yb ratios), common among the eclogite facies
xenoliths and terrains, do not seem to be relegated to specific regions. It is possible that
eclogite facies terrains are biased by alteration and subduction processes despite how
closely they resemble xenoliths because of our limited dataset (Tsujimori et al., 2006).
65
P.M. Normalized Abundance
As mentioned earlier, many of our eclogite facies terrains samples originate from areas
that have evidence of significant subduction, such as the western United States. Eclogite
facies terrains from the Caledonides in Norway (Rockow et al., 1997; Svensen et al.,
2001) also show REE variation even within the same formation. Sampling exposures of
eclogite facies terrains in regions that have not been subjected to significant amounts of
subduction would provide more clarity ? if such terrains exist. The apparent enrichment
in Gd, Dy, Ho, Er, and Tm in eclogite facies terrains is due to our limited dataset, with
those elements having only 1 to 5 datapoints each.
On average the crust shows a systematic vertical concentration gradient in REE
abundances (i.e., UC ? MC ? LC, showing La 36 ? 18 ? 12; Yb 3.1 ? 2.2 ?
1.7 ppm), with a mildly fractionated downward decrease in the LREE (factor of 3) and
HREE (factor of 2). Likewise, Eu/Eu? changes from a 30% negative anomaly in the upper
crust to essentially no anomaly in the lower crust. These compositional gradient are most
likely products of intracrustal differentiation, with granite magmas moving upward and
residuals being stored in the lower crust or lost to the mantle via gravitational processes.
3.7.2 Heat Producing Elements
Heat producing elements (HPEs: K, Th and U) are of particular interest because
they are crucial to understanding Earth?s radioactive heat budget (these three elements
produce 99.5% on the radiogenic heat) through time as well as the temperature and
strength of the crust. Rudnick and Gao (2014) estimate the continents host 35 to 40% of
the Earth?s budget of HPE. Constraining Earth?s HPE abundances (especially the abun-
66
dance of the refractory lithophiles, e.g., U and Th) also constrains ?26 other elements
(McDonough and Sun, 1995) that are in conserved, chondritic ratios relative to U and Th.
3.7.2.1 Th, U, and K
The behavior HPE can often be understood through comparisons of elemental ra-
tios. About 80% of the Earth?s total heat production comes from Th and U and thus the
Th/U ratio is key. Wipperfurth et al. (2018) recently reviewed Th/U values for ?150,000
crustal rocks and sediments and found that the median values for igneous and metamor-
phic rocks were close to the bulk Earth?s value of 3.8. We find that amphibolites have
a median Th/U of 3.7, whereas granulite terrains appear to have lost U (see below) rel-
ative Th (median Th/U for Archean and Post-Archean granulite terrains are 7.3 and 6.6,
respectively). In contrast, the median Th/U of granulite xenoliths (3.4) appears to be nor-
mal (i.e., no U loss). There is little correlation between K content and K/U, with median
K/U values for all metamorphic lithologies ranging upwards of one to three times that
of upper continental crust. We observe K/U values from 10,000 to 100,000 with uncer-
tainties on the order of 60%. Whether K behaves as a trace element or thermodynamic
component (i.e., mineral) controls the K abundance in our samples. That is, K abundance
is high if K-feldspar is present in the system, whereas K is low if K-feldspar is not present.
The K/Th value is relatively constant in deep crustal lithologies and similar to that of the
upper crust (Alessio et al., 2018; Rudnick and Gao, 2014), implying K/U fractionation
is due to U loss, not K loss, and supporting the earlier finding of Rudnick and Presper
(1990).
67
A question that remains is, how long ago did this uranium loss occur in the gran-
ulite terrains? To address this issue we combined the two separate measures of Th/U.
The isotopic ratio of 232Th/238U value is referred to as ? (? ? Th/U ? 1.033), while
the time-integrated Pb isotopic ratio (208Pb?/206Pb?, the decay products of 232Th/238U) is
?Pb, which serves as a proxy for Th/U. [ ?Pb values are calculated from the measured lead
isotopic composition of the sample minus its primordial lead contribution; see details in
Wipperfurth et al. (2018).] The ?Pb provides a measure of the time-integrated Th/U value
and is resistant to recent resetting. The average (and median) ?Pb for the amphibolites,
granulite xenoliths, and Archean and Post-Archean granulite terrains is 4.1?0.1 (4.0?0.1;
n=165, 357, 33, and 4, respectively), while eclogite xenoliths are 5.5 (5.8; n=21) and no
data for eclogite terrains; see Table S3 for further details. The ?70% difference between
Th/U and ?Pb values for granulite facies terrains is consistent with a relatively recent
(< 450 Ma) uranium loss. On average it appears that exhumation or surficial weather-
ing processes result in the loss of U from the granulite and less so due to dehydration
metamorphism.
In general, U and Th show positive correlations with SiO2. Mean and median U
and Th values increase with increasing SiO2 abundance for amphibolite and granulite
facies rocks. While the relationship between SiO2 and U or Th is log-normal within
uncertainties, the concentration of U and Th could potentially also be derived from SiO2
through a probability analysis (Gard et al., 2019; Hasterok et al., 2019a).
Deep crustal heat production is but a fraction of upper crustal heat production. The
median heat production in the deep crustal lithologies ranges from 0.04 to 0.41 nW/kg
(roughly 0.1 to 1.2 ?W/m3, assuming a density of 2900 kg/m3). Post Archean granulite
68
facies terrains have the highest heat production to 0.4 ? 0.5 nW/kg (?1.2 ?W/m3). How
we calculate our heat production is significant: the mean of the averages does not equal
the average of the means. Using the median K2O, Th, and U abundances, we calculate
heat production for Post-Archean and Archean granulite facies terrains to be 0.14 and
0.15 nW/kg (0.41 and 0.42?W/m3), respectively. The answer lies in the shape of the dis-
tributions, which are neither normal nor log-normal. In this case, our simplified statistics
are sub par descriptors of the datasets. Yet, we see that median deep crustal heat produc-
tion should be minimal unless there is significant incorporation of granulite facies terrain
material.
3.8 Distributions that Trend, Periodically
The periodic table is a wonderful tool to use when analyzing elemental trends be-
cause of its structure and organization. It shows that elements of similar valence states
and radii behave predictably and highlights anomalies caused by specific minerals or sam-
pling methods. We expect to see more skewed distributions for elements that change
abundance with depth, since our dataset includes samples from a range of depths (Bohlen
and Mezger, 1989). The difference between mean and median is one metric for quickly
assessing the shape of non-normal distributions. McDonough (1990) found that major
and compatible trace elements have similar average and median values, whereas median
values are systematically lower than average values for the incompatible trace elements
(e.g., LREE, K, Rb), with differences between average and median values increasing with
increasing incompatibility. Figures 3.5 and 3.6 are color coded to show the % difference
69
between mean and median for amphibolite and granulite facies lithologies. The same
methodology can be applied to eclogite facies lithologies, but the distributions are more
discontinuous, and trends are less clear due to having an order of magnitude fewer sam-
ples. Even in the amphibolite and granulite datasets, elements with relatively few data
points (such as the highly siderophile elements) appear highly skewed.
Al and Ga are unimodal, with little variation in their abundances compared to other
elements, and Na is relatively constant, with some possible bimodality. The mean and
median values for Si are similar because of its bimodal distribution, but we do not find
conclusive evidence for bimodality in other oxides. Fe, Mg, and Ca show some degree
of multi-modality. Most of other elements in the table are unimodal. The rare earth
elements exhibit consistent patterns between amphibolite and granulite facies lithologies,
though amphibolites have greater differences between mean and medium in the light rare
earth elements. The homogeneity of the rare earth element distributions underscores their
predictable behavior, with the greatest skewness occurring in the light rare earths which
tapers off to a steady ?10% difference between mean and median towards the heavy rare
earths.
Both Th and U have highly skewed distributions that verge on log-normal for both
amphibolite and granulite facies lithologies. The distributions of U and Th in granulite
facies terrains are indistinguishable from both a gamma and a log-normal distribution
(using a Wilcoxon rank sum test of median values). The distributions of U and Th in
granulite facies xenoliths and amphibolite facies lithologies, however are more accurately
described by log-normal distributions. That is to say, the misfit between either the Th or U
distribution and log-normal distributions with the same ? and ? is negligible according to
70
71
hydrogen helium 90
0 0
H He
0 0
lithium beryllium boron carbon nitrogen oxygen fluorine lithium 80
18 1.5 12 0 0 0 230 0
Li Be B C N O F Ne
653 336 326 0 0 0 243 0
sodium magnesium Amphibolite Facies Lithologies aluminum silicon phosphorus sulfur chlorine argon 70
2.7 5.4 15 52 0.11 60 12 0
Na Mg Al Si P S Cl Ar
4921 4940 4942 4943 149 262 172 0
potassium calcium scandium titanium vanadium chromium manganese iron cobalt nickel copper zinc gallium germanium arsenic selenium bromine krypton 60
0.96 7.8 28 0.79 170 120 0.16 8.4 38 55 36 81 18 0 1.3 50 0 0
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr
4887 4933 2843 4926 3317 3784 4723 4888 3129 3862 2646 2970 2070 0 324 142 0 0
rubidium strontium yttrium zirconium niobium molebdenum technetium ruthenium rhodium palladium silver cadmium indium tin antimony tellurium iodine xenon 50
30 180 22 100 6.4 0.65 0 0 0 0.5 43 50 0.07 1.6 0.21 0.05 0 0
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
3618 4255 4000 4182 3401 308 0 0 0 195 179 171 17 320 447 11 0 0
caesium barium hafnium tantalum tungsten rhenium osmium iridium platinum gold mercury thallium lead bismuth polonium astatine radon 40
1 230 2.8 0.4 1.1 0.35 0.038 0.15 0.6 0.8 0.04 0.29 9.1 0.062 0 0 0
Cs Ba Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn
1492 3953 2548 2271 368 4 4 14 193 233 9 50 2230 179 0 0 0
francium radium 30
0 0
Fr Ra
0 0 20
lanthanum cerium praseodymium neodymium promethium samarium europium gadolinium terbium dysprosium holmium erbium thulium ytterbium lutetium
element name 14 29 3.6 16 0 3.6 1 3.8 0.62 3.7 0.77 2.2 0.33 2.1 0.32
median La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 10
X 3523 3478 1905 3144 0 3120 3122 2430 2564 2096 1893 2019 1763 3361 2894actinium thorium proctactinium uranium neptunium plutonium americium berkelium californium eisteinium fermium nobelium lawrencium0 2.1 0 0.83 0 0 0 0 0 0 0 0 0 0 0
# of samples Ac Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No Lr 0
0 2996 0 2418 0 0 0 0 0 0 0 0 0 0 0
-10
Figure 3.5: The percent difference between the mean and median for elemental distributions is an indicator of the symmetry of the
distribution. On this periodic table, color correlates to that percent difference. Elements that are grayed out have insufficient data. The
number at the bottom of each element box is the number of available samples, and the top number is the median concentration. Median
concentrations are given in wt.% oxide for Na, Mg, Al, Si, K, Ca, Ti, Mn, and Fe. All other elements are reported in ppm. Similar
trends emerge among elements of the same family and among the lathanides. Heavier, larger elements vary more in amphibolite facies
lithologies than in granulite facies lithologies.
( M e a n  -  M e d i a n ) / M e a n  ( % )
72
hydrogen helium 90
0 0
H He
0 0
lithium beryllium boron carbon nitrogen oxygen fluorine lithium 80
8.4 1 0 0 0 0 440 0
Li Be B C N O F Ne
278 50
sodium magnesium Granulite Facies Lithologies 0 0 0 0 40 0aluminum silicon phosphorus sulfur chlorine argon 70
2.9 4.1 15 55 0.09 34 24 0
Na Mg Al Si P S Cl Ar
4156 4142 4166 4182 13 73 36 0
potassium calcium scandium titanium vanadium chromium manganese iron cobalt nickel copper zinc gallium germanium arsenic selenium bromine krypton 60
0.94 6.4 23 0.67 140 92 0.12 7.2 31 47 22 71 18 0 0.33 0.11 0 0
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr
4151 4169 1768 4131 2750 2984 3901 4030 1815 2882 2042 2313 1652 0 9 4 0 0
rubidium strontium yttrium zirconium niobium molebdenum technetium ruthenium rhodium palladium silver cadmium indium tin antimony tellurium iodine xenon 50
18 290 19 110 6 1 0 0 0 5.5 0 0 0 1.8 0.04 0 0 0
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
3376 3601 2966 3292 2535 79 0 0 0 6 0 0 0 344 70 0 0 0
caesium barium hafnium tantalum tungsten rhenium osmium iridium platinum gold mercury thallium lead bismuth polonium astatine radon 40
0.32 390 2.5 0.34 1.4 0 0.18 0.28 3 1.3 0 0.93 8 0.08 0 0 0
Cs Ba Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn
979 3243 1550 1148 46 0 2 5 7 6 0 16 1934 11 0 0 0
francium radium 30
0 0
Fr Ra
0 0 20
lanthanum cerium praseodymium neodymium promethium samarium europium gadolinium terbium dysprosium holmium erbium thulium ytterbium lutetium
element name 16 34 3.6 16 0 3.4 1.1 3.5 0.56 3.5 0.65 1.9 0.26 1.7 0.26
median La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
10
X 2800 2809 1346 2283 0 2299 2205 1649 1891 1587 1397 1587 447 2206 1897actinium thorium proctactinium uranium neptunium plutonium americium berkelium californium eisteinium fermium nobelium lawrencium0 1.1 0 0.34 0 0 0 0 0 0 0 0 0 0 0 0
# of samples Ac Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No Lr
0 1975 0 1597 0 0 0 0 0 0 0 0 0 0 0
-10
Figure 3.6: Color correlates to the percent difference between the mean and median concentration of a given element in granulite facies
lithologies. Elements that are grayed out have insufficient data. The number at the bottom of each element box is the number of available
samples, and the top number is the median concentration. Median concentrations are given in wt.% oxide for Na, Mg, Al, Si, K, Ca,
Ti, Mn, and Fe. All other elements are reported in ppm. Terrains and xenoliths are treated as one dataset since there is no observable
difference between terrains and xenoliths of comparable SiO2. As with amphibolite facies lithologies, Al and Ga are the narrowest, least
variable distributions. Similar trends emerge among elements of the same family and among the lathanides.
( M e a n  -  M e d i a n ) / M e a n  ( % )
the (admittedly simplistic) statistical test mentioned above. On the other hand, the misfit
between Th or U and the corresponding gamma distributions with the same shape and rate
parameters is significant according to the same statistical test. Th and U are expected to
be skewed because their abundance changes rapidly as a function of depth (Huang et al.,
2013; Rudnick and Fountain, 1995; Rudnick and Gao, 2014).
3.9 A Basalt, by Any Other Name
If the deep (particularly the lower) continental crust looks like granulite facies xeno-
liths, or to an extent amphibolite facies lithologies, then it looks like a basalt. Similarities
between these deep crustal samples and mid ocean ridge basalt (MORB) and ocean island
basalt (OIB) span SiO2, Al2O3, MgO, and MnO (Fig. 3.7, Table 3.2). CaO and Na2O
differ by about 10% among the three different basalts, with MORBs having the highest
concentration of both. An important note, though, is that all of our deep crustal analogues
are depleted in Ti compared to MORB or OIB. Since there is no complementary Ti enrich-
ment in the upper crust (Rudnick and Gao, 2014; Taylor and McLennan, 1985; Wedepohl,
1995), there may exist an unsampled Ti reservoir on Earth (McDonough, 1991; Rudnick
et al., 2000).
If the deep crust resembles granulite facies terrains, then it differs more substantially
from MORB and OIB (Fig. 3.8). Granulite facies terrains have 16-27% higher SiO2
content than MORB and 3 to 4 times higher concentrations of incompatible elements as a
result. However, while comparable in major element space, both granulite facies xenoliths
and terrains have on average 13 to
1
2 the concentration of LREEs of OIBs. The undepleted
73
Amph. Gran. PA A Gran. Ecg. Ecg. MORB OIB
Lith. Xen. Gran. Ter. Xen. Ter.
Ter.
SiO2 62.2 53.3 63.18 65.4 48.0 50.1 50.7 50.1
TiO2 0.79 0.98 0.78 0.65 0.63 1.16 1.69 2.53
Al2O3 16.1 17.2 15.9 15.8 16.3 15.9 14.8 14.1
FeOT 5.73 7.12 5.74 4.39 9.09 9.11 10.47 11.37
MnO 0.15 0.16 0.13 0.11 0.18 0.20 0.18 0.16
MgO 4.03 7.63 4.21 3.96 12.06 8.47 7.61 8.53
CaO 6.20 10.15 5.56 4.77 11.75 11.78 11.44 10.44
Na2O 3.23 2.63 2.57 3.45 1.64 2.72 2.80 2.16
K2O 1.44 0.70 1.78 1.28 0.21 0.45 0.16 0.38
P2O5 0.15 0.16 0.12 0.17 0.07 0.11 0.18 0.25
Mg# 46.8 57.1 45.8 47.0 47.9 40.8 - -
Mg# 55.6 65.6 56.7 61.6 70.3 62.4 56.4 57.2
Calc.
Table 3.2: Median major oxide compositions for our sample sets, mid ocean ridge basalts
(MORB, (Gale et al., 2013)), and ocean island basalts (OIB, (MacDonald and Katsura,
1964))
3.5
Hawaiian Tholeiites
3 Granulite Facies Xenoliths
2.5
2
1.5
1
0.5
0SiO2 TiO2 Al2O3 FeOT MnO MgO CaO Na2O K2O
Figure 3.7: Ocean island basalt Hawaiian tholeiites (MacDonald and Katsura, 1964)
and granulite facies xenoliths have similar major element patterns except in TiO2. Both
resemble mid ocean ridge basalts (Gale et al., 2013) except in TiO2 and K2O.
74
MORB Normalized Abundance
8
7 OIB averageAmphibolite Facies Lithologies
6 Granulite Facies XenolithsPost Archean Granulite Facies Terrains
Archean Granulite Facies Terrains
5 Eclogite Facies Xenoliths
Eclogite Facies Terrains
4
3
2
1
0
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Figure 3.8: Rare earth element concentrations for granulite and amphibolite facies
lithologies fall between MORB (Gale et al., 2013) and OIB (Arevalo and McDonough,
2010) abundances. The shape of their REE patterns, however, resemble OIB moreso than
MORB. Amphibolite, granulite, eclogite, and OIB rare earth element patterns converge
towards the heavy rare earths.
or otherwise uniquely enriched source of OIB material is not reflected in crustal basalts.
Eclogite facies lithologies are essentially basaltic in their bulk compositions and
comparable to that of MORBs & OIBs (Table 3.2) when normalized to 100 wt.%. With
16 wt.% Al2O3, 9 wt.% FeOT , and varying amounts of MgO, however, we expect eclog-
ite facies lithologies to maintain their traditionally-higher-than-basaltic densities. Differ-
ences in expected deep crustal densities, discussed below, suggest that eclogites will by
and large be gravitationally unstable in most lower crustal models.
75
MORB Normalized
3.10 Constructing the Continental Crust
The composition of the deep continental crust is a direct result of the many pro-
cesses through which it was constructed and evolved. The abundance of incompatible
elements in our dataset provides a tool for analyzing the probability of different deep
crustal compositions.
Using a simplified, 3-layer modeling approach, we identify the elements that stand
out as markers of different crustal compositions and formation processes, and which con-
tribute significantly to estimates of bulk silicate Earth (BSE) composition. Our bulk con-
tinental crust composition is calculated by weighting the elemental abundances from the
upper, middle, and lower crust by each layer?s mass fraction (Table 3.3). The mass frac-
tions of these crustal layers are from Wipperfurth et al. (2020). The upper crust?s com-
position uses Gaschnig et al. (2016)?s concentrations for all elements for the top 13 except
Sr and Mo (Rudnick and Gao, 2014), and Rudnick and Gao (2014) for the bottom 23 of
the upper crust. We take the upper crust HSEs from (Chen et al., 2016), and Ag from
(Chen et al., 2020). We use amphibolite facies lithologies as representatives of a middle
crustal composition and split the lower crust 50/50 between granulite facies xenoliths and
terrains. Debate remains regarding models for the composition of the lower crust (Hacker
et al., 2015; Rudnick and Gao, 2014); there is no obvious solution to determining the
composition of the lower crust. Our solution for deciding on a model for the lower crust?s
composition recognizes that granulite facies terrains come from on average ?0.8 GPa
(Brown and Johnson, 2019), approximately the upper portion of the lower crust, whereas
granulite facies xenoliths appear to dominate the bottom of the lower crust (Rudnick and
76
Gao, 2014).
Table 3.3: Recommended Continental Crust Composition
Upper Crust Middle Crust Lower Crust Deep Crust Bulk Crust
SiO2 68.0 62.2 53.3 57.6 61.1
TiO2 0.663 0.795 0.980 0.889 0.812
Al2O3 15.1 16.1 17.2 16.66 16.1
FeOT 5.21 5.73 7.12 6.44 6.02
MnO 0.100 0.150 0.156 0.153 0.135
MgO 2.29 4.03 7.63 5.87 4.66
CaO 2.75 6.20 10.15 8.21 6.36
Na2O 2.63 3.23 2.63 2.93 2.82
K2O 3.11 1.44 0.698 1.06 1.75
P2O5 0.169 0.150 0.162 0.156 0.161
Li 26.8 15.0 6.89 10.9 16.0
Be 2.18 1.44 0.478 0.956 1.35
B 10.6 9.00 - - -
N 51.1 - - - -
F 343 399 - - -
S 382 22.0 140 81 178
Cl** 181 29.3 151 90.5 120
Sc 13.3 21.0 28.0 24.5 20.9
V 88.1 134 186 160 137
77
Cr 77.2 81.0 168 125 109
Co 15.2 29.9 46.8 38.4 30.9
Ni 39.2 39.7 100 70.1 60.1
Cu 26.3 30.0 37.8 33.9 31.5
Zn 69.3 78.0 81.1 79.6 76.2
Ga 17.8 18.0 17.3 17.6 17.7
Ge 1.53 - - - -
As 2.92 1.30 - - -
Se 0.058 0.0530 - - -
Br 0.98 - - - -
Rb 92.8 43.8 10.6 27.1 48.3
Sr? 320 201 465 334 329
Y 25.7 22.5 19.0 20.7 22.3
Zr 203 123 83.3 103 135
Nb 12.1 7.20 7.00 7.10 8.73
Mo?** 1.12 0.520 1.90 1.21 1.18
Ru 0.240 - - - -
Rh - - - - -
Pd** 0.000800 0.000850 0.00554 0.00321 0.00214
Ag? 0.0329 0.0480 - - -
Cd 0.097 0.0600 - - -
In 0.058 0.0710 - - -
Sn 2.23 1.60 1.58 1.59 1.80
Sb 0.39 0.200 - - -
78
Te - 0.0500 - - -
I 0.862 - - - -
Cs 4.18 1.19 0.390 0.787 1.88
Ba 665 330 393 362 460
La 33.3 18.1 11.6 14.8 20.8
Ce 67.0 36.5 27.0 31.7 43.1
Pr 7.65 4.68 3.48 4.08 5.23
Nd 29.4 18.3 14.7 16.5 20.7
Sm 5.38 4.10 3.57 3.83 4.33
Eu 1.15 1.09 1.29 1.19 1.18
Gd 4.79 3.91 3.77 3.84 4.15
Tb 0.780 0.664 0.570 0.618 0.670
Dy 4.55 3.91 3.60 3.75 4.01
Ho 0.95 0.822 0.640 0.730 0.800
Er 2.67 2.29 1.89 2.09 2.28
Tm 0.380 0.350 0.254 0.302 0.327
Yb 2.44 2.19 1.70 1.94 2.10
Lu 0.370 0.330 0.245 0.287 0.315
Hf 5.73 3.42 2.05 2.73 3.70
Ta 0.900 0.540 0.597 0.569 0.677
W?** 1.59 0.440 0.310 0.374 0.773
Re? 0.000220 - - - -
Os?** 0.0000418 - 0.0000210 - -
Ir?** 0.0000274 0.0000154 0.0000290 0.0000222 0.0000160
79
Pt? 0.000618 0.000650 0.000249 0.000158 0.000127
Au 0.000938 0.000800 - - -
Hg** 0.0337 0.0300 - - -
Tl 0.805 0.500 - - -
Pb 16.5 11.7 4.50 8.08 10.8
Bi 0.190 0.0700 - - -
Th 10.6 3.68 0.767 2.21 4.90
U 2.64 1.00 0.248 0.621 1.27
Eu/Eu* 0.693 0.830 1.07 0.946 0.848
Heat prod. 0.623 0.235 0.0640 0.149 0.303
(nW/kg)
Heat prod. 1.81 0.682 0.186 0.432 0.878
(?W/m3)
Nb/Ta 13.4 13.3 11.7 12.5 12.9
Zr/Hf 35.5 36.0 40.6 37.7 36.6
Th/U 4.00 3.68 3.09 3.56 3.85
K/U 10500 12800 25100 15200 12300
La/Yb 13.7 8.27 6.82 7.63 9.89
Rb/Cs 22.2 36.8 27.2 34.4 25.7
K/Rb 278 272 547 325 301
La/Ta 36.9 33.5 19.4 26.1 0.71
Oxides reported in wt.%. All other abundances reported in ppm.
80
** denotes elements for which N <6 for middle and/or lower crust
? denotes Upper crustal values from sources other than Gaschnig et al. (2016). Please see text for
sources of ? abundances. W values from Archean granulite facies terrains (n = 3) excluded as
outliers
Though amphibolite facies lithologies are held to represent the middle crust, the
median SiO2 for amphibolite lithologies is ?10% lower than existing estimates. This
makes our bulk continental crustal model more mafic in the middle crust, leaving a poten-
tial mid-crustal deficit in many incompatible elements. A mafic bias in our amphibolite
facies lithologies may exist and seems to stem from:
1) biasedly assigning medium grade mafic metamorphic samples to the ?amphibolite?
facies (due to the hallmark abundance of amphiboles) but not assigning medium grade
felsic metamorphics to the amphibolite facies; and
2) the oversampling of mafic amphibolite facies locations.
Figure 3.9 orders elements left to right from the most to least abundant in the conti-
nental crust, relative to a BSE model. In doing so, we highlight the enrichments of Cs, Rb,
Ba, Pb, U, Th, K, W, and La in the continental crust and identify the crust as an important
host for these elements in the bulk silicate Earth (BSE), despite its insignificant mass con-
tribution (0.55% of the BSE mass). Assuming a BSE composition (McDonough and Sun,
1995; Palme and O?Neill, 2014), the estimated crustal contribution represents 15 to 50%
of Earth?s total budget of these elements sequestered in the continental crust. Importantly,
?35% of the heat producing elements are stored in the crust and not available for driving
mantle convention.
81
The continental crust is often viewed as the complementary reservoir of the De-
pleted Mantle, particularly for the incompatible elements. In Figure 3.9 we compare the
composition of the continental crust with that of the average MORB, a representation for
the upper portion of the oceanic crust. A crude comparison of the composition of the De-
pleted Mantle (DM) can be taken as? 110 the value of MORB. If the upper mantle (mantle
above the 670 km seismic discontinuity; 26% of the mass of the BSE), often considered
the Depleted Mantle, was uniformly depleted to create the continents, then it cannot be
the complement to the continental crust. Elements whose crustal mass contribution ex-
ceeds 26% (i.e., Cs, Rb, Ba, Pb, U, Th, K, and W) require that the lower mantle has been
accessed during the production of continental crust. Therefore, the part of the BSE re-
ferred to as the Depleted Mantle extends considerably into the lower mantle (nearly half
a mantle mass is necessary to account for all of the Rb & Cs in the continental crust). The
implications of this finding demand that all or most of the mantle has been involved, to
some extent, in the production of the continents.
Apparent non-complementary relationships are found for K, Sr, and Li. The average
MORB (Gale et al., 2013) pattern (Figure 3.9) shows marked depletion in the primitive
mantle normalized abundance for K, Sr, and Li relative to adjacent elements. The rela-
tive incompatibility of these elements during mantle melting are K?U, Sr?Pr-Nd, and
Li?Dy (Sun and McDonough, 1989). The K/U values of continental crust and MORB
are complementary relative to that of the BSE (Arevalo et al., 2009; Farcy et al., 2020),
consistent with these patterns (Figure 3.9). The marked depletion in Sr seen in the MORB
pattern is due to these basalts having experienced considerable plagioclase fractionation.
However, as Tang et al. (2017) noted, MORBs with ?10wt.% MgO, primitive MORBs
82
102
Continental Crust, Rudnick & Gao `14
Weighted Continental Crust, This Study
Average MORB, Gale et al.`13
101
80
40
100 Cs Pb Th W* La Mo* Be Sr* Nb Zr Li Gd Dy Er Tm Lu Ga
Rb Ba U K Ce Pr Ta Nd Hf Sm Eu Tb Ho Y Yb Ti Al
Figure 3.9: Weighted proportions of amphibolite and granulite facies lithologies were
combined with upper crustal values (see text) to generate bulk silicate Earth incompat-
ible element abundance estimates. Elements are ordered from left to right by relative
enrichment in the continental crust. Alternatively, this ordering reflects element compati-
bility during mantle melting. Colored circles show our bulk continental crust model while
empty circles plot MORB abundances. Black bars follow the right-hand y-axis and show
the % of each element sequestered in the continental crust relative to BSE. Normalized to
(McDonough and Sun, 1995) with W from (Arevalo and McDonough, 2008).
83
P.M. Normalized Abundance
% of Element in Crust
(Crust/BSE %)
that have yet to experience plagioclase fractionation, do not show any depletion in Sr.
Finally, the depletion in Li is somewhat more challenging to explain. Lithium?s position
(established by crustal abundances) suggests its relative incompatibility is enhanced due,
most likely, to a combination of melting and weathering processes. That said, however,
the continental crust, which hosts only ?6% of the BSE?s Li budget, cannot account for
the marked depletion seen in the MORB pattern. (Moving the MORB Li data point over
to the heavy REE still does not account for its depletion.) We offer no explanation for this
enigmatic observation.
Another method to determine deep crustal lithology is to test its composition with
seismic (Christensen and Mooney, 1995; Rudnick and Fountain, 1995) and gravity data.
Seismic velocities and densities are controlled by mineral forming, major elements (e.g.,
SiO2 and CaO), not the highly incompatibles. However, highly incompatibles correlate to
a sample?s degree of differentiation (i.e., SiO2 content); therefore, incompatible element
abundances can still be derived from major element concentrations. Table 3.4 compares
the continent?s expected velocities and densities, computed by Perple X via Gibbs free
energy minimization, for the middle (MC) and lower (LC) crust based on different geo-
physical and geochemical models. Our MC model has a middle crust of Vp = 6.8 km/s,
Vs = 4.0 km/s, and density of 2980 kg/m3, values higher than other models. A middle
crust calculated from Rudnick and Gao (2014)?s average composition also has a high den-
sity (Brocher, 2005) when compared to the Vp and surface wave predictions of CRUST
1.0 (Laske et al., 2013) and LITHO 1.0 (Pasyanos et al., 2014). These inconsistencies
between geochemical and geophysical predictions extend into the deep crust, again with
all but the lowest density and Vp geochemical estimates (Hacker et al., 2015) exceeding
84
Model Vp (km/s) Vs(km/s) Vp/Vs Poiss. Density (kg/m3)
# of Samples
Our MC* 6.84 4.04 1.69 0.233 2980
RG MC* 6.74 3.97 1.70 0.235 2940
Low Vp MC? 6.57 3.80 1.73 0.235 2720
CRUST 1.0 MC 6.47 3.70 1.75 0.257 2830
LITHO 1.0 MC 6.51 3.75 1.74 0.252 2840
Our LC** 7.05 4.13 1.71 0.239 3090
RG LC** 7.00 4.01 1.74 0.255 3050
Low Vp LC? 6.80 3.92 1.74 0.235 2920
CRUST 1.0 LC 7.04 4.01 1.76 0.261 3010
LITHO 1.0 LC 7.05 4.00 1.76 0.263 2990
MORB (LC)** 7.40 4.23 1.75 0.258 3260
OIB (LC)** 7.46 4.28 1.74 0.255 3330
? Hacker et al. (2015) Table 4 Middle Crust Vp 6.5-6.6; Lower Crust 6.7-6.9
*MC Conditions - 300 C, 0.4 GPa
**LC Conditions - 500 C, 0.85 GPa
?RG? = Rudnick and Gao (2014)
Table 3.4: Comparison of Deep Crustal Physical Properties
seismic expectations. Though these global averaged seismic models are not infallible,
a reconciliation is still required between the geochemical based and geophysical based
models for Earth?s deep crust.
3.11 Conclusion
The deep continental crust will remain a topic of intense debate for years to come
due to its inaccessibility. Amphibolite, granulite, and eclogite facies metamorphic terrains
and xenoliths serve as our windows to middle and lower continental crust. Compositional
variability even within these facies underscores the potential for deep crustal heterogene-
ity, though certain elements and patterns anchor our understanding of the chemistry of
the crust. The Daly Gap, evident in amphibolite facies lithologies and granulite facies ter-
85
rains, dominates the relative abundance of SiO2 in the deep crust, especially its lowermost
portions. The more homogeneous mineralogy of eclogite facies lithologies distinguishes
them from amphibolite or granulite facies lithologies, though all three could plausibly
contribute to previous deep crustal compositional estimates. Constraining the proportion
of mafic to felsic material in the deep crust will in turn constrain its trace element content
and distribution, because its enrichment or depletion in these elements is most heavily
influenced by differentiation processes.
Thankfully, all is not lost when using these samples to parse out the composition of
the deep crust. The amount of Al2O3 is similar among all of the samples, as are the heavy
rare earth elements. We find that the concentrations of Er, Tm, Yb, and Lu in particular
show little variation among samples of different metamorphic grades. The controlling
factor in incompatible element abundance among our deep crustal lithologies is how dif-
ferentiated the material is, not metamorphic grade. This means that igneous processes and
protolith composition rather than metamorphic processes dictate the chemical signatures
of the deep crust.
Although the deep continental crust has been studied at length, many elements still
lack sufficient concentration data (such as the highly siderophile elements). Future stud-
ies will be challenged to reduce the size of the uncertainty on element concentrations,
and since instrumental precision is not the main source of uncertainty, inquiry into the
processes that alter elemental abundances in different samples will have to be identified
and explained. As it stands, there is also a density discrepancy between highly cited
geochemical models (e.g. Rudnick and Gao (2014) versus Hacker et al. (2015)) and com-
mon geophysical crust models, with geochemical samples suggesting a deep crust that
86
has higher overall density than what is seismically observed. Ultimately, the future of
deep crustal modeling will depend on the integration of multiple types of datasets, such
as geochemical and seismological measurements and gravity analyses.
87
Chapter 4: Localized Lower Crustal Composition from Joint-Inversion
of Geochemical and Geophysical Datasets
4.1 Overview
In this chapter, we develop a method for quantitatively determining the composition
of the deep continental crust. We chose to test our method on an area that has been ex-
tensively studied, the southwestern United States, so that we could incorporate abundant
geochemical, seismological, and geophysical data. We found that we could determine the
major element composition of the crust in this area with ?10% uncertainty, and we used
our compositional model to identify the most likely crustal deformation processes that
were taking place.
[1] This chapter has been published as:
Sammon, L. G., Gao, C., & McDonough, W. F. (2020). Lower crustal composition
in the southwestern United States. Journal of Geophysical Research: Solid Earth,
125(3), e2019JB019011.
[2] LGS wrote this text and modeled the lower crustal compositions with help from
CG. CG provided ambient noise inversions. WFM contributed significantly to data anal-
ysis and text.
88
4.2 Abstract
The composition of the lower continental crust is well-studied, but poorly under-
stood because of the difficulty of sampling large portions of it. Petrological and geochem-
ical analyses of this deepest portion of the continental crust are limited to the study of high
grade metamorphic lithologies, such as granulite. In situ lower crustal studies require geo-
physical experiments to determine regional-scale phenomena. Since geophysical proper-
ties, such as shear wave velocity (Vs), are nonunique among different compositions and
temperatures, the most informative lower crustal models combine both geochemical and
geophysical knowledge. We explored a combined modeling technique by analyzing the
Basin and Range and Colorado Plateau of the United States, a region for which plentiful
geochemical and geophysical data are available. By comparing seismic velocity predic-
tions based on composition and thermodynamic principles to ambient noise inversions,
we identified three compositional trends in the southwestern United States that reflect
three different geologic settings. The Colorado Plateau (thick crust), Northern Basin and
Range (medium crust), and Southern Basin and Range (thin crust) have intermediate,
intermediate-mafic, and mafic deep crustal compositions. Identifying the composition of
the lower crust depends heavily on its temperature because of the effect it has on rock
mineralogy and physical properties. In this region, we see evidence for a lower crust
that overall is intermediate-mafic in composition (53.7 ? 7.2 wt.% SiO2), and notably
displays a gradient of decreasing SiO2 with depth.
89
4.3 Introduction
The composition of the lower continental crust, despite its influence over crust for-
mation and geologic hazards, remains a mystery. Though as thin as 10 km in some re-
gions (Rudnick and Gao, 2014), the lower crust contributes critically to the temperature,
structure, and stress state of the continent (e.g. Hacker et al. (2015a); Jackson (2002);
Shinevar et al. (2018)). Lower crustal deformation models are heavily informed by deep
crust silica content, water, and mineralogy (Jackson, 2002). However, because of the
relative scarcity (<1% of all samples listed on http://www.EarthChem.org/) and the
compositional heterogeneity of deep crustal samples, it is difficult to constrain the bulk
composition of the lower crust purely through geochemical or petrological measures.
Because the lower crust resides at depths >20 km, its composition can only be
sampled indirectly. Granulite facies lithologies serve as metamorphic analogues for the
lower crust due to their appearance in exposed crustal cross-sections (Rudnick and Gao,
2014). High grade metamorphic terrains, which have been tectonically emplaced in areas
such as the Ivrea-Verbano Zone in Italy or the Fraser Range in western Australia (Fountain
and Salisbury, 1981), and granulite facies xenoliths serve as two geochemical windows
to the lower crust. As a metamorphic facies, characterized by the dehydration of hydrous
minerals (Semprich and Simon, 2014), granulites span a confounding range of mafic (<
52 wt. % SiO2) to felsic (> 68 wt. % SiO2) compositions. Such wide variation leads
to competing models for the lower crust?s composition and density structure, as outlined
recently by Dumond et al. (2018).
Combined modeling of high resolution geophysical and geochemical data can place
90
N = 128 Granulite N = 100 Seismic 30
Samples Stations
25
Northern 
Basin and Range
20
Colorado 
Plateau
15
10
Southern
Basin and Range 5
Figure 4.1: The southwestern United States has been sampled at high resolution through
geochemical analyses and ambient noise seismology. The black triangles represent the
placement of 100 Earthscope Transportable Array stations whose data were used in this
study. Colored squares indicate the location of 128 granulite xenolith and terrain samples
used as possible chemical compositions for the lower crust. The color of the squares
indicates how many samples were collected from the area covered by the square. The
overlaid blue lines demark three geologically distinct sub-regions within the study area.
91
Number of Samples per Area
as
Nev
ad
Sier
ra 
tighter constraints on lower crustal composition. Seismic velocity measurements help dif-
ferentiate among possible lower crustal compositions when compared to laboratory exper-
iments (Holbrook et al., 1992). We use seismic inversions in conjunction with petrological
data in an effort to form a less biased lower crustal composition model. In this study, we
target the southwestern United States (Fig. 4.1) as a demonstration of such joint model-
ing efforts because of the variety of data available for the Basin and Range and Colorado
Plateau physiographic provinces.
Global scale models (Bassin et al., 2000; Laske et al., 2013) predict seismic veloci-
ties in the Basin and Range that are 10% slower and densities 5% lower than those of ad-
jacent tectonic regions. Slower seismic velocities could suggest that the Basin and Range
has a more felsic lower crust than surrounding areas and stands in contrast to local velocity
studies (Gao and Lekic?, 2018; Olugboji et al., 2017; Plank and Forsyth, 2016; Shen et al.,
2013). Schulte-Pelkum et al. (2017) provides an in depth reconciliation of various seis-
mic data and lower crustal xenoliths, and proposes a mafic lower crust in the southwestern
United States and throughout the country. Both mafic and felsic granulite facies terrains
and xenoliths have been extensively characterized in the southwestern US, providing us
with a geochemical dataset of 128 samples (http://www.EarthChem.org/). We incor-
porate high resolution, ambient noise, dispersion measurements (Ekstro?m, 2014; Olugboji
et al., 2017) from the Earthscope USArray (http://www.usarray.org/) project; Moho
temperature models from Pn velocities (Schutt et al., 2018); and thermal gradient calcula-
tions to derive a distribution of compositions and compositional trends for the lower crust,
addressing current model discrepancies.
92
4.4 Background
4.4.1 Compositional Modeling of the Lower Crust
The depth and thickness of the lower continental crust varies regionally and in the
context of different studies. The Conrad discontinuity defines the lower crust seismically,
but it is not ubiquitous. When the continental crust is split into thirds, the average lower
crustal composition is typically? 53 wt.% SiO2 (Rudnick and Gao, 2014). In some areas,
however, the ?lower crust? may refer to the bottom half of the continental crust, in which
case the average SiO2 becomes more felsic (Hacker et al., 2015b). The abundance of
SiO2 in the lower crust is not only a function of lower crustal composition, but also of
one?s definition of the lower crust. For the purposes of this study, we define the lower
crust as simply the bottom half of the crust between 11 km and the Moho after (Schmandt
et al., 2015). For example, if the Moho depth were 31 km, the lower crust would have
a thickness of (31?11)2 = 10 km, and range from 21 km - 31 km depth. We designate 11
km as the thickness of the upper crust because of changes seen in regional Rayleigh wave
models from Lin et al. (2014). Though 11 km of upper crust and sediment throughout the
entire southwestern United States is a sweeping generalization, it is similar to Roy et al.
(1968) 7-11 km thick heat producing layer and Rudnick and Gao (2014)?s 12 km thick
upper crust. Keep in mind that our compositional trends are more consequential than our
somewhat arbitrary layer thicknesses.
Petrological and geochemical studies of the deep continental crust have sought to
define composition through analysis of granulite facies xenoliths and terrains where avail-
93
able, usually analyzing in detail a small (5 - 20) set of samples. Similar practices have
been used by many (Al-Safarjalani et al., 2009; Halliday et al., 1993; Parsons et al., 1995;
Rudnick and Taylor, 1987; Schaaf et al., 1994) to determine the deep crustal structure in
regions where samples are available, but it is hard to gauge if these isolated samples are
representative of the whole lower crust. While studies of xenoliths provide insight into
specific areas of the lower crust, limited sample sets and even smaller sample sizes prove
to be recurring obstacles for geoscientists who seek to uncover the composition of the
deep crust as it relates to global processes. Seismological crust models, on the other hand,
are typically used to describe wide scale crustal phenomena. The use of seismic models
for determining lower crust composition requires a conversion between seismic wave ve-
locities and bulk rock compositions, typically achieved through laboratory experiments
(Christensen and Fountain, 1975; Christensen and Mooney, 1995; Holbrook et al., 1992).
Recent studies (Hacker et al., 2015b) give comprehensive assessments of shear and com-
pressional waves velocities of granulite facies lithologies through thermodynamic model-
ing (calculations are based on empirical, composition-pressure-temperature relationships
derived from rock mechanics and mineral physics experiments, and thermodynamic the-
ory).
4.4.2 Geologic Setting
The southwestern United States has undergone multiple episodes of compression
and extension since the Mesozoic (Coney and Harms, 1984). The elevated Colorado
Plateau remains relatively undeformed (Levander et al., 2011) despite being sandwiched
94
4.6 A 8 B
4.4 7.5
4.2
7
4
3.8 6.5
3.6 6
40 50 60 70 80 40 50 60 70 80
SiO2 (wt.%) SiO2 (wt.%)
1.9 C 3600 D
1.8 3400
3200
1.7
3000
1.6 2800
1.5 2600
40 50 60 70 80 40 50 60 70 80
SiO2 (wt.%) SiO2 (wt.%)
Figure 4.2: An example Perple X velocity and density calculation shows a correlation
with SiO2. The 128 samples from the southwestern United States span the range of mafic
to felsic, showing especially strong correlations between Vp and SiO2 and Vp/Vs and
SiO2. These calculations were done at one temperature (650?C) and one pressure (0.85
GPa). Increasing temperature systematically shifts the distributions to lower Vs, Vp, and
density, and, in general, higher Vp/Vs ratio.
95
Vp/Vs Vs (km/s)
Density (kg/m3)
Vp (km/s)
between North American Cordillera and the Basin and Range. The Basin and Range
province, on the other hand, is characterized by abruptly alternating basins and narrow
mountain chains that arose from tensional stress and normal faulting in the Early Miocene
(17 Ma) (Coney, 1980). The Basin and Range extended crust, in conjunction with the
Colorado Plateau, houses Cenozoic volcanics that are thought to be linked to changes
in plate interactions after the conclusion of the Laramide Orogeny (McKee, 1971). The
deep crustal xenoliths delivered through Cenozoic volcanic eruptions provide one of our
sources of geochemical data. A second data source are the Basin and Range?s metamor-
phic core complexes - a belt of medium- to high-grade metamorphic terrains exhumed
through crustal extension (Crittenden et al., 1980). A suite of crust deformation models
have been proposed to produce these core complexes (Cooper et al., 2010), each model a
different combination of brittle faulting and ductile extension.
4.5 Methods
We used a three-step joint geochemical-geophysical modeling process to constrain
composition. Figure 4.3 provides a schematic walk-through of the inputs and outputs of
each step.
First, we calculated physical properties over a range of pressures and tempera-
tures for local granulite facies samples through the thermodynamic Gibbs free energy
minimization software Perple X (Connolly, 2005). Second, we determined pressure-
temperature conditions at 1 km intervals within the lower crust of the southwestern United
states, making the assumption that pressure uniformly increases 1 GPa per 35 km depth,
96
Sample 1
SiO2 = 51 wt.%
Overlap
3.4 3.5 3.6 3.7 3.8 3.9 4 4.1 4.2 4.3
Vs (km/s)
Probability that Sample 1 
produced seismic signal : Seismic Vs
Geochemical Vs
Sample 2
SiO2 = 48 wt.%
Overlap
3.4 3.5 3.6 3.7 3.8 3.9 4 4.1 4.2 4.3
Vs (km/s)
Distribution of 
composition, depth,
temperature combinations Probability that Sample 2 
that satisfy inputs produced seismic signal :
Seismic Vs: Vs from seismic inversion at given depth
Geochemical Vs: Vs from Perple_X calculation at given depth and temperature
Figure 4.3: Crust modeling flowchart showing our procedure for finding consistent mod-
els based on Vs, temperature, depth, and composition. Seismic velocity map from Olug-
boji et al. (2017); Moho temperatures based on Schutt et al. (2018); Moho depths from
Shen and Ritzwoller (2016). Colors in both panels match. The tan probability distribution
reflects the spread of Vs solutions from seismic inversions at a given location and depth.
The teal Vs is generated from the Perple X calculations. The overlap between the two Vs
distributions is shown in dark green, reflecting the mutually consistent Vs values.
or roughly 28.6 MPa (286 bars) per kilometer. Temperature inputs at the top and base
of the crust allowed us to calculate a geothermal gradient and therefore a temperature for
each kilometer within the crust. We assumed that the top of the crust resides at 5?5?C
and temperature at the Moho follows Schutt et al. (2018).
Third, we compared the Perple X-calculated shear wave velocities (Vs) of each
sample to seismic inversions for Vs. The seismic data were provided by Olugboji et al.
(2017), and one-dimensional Vs profiles were calculated through Bayesian inversion of
Rayleigh and Love wave data following Gao and Lekic? (2018). We calculated the prob-
ability of each sample producing the observed seismic signal by convolving the two
datasets. The overlap between the seismic Vs distribution and the geochemical Vs distri-
97
Relative Probability Relative Probability
bution at a given pressure and temperature was totaled for each geochemical sample (Fig.
4.2).
In general, we favored what we considered to be the simplest parameter space that
could explain the geochemical observations in our dataset, such as equilibrium conditions
and common mineral assemblages. We used the hpha11.ver thermodynamic database,
which contains the self-consistent (Holland and Powell, 2004) database but is augmented
by Hacker et al. (2015a) to account for the a ?- ? -quartz transition. A full explanation of
the Perple X parameters we used and our rationale is given in Appendix C. Olugboji et al.
(2017) and Ekstro?m (2014) explain the inversion techniques that produced our seismic
profiles.
We evaluated the uncertainties associated with each step of our combined model,
allowing for variations in lower crustal thickness, temperature, and seismic velocity (Ta-
ble 4.1). Moho depths were assigned a 2 km uncertainty (Shen and Ritzwoller, 2016),
and Moho temperature uncertainties range from 50 to 80C depending on location (Schutt
et al., 2018). Combined variations in Moho temperature and depth, and a linear extrapo-
lation of temperature through the crust gave us variable temperature gradients throughout
the area of study, which we calculated via Monte Carlo simulation. The result was a dis-
tribution of possible lower crustal pressure-temperature conditions, which translated to a
probability distribution of compositions. Convolving the distribution of Perple X gener-
ated velocities with the seismic shear wave velocities produced our final distribution.
Systematic uncertainties may exist if our fundamental assumption of a dry, granulite
facies lower crust is inaccurate. The accuracy of Perple X?s velocity calculations depends
largely on this assumption, as a lack of water restricts our compositions to anhydrous
98
Parameter Uncertainty
Seismic velocity inversions full distribution compared to geochem-
ical results, uncertainties on seismic
inversion methods given from Gao and
Lekic? (2018)
Perple X calculations subject to unknown systematic uncer-
tainty, but < 1% uncertainty on Gibbs
free energy calculations
Moho temperature 5%-10% (Schutt et al., 2018)
Lower crustal thickness 13% - 25%, assuming absolute uncer-
tainty of 2 km (Shen and Ritzwoller,
2016)
Table 4.1: Uncertainties Associated with Methods
minerals. Connolly (2005) offers an overview of the software?s free energy minimization
technique for calculating mineral assemblages.
4.6 Results
Overall, the hot lower crust of the southwestern United States trends towards inter-
mediate and mafic compositions. When investigating sub-regional scale variations, how-
ever, three separate trends of composition emerge. Joint modeling of surface wave veloci-
ties and geochemical and petrological data yields a variety of compositions that depend on
temperature. An iterative approach allows us to construct a distribution of probable com-
positions at each of 100 seismic stations, to account for uncertainties in temperature and
composition. Any granulite compositions that were duplicates (i.e. samples whose Vs?s
or compositions were indistinguishable from another sample?s) were removed to avoid
99
artificially weighting our results towards redundantly-sampled lithologies. Excluding all
samples with similar silica content, though, would render the geochemical inputs to this
model useless, since we would be ignoring geochemical trends in our samples.
Similar velocities and compositions are evident among three sub-provinces of the
study area: the Colorado Plateau to the east, the beginnings of the Northern Basin and
Range in the northwest, and the Southern Basin and Range in the southwest. As a whole,
the shear wave velocities of all three regions range from 3.8 km/s to 4.2 km/s, with about
half of the lower crust being faster than 4.0 km/s (Fig. 4.4). Vp, calculated from Perple X,
often exceeds 7.0 km/s in the Southern Basin and Range and in deeper portions of the
Northern Basin and Range and Colorado Plateau. The Southern Basin and Range, which
has experienced the most recent tectonic activity, is marked by the thinnest, hottest crust,
while the Colorado Plateau has the thickest, coolest crust. Despite comparatively slow Vs
in the Southern Basin and Range (3.9 ? 0.1 km/s), its high temperatures (often > 800?C)
require a Vp of 7.1 ? 0.1 km/s and a density of 3000 ? 190 kg/m3 at the base of the crust
to satisfy the geophysical model (Fig. 4.4). The Vp/Vs ratio remains poorly constrained,
with uncertainties upwards of 10% encompassing most lithologies (Brocher, 2005). Fig-
ure 4.4 illustrates a change in the median Vp/Vs from ? 1.79 to ? 1.72 separating the
Colorado Plateau from the Basin and Range, a shift that reflects compositional variation.
Not surprisingly, compositional trends follow velocity trends, forming three distinct
compositional provinces. Figure 4.5 shows representative distributions of SiO2 content
that result from our model. The Colorado Plateau, which has the coolest crust and the
lowest Vp/Vs ratio, also has the widest distribution of possible compositions (Fig. 4.5A),
which range from 45 to roughly 75 wt.% SiO2. The Basin and Range favors narrower,
100
more mafic distributions (Fig. 4.5B and C). Regardless of location, though, mafic litholo-
gies can explain the lower crust?s seismic profile more frequently with increasing depth,
as shown by the increasing blueness with depth of Figure 4.5. Figure 4.6 (and Figure S7)
maps reveal clear compositional distinctions among the three sub-provinces. The differ-
ences in SiO2 between the intermediate Colorado Plateau, intermediate-mafic Northern
Basin and Range, and mafic Southern Basin and Range are most apparent in the shallow
lower crust. Both the Colorado Plateau and the Northern Basin and Range increase in
MgO and FeO content and decrease in SiO2 content at greater depths, but the Colorado
Plateau does not reach truly mafic compositions until 35 - 40 km depth. Caution is rec-
ommended when interpreting fine-scale variations in Vs, and therefore in composition.
Since the seismic dispersion curves are sensitive to a range of depths (Ekstro?m, 2014),
values for the deepest regions of the crust might be elevated by averaging high mantle Vs
into the crustal Vs estimates. However, the difference between the seismic velocity model
and the combined velocity model (Fig. S11 and Fig. S12) suggests that the weight the
geochemical inputs have on the combined model may override any averaging in of mantle
velocities. The abundance of distinguishably different mafic granulite lithologies causes
the combined model to favor higher velocities.
Six mineral groups dominate the modeled lower crustal mineralogy. Clinopyrox-
ene and garnet grow at the expense of quartz and plagioclase and K-feldspars in deeper
portions of the crust. Orthopyroxene abundances also decrease by a few weight % with
depth. The high mode proportion (?3.5 wt.% to 16 wt.%) of K-feldspars (which pri-
marily manifests sanidine under simulated pressure and temperature conditions) reflects
the alkali-rich, latite-like compositions of crystalline rocks from the southwestern United
101
A Vs (km/s) B Vp (km/s)
4.15 7.2
38 Nevada Utah 4.1 38 Nevada Utah 7.1
4.05
36 36 7
California Arizona 4 California Arizona
6.9
34 3.95 34
3.9 6.8
32 32
3.85 6.7
30 3.8 30 6.6
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
C Vp/Vs D Density (kg/m3)
1.85
38 Nevada Utah 38 Nevada Utah 3200
1.8
3100
36 36
California Arizona California Arizona
1.75 3000
34 34
2900
32 1.7 32
2800
30 1.65 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure 4.4: The Vs in (A) is a convolution of seismic and geochemical data. Panels (B)
- (D) predict Vp, Vp/Vs, and density from the geochemical samples that were consistent
with seismic Vs data. The Colorado Plateau is differentiated from the Basin and Range
in Vp and Vp/Vs. Hotter temperatures in the south lead to slower Vs but faster Vp and
higher densities in the Southern Basin and Range. Blue regions in B, C, and D correspond
to more mafic compositions.
A Station 54 - Eastern Utah B Station 48 - SE Nevada C Station 78 - SW Arizona
37.64?N, -110.80?E 37.64?N, -115.75?E 35.30?N, -114.87?E
1
25 21 20
26
22
27 21
28 23 0.8
22
29 24
30
25 23
31 0.6
32 26 24
33 27 25
34
28 0.4
35 26
36 29
27
37 30 0.2
38
31 28
39
40 32 29
0
35 40 45 50 55 60 65 70 75 80 85 90 35 40 45 50 55 60 65 70 75 80 85 90 35 40 45 50 55 60 65 70 75 80 85 90
SiO2 (wt.%) SiO2 (wt.%) SiO2 (wt.%)
Figure 4.5: Representative histograms (A-C) of the three ?sub-provinces? within the
study area show increasing probability of a mafic crust with increasing depth. Color
indicates the relative probability of a given SiO2 abundance explaining the seismic signal
at a given depth. The thicker, cooler Colorado Plateau (A) can, on average, accommodate
a higher percentage of SiO2 than the Northern (B) or Southern (C) Basin and Range.
102
Depth (km)
Depth (km)
Depth (km)
Relative Probability
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
32 32 65
62
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110 59
C 30 km Depth D 35 km Depth 56
38 Nevada Utah 38 Nevada Utah 53
36 36 50
California Arizona California Arizona
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure 4.6: Variability in median SiO2 abundance in the southwestern United States
tracks the Colorado Plateau (high SiO2), Northern Basin and Range (medium SiO2), and
Southern Basin and Range (low SiO2). SiO2 abundance overall decreases with increasing
depth (A - D). Color scale indicates wt.% SiO2. Mantle compositions are not shown in
this figure, and therefore deeper profiles (e.g. C - D) show only regions with greater
crustal thickness.
103
SiO  (wt.%)2
States (Tyner and Smith, 1986). At shallower pressures and colder temperatures, miner-
als such as kyanite, sillimanite, or ilmenite can comprise anywhere from 5 - 15 wt.% of
the ?lower crust?. Mineral assemblages simplify at greater depths, with clinopyroxene,
garnet, plagioclase, and quartz often controlling >80% of the mineralogy.
Though it is convenient to report one number and an uncertainty as representative
for composition, we are mindful that the shapes of these major oxide and mineral dis-
tributions are non-normal and cannot be fully described by simple summary statistics.
That being said, whether reporting mean or median value as representative of the lower
crust, the depth trend of change in composition holds true for the Colorado Plateau and
Northern Basin and Range (see Tables 4.2 - 4.3). The Southern Basin and Range mean
composition shows this gradient to a lesser extent, while the median is homogeneously
mafic. For the sake of convenience, our interpretations will reference the median ? 12 the
inter-quartile range (IQR) compositions unless stated otherwise. We favor the median and
IQR because they are more resistant to outliers than the mean.
4.7 Discussion
4.7.1 Lower Crust Composition
One value, one composition, is insufficient for describing the entirety of the conti-
nental lower crust. We can describe the lower crust more accurately by reporting changes
in velocity, density, and composition as a function of depth and location. The lower con-
tinental crust, though less than 8 km thick in some sections of the southwestern United
States (Buehler and Shearer, 2017), undoubtedly displays lateral and vertical heterogene-
104
Isochemical Projection vs. Observed
26 5.0
Seismic Mean Vs 4.5
28 Observed 
Seismic 1? seismic Vs 4.0
Isochemical 
30 projection from 3.5
26 km
3.0
32
2.5
34 Geochemical-based 2.0
Vs calculation if  
36 no compositional 1.5
changes 1.0
38
0.5
40 0
2 2.5 3 3.5 4 4.5 5 5.5
Vs (km/s)
Figure 4.7: An example of granulite lithologies fitting the seismic signal at the top
of the lower crust projected to higher temperature and pressure conditions (red field).
This isochemical predicted Vs projection deviates substantially from the mean seismic
Vs to the extent that by 38 km depth the distributions are distinguishable at 1? . (For
comparison, seismic Vs is typically reported as mean ? 1 standard error of the mean.
Using this metric, the distributions become distinguishable at ?30 km.)
105
Depth (km)
% of Inversions
Table 4.2: Colorado Plateau SiO2 Content
Depth Mean Median Standard 1st Quartile 3rd Quar-
Devia- tile
tion
26 60.0 58.8 11.6 50.0 69.7
27 59.8 58.6 11.8 49.5 69.9
28 59.6 58.3 11.7 49.5 69.5
29 59.3 57.8 11.7 49.3 69.2
30 59.0 57.4 11.7 49.1 68.7
31 58.8 56.8 11.8 48.8 68.6
32 58.6 56.2 11.8 48.6 68.1
33 58.3 55.8 11.8 48.5 67.3
34 58.1 55.2 11.7 48.4 66.8
35 57.9 54.7 11.7 48.3 66.4
36 57.7 54.3 11.7 48.3 65.9
37 57.5 53.7 11.6 48.2 65.3
38 57.3 53.4 11.6 48.1 64.8
39 57.1 53.1 11.5 48.1 64.4
40 56.9 52.8 11.5 48.0 64.0
Oxide abundances reported in wt.%. 
Table 4.3: Northern Basin and Range SiO2 Content
Depth Mean Median Standard 1st Quartile 3rd Quar-
Devia- tile
tion
22 58.3 55.1 10.8 49.9 65.7
23 58.3 55.1 10.9 50.0 65.4
24 58.1 54.7 10.9 49.7 65.0
25 58.0 54.4 11.0 49.7 64.9
26 57.7 54.0 10.9 49.5 64.3
27 57.7 53.9 11.1 49.2 64.2
28 57.4 53.6 11.0 49.1 63.8
29 57.4 53.6 11.1 48.9 64.0
30 57.2 53.4 11.2 48.8 63.6
31 57.0 52.9 11.2 48.6 63.4
32 57.0 52.8 11.4 48.3 63.7
Oxide abundances reported in wt.%. 
106
Table 4.4: Southern Basin and Range SiO2 Content
Depth Mean Median Standard 1st Quartile 3rd Quar-
Devia- tile
tion
19 54.6 51.2 10.0 48.0 58.1
20 54.3 51.1 9.9 48.0 57.6
21 54.1 51.1 9.8 47.9 57.0
22 54.0 51.2 9.7 47.9 56.8
23 54.0 51.2 9.7 48.0 56.6
24 53.9 51.1 9.6 47.9 56.3
25 53.9 51.2 9.6 47.8 56.4
26 53.8 51.3 9.6 47.9 56.2
27 54.0 51.4 9.7 47.9 56.7
Oxide abundances reported in wt.%. 
Table 4.5: Southwestern United States Lower Crust Major Oxide Content
Mean Median Standard 1st Quartile 3rd Quar-
Devia- tile
tion
SiO2 56.9 53.7 10.6 49.1 63.5
Al2O3 16.7 16.1 4.3 14.1 19.5
MgO 5.4 3.8 4.2 2.5 7.3
FeO 8.6 7.7 4.1 5.5 11.4
CaO 7.0 5.6 4.9 2.2 10.6
K2O 1.6 1.3 1.5 0.4 2.2
Na2O 2.7 2.7 1.4 1.6 3.8
TiO2 1.1 0.9 0.8 0.5 1.4
Oxide abundances reported in wt.%. 
107
Table 4.6: Summary of Lower Crust Seismic Properties and Mineralogy
Mean Median Standard 1st Quartile 3rd Quar-
Devia- tile
tion
Vs* 3.62 3.61 0.36 3.30 3.64
Vs** 4.02 3.99 0.23 3.86 4.15
Vp 6.94 6.91 0.44 6.59 7.27
Vp/Vs 1.73 1.75 0.07 1.68 1.78
Density 3010 3000 220 2820 3190
Clinopyroxene 17.5 13.3 16.6 3.4 31.5
Garnet 13.0 11.8 8.7 6.5 18.1
K-feldspars 10.9 8.2 6.7 3.4 16.0
Kyanite 3.5 2.5 4.1 0.5 5.3
Olivine 1.3 0.03 5.8 0 0.3
Orthopyroxene 3.0 1.5 4.3 0.2 5.0
Plagioclase 30.3 27.6 14.0 15.5 43.5
Quartz 12.7 10.9 16.1 0.7 27.0
*Vs from surface wave inversions only
**Vs from combined surface wave and geochemical 
model
108
ity (Fig. 4.6 and S7). Temperature plays a crucial role in determining lower crustal
composition. Both cold intermediate and hot mafic granulites can produce the shear wave
velocities of >3.9 km/s observed across the southwestern United States (Christensen and
Mooney, 1995). The thicker, cooler crust of the Colorado Plateau (average Moho tem-
perature 700?C, constant gradient of 17.5?C/km) and Northern Basin and Range (average
Moho temperature 740?C, constant gradient of 22.4?C/km) can therefore accommodate
55.8 ?9.4 and 53.9 ?7.5% SiO2, respectively. The Southern Basin and Range, in con-
trast, must have a predominantly mafic composition of 51.2? 4.3% SiO2 to reach similar
Vs because of its thin crust and 800?C temperatures.
Our model does not account for contributions to the geotherm from heat produc-
tion. The first order approximation of a linear, conductive geotherm could cause us to be
underestimating temperatures in regions of high heat production, particularly at shallower
depths in the crust. Correcting for heat production, however, is nontrivial. If we are under-
estimating temperature by excluding heat production, then it?s possible that the deep crust
is more mafic than predicted here. A more mafic crust, however, should have low heat
production, causing a negative feedback loop: that is, by accounting for heat production,
we are lowering the contribution from heat production. Calculating Moho temperature
from surface heat flow measurements (Davies, 2013), literature heat production (Rudnick
and Gao, 2014), and thermal conductivity (Jaupart et al., 2016) values yields results that
vary by hundreds of degrees and are inconsistent at best. Future calculations regarding
heat production and the geotherm will depend also on crustal heat producing element
abundances, which are currently debated (Hacker et al., 2015b).
The temperature gradient in the lower crust necessitates a vertical gradient in miner-
109
alogy and composition. The crust becomes increasingly mafic with increasing depth. This
trend is observed most prominently in areas of thicker crust. The increase in Vs cannot be
explained by isochemical chemical changes in the lower crust - that is, we cannot explain
the observed Vs by simply projecting mid-crustal compositions to higher pressures and
temperatures (Figure 4.7). As noted by Christensen and Mooney (1995), we must invoke
a compositional gradient within the lower crust to explain the increase in seismic velocity.
In the topmost portions of the Colorado Plateau?s lower crust, our model can ac-
commodate over 59 wt.% SiO2 (Table 4.2). However, such intermediate-felsic material
cannot reach high enough velocities to match the seismic signal deeper in the crust, where
temperatures increase above 700?C (Schutt et al., 2018). Furthermore, our set of gran-
ulites can explain the seismic signal at the base of the crust more often than at the top,
whereas we might expect equal probabilities at all depths if the lower crust were com-
positionally uniform (shown by the colors of Fig. 4.5). The Northern Basin and Range
and Colorado Plateau (Fig. 4.6) show 3 wt.% to 6 wt.% decrease in SiO2 and an increase
in MgO, FeO, and CaO with increasing depth. The Southern Basin and Range, though,
seems to lack this trend, the lower crust remaining consistently at 51 wt.% SiO2. This is
possibly due to removal of more felsic material from the top of the crustal column, which
we discuss in section 4.7.1.1.
The specific mineralogy of the lower crust is trickier to constrain than the bulk com-
position because of its strong dependence on our initial assumptions. Provided that our
lower crust is dry and equilibrated in the granulite metamorphic facies, we expect to see
mineral assemblages that are rich in clinopyroxenes, garnets, and plagioclase (Rudnick
and Fountain, 1995). Few studies that characterize the whole rock compositions of gran-
110
ulite quantitatively report mineralogy. This makes comparison between our results and
petrological studies of our samples difficult. Though Perple X builds bulk rock veloci-
ties from mineral constituents, many mafic rock-forming minerals have similar Vs under
lower crustal pressure and temperature conditions (e.g. at 650?C and 0.85 GPa diopside:
4.60 km/s; almandine: 4.57 km/s; spessartine: 4.65 km/s; anorthite: 3.65 km/s; sanidine:
3.49 km/s). A sample may therefore change mineralogy without drastically changing its
bulk rock properties or composition. In addition, our model?s mineralogy predictions are
more sensitive to temperature than its seismic velocity predictions are, due to the abrupt
and complete phase changes implemented by Perple X. We do not have the seismic reso-
lution to see such sharp changes in reality (Olugboji et al., 2017), if they exist at all.
Though we assume that the lower crust is predominantly granulite facies, projec-
tions of the geotherm through the Colorado Plateau leave the deep crust at temperatures
cooler than what is required for granulite facies metamorphism. Disequilibrium likely
exists here, but the assumption that the lower crust has, at some point, reached gran-
ulite facies metamorphism is based on observations of granulite grade metamorphic core
complexes and xenoliths (Coney and Harms, 1984). Amphibolite facies lithologies could
comprise shallower portions of the crust, or even portions of the lower crust where the
crust is thin or cool. Ample amphibolite facies xenolith samples exist for the western
United States. In the future, similar combined modeling assessments should consider the
role amphibolite and other medium- to high-grade metamorphic lithologies play in the
lower crust. However, if the lower crust reached granulite facies at some point in its his-
tory, retrograde metamorphism is unlikely to occur due to the thermodynamic barrier of
rehydration (Semprich and Simon, 2014), and the base of the lower crust must be mafic
111
in our model no matter which mafic minerals specifically are present. Broadly speaking,
the abundance of garnet and clinopyroxene increases with depth, driving the increase in
Vs. Mineral assemblages simplify with increasing depth and temperature, leaving little
room for accessory phases, such as ilmenite and kyanite, at the base of the crust.
Further seismic constraints could reduce the uncertainty on our compositions. The
Vp/Vs ratio can often distinguish mafic from felsic compositions (Holbrook et al., 1992).
A Vp/Vs of >1.65 would reduce the probability of lower crustal compositions >65 wt.%
SiO2 (Holbrook et al., 1992) (or, conversely, Vp/Vs of <1.65 would indicate that geo-
chemical studies over-sample mafic compositions). Given that most crystalline rocks
exhibit Vp/Vs between 1.6 - 1.9 at standard experimental conditions (Brocher, 2005),
the ratio would have to be tightly constrained at <? 7% variation. Future quantitatively
robust modeling efforts of the southwestern United States should also investigate the pres-
ence of hydrous minerals (Dixon et al., 2004; Valentine and Perry, 2007) and melt (Rey
et al., 2009) in the lower crust. The presence of fluids could lower the deep crust?s Vs,
requiring compositions that are even more mafic than those reported here. Alternatively,
melt could cause the temperatures implemented in this study to be over-predicted (Schutt
et al., 2018).
4.7.1.1 Implications for Crust Formation
Ductile spreading and uplift of the lower crust could explain the correlation be-
tween crustal thickness and composition. Brittle thinning of the upper and/or middle
crust through normal faulting allows for isostatic uplift of ductile, deeper crust with little
112
Detachment fault
Upper Crust
Lower Crust
Ductile deformation
Figure 4.8: The west-east schematic cross-section of the southwestern United States de-
formation that our model supports. Crustal extension is accommodated primarily through
brittle faulting in the upper crust, which is separated from the lower crust by a rolling
hinge detachment Cooper et al. (2010).The ductilely deformed lower crust is uplifted but
not substantially thinned.
change in lower crustal thickness (Cooper et al., 2010), illustrated by Figure 4.8. Because
the thinnest regions of the southwestern United States are also the most mafic, what was
once ?lower crust? now likely comprises a greater volume of the 25-28 km thick crustal
column. Crustal thickness was lost as intermediate and felsic material was removed from
the top, rather than through lower crustal delamination (Rudnick and Gao, 2014). The
composition of the deepest layers of Northern Basin and Range are similar to the deep
Colorado Plateau 53 wt.% SiO2, further suggesting that the extended crust has not lost a
mafic root relative to the thicker crust. Had crustal thinning been caused by delamintaion,
the thinnest Southern Basin and Range crust would be the most felsic rather than the most
mafic. Based on the lower crust?s mafic composition, the eclogitization process required
for delamination would likely only be triggered at high pressures (Semprich and Simon,
2014). While the Basin and Range is certainly hot enough to undergo delamination (Jull
and Kelemen, 2001), it would have had to occur during a time of crustal thickening (for
instance, during the Laramide Orogeny (Bird, 1984; Livaccari, 1991), not during exten-
sion.
113
4.8 Conclusion
Joint modeling of geophysical and geochemical properties of the lower crust can
help constrain lower crustal composition. As an example, in the southwestern United
States, seismic velocities, when paired with Moho temperatures and thermodynamic cal-
culations, indicates that the lower crust transitions from intermediate to mafic composi-
tion, SiO2 content decreasing by up to 6% with increasing depth. Temperature gradients
cause compositional distinctions to arise among the relatively cool Colorado Plateau, the
warm Northern Basin and Range, and the hot Southern Basin and Range. The predomi-
nantly mafic composition of the lower crust reflects the tectonic history of this region and
can help distinguish between different crustal deformation mechanisms.
Though global-scale models give a generalized view of the lower crust, nonunique
solutions to composition can be better constrained in regional-scale studies by combin-
ing high resolution local datasets and compositional proxies. Combining seismological,
petrological, and thermodynamic data opens avenues for future detailed investigation into
deep crust composition and structure.
114
Chapter 5: The composition of the deep continental crust inferred from
geochemical and geophysical data
5.1 Overview
Combining many different geophysical and geochemical techniques together helps
us understand the composition of the bottom two-thirds of the continental crust. Build-
ing on the model developed in the previous chapter, we expand our joint geophysical-
geochemical modeling techniques to a global scale. We account for various factors, such
as crust temperature and tectonic setting, to create a large-scale model for the compo-
sition of the deep crust. Then, using joint probability analyses between major and trace
elements, we predict heat production in the continental crust and heat flux across the
Moho.
[1] This chapter has been accepted in the Journal of Geophysical Research: Solid
Earth as: Sammon, L. G., McDonough, W. F., & Mooney, W. D. (2022). The composition
of the deep continental crust inferred from geochemical and geophysical data.
[2] LGS, WFM and WDM contributed to the conceptualization and methodological
development of this project. LGS did software development, modeling, visualization, and
115
writing. WFM and WDM contributed input and discussion throughout, as well as with
the revising and editing. WDM provided the seismic surveys used to build this model.
5.2 Abstract
The combination of geochemical and seismological results places tighter constraints
on the composition of the middle and lower continental crust than either field can achieve
alone. The inaccessible depth of the deep crust (typically >15 km) forces reliance on
analogue samples and modeling results to interpret its bulk composition, evolution, and
physical properties. A common practice relates major oxide compositions of small- to
medium-scale samples (e.g. medium to high metamorphic grade terrains and xenoliths)
to large scale measurements of seismic velocities (Vp, Vs, Vp/Vs) to determine the com-
position of the deep crust. We provide a framework for building crustal models with
multidisciplinary constraints on composition. We present a global deep crustal model
that documents compositional changes with depth and accounts for uncertainties in Moho
depth, temperature, and physical and chemical properties. Our 3-D deep crust global
compositional model uses the USGS global seismic database (Mooney, 2015) and a com-
pilation of geochemical analyses on amphibolite and granulite facies lithologies (Sammon
and McDonough, 2021). We find that SiO2 abundance decreases from 61.2 ? 7.3 to 53.8
? 3.0 wt.% SiO2 from the middle to the base of the crust. In addition, we calculate
trace element abundances as a function of depth from their relationships to major oxides.
From here, other lithospheric properties, such as Moho heat flux, are derived (18.8 ? 8.8
mW/m2). This study provides a global assessment of major element composition in the
116
deep continental crust.
5.3 Introduction
The deepest parts of Earth?s crust are widely inaccessible to traditional geochemical
sampling and so their composition is poorly understood. Only in areas where eruptions
have brought xenoliths to the surface or where tectonic activity has exhumed medium and
high grade metamorphic terrains are we able to partially determine the composition of
the deep (middle and lower) continental crust. Even so, these ex situ, aged, weathered,
and transported rocks may not adequately represent the overall, current composition of
the deep crust. Such inaccessibility has challenged geochemists for decades, leading to
competing models for continental crust and bulk silicate Earth (BSE) compositions, for-
mation, and evolution. Dissonance in the geochemical community stems from known and
unknown unknowns; that is, we are mostly certain of the uncertainties in our geochemical
and petrological measurements, but we are uncertain if our samples are truly representa-
tive of large swathes of the deep crust or if they are merely point samples. Xenoliths and
terrains are the sum of the processes that form them, which may cause them to differ from
what is presently at 15-45 km and deeper. The deep crust is an enigma, and compositions
of xenoliths and high grade metamorphic terrains provide only an incomplete cipher.
Seismological techniques, however, provide another piece of the cipher by directly
measuring the physical state of large sections of the deep crust. Physical properties (e.g.
density, Poisson?s ratio, Vp, and Vs) determined from these in situ geophysical experi-
ments can be compared to laboratory experiments on rocks of known compositions, par-
117
ticularly medium to high grade metamorphic lithologies (amphibolite and granulite facies
rocks) to place constraints on estimates of deep crustal composition. Integrating geo-
chemical and geophysical observations, related to each other by empirically (laboratory)
derived thermodynamic properties, provides a reinforced, clearer, consistent picture of
middle and lower crustal composition.
This study uses geophysical and geochemical datasets to build a global composi-
tional model of the lower two-thirds of the continental crust. We generate a composition
versus depth model for the middle and lower continental crust by applying thermody-
namic modeling software to medium and high grade lithologies. We then compare the
thermodynamically-generated seismic velocities to velocities obtained from seismologi-
cal measurements to produce a jointly constrained geochemical-seismological composi-
tional model.
5.4 Methods
Our model calculations are conducted in two steps: 1) assembling the data and per-
forming thermodynamic calculations, and 2) adjusting model parameters to generate deep
crustal compositional models with uncertainties. These calculations require seismic ve-
locity depth profiles, Moho depths, and crustal temperature gradients for areas of interest.
Using the thermodynamic modeling software Perple X (Connolly, 2005), we calculate
the probability that different deep crustal compositions could produce the observed seis-
mic signal. These calculations are conducted using our modeling software, CrustMaker,
which is provided as an electronic supplement. The calculation adopts a subdivision of the
118
Tectonic Regimes
)
lay
a
60 
km
ma t (>
in 
Hi us
h ick 
Cr s
T h An
de s
T Tet
hy
Rift
A
oic
roz
an
e
Ph
rozoic
Prote
Oceanic C Ao rchn eantinental Margin
a  km
)
laya t (>
60
n H
im rus
hi ick 
C s
T Th An
de
eth
ys
T
B
Arc Rift
Archea dn Extende
Platform
Phanerozoic
Proterozoic
Figure 5.1: The weighted area proportion of crustal types, or ?tectonic regimes?, used for
our model as A) a fraction of total crust and B) a fraction of continental crust. Proterozoic
crust is most abundant (32% of the continental crust), followed by continental margins
(16%) and Archean crust (12%). Modern and paleo-orogens, including arcs, make up a
combined 19% of the continental crust in our model.
119
Pa Pal lee o-oo rog- eo nrogen Ex Art cPla entf do er dm
rgin
Ma
tal 
ntin
en
Co
global continental crust into 13 tectonic provinces (Figures 5.1 and 5.2) to speed calcula-
tions and extrapolate results to areas with lower data coverage. Individual tile resolution
of this global model is set to 1?latitude x 1?longitude x 3 km depth as a default, but can
be changed in the model to suit user needs. We chose this default resolution for our
global model based on the resolution of our crustal categories (each 1?x 1?of crust was
assigned a tectonic province based on models such as CRUST1.0, Litho1.0, and modifi-
cations discussed further in Section 5.4.1), and the resolution of our crustal thickness and
temperature data, the ramifications of which are discussed further in the Results section.
For considering higher resolution, regional scale data, the same methods can be used. In-
stead of simplifying the crust into tectonic provinces, calculations are run for individual
seismic velocity profiles, so that if there are, for example, 34 seismic velocity profiles as
inputs, there will be 34 locations for which compositional profiles are generated.
We calculated the overlapping probability between measured seismic velocities and
the Perple X-derived velocities for amphibolites and granulites equilibrated at middle
and lower crustal pressures and temperatures. A reference average crustal density of
2900 ? 200 kg/m3 is used to calculate lithostatic pressure (Wipperfurth et al. (2020), cf.,
Christensen and Mooney (1995)). Integrating the area under both curves, the area shown
as magenta in Figure 5.3, for a sample of composition X yields the total probability of
sample X producing the observed seismic signal. Repeating this technique for a multitude
of sample compositions at various depths and temperatures yields a final Monte Carlo
model for a deep crustal composition. Probability distributions are generated for Vp, Vs,
and Vp/Vs and then multiplied together to further constrain the final probability.
120
A
nic
in
arg an i
c ic ms en ust yan
cea t. M rch
e tero
zo rozo atfo
r
nde
d g
ted oro rc yan ean
r la
O n A o ane . Pl e if - A t
h st Cxt R leo Te An
d cke  Hi
ma
Co Pr Ph Sed E Pa Thi Thin
B
hy m d er dp k
e y
ra efo
r l se iv n se a l uo
g v od
e ve
r ion cet e tio
r n u e
a v
e tio bhq od o
mo W M nr
e
fra
c R nc Re rac artu f M o
n
S
. To
U Re F Re E
A.N
Figure 5.2: A) The mapped distribution of our 13 crust types and B) the seismic velocity
profile data distribution from the USGS database. Data coverage is greatest in the northern
hemisphere while places with less coverage, like Africa and Antarctica, rely more heavily
on extrapolation of crust type.
121
Seismic Velocity
Sample/Perple_X Velocity
Probable Velocity overlap
Relative 
Probability 
What are these samples? 
Distribution compositions?
Velocity or Ratio
Figure 5.3: A conceptual illustration of overlapping velocity distributions used to identify
probable crust compositions. The central pink region of the diagram, where the measured
seismic velocity distribution (purple) overlaps the Perple X-generated velocity distribu-
tion (tan), are the velocities that are considered the best-fit by the model. The model
records the compositions of the samples that can produce the best-fit velocities.
122
5.4.1 Model Inputs
A global model of Vp, Vs, and Vp/Vs was generated from a compilation of over
5000 (Vp) and 1000 (Vs) 1-D seismic velocity profiles obtained from the Global Seismic
Catalog (GSC) database (Mooney, 2015). Both controlled and passive source seismic
velocity profiles were included to increase data coverage. We included profiles with Vp
and Vs data that had been sampled at a minimum of 5 depth intervals within the crust.
Figure 5.2 shows our tectonic province designations and the location of each seismic
velocity profile used. We used global Moho depths from Litho1.0, except on the conti-
nental margins, where we reference Szwillus et al. (2019) Moho values. In comparison
to Litho1.0, Szwillus et al. (2019) incorporated a larger dataset on the continental mar-
gins (?1600 profiles) and did not average depths across the continent-ocean transition.
Global Moho temperatures were generated from the TC15 global temperature model of
Artemieva (2006). We assumed a linear temperature gradient within the continental crust,
though we address the contributions from crustal heat production in a later section of this
paper. The Szwillus et al. (2019) and Artemieva (2006) models are based on the same
USGS-maintained database as Mooney (2015). In most areas, the middle and lower crust
are not in thermal equilibrium with the conditions of amphibolite and granulite facies,
respectively. Temperature profiles suggest that the deep crust has cooled off since reach-
ing peak metamorphic grade. To account for this in our composition calculations, we
adopt the assumptions of Hacker et al. (2015), which uses an equilibration temperature of
500?C for amphibolite facies in the middle crust and 750?C for the granulite facies in the
lower crust. We performed a temperature correction of 4x10?4 km s?1C?1 (Christensen
123
and Mooney, 1995; Rudnick and Fountain, 1995) to account for the crust?s cooling. We
also perform a pressure correction of -2x10?4km s?1MPa?1 on both facies (Christensen
and Mooney, 1995; Rudnick and Fountain, 1995).
The foundation of the tectonic provinces chosen for this global model are the classi-
fications of crust provided by the Crust family of models (Mooney et al., 1998). To further
identify tectonic provinces and group together geophysically similar crust, we incorpo-
rated crustal thickness, seismic velocity (Vp, Vs), gravity anomaly, sediment thickness,
crust elevation, and surface heat flux observations in a tSNE test (t-distributed stochastic
neighbor embedding). Results generally favored grouping the continental crust into 8 -
12 regimes, mostly matching the designations already given in Crust1.0. We augmented
these regimes with additional groupings, such as ?Thinner Himalyan? crust, when it be-
came clear that the seismic velocity structure of the perimeter of the Himalayas differed
from the thickest Himalayan crust. Areas with sparse seismic coverage such as central
South America and northern Africa, rely heavily on extrapolation of measurements from
similar tectonic provinces. Average Vp and Vs profiles for most tectonic provinces were
created from a distribution of tens to hundreds of individual measurements. A notable
exception is the ?Continental Margins? province, which was represented by > 1,600 pro-
files. Highly localized regimes, such as Andean or Himalayan crust, tended to have <
100 profiles due to the uniqueness of their crustal profiles.
Figure 5.1 and Table 5.1 show the proportion of different crustal provinces by sur-
face area coverage. These tectonic provinces consider only crust exposed at the surface,
so that provinces such as ?Platform? have underlying crystalline crust that may be Pro-
terozoic or Archean in age. The Proterozoic crust covers the largest fraction (32%) of
124
Crustal Regime % All Crust % Continental Crust Number of Profiles
Oceanic 63 - -
Continental Margin 6 16 1693
Archean 5 12 416
Proterozoic 12 32 919
Phanerozoic 3 9 353
Platform 2 5 318
Extended 2 6 403
Rifted <1 1 148
Arc 2 5 262
Paleo-orogenic 3 9 565
Tethyan <1 2 59
Andean <1 1 31
Thick Crust <1 1 106
Thin Himalayan <1 1 28
Table 5.1: Tectonic Regimes by Surface Area
the continental crust, followed by continental margins (17%). The thicknesses of the
crust and lithospheric mantle in Table 5.1 were calculated from the Litho1.0 Moho and
lithosphere-asthenosphere boundary depths. The masses and volumes of the lithospheric
mantle are from Wipperfurth et al. (2020)?s physical properties associated with Litho1.0.
Sammon and McDonough (2021) serves as our geochemical constraint on the deep
(middle and lower) continental crust. That study represents a compilation of major and
trace element abundances for amphibolite and granulite facies rocks. We modeled amphi-
bolite facies lithologies for the middle third of the crust and granulite facies lithologies for
the bottom third, in agreement with the depth assignment of Rudnick and Gao (2014). We
cannot confidently determine a priori which portions of the deep crust are more appropri-
ately represented by amphibolite versus granulite facies data with our current model. In
theory, one facies-classified lithology would have greater overall overlap with the seismic
velocity profile(s), thus determining which is the more accurate rock type to use. In prac-
125
tice, however, amphibolite and granulite facies lithologies of the same SiO2 abundance
tend to have similar seismic velocities (see Section 5.5.1). As such, we have assumed
a metamorphic facies switch from amphibolite to granulite at 2/3 the crustal depth. Fu-
ture investigations should use seismic anisotropy of the deep crust to further constrain its
mineralogy.
Though trace elements do not participate in thermodynamic calculations, we were
able to estimate trace element abundances based on a joint probability analysis with the
mineral-forming major oxides. Samples were placed into bins based on the abundance of
the oxide and trace element of interest (e.g. SiO2 and U). Bin width was selected using
Sturge?s rule (Nbins = log2(N) + 1). For each major oxide composition bin, there was then
a correlated trace element abundance distribution.
5.4.2 Model Uncertainties
Errors in seismic and/or geochemical inputs will skew results. It is imperative to
understand the uncertainties in the input datasets, if we want an accurate picture of the
uncertainty of our crust compositional models.
The program also will not assess modeling errors stemming from foundational as-
sumptions about what types of lithologies should be used as geochemical inputs and the
tectonic provinces assigned to the global crust. These two assumptions are expected to
control the systematic error of the model, which is why we made the program flexible
and modular. Our approach facilitates testing different crustal models and highlights the
projected differences in crust composition.
126
The primary sources of model error stem from uncertainty in the assumed crustal
temperature gradient and Moho depth. Again, these are parameters that can be set by
the user. For our preferred model, the uncertainty on Moho depth is on the order of
10% or less in most areas of the global model (Szwillus et al., 2019). The temperature
uncertainty is much greater. Global Moho temperatures are taken from Artemieva (2006),
which reports no uncertainties. Therefore, uncertainty is taken as the standard deviation
of all temperatures found within a given crustal tectonic province (discussed below), and
the model runs a Monte Carlo of at least 10,000 iterations to produce a distribution of
Moho depths and temperatures. Future results could be improved with Moho temperature
models that quantify uncertainty more directly.
We have also attempted to mitigate the bias introduced by oversampling of partic-
ular geochemical compositions. An over-sampled composition, such as 100 input com-
positions with nearly identical major oxide content artificially inflates the probability of
that composition in our final combined model. However, we do consider the reporting of
compositions to be at least somewhat reflective of the proportion of rock types present in
the deep crust, i.e. if the distribution of reported compositions is bimodal, the rocks in
the deep crust are likely bimodal in composition. Therefore, we only considered a sample
redundant if its oxide content differed from?another?s by < 3 wt.% (9 major oxides, using
the distance between vectors formula d = x12 + x22 + ...+ xn2, where xn is the differ-
ence in wt.% of an oxide between two samples), and its Perple X generated values for
Vp, Vs, and Vp/Vs were within uncertainty of each other.
The internal error contributed by calculational uncertainty is minimal. The over-
lap between of seismic velocity measurements and Perple X-derived seismic velocities
127
is calculated via trapezoidal numerical integration at intervals determined by the uncer-
tainty in the seismological data. When this interval is too large to use for the integration,
the program reduces the interval by half. The precision errors of Perple X are generally
negligible compared to the uncertainty on our other inputs (Connolly, 2005). The sys-
tematic uncertainty of Perple X, however, remains difficult to assess. As the software
is built upon empirical measurements and thermodynamic equations, the experimental
database for crustal minerals is incomplete and relies on ideal thermodynamic relation-
ships to compute expected physical properties.
5.4.3 Quality, Expense, and Time: Global vs. Local Models
In numerical modeling, there is often a tradeoff between computation time and
model resolution. For a global view of the continental crust, breadth and total model
coverage may be more valuable than high data resolution, especially if results can be av-
eraged over large areas. A large-scale, globe-encompassing model, however, comes with
the choice of either short computation time and low resolution or longer computation
time and higher resolution. Alternatively, those interested in a more in-depth analysis of
a localized region may be able to accommodate higher resolution models. We suggest
considering the following when determining whether to use a global or local scale model:
data resolution (especially in seismic velocity profiles), data coverage, and model appli-
cation. Those with data resolution on the scale of > 0.5?x 0.5?should consider using the
global version of the script. Data sources with higher resolution, such as that provided by
the Earthscope USArray, the AUSArray, or the J-ARRAY, should use the regional scale
128
model. For the remainder of this study, we will analyze global model results. Sammon
et al. (2020) presents an example of a local-scale composition analysis using a nascent
version of this method.
5.5 Results
5.5.1 Empirical Composition-Velocity Trends
For granulite and amphibolite facies lithologies seismic velocities correlate with
SiO2 content because of its high abundance, as compared to all other oxides. Perple X-
calculated Vp and Vs values at given pressure-temperature conditions show a quadratic
relationship between SiO2 and velocity (Figures 5.4 and 5.5). The coefficients of the
quadratic vary for different pressures and temperatures, and are ultimately correlated to
the empirical mineral physics datasets used in the Perple X Gibbs free energy minimiza-
tion. Amphibolite and granulite facies lithologies span similar Vp and Vs values, though
the shapes of their distributions are marginally different. This is because the velocities of
many amphibolite and granulite facies minerals (e.g., garnet, pyroxene, plagioclase) have
similar Vp (? 7 km/s) and Vs (? 3.6 km/s) values. Despite considerable scatter in the
Vs data, when paired with Vp, a clear trend emerges: increasing SiO2 abundance leads to
decreasing velocities.
Higher Vp/Vs ratios have lower silica content (Figure 5.6), though for a given SiO2,
there is roughly a 10% spread in Vp/Vs. For both amphibolite and granulite lithologies,
increasing Vs can lead to either an increase or a decrease in Vp/Vs ratio. The maximum
Vp/Vs for amphibolite facies lithologies at typical middle crustal P-T conditions, is ex-
129
Figure 5.4: Vp as a function of SiO2 wt.% for amphibolite (A) and granulite (B) facies
litholoiges at expected deep crustal pressures and temperatures. The color of the data
points indicates percent data point density, with the brighter colors indicating more data
points. The red line shows the best fit quadratic regression between Vp and SiO2 and
changes for different temperatures and pressures.
130
Figure 5.5: Vs as a function of SiO2 wt.% for amphibolite (A) and granulite (B) facies
litholoiges at expected deep crustal pressures and temperatures, generated through Per-
ple X. The color of the data points indicates percent data point density, with the brighter
colors indicating more data points. The red line shows the best fit quadratic regression
between Vs and SiO2 and changes for different temperatures and pressures. There is
more scatter between SiO2 and Vs than SiO2 and Vp, but can be combined for a tighter
constraint on composition than either compressional or shear velocity alone.
131
Figure 5.6: Perple X-generated Vp/Vs plotted against SiO2 for amphibolite (A) and
granulite facies lithologies. Amphibolite results are shown for 450?C and 0.5 GPa. Gran-
ulite results are shown for 650?C and 0.8 GPa. Color indicates relative data point density.
132
pected at a Vs of about 3.5-3.8 km/s, a Vp of 6.5-7 km/s, and SiO2 of 55 wt.%. For
granulite, this maximum is expected at compositions closer to 60-63 wt.% SiO2. Inter-
estingly, the maximum Vp/Vs in granulite lithologies corresponds to the lowest Vs rather
than the highest Vp, suggesting that Vs variations exert a stronger control on Vp/Vs ratios.
5.5.2 Deep Crustal Density
We calculated deep crustal density by tracking the Vp and Vs values from Perple X
that overlapped with our seismological database back to their input samples. Then, we
report the Perple X-derived density of those input samples. We found that the depth-
dependent variation deep crustal densities among the different tectonic provinces appear
more similar when normalized to crustal thickness (Figure 5.7). Uncertainties in density
are ? 3% for each tectonic province, a result which is likely controlled by the ? 3%
uncertainty on velocities. Deep crustal density ranges from 2700-2780 kg/m3 at 13 km
depth to 3290-3340 kg/m3 at the Moho.
We note that, in order to calculate deep crustal pressure, and thus mineralogy and
composition, we already assumed a bulk crustal density of 2900 kg/m3 to calculate litho-
static pressure. This initial assumption, though, does not significantly affect our compo-
sition results because there is, at most, a calculated pressure difference of <15% caused
by using the 2900 kg/m3 a-priori density vs. our model-generated density. This <15%
pressure difference does not greatly change the stable mineral assemblages or velocities
calculated by Perple X for the deep crust.
The average densities and total masses of each tectonic province listed in Table 5.1
133
A 0 B 0 Continental Margin
Archean
20 Median Density 20 Proterozoic
All Types Phanerozoic
Platform
40 40 Extended
Rifted
Arc
60 60 Paleo-orogen
Tethyan
80 80 AndeanThick Crust (>60 km)
Thin Himalaya
100 100 Mantle Density
2600 2800 3000 3200 3400 2600 2800 3000 3200 3400
Density (kg/m3) Density (kg/m3)
Figure 5.7: Calculated density normalized to depth for (A) average continental crust and
(B) our different tectonic regimes with an imposed lithology transition (amphibolite to
granulite facies) at 2/3 total crust depth. By this method, the bottom ? 20?30% of the
crust approaches or exceeds mantle density.
are calculated by this study. The average thicknesses, total surface area, and total volume
of each tectonic province are calculated from Litho1.0?s Moho and LAB depths, whose
uncertainties are ?12% (c.f. Dziewonski and Anderson (1981); Huang et al. (2013);
Pasyanos et al. (2014)). Continental Margin Moho depths are an exception and are from
Szwillus et al. (2019) (Section 5.4.1). The densities of the oceanic crust and lithospheric
mantle, not calculated by this study, were assumed to be 2900 ? 116 kg/m3 and 3400
? 136 kg/m3, respectively (Carlson and Herrick, 1990; Huang et al., 2013). The total
volume of continental crust is 9.2 x 109 km3. The continental crust, including submerged
continental margins, comprises 44% of Earth?s surface area. Multiplying the density of
each tectonic province by its volume yields the mass of crust predicted in each tectonic
province. We assumed one-third of the crustal column was upper crust with a density of
2650 kg/m3 (Christensen and Mooney, 1995). The total mass of the continental crust,
2.32 x 1022 kg, is 4% higher than Albare?de (1998); Cogley (1984); Rudnick (1995);
Wipperfurth et al. (2020).
134
% Total Depth
% Total Depth
5.5.3 Composition
Our main analysis focuses on SiO2 abundance and its uncertainties because of its
strong correlation to seismic velocities, particularly Vp. The SiO2 content at typical mid-
dle and lower crustal depth intervals (Figure 5.8) is given in Table 5.2. All 9 major oxide
inputs (SiO2, TiO2, Al2O3, CaO, MgO, FeOT , MnO, K2O, Na2O) can be found in Table
5.3 and corresponding maps in Supplement Section 3. We use the notation ?Mx%?, where
x% is the percent distance to the Moho (M) from the surface. This notation normalizes
comparisons among tectonic provinces of various thicknesses. We found that when we
normalize depth in this manner, the composition vs. depth profiles for each province
become more similar. The top of the deep crust starts at an intermediate composition,
globally ranging from 58 - 68 wt.% SiO2, and the composition gradually decreases to
50-55 wt.% SiO2 as it approaches the Moho. Global scale SiO2 composition of the con-
tinental crust mostly decreases (or remains steadily mafic) with increasing depth for all
tectonic provinces (Figure 5.9). Uncertainty in global SiO2 also decreases with increas-
ing depth due to fewer samples fitting the seismic signal at depth. However, the Andean
and Himalayan tectonic provinces have larger uncertainties at depth because of the varia-
tion in geochemical data fitting the seismic signal and the sparsity of deep seismological
profiles.
CaO content of the deep crust is also of interest due to its absolute abundance and
significance as a contributor to sedimentary deposits, though only siliciclastic rocks and
not carbonates were considered viable deep crustal components (Hartmann et al., 2012;
Wilkinson et al., 2009). In our model, Ca is mostly contained in plagioclase, pyroxene,
135
75
A SiO2 at 30 km 70
65
 45? N  60
55
  0?  
50
 45? S  45
40
SiO2 at M85%
B 75
70
 45? N  
65
  0?  60
55
 45? S  50
45
40
Figure 5.8: Global SiO2 composition at a depth of 30 km shows regional distinctions
whereas measuring composition at a crustal depth relative to the Moho (M85% notation
= 85% of the total crustal depth) produces a view of a deep crust that is contiguous and
decreases in SiO2 gradually with depth. Areas of high projected SiO2 include the Hi-
malayas, Andes, East African rift, and some continental margins. While the Himalayas
and Andes may show compositional features, the high SiO2 in some rifts and continental
margins are likely from model input inaccuracies
136
wt.% wt.%
SiO2 at M50% Uncertainty SiO2 at M85% Uncertainty
(? middle crust) ? (? lower crust) ?
Continental Margin 58.8 10.2 50.8 2.7
Archean 59.1 7.8 50.3 2.9
Proterozoic 58.0 7.3 50.9 3.7
Phanerozoic 59.2 9.2 49.9 2.5
Platform 56.9 7.6 50.0 2.7
Extended 66.1 7.6 50.6 2.4
Rifted 63.9 7.3 55.3 6.5
Arc 65.9 9.8 55.2 6.9
Paleo-orogenic 61.6 9.2 50.8 3.4
Tethyan 61.7 9.4 50.5 3.1
Andean 57.7 9.1 54.0 6.9
Thick Crust 65.4 8.8 56.0 8.6
Thin Himalayan 68.9 5.9 50.0 2.2
Table 5.2: SiO2 in wt.% for different tectonic regimes
and garnet. The CaO abundance tends to increase with depth because of the increasingly
mafic nature of the deep crust, and therefore regions of low SiO2 correlate with regions of
high CaO. Globally, the median CaO at crustal depths of M85% is 9.1 ? 3.1 wt.% (Figure
5.10).
We can also derive the global distribution of a trace element if that trace element
has a quantifiable relationship to one of the thermodynamic components (major oxides)
used in our model. We used a geochemical database of samples with both major and
trace element concentrations (Sammon and McDonough, 2021) to generate trace element
maps as a function of major oxide abundances. We used a bivariate probability analysis
to generate trace element distributions from a major oxide abundance, such as SiO2, at a
specific depth or location. Although we suggest using regional analyses for high resolu-
tion interpretations of trace element abundances, we present here global predictions and
uncertainties for Sr (Figure 5.11) and U (Figure 5.12) content based on their relationships
137
Composition at M50%
 45? N  
75
  0?  
70
 45? S  
65
Composition at M70%
60
 45? N  
55
  0?  
50
 45? S  
45
Composition at M90% 40
 45? N  35
  0?  
 45? S  
Figure 5.9: Global SiO2 decreases with increasing depth from the middle to the bottom
of the continental crust. The middle crust M50% ranges from 60 to 65 wt.% SiO2 in most
areas and increases at a rate of about wt.% per km until reaching the base of the crust.
138
SiO2 (wt.%)
Composition at M50% Uncertainty Composition at M85% Uncertainty
(? middle crust) ? (? lower crust) ?
SiO2 59.1 7.31 50.9 2.98
TiO2 0.74 0.38 0.82 0.40
Al2O3 15.8 1.68 16.4 3.46
FeO 7.26 2.93 9.23 2.25
MnO 0.12 0.06 0.16 0.06
MgO 2.93 1.73 5.60 2.81
CaO 5.52 2.05 8.58 3.08
Na2O 3.64 0.81 2.16 1.02
K2O 1.41 0.97 0.77 0.96
Table 5.3: Middle and Lower Crust Bulk Composition in wt.%
with CaO and SiO2, respectively, as examples. Global average Sr increases with increas-
ing CaO until plagioclase is no longer the dominant Ca-bearing mineral. Uncertainties
on global U concentration span an order of magnitude because the abundance of U in a
given metamorphic sample set ranges from a few hundreds of ppb to a few ppm. U and
SiO2 abundances, however, are positively correlated, with median U increasing as median
SiO2 increases.
5.6 Discussion
5.6.1 SiO2 and Overall Deep Crustal Composition
Figure 5.9 shows steady or decreasing SiO2 with increasing depth in the deep crust.
Figure 5.8 also makes it apparent, though, that the absolute SiO2 at a given depth is not
equal across different crustal types. For example, ?Extended? crust appears mafic at 30
km depth while the ?Thick Himalayan? crust is felsic at that depth, and ?Proterozoic?
crust falls in between. However, a more laterally consistent trend appears when compar-
139
CaO at M 14A 85%
12
10
 45? N  
8
  0?  
6
 45? S  4
2
B CaO Uncertainty at M85% 5
 45? N  4
3
  0?  
2
 45? S  
1
0
Figure 5.10: Global CaO abundance and uncertainty at 85% of the total crustal depth.
Areas of low CaO correlate to areas of high SiO2. There does not appear to be any
correlation between CaO content and uncertainty, with most regions having 3 to 4 wt.%
uncertainty regardless of CaO abundance.
140
Uncertainty (wt.%) wt.%
Figure 5.11: Global Sr abundance and uncertainty was derived from a joint probability
analysis with CaO at 85% of the total crustal depth.
141
ing percent of the crustal column traversed rather than absolute depth (Figure 5.8). Most
regions show a 5-10 wt.% decrease in median SiO2 through the deep crust regardless of
crustal thickness, so that SiO2 decreases more rapidly in areas of thin crust than in areas of
thick crust. We predict the global median SiO2 at 50% above the Moho (or, alternatively,
50% crustal column thickness) to be 61.2 ? 7.3 wt.% SiO2 with CIPW normative min-
eralogy of <10 wt.% alkali feldspar and <15 wt.% quartz. The average composition of
middle continental crust is therefore expected to resemble that of a quartz monzonite; the
lower crust average composition, with 50.5 ? 3.4 wt.% SiO2 and 9.1 ? 3.1 wt.% CaO, is
expected to resemble a gabbronorite. It must be noted, however, that these compositions
can be the result of seismic velocity profiles averaging felsic and mafic endmembers of
the deep crust, and need not be produced by a monzonite or norite.
Density sorting provides a simple mechanism for producing the compositional struc-
ture of the continental crust. The process of crustal genesis leaves mafic, restitic material
at the base of the crust regardless of crustal thickness except in the few cases discussed in
the next paragraph. More buoyant, felsic material ascends to the top of the crust, produc-
ing a gradient of SiO2 that scales with crustal thickness. In addition to density sorting,
the deep crust could be more mafic because it is simply closer to the mantle and therefore
has a greater number of mafic intrusions. Our results do not indicate any need for sharp
compositional boundaries in the deep crust. The MX% notation reinforces the importance
of scaled, relative depth in the crust rather than absolute depth for making compositional
comparisons.
Two regions that appear conspicuously more felsic than the global deep crustal
median are the Andes and the Thin Himalayan crust (Figure 5.9). A low temperature
142
gradient could be the cause of the compositional difference between these two tectonic
provinces an the global average, but we also must consider two other possibilities, par-
ticularly around the northern and northeastern Tibetan Plateau and Himalayan ramp. The
first is that thick, convergent margins, especially in the Himalayas, might have layers of
upper crustal material thrust deeper within the crust, incorporating more felsic material
into the deep crust. In contrast, underthrust upper crustal material is less likely to appear
in the Andes, which is a continent-ocean subduction zone. Alternatively, pockets of melt
and partially melted material in the Andean and Himalayan middle and lower crust could
reduce the shear wave velocity (Caldwell et al., 2009; Hacker et al., 2014; Nelson et al.,
1996; Regis et al., 2016; Schilling and Partzsch, 2001; Schmitz et al., 1997; Searle et al.,
2009). Because our current model does not factor in melt, slower Vs speeds would be
attributed to a more felsic composition instead.
Other anomalous regions in Figure 5.8, particularly the continental margins of
Antarctica, the East African rift zone, and the Sea of Japan, are likely caused by inac-
curate temperature and Moho inputs. The East African Rift could appear felsic because
the model?s temperature gradient for that actively rifting region is too low; a cooler fel-
sic composition can produce the same velocities as a warmer mafic composition. On
the other hand, the highly localized, extremely felsic borders around Antarctica and be-
tween Japan and China likely indicate a misclassification of crust type and/or Moho depth.
Oceanic-type crust has been documented in both regions (Cho et al., 2004; Gohl, 2008;
Hirata et al., 1992; McCarthy et al., 2020). Better Moho and temperature resolution of the
ocean-continent transition should increase the accuracy of compositional models in these
regions.
143
Mafic granulite lithologies reach gravitational instability in the lower 10-20% of
the average crustal column (Jagoutz et al., 2011), surpassing the upper mantle?s density
of 3300 kg/m3. Therefore, according to Figure 5.7, most of the granulite facies lower
crust for continental margins, Andean crust, Tethyan crust, and Phanerozoic crust should
be gravitationally unstable. On the other hand, most other tectonic provinces would just
reach mantle-like densities around the Moho depths. Thinner Himalayan type crust has a
middle crustal density?9% lower than other provinces, correlating with negative seismic
velocity anomalies. Arcs have the next lowest densities on average, suggesting that a
potentially denser lower crust beneath some arcs has already foundered (Jagoutz et al.,
2011). Figure 5.7B shows that accreted arcs of Andean type crust in particular display a
stark decrease in density interpreted as being due to delamination of the lowermost crust
(Ducea, 2011; Gao et al., 2021; Kay and Kay, 1993).
Forming continental crust via island arc processes, however, would then require
the deep crust to become denser over time, since most of our tectonic provinces have
a lower crust that is denser than arclower crust. This can be achieved by cooling the
crust, thickening it further, intra-crustal differentiation, or by mafic igneous injections
into the lower crust. Moreover, if our Moho temperature model is too hot, though, it
requires denser, more mafic lower crust than is actually present to explain the observed
Vp and Vs values. As such, our compositional models are intrinsically tied to Moho
temperature, and may be skewed towards mafic granulites. Reducing the assumed crustal
Moho temperatures would bring the estimated average crustal density closer to arc crust
density and increase the average weight percentage of SiO2.
There is a tradeoff between temperature and composition. Vp and Vs both carry
144
a temperature dependency through their bulk and shear moduli, so accurate temperature
estimates are imperative for modeling the crust; decreased seismic velocities can be the
result of either higher temperatures or greater SiO2 content. As a reminder, this study
implements a linear temperature gradient through the crust from the TC15 global temper-
ature model (Artemieva, 2006).
Table 5.4 reports compositional models for the middle and lower continental crust,
a practice that is required to make meaningful comparisons to previous crustal models.
While we recognize the assumption of a three-layer crust as an oversimplification of the
diversity of crustal compositions, it is useful for some calculations to have average com-
position numbers for the crust. A few examples include mantle tomography studies which
require crustal correction, crustal corrections for geoneutrino studies; models of Earth?s
thermal history; and planetary scale compositional model for comparison with other rocky
bodies. Compositional models in Table 5.4 have been normalized to 100 wt.%. Our mid-
dle crustal composition falls between two possible compositions given by Hacker et al.
(2015): the fastest Vp endmember composition for the middle crust (62.7 wt.% SiO2),
and the middle crustal composition expected when the crust takes on a two compositional
layer (upper and lower) structure, instead of three, (57.3 wt.% SiO2). These SiO2 esti-
mates also overlap with the 62 wt.% SiO2 reported by Christensen and Mooney (1995)
and are more mafic than than the 63.5 wt.% SiO2 middle crust reported by Rudnick and
Gao (2014). Our proposed lower crust composition is agrees with Rudnick and Gao
(2014) and other mafic models (e.g. Hacker et al. (2015)?s fast Vp lower crust; Jagoutz
and Schmidt (2012)). Models which predict a more intermediate-felsic lower crust, such
as the North China craton lower crustal model of Liu et al. (2001) or the higher SiO2,
145
lower Vp options listed by Hacker et al. (2015), are not consistent with our global aver-
age, though isolated regions of more felsic lower crust may exist.
5.6.2 CaO and Sr
Bulk CaO concentration increases with depth (Figure 5.10), but as a component
of mafic, siliciclastic rocks, not carbonate. The lack of carbonate is due in part to our
imposed amphibolite/granulite grade lithology restrictions on possible deep crust com-
position, but is reinforced by high density and Vp values observed in the deep crust.
Carbonates, with densities of approximately 2750 kg/m3 and Vp?s of 6.6 ? 6.8 km/s at
deep crustal conditions (Christensen and Mooney, 1995), cannot substantially contribute
to the observed deep crustal velocities. Also, there are few carbonate-dominated gran-
ulite facies xenoliths and terrains as compared to silicate granulites. A comparison of
Figures 5.8B and 5.10 shows good correlation globally between regions of high SiO2 and
low CaO. Uncertainties in CaO estimates follow the same trends as SiO2 uncertainties,
though the former?s uncertainty is roughly 10% higher. The relationship between Vp or
Vs and CaO content has more scatter than with SiO2, which increases the % uncertainty
on CaO estimates.
The CaO content does, however, predictably track with Sr concentration (Figures
5.10 and 5.11). Sr abundances cannot be directly derived from velocity calculations, but
they can be predicted from its geochemical relationship with CaO. Patterns emerge when
comparing the global distribution of Sr and CaO from two distinct sources: equilibrium
mineralogy and data binning. First, Sr abundance increases for CaO contents between 2
146
Christen Liu Jagoutz & Rudnick Hacker Hacker This Study
& et al., Schmidt, & Gao, et al., et al.,
Mooney, 2001 2012 2014 2015? 2015?
1995
Middle Crust
SiO2 62 - - 63.5 62.7 57.3 61.2
TiO2 - - - 0.69 0.8 0.99 0.77
Al2O3 - - - 15 15.7 16.8 16.4
FeOT - - - 6.02 6.76 8.15 7.52
MnO - - - 0.10 0.13 0.16 0.12
MgO - - - 3.59 3.51 4.46 3.04
CaO - - - 5.25 5.27 6.63 5.72
Na2O - - - 3.39 3.42 3.89 3.77
K2O - - - 2.3 1.6 1.42 1.46
Mg# - - - 51.5 48.1 43.4 41.9
Lower Crust
SiO2 47 58.3 52.16 53.4 50.7 57.3 53.8
TiO2 - 0.59 0.78 0.82 1.24 0.99 0.87
Al2O3 - 13.6 18.68 16.9 16.5 16.8 16.3
FeOT - 5.32 8.41 8.57 10.39 8.15 9.75
MnO - 0.08 0.17 0.10 0.19 0.16 0.17
MgO - 9.58 5.86 7.24 7.03 4.46 5.92
CaO - 4.54 10.79 9.59 10.1 6.63 9.07
Na2O - 2.54 2.56 2.65 2.8 3.89 2.28
K2O - 3.23 0.41 0.61 0.79 1.42 0.81
Mg# - 76.2 55.4 60.1 54.7 49.4 52.0
? Hacker et al. (2015) fast Vp crustal model
? Hacker et al. (2015) middle crust composition = lower crust composition model
Table 5.4: Continental crust composition estimates
147
to 6 wt.%, reaching a maximum at about 500 ppm Sr. However, Sr gradually decreases to
350 ppm as CaO increases to >6 wt.%. This shift in Sr abundance corresponds with the
transition from plagioclase as the only Ca-bearing mineral phase to the addition of garnet
and clinopyroxene as stable Ca-bearing phases.
Second, we see sharp jumps in Sr abundance in neighboring tectonic regions as
a consequence of our data binning. The uncertainty on CaO content dictates that the
compositional bin-widths are as wide as 2-3 wt.% for a total of six bins. Each bin has
a central Sr value and distribution, leading to six possible median Sr abundances. The
uncertainties on Sr are a combination of the systematic uncertainty (which CaO bin) and
the statistical uncertainty (Sr variation within each bin).
5.6.3 Heat Production and Moho Heat Flux
Radioactive elements produce heat inside Earth, with K, Th, and U, the heat pro-
ducing elements (HPEs), accounting for 99.5% of the total radiogenic energy output (Mc-
Donough et al., 2020). The few remaining radioactive elements in the planet contribute
<0.5% to Earth?s total heat production. We can model the abundance and distribution of
HPEs in the crust, and Moho heat flux, with our model and supplementary geochemical
relationships.
HPE abundance data are provided in the (Sammon and McDonough, 2021) geo-
chemical dataset. Using a joint probability analysis, we derived U abundances with un-
certainties from SiO2. The middle continental crust has a median U concentration of 1.3
? 0.2 ppm and the lower crust has 0.22 ? 0.03 ppm. The uncertainty on these median
148
A U Composition at M85%
0.7
 45? N  
0.5
  0?  
0.3
 45? S  
0.1
B U Uncertainty at M 1.8 x10?85%
1.1 x10?
 45? N  
6.7 x10?
  0?  
4.0 x10?
 45? S  
2.4 x10?
1.5 x10?
Figure 5.12: Global U abundance derived from a joint probability analysis with SiO2
at 85% of the total crustal depth. Uncertainties span orders of magnitude because of the
range of possible U values, but the global median at this depth is?0.2 ppm U. Regions of
high SiO2, especially the potentially inaccurate continental margin of Antarctica correlate
with high U and the highest uncertainties.
149
ppm
% uncertainty
values is reported as the standard error of the median. We do this to highlight that this
is the confidence in the central value, the median. The full range of possible abundances
spans five orders of magnitude (Fig. 5.12B) Th abundances could also be generated with
a joint Th-SiO2 probability analysis, but we use a constant Th/Umass ratio of 3.77 ? 0.1
(Sammon and McDonough, 2021; Wipperfurth et al., 2018) which provides a narrower
constraint on Th concentration. K2O abundance is calculated directly as a thermody-
namic component in our compositional model. We calculate heat production on the same
1?latitude by 1?longitude cell scale that we use for calculating major element abundances.
The concentrations of K, Th, and U in each cell are multiplied by each cell?s density (cal-
culated in previous section) and volume. The resulting masses of K, Th, and U are then
multiplied by the heat produced during their respective decays (McDonough et al., 2020),
resulting in a heat production value for each cell in units of W/kg. Heat production values
can be converted to W/m3 via multiplying by cell density once more. Uncertainties are
calculated through standard propagation of errors on concentration and density. Figure
5.13 maps the total (depth-integrated) median contribution to surface heat flux from the
upper, middle, and lower continental crust in mW/m2.
We calculated a median heat production of 0.3 nW/kg (0.8 ?W/m3) for the middle
third of the crust and 0.05 nW/kg (0.1 ?W/m3) for the lower third of the crust. Previous
studies have also found low heat production in the deep continental crust (Fountain et al.,
1987; Jaupart et al., 2016; Kukkonen et al., 1997). Our model is also consistent with local
studies of HPE analyses of deep crustal xenoliths, such as Gruber et al. (2021); Pinet and
Jaupart (1987); and Ashwal et al. (1987).
The uncertainties on this global model are dominated by uncertainties on U abun-
150
Contribution to Crustal Heat Production
A Upper Crust B Middle Crust 18
40
16
 45? N  ?30  45  N  14
  0?    0?  12
20
 45? S   45? S  
10
10
8
0 6
C Lower Crust D Total Continental Crust 70
10
60
? 8 45  N   45? N  50
6 ? 40  0?    0   
4 30
 45? S   45? S  
20
2
10
0
Figure 5.13: Panels A - C show the contribution from each crustal layer (upper, middle,
lower) to the total continental surface heat flux (D). Continental crustal heat production
is dominated by upper crustal heat production but with significant contributions from the
middle crust.
151
mW/m? mW/m?
mW/m? mW/m?
dances, which span an order of magnitude. Even so, our uncertainty on the median or cen-
tral value (standard error of the median) of HPEs or heat production is well constrained 
at pm 0.1%. While the range of possible heat production values span an order of mag-
nitude, the median/average heat production value is tightly constrained. Provinces with 
high predicted SiO2 content, such as the Andes, have estimated U content up to four times 
higher than the global lower crustal median because of the correlation between high SiO2
samples and high U. We recommend using regional HPE data for understanding smaller 
scale variations and reserve this study?s results for continent- or global-scale models.
We took our calculations a step further by estimating Moho heat flux. O nce we 
determined deep crustal heat production, we calculated a subcontinental Moho heat flux 
by subtracting continental heat production from the global surface heat flux of Lucazeau 
(2019), supplemented with Shen et al. (2020)?s Antarctica surface heat flux. We assumed 
a Gaschnig et al. (2016) upper crustal heat production of 1.75 ?W/m3 to complete the 
initial calculation. Pairing an upper crustal composition of Gaschnig et al. (2016) with 
our deep crustal composition yielded Moho heat fluxes for tectonically stable regions that 
agree with Jaupart et al. (2007) and marginally agree with (Hacker et al., 2015)?s models 
of crust with slow Vp. However, Moho heat flux calculations depend substantially on the 
assumed HPE abundance model for the upper crust, which contributes ?70% of the total 
crustal heat production in most regions. Using Gaschnig et al. (2016)?s upper crustal U 
and Th abundances in low heat flux, cratonic regions, results in roughly 6% (by area) of 
the continents having a negative heat flux across the Moho (an unreasonable condition) 
? or more likely, other factors, such as heat dissipation through fluid circulation in the
near surface, would be needed to explain these low surface heat flux (e.g., 20-40 mW/m2) 
152
Parameter Value
Global Surface Heat Flux inputs from Lucazeau (2019)
Antarctica Surface Heat Flux inputs from Shen et al. (2020)
Upper Crust Heat Production 1.75 ?W/m3 (Gaschnig et al., 2016)
Upper Crust Heat Production (cratonic) 0.8 ?W/m3 (see discussion for sources)
Deep Crustal Densities inputs from this study
Table 5.5: Heat production calculation parameters
regions. Alternatively, the assumed upper crustal heat production values may need to be
lowered. Most low heat flux areas coincide with stable cratonic lithosphere, where low
heat flux and low heat production is not a new observation (e.g., Cammarano and Guerri
(2017); Jaupart et al. (2007); Kukkonen et al. (1997); Nyblade and Pollack (1993)). Es-
timates of upper crustal heat production in cratonic regions range from 0.6 to 1 ?W/m3
(Gruber et al., 2021; Jaupart et al., 2014, 2016; Mareschal and Jaupart, 2013; Phaneuf
and Mareschal, 2014), so we approximated Archean upper crustal heat production at 0.8
?W/m3, which is the maximum permissible heat production value found by Rudnick and
Nyblade (1999) for the Kalahari craton and also the maximum average crustal heat pro-
duction expected for crust ? 2.5 Ga (Jaupart et al., 2016). Our final model for Moho
heat flux uses the data and resources listed in Table 5.5. The median global continental
Moho heat flux, shown in Figure 5.15, is 21.5 ? 11.6 mW/m2. Figure 5.14 summarizes
the distribution of Moho heat flux values for each tectonic province.
Table 5.6 compares our average crustal heat production and Moho heat flux val-
ues to the average values calculated from other studies. We calculated the Moho heat
fluxes from the previous studies by taking the average surface heat flux and subtracting
the crust?s average radiogenic contribution. For consistency, we recalculated our study?s
Moho heat flux in this simplified manner as well. Total crustal HPE concentrations av-
153
154
Models of the continental crust
RG13 HKB11 HKB15 low HKB15 high WSM20 This Study
K (ppm) 15000 15800 11900 18800 11700 15400
Th (ppm) 5.60 7.30 4.20 5.31 5.64 5.40
U (ppm) 1.30 1.40 1.09 1.33 1.27 1.40
Heat production K (nW/kg) 0.0508 0.0535 0.0403 0.0637 0.0396 0.0522
Heat production Th (nW/kg) 0.147 0.191 0.110 0.139 0.148 0.141
Heat production U (nW/kg) 0.128 0.138 0.107 0.131 0.125 0.138
Total crustal heat production (nW/kg) 0.326 0.383 0.258 0.334 0.312 0.331
Total crustal heat production (mW/m3) 0.945 1.111 0.748 0.969 0.905 0.960
Crustal radiogenic heat production (TW) 7.56 8.89 5.99 7.75 7.24 7.68
Upper crust contribution 59.4% 50.5% 75.0% 57.9% 62.0% 58.4%
to total heat production
Radiogenic contribution to 33.7 39.6 26.7 34.6 32.3 34.2
surface heat flux (mW/m2)
Average crustal Moho heat flux (mW/m2) 29.3 23.4 36.3 28.4 30.7 28.9
Data sources: RG13 = (Rudnick and Gao, 2014), HKB11 = (Hacker et al., 2011), HKB15 = (Hacker et al., 2015), WSM20 =
(Wipperfurth et al., 2020), S?ra?mek et al. 13 = (S?ra?mek et al., 2013). ?Moho heat flux is calculated by taking the average surface heat
flux for the continents 63 mW/m2 (and the oceans 65 mW/m2) (Lucazeau, 2019) and subtracting the continental crust?s (and ocean)
radiogenic contribution. ?Upper crust contribution is calculated using the upper crust composition from Rudnick and Gao (2014)
relative to the total crust production; note all models assume this same upper crust composition.
Table 5.6: Heat production in the continental crust
62 72 64 73 69 64 74 65 69 76 98 63 65
150
100
50
0
150 24 21 29 32 27 23 34 25 33 22 32 21 24
100
50
0
rgin ean ic ic ms ed ed en rc ana h a
n ust
c rozo rozo tfor end Rift rog A thy nde Cr ay
an
nt. M Ar e e la t -o Te Ao Prot han . P Ex leo icke
st 
d  Him
al
C P Se Pa Th Thin
Figure 5.14: The median global Moho heat flux is 21.6 ? 11.6 mW/m2, shown by the
blue shaded area at the bottom of the graph. Global surface heat flux is 63.8 ? 12.3
mW/m2, indicated by the red shaded band towards the top. Although variations exist
among the tectonic provinces, they show remarkably consistent central heat flux values.
These statistics were produced by calculating the median heat flux for each latitude x
longitude cell. Outliers are not shown on the graphs. The white line in the middle of each
box and whisker plot is the median value. This value is written above the box and whisker
plot in blue for the Moho heat flux and in red for the surface heat flux for each province.
erage upper, middle, and lower crustal abundances. Figure 5.16 compares the different
models of U abundance in the bulk continental crust from Table 5.6 to their corresponding
Moho heat flux values.
Notably, all models reported in Table 5.6 assume a similar compositional model for
the upper crust (Gaschnig et al., 2016; Rudnick and Gao, 2014). A 1% change in upper
crustal U has eight times the effect on the Moho heat flux as a 1% change in lower crustal
U content. Thus, having an accurate model for the upper crust is of paramount importance
to determining Moho heat flux. Only after upper crustal heat production is constrained
will the uncertainties on deep crustal heat production impact Moho heat flux calculations.
155
Moho Heat Flux Surface Heat Flux
mW/m mW/m
Figure 5.15: Global heat flux across the Moho calculated by subtracting crustal heat
production from measurements of surface heat flux. The median subcontinental Moho
heat flux is 21.5? 11.6 mW/m2 globally. This result is heavily sensitive to the assumption
of a uniform upper crustal heat production of 0.8 ?W/m3 for cratonic and 1.65 ?W/m3
for non-cratonic regions.
156
1.4
1.3
1.2
Linear:  y = - 0.016*x + 1.6
1.1             R
2 = 0.9395
5 10 15 20 25 30
Moho Heat Flux (mW/m2)
Figure 5.16: There is a linear dependency between crustal U concentration and calculated
Moho heat flux. When the different models of Table 5.6 are compared, differences in each
models? predicted Moho heat flux are mostly the result of differences in upper crustal HPE
concentrations.
157
Bulk Crustal U Concentration (ppm)
5.7 Conclusions
We have constructed a global compositional model of the deep continental crust
by synthesizing seismic, temperature, heat flux, and geochemical data. We predict deep
crustal compositions on the global scale using major and trace element compositions from
amphibolite and granulite facies lithologies, and seismic velocity profiles. Our proposed
global compositional model reconciles the USGS Global Seismic Catalog of crustal ve-
locities, compositions for thousands of medium and high grade metamorphic rocks, and
constraints on Moho depth (Pasyanos et al., 2014; Szwillus et al., 2019), crust temperature
(Artemieva, 2006), and surface heat flux (Lucazeau, 2019; Shen et al., 2020).
Vp values, and to a lesser extent Vs and Vp/Vs values, correlate with bulk rock
SiO2 content, and SiO2 can be used as a predictor of seismic velocities if temperature can
be estimated accurately. Globally, SiO2 concentrations tend to decrease with increasing
depth, leading to a predominantly mafic to intermediate-mafic base of the crust. The de-
creased density and less mafic nature of the lower crust in younger and tectonically active
crust, such as arcs and active mountain ranges, suggests that those tectonic provinces are
hotter than our temperature model predicts, that they have undergone lower crustal delam-
ination, or both. Global median SiO2 concentrations for the middle and lower crust are
61.2 ? 7.31 and 50.5 ? 3.4 wt.%, respectively, though steady composition and velocity
gradients in the deep crust urge us to embrace a less distinctly layered view of the crust.
This mid-to-deep crustal gradient in wt.% SiO2 is the equivalent of a lithological gradient
ranging from quartz monzonite to gabbronorite. We predict the abundances of multiple
major oxides, many of which are correlated to trace element abundances. This correlation
158
allows us to derive expected heat production in the deep crust. We also predict a Moho
heat flux of 21.5 ? 11.6 mW/m2.
159
Chapter 6: Conclusions and Future Work
6.1 Key conclusions
The chemical composition of Earth?s continental crust is the sum of its evolutionary
processes throughout time. Radiogenic heat production in Earth?s interior is a particularly
important feature to understand if we want to decipher Earth?s starting materials and for
how long our mantle will maintain convection. Once the planet runs out of heat energy in
its interior, its core will fully solidify, its protective magnetic field will dissipate followed
by a significant loss of atmosphere, and the surface-renewing forces of plate tectonics will
cease.
The concentrations of heat producing elements (HPEs: K, U, and Th) and their dis-
tribution throughout the crust can be inferred from many different geological, geochem-
ical, and physical methods. Ultimately, using all of these techniques in tandem provides
the most clarity, constraint, and quantifiability on the chemical attributes of the crust and
mantle.
Geoneutrino detection remains the only method to directly sample the HPE con-
tent of Earth?s inaccessible mantle, but without accurate knowledge of the geology sur-
rounding geoneutrino detectors, geoneutrino signal measurements alone constrain noth-
ing. Chapter 2 shows the importance of applying an accurate, precise lithospheric HPE
160
correction to the geoneutrino signal. Changes in U and Th content surrounding a detector
balloon into vastly different results for Earth?s total radiogenic heat production.
While up to 70% of the geoneutrino signal originates from the uppermost portions
of the continental crust, the 30% from the deep crust proves more challenging to constrain.
Moreover, the composition of the deeper portion of the continental crust holds crucial in-
formation about how Earth?s outermost layer forms, deforms, and evolves through time.
The chemical makeup of the deep crust controls not only its contribution to the geoneu-
trino signal (through its HPE content), but its stability and strength overall.
The deep continental crust is unreachable by traditional geological sampling meth-
ods. However, we can studying medium- and high- grade metamorphic lithologies, which
serve as analogues or windows into the higher pressure and temperature regions of the
crust (Chapter 3). Analyzing the composition of thousands of amphibolite, granulite, and
eclogite facies xenoliths and terrains gives insight into the chemical and petrological be-
haviors of major element oxides and trace elements. In Chapter 3, we learned that the
?Daly Gap? phenomenon is strongly reflected in the compositions of deep crustal ana-
logue samples, leading to an abundance of mafic (<52 wt.% SiO2) and felsic (>68 wt.%
SiO2) rocks, and few rocks of intermediate composition. Though metamorphic grade
does play a role in the composition of these medium- and high-grade metamorphic rocks,
the degree of melt evolution/differentiation has an even stronger correlation with trace el-
ement abundances. In short, it is just as important to constrain major element composition
as it is metamorphic grade.
Because deep crustal lithologies range from mafic to felsic in composition, study-
ing the geochemistry of these rocks alone leaves multiple possibilities for deep crustal
161
major and trace element content. Similarly, geophysical applications, such as seismol-
ogy, provides multiple nonunique solutions to crustal composition, since different rock
types can have the same seismic velocities. The key to determining composition, both
on the regional and global scale, lies in reconciling geochemical trends with geophysical
observations. Creating joint geochemical-geophysical models sheds more light on deep
crustal composition.
We built a probabilistic model of deep crustal composition by calculating the ex-
pected physical properties of thousands of amphibolite and granulite samples at specific
pressures and temperatures. We then used a Monte Carlo numerical simulation to find
which compositions best reproduced the seismic properties (i.e., Vp, Vs, and Vp/Vs ra-
tios) observed by regional velocity surveys. In Chapter 4, we applied a geochemical-
geophysical data and models to a specific region, where pairing our model with obser-
vations of surface geology allowed us to infer crustal deformation schemes. We adapted
and expanded our joint model by incorporating global seismic datasets and splitting the
crust into tectonic provinces in Chapter 5. The resulting code gave us a lower resolution
but larger scale map of the middle and lower continental crust (Chapter 5).
In both the local and global models, we observed that the deep crust displays a
gradient of decreasing SiO2 with increasing depth, though the severity of this gradient
depends on specific location. In most regions, the deep crust reaches an intermediate
composition of 61.2 ? 7.3 wt.% SiO2 at about half its total thickness. At the base of the
continental crust, bulk composition is as mafic as 53.8 ? 3 wt.% SiO2.
An increasingly mafic composition in the crust with depth has three significant im-
plications. First, there are no universal, sharp compositional changes within the crust, and
162
we should therefore move away from thinking of the continental crust as a three-layer
ensemble. We have enough technology and data to in most cases to take a more gradual
composition change within the crust into consideration. Second, The lowermost parts of
the continental crust are, by and large, mafic in nature. This means that relamination of
sediments onto the base of the crust is likely not a dominant crustal evolutionary pro-
cess, though it is currently impossible to differentiate between mafic sediments and mafic
restites. Third, the mafic composition of the deep continental crust correlates to lower
abundances of HPEs in that portion of the crust. This means that for geoneutrino stud-
ies, the uppermost continental crust remains the major focus of chemical modeling, as
changes in the HPE-depleted deep crust do not affect the lithospheric correction to the
geoneutrino signal as much as the abundance of HPEs in the uppermost crust.
6.2 Future geoneutrino project locations
The under-development Jiangmen Underground Neutrino Observatory (JUNO, 20
kton) experiment in Jiangmen, Guangdong province, China (An et al., 2016) is currently
being built. This detector will be 20 times bigger than the any existing detector. It will
begin detecting geoneutrinos in 2023 or 2024. Because of its large size, one year of JUNO
data will present us with more detected geoneutrino events than what has been collected
by all the previous experiments. The current Slithosphere predictions for the for the JUNO
detector differ significantly (Gao et al., 2020; Strati et al., 2015). A more accurate and
precise estimate of the distribution of K, Th, and U in the continental lithosphere sur-
rounding the JUNO detector is needed. Implications for these findings go far beyond
163
this immediate applications of neutrino geosciences for southeastern China in particular.
An understanding of the continental lithosphere in this region will constrain the crustal
growth history and architecture of southeastern China, and the nature of its granitic mag-
matism, geothermal energy, and ore potential.
An under development detector adjacent to the Himalayas, at the China Jinping
Underground Laboratory (CJPL), poses unique challenges for crustal modeling. The
principle goals of this instrument will be detecting geoneutrinos and solar neutrinos,
simultaneously revealing the secrets of the fuel powering both the Earth and the Sun.
At 2.4 kilometers beneath Mount Jinping and 1000 km away from the closest commer-
cial nuclear reactor, CJPL is the deepest of the world?s underground laboratories, shield-
ing it against unwelcome background signals from reactor antineutrinos and atmospheric
muons. Sited on the eastern slope of the Tibetan plateau, Jinping will sense geoneutrinos
from the planet?s thickest crust. The Sichuan region of southern China is geologically
one of the most active regions of the world with deadly earthquakes, some of the fastest
moving crust (5 - 20 mm/yr), and nearly unrivaled topographic gradients (10%) between
the Sichuan basin and the Tibetan plateau. The composition of the Himalayan crust and
Tibetan Plateau remains highly debated, and global velocity models conflict with local
assessments, leaving geoscientists with competing hypotheses for the strength and defor-
mation mechanisms of the region?s crust (Jackson, 2002).
Last but not least, teams have proposed to remove the issue of the continental crust
altogether by building an ocean bottom detector (OBD) (Sakai et al., 2021). Because the
oceanic crust is thin and depleted in HPEs, 70% of the OBD signal would come directly
from mantle geoneutrinos, with minimal crustal corrections needed. A team of scientists
164
and engineers have proposed working with JAMSTEC to build a portable prototype detec-
tor that would be moved to different ocean bottom locales every few years. This project
is in its infancy but represents a massive feat in engineering.
165
Appendix A: Supporting Information for Chapter 2
A.1 Full electron antineutrino flux equation
Table S1 explains the meaning of each symbol and its units.
dN(E ???e,~r) NA? dn(E?? ) a(~r)?(~r?)= ? ?P (E?e) e Pee (E??e , |~r ? ~r?|)d~r? (A.1)d(E??e) ? d(E??e) ? 4?|~r ? ~r?|2
Symbol Description Units
dN(E??e ,~r)
d(E ) ??e detection spectrum ???? ee
? 1032 protons x 3.154 x 107s x 100%* proton? s
NA Avogadro?s number atommol
? Decay constant decays?atom
? Atomic mass kgmol *detector
2
?P (E?e) ??e cross-section (function of E??e)
m
proton
dn(E??e) ?? emission spectrum ??ed(E??e) e decay
Pee (E??e , |~r ? ~r?|) Oscillation probability (function of E??e) unitless
a(~r) Concentration of radionuclide in cell kgkg
?(~r?) Density of rock in cell kgm3
|~r ? ~r?|2 Distance from cell to detector m
size and efficiency normalization factor
Table A.1: Heat production and geoneutrino flux results
166
A.2 Heat production from K, Th, and U decay
Radionuclide Mole Fraction (%) ? (a?1) Q(MeV) Q(pJ)
232Th 100 4.916 x 10?11 42.646 6.8326 *Total
235U 0.72049 9.8531 x 10?10 46.397 7.4336
238U 99.2740 1.5513 x 10?10 51.694 8.2823
40K 0.01167 5.491 x 10?10* 1.331* 2.132*
from all 40K decay modes
Table A.2: Radionuclide heat production
167
Appendix B: Supporting Information for Chapter 3
B.1 Eu Anomalies
All amphibolite, granulite, and eclogite facies lithologies have similar median Eu
anomalies (? EuSm ?Gd ). While granulite facies lithologies tend to show more variation inN N
Eu anomalies (Figure B.1), the median Eu/Eu* values for all of our groups fall between
0.89 and 1.2. The distributions of Eu/Eu*, shown in Figure B.1, are skewed towards the
right, moreso for granulite facies than for other facies, favoring Eu/Eu* values around 1
but reaching as high as 2-3. These values of Eu/Eu* < 1 (which translate to negative
median Eu anomalies) are in line with previous estimates for the middle and lower crust
and could indicate voluminous crustal recycling (Tang et al., 2017). The lack of strong,
positive Eu anomalies in the deep crust counters the argument that the deep (or lower)
crust is complementary to the upper crust, which also has a negative Eu anomaly (Taylor
and McLennan, 1985).
168
600 300
Amphibolite Granulite 
400 Facies Lithologies 200 Facies Xenoliths
200 100
N = 2412 N = 633
0 0
0 1 2 3 4 0.5 1 1.5 2 2.5 3 3.5 4
0.91 Eu/Eu* 1.0 Eu/Eu*
300 200
Post Archean Granulite Archean Granulite 
Facies Terrains 150 Facies Terrains
200
100
100 N = 597 N = 400
50
0 0
0 1 2 3 4 0 1 2 3 4
0.96 Eu/Eu* 0.99 Eu/Eu*
6
10
Eclogite Eclogite 
8 FaciesXenoliths 4 Facies Terrains
6
4 2
2 N = 27 N = 8
0 0
0 1 2 3 4 0 1 2 3 4
1.2 Eu/Eu* 0.89 Eu/Eu*
Figure B.1: The right-skewed Eu/Eu* distributions of our datasets all find a median value
close to 1, which is 15 - 30% less than previous estimates. Because of the asymmetry in
the distributions, both granulite facies xenoliths and terrains have similar values, though
the more felsic Archean granulite facies lithologies have a higher percentage of Eu/Eu*
values >1 than their more mafic Post Archean counterparts.
169
# of Samples # of Samples # of Samples
# of Samples # of Samples
# of Samples
B.2 Fluid Mobile Elements
B.2.1 K, Rb, and Cs
The transition from amphibolite to granulite facies can be considered, for the major
elements, an isochemical dehydration reaction, but this is not the case for the most fluid
mobile elements: K, Rb, and Cs. Consistent with the observations of (Rudnick and Pres-
per, 1990), K2O/Rb ratios in amphibolite and granulite facies lithologies are negatively
correlated with K2O content, especially at K2O <1.2 wt.%, indicating Rb depletion rel-
ative to K2O. We restrict our analysis of K2O/Rb to compositions to rock with > 55
wt.% SiO2 because the K2O/Rb ratios of unmetamorphosed basalts are highly variable,
making it difficult to evaluate Rb depletion in mafic lithologies (Rudnick and Presper,
1990). K2O/Rb ratios reach a maximum at about 1 in most of the metamorphic datasets.
Although we omit them, mafic lithologies follow the same trend, though lower K2O val-
ues are reached. As suggested by various studies, this trend likely reflects a mineralogical
control on the partitioning of Rb and K between minerals and a fluid phase (Fowler, 1986;
Rudnick et al., 1985). Low concentrations of K2O and Rb, and high K2O/Rb ratios, may
reflect igneous processes rather than metamorphic processes (Van Calsteren et al., 1986).
However, several mafic granulite facies and amphibolite facies lithologies have high K/Rb
ratios compared to basalts, which suggests that they have experienced some Rb depletion
during metamorphism (Stosch et al., 1986). The similarities between the amphibolite and
granulite facies K/Rb are striking, because high K2O/Rb was thought to be the provi-
dence of dehydration metamorphism. Partial dehydration of amphibolite facies litholo-
170
90
Fluid Mobile Amphibolite Facies Lithologies
80 Elements - Medians Granulite Facies Xenoliths
Post Archean Granulite Facies Terrains
70 Archean Granulite Facies Terrains
60 Eclogite Facies Xenoliths
Eclogite Facies Terrains
50
40
30
20
10
0
Cs Rb Ba U K2O Pb Sr La Ce Pr Nd
Figure B.2: Fluid mobile element concentrations from amphibolite and granulite facies
lithologies are comparable with previous estimates for the lower crust. Eclogite facies
lithologies are up a half an order of magnitude more depleted. Variable labile element
depletions cause the amphibolite facies lithologies to overlap with the granulite facies
terrains, suggesting that either the amphibolite facies samples have experienced some
fluid depletion or that the behavior of the labile elements is caused by an igneous process
and not metamorphic dehydration. The black line and gray shaded region surrounding it
is the Rudnick and Gao (2014) lower crustal composition ? 15%.
gies might lead to high K2O/Rb ratios while still leaving behind enough hydrous mineral
assemblages to classify the rock as amphibolite. Future studies should explore the quan-
tification of dehydration as it relates to K2O/Rb ratios. Alternatively, metamorphosing
gabbro to the amphibolite facies can produce uneven increases in K and Rb (Field and
Elliott, 1974), leading to lower K2O/Rb ratios. Low K2O/Rb ratios in amphibolites could
also be inherited from the retrograde metamorphism of granulite with low K2O/Rb (Field
and Clough, 1976).
171
P.M. Normalized Abundance
Amphibolite Facies Granulite Facies
6 Lithologies 6 Xenoliths
4 4 0.1
2 0.2 2
0 0
-2 0 -2 0
-5 0 5 -5 0 5
Ln(Th/U) Ln(Th/U)
Post Archean Granulite Archean Granulite 
6 Facies Terrains 6 Facies Terrains
4 4 0.1
2 0.1 2 0.05
0 0
-2 0 -2 0
-5 0 5 -5 0 5
Ln(Th/U) Ln(Th/U)
Eclogite Facies Eclogite Facies
6 Xenoliths 6 Terrains 0.12
4 0.5 4 0.1
0.08
2 2 0.06
0 0 0.04
-2 0 -2 0.02
-5 0 5 -5 0 5
Ln(Th/U) Ln(Th/U)
Figure B.3: La/Th vs. Th/U in natural log space. Color indicates the data point density.
High La/Th values correlate with low Th/U concentrations, indicating that low La/Th is
due to depletions in Th instead of enrichments in La.
172
Ln(La/Th) Ln(La/Th) Ln(La/Th)
% Data Density % Data Density % Data Density
Ln(La/Th) Ln(La/Th) Ln(La/Th)
% Data Density % Data Density % Data Density
B.2.2 La/Th and Th/U
Granulite facies lithologies and eclogite facies terrains show greater Th and U de-
pletion than amphibolite facies lithologies or eclogite facies xenoliths. While the granulite
and amphibolite facies lithologies show comparable amounts of U loss, indicated by el-
evated Th/U and low La/Th values, the granulite facies lithologies can reach almost an
order of magnitude greater La/Th values with low Th/U, indicating a loss of both U and
Th. U can be depleted through loss of water due to its redox sensitivity (U6+ is fluid
mobile; U4+ is not), though both U and Th depletions typically indicate a lack of U and
Th-bearing accessory phases (Fowler, 1986; Rudnick et al., 1985). Archean granulite
facies terrains reach higher levels of U depletion than Post-Archean terrains, a curious
observation since many Archean terrains contain monazite, a Th and U-bearing accessory
phase (Rudnick and Presper, 1990). Eclogite facies lithologies on the whole have Th/U
ratios around 2 and a wider range (0.1 to 20) in La/Th.
B.2.3 Alkalis and Alkaline Earth Metals, in General
The alkali and alkaline earth metals are, generally, large ion lithophiles due to their
low charges and high ionic radii. The abundance of these fluid mobile elements becomes
more variable as ionic radius increases, with Rb and Cs having the greatest variability
and greatest fluid mobility. Samples that lack K-feldspar typically display depletions in
K, Rb, and Cs while Na and Ca remain rather remarkably invariable outside of their felsic
and mafic modes. Ba is also quite variable with relative uncertainties ranging from 100
to 200% (similar to Rb, Cs, and La, though Ba is more normally distributed than those
173
three). Many of the amphibolite and granulite facies lithologies are also metaluminous,
though our samples span the range of peraluminous to peralkaline.
As alluded to in Section B.2.3, the alkali metals show considerable variability. K,
Rb, and Cs increase in skewness as their ionic radius increases, with Cs being the largest
and most labile. Rb is also incompatible and fluid mobile. We suspect much of the
variation in K stems from the presence or absence of K-feldspar. Lithium also shows a
moderate to high level of variability in amphibolite and granulite facies, which could be
the result of Li?s fluid mobility. Ba and Sr, also incompatible and fluid mobile elements,
vary more than their smaller radii counterparts, Mg and Ca, but not nearly as much as the
alkalis.
B.3 High Field-Strength Elements
Eclogite facies lithologies display the most fractionation of the generally insoluble
high field strength elements (HFSEs). Of the HFSEs, Ta and Mo display the largest spread
and highest uncertainties for most of our lithologies. There is not enough data to plot Mo
or Ta for eclogite facies lithologies. Once more Mo data are reported, paired spikes in
both Mo and Nb content in eclogites could indicate that Mo and Nb, which have similar
ionic radii, are both being retained in rutile (Rudnick et al., 2000). With 64 data points,
granulite facies xenoliths? elevated Mo is more of a conundrum. High abundances of
rutile are not expected for granulite facies xenoliths (Rudnick and Presper, 1990), and the
xenoliths do not have a complementary Nb spike. Since titanite (sphene), ilmenite, and
magnetite typically host the greatest abundance of Mo in crustal rocks (Greaney et al.,
174
25
Amphibolite Facies Lithologies High Field Strength 
Granulite Facies Xenoliths
Post Archean Granulite Facies Terrains Elements - Medians
20 Archean Granulite Facies Terrains
Eclogite Facies Xenoliths
Eclogite Facies Terrains
15
10
5
0
TiO2 Zr Hf Nb Ta Mo W
Figure B.4: As with the fluid mobile elements, the high field strength element concen-
trations are broadly in agreement with previous estimates. Granulite facies xenoliths and
eclogite facies lithologies show elevated concentrations of Mo, partly due to lack of data
but also potentially because of the abundance of rutile, titanite, ilmenite, and magnetite in
these predominantly mafic samples. Eclogite facies lithologies have negative Ta spikes,
which are also potentially caused by the presence of rutile. The black line and gray shaded
region surrounding it is the Rudnick and Gao (2014) lower crustal composition ? 15%.
2018), along with rutile, it is possible that these minerals exist in higher abundances in
granulite facies xenoliths and/or the deep crust than previously thought.
The data also show variable depletions in Ta compared to (Rudnick and Gao, 2014)?s
value. Nb/Ta values also exceed previous lower crustal estimates (Rudnick and Gao,
2014), mimicking more the middle crust?s ratio, suggesting that the deep crust might have
a super-chondritic Nb/Ta ratio. Plotting Nb vs. Ta (Figure B.7) yields a linear log-log plot
for amphibolite, granulite, and eclogite facies lithologies. In many cases, we do not have
high resolution measurements for Ta concentrations at the sub-ppm level, evidenced by
the diagonal lines of data on the Nb vs. Nb/Ta plots (ratios dominated solely by changing
Nb) and horizontal array of dots on the Ta vs. Nb/Ta plots. Concentrations which have
175
P.M. Normalized Abundance
been rounded to the nearest 100 ppb or ppm when reported. Eclogite facies xenoliths and
terrains have notably different Nb/Ta ratios: 15.6 ? 1.9 and 22.4 ? 10.2, respectively.
Studies have hypothesized that Nb can be sequestered by rutile in eclogite formation, but
Ta is not expected to increase (Rudnick et al., 2000). Eclogitic material in down-going
crustal slabs could be a complementary reservoir to the apparent Nb depletion in the rest
of the observable bulk silicate Earth.
B.4 Transition Metals
Transition metal data exists in more abundance for amphibolite facies lithologies
than for granulite or eclogite. Trends in the first row transition elements are discussed
in depth in Section 6 of the main text, but we wanted to briefly mention them in context
with the other elements here. The dearth of data on moderately and highly siderophile
elements is because they exist in such low abundances in the silicate portion of Earth that
only recently have we been able to obtain high precision measurements.
Ti, Zr, and Hf have similar variability, with more mafic lithologies having lower
absolute concentrations of Zr. Eclogite facies lithologies have the lowest concentrations
of all three elements, a curious finding considering that eclogite facies lithologies tend
to have an abundance of rutile (FeTiO2). Though they track similarly, what controls the
abundance of Ti family elements is unclear.
The Cr family, on the other hand, shows much greater variability, perhaps due to
Cr concentrating in garnets, spinels, chromite, or even olivine. Olivine fractionation in
the source melt could also explain the variations also seen in Ni (Arndt, 1994) among
176
amphibolite and granulite facies lithologies. However, Mo?s slightly chalcophile behavior
(Barnes, 2016) could be lending the Cr-family a more discontinuous and variable look
than, for example, the Sc, Ti, or Mn families. The Ni and Cu display more variable
abundances as well, especially in amphibolite facies lithologies. Their behavior could be
attributed to S-affinity and/or the presence of Ni- or Cu-bearing accessory mineral phases
(e.g., olivine, dioptase). Weathering has been shown to (Arndt, 1994; Polat et al., 2002),
making it more likely that the spread of Cr and Ni abundances is controlled by igneous
factors.
A stepped increase in variability is visible from the top to the bottom of the V
family for both amphibolite and granulite facies lithologies. This trend, notable in other
parts of the periodic table is discussed in Section 6 of the main text. Of course, limited
sample numbers may be causing the fluctuation in third-row transition metal abundances.
More data is always welcomed to better constrain the abundances and behaviors of the
transition metals.
Not much can be said about the elements towards the bottom and right of the peri-
odic table for granulite facies lithologies due to a lack of data, though we can devise a few
hypotheses to explain the increased spread in abundances of heavier elements compared
to their lighter counterparts. Especially in heavy elements that do not lack for data, such
as Ta, Pb, Th, and U, the continuous but wide distribution of data could be caused by the
vertical fractionation of these elements within the continental crust. Vertical fractionation
of elements within the crust has been discussed in depth for decades (Rudnick and Gao,
2014; Sighinolfi, 1971; Taylor and McLennan, 1985). One means for testing this hypoth-
esis is comparing trace element abundance to inferred maximum pressure conditions -
177
the higher the pressure, the deeper from within the crust the sample originated. Sadly,
most of our data does not have precise pressures and/or temperatures associated with it,
though some studies have explored these composition-pressure relationships with quan-
titative methods (Barnhart et al., 2012; Bohlen and Mezger, 1989; Rudnick and Jackson,
1995).
Cr, Ni, and Cu break the homogeneity of the first row transition elements. Cr and
the two other elements it shares a family with, Mo and W, are more variable than any
other transition metals, though we must consider that Mo and W have fewer analyses
than many of the others (89 and 42 data points, respectively, compared to the 1000?s of
analyses done on first row transition elements and Y, Zr, Nb, Hf, and Ta). The Cr family
takes on an abnormal valence electron structure (4s13d5). This causes Cr to have a range
of valence states, though it commonly exists in a 3+ valence state as a substitution in
garnet (uvarovite, Ca3Cr2(SiO4)3). Mo and W are high field strength elements with high
valences and large radii, which make it harder for them to substitute into common min-
erals and may contribute to their variability. Mo is typically hosted in Ti-oxides though
displays much more variation than Ti.
The Ni and Cu families might also be special cases, not because of their valence
structure, but because of their metal and sulfide affinity. The variability in S content
may cause fluctuations chalcophile-siderophile elements (CSEs), including Ni and Cu.
Of the other CSE?s, Zn, a moderately chalcophile element (Jenner, 2017), is much more
normally distributed than either Ni or Cu. There are too few measurements on Au and Ag
to determine whether their apparent variation is caused by CSE (Jenner, 2017) behavior
or if they are under-sampled. The Cu and Ni distributions have far more outliers than any
178
other first row transition metals. Cross-referencing the original data sources show that >
1100 amphibolite and > 900 granulite measurements come from samples with 0.01 - 0.1
wt.% Cu or Ni. The spread in Cu and Ni compositions could be due to cumulate formation
and recycling processes, whereby CSE?s can be partitioned by magnetite-induced sulfide
fractionation in lower crustal settings. While we expect enrichment in CSE?s in these
lower crustal cumulates, delamination of such material could also cause depletion and
contribute to the bulk crust?s low Cu concentrations.
B.5 Halogens
Limited data is available for the halogens in these medium and high grade meta-
morphic lithology datasets. F and Cl are the only elements measured for amphibolite
and granulite facies lithologies. No halide data is available for eclogite facies lithologies
in our dataset. Halides can be transported in fluids or incorporated into select mineral
phases, such as Cl-rich biotite and amphibole, apatite, and fluorite (Kusebauch et al.,
2015). High solubility in water means these elements are often mobilized by fluid fluxing
and can often be found as chemical weathering and alteration byproducts in granulites,
despite granulites being nominally anhydrous (Banks et al., 2000; Kusebauch et al., 2015;
Markl et al., 1997). Stable Cl isotopes have been studied in granulites to track mantle par-
ticipation in silicate Earth?s Cl cycle and fluid infiltration processes (Markl et al., 1997).
However, the volatility of halogens and their incompatibility in many melt systems has
resulted in relatively low halide abundances in the deep crust outside of fluid processes.
179
B.6 Supplementary Tables
Table B.1: Amphibolite Facies Lithologies
Mean Median Geometric ? Mean STD IQR Geo ? STD N (filtered) N (original)
Mean STD
Li 17.9 15.0 14.8 17.9 10.7 14.5 1.89 10.6 455 727
Be 1.57 1.44 1.21 1.57 1.02 1.10 2.22 1.08 212 374
B 14.0 9.00 8.44 14.0 14.5 15.6 2.81 13.1 184 362
N - - - - - - - - 0 0
F 435 399 60 435 351 512 48.2 746 180 269
S 41.2 22.0 25.2 41.2 46.1 44.0 2.6 38.2 93 292
Cl 51.0 29.3 1.2 51.0 62.2 86.7 83.3 114 81 196
Sc 23.1 21.0 18.3 23.1 13.6 24.4 2.11 15.2 1360 3160
V 152 134 104 152 108 192 2.71 125 1840 3690
Cr 119 81.0 63.5 119 121 151 3.57 124 1820 4200
Co 30.1 29.9 22.2 30.1 19.1 33.1 2.42 22.4 1420 3480
Ni 51.8 39.7 31.2 51.8 47.1 62.0 3.08 49.0 1810 4290
Cu 41.3 30.0 25.8 41.3 36.4 49.5 2.89 37.6 1290 2970
Zn 80.7 78.0 73.0 80.7 34.6 49.5 1.60 35.7 1550 3310
Ga 18.1 18.0 17.7 18.1 3.66 4.32 1.23 3.69 1120 2320
Ge - - - - - - - - 0 0.0
As 2.9 1.3 1.4 2.9 4.2 2.4 3.1 3.2 200 368
Se 77.5 53.0 57.1 77.5 63.9 69.0 2.2 57.8 136 158
180
Br - - - - - - - - 0 0
Rb 59.1 43.8 35.1 59.1 51.8 80.0 3.16 56.5 1910 4030
Sr 245 201 201 245 151 204 1.91 149 2100 4730
Y 24.3 22.5 21.9 24.3 11.0 15.0 1.60 11.0 1970 4460
Zr 133 123 115 133 69.5 97.0 1.77 70.9 2070 4650
Nb 8.61 7.20 6.90 8.61 5.47 7.93 2.02 5.55 1790 3790
Mo 1.29 0.520 0.670 1.29 1.56 1.48 3.10 1.37 208 354
Ru - - - - - - - - 0 0
Rh 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
Pd 1.71 0.85 0.89 1.71 2.20 1.70 3.09 1.81 118 218
Ag 47.7 48.0 34.2 47.7 22.1 25.0 3.4 37.1 150 200
Cd 66.2 60.0 49.1 66.2 36.4 58.0 3.0 49.0 146 192
In 0.0712 0.0712 0.0712 - 0.0 0.0 1.0 - 1 28
Sn 3.08 1.60 1.85 3.08 3.58 2.75 2.67 2.90 215 377
Sb 0.30 0.20 0.21 0.30 0.27 0.32 2.38 0.25 253 497
Te 0.08 0.05 0.05 0.08 0.11 0.05 2.76 0.08 8 14
I - - - - - - - - 0 0
Cs 1.8 1.2 1.1 1.8 1.8 2.0 2.9 1.7 752 1680
Ba 399 330 268 399 306 486 2.7 337 2030 4400
La 22.3 18.1 15.8 22.3 16.7 25.5 2.5 17.5 1820 3920
Ce 44.4 36.5 33.1 44.4 31.8 46.4 2.3 32.7 1790 3870
Pr 5.6 4.7 4.3 5.6 3.8 5.4 2.1 3.9 1110 2120
Nd 21.6 18.3 17.7 21.6 13.5 18.0 1.9 13.3 1710 3500
Sm 4.6 4.1 4.0 4.6 2.4 3.0 1.7 2.3 1630 3470
181
Eu 1.1 1.1 1.0 1.1 0.5 0.6 1.5 0.5 1640 3470
Gd 4.4 3.9 3.9 4.4 2.1 2.7 1.6 2.1 1390 2700
Tb 0.7 0.7 0.7 0.7 0.3 0.4 1.6 0.3 1250 2860
Dy 4.3 3.9 3.8 4.3 2.0 2.8 1.6 2.0 1150 2330
Ho 0.9 0.8 0.8 0.9 0.4 0.6 1.6 0.4 1050 2100
Er 2.5 2.3 2.2 2.5 1.2 1.7 1.7 1.2 1150 2250
Tm 0.4 0.4 0.3 0.4 0.2 0.2 1.6 0.2 821 1960
Yb 2.4 2.2 2.1 2.4 1.1 1.5 1.7 1.1 1740 3740
Lu 0.4 0.3 0.3 0.4 0.2 0.2 1.7 0.2 1550 3240
Hf 3.9 3.4 3.3 3.9 2.3 3.1 1.9 2.3 1320 2830
Ta 0.7 0.5 0.5 0.7 0.5 0.7 2.3 0.5 1120 2540
W 1.0 0.4 0.5 1.0 2.0 0.8 2.9 1.1 182 454
Re - - - - - - - - 0 6
Os - - - - - - - - 0 6
Ir 0.2 0.2 0.2 0.2 0.1 0.1 1.8 0.1 6 19
Pt 1.7 0.7 0.9 1.7 2.0 1.9 3.1 1.8 120 217
Au 2.0 0.8 1.0 2.0 2.7 1.9 3.0 2.2 160 271
Hg 0.0 0.0 0.0 0.0 0.0 0.0 2.2 0.0 4 12
Tl 0.6 0.5 0.3 0.6 0.5 0.8 3.9 0.6 20 60
Pb 13.2 11.7 10.5 13.2 8.3 11.5 2.1 8.6 1260 2490
Bi 0.1 0.1 0.1 0.1 0.1 0.1 2.2 0.1 143 200
Th 5.6 3.7 3.0 5.6 5.4 8.2 3.6 5.9 1600 3330
U 1.4 1.0 0.9 1.4 1.1 1.5 2.6 1.1 1390 2700
182
Table B.2: Granulite Facies Xenoliths
Mean Median Geometric ? Mean STD IQR Geo ? STD N (filtered) N (original)
Mean STD
Li 11.8 6.9 9.2 11.8 10.5 6.5 1.9 8.0 12 154
Be 0.5 0.5 0.5 0.5 0.0 0.0 1.1 0.0 2 20
B - - - - - - - - 0 0
N - - - - - - - - 0 0
F - - - - - - - - 0 35
S 143 140 89.8 143.0 110.8 220.6 2.85 129.8 4 67
Cl 151.0 151.0 151.0 - 0.0 0.0 1.0 - 1 28
Sc 28.0 28.0 26.5 28.0 9.0 11.9 1.4 9.1 97 863
V 186.0 186.0 166.0 186.0 79.4 100.0 1.7 88.1 92 1030
Cr 210.0 168.0 165.0 210.0 145.0 162.0 2.1 140.0 120 1140
Co 48.9 46.8 46.5 48.9 15.1 23.5 1.4 15.5 87 759
Ni 103.0 100.0 91.6 103.0 46.7 65.0 1.7 48.3 100 1080
Cu 43.4 37.8 36.0 43.4 27.4 33.1 1.9 25.9 72 774
Zn 77.6 81.1 73.0 77.6 25.2 43.3 1.4 26.9 72 751
Ga 17.1 17.3 16.8 17.1 3.1 4.6 1.2 3.2 48 391
Ge - - - - - - - - 0 0
As - - - - - - - - 0 0
Se - - - - - - - - 0 0
Br - - - - - - - - 0 0
Rb 15.5 10.6 10.0 15.5 13.0 20.2 2.8 13.6 112 1180
183
Sr 482.0 465.0 437.0 482.0 207.0 315.0 1.6 210.0 121 1280
Y 19.9 19.0 17.3 19.9 9.9 16.6 1.8 10.5 102 1030
Zr 91.3 83.3 70.5 91.3 58.9 88.7 2.2 63.1 105 1060
Nb 8.8 7.0 7.0 8.8 5.8 7.7 2.0 5.8 99 888
Mo 2.3 1.9 2.1 2.3 1.2 0.8 1.6 1.1 6 62
Ru - - - - - - - - 0 0
Rh 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
Pd 5.54 5.54 5.54 - 0.00 0.00 1.00 - 1 5
Ag - - - - - - - - 0 0
Cd - - - - - - - - 0 0
In - - - - - - - - 0 0
Sn 1.70 1.58 1.62 1.70 0.52 0.96 1.39 0.54 9 77
Sb - - - - - - - - 0 39
Te - - - - - - - - 0 0
I - - - - - - - - 0 0
Cs 0.752 0.390 0.316 0.752 0.960 0.871 4.280 0.901 44 340
Ba 470 393 358 470 330 428 2 334 107 1120
La 14.1 11.6 11.5 14.1 8.6 12.9 2.0 8.7 118 1080
Ce 30.0 27.0 25.0 30.0 17.2 27.4 1.9 17.6 119 1090
Pr 3.6 3.5 3.0 3.6 2.1 3.3 2.0 2.2 63 535
Nd 15.8 14.7 13.7 15.8 8.0 12.3 1.8 8.4 115 1080
Sm 3.8 3.6 3.4 3.8 1.7 2.3 1.6 1.7 114 1070
Eu 1.2 1.3 1.2 1.2 0.4 0.6 1.4 0.4 108 983
Gd 3.8 3.8 3.4 3.8 1.7 2.4 1.7 1.8 79 705
184
Tb 0.6 0.6 0.5 0.6 0.3 0.3 1.6 0.3 93 778
Dy 3.7 3.6 3.2 3.7 1.8 3.0 1.8 2.0 73 656
Ho 0.7 0.6 0.6 0.7 0.3 0.6 1.7 0.3 63 538
Er 2.0 1.9 1.7 2.0 1.1 1.9 1.9 1.2 71 645
Tm 0.3 0.3 0.2 0.3 0.2 0.3 2.0 0.2 41 344
Yb 1.9 1.7 1.6 1.9 1.0 1.6 1.7 1.0 109 973
Lu 0.3 0.2 0.2 0.3 0.2 0.3 1.7 0.1 99 815
Hf 2.5 2.1 2.0 2.5 1.6 1.3 1.8 1.4 82 680
Ta 0.8 0.6 0.6 0.8 0.5 0.7 2.1 0.5 62 444
W 2.4 1.9 1.8 2.4 1.7 2.6 2.2 1.7 4 30
Re - - - - - - - - 0 1
Os 0.2 0.2 0.2 - 0.0 0.0 1.0 - 1 4
Ir 0.3 0.3 0.3 - 0.0 0.0 1.0 - 1 5
Pt 2.5 2.5 2.5 - 0.0 0.0 1.0 - 1 6
Au - - - - - - - - 0 3
Hg - - - - - - - - 0 0
Tl - - - - - - - - 0 0
Pb 5.6 4.5 4.6 5.6 3.6 4.9 2.0 3.5 77 690
Bi - - - - - - - - 0 0
Th 1.8 0.8 0.9 1.8 2.4 1.5 3.0 1.9 94 750
U 0.4 0.2 0.3 0.4 0.5 0.3 2.5 0.4 77 604
185
Table B.3: Post-Archean Granulite Terrains
Mean Median Geometric ? Mean STD IQR Geo ? STD N (filtered) N (original)
Mean STD
Li 7.9 7.2 7.3 7.9 3.2 4.7 1.5 3.1 6 74
Be 2.0 1.6 1.9 2.0 0.7 1.1 1.4 0.6 3 36
B - - - - - - - - 0 0
N - - - - - - - - 0 0
F - - - - - - - - 3 9
S 300.0 300.0 300.0 - - - - - 1 10
Cl 100.0 100.0 100.0 - 0.0 0.0 1.0 - 2 6
Sc 25.4 25.6 22.7 25.4 10.5 14.3 1.7 11.9 50 635
V 163.0 157.0 145.0 163.0 68.9 90.9 1.7 78.9 54 1030
Cr 198.0 132.0 126.0 198.0 209.0 183.0 2.6 176.0 81 1130
Co 32.3 31.8 27.8 32.3 16.5 21.6 1.8 17.2 41 677
Ni 80.4 59.0 55.7 80.4 78.4 64.8 2.3 65.5 82 1110
Cu 34.2 29.0 28.2 34.2 20.4 27.7 1.9 20.6 45 748
Zn 81.1 77.7 76.4 81.1 29.4 37.6 1.4 27.7 67 949
Ga 18.6 18.0 18.4 18.6 2.8 3.9 1.2 2.8 38 774
Ge - - - - - - - - 0 0
As - - - - - - - - 0 0
Se - - - - - - - - 0 0
Br - - - - - - - - 0 0
Rb 65.2 59.0 42.6 65.2 48.1 79.0 3.0 56.9 127 1340
186
Sr 280.0 267.0 240.0 280.0 144.0 190.0 1.8 152.0 134 1420
Y 28.7 24.2 24.9 28.7 15.8 17.3 1.7 14.9 67 1170
Zr 166.0 136.0 127.0 166.0 114.0 147.0 2.2 118.0 90 1330
Nb 9.9 9.5 7.7 9.9 6.2 7.9 2.3 6.9 63 948
Mo 2.4 3.2 1.8 2.4 1.3 2.1 2.3 1.7 3 37
Ru - - - - - - - - 0 0
Rh 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
Pd - - - - - - - - 0 0
Ag - - - - - - - - 0 0
Cd - - - - - - - - 0 0
In - - - - - - - - 0 0
Sn 3.6 2.9 3.0 3.6 2.6 2.7 1.8 2.3 5 223
Sb 0.1 0.1 0.0 0.1 0.1 0.1 2.9 0.1 6 60
Te - - - - - - - - 0 0
I - - - - - - - - 0 0
Cs 1.7 0.7 0.6 1.7 2.0 2.5 6.0 2.3 24 530
Ba 470.0 461.0 360.0 470.0 280.0 447.0 2.4 330.0 83 1260
La 26.8 24.5 20.9 26.8 16.7 24.4 2.2 18.3 63 1050
Ce 55.6 50.5 45.0 55.6 33.3 42.5 2.0 35.0 62 1060
Pr 6.0 6.1 4.3 6.0 3.9 5.4 2.7 4.7 27 588
Nd 25.9 24.8 21.6 25.9 15.3 17.1 1.9 15.1 57 878
Sm 5.2 4.9 4.5 5.2 2.8 3.0 1.8 2.8 55 842
Eu 1.3 1.4 1.2 1.3 0.4 0.7 1.4 0.4 53 836
Gd 4.9 4.4 4.3 4.9 2.6 2.3 1.7 2.4 42 683
187
Tb 0.9 0.7 0.8 0.9 0.5 0.5 1.8 0.5 44 759
Dy 4.3 3.7 3.9 4.3 2.0 2.4 1.6 1.9 32 634
Ho 0.9 0.8 0.8 0.9 0.4 0.6 1.6 0.4 31 607
Er 2.5 2.1 2.3 2.5 1.2 1.3 1.6 1.2 33 642
Tm 0.5 0.4 0.4 0.5 0.2 0.4 1.6 0.2 13 67
Yb 2.5 2.0 2.2 2.5 1.3 1.5 1.7 1.3 61 862
Lu 0.4 0.4 0.3 0.4 0.2 0.2 1.7 0.2 48 802
Hf 4.1 4.1 3.3 4.1 2.3 3.0 2.0 2.5 41 666
Ta 0.7 0.5 0.5 0.7 0.5 0.7 2.4 0.5 29 578
W - - - - - - - - 0 29
Re - - - - - - - - 0 0
Os - - - - - - - - 0 0
Ir - - - - - - - - 0 0
Pt - - - - - - - - 0 0
Au - - - - - - - - 0 0
Hg - - - - - - - - 0 0
Tl - - - - - - - - 0 18
Pb 13.2 11.7 10.4 13.2 8.2 12.1 2.1 8.7 35 728
Bi - - - - - - - - 0 14
Th 9.9 5.4 4.2 9.9 10.2 16.2 4.9 11.7 90 816
U 1.0 0.8 0.7 1.0 0.9 1.0 2.8 0.9 83 732
Units: Pd,W, Re, Os, Ir, Pt, and Au are in ppb. All other elements are in ppm.
188
Table B.4: Archean Granulite Terrains
Mean Median Geometric ? Mean STD IQR Geo ? STD N (filtered) N (original)
Mean STD
Li 15.3 16.3 14.4 15.3 4.6 7.6 1.4 5.0 4 84
Be - - - - - - - - 0 0
B - - - - - - - - 0 0
N - - - - - - - - 0 0
F - - - - - - - - 0 1
S 1772 1772.0 698 1772 2303 3260 4.9 2190 2 5
Cl 90.0 90.0 90.0 - 0.0 0.0 1.0 - 1 6
Sc 16.6 16.1 13.1 16.6 9.8 13.9 2.2 11.0 33 468
V 119.0 115.0 102.0 119.0 62.0 81.0 1.8 64.0 48 1000
Cr 146.0 117.0 104.0 146.0 108.0 170.0 2.4 114.0 67 1060
Co 41.3 35.7 33.6 41.3 25.4 38.1 1.9 25.7 35 604
Ni 72.4 55.0 54.9 72.4 52.1 76.7 2.2 51.7 69 1090
Cu 28.5 24.7 22.4 28.5 17.5 30.3 2.1 18.9 32 736
Zn 66.0 65.9 61.9 66.0 22.0 32.6 1.5 23.6 40 875
Ga 20.2 19.5 19.9 20.2 3.7 4.9 1.2 3.6 27 703
Ge - - - - - - - - 0 0
As 30.0 30.0 30.0 - 0.0 0.0 1.0 - 1 11
Se - - - - - - - - 0 6
Br - - - - - - - - 0 0
Rb 44.2 37.0 23.9 44.2 39.9 63.2 3.7 45.5 89 1340
189
Sr 307.0 295.0 237.0 307.0 185.0 333.0 2.2 212.0 87 1360
Y 18.4 17.8 15.8 18.4 9.4 13.0 1.8 9.9 70 1130
Zr 155.0 137.0 135.0 155.0 86.4 75.7 1.7 79.8 84 1270
Nb 9.2 7.3 7.5 9.2 6.6 5.7 1.9 5.7 66 973
Mo - - - - - - - - 0 2
Ru - - - - - - - - 0 0
Rh 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
Pd - - - - - - - - 0 3
Ag - - - - - - - - 0 0
Cd - - - - - - - - 0 0
In - - - - - - - - 0 0
Sn 2.6 2.5 2.5 2.6 0.7 1.2 1.4 0.8 23 95
Sb - - - - - - - - 0 6
Te - - - - - - - - 0 0
I - - - - - - - - 0 0
Cs 1.8 0.9 0.8 1.8 2.2 2.4 4.2 2.1 38 229
Ba 561.0 559.0 459.0 561.0 316.0 425.0 2.0 346.0 69 1220
La 27.3 20.0 21.8 27.3 20.6 24.5 1.9 17.6 66 993
Ce 49.8 40.0 41.6 49.8 33.8 34.5 1.8 28.9 65 987
Pr 5.0 4.4 4.4 5.0 2.6 2.7 1.7 2.4 39 376
Nd 18.1 16.4 15.8 18.1 9.2 8.6 1.7 9.3 64 594
Sm 3.6 3.3 3.3 3.6 1.5 2.3 1.5 1.5 55 661
Eu 1.1 1.1 1.1 1.1 0.3 0.5 1.4 0.33 53 641
Gd 3.5 3.3 3.2 3.5 1.5 2.2 1.6 1.56 45 455
190
Tb 0.5 0.5 0.5 0.5 0.2 0.3 1.6 0.22 50 571
Dy 3.3 3.0 3.0 3.3 1.2 1.9 1.5 1.28 39 481
Ho 0.6 0.6 0.6 0.6 0.3 0.4 1.7 0.30 40 417
Er 1.8 1.7 1.5 1.8 0.8 0.9 1.8 0.93 42 484
Tm 0.3 0.3 0.3 0.3 0.1 0.1 1.2 0.05 9 86
Yb 1.6 1.6 1.4 1.6 0.8 1.3 1.9 0.91 53 630
Lu 0.3 0.3 0.2 0.3 0.1 0.1 1.8 0.13 35 524
Hf 3.2 3.1 3.0 3.2 0.9 1.4 1.4 0.95 44 377
Ta 0.7 0.4 0.4 0.7 0.9 0.3 2.5 0.61 26 258
W 6.7 6.7 6.7 - 0.0 0.0 1.0 - 1 7
Re - - - - - - - - 0 0
Os - - - - - - - - 0 0
Ir - - - - - - - - 0 7
Pt - - - - - - - - 0 3
Au - - - - - - - - 0 6
Hg - - - - - - - - 0 0
Tl - - - - - - - - 0 0
Pb 10.0 8.0 8.8 10.0 5.4 5.4 1.6 4.77 61 758
Bi - - - - - - - - 0 0
Th 7.4 2.7 2.9 7.4 14.6 5.2 3.7 9.21 73 772
U 1.2 0.5 0.6 1.2 1.7 1.4 3.3 1.28 66 562
Units: Pd,W, Re, Os, Ir, Pt, and Au are in ppb. All other elements are in ppm.
191
Table B.5: Eclogite Facies Xenoliths
Mean Median Geometric ? Mean STD IQR Geo ? STD N (filtered) N (original)
Mean STD
Li 28.7 28.7 28.7 - 0.0 0.0 1.0 - 1 8
Be - - - - - - - - 0 0
B - - - - - - - - 0 0
N - - - - - - - - 0 0
F - - - - - - - - 0 4
S - - - - - - - - 0 0
Cl - - - - - - - - 0 4
Sc 56.2 57.0 55.6 56.2 7.5 13.8 1.15 7.6 3 33
V 320 274 314 320 62.4 120 1.21 61.0 5 60
Cr 626 764 552 626 278 554 1.70 308 5 60
Co 62.7 65.2 61.6 62.7 11.4 20.5 1.21 11.9 4 35
Ni 182 157 174 182 53.5 110 1.34 53.2 5 60
Cu 55.2 55.2 55.1 55.2 4.13 8.26 1.08 4.1 2 31
Zn 72.3 68.9 70.9 72.3 14.3 25.5 1.21 14.0 4 58
Ga 11.4 11.4 11.4 - 0.0 0.0 1.00 - 1 46
Ge - - - - - - - - 0 0
As - - - - - - - - 0 0
Se - - - - - - - - 0 0
Br - - - - - - - - 0 0
Rb 7.4 7.1 5.4 7.4 4.9 11.0 2.4 5.6 5 41
192
Sr 165.0 178.0 157.0 165.0 53.5 72.0 1.4 52.0 7 67
Y 19.6 20.7 19.1 19.6 4.1 7.2 1.2 4.2 7 66
Zr 40.1 41.1 34.4 40.1 20.6 29.3 1.8 21.7 6 65
Nb 10.4 6.3 5.1 10.4 9.9 19.2 3.9 11.3 6 58
Mo - - - - - - - - 0 2
Ru - - - - - - - - 0 0
Rh - - - - - - - - 0 0
Pd - - - - - - - - 0 18
Ag - - - - - - - - 0 0
Cd - - - - - - - - 0 0
In - - - - - - - - 0 0
Sn - - - - - - - - 0 2
Sb - - - - - - - - 0 2
Te - - - - - - - - 0 0
I - - - - - - - - 0 0
Cs - - - - - - - - 0 6
Ba 370.0 239.0 242.0 370.0 350.0 382.0 2.6 323.0 7 67
La 5.6 6.0 4.3 5.6 3.5 6.0 2.2 3.9 7 62
Ce 16.5 13.9 12.3 16.5 10.9 20.7 2.3 12.3 7 62
Pr 2.2 1.9 1.9 2.2 1.3 1.9 1.9 1.3 4 30
Nd 9.7 7.9 7.9 9.7 6.2 9.5 1.9 6.0 6 37
Sm 2.6 2.3 2.5 2.6 1.1 1.0 1.4 1.0 5 34
Eu 1.0 0.8 1.0 1.0 0.2 0.4 1.3 0.2 5 34
Gd 2.8 2.4 2.7 2.8 0.8 1.1 1.3 0.8 4 30
193
Tb 0.4 0.4 0.4 0.4 0.1 0.1 1.2 0.1 5 34
Dy 3.0 3.0 2.9 3.0 0.5 0.9 1.2 0.5 4 30
Ho 0.6 0.6 0.6 0.6 0.1 0.2 1.2 0.1 4 30
Er 1.6 1.7 1.6 1.6 0.2 0.3 1.2 0.2 4 30
Tm 0.3 0.3 0.3 0.3 0.0 0.0 1.1 0.0 2 27
Yb 2.1 1.9 2.0 2.1 0.8 1.3 1.5 0.8 5 34
Lu 0.3 0.3 0.3 0.3 0.1 0.1 1.3 0.1 4 32
Hf 1.5 1.4 1.3 1.5 0.7 1.2 1.6 0.7 4 31
Ta 0.6 0.7 0.6 0.6 0.1 0.1 1.1 0.1 3 29
W - - - - - - - - 0 0
Re - - - - - - - - 0 0
Os - - - - - - - - 0 0
Ir - - - - - - - - 0 18
Pt - - - - - - - - 0 18
Au - - - - - - - - 0 18
Hg - - - - - - - - 0 0
Tl - - - - - - - - 0 0
Pb 2.9 2.9 2.3 2.9 1.7 3.3 2.0 1.8 4 38
Bi - - - - - - - - 0 0
Th 0.4 0.3 0.4 0.4 0.2 0.3 1.7 0.2 5 55
U 0.2 0.2 0.2 0.2 0.2 0.3 1.9 0.1 4 54
Units: Pd,W, Re, Os, Ir, Pt, and Au are in ppb. All other elements are in ppm.
194
Table B.6: Eclogite Facies Terrains
Mean Median Geometric ? Mean STD IQR Geo ? STD N (filtered) N (original)
Mean STD
Li - - - - - - - - 0 2
Be - - - - - - - - 0 0
B - - - - - - - - 0 0
N - - - - - - - - 0 0
F - - - - - - - - 0 2
S - - - - - - - - 0 0
Cl - - - - - - - - 0 1
Sc 37.9 30.3 31.0 37.9 23.2 41.4 1.9 23.4 3 33
V 239.0 244.0 224.0 239.0 75.4 86.5 1.5 84.3 5 14
Cr 167.0 187.0 140.0 167.0 76.6 127.0 2.0 98.2 8 39
Co 38.0 38.1 36.2 38.0 11.5 13.2 1.4 11.7 7 38
Ni 83.7 86.0 77.9 83.7 27.0 35.8 1.5 31.1 8 37
Cu 26.8 17.8 21.7 26.8 18.6 26.5 1.9 17.0 6 13
Zn 108.0 105.0 106.0 108.0 19.8 36.5 1.2 19.7 4 30
Ga 20.3 19.0 19.7 20.3 5.0 9.0 1.3 4.9 3 9
Ge - - - - - - - - 0 0
As - - - - - - - - 0 23
Se - - - - - - - - 0 0
Br - - - - - - - - 0 0
Rb 7.6 4.4 5.5 7.6 6.0 10.2 2.2 5.8 3 35
195
Sr 207.0 183.0 165.0 207.0 118.0 208.0 2.1 134.0 9 45
Y 27.6 26.2 24.8 27.6 12.0 22.0 1.6 12.3 6 13
Zr 89.2 107.0 81.4 89.2 33.1 60.3 1.6 37.6 7 36
Nb 6.5 5.4 6.1 6.5 2.5 4.3 1.4 2.4 3 7
Mo - - - - - - - - 0 2
Ru - - - - - - - - 0 0
Rh - - - - - - - - 0 0
Pd - - - - - - - - 0 0
Ag - - - - - - - - 0 0
Cd - - - - - - - - 0 0
In - - - - - - - - 0 0
Sn - - - - - - - - 0 2
Sb - - - - - - - - 0 25
Te - - - - - - - - 0 0
I - - - - - - - - 0 0
Cs 0.1 0.1 0.1 - 0.0 0.0 1.0 - 1 26
Ba 212.0 135.0 136.0 212.0 188.0 319.0 2.7 189.0 9 39
La 7.9 6.4 6.3 7.9 5.3 9.4 2.0 5.2 4 35
Ce 15.6 14.8 12.9 15.6 8.8 16.9 1.9 9.3 4 35
Pr 0.8 0.8 0.8 - 0.0 0.0 1.0 - 1 4
Nd 6.7 5.8 6.3 6.7 2.5 3.5 1.4 2.3 4 36
Sm 2.5 2.0 2.4 2.5 1.0 1.4 1.4 0.9 4 36
Eu 1.0 1.0 1.0 1.0 0.1 0.1 1.1 0.1 3 34
Gd 5.0 5.0 5.0 5.0 0.2 0.5 1.1 0.2 2 8
196
Tb 0.7 0.8 0.7 0.7 0.2 0.4 1.4 0.2 3 34
Dy 6.3 6.3 6.3 - 0.0 0.0 1.0 - 1 4
Ho 1.3 1.3 1.3 1.3 0.1 0.1 1.0 0.1 2 8
Er 3.8 3.8 3.8 3.8 0.2 0.3 1.0 0.2 2 8
Tm 0.5 0.5 0.5 - 0.0 0.0 1.0 - 1 6
Yb 3.4 3.8 3.0 3.4 1.5 2.5 1.7 1.6 6 40
Lu 0.2 0.2 0.2 - 0.0 0.0 1.0 - 1 31
Hf 1.8 1.8 1.8 - 0.0 0.0 1.0 - 1 32
Ta 0.1 0.1 0.1 - 0.0 0.0 1.0 - 1 29
W - - - - - - - - 0 0
Re - - - - - - - - 0 0
Os - - - - - - - - 0 0
Ir - - - - - - - - 0 0
Pt - - - - - - - - 0 0
Au - - - - - - - - 0 0
Hg - - - - - - - - 0 0
Tl - - - - - - - - 0 0
Pb 2.3 2.3 2.2 2.3 0.7 1.4 1.4 0.7 2 8
Bi - - - - - - - - 0 0
Th 0.2 0.2 0.2 0.2 0.1 0.3 2.1 0.1 2 32
U 0.1 0.1 0.1 - 0.0 0.0 1.0 - 1 32
Units: Pd,W, Re, Os, Ir, Pt, and Au are in ppb. All other elements are in ppm.
197
198
Mean Median Geo ? Mean STD IQR Geo ? STD N(filtered) N(original)
Mean STD
La/Yb 11.6 8.16 7.15 11.6 10.9 13.40 2.88 10.7 1740 3740
Nb/Ta 14.4 14.1 12.6 14.4 5.99 6.33 1.99 7.37 719 1360
Zr/Hf 35.5 35.6 35.0 35.5 6.08 7.70 1.20 6.27 1320 2820
Rb/Cs 48.5 34.0 31.4 48.5 45.4 41.90 2.80 42.7 753 1580
K/Rb 331 294 299 331 160 183.00 1.56 147 1900 3960
La/Th 6.62 5.02 5.30 6.62 4.75 5.42 1.93 4.27 1590 3200
Th/U 4.20 3.65 3.35 4.20 2.67 3.10 2.08 2.73 1380 2600
K/U 15900 12100 11300 15900 13000 14200.00 2.42 12600 1380 2600
Eu/Eu* 0.859 0.884 0.826 0.859 0.229 0.32 1.34 0.241 1372 2674
?Pb 3.89 3.86 3.84 3.89 0.646 0.48 1.17 0.621 - 165
Heat Prod. (nW/kg) 0.470 0.278 0.240 0.470 1.99 0.46 3.16 0.503 490 1467
Heat Prod. (mW/m3) 1.36 0.81 0.70 1.36
Table B.7: Amphibolite Facies Lithologies
199
Mean Median Geo ? Mean STD IQR Geo ? STD N(filtered) N(original)
Mean STD
La/Yb 8.71 7.52 7.29 8.71 5.33 6.37 1.83 5.05 103 957
Nb/Ta 18.0 18.6 15.3 18 7.54 9.1 2.14 9.98 41 424
Zr/Hf 39.9 36.3 36.4 39.9 17.8 11.5 1.56 16.9 62 669
Rb/Cs 73.4 40.6 37.2 73.4 84.1 52 3.83 79 39 317
K/Rb 788 643 653 788 491 685 1.86 470 104 1140
La/Th 27.0 17.1 18.0 27.0 25.0 25.8 2.51 23.0 89 683
Th/U 3.79 3.39 3.05 3.79 2.47 2.67 2.03 2.42 67 571
K/U 46300 36900 32300 46300 36100 43000 2.66 37400 72 571
Eu/Eu* 1.28 1.09 1.21 1.28 0.479 0.38 1.38 0.425 79 703
?Pb 4.17 4.04 4.14 4.17 0.520 0.24 1.11 0.464 - 357
Heat Prod. 0.147 0.0716 0.0861 0.147 0.187 0.11 2.68 0.142 62 79
Heat Prod. (mW/m3) 0.43 0.21 0.25 0.43
Table B.8: Granulite Facies Xenoliths
200
Mean Median Geo ? Mean STD IQR Geo ? STD N(filtered) N(original)
Mean STD
La/Yb 11.7 10.1 8.54 11.7 7.75 8.96 2.77 8.84 41 658
Nb/Ta 14.7 13.9 9.46 14.7 10.4 8.57 4.00 13.0 22 416
Zr/Hf 35.9 35.7 35.1 35.9 7.2 11 1.24 7.46 29 479
Rb/Cs 100 44.6 44.8 100 159 81.4 3.50 116 18 386
K/Rb 423 291 346 423 335 265 1.78 259 109 1090
La/Th 23.7 11.2 12.3 23.7 30.2 18 3.05 25.0 36 777
Th/U 10.1 6.58 6.50 10.1 9.91 10 2.61 8.90 73 564
K/U 37500 31000 27500 37500 32900 32200 2.22 28300 73 564
Eu/Eu* 1.02 1.03 0.94 1.02 0.411 0.46 1.48 0.395 40 673
?Pb 3.99 4.00 3.99 3.99 0.159 0.19 1.04 0.158 - 33
Heat Prod. 0.514 0.407 0.267 0.514 0.490 0.62 4.41 0.544 70 91
Heat Prod. (mW/m3) 1.49 1.18 0.77 1.49
Table B.9: Post Archean Granulite Facies Terrains
201
Mean Median Geo ? Mean STD IQR Geo ? STD N(filtered) N(original)
Mean STD
La/Yb 25.4 16 18.2 25.4 20.3 31.1 2.33 19.8 45 572
Nb/Ta 17.5 16.3 16.4 17.5 6.24 6.92 1.45 6.26 16 218
Zr/Hf 35.4 35.8 34.7 35.4 7.21 9.54 1.22 7.00 39 341
Rb/Cs 86.0 61.0 44.7 86.0 94.0 93.7 3.61 90.9 30 196
K/Rb 640 433 506 640 477 515 1.95 424 81 1250
La/Th 13.4 7.32 7.8 13.4 16.1 14.2 2.88 13.1 47 728
Th/U 9.17 7.26 6.58 9.17 7.00 9.37 2.39 7.13 60 527
K/U 38100 29300 22900 38100 31900 44900 3.30 36100 60 527
Eu/Eu* 1.16 1.09 1.11 1.16 0.331 0.46 1.34 0.330 44 451
?Pb 4.16 4.11 4.16 4.16 0.164 0.23 1.03 0.141 - 4
Heat Prod. 0.558 0.179 0.207 0.558 1.04 0.36 3.73 0.708 54 70
Heat Prod. (mW/m3) 1.62 0.52 0.60 1.62
Table B.10: Archean Granulite Facies Terrains
202
Mean Median Geo ? Mean STD IQR Geo ? STD N(filtered) N(original)
Mean STD
La/Yb 6.53 4.69 4.87 6.53 5.16 7.37 2.14 4.8 7 34
Nb/Ta 16.70 17.5 7.89 16.7 10.1 16 6.43 18.8 5 29
Zr/Hf 30.20 31.7 28.4 30.2 9.24 13.5 1.47 10.6 5 31
Rb/Cs 26.60 26.6 18.5 26.6 19.1 38.3 2.48 21.6 2 6
K/Rb 542 503 458 542 287 476 1.85 306 7 41
La/Th 9.22 9.56 8.77 9.22 2.66 4.61 1.4 2.91 6 31
Th/U 2.18 2.17 2.03 2.18 0.798 1.47 1.45 0.798 5 29
K/U 16100 5190 7040 16100 17400 25600 4.29 18800 5 29
Eu/Eu* 1.01 0.968 1.01 1.01 0.095 0.14 1.09 0.0926 4 30
?Pb 5.48 5.75 5.44 5.48 0.668 1.01 1.13 0.668 - 21
Heat Prod. 0.0792 0.0433 0.0585 0.0792 0.0771 0.07 2.08 0.0590 3 5
Heat Prod. (mW/m3) 0.23 0.13 0.17 0.23
Table B.11: Eclogite Facies Xenoliths
203
Mean Median Geo ? Mean STD IQR Geo ? STD N(filtered) N(original)
Mean STD
La/Yb 7.46 5.48 3.96 7.46 6.9 10.4 3.69 7.77 5 40
Nb/Ta 29.40 29.4 29.4 - 0 0 1 - 1 25
Zr/Hf 27.20 27.2 20.6 27.2 17.7 35.4 2.18 19.4 2 32
Rb/Cs 53.30 53.3 52.7 53.3 7.52 15 1.15 7.55 2 23
K/Rb 693 526 564 693 498 435 1.84 431 5 35
La/Th 60.50 60.5 31.7 60.5 51.5 103 3.53 63.6 2 19
Th/U 2.73 2.56 2.7 2.73 0.424 0.746 1.16 0.415 3 26
K/U 68000 44300 24700 68000 65600 117000 6.04 87100 3 26
Eu/Eu* 0.875 0.875 0.875 0.875 0.0182 0.036 1.02 0.0182 2 8
?Pb - - - - - - - - - 0
Heat Prod. 0.0647 0.0437 0.0539 0.0647 0.0489 0.0681 1.81 0.0380 1 3
Heat Prod. (mW/m3) 0.19 0.13 0.16 0.19
Table B.12: Eclogite Facies Terrains
Amphibolite Facies Lithologies Granulite Facies Xenoliths
Post Archean Granulite Facies Terrains Archean Granulite Facies Terrains
Eclogite Facies Xenoliths Eclogite Facies Terrains
20 40 60 80
SiO2 (wt.%)
Figure B.5: Maps of the metamorphic sample datasets that we have extensive sample
coverage for amphibolite and granulite facies lithologies with more limited exposures of
eclogite facies lithologies. Each circle is a single sample whose color is determined by
weight percent SiO2. Many circles overlap. There is little correlation between SiO2 and
location.
204
wt.% SiO wt.% SiO2 wt.% SiO2 2
Amphibolite Facies Lithologies Granulite Facies Xenoliths
Post Archean Granulite Facies Terrains Archean Granulite Facies Terrains
Eclogite Facies Xenoliths Eclogite Facies Terrains
20 40 60 80
Mg#
Figure B.6: Maps of the metamorphic sample datasets that we have extensive sample
coverage for amphibolite and granulite facies lithologies with more limited exposures of
eclogite facies lithologies. Each circle is a single sample whose color is determined by
weight percent SiO2. Many circles overlap. There is no correlation between Mg# and
location.
205
5 Amphibolite Facies 5 Granulite Facies 5
0.3
Post Archean 
Lithologies Xenoliths Granulite Facies 
Terrains
0 0 0
0.2
-5 -5 -5
0 2 4 0 2 4 0 2 4
5Archean Granulite 5 Eclogite Facies 5 Eclogite Facies 
Facies Terrains Xenoliths Terrains 0.1
0 0 0
-5 -5 -5
0 2 4 0 2 4 0 2 4 0
Ln(Nb) (ppm)
Figure B.7: Nb vs. Ta forms a log-linear relationship with consistent Nb/Ta values for
amphibolite and granulite facies lithologies. Color is relative data point density.
206
Ln(Ta) (ppm)
% Data Density
0.15
Amphibolite Facies Granulite Facies Post Archean 
2 Lithologies 2 Xenoliths 2 Granulite 
Facies Terrains
0 0 0
-2 -2 -2 0.1
-4 -2 0 2 -4 -2 0 2 -4 -2 0 2
2 2 Eclogite Facies Xenoliths 2 Eclogite Facies Terrains 0.05
0 0 0
-2 Archean -2 -2
Granulite Facies Terrains
-4 -2 0 2 -4 -2 0 2 -4 -2 0 2 0
Ln(U) (ppm)
Figure B.8: Ln[Th] vs. Ln[U] for amphibolite and granulite facies lithologies forms a
linear trend. Color indicates relative data point density. The Th/U ratio for all lithologies
forms a linear trend in log-log space. Scatter in U concentration could be caused by U?s
mobility in certain oxidation states since amphibolites can contain hydrous minerals.
207
Ln(Th) (ppm)
% Data Density
Appendix C: Supporting Information for Chapter 4
1000
4.5
800
4
600
3.5
400
3
200
2.5
0
30 40 50 60 70 80 90
SiO2 (wt.%)
Figure C.1: Vs varies primarily as a function of temperature and composition. The
figure shows the calculated Vs for each sample from the southwestern US at different
temperature intervals. For this example, pressure is kept constant at 0.858 GPa (?30 km
depth). Lowering temperature or SiO2 content generally decreases Vs.
C.1 Compositional Maps and Uncertainties
208
Temperature (?C)
Vs (km/s)
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
32 32 14
12
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110 10
C 30 km Depth D 35 km Depth 8
38 Nevada Utah 38 Nevada Utah 6
36 36 4
California Arizona California Arizona
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure C.2: Uncertainty in wt.% associated with our SiO2 calculations. Uncertainty is
calculated as 12 the inter-quartile range at various depths. Uncertainty is lowest in the
Southern Basin and Range and remains relatively consistent with depth.
209
SiO2 (wt.%)
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
18
32 32
16
30 30 14
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
C D 1230 km Depth 35 km Depth
10
38 Nevada Utah 38 Nevada Utah
8
36 36
California Arizona California
6
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure C.3: Variations in median MgO + FeO abundance across the Basin and Range
and Colorado Plateau. Where SiO2 abundance decreases, amount of mafics increases
with depth. The Southern Basin and Range has a consistently high mafic content while
the Colorado Plateau is more intermediate.
210
MgO + FeO (wt.%)
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
8
32 32
7
6
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
5
C 30 km Depth D 35 km Depth 4
38 Nevada 3Utah 38 Nevada Utah
2
36 36
California Arizona California
1
Arizona
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure C.4: Uncertainty in wt.% associated with our MgO + FeO calculations. Uncer-
tainty is calculated as 12 the inter-quartile range at various depths, and increases by a few
wt.% with increasing depth.
211
MgO + FeO (wt.%)
Change in Composition
Relative to Uncertainty
A SiO2 200
38 Nevada 180Utah
160
140
36
California 120Arizona
100
34
80
60
32 40
20
30 0
-120 -118 -116 -114 -112 -110
B MgO + FeO
500
38 Nevada 450Utah
400
350
36
California 300Arizona
250
34
200
150
32 100
50
30 0
-120 -118 -116 -114 -112 -110
Figure C.5: The Colorado Plateau shows the greatest % change in composition when
compared to the uncertainty in our measurements. There is almost no composition change
from the top to the bottom of the lower crust in the Southern Basin and Range. The color
scale indicates the difference in (A) SiO2 or (B) mafics between 20 km and at Moho
depth, divided by the uncertainty in the composition at Moho depth.
212
Top of Lower Crust - Moho Top of Lower Crust - Moho
x 100 (%) x 100 (%)
Composition Uncertainty Composition Uncertainty
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
7.2
32 32
7.1
30 30 7.0
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110 6.9
C 6.830 km Depth D 35 km Depth 6.7
38 Nevada Utah 38 Nevada Utah 6.6
6.5
36 36 6.4
California Arizona California Arizona
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure C.6: Median Vp calculated from the joint geochemical-geophysical model. Orig-
inal Vp?s were calculated in Perple X. The Colorado Plateau is clearly visible as a slower
Vp region to the east.
213
Vp (km/s)
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
32 32 4.2
30 30 4.0
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
3.8
C 30 km Depth D 35 km Depth 3.6
38 Nevada Utah 38 Nevada Utah 3.4
3.2
36 36
California Arizona California Arizona
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure C.7: Joint model selections of median Vs. The geochemical data favor faster Vs
solutions and therefore weight the model.
214
Vs (km/s)
A 20 km Depth B 25 km Depth
38 Nevada Utah 38 Nevada Utah
36 36
California Arizona California Arizona
34 34
32 32 4.2
4.0
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
3.8
C 30 km Depth D 35 km Depth 3.6
38 Nevada Utah 38 Nevada Utah 3.4
3.2
36 36
California Arizona California Arizona
34 34
32 32
30 30
-120 -118 -116 -114 -112 -110 -120 -118 -116 -114 -112 -110
Figure C.8: Median Vs from inversion of Earthscope USArray seismic data at various
depths. Vs increases noticeably with increasing depth.
A Station: A17A-TA B Station: P08A-TA C Station: X14A-TA
5 700 5 500 5 500
10 600 10 10
400 400
15 500 15 15
20 400 20 300 20 300
25 300 25 200 25 200
30 200 30 30
100 100
35 100 35 35
40 0 40 0 40 0
3 3.5 4 4.5 5 3 3.5 4 4.5 5 3.5 4 4.5 5
Vs (km/s) Vs (km/s) Vs (km/s)
Figure C.9: Examples of the results of the dispersion wave inversions from three different
TA stations used to produce the ?seismic Vs? in the model. At each depth (y-axis) an en-
semble of results shows the distribution of possible Vs values from the seismic inversions
alone.
215
Depth (km)
Number of Inversions
Depth (km)
Number of Inversions
Depth (km)
 (km/s)
Number of Inversions Vs
Bibliography 
RN Abbott and JP Greenwood. Retrograde metamorphism of eclogite in the southern 
appalachian mountains, usa-a case involving seamount subduction? Journal of Meta-
morphic Geology, 19(4):433-443, 2001. 
M. Agostini, S. Appel, G. Bellini, J. Benziger, D. Bick, G. Bonfini, D. Bravo, B. Caccian?
iga, F. Calaprice, A. Caminata, P. Cavalcante, A. Chepumov, K. Choi, D. D' Angelo,
S. Davini, A. Derbin, L. Di Noto, I. Drachnev, A. Empl, A. Etenko, G. Fioren?
tini, K. Fomenko, D. Franco, F. Gabriele, C. Galbiati, C. Ghiano, M. Giammarchi, 
M. Goeger-Neff, A. Goretti, M. Gromov, C. Hagner, T. Houdy, E. Hungerford, Aldo
Ianni, Andrea Ianni, K. Jedrzejczak, M. Kaiser, V. Kobychev, D. Korablev, G. Ko?
rga, D. Kryn, M. Laubenstein, B. Lehnert, E. Litvinovich, F. Lombardi, P. Lombardi, 
L. Ludhova, G. Lukyanchenko, I. Machulin, S. Manecki, W. Maneschg, F. Manto?
vani, S. Marcocci, E. Meroni, M. Meyer, L. Miramonti, M. Misiaszek, M. Mon?
tuschi, P. Mosteiro, V. Muratova, B. Neumair, L. Oberauer, M. Obolensky, F. Or?
tica, K. Otis, L. Pagani, M. Pallavicini, L. Papp, L. Perasso, A. Pocar, G. Ranucci, 
A. Razeto, A. Re, B. Ricci, A. Romani, R. Roncin, N. Rossi, S. Schonert, D. Se?
menov, H. Simgen, M. Skorokhvatov, 0. Smirnov, A. Sotnikov, S. Sukhotin, Y. Su?
vorov, R. Tartaglia, G. Testera, J. Thurn, M. Toropova, E. Unzhakov, R. B. Vogelaar, 
F. von Feilitzsch, H. Wang, S. Weinz, J. Winter, M. Wojcik, M. Wurm, Z. Yokley,
216 
0. Zaimidoroga, S. Zavatarelli, K. Zuber, and G. Zuzel. Spectroscopy of geoneu?
trinos from 2056 days of borexino data. Phys. Rev. D, 92:031101, Aug 2015. doi: 
10.1103/PhysRevD.92.031101. 
M. Agostini, K. Altenmiiller, S. Appel, V. Atroshchenko, Z. Bagdasarian, D. Basil?
ico, G. Bellini, J. Benziger, D. Bick, G. Bonfini, D. Bravo, B. Caccianiga,
F. Calaprice, A. Caminata, L. Cappelli, P. Cavalcante, F. Cavanna, A. Chepurnov,
K. Choi, D. D' Angelo, S. Davini, A. Derbin, A. Di Giacinto, V. Di Marcello,
X. F. Ding, A. Di Ludovico, L. Di Noto, I. Drachnev, G. Fiorentini, A. Formozov,
D. Franco, F. Gabriele, C. Galbiati, M. Gschwender, C. Ghiano, M. Giammarchi,
A. Goretti, M. Gromov, D. Guffanti, C. Hagner, E. Hungerford, Aldo Ianni, An?
drea Ianni, A. Jany, D. Jeschke, S. Kumaran, V. Kobychev, G. Korga, T. Lachen?
maier, T. Lasserre, M. Laubenstein, E. Litvinovich, P. Lombardi, I. Lomskaya, L. Lud?
hova, G. Lukyanchenko, L. Lukyanchenko, I. Machulin, F. Mantovani, G. Manuzio, 
S. Marcocci, J. Maricic, J. Martyn, E. Meroni, M. Meyer, L. Miramonti, M. Misi?
aszek, M. Montuschi, V. Muratova, B. Neumair, M. Nieslony, L. Oberauer, A. Onillon, 
V. Orekhov, F. Ortica, M. Pallavicini, L. Papp, 6. Penek, L. Pietrofaccia, N. Pilipenko,
A. Pocar, G. Raikov, M. T. Ranalli, G. Ranucci, A. Razeto, A. Re, M. Redchuk,
B. Ricci, A. Romani, N. Rossi, S. Rottenanger, S. Schonert, D. Semenov, M. Sko?
rokhvatov, 0. Smimov, A. Sotnikov, V. Strati, Y. Suvorov, R. Tartaglia, G. Testera, 
J. Thum, E. Unzhakov, A. Vishneva, M. Vivier, R. B. Vogelaar, F. von Feilitzsch,
M. Wojcik, M. Wurm, 0. Zaimidoroga, S. Zavatarelli, K. Zuber, and G. Zuzel. Com?
prehensive geoneutrino analysis with Borexino. Phys. Rev. D, 101(1):012009, 2020. 
217 
doi: 10.1103/PhysRevD.101.012009. 
Abdulrahman Al-Safarjalani, Sobhi Nasir, Thomas Fockenberg, and Hans-Joachim Mas?
sonne. Chemical composition of crustal xenoliths from southwestern Syria: Character?
ization of the upper part of the lower crust beneath the Arabian plate. Geochemistry, 
69(4):359-375, 2009. ISSN 0009-2819. doi: 10.1016/j.chemer.2009.05.005. 
Francis Albarede. The growth of continental crust. Tectonophysics, 296(1-2):1-14, 1998. 
Kiara L Alessio, Martin Hand, David E Kelsey, Megan A Williams, Laura J Morrissey, 
and Karin Barovich. Conservation of deep crustal heat production. Geology, 46(4): 
335-338, 2018.
Fengpeng An, Guangpeng An, Qi An, Vito Antonelli, Eric Baussan, John Beacom, 
Leonid Bezrukov, Simon Blyth, Riccardo Brugnera, Margherita Buizza Avanzini, et al. 
Neutrino physics with juno. Journal of Physics G: Nuclear and Particle Physics, 43 
(3):030401, 2016. 
T Araki, S Enomoto, K Furuno, Y Gando, K Ichimura, H Ikeda, K Inoue, Y Kishimoto, 
M Koga, Y Koseki, et al. Experimental investigation of geologically produced antineu?
trinos with kamland. Nature, 436(7050):499-503, 2005. 
Ricardo Arevalo and William F McDonough. Tungsten geochemistry and implications 
for understanding the earth's interior. Earth and Planetary Science Letters, 272(3-4): 
656-665, 2008.
Ricardo Arevalo and William F McDonough. Chemical variations and regional diversity 
218 
observed in MORB. Chemical Geology, 271(1):70-85, 2010. ISSN 0009-2541. doi: 
10.1016/j.chemgeo.2009.12.013. 
Ricardo Arevalo, William F McDonough, and Mario Luong. The K/U ratio of the silicate 
earth: Insights into mantle composition, structure and thermal evolution. Earth and 
Planetary Science Letters, 278(3-4):361-369, 2009. 
NT Arndt. Archean komatiites. Archean crustal evolution, 10:11-44, 1994. 
Irina M Artemieva. Global 1 x 1 thermal model TC 1 for the continental lithosphere: im?
plications for lithosphere secular evolution. Tectonophysics, 416(1-4):245-277, 2006. 
LD Ashwal, P Morgan, SA Kelley, and JA Percival. Heat production in an Archean crustal 
profile and implications for heat flow and mobilization of heat-producing elements. 
Earth and Planetary Science Letters, 85(4):439-450, 1987. 
Lewis D Ashwal, Robert D Tucker, and Ernst K Zinner. Slow cooling of deep crustal 
granulites and pb-loss in zircon. Geochimica et Cosmochimica Acta, 63(18):2839-
2851, 1999. 
D A B anks, R Green, RA Cliff, and B W D Yardley. Chlorine isotopes in fluid inclusions: 
determination of the origins of salinity in magmatic fluids. Geochimica et Cosmochim-
ica Acta, 64(10):1 785-1789, 2000. 
Sarah-Jane Barnes. Chalcophile elements. Springer International Publishing, 2016. 
Katherine R Barnhart, Kevin H Mahan, Terrence J Blackburn, Samuel A Bowring, and 
Francis O Dudas. Deep crustal xenoliths from central montana, usa: Implications for 
219 
the timing and mechanisms of high-velocity lower crust formation. Geosphere, 8(6): 
1408-1428, 2012. 
C Bassin, G Laske, and G Masters. The current limits of resolution for surface wave 
tomography in North America. Eos, 81, 2000. 
John F. Beacom, Shaomin Chen, Jianping Cheng, Sayed N. Doustimotlagh, Yuanning 
Gao, Guanghua Gong, Hui Gong, Lei Guo, Ran Han, Hong-Jian He, Xingtao Huang, 
Jianmin Li, Jin Li, Mohan Li, Xueqian Li, Wei Liao, Guey-Lin Lin, Zuowei Liu, 
William McDonough, Ondrej Sramek, Jian Tang, Linyan Wan, Yuanqing Wang, Zhe 
Wang, Zongyi Wang, Hanyu Wei, Yufei Xi, Ye Xu, Xun-Jie Xu, Zhenwei Yang, 
Chunfa Yao, Minfang Yeh, Qian Yue, Liming Zhang, Yang Zhang, Zhihong Zhao, 
Yangheng Zheng, Xiang Zhou, Xianglei Zhu, and Kai Zuber. Physics prospects 
of the Jinping neutrino experiment. Chinese Physics C, 41(2):023002, 2017. doi: 
10.1088/1674-l 137/41/2/023002. 
G Bellini, A Ianni, L Ludhova, Fabio Mantovani, and William F McDonough. Geo?
neutrinos. Progress in Particle and Nuclear Physics, 73:1-34, 2013. 
Gianpaolo Bellini, Kunio Inoue, Fabio Mantovani, Andrea Serafini, Virginia Strati, 
and Hiroko Watanabe. Geoneutrinos and geoscience: an intriguing joint-venture. 
La Rivista del Nuovo Cimento, pages 1-105, 2021. doi: https://doi.org/10.1007/ 
s40766-021-00026-7. 
Peter Bird. Laramide crustal thickening event in the Rocky Mountain Foreland and 
220 
Great Plains. Tectonics, 3(7):741-758, 1984. ISSN 1944-9194. doi: 10.1029/ 
TC003i007p00741. 
S R Bohlen and Klaus Mezger. Origin of granulite terranes and the formation of the 
lowermost continental crust. Science, 244(4902):326-329, 1989. 
Thomas M Brocher. Empirical Relations between Elastic Wavespeeds and Density in 
the Earth's Crust. Bulletin of the Seismological Society of America, 95(6):2081-2092, 
2005. ISSN 0037-1106. doi: 10.1785/0120050077. 
Michael Brown and Tim Johnson. Time's arrow, time's cycle: Granulite metamorphism 
and geodynamics. Mineralogical Magazine, 83(3):323-338, 2019. 
JS Buehler and PM Shearer. Uppermost mantle seismic velocity structure beneath usarray. 
Journal of Geophysical Research: Solid Earth, 122(1):436-448, 2017. 
Roland Biirgmann and Georg Dresen. Rheology of the lower crust and upper mantle: 
Evidence from rock mechanics, geodesy, and field observations. Annu. Rev. Earth 
Planet. Sci., 36:531-567, 2008. 
Warren B Caldwell, Simon L Klemperer, Shyam S Rai, and Jesse F Lawrence. Partial 
melt in the upper-middle crust of the northwest himalaya revealed by rayleigh wave 
dispersion. Tectonophysics, 477(1-2):58-65, 2009. 
Fabio Cammarano and Mattia Guerri. Global thermal models of the lithosphere. Geo?
physical Journal International, 210(1):56-72, 2017. 
R L Carlson and C N Herrick. Densities and porosities in the oceanic crust and their 
221 
variations with depth and age. Journal of Geophysical Research: Solid Earth, 95(B6): 
9153-9170, 1990. 
DA Carswell and SJ Cuthbert. Eclogite facies metamorphism in the lower continental 
crust. Geological Society, London, Special Publications, 24(1):193-209, 1986. 
Bruce W Chappell, Colleen J Bryant, and Doone Wyborn. Peraluminous i-type granites. 
Lithos, 153:142-153, 2012. 
Bernard Charlier, Olivier Namur, Michael J Toplis, Pierre Schiano, Nicolas Cluzel, 
Michael D Higgins, and Jacqueline Vander Auwera. Large-scale silicate liquid im?
miscibility during differentiation of tholeiitic basalt to granite and the origin of the daly  
gap. Geology, 39(10):907-910, 2011. 
Kang Chen, Richard J Walker, Roberta L Rudnick, Shan Gao, Richard M Gaschnig, Igor S 
Puchtel, Ming Tang, and Zhao-Chu Hu. Platinum-group element abundances and re-os 
isotopic systematics of the upper continental crust through time: Evidence from glacial 
diamictites. Geochimica et CosmochimicaActa, 191:1-16, 2016. 
Kang Chen, Roberta L Rudnick, Zaicong Wang, Ming Tang, Richard M Gaschnig, Zongqi 
Zou, Tao He, Zhaochu Hu, and Yongsheng Liu. How mafic was the archean upper 
continental crust? insights from cu and ag in ancient glacial diamictites. Geochimica 
et CosmochimicaA cta, 278: 16-29, 2020. 
Hyun-Moo Cho, Han-Joon Kim, Hyeong-Tae Jou, Jong-Kuk Hong, and Chang-Eob Baag. 
Transition from rifted continental to oceanic crust at the southeastern korean margin in 
the east sea (japan sea). Geophysical Research Letters, 31(7), 2004. 
222 
Nikolas I. Christensen and David M. Fountain. Constitution of the Lower Continen?
tal Crust Based on Experimental Studies of Seismic Velocities in Granulite. Ge?
ological Society of America Bulletin, 86(2):227, 1975. ISSN 0016-7606. doi: 
10.1130/0016-7606(1975)86(227:COTLCC)2.0.CO;2. 
Nikolas I Christensen and Walter D Mooney. Seismic velocity structure and composition 
of the continental crust: A global view. Journal of Geophysical Research: Solid Earth, 
lO0(B6):9761-9788, 1995a. 
Nikolas I Christensen and Walter D Mooney. Seismic velocity structure and composition 
of the continental crust: A global view. Journal of Geophysical Research: Solid Earth, 
lO0(B6):9761-9788, 1995b. ISSN 2156-2202. doi: 10.1029/95JB00259. 
J Graham Cogley. Continental margins and the extent and number of the continents. 
Reviews of Geophysics, 22(2):101-122, 1984. 
Massimo Coltorti, Riccardo Boraso, Fabio Mantovani, Michele Morsilli, Giovanni 
Fiorentini, Alberto Riva, G Rusciadelli, Renzo Tassinari, C Tomei, G Di Carlo, et al. 
U and Th content in the Central Apennines continental crust: A contribution to the 
determination of the geo-neutrinos flux at LNGS. Geochimica et Cosmochimica Acta, 
75(9):2271-2294, 2011. doi: 10.1016/j.gca.2011.01.024. 
Peter J Coney. Cordilleran metamorphic core complexes: An overview. Cordilleran 
metamorphic core complexes: Geological Society ofA merica Memoir, 153:7-31, 1980. 
Peter J Coney and Tekla A Harms. Cordilleran metamorphic core complexes: Cenozoic 
extensional relics of mesozoic compression. Geology, 12(9):550-554, 1984. 
223 
J A D Connolly. Computation of phase equilibria by linear programming: A tool for 
geodynamic modeling and its application to subduction zone decarbonation. Earth and 
Planetary Science Letters, 236(1):524-541, 2005. ISSN 0012-821X. doi: 10.1016/j. 
epsl.2005.04.033. 
Sandro Conticelli, Richard W Carlson, Elisabeth Widom, and Giancarlo Serri. Chemical 
and isotopic composition (Os, Pb, Nd, and Sr) of Neogene to Quaternary calc-alkalic, 
shoshonitic, and ultrapotassic mafic rocks from the Italian peninsula: Inferences on the 
nature of their mantle sources. Cenozoic volcanism in the Mediterranean area, 418: 
171, 2007. 
F. J. Cooper, J. P. Platt, R. Anczkiewicz, and M. J. Whitehouse. Footwall dip of a core 
complex detachment fault: Thermobarometric constraints from the northern Snake 
Range (Basin and Range, USA). Journal of Metamorphic Geology, 28(9):997-1020, 
2010. ISSN 1525-1314. doi: 10.llll/j.1525-1314.2010.00907.x. 
Domenico Cosentino, Paola Cipollari, Pietro Marsili, and Davide Scrocca. Geology of 
the central apennines: a regional review. Journal of the virtual explorer, 36(11):1-37, 
2010. 
Max D Crittenden, Peter J Coney, George Herbert Davis, and George Hamilton Davis. 
Cordilleran metamorphic core complexes, volume 153. Geological Society of America, 
1980. 
RA Daly. Igneous rocks and their origin. Preliminary Note on Alkaline Rhyolites from 
Tokati, Hokkaido, 319:20, 1914. 
224 
J Huw Davies. Global map of solid earth surface heat flow. Geochemistry, Geophysics, 
Geosystems, 14(10):4608-4622, 2013. 
R De Argollo and J-G Schilling. Ge/si and ga/al fractionation in hawaiian volcanic rocks. 
Geochimica et Cosmochimica Acta, 42(6):623-630, 1978. 
de H De La Roche, J t Leterrier, Philippe Grandclaude, and M Marchal. A classification 
of volcanic and plutonic rocks using rlr2-diagram and major-element analyses-its 
relationships with current nomenclature. Chemical geology, 29(1-4):1 83-210, 1980. 
E Di Luzio, G Mele, MM Tiberti, GP Cavinato, and M Parotto. Moho deepening and 
shallow upper crustal delamination beneath the central apennines. Earth and Planetary 
Science Letters, 280(1 -4 ): 1-12, 2009. doi: https :// doi.org/10.1016/j .epsl.2008. 09.018. 
Jacqueline E Dixon, T. H Dixon, D. R Bell, and R Malservisi. Lateral variation in upper 
mantle viscosity: Role of water. Earth and Planetary Science Letters, 222(2):451-467, 
2004. ISSN 0012-821X. doi: 10.1016/j.epsl.2004.03.022. 
Ksenia Dokukina and Michael Mints. Subduction of the mesoarchaean spreading ridge 
and related metamorphism, magmatism and deformation by the example of the gridino 
eclogitized mafic dyke swarm, the belomorian eclogite province, eastern fennoscandian 
shield. Journal of Geodynamics, 123:1-37, 2019. 
Mihai N Ducea. Fingerprinting orogenic delamination. Geology, 39(2):191-192, 2011. 
Josef Dufek and Olivier Bachmann. Quantum magmatism: Magmatic compositional gaps 
generated by melt-crystal dynamics. Geology, 38(8):687-690, 2010. 
225 
Gregory Dumond, Michael L Williams, and Sean P Regan. The athabasca granulite ter?
rane and evidence for dynamic behavior of lower continental crust. Annual Review of 
Earth and Planetary Sciences, 46:353-386, 2018. 
Adam M Dziewonski and Don L Anderson. Preliminary reference earth model. Physics 
of the earth and planetary interiors, 25(4):297-356, 1981. 
Goran Ekstrom. Love and rayleigh phase-velocity maps, 5-40 s, of the western and 
central usa from usarray data. Earth and Planetary Science Letters, 402:42-49, 2014. 
Sanshiro Enomoto. Experimental study of geoneutrinos with kamland. In Neutrino Geo?
physics: Proceedings of Neutrino Sciences 2005, pages 131-146. Springer, 2006. 
Sanshiro Enomoto, Eiji Ohtani, K Inoue, and A Suzuki. Neutrino geophysics with Kam?
LAND and future prospects. Earth and Planetary Science Letters, 258(1-2):147-159, 
2007. doi: 10.1016/j.epsl.2007.03.038. 
B Farcy, R Arevalo, and William F McDonough. K/U of the morb source and silicate 
earth. Journal of Geophysical Research: Solid Earth, 125(12):e2020JB020245, 2020. 
D Field and RB Elliott. The chemistry of gabbro/amphibolite transitions in south norway. 
Contributions to Mineralogy and Petrology, 47(1):63-76, 1974. 
Dennis Field and Peter W L Clough. K/Rb ratios and metasomatism in metabasites from a 
precambrian amphibolite-granulite transition zone. Journal of the Geological Society, 
132(3):277-288, 1976. 
226 
David M Fountain and Matthew H Salisbury. Exposed cross-sections through the con?
tinental crust: implications for crustal structure, petrology, and evolution. Earth and 
Planetary Science Letters, 56:263-277, 1981. 
David M Fountain, Kevin P Furlong, and Matthew H Salisbury. A heat production model 
of a shield area and its implications for the heat flow-heat production relationship. 
Geophysical Research Letters, 14(3):283-286, 1987. 
M B Fowler. Large-ion lithophile element characteristics of an amphibolite facies to 
granulite facies transition at gruinard bay, north-west scotland. Journal of Metamorphic 
Geology, 4(3):345-359, 1986. 
Allison Gale, Muriel Laubier, Stephane Escrig, and Charles H Langmuir. Constraints on 
melting processes and plume-ridge interaction from comprehensive study of the famous 
and north famous segments, mid-atlantic ridge. Earth and Planetary Science Letters, 
365:209-220, 2013. 
Azusa Gando, Y Gando, H Hanakago, H Ikeda, K Inoue, K Ishidoshiro, H Ishikawa, 
M Koga, R Matsuda, S Matsuda, et al. Reactor on-off antineutrino measurement with 
kamland. Physical Review D, 88(3):033001, 2013. 
Chao Gao and Vedran Lekic. Consequences of parametrization choices in surface wave 
inversion: Insights from transdimensional Bayesian methods. Geophysical Journal 
International, 215(2): 1037-1063, 2018. ISSN 0956-540X. doi: 10.1093/gji/ggy310. 
Ruohan Gao, Zhiwei Li, Ran Han, Andong Wang, Yufeng Li, Yufei Xi, Jingao Liu, Xin 
Mao, Yao Sun, and Ya Xu. Juloc: A local 3-d high-resolution crustal model in south 
227 
china for forecasting geoneutrino measurements at juno. Physics of the Earth and 
Planetary Interiors, 299:106409, 2020. 
Shan Gao, Ben-Ren Zhang, Zhen-Ming Jin, Hartmut Kern, Ting-ChuanL uo, and Zi-Dan 
Zhao. How mafic is the lower continental crust? Earth and Planetary Science Letters, 
161(1-4):101-117, 1998. 
Yajian Gao, Frederik Tilmann, Dirk-Philip van Herwaarden, Solvi Thrastarson, Andreas 
Fichtner, Benjamin Heit, Xiaohui Yuan, and Bernd Dieter Schurr. Full waveform in?
version beneath the central andes: Insight into the dehydration of the nazca slab and 
delamination of the back-arc lithosphere. Earth and Space Science Open Archive ES-
SOAr, 2021. 
M Gard, D Hasterok, M Hand, and G Cox. Variations in continental heat production from 
4 ga to the present: Evidence from geochemical data. Lithos, 342:391-406, 2019. 
Richard M Gaschnig, Roberta L Rudnick, William F McDonough, Alan J Kaufman, 
John W Valley, Zhaochu Hu, Shan Gao, and Michelle L Beck. Compositional 
evolution of the upper continental crust through time, as constrained by ancient 
glacial diamictites. Geochimica et Cosmochimica Acta, 186:316-343, 2016. doi: 
10.1016/j.gca.2016.03.020. 
TV Gerya, LL Perchuk, and J-P Burg. Transient hot channels: perpetrating and regurgi?
tating ultrahigh-pressure, high-temperature crust-mantle associations in collision belts. 
Lithos, 103(1-2):236-256, 2008. 
Taras V Gerya, Leonid L Perchuk, Walter V Maresch, Arne P Willner, Dirk D Van 
228 
Reenen, and C Andre Smit. Thermal regime and gravitational instability of multi?
layered continental crust: implications for the buoyant exhumation of high-grade meta?
morphic rocks. European Journal of Mineralogy, 14(4):687-699, 2002. 
Karsten Gohl. Antarctica's continent-ocean transitions: consequences for tectonic recon?
structions. In Antarctica: A Keystone in a Changing World: Proceedings of the 10th 
International Symposium on Antarctic Earth Sciences/Alan K. Cooper, Peter Barrett, 
Howard Stagg, Bryan Storey, Edmund Stump, Woody Wise, and the 10th ISAES edito?
rial team, Polar Resear, pages 29-38, 2008. 
Allison T Greaney, Roberta L Rudnick, Richard M Gaschnig, Joseph B Whalen, Beatrice 
Luais, and John D Clemens. Geochemistry of molybdenum in the continental crust. 
Geochimica et Cosmochimica Acta, 238:36-54, 2018. 
Benjamin Gruber, Thomas Chacko, D Graham Pearson, Claire Currie, and Andrew Men?
zies. Heat production and m oho temperatures in cratonic crust: evidence from lower 
crustal xenoliths from the slave craton. Lithos, 380:105889, 2021. 
BR Hacker, MH Ritzwoller, and J Xie. Partially melted, mica-bearing crust in central 
tibet. Tectonics, 33(7):1408-1424, 2014. 
Bradley R Hacker, Torgeir B Andersen, Scott Johnston, Andrew RC Kylander-Clark, 
Emily M Peterman, Emily O Walsh, and David Young. High-temperature deformation 
during continental-margin subduction & exhumation: The ultrahigh-pressure western 
gneiss region of norway. Tectonophysics, 480(1-4):149-171, 2010. 
229 
Bradley R Hacker, Peter B Kelemen, and Mark D Behn. Differentiation of the continental 
crust by relamination. Earth and Planetary Science Letters, 307(3-4):501-516, 2011. 
Bradley R Hacker, Peter B Kelemen, and Mark D Behn. Continental lower crust. Annual 
Review of Earth and Planetary Sciences, 43: 167-205, 2015a. 
Bradley R Hacker, Peter B Kelemen, and Mark D Behn. Continental Lower Crust. An?
nual Review of Earth and Planetary Sciences, 43(1):167-205, 2015b. doi: 10.1146/ 
annurev-earth-050212-124117. 
Bradley R. Hacker, Peter B. Kelemen, and Mark D. Behn. Continental Lower Crust. 
Annual Review of Earth and Planetary Sciences, 43(1):167-205, 2015c. doi: 10.1146/ 
annurev-earth-050212-124117. 
Alex N Halliday, Alan P Dickin, Robert N Hunter, Gareth R Davies, Tim J Demp?
ster, P Joseph Hamilton, and Brian G J Upton. Formation and composition of the 
lower continental crust: Evidence from Scottish xenolith suites. Journal of Geo-
physical Research: Solid Earth, 98(Bl):581-607, 1993. ISSN 2156-2202. doi: 
10.1029/92JB02276. 
Jens Hartmann, Hans H Dlirr, Nils Moosdorf, Michel Meybeck, and Stephan Kempe. The 
geochemical composition of the terrestrial surface (without soils) and comparison with 
the upper continental crust. International Journal of Earth Sciences, 101(1):365-376, 
2012. 
D Hasterok, M Gard, G Cox, and M Hand. A 4 ga record of granitic heat production: 
230 
Implications for geodynamic evolution and crustal composition of the early earth. Pre?
cambrian Research, 331:105375, 2019a. 
D Hasterok, Matthew Gard, C M B Bishop, and David Kelsey. Chemical identification of 
metamorphic protoliths using machine learning methods. Computers & Geosciences, 
132:56-68, 2019b. 
Christopher J Hawkesworth and AIS Kemp. Evolution of the continental crust. Nature, 
443(7113):811-817, 2006. 
Naoshi Hirata, Boris Ya Karp, Takashi Yamaguchi, Toshihiko Kanazawa, Kiyoshi Suye?
hiro, Junzo Kasahara, Hajime Shiobara, Masanao Shinohara, and Hajimu Kinoshita. 
Oceanic crust in the japan basin of the japan sea by the 1990 japan-ussr expedition. 
Geophysical Research Letters, 19(20):2027-2030, 1992. 
W Steven Holbrook, Walter D Mooney, and Nikolas I Christensen. The seismic velocity 
structure of the deep continental crust. Continental lower crust, 23: 1-43, 1992. 
T J B Holland and R Powell. An internally consistent thermodynamic data set for phases 
of petrological interest: An Internally Consistent Thermodynamic Dataset. Journal 
of Metamorphic Geology, 16(3):309-343, 2004. ISSN 02634929, 15251314. doi: 
10.1111/j.1525-1314.1998.00140.x. 
Yu Huang, Viacheslav Chubakov, Fabio Mantovani, Roberta L Rudnick, and William F 
McDonough. A reference Earth model for the heat-producing elements and associated 
geoneutrino flux. Geochemistry, Geophysics, Geosystems, 14(6):2003-2029, 2013. 
ISSN 1525-2027. doi: 10.1002/ggge.20129. 
231 
JA Jackson. Strength of the continental lithosphere: time to abandon the jelly sandwich? 
GSA today, 12:4-10, 2002. 
M D Jackson, J Bluntly, and R S J Sparks. Chemical differentiation, cold storage and 
remobilization of magma in the earth's crust. Nature, 564(7736):405-409, 2018. 
Oliver Jagoutz and Max W Schmidt. The formation and bulk composition of modem 
juvenile continental crust: The kohistan arc. Chemical Geology, 298:79-96, 2012. 
Oliver Jagoutz, Othmar Miintener, Max W Schmidt, and Jean-Pierre Burg. The roles of 
flux-and decompression melting and their respective fractionation lines for continental 
crust formation: Evidence from the kohistan arc. Earth and Planetary Science Letters, 
303(1-2):25-36, 2011. 
C Jaupart and J C Mareschal. Constraints on heat production from heat flow data, treatise 
on geochemistry, vol. 3, the crust, edited by r. rudnick, 2003. 
C Jaupart, S Labrosse, F Lucazeau, and JC Mareschal. 7 .06-temperatures, heat and energy 
in the mantle of the earth. Treatise on geophysics, 7:223-270, 2007a. 
C Jaupart, JC Mareschal, and AB Watts. Heat flow and thermal structure of the litho?
sphere. Treatise on geophysics, 6:217-252, 2007b. 
C Jaupart, J-C Mareschal, H Bouquerel, and C Phaneuf. The building and stabilization 
of an archean craton in the superior province, canada, from a heat flow perspective. 
Journal of Geophysical Research: Solid Earth, 119(12):9130-9155, 2014. 
C. Jaupart, S. Labrosse, F. Lucazeau, and J.-C. Mareschal. Temperatures, heat, and energy
232 
in the mantle of the earth. In Gerald Schubert, editor, Treatise on Geophysics (Second 
Edition), volume 7, pages 223-270. Elsevier, Oxford, 2015. ISBN 978-0-444-53803-1. 
doi: 10.1016/B978-0-444-53802-4.00126-3. 
Claude Jaupart, Jean-Claude Mareschal, and Lidia Iarotsky. Radiogenic heat production 
in the continental crust. Lithos, 262:398-427, 2016. 
M J avoy. The integral enstatite chondrite model of the earth. Geophysical Research 
Letters, 22(16):2219-2222, 1995. 
Marc Javoy, Edouard Kaminski, Frarn;ois Guyot, Denis Andrault, Chrystele Sanloup, 
Manuel Moreira, Stephane Labrosse, Albert Jambon, Pierre Agrinier, Anne Davaille, 
and Claude J au part. The chemical composition of the Earth: Enstatite chondrite mod?
els. Earth and Planetary Science Letters, 293(3):259-268, 2010. doi: 10.1016/j.epsl. 
2010.02.033. 
Frances Elaine Jenner. Cumulate causes for the low contents of sulfide-loving elements 
in the continental crust. Nature Geoscience, 10(7):524-529, 2017. 
M. Jull and P. B. Kelemen. On the conditions for lower crustal convective instability.
Journal of Geophysical Research: Solid Earth, 106(B4):6423-6446, 2001. ISSN 2156-
2202. doi: 10.1029/2000JB900357.
Robert Woodbury Kay and S Mahlburg Kay. Delamination and delamination magmatism. 
Tectonophysics, 219(1-3):177-189, 1993. 
Peter B Kelemen and Mark D Behn. Formation of lower continental crust by relamination 
of buoyant arc lavas and plutons. Nature Geoscience, 9(3): 197-205, 2016. 
233 
C Brenhin Keller, Blair Schoene, Melanie Barboni, Kyle M Samperton, and Jon M Hus?
son. Volcanic-plutonic parity and the differentiation of the continental crust. Nature, 
523(7560):301-307, 2015. 
Jun Korenaga. Plate tectonics, flood basalts and the evolution of earth's oceans. Terra 
Nova, 20(6):419-439, 2008. 
EJ Krogh, CW Oh, and JC Liou. Polyphase and anticlockwise p-t evolution for franciscan 
eclogites and blueschists fromjenner, califomia, usa. Journal ofM etamorphic Geology, 
12(2):121-134, 1994. 
IT Kukkonen, IV Golovanova, Yu V Khachay, VS Druzhinin, AM Kosarev, and 
VA Schapov. Low geothermal heat flow of the urals fold belt-implication of low 
heat production, fluid circulation or palaeoclimate? Tectonophysics, 276(1-4):63-85, 
1997. 
Christof Kusebauch, Timm John, Jaime D Barnes, Andreas Kliigel, and Hakon O Aus?
trheim. Halogen element and stable chlorine isotope fractionation caused by fluid-rock 
interaction (bamble sector, se norway). Journal of Petrology, 56(2):299-324, 2015. 
N J Kusznir and R G Park. The extensional strength of the continental lithosphere: its 
dependence on geothermal gradient, and crustal composition and thickness. Geological 
Society, London, Special Publications, 28(1):35-52, 1987. 
s Labrosse. Core-mantle boundary, heat flow across. In D Gubbins and E Herrero?
Bervera, editors, Encyclopedia of Geomagnetism and Paleomagnetism. Springer, Dor-
drecht, 2007. doi: https://doi.org/10.1007/978-1-4020-4423-6_51. 
234 
Stephane Labrosse. Hotspots, mantle plumes and core heat loss. Earth and Planetary 
Science Letters, 199(1-2):147-156, 2002. 
Stephane Labrosse, Jean-Paul Poirier, and Jean-Louis Le Mouel. The age of the inner 
core. Earth and Planetary Science Letters, 190(3-4):111-123, 2001. 
Gabi Laske, Guy Masters, Zhitu Ma, and Mike Pasyanos. Update on crustl.0 - a 1-
degree global model of earth's crust. In Geophys. res. abstr, volume 15, page 2658, 
2013. 
Mary L Leech. Arrested orogenic development: Eclogitization, delamination, and tec?
tonic collapse. Earth and Planetary Science Letters, 185(1):149-159, 2001. ISSN 
0012-821X. doi: 10.1016/S0012-821X(00)00374-5. 
A. Levander, B. Schmandt, M. S. Miller, K. Liu, K. E. Karlstrom, R. S. Crow, C.-T. A.
Lee, and E. D. Humphreys. Continuing Colorado plateau uplift by delamination-style
convective lithospheric downwelling. Nature, 472(7344):461-465, 2011. ISSN 1476-
4687. doi: 10.1038/naturelO00l.
A Lima, S Albanese, and D Cicchella. Geochemical baselines for the radioelements k, 
u, and th in the campania region, italy: a comparison of stream-sediment geochemistry 
and gamma-ray surveys. Applied Geochemistry, 20(3):611-625, 2005. 
Fan-Chi Lin, Victor C Tsai, and Brandon Schmandt. 3-d crustal structure of the west?
ern united states: application of rayleigh-wave ellipticity extracted from noise cross?
correlations. Geophysical Journal International, 198(2):656-670, 2014. 
235 
Yong-Sheng Liu, Shan Gao, Shu-Yan Jin, Sheng-Hong Hu, Min Sun, Zu-Bin Zhao, 
and Jia-Lin Feng. Geochemistry of lower crustal xenoliths from neogene hannuoba 
basalt, north china craton: Implications for petrogenesis and lower crustal composi?
tion. Geochimica et Cosmochimica Acta, 65(15):2589-2604, 2001. 
Richard F. Livaccari. Role of crustal thickening and extensional collapse in the tec?
tonic evolution of the Sevier-Laramide orogeny, western United States. Geology, 19 
(11):1 104-1107, 1991. ISSN 0091-7613. doi: 10.1130/0091-7613(1991)019( 1104: 
ROCTAE)2.3.CO;2. 
Bruno Lombardo and Franco Rolfo. Two contrasting eclogite types in the Himalayas: 
Implications for the Himalayan orogeny. Journal of Geodynamics, 30(1):37-60, 2000. 
ISSN 0264-3707. doi: 10.1016/S0264-3707(99)00026-5. 
F Lucazeau. Analysis and mapping of an updated terrestrial heat flow data set. Geochem?
istry, Geophysics, Geosystems, 20(8):4001-4024, 2019. 
Gordon A MacDonald and Takashi Katsura. Chemical composition of hawaiian lavas. 
Journal of petrology, 5(1):82-133, 1964. 
Jean-Claude Mareschal and Claude Jaupart. Radiogenic heat production, thermal regime 
and evolution of continental crust. Tectonophysics, 609:524-534, 2013. 
Gregor Markl, Masaaki Musashi, and Kurt Bucher. Chlorine stable isotope composition 
of granulites from lofoten, norway: Implications for the cl isotopic composition and 
for the source of cl enrichment in the lower crust. Earth and Planetary Science Letters, 
150(1-2):95-102, 1997. 
236 
H Martin. Effect of steeper archean geothermal gradient on geochemistry of subduction?
zone magmas. Geology, 14(9):753-756, 1986. 
A McCarthy, TJ Falloon, I Sauermilch, JM Whittaker, K Niida, and DH Green. Re?
visiting the australian-antarctic ocean-continent transition zone using petrological and 
geophysical characterization of exhumed subcontinental mantle. Geochemistry, Geo?
physics, Geosystems, 21(7):e2020GC009040, 2020. 
W F McDonough and S-s Sun. The composition of the Earth. Chemical Geology, 120 
(3):223-253, 1995a. ISSN 0009-2541. doi: 10.1016/0009-2541(94)00140-4. 
William F McDonough. Constraints on the composition of the continental lithospheric 
mantle. Earth and Planetary Science Letters, 101(1):1-18, 1990. 
William F McDonough. Partial melting of subducted oceanic crust and isolation of its 
residual eclogitic lithology. Philosophical Transactions of the Royal Society of London.
Series A: Physical and Engineering Sciences, 335(1638):407-418, 1991. 
William F McDonough and S-S Sun. The composition of the earth. Chemical geology,
120(3-4):223-253, 1995b. 
William F McDonough, Ondrej Sramek, and Scott A Wipperfurth. Radiogenic power 
and geoneutrino luminosity of the earth and other terrestrial bodies through time. Geo?
chemistry, Geophysics, Geosystems, 21(7):e2019GC008865, 2020. 
EH McKee. Tertiary igneous chronology of the great basin of western united states?
implications for tectonic models. Geological Society ofA merica Bulletin, 82(12):3497-
3502, 1971. 
237 
W D Mooney. Crust and Lithospheric Structure-Global Crustal Structure, volume 1. 
Oxford: Elsevier, 2015. 
Walter D Mooney, Gabi Laske, and T Guy Masters. CRUST 5.1: A global crustal model 
at 5? x 5?. Journal of Geophysical Research: Solid Earth, 103(Bl):727-747, 1998a. 
Walter D Mooney, Gabi Laske, and T Guy Masters. CRUST 5.1: A global crustal model 
at 5?x 5?. Journal of Geophysical Research: Solid Earth, 103(Bl):727-747, 1998b. 
ISSN 01480227. doi: 10.1029/97JB02122. 
Takashi Nakagawa and Paul J Tackley. Influence of initial cmb temperature and other pa?
rameters on the thermal evolution of earth's core resulting from thermochemical spher?
ical mantle convection. Geochemistry, Geophysics, Geosystems, 11(6), 2010. 
K Douglas Nelson, Wenjin Zhao, LD Brown, J Kuo, Jinkai Che, Xianwen Liu, SL Klem?
perer, Y Makovsky, RJJM Meissner, James Mechie, et al. Partially molten middle 
crust beneath southern tibet: synthesis of project indepth results. Science, 274(5293): 
1684-1688, 1996. 
Andrew A Nyblade and Henry N Pollack. A global analysis of heat flow from precam?
brian terrains: implications for the thermal structure of archean and proterozoic litho-
sphere. Journal of Geophysical Research: Solid Earth, 98(B7):12207-12218, 1993. 
Tohru Ohta and Hiroyoshi Arai. Statistical empirical index of chemical weathering in 
igneous rocks: A new tool for evaluating the degree of weathering. Chemical Geology, 
240(3-4):280-297, 2007. 
238 
TM Olugboji, V Lekic, and W McDonough. A statistical assessment of seismic models 
of the us continental crust using bayesian inversion of ambient noise surface wave 
dispersion data. Tectonics, 36(7):1232-1253, 2017. 
Herbert Palme and H St C O'Neill. Cosmochemical estimates of mantle composition. In 
Heinrich D Holland and Karl K Turekian, editors, Treatise on Geochemistry (Second 
Edition), volume 3, pages 1-39. Elsevier, Oxford, 2014. ISBN 978-0-08-098300-4. 
doi: 10.1016/B978-0-08-095975-7 .00201-1. 
Tom Parsons, Nikolas I. Christensen, and Howard G. Wilshire. Velocities of south?
ern Basin and Range xenoliths: Insights on the nature of lower crustal reflectiv?
ity and composition. Geology, 23(2):129-132, 1995. ISSN 0091-7613. doi: 
10.1130/0091-7613(1995)023(0129:VOSBAR)2.3.CO;2. 
Michael E Pasyanos, T Guy Masters, Gabi Laske, and Zhitu Ma. LITHOl.0: An updated 
crust and lithospheric model of the Earth. Journal of Geophysical Research: Solid 
Earth, 119(3):2153-2173, 2014. ISSN 2169-9356. doi: 10.1002/2013JB010626. 
Etta Patacca, Paolo Scandone, Emiliano Di Luzio, Gian Paolo Cavinato, and Maurizio 
Parotto. Structural architecture of the central apennines: Interpretation of the CROP 11 
seismic profile from the adriatic coast to the orographic divide. Tectonics, 27(3), 2008. 
doi: https://doi.org/10.1029/2005TC001917. 
C Pauselli, G Gola, P Mancinelli, E Trumpy, M Saccone, A Manzella, and Giorgio 
Ranalli. A new surface heat flow map of the northern apennines between latitudes 42.5 
239 
and 44.5?N. Geothermics, 81:39-52, 2019. doi: https://doi.org/10.1016/j.geothermics. 
2019.04.002. 
Catherine Phaneuf and Jean-Claude Mareschal. Estimating concentrations of heat pro?
ducing elements in the crust near the sudbury neutrino observatory, ontario, canada. 
Tectonophysics, 622:135-144, 2014. 
Christophe Pinet and Claude J au part. The vertical distribution of radiogenic heat produc?
tion in the precambrian crust of norway and sweden: geothermal implications. Geo-
physical Research Letters, 14(3):260-263, 1987. 
Terry Plank and Donald W Forsyth. Thermal structure and melting conditions in the man?
tle beneath the basin and range province from seismology and petrology. Geochemistry, 
Geophysics, Geosystems, 17(4):1312-1338, 2016. 
A Polat, AW Hofmann, and Minik Thorleif Rosing. Boninite-like volcanic rocks in 
the 3.7-3.8 ga isua greenstone belt, west greenland: geochemical evidence for intra?
oceanic subduction zone processes in the early earth. Chemical geology, 184(3-4): 
231-254, 2002.
Jason R Price and Michael A Velbel. Chemical weathering indices applied to weather?
ing profiles developed on heterogeneous felsic metamorphic parent rocks. Chemical 
geology, 202(3-4):397-416, 2003. 
Daniele Regis, CJ Warren, Catherine Mary Mottram, and Nick MW Roberts. Using 
monazite and zircon petrochronology to constrain the P-T-t evolution of the middle 
crust in the bhutan himalaya. Journal of Metamorphic Geology, 34(6):617-639, 2016. 
240 
Olivier Reubi and Jon Bluntly. A dearth of intermediate melts at subduction zone volca?
noes and the petrogenesis of arc andesites. Nature, 461(7268):1269-1273, 2009. ISSN 
1476-4687. doi: 10.1038/nature08510. 
P. F. Rey, C. Teyssier, and D. L. Whitney. Extension rates, crustal melting, and core 
complex dynamics. Geology, 37(5):391-394, 2009. ISSN 0091-7613. doi: 10.1130/ 
G25460A.1. 
Paul H Roberts, Christopher A Jones, and Arthur R Calderwood. Energy fluxes and ohmic 
dissipation in the earth's core. Earth's core and lower mantle, pages 100-129, 2003. 
Kaylynn M Rockow, Larry A Haskin,* BL Jollif, and David M Fountain. Constraints on 
element mobility associated with the conversion of granulite to eclogite along fractures 
in an anorthositic complex on holsn?y, norway. Journal of Metamorphic Geology, 15 
(3):401-418, 1997. 
NW Rogers and C J Hawkesworth. Proterozoic age and cumulate origin for granulite 
xenoliths, lesotho. Nature, 299(5882):409-413, 1982. 
Robert F Roy, David D Blackwell, and Francis Birch. Heat generation of plutonic rocks 
and continental heat flow provinces. Earth and Planetary Science Letters, 5:1-12, 
1968. 
R. L. Rudnick and S. R. Taylor. The composition and petrogenesis of the lower crust: A
xenolith study. Journal of Geophysical Research: Solid Earth, 92(B13):13981-14005,
1987a. ISSN 01480227. doi: 10.1029/JB092iB13p13981.
Roberta L Rudnick. Making continental crust. Nature, 378(6557):571-578, 1995. 
241 
Roberta L Rudnick and David M Fountain. Nature and composition of the continental 
crust: A lower crustal perspective. Reviews of Geophysics, 33(3):267-309, 1995. ISSN 
1944-9208. doi: 10.1029/95RG01302. 
Roberta L Rudnick and S Gao. Composition of the Continental Crust. In Treatise on 
Geochemistry, pages 1-51. Elsevier, 2014. ISBN 978-0-08-098300-4. doi: 10.1016/ 
B978-0-08-095975-7.00301-6. 
Roberta L Rudnick and Ian Jackson. Measured and calculated elastic wave speeds in 
partially equilibrated mafic granulite xenoliths: Implications for the properties of an 
underplated lower continental crust. Journal of Geophysical Research: Solid Earth, 
lO0(B6):10211-10218, 1995. 
Roberta L Rudnick and Andrew A Nyblade. The thickness and heat production of archean 
lithosphere: constraints from xenolith thermobarometry and surfa ce heat flow. Mantle 
petrology: field observations and high pressure experimentation: a tribute to Francis 
R.(Joe) Boyd, 6:3-12, 1999. 
Roberta L Rudnick and T Presper. Geochemistry of intermediate/-to high-pressure gran?
ulites. In Granulites and crustal evolution, pages 523-550. Springer, 1990. 
Roberta L Rudnick and S R T aylor. The composition and petrogenesis of the lower crust: a 
xenolith study. Journal of Geophysical Research: Solid Earth, 92(B13):13981-14005, 
1987b. 
Roberta L Rudnick, S M McLennan, and S R Taylor. Large ion lithophile elements in 
242 
rocks from high-pressure granulite facies terrains. Geochimica et Cosmochimica Acta, 
49(7):1645-1655, 1985. 
Roberta L Rudnick, Matthias Barth, Ingo Horn, and William F McDonough. Rutile?
bearing refractory eclogites: missing link between continents and depleted mantle. 
Science, 287(5451):278-281, 2000. 
James K Russell, Lucy A Porritt, Yan Lavallee, and Donald B Dingwell. Kimberlite 
ascent by assimilation-fuelled buoyancy. Nature, 481(7381):352-356, 2012. 
T Sakai, K Inoue, H Watanabejg, WF McDonough, N Abe, E Araki, T Kasaya, M Kyo, 
N Sakurai, K Uek, et al. Study of ocean bottom detector for observation of geo-neutrino 
from the mantle. In Journal of Physics: Conference Series, volume 2156, page 012144. 
IOP Publishing, 2021. 
T Sakai, K Inoue, H Watanabe, W F McDonough, N Abe, E Araki, T Kasaya, M Kyo, 
N Sakurai, K Ueki, and H Yoshida. Study of Ocean Bottom Detector for observation 
of geo-neutrino from the mantle. Proceedings for TAUP, 2022. 
L G Sammon, C Gao, and W F McDonough. Lower crustal composition m the 
southwestern united states. Journal of Geophysical Research: Solid Earth, 125(3): 
e2019JB019011, 2020. 
Laura G Sammon and William F McDonough. A geochemical review of amphibolite, 
granulite, and eclogite facies lithologies: Perspectives on the deep continental crust. 
Journal of Geophysical Research: Solid Earth, page e2021JB022791, 2021a. 
243 
Laura G Sammon and William F McDonough. A geochemical review of amphibolite, 
granulite, and eclogite facies lithologies: Perspectives on the deep continental crust. 
Journal of Geophysical Research: Solid Earth, 2021b. 
Laura G Sammon, William F McDonough, and Walter D Mooney. The composition of 
the deep continental crust inferred from geochemical and geophysical data. Journal of 
Geophysical Research: Solid Earth, page in review, 2022. doi: https:/ /doi.org/10.1002/ 
essoar. l 0507 582.1. 
P. Schaaf, W. Heinrich, and T. Besch. Composition and Sm-Nd isotopic data of the
lower crust beneath San Luis Potosf, central Mexico: Evidence from a granulite-facies
xenolith suite. Chemical Geology, 118(1):63-84, 1994. ISSN 0009-2541. doi: 10.
1016/0009-2541(94)90170-8.
FR Schilling and GM Partzsch. Quantifying partial melt fraction in the crust beneath the 
central andes and the tibetan plateau. Physics and Chemistry of the Earth, Part A: Solid 
Earth and Geodesy, 26(4-5):239-246, 2001. 
Brandon Schmandt, Fan-Chi Lin, and Karl E Karlstrom. Distinct crustal isostasy trends 
east and west of the rocky mountain front. Geophysical Research Letters, 42(23): 10-
290, 2015. 
M Schmitz, W-D Heinsohn, and FR Schilling. Seismic, gravity and petrological evidence 
for partial melt beneath the thickened central andean crust (21-23 s). Tectonophysics, 
270(3-4):313-326, 1997. 
Vera Schulte-Pelkum, Kevin H Mahan, Weisen Shen, and Josh C Stachnik. The distri-
244 
bution and composition of high-velocity lower crust across the continental us: Com?
parison of seismic and xenolith data and implications for lithospheric dynamics and 
history. Tectonics, 36(8):1455-1496, 2017. 
Derek L. Schutt, Anthony R. Lowry, and Janine S. Buehler. Moho temperature and mo?
bility of lower crust in the western United States. Geology, 46(3):219-222, 2018. ISSN 
0091-7613. doi: 10.1130/G39507.l. 
MP Searle, JM Cottle, MJ Streule, and DJ Waters. Crustal melt granites and migmatites 
along the himalaya: melt source, segregation, transport and granite emplacement mech?
anisms. Earth and Environmental Science Transactions of the Royal Society of Edin-
burgh, 100(1-2):219-233, 2009. 
J Semprich and N S C Simon. Inhibited eclogitization and consequences for geo?
physical rock properties and delamination models: Constraints from cratonic lower 
crustal xenoliths. Gondwana Research, 25(2):668-684, 2014. ISSN 1342-937X. doi: 
10.1016/j.gr.2012.08.018. 
Weisen Shen and Michael H. Ritzwoller. Crustal and uppermost mantle structure beneath 
the United States. Journal of Geophysical Research: Solid Earth, 121(6):4306-4342, 
2016. ISSN 2169-9356. doi: 10.1002/2016JB012887. 
Weisen Shen, Michael H. Ritzwoller, and Vera Schulte-Pelkum. A 3-D model of the 
crust and uppermost mantle beneath the Central and Western US by joint inversion 
of receiver functions and surface wave dispersion. Journal of Geophysical Research: 
Solid Earth, 118(1):262-276, 2013. ISSN 2169-9356. doi: 10.1029/2012JB009602. 
245 
Weisen Shen, Douglas A Wiens, Andrew J Lloyd, and Andrew A Nyblade. A geothermal 
heat flux map of antarctica empirically constrained by seismic structure. Geophysical
Research Letters, 47(14):e2020GL086955, 2020. 
William J Shinevar, Mark D Behn, Greg Hirth, and Oliver Jagoutz. Inferring crustal 
viscosity from seismic velocity: Application to the lower crust of Southern California. 
Earth and Planetary Science Letters, 494:83-91, 2018. ISSN 0012-821X. doi: 10. 
1016/j.epsl.2018.04.055. 
G Paolo Sighinolfi. Investigations into deep crustal levels: fractionating effects and geo?
chemical trends related to high-grade metamorphism. Geochimica et Cosmochimica
Acta, 35(10):1005-1021, 1971. 
Ondrej Sramek, William F. McDonough, Edwin S. Kite, Vedran Lekic, Stephen T. Dye, 
and Shijie Zhong. Geophysical and geochemical constraints on geoneutrino fluxes 
from Earth's mantle. Earth and Planetary Science Letters, 361:356-366, 2013. doi: 
10.1016/j.epsl.2012.11.001. 
H-G Stosch, G W Lugmair, and HA Seek. Geochemistry of granulite-facies lower crustal
xenoliths: implications for the geological history of the lower continental crust below
the eifel, west germany. Geological Society, London, Special Publications, 24(1):309-
317, 1986.
Virginia Strati, Marica Baldoncini, Ivan Callegari, Fabio Mantovani, William F Mc?
Donough, Barbara Ricci, and Gerti Xhixha. Expected geoneutrino signal at juno. 
Progress in Earth and Planetary Science, 2(1): 1-7, 2015. 
246 
S-s Sun and William F McDonough. Chemical and isotopic systematics of oceanic
basalts: Implications for mantle composition and processes. Geological Society, Lon-
don, Special Publications, 42(1):313-345, 1989. ISSN 0305-8719, 2041-4927. doi:
10.1144/GSL.SP.1989.042.01.19.
H Svensen, Bjorn Jamtveit, D Austrheim Banks, and Hakon Austrheim. Halogen contents 
of eclogite facies fluid inclusions and minerals: Caledonides, western norway. Journal 
of Metamorphic Geology, 19(2):165-178, 2001. 
Wolfgang Szwillus, Juan Carlos Afonso, Jorg Ebbing, and Walter D Mooney. Global 
crustal thickness and velocity structure from geostatistical analysis of seismic data. 
Journal of Geophysical Research: Solid Earth, 124(2):1626-1652, 2019. 
Ming Tang, William F McDonough, and Richard D Ash. Europium and strontium anoma?
lies in the morb source mantle. Geochimica et Cosmochimica Acta, 197:132-141, 
2017. 
Stuart Ross Taylor and Scott M McLennan. The continental crust: its composition and 
evolution. Blackwell Scientific Pub., Palo Alto, CA, 1985. 
Tatsuki Tsujimori, Virginia B Sisson, Juhn G Liou, George E Harlow, and Sorena S 
Sorensen. Very-low-temperature record of the subduction process: A review of world-
wide lawsonite eclogites. Lithos, 92(3-4):609-624, 2006. 
Donald Turcotte and Gerald Schubert. Geodynamics. Cambridge University Press, 2014. 
G. Nell Tyner and Douglas Smith. Peridotite xenoliths in silica-rich, potassic latite from
the transition zone of the Colorado Plateau in north-central Arizona. Contributions
247 
to Mineralogy and Petrology, 94(1):63-71, 1986. ISSN 1432-0967. doi: 10.1007/ 
BF00371227. 
Greg A. Valentine and Frank V. Perry. Tectonically controlled, time-predictable basaltic 
volcanism from a lithospheric mantle source (central Basin and Range Province, USA). 
Earth and Planetary Science Letters, 261(1):201-216, 2007. ISSN 0012-821X. doi: 
10.1016/j.epsl.2007 .06.029. 
PWC Van Calsteren, NBW Harris, CJ Hawkesworth, MA Menzies, and NW Rogers. 
Xenoliths from southern africa: a perspective on the lower crust. Geological Society, 
London, Special Publications, 24(1):351-362, 1986. 
Massimo Verdoya, Paolo Chiozzi, and Gianluca Gola. Unravelling the terrestrial heat 
flow of a young orogen: The example of the northern Apennines. Geothermics, 90: 
101993, 2021. doi: https://doi.org/10.1016/j.geothermics.2020.101993. 
G Vignaroli, M Mancini, F Bucci, M Cardinali, GP Cavinato, M Moscatelli, ML Putig?
nano, P Sirianni, M Santangelo, F Ardizzone, et al. Geology of the central part of the 
amatrice basin (central apennines, italy). Journal of Maps, 15(2):1 93-202, 2019. 
K Hans Wedepohl. The composition of the continental crust. Geochimica et cosmochim-
ica Acta, 59(7):1 217-1232, 1995. 
Bruce H Wilkinson, Brandon J McElroy, Stephen E Kesler, Shanan E Peters, and Ed?
ward D Rothman. Global geologic maps are tectonic speedometers-rates of rock 
cycling from area-age frequencies. Geological Society of America Bulletin, 121(5-6): 
760-779, 2009.
248 
Scott A Wipperfurth, Meng Guo, Ondrej Sramek, and William F McDonough. Earth's 
chondritic Th/U: Negligible fractionation during accretion, core formation, and 
crust-mantle differentiation. Earth and Planetary Science Letters, 498:1 96-202, 2018. 
ISSN 0012-821X. doi: 10.1016/j.epsl.2018.06.029. 
Scott A Wipperfurth, Ondrej Sramek, and William F McDonough. Reference models for 
lithospheric geoneutrino signal. Journal of Geophysical Research: Solid Earth, 125 
(2):e2019JB018433, 2020. 
M Xhixha, Marica Baldoncini, GP Bezzon, GP Buso, L Carmignani, L Casini, I Callegari, 
T Colonna, S Cuccuru, E Guastaldi, et al. A detailed gamma-ray survey for estimating 
the radiogenic power of sardinian variscan crust. 2014. 
Toru Yamasaki. The role of bimodal magmatism in seafloor massive sulfide (sms) ore?
forming systems at the middle okinawa trough, japan. Ocean Science Journal, 53(2): 
413-436, 2018.
Takashi Yoshizaki and William F McDonough. Earth and mars-distinct inner solar system 
products. Geochemistry, 81(2):125746, 2021. 
E-An Zen. Phase relations of peraluminous granitic rocks and their petrogenetic implica?
tions. Annual Review of Earth and Planetary Sciences, 16(1):21-51, 1988.
249