ARCHEAN EVOLUTION OF THE SOUTHERN SÃO

FRANCISCO CRATON (SE BRAZIL)

i
ii
FUNDAÇÃO UNIVERSIDADE FEDERAL DE OURO PRETO
Reitor
Cláudia Aparecida Marliére de Lima
Vice-Reitor
Hermínio Arias Nalini Júnior
Pró-Reitor de Pesquisa e Pós-Graduação
Sérgio Francisco de Aquino

ESCOLA DE MINAS
Diretor
Issamu Endo
Vice-Diretor
José Geraldo Arantes de Azevedo Brito

DEPARTAMENTO DE GEOLOGIA
Chefe
Luis Antônio Rosa Seixas

iii
EVOLUÇÃO CRUSTAL E RECURSOS NATURAIS

iv
 CONTRIBUIÇÕES ÀS CIÊNCIAS DA TERRA - VOL. 349

TESE DE DOUTORAMENTO

ARCHEAN EVOLUTION OF THE SOUTHERN SÃO FRANCISCO

CRATON (SE BRAZIL)

Capucine Albert

Orientador

Dr. Cristiano de Carvalho Lana

Co-orientador

Dr. Federico Farina

Tese apresentada ao Programa de Pós-Graduação em Evolução Crustal e Recursos Naturais do

Departamento de Geologia da Escola de Minas da Universidade Federal de Ouro Preto como requisito
parcial à obtenção do Título de Doutor, Área de Concentração: Tectônica, Petrogênese e Recursos
Minerais

OURO PRETO

2017

v
vii
viii
 Acknowledgements
Many people have contributed toward the realization of this project, bringing their geological
knowledge, financial or logistic support and moral encouragements, and I have every person to
acknowledge and thank for their contribution, whether small or big. Thank you!

Cristiano, I am indebted to you for providing me with the opportunity to do research, and for
all your support throughout the years. I am particularly grateful for your enthusiasm, your trust, and
your availability. Thank you for the time and effort you dedicate to your students, providing us with
the opportunity to be involved in scientific discussions, and for letting us follow our own research
interests and ideas. I feel truly privileged to have been a part of the dynamic group you have built, and
to have had you as a mentor.

Federico, your patience and your advices as I was taking my first steps into the world of
research have really been priceless. Thank you for your support, your friendship, all the good laughs,
and for your blunt honesty, which has me question and improve myself every time.

I would like to acknowledge all those who, by sharing their knowledge have contributed one
way or another to this project. I want to give a special mention to Craig Storey and Axel Gerdes, for
your valuable collaboration and the stimulating discussions every time our paths have crossed. Thanks
to the both of you, as well as Márcio Pimentel and Ricardo Scholz for having accepted to read and
evaluate this work. I would like to extend my thanks to Gary Stevens, Jean-François Moyen, Fernando
Alkmim and Ian Buick, with whom I have had the chance to discuss and exchange ideas on a regular
basis.

I am grateful to all the people I have met during these four years, who have enriched my life
and experience in the lab and outside, I couldn‘t have done it without all of you. Special mention to G
and Cynthia, for all the adventures and travels, the laughter, the encouragements and your friendship,
see you soon! To all my fellow officemates, Francesco, Mari, Fabiana, Marcha and Kai, and all the
others, part of the group in Ouro Preto, Hugo Boss, Dr. C, Leo, Carmen 1, Carmen 2, Ana, Lorena,
Jessica and many others, as well as Camille et Julien (alias Gulen) sur la fin. You guys rock! Thanks
to all the crew in Stellenbosch, Shawn Kittchy, Tolene, Stephan the Red, Duncan, Andrea, Simon,
Marcos and Valby, for all the wine and the good times (Ons gaan nou braai!). And finally, last but not
least, Matzi, I am so lucky to have you by my side.

Mum and Dad, thank you for your unfailing support before and during my PhD, for cultivating
our curiosity, and for letting us pursue what we are passionate about.

ix
x
 Summary
ACKNOWLEDGEMENTS ................................................................................................................ IX
SUMMARY ......................................................................................................................................... XI
LIST OF FIGURES ........................................................................................................................... XV
LIST OF TABLES ..........................................................................................................................XVII
ABSTRACT ...................................................................................................................................... XIX
RESUMO .......................................................................................................................................... XXI
CHAPTER 1 - INTRODUCTION ....................................................................................................... 1
1.1 GENERAL INTRODUCTION ........................................................................................... 1
1.2 THE SECULAR EVOLUTION OF THE CONTINENTAL CRUST................................. 5
1.2.1 Crustal growth ........................................................................................................ 5
1.2.2 Geodynamic evolution .......................................................................................... 10
1.3 ARCHEAN CRATONS ................................................................................................... 15
1.3.1 Archean granitoids ................................................................................................ 15
1.3.2 Greenstone belts.................................................................................................... 24
1.3.3 Architecture of Archean granite-greenstone terranes ........................................... 26
1.4 CONCLUSIONS .............................................................................................................. 28
CHAPTER 2 - THE ARCHEAN-PALEOPROTEROZOIC EVOLUTION OF THE
QUADRILÁTERO FERRÍFERO (BRAZIL): CURRENT MODELS AND OPEN QUESTIONS
............................................................................................................................................................... 31
ABSTRACT................................................................................................................................ 31
2.1 INTRODUCTION ............................................................................................................ 32
2.2 HISTORICAL BACKGROUND AND GEOGRAPHIC BOUNDARIES ....................... 32
2.3 MAIN GEOLOGICAL FEATURES ................................................................................ 34
2.4 THE ARCHEAN METAMORPHIC COMPLEXES........................................................ 36
2.4.1 Field relationships and petrography...................................................................... 36
2.4.2 Geochemistry: medium- and high-K granitoids ................................................... 38
2.4.3 Geochronology ..................................................................................................... 41
2.5 THE RIO DAS VELHAS SUPERGROUP ....................................................................... 43
2.5.1 Stratigraphy........................................................................................................... 43
2.5.2 Geochronology ..................................................................................................... 45
2.6 THE MINAS SUPERGROUP .......................................................................................... 47
2.6.1 Main stratigraphic features ................................................................................... 47
2.6.2 Tamanduá and Caraça Groups .............................................................................. 47
2.6.3 Itabira Group......................................................................................................... 49
2.6.4 Piracicaba Group .................................................................................................. 50
2.6.5 Sabará Group ........................................................................................................ 50
2.7 THE ITACOLOMI GROUP ............................................................................................. 53
xi
 2.8 MAFIC AND INTERMEDIATE DIKES, BOUDINS AND LENSES ............................. 54
2.9 MAIN STRUCTURES AND STRUCTURAL EVOLUTION ......................................... 55
2.10 DISCUSSION: MODELS AND OPEN PROBLEMS ...................................................... 56
2.10.1 Building up the continental crust: the Paleo- to Mesoarchean rock record ........ 56
2.10.2 The record of plate tectonics: the Meso-Neoarchean .......................................... 57
2.10.3 Evolution of the Minas Basin.............................................................................. 59
2.10.4 The Minas accretionary orogeny ......................................................................... 59
2.10.5 Questions for future research .............................................................................. 62
CHAPTER 3 - THE NEOARCHEAN TRANSITION BETWEEN MEDIUM- AND HIGH-K
GRANITOIDS: CLUES FROM THE SOUTHERN SÃO FRANCISCO CRATON (BRAZIL) . 65
ABSTRACT ................................................................................................................................ 65
3.1 INTRODUCTION ............................................................................................................ 66
3.2 REGIONAL GEOLOGY .................................................................................................. 67
3.2.1 The Southern São Francisco craton....................................................................... 67
3.2.2 Gneiss-granitoid complexes and supracrustal sequences ...................................... 68
3.2.3 Tectono-metamorphic and magmatic evolution.................................................... 69
3.3 ANALYTICAL TECHNIQUES ....................................................................................... 71
3.4 RESULTS ......................................................................................................................... 72
3.4.1 Field relationships ................................................................................................. 72
3.4.2 Petrography ........................................................................................................... 74
3.4.3 Whole-rock geochemistry ..................................................................................... 76
3.4.4 Geochronology: cathodoluminescence images and U–Pb ages ............................ 82
3.5 DISCUSSION ....................................................................................................................... 87
3.5.1 Medium- and high-K granites in the Southern São Francisco craton ................... 87
3.5.2 Petrogenesis of gneisses and granitoids ................................................................ 88
3.5.3 Chemical evolution of the continental crust.......................................................... 94
3.5.4 The emergence of sedimentary basins in the Neoarchean .................................... 95
3.6 CONCLUSIONS............................................................................................................... 97
CHAPTER 4 - ARCHEAN CRUSTAL EVOLUTION IN THE SOUTHERN SÃO FRANSISCO
CRATON, BRAZIL: CONSTRAINTS FROM U-PB, LU-HF AND O ISOTOPE ANALYSES . 99
ABSTRACT ................................................................................................................................ 99
4.1 INTRODUCTION .......................................................................................................... 100
4.2 GEOLOGICAL SETTING ............................................................................................. 101
4.3 ANALYTICAL TECHNIQUES ..................................................................................... 103
4.4 RESULTS ....................................................................................................................... 105
4.4.1 Geochronological results..................................................................................... 116
4.4.2 Lu-Hf isotopes .................................................................................................... 117
4.4.3 O isotopes............................................................................................................ 122

xii
 4.5 DISCUSSION ................................................................................................................. 123
4.5.1 Origin of the Meso- and Neoarchean granitoids ................................................. 123
4.5.2 The growth of the continental crust in the SSFC ................................................ 133
4.5.3 A geodynamic model for the SSFC .................................................................... 137
4.6 CONCLUSION .............................................................................................................. 139
CHAPTER 5 - BORON ISOTOPE COMPOSITION (LA MC-ICP-MS) OF TOURMALINE
FROM THE QUADRILÁTERO FERRÍFERO DISTRICT, SE BRAZIL: CONSTRAINTS ON
ARCHEAN MAGMATIC-HYDROTHERMAL EVOLUTION .................................................. 141
ABSTRACT.............................................................................................................................. 141
5.1 INTRODUCTION .......................................................................................................... 142
5.2 REGIONAL GEOLOGY ................................................................................................ 142
5.3 SAMPLE LOCATIONS AND TOURMALINE OCCURRENCES............................... 144
5.4 ANALYTICAL METHODS .......................................................................................... 152
5.5 RESULTS ....................................................................................................................... 154
5.5.1 Tourmaline classification and major element variations .................................... 154
5.5.2 Boron isotope compositions ............................................................................... 157
5.6 DISCUSSION ................................................................................................................. 158
5.6.1 Tourmaline growth ............................................................................................. 158
5.6.2 Boron isotope variations and fluid evolution ...................................................... 159
5.6.3 B isotope constraints on fluid mixing ................................................................. 163
5.7 CONCLUSIONS ............................................................................................................ 164
CHAPTER 6 - GENERAL CONCLUSIONS ................................................................................. 167
REFERENCES .................................................................................................................................. 169
APPENDICES ................................................................................................................................... 193

xiii
xiv
 List of figures
Figure 1.1 – Map showing the distribution of exposed Archean crust and craton boundaries ............... 2
Figure 1.2 – Variations of different geological and geochemical parameters through time .................. 9
Figure 1.3 – Histograms showing ages of preserved plate tectonic indicators for the last 3 Ga .......... 11
Figure 1.4 – Classification of orogens.................................................................................................. 13
Figure 1.5 – Schematic cross sections of early Earth and young Earth................................................ 14
Figure 1.6 – Field photographs of TTGs. ............................................................................................. 16
Figure 1.7 – Major and trace element characteristics of TTGs ............................................................ 17
Figure 1.8 – Schematic summary of the most accepted mechanism of TTG production ..................... 18
Figure 1.9 – Diagrams illustrating the two main ideas for the geodynamic sites of TTG generation .. 20
Figure 1.10 – Field photographs of late Archean granitoids ................................................................ 22
Figure 1.11 – Trace-element geochemistry of late Archean granitoids and TTGs .............................. 23
Figure 1.12 – General stratigraphic column of the Swaziland Supergroup, Barberton greenstone belt,
South Africa .................................................................................................................................... 26
Figure 1.13 – Schematic structural map of the East Pilbara craton...................................................... 27
Figure 1.14 – Schematic model of partial convective overturn for the east Pilbara craton .................. 28
Figure 2.1 – Geological map of the Quadrilátero Ferrífero.................................................................. 33
Figure 2.2 – Stratigraphic column of the supracrustal sequences in Quadrilátero Ferrífero ................ 35
Figure 2.3 – Field photographs of the basement .................................................................................. 37
Figure 2.4 – Normative An-Ab-Or triangle.......................................................................................... 39
Figure 2.5 – Harker diagrams for the igneous rocks of the Quadrilátero Ferrífero .............................. 40
Figure 2.6 – Average chondrite-normalized REE patterns for high-K granites and medium-K gneisses
and granitoids .................................................................................................................................. 41
Figure 2.7 – Timeline showing 207Pb/206Pb zircon ages for intrusive and volcanic rocks .................... 42
Figure 2.8 – Field photographs of the supracrustal rocks .................................................................... 44
Figure 2.9 – Frequency histogram showing the age distribution of detrital zircon grains in the rocks of
the Rio das Velhas Supergroup ....................................................................................................... 46
Figure 2.10 – Frequency histogram for the Minas Supergroup............................................................ 53
Figure 2.11 – Sketch of the geodynamic evolution of the Quadrilátero Ferrífero ............................... 61
Figure 3.1 – Geological map of the study area ..................................................................................... 68
Figure 3.2 – Field photographs............................................................................................................. 73
Figure 3.3 – Normative An-Ab-Or triangle.......................................................................................... 76
Figure 3.4 – Harker diagrams showing the major element features of gneisses, granites and
leucogranites ................................................................................................................................... 78
Figure 3.5 – Harker diagrams showing the trace element composition of gneisses, granites and
leucogranites ................................................................................................................................... 79
Figure 3.6 – Average chondrite-normalized REE patterns................................................................... 79
Figure 3.7 – Cathodoluminescence images of zircon grains. ............................................................... 82
Figure 3.8 – Representative Concordia diagrams................................................................................. 86
xv
Figure 3.9 – Ternary classification diagram for late Archean granitoids ............................................. 87
Figure 3.10 – Sketch of the model proposed for the petrogenesis of gneisses and granitoids in the
Bação and Bonfim complexes ......................................................................................................... 91
Figure 3.11 – Ternary plots comparing the composition of gneisses, granitic rocks and leucogranites
with the composition of experimental glasses ................................................................................. 93
Figure 3.12 – Timeline showing 207Pb/206Pb zircon ages for intrusive and volcanic rocks .................. 95
Figure 4.1 – Geological map of the Southern São Francisco craton with sample locations ............... 102
Figure 4.2 – 176Hf/177Hft vs. apparent 207Pb/206Pb age diagrams for each magmatic event ................. 120
Figure 4.3 – δ18O vs. intrusion age diagram ....................................................................................... 122
Figure 4.4 – εHft vs. age diagrams ..................................................................................................... 124
Figure 4.5 – Plots of εHft vs. geochemical parameters ....................................................................... 127
Figure 4.6 – δ18O vs. U-Th-Pb systematics diagrams ......................................................................... 129
Figure 4.7 – (a) U-Pb age distribution for detrital zircons from the SSFC. (b) εHft vs. age diagram. 136
Figure 4.8 – Sketch illustrating the evolution of the Archean continental crust................................. 138
Figure 5.1 – Geological map of the Quadrilátero Ferrífero and Bação complex................................ 144
Figure 5.2 – Field photographs ........................................................................................................... 146
Figure 5.3 – Photomicrographs of tourmaline occurrences in thin sections ....................................... 148
Figure 5.4 – Back-scattered electron (BSE) images of representative tourmaline. ............................ 149
Figure 5.5 – B isotope compositions obtained for Reference Materials (RM) ................................... 153
Figure 5.6 – Ternary diagrams showing tourmaline compositions..................................................... 155
Figure 5.7 – Chemical compositions of tourmaline ............................................................................ 156
Figure 5.8 – Plots of (a) MgO (wt.%) vs CaO (wt.%) and (b) Al2O3 (wt.%) vs TiO2 (wt.%) ............ 157
Figure 5.9 –Frequency distribution diagrams of tourmaline B isotope composition.......................... 162

xvi
 List of tables
Table 3.1 – Major and trace elements composition of orthogneisses and granites from the Bação,
Bonfim and Belo Horizonte complexes. ......................................................................................... 80
Table 3.2 – Summary of U–Pb ages for gneisses, granitoids and leucogranites of the Bação, Bonfim
and Belo Horizonte complexes ....................................................................................................... 84
Table 4.1 – Operating conditions and instrumental settings ............................................................... 104
Table 4.2 – Selection of representative analyses of U-Pb, Lu-Hf and O isotopes.............................. 106
Table 4.3 – Summary of ages and isotopic compositions for samples analysed in this study............ 118
Table 5.1 – Summary of tourmaline samples investigated during this study ..................................... 150
Table 5.2 – Summary of LA MC-ICP-MS B isotope analyses of reference tourmaline. ................... 154

xvii
xviii
 Abstract
Modern continents are fundamentally different from their Archean counterparts, owing to the
influence of a substantially hotter mantle on the production and rheological properties of the crust in
the early Earth. Particularly, the nature and composition of the igneous rock record underwent
significant modifications at the end of the Archean eon, attesting to important changes in geodynamic
processes. This work focuses on the granitoid rocks that were emplaced in the southern tip of the São
Francisco craton in southeast Brazil. Combined field and petrographic observations, whole rock
geochemistry and isotope (U-Pb, Lu-Hf, O and B) analyses were performed in order to place better
constraints on the mechanisms that led to the formation and stabilization of this cratonic block, and
understand the global geodynamic modifications that occurred during this period. The basement of the
Quadrilátero Ferrífero district, in the southern São Francisco craton, consists of orthogneisses,
intruded by abundant granitoid plutons and associated products of granitic magmatism (i.e.
pegmatitic/aplitic veins). The basement in the Quadrilátero Ferrífero records three main periods of
magmatism, namely the Rio das Velhas I (RVI) and II (RVII) events at 2.92-2.85 and 2.80-2.76 Ga
respectively, and the latest Mamona event (2.75-2.68 Ga). Granitoids and gneisses of the three studied
basement complexes (Bação, Bonfim and Belo Horizonte) have been subdivided into two groups,
reflecting different petrogenetic processes and/or sources: medium- and high-K rocks. Medium-K
gneisses and granitoids were formed during the Rio das Velhas I and II events. These rocks have some
similarities with Archean TTGs, despite a small enrichment in SiO2, K2O and depletion in Al2O3, and
are thought to result from mixing between an end-member derived by partial melting of a metabasaltic
rock and a melt resulting from recycling of older TTG crust. On the other hand, high-K granites
emplaced during the later Mamona event resemble typical late Archean biotite- granites, and are
inferred to result from the partial melting of immature metasediments. The involvement of
metasediments in the petrogenesis of the Neoarchean high-K magmas is further supported by: (i) a
trend towards heavy δ18O(Zrc) values, indicating O isotope equilibration at near-surface conditions, and
(ii) the presence of boron-rich magmatic melts and fluids emplaced at 2.70-2.60 Ga, likely derived
from a boron-rich metasedimentary protolith. Overall, we propose that the magmatic and geodynamic
evolution of the southern São Francisco craton unfolded in the following way. From ~ 3.50 to 2.90 Ga,
fragments of juvenile crust (possibly TTGs) were produced, forming the protolith of the present-day
southern São Francisco craton. This period is one of net crustal growth, as evidenced by hafnium
isotope modelling from detrital zircons. At ~ 2.90 Ga, the crust witnessed significant modifications,
owing to the onset of collisional tectonics. Several crustal fragments with distinct histories were
progressively assembled, ultimately building a strong and emerged continental nucleus. This is
supported by: (i) the transition into a regime dominated by crustal reworking, with decreasing mantle
input into newly generated magmas, (ii) the involvement of older TTG components in the generation
of medium-K magmas during the RVI and RVII events, and the paralleled absence of preserved

xix
crustal materials older than ~ 2.90 Ga in the southern São Francisco craton, (iii) the abundance of
upper-crustal hydrothermal systems evidenced by light δ18O(Zrc) values, indicating the emergence of
large portions of the continental crust in the Neoarchean, (iv) the progressive crustal maturation,
recorded by a trend towards more potassic and HFSE (High Field Strength Element)-rich magmatism.
The emergence and rapid burial of large clastic sedimentary basins in the Neoarchean (~ 2.75 Ga) led
to the production of high-K granitoids during the Mamona event. Contemporaneous widespread
tourmaline-bearing differentiated magmatic fluids percolated the crust and interacted with the nearby
greenstone belt. The presence of hydrothermal tourmaline of magmatic (and Neoarchean) origin in
close vicinity of the basement complexes suggests that the formation of the dome-and-basin structures
occurred prior to the last magmatic event (i.e. before 2.70 Ga), which is at odds with the current idea
that the region acquired its present-day architecture in the Paleoproterozoic.

xx
 Resumo
Os continentes modernos são fundamentalmente diferentes de suas contrapartes Arqueanas,
devido à influência de um manto substancialmente mais quente sobre a produção e as propriedades
reológicas da crosta no início da Terra. Particularmente, a natureza e a composição do registro de
rochas ígneas sofreram modificações significativas no final do eon Arqueano, atestando mudanças
importantes nos processos geodinâmicos. Este trabalho concentra-se nas rochas granitóides
posicionadas na porção sul do craton São Francisco no sudeste do Brasil. Uma combinação de
observações de campo e petrográficas, geoquímica de rocha total e análises de isotópicas (U-Pb, Lu-
Hf, O e B) foram realizadas para um melhor entendimento nos mecanismos que levaram à formação e
estabilização deste bloco cratônico, além de compreender as modificações geodinâmicas globais
ocorridas durante esse período. O embasamento do Quadrilátero Ferrífero, no sul do cráton São
Francisco, consiste em ortognaisses, intrudidos por abundantes plútons granitóides e associados
produtos magmáticos (veios pegmatíticos/aplíticos, por exemplo). O embasamento do Quadrilátero
Ferrífero registra três períodos principais de magmatismo, denominados de Rio das Velhas I (RVI) e II
(RVII) de 2.92-2.85 e 2.80-2.76 Ga, respectivamente, e o último evento Mamona (2.75-2.68 Ga).
Granitóides e gnaissses dos três complexos do embasamento estudados (Bação, Belo Horizonte e
Bonfim) foram subdivididos em dois grupos, refletindo diferentes processos petrogenéticos e/ou
fontes: rochas de médio- e alto-K. Os gnaisses e granitóides de médio-K foram formados durante os
eventos Rio das Velhas I e II. Essas rochas apresentam algumas semelhanças com os TTG Arqueanos,
apesar de possuírem um pequeno enriquecimento em SiO2 e K2O e depleção em Al2O3, e foram
interpretadas como o resultado da mistura entre um membro final derivado da fusão parcial de uma
rocha metabasáltica e um fundido resultante da reciclagem de uma crosta mais antiga de composição
TTG. Por outro lado, os granitos de alto-K colocados na crosta durante subsequente evento Mamona,
assemelham-se aos típicos granitóides à biotita do Arqueano tardio, com origem interpretada da fusão
parcial de metassedimentos imaturos. O envolvimento de metassedimentos na petrogênese destes
magmas de alto-K Neoarqueanos é ainda suportado por: (i) uma tendência para valores pesados de
δ18O(Zrc), indicando equilíbrio isótopico de O em condições de subsuperfície, e (ii) a presença fundidos
magmáticos ricos em boro e fluídos intrudidos em 2.70-2.60 Ga, provavelmente derivados de um
protólito metassedimentar rico em boro. Em geral, propomos que a evolução magmática e
geodinâmica da porção sul do craton São Francisco ocorreu da seguinte maneira. De ~ 3.50 a 2.90 Ga,
fragmentos de crosta juvenil (possivelmente TTGs) foram produzidos, formando o protólito do atual
sul do cráton São Francisco. Este é um período de crescimento crustal, como evidenciado pela
modelagem isotópica de hafnium à partir de zircões detríticos. Em ~ 2.90 Ga, a crosta sofreu
significativas modificações, devido ao início de uma tectônica colisional. Diversos fragmentos crustais
com histórias distintas foram acrescidos progressivamente, construindo, por fim, um núcleo
continental rígido e soerguido. Isto é suportado por: (i) a transição para um regime dominado por

xxi
reciclagem crustal, com diminuição de contribuição mantélica em magmas recém-gerados, (ii) o
envolvimento de componentes TTG mais antigos na geração de magmas de médio-K durante os
eventos RVI e RVII e a simultânea ausência de materiais crustais preservados mais velhos que ~ 2.90
Ga na porção sul do craton São Francisco, (iii) a abundância de sistemas hidrotermais na porção
superior da crosta evidenciados por valores leves de δ18O(Zrc), indicando o soerguimento de grandes
porções da crosta continental durante o Neoarqueano, (iv) a progressiva maturação crustal, registrada
por uma tendência para o magmatismo mais potássico e rico em HFSE (High Field Strength Element).
O soerguimento e rápido soterramento de grandes bacias sedimentares dominadas por componentes
clásticos no Neoarqueano (~ 2.75 Ga) levaram à produção de granitóides de alto-K durante o evento
Mamona. Fluidos magmáticos diferenciados portadores de turmalina percolaram a crosta e interagiram
de forma generalizada com o greenstone belt adjacente. A presença de turmalina hidrotermal de
origem magmática (e Neoarqueana) em associação espacial próxima aos complexos do embasamento
sugere que a formação das estruturas domo e quilhas ocorreu antes do último evento magmático (ou
seja, antes de 2.70 Ga), o que está em desacordo com a idéia de que a região adquiriu sua arquitetura
atual durante o Paleoproterozóico.

xxii
 CHAPTER 1
INTRODUCTION
1

1.1 GENERAL INTRODUCTION

Fingerprinting the origin of the Earth‘s continental crust relies on the incomplete dataset given
by small remnants of Archean (and Hadean) crust, representing an estimated < 10 % of the present-
day exposed continental surface area (Goodwin, 1996). These vestiges, which withstood billions of
years of active tectonism, melting and surficial erosion, hold the key to understanding the processes of
terrestrial differentiation. Some of the underlying issues include the rates and mechanisms of growth
and recycling of the continental crust, and the geodynamic settings in which it formed.

Our understanding of the early Earth is largely obliterated by the scarcity of the ambiguous
Archean geological record. Rocks older than 2.5 Ga are relatively rare, but present on every continent
(Fig. 1.1). In the Nuvuaggittuq greenstone belt of the northeastern Superior Province (Canada), O‘Neil
et al. (2008) dated a cummingtonite-amphibolite using the short-lived 146Sm-142Nd radioisotope system
and obtained an isochron age of 4280 +53/-81 Ma, thus possibly representing the oldest known rocks on
Earth. The oldest felsic rocks, to date, consist of small domains within the migmatitic Acasta Gneisses
of the Slave craton (northwestern Canada). Two metatonalites and a metagranodiorite yielded
SHRIMP U-Pb-Th zircon ages of 4002 ± 4, 4012 ± 6 and 4031 ± 3 Ma respectively (Bowring &
Williams, 1999), testifying of the presence of terrestrial differentiation as early as 4.0 Ga. Jack Hills
metaquartzites in the Narryer Gneiss Terrane of the Yilgarn craton (western Australia) yielded detrital
zircon grains with abundant ages ranging from ~ 4.0 Ga to the world-record 4404 ± 8 Ma result (e.g.
Froude et al., 1983; Wilde et al., 2001). The presence of quartz and feldspar inclusions together with
characteristic trace element patterns may indicate that these zircons crystallized from magmas of
granitic composition (Maas et al., 1992). Dispute however remains as to whether models that invoke
an early (i.e. Hadean-Paleoarchean) continental growth can find support in the existence of a small
population of zircons from one area in western Australia.

Decades of geoscience research have uncovered important differences in the geological record
between modern continents and their Archean analogues. Changes in parameters such as metamorphic
gradients, sediment composition, structural aspects and nature and composition of plutonic and
volcanic rocks attest the presence of different processes operating in the Archean. A major point of
contention lies in identifying the geodynamic context in which these processes have taken place. There
is a consensus on the view that the Archean mantle was hotter than today, which resulted in a different
style of convection, contrasting with modern-day plate tectonics. The transition from an Archean
regime to modern geodynamic settings took place between 3.0 and 2.5 Ga, and is generally interpreted
to result from the progressive cooling of the Earth.

1
Figure 1.1 – Map showing the distribution of exposed Archean crust and craton boundaries. Modified after Bleeker (2003).

2
 This thesis is built within the framework of the secular evolution of the continental crust and
terrestrial geodynamic processes during the Archean. This work focuses on the southern portion of the
São Francisco craton in southeast Brazil, and aims to reconstruct the geological history of the
basement in order to understand the mechanisms that led to the formation and final stabilization of this
segment of Archean crust. The intentionally vague title to this thesis (―Archean evolution of the
southern São Francisco craton (SE Brazil)‖) allows for a certain flexibility in the geological aspects
investigated over the course of this work.

The southern São Francisco craton, which includes the Quadrilátero Ferrífero mining district
(―Iron Quadrangle‖), is one of the world‘s largest mineralized provinces. Its exceptional wealth of
economic resources, including world-class gold and iron-ore deposits, has attracted numerous studies
since the turn of the eighteenth century (e.g. Machado, 2009). As hosts to the vast majority of the ore
deposits in the region, the volcano-sedimentary sequences have stood under the spotlight of scientists
for most of the time, and comparatively little attention has been given to the (barren) crystalline
basement. Since the beginning of the 1990s, several Brazilian scientists have built up a general
geological framework for the Archean basement of the southern portion of the craton, based on an
integrated interpretation of zircon U-Pb ages and major and trace element geochemical data for
granites and gneisses (e.g. Machado et al., 1992; Carneiro et al., 1997; Noce et al., 1998). However, to
date, most studies have had only limited scope and lack regional-scale correlations, and many
questions surrounding the formation of this crustal nucleus remain unanswered. In particular, this
work has the following aims:

- The mineralogical, chemical and isotope compositions of granitoid rocks can place
important constraints on their petrogenesis. Such data can be used to define the conditions
of partial melting, the nature of the sources undergoing melting, and the geodynamic
settings in which these magmas were generated.
- Building a comprehensive and complete set of geochronological data for the granitoids
and gneisses of the basement of the craton allows establishing hypotheses regarding the
rate and timing of crust formation in this region.
- The presence of various signs of hydrothermal activity and alteration (i.e. numerous
pegmatitic/aplitic dikes and quartz veins) testify of the polymetamorphic history of the
region. Examining these rocks will place constraints on the timing of hydrothermal
activity and the origin of the fluids they crystallized from, in order to reconstruct part of
the complex tectono-metamorphic-hydrothermal evolution of the southern São Francisco
craton.

This thesis consists of a compilation of articles that aim to clarify the issues mentioned above,
each one using a different scientific approach. The manuscript is laid out as follows:

3
- Chapter 1 lays out the main goals and introduces the general geological framework of this
study. It presents a summary of the main geological aspects associated with the secular
evolution of the continental crust, and outlines a number of characteristic features specific
to the Archean eon.
- Chapter 2 synthesizes our current knowledge of the geology and evolution of the southern
part of the São Francisco craton, and more particularly of the region of the Quadrilátero
Ferrífero. It is presented as an article published in Journal of South American Earth
Sciences (2016, vol. 68, pp. 4-21), written within the framework of a pre-conference field
trip organised by F. Farina in the southern São Francisco craton in September 2015. All
co-authors contributed to the paper either by drafting individual chapters or as standard
reviewers. F. Farina assembled the article and wrote the discussion. I assisted with the
field trip, and drafted all the figures of the article. All authors reviewed the manuscript.
- Chapter 3 presents a comprehensive set of field, geochemical and U-Pb geochronological
data for the granitoids and gneisses of the basement of the southern São Francisco craton,
and discusses the results in terms of sources, mechanisms, and geodynamic sites of
magma generation. This work is published in Precambrian Research (2015, vol. 266, pp.
375-394). F. Farina led all aspects of the research and wrote the article. I took part in the
majority of the field work, performed half of the U-Pb geochronology and assisted in the
interpretation of the data. All authors reviewed the manuscript.
- Chapter 4 presents the results of a combined U-Pb-Hf-O isotope zircon study from
granitoids and gneisses of the basement, in order to place new petrogenetic constraints on
the production, maturation and stabilization of the crust. These results, coupled with
isotopic modelling are used to discuss the geodynamic context of formation of these rocks
and the rates of crustal growth. This study is published in Lithos (2016, vol. 266-267, pp.
64-86). I performed half of the isotope analyses, collected and interpreted all the data
under the supervision of F. Farina and C. Lana, and wrote the paper. All authors reviewed
the manuscript.
- Tourmaline is present in numerous felsic intrusions, pegmatitic dikes and quartz veins as
well as in hydrothermally altered rocks in the vicinity of the basement of the southern São
Francisco craton. Chapter 5 presents the chemical and boron isotope composition of these
tourmalines, used to discuss the timing and origin of these widespread hydrothermal
fluids. This work is currently under review in Geochimica et Cosmochimica Acta. I
performed the field and petrographic observations, mineral geochemical analyses, and
conducted the isotope analyses with the supervision of C. Lana. I interpreted the data and
wrote the manuscript. All authors reviewed the manuscript.
- Chapter 6 summarises the main results outlined in this work.

4
1.2 THE SECULAR EVOLUTION OF THE CONTINENTAL CRUST

The continental crust represents the archive of the Earth, effectively recording more than 4
Gyr of its evolution. What are the processes responsible for the generation of the continental crust, and
how have they evolved throught time? This section aims to lay out an up-to-date perspective of the
secular geodynamic evolution of the continental crust. A summary of the various points raised in this
section is illustrated in Fig. 1.2.

1.2.1 Crustal growth

As an ubiquitous, refractory and readily datable mineral, zircon is an ideal mineral tracer of
geological times and represents a unique window into the early Earth. It is commonly used as a proxy
for magmatic and metamorphic processes from individual units to entire orogens, as well as for the
evolution of continental crust production and composition through time. Large compilations of zircon
U-Pb ages show a non-uniform distribution over the course of Earth‘s history, with crystallization ages
scattered as a succession of peaks and troughs (Fig. 1.2a). Peaks are identified at approximately 2.7,
1.8, 1.1, 0.6 and 0.2 Ga by numerous studies using data from both detrital and igneous zircons,
obtained using multiple precise dating methods and from all continents (e.g. Rino et al., 2004; Wang
et al., 2009; Belousova et al., 2010; Condie & Aster, 2010; Voice et al., 2011). Two different
interpretations arise from this distribution of crystallization ages. The first one relates the peaks to
short pulses of accelerated crustal growth associated with episodic mantle convection (e.g. Reymer &
Schubert, 1986; Stein & Hofmann, 1994; Condie, 1998), whereas the other argues that they represent
periods of enhanced crustal preservation associated with the global plate-tectonic cycle (e.g.
Hawkesworth et al., 2009, 2010; Condie & Aster, 2010; Bradley, 2011). The implication of the latter
hypothesis is that the supercontinent cycle tends to bias the rock record. Advocates for the
preservation model explain that magmas generated in subduction environments or island arcs have far
better chances of being destroyed by erosion, subduction or delamination than those produced in the
late stages of continental collision, and that these peaks coincide with the ages of assembly of
supercontinents. Moreover, peaks and troughs in the δ18O zircon record appear to correlate well with
periods of crustal thickening and the development of supercontinents (Fig. 1.2b). Indeed, high δ18O
values are taken as indicator for sediment reworking, which is readily achieved during continental
collision.

The development of radiogenic isotope methods of analyses has brought up a range of models
describing the rates of crustal growth through time. Most models predict an increase in the volume of
continental crust from planetary accretion to present, but disagree on the timing and rates of crust
formation (Fig. 1.2a). Early models were largely based on the geographic distribution of K-Ar and Rb-
Sr ages (e.g. Hurley et al., 1962; Hurley & Rand, 1969; Veizer & Jansen, 1979; Goodwin, 1996).
However, these isotopic systems are very sensitive to more recent orogenic events, and the resulting

5
models are strongly biased towards younger ages (i.e. they predict exponentially increasing crustal
growth rates through time). More recently, Poupinet & Shapiro (2009) used an age distribution for the
lithosphere derived from its seismological properties to deduce the rates of crustal growth, and
obtained results similar to those of Hurley & Rand (1969). On the contrary, the models of Fyfe (1978),
Armstrong (1981) and Reymer & Schubert (1984) propose a very rapid and very early (i.e. Hadean)
burst of crustal growth, followed respectively by a regime of decreasing, stable or steadily increasing
crustal volume. However, a major issue associated with these models is the lack of geological
evidence supporting the presence of such large volumes of early continental crust.

Several models invoke a somewhat episodic growth of the crust with enhanced production at
specific periods, corresponding with phases of the supercontinent cycle or mantle geodynamics (e.g.
Condie, 1998; Campbell & Allen, 2008; Condie & Aster, 2010; Voice et al., 2011). This interpretation
is largely based on the episodic nature of the rock record, including ages of mineral deposits (Gastil,
1960), zircon U-Pb ages (e.g. Condie, 1998; Rino et al., 2004), He isotope ages (Parman, 2007) and
mantle depletion events deduced from Re-Os analyses (Pearson et al., 2007). Taylor & McLennan
(1985, 1995) used coupled geochemical and geochronological data integrated with tectonic models of
the continental crust to understand the interrelationship between crustal and mantle reservoirs, and
propose a model for the origin and rate of crustal growth. Their model notably predicts a big pulse of
crustal growth (up to ~ 60 % of the present-day volume) during the late Archean period (2.5-3.0 Ga)
followed by a series of more subordinate events in the Proterozoic.

More recent, and somewhat seemingly more accepted models of crustal growth feature
increasing volumes of continental crust, but overall decreasing growth rates through time (e.g.
Belousova et al., 2010; Dhuime et al., 2012; Guitreau, 2012; Roberts & Spencer, 2014). A common
approach is to use the distribution of Hf model ages of zircons to infer the timing of extraction of the
continental crust from the mantle. One concern associated with zircon isotope data is the likelihood
that it might represent metamorphic or crustal reworking events (i.e. model ages do not necessarily
reflect times of juvenile crust production). It is thus essential to correct the observed distribution of
model ages for crustal reworking. Belousova et al. (2010) calculated a degree of crustal reworking by
looking at the temporal distribution of zircons with Hf model ages greater than their crystallization
ages (zircons with similar crystallization and model ages are thought to represent juvenile crustal
material). Dhuime et al. (2012) used O isotope data to screen their detrital zircon compilation and
exclude the data representing reworked crust (i.e. zircon with non-mantle-like O isotope signatures).
They observe an inflection point at ~ 3.0 Ga, interpreted as a transition into a regime with enhanced
crustal reworking. Roberts & Spencer (2014) propose a minimum estimate for the growth of the
continental crust, based on a reflection on the balance between continental growth and destruction, the
geodynamic context of crust formation in the early Earth and the role of the preservation bias in the
zircon record. Hawkesworth et al. (2016) predict a stepwise increase of continental volume: one

6
period of enhanced crustal production during the Archean, similarly to the models discussed above,
although somewhat faster, and a second one between 2.0 and 1.0 Ga. The peculiarity of this model lies
in the fact that it suggests that the volume of continental crust was 20 % higher in the past than at
present. The authors make the assumption that crustal thickness may be a reliable proxy for crustal
volume – at least in recent times – and that a decrease in crustal thickness since ~ 1 Ga, as predicted
by Dhuime et al. (2015) (Fig. 1.2b) would indicate that during the past billion years, the rates of
crustal destruction exceeded those of crust formation.

Overall, the recent and generally more accepted models in the literature predict that two thirds
of the present-day volume of continental crust was extracted from the mantle prior to 2.5 Ga, and that
the end of the Archean eon featured a change from a regime dominated by crustal growth to one with
balanced growth and destruction (Fig. 1.2a). To understand this transition, we need to consider the
secular geodynamic evolution of the Earth.

7
8
Figure 1.2 – Variations of different geological and geochemical parameters through time. Also highlighted are
the timing of supercontinent assembly and breakup and transition periods at the Archean-Proterozoic and
Proterozoic-Phanerozoic boundaries. (a) Crustal growth models (plain bold lines, grey scale variations indicating
the year of publication (darker = more recent)) and present-day surface age distribution of Goodwin (1996)
(dashed line), plotted against a histogram of detrital zircon U-Pb ages (> 95 % concordant, n = 100 445; Voice et
al., 2011). Abbreviations are: HR69 – Hurley & Rand (1969); F78 – Fyfe (1978); A81 – Armstrong (1981);
RS84 – Reymer & Schubert (1984); TM95 – Taylor & McLennan (1995); R04 – Rino et al. (2004); PS09 –

9
Poupinet & Shapiro (2009); B10 – Belousova et al. (2010); CA10 – Condie & Aster (2010); D12 – Dhuime et al.
(2012); RS14 – Roberts & Spencer (2014); H16 – Hawkesworth et al. (2016); (b) Mean of initial εHf in ~ 7000
detrital zircons (Cawood et al., 2013) and mean of ~ 3300 δ18O zircon analyses (Spencer et al., 2014); (c)
Variation in the estimated Rb/Sr ratio, SiO2 content and thickness of juvenile continental crust (Dhuime et al.,
2015); (d) Average burial rates of various metasedimentary units and associated apparent metamorphic gradients
through time (Nicoli et al., 2016). Circles and diamonds are accretionary and collisional orogens respectively.
The Pilbara craton is indicated by an arrow. Boxes illustrate the range of burial rate values for each period; (e)
Hypothesized timing of global geodynamic regimes and metamorphic conditions (see text for discussion). UHP
– Ultra-high pressure; UHT – Ultra-high temperature; E-HP – Eclogite-high pressure; (f) Compositional
evolution of felsic and mafic lithologies through time (Keller & Schoene, 2012). A shift in igneous rocks
composition is inferred at 2.5 Ga; (g) Proportions of different sedimentary rock types through time (Spencer et
al., 2014).

1.2.2 Geodynamic evolution

The questions of why and when did plate tectonics begin on Earth? – and the complementary:
what was there before? – is one of the most important and unresolved issue in the history of the solid
Earth, resulting in an enormous quantity of dedicated articles, books and special issues.

Why?

The most commonly accepted explanation for the appearance of plate tectonics on Earth is the
progressive cooling of the Earth with decreasing radiogenic heat production. Sizova et al. (2010)
performed 2D petrological-thermomechanical numerical subduction models, and argued that the
degree of lithospheric weakening induced by percolating sub-lithospheric melts at high upper-mantle
temperature (> 250 °C above present) hampers the development of modern subduction style. With
cooling and lower melt fluxes, the results indicate a transition to a ―pre-subduction‖ tectonic regime (T
between 250 and 200 °C above present) characterized by shallow underthrusting, and to the modern
style of subduction (T between 175 and 160 °C above present). Similarly, the results of a handful of
other thermomechanical numerical experiments have suggested that a decrease of Moho temperature
would have resulted in rheological strengthening of the lithosphere and a gradual change towards
modern subduction-collision settings (e.g. van Thienen et al., 2004; Davies, 2006; O‘Neill et al., 2007;
van Hunen & van den Berg, 2008; Halla et al., 2009; Moyen & van Hunen, 2012; Gerya, 2014; Sizova
et al., 2014). The style of the tectonic regime in play during the Precambrian is still however a matter
of debate and will be discussed later.

When?

Estimates range from ~ 800 Ma (Hamilton, 2008, 2011; Stern, 2008) to > 4.0 Ga (Harrison et
al., 2008), with a clear predilection for the end of the Archean (3.0-2.5 Ga; e.g. Condie & Kröner,
2008; Gerya, 2014). Identifying the timing of the onset of plate tectonics implies tracking the
appearance of specific petrotectonic tracers (i.e. rock ―packages‖ originating in, and characteristic of,
specific tectonic settings) in the geological record. The question remains: what imprint does one use to
undoubtedly identify the action of plate tectonics? The subjectivity of the answer to this problem is

10
responsible for the lack of consensus on the topic. Petrotectonic assemblages characteristic of plate
tectonics include ophiolites, arc and back-arc assemblages, accretionary prisms, ocean plate
stratigraphy, foreland basins, blueschists and Ultra-High Pressure (UHP) rocks, passive margins,
continental rifts and specific mineral deposits. Other indicators comprise UHP metamorphic terranes,
paired metamorphic belts, large transcurrent faults and suture zones, collisional and accretionary
orogens, paleomagnetic data, geochemical and isotopic data, and the existence of continental crust.

The main reasoning used by supporters of a Neoproterozoic plate tectonics appearance is that
three indicators are essentially missing in the Archean (and most of the Proterozoic) rock record:
ophiolites (Stern, 2005), UHP metamorphism (Fig. 1.2d; Brown, 2006) and blueschists (Maruyama et
al., 1996) (Fig. 1.3), leading them to the conclusion that their appearance at < 1 Ga is important
evidence for when modern subduction began (e.g. Stern, 2005, 2008).

Figure 1.3 – Histograms showing ages of preserved plate tectonic indicators for the last 3 Ga. A: Oceanic
lithosphere (ophiolites); B: Subduction zone metamorphic products (jadeitites, blueschists and lawsonite
eclogites); C: Continental margins and collision zones (gem corundum, UHP metamorphic rocks and passive
continental margins). Gray bins represent modern passive margins. Stern et al. (2013).

On the other hand, several researchers argue that settings analogous to modern convergent
margins existed at the surface of the Earth at ~ 4 Ga. Harrison et al. (2008) and Hopkins et al. (2010)
obtained isotope and trace element compositions and mineral inclusion data from the Hadean Jack

11
Hills zircons, and suggested the existence of portions of continental crust possibly generated by
mechanisms similar to plate tectonics as early as ~ 4.3 Ga. Arndt & Chauvel (1991) commented on the
remarkable survival of the Jack Hills zircons: ―The fact that such old zircons constitute 2 % of the
zircon population in a sediment deposited some 1 Ga after they formed is astonishing: either there was
a remarkable preservation mechanism that enabled these zircons to survive for a billion years at the
surface of an unstable early Earth, or the source of the old zircons was abundant and voluminous.
Since the most likely platform that could have preserved the zircons is buoyant, low density, felsic
crust, the very existence of old zircons is strong evidence that voluminous felsic material in one form
or another existed on the Hadean Earth.‖ Other authors for example plead for a change in geodynamic
style at ~ 3.8 Ga, based for example on the occurrence of sheeted dikes and pillows basalts resembling
ophiolite sequences in the Isua greenstone belt (SW Greenland) (Dilek & Polat, 2008), or a shift
towards more radiogenic Hf signatures in the detrital zircon record of Jack Hills, signalling renewed
juvenile material input at the time (Nebel-Jacobsen et al., 2010; Arndt, 2013).

However, as stressed by Condie & Kröner (2008): ―It is important to point out that “single
lines of evidence” (including single petrotectonic assemblages) may not be definitive of modern-style
plate tectonics, but it is the convergence of evidence at any period of time that is most useful in
tracking plate tectonics into the past.‖ Thus, evidence seems to converge towards one specific period:
the Meso- and Neoarchean. Indeed, most plate tectonic tracers mentioned above suggest that
significant gradual changes occurred between 3.0 and 2.5 (and 2.0) Ga. Important modifications
notably include: (1) increasing proportions of basalts derived from enriched, depleted or hydrated
mantle sources, consistent with enhanced crustal recycling into the mantle (Smithies et al., 2004;
Condie, 2015), (2) a change in the source of crustal granitoids evidenced by major and trace element
geochemistry and O isotopes (Fig. 1.2b and f; Valley et al., 2005; Keller & Schoene, 2012), (3) the
rapid thickening of juvenile crust, indicated by a dramatic increase in Rb/Sr ratio in igneous rocks
(Fig. 1.2c; Dhuime et al., 2015), (4) a decrease in the diversity of tectonic processes, indicated by a
decrease in the range of burial rates (Fig. 1.2d; Nicoli et al., 2016), (5) the appearance of Ultra-High
Temperature (UHT) granulite metamorphism (Fig. 1.2e; Brown, 2006), (6) the decrease in mantle
temperature and degree of melting (Fig. 1.2f; Keller & Schoene, 2012; Condie et al., 2016; Ganne &
Feng, 2017), (7) lateral plate motions evidenced by paleomagnetic data (Mitchell et al., 2014), (8) the
appearance of eclogite inclusions in diamonds (Shirey & Richardson, 2011), (9) the appearance of
collisional orogens (Cawood et al., 2013), (10) a change in the style of crustal deformation from hot
accretionary orogens involving weak lithosphere and diffuse deformation to modern orogens
evidencing strongly localised strain (Fig. 1.4; Chardon et al., 2009; Cagnard et al., 2011), (11) the
appearance of mature sediments and large shelf sequences, reflecting the initiation of Wilson cycles
(Fig. 1.2g; Veizer & McKenzie, 2003), and (12) the rapid increase in volume of continental crust (Fig.
1.2a; see previous section).

12
Figure 1.4 – Classification of orogens. (a) Schematic orogen scenarios. C = Crust; LM – Lithospheric mantle;
LM1 – Stiff upper mantle lithosphere; LM2 – Ductile, lower viscosity, lower lithospheric mantle (Chardon et al.,
2009); (b) Schematic cross sections showing different orogenic styles through time (Cagnard et al., 2011).

The geological record appears to indicate a second important change at the Proterozoic-
Phanerozoic transition. In particular, the appearance of blueschists and ophiolite sequences (Fig. 1.3),
changes in the metamorphic record exemplified by the appearance of UHP rocks (Fig. 1.2e; Brown,
2006) and numerical experiments (Sizova et al., 2010, 2014) imply yet a different style of tectonics in
the Phanerozoic eon compared to the Proterozoic. Evidence for modern continental collision is also

13
more pronounced in the Phanerozoic. This transition is attributed to the onset of cold, deep and steep
subduction typical of the present-day.

What was there before plate tectonics began?

The geological record becomes exponentially scarcer as we go back in time (Fig. 1.2a;
Goodwin, 1996). Therefore, one of the most valuable windows into uncovering the geodynamic
conditions operating in the past is thermomechanical numerical modelling (van Thienen et al., 2004;
Davies, 2006, O‘Neill et al., 2007; Sleep, 2007; van Hunen & van der Berg, 2008; Halla et al., 2009;
Moyen & van Hunen, 2012; Sizova et al., 2010, 2014; Gerya, 2014; Fischer & Gerya, 2016). A
plume-induced tectonics hypothesis is often involved for the early Earth, in which lithosphere
deformation is produced by mantle upwellings and downwellings rather than plate tectonics (Fig. 1.5).
Geochemical and geological evidence point towards higher mantle temperatures during the Archean (~
200-300 °C; Korenaga, 2013). At increased temperature, mantle-derived magmas convey large heat
and mass advection, resulting in a number of characteristics predicted by numerical 2D and 3D
models: (1) a weak, internally deformable, highly heterogenous lithosphere, (2) huge juvenile crust
production from those mantle-derived melts, (3) mantle flow-driven crustal deformation, (4) episodic
magma-assisted crustal overturns, (5) the widespread development of lithospheric delamination and
eclogitic drips, removing portions of overthickened unstable crust, and (6) continental crust production
and segregation.

Figure 1.5 – Schematic cross sections of early Earth (> 3 Ga; A) and young Earth (< 3 Ga; B) showing different
modes of crust formation. In the early Earth, crust forms in two settings: (i) above upwelling hot mantle, where it
forms thick crust with a depleted keel of peridotitic subcontinental lithospheric mantle (SCLM), and (ii) above
areas of down going mantle, where oceanic lithosphere is imbricated and extensively melted to form high-grade
gneiss terrains. In the young Earth with larger plates, mantle convection generates steep subduction zones and
the return of oceanic lithosphere to the mantle. Crust grows via subduction-generated arc magmatism and at hot
spots. Subcretion of oceanic lithosphere to the base of continents results in SCLM with mixed peridotitic and
eclogitic compositions. (Van Kranendonk, 2011).

14
 Overall, geological and geochemical observations combined with numerical experimental data
allow us to identify some key features for the geodynamic evolution of the continental crust in a
cooling Earth, as summarised below:

- Prior to ~ 3.0 Ga, Archean geodynamics was characterized by plume tectonics and
lithospheric dripping (delamination). Melt percolation from a hotter sublithospheric
mantle rheologically weakened the overlying lithosphere, which hampered the
sustainment of high topography in accretionary orogens (Rey & Coltice, 2008).
Deformation patterns in the continental crust archive attest the dominant role of
gravitational tectonics.
- A marked change appears to have occurred between 3.0 and 2.5 (and 2.0) Ga, commonly
attributed to the widespread development of subduction on Earth as a response to the
rheological strengthening of the cooling lithosphere. Hot mantle temperatures (175-250 °C
higher than present) led to weaker slabs subjected to frequent break-off, which resulted in
a more episodic style of plate tectonics (van Hunen & van der Berg, 2008). Consequently,
continental crust in the Precambrian is thought to have formed in short, repeated events
rather than in stable and laterally extensive magmatic arc environments (Moyen & van
Hunen, 2012).
- Development of modern-style, cold and deep subduction started in the Neoproterozoic, as
evidenced by the preservation of LT-HP metamorphic assemblages in Phanerozoic
orogens.

1.3 ARCHEAN CRATONS

In the previous section, I discussed the secular evolution of the Earth in response to
progressive cooling. Evidently, the peculiar conditions operating in the Archean resulted in very
specific lithological associations and structures. Schematically, Archean terranes consist of three
contrasting lithologies: (1) deformed and often migmatized, sodic metaigneous rocks belonging to the
tonalite-trondhjemite-granodiorite series (the so-called ―grey gneisses‖), (2) sequences of
metavolcanic and metasedimentary units, generally metamorphosed to greenschist and amphibolite
facies conditions (greenstone belts), and (3) late and generally more K-rich granitoids, intruding both
previous lithologies. In addition to specific lithologies, sections of Archean crust display very
distinctive ―dome-and-basin‖ structures, characterized by multi-kilometre scale basement domes
separated by belts containing the supracrustal rocks. This section aims to introduce their main
characteristics and review their models of formation.

1.3.1 Archean granitoids

1.3.1.1 Tonalites, trondhjemites, granodiorites

15
 Jahn et al. (1981) first coined the acronym TTG in reference to closely associated plutonic
rocks with sodic compositions (tonalites, trondhjemites and granodiorites) occurring as grey gneisses.
TTGs represent the main component of Archean cratons (for reviews, see for example Condie, 2005
and Moyen & Martin, 2012 and references therein). They often crop out as banded or migmatized
orthogneisses (Fig. 1.6a) or, more rarely, as nearly pristine batholiths (Fig. 1.6b). Typical
mineralogical assemblages consist of quartz + oligoclase + biotite ± K-feldspar ± hornblende, with
accessory allanite, pistacite, apatite, zircon, titanite and titanomagnetite.

Figure 1.6 – Field photographs of TTGs. (a) Ca. 3.2 Ga migmatitic orthogneiss (Sand River Gneiss, Limpopo
Belt, South Africa) (picture from J.-F. Moyen; http://jfmoyen.free.fr/); (b) 3.45 Ga homogeneous Stolzburg
pluton (Barberton, South Africa) (Moyen, 2011).

16
 The geochemical definition of TTGs is precise: they are sílica-rich (SiO2 > 64 wt.%,
commonly > 70 wt.%), sodic (Na2O > 3.0 wt.% and associated K2O/Na2O < 0.5) plutonic rocks,
generally metaluminous to slightly peraluminous (1.0 < A/CNK < 1.1), with low ferromagnesian
contents (FeOt + MgO + MnO + TiO2 ≤ 5 wt.%). Archean TTGs do not show clear trends of
differentiation, contrary to the modern calc-alkaline series which evolve through K-enrichment during
differentiation (Fig. 1.7a and b). TTGs generally exhibit very distinctive trace element patterns, with
high Light Rare Earth Element (LREE) and low Heavy Rare Earth Element (HREE) contents and
associated high La/Yb ratios, negative Nb-Ta and Ti anomalies, but no significant Sr and Eu
anomalies (Fig. 1.7c and d). These characteristics differ strongly to those of modern granitoids (Fig.
1.2f).

Figure 1.7 – Major and trace element characteristics of TTGs. (a) Normative Na-Ab-Or triangle (O‘Connor,
1965) showing the composition of TTGs (blue dots) against that of modern continental crust (yellow field); (b)
K-Na-Ca triangle comparing the compositional evolution trend of modern calc-alkaline magmas during
differentiation (brown line), and TTG magma compositions, lacking a differentiation trend; (c) (La/Yb) N vs. YbN
diagram, illustrating the compositional contrast between TTGs (low-HREE, strongly fractionated patterns) and
post-2.5 Ga granitoids (high-HREE, moderately fractionated); (d) Trace element patterns for TTGs and modern
continental crust normalized to the primitive mantle. (Moyen & Martin, 2012).

17
 Importantly, significant trace element variations between TTGs indicate a strong pressure
dependance. In particular, ratios such as Sr/Y and La/Yb show systematic variations, and are used a
proxies for depth of melting (Halla et al., 2009; Almeida et al., 2011; Moyen, 2011; Moyen & Martin,
2012). TTGs largely range between ―high pressure‖ (with low HREE, Y, Nb, Ta and high Sr contents)
and ―low pressure‖ (high HREE, Y, Nb, Ta and low Sr contents) end-members (Moyen, 2011). These
lines of evidence have been taken to reflect the mineralogical composition of the residue, with the
―high pressure‖ group bearing characteristics of melts in equilibrium with garnet (rich in HREE) and
rutile, but without plagioclase, and the ―low pressure‖ group having compositions implying the
presence of plagioclase (in which Sr is compatible) in the residue of melting but little garnet and no
rutile. TTGs belonging to the ―medium pressure‖ have characteristics intermediate between the
previous two.

TTGs are generally accepted to be the differentiation products of a basaltic parent. Their
generation follows a general three-step model: (1) partial melting of the mantle to generate a basaltic
magma, (2) partial melting of this hydrated basalt, metamorphosed into eclogite or garnet-bearing
amphibolite, giving rise to a TTG parental magma, and (3) fractional crystallization (optional) to
generate the more differentiated TTG products (fractionation probably limited to < 25 % where
present; Moyen et al., 2007) (Fig. 1.8; Atherton & Petford, 1993; Barker & Arth, 1976; Ellam &
Hawkesworth, 1988; Martin, 1986, 1987, 1994; Rapp et al., 1991, 2003; Foley et al., 2002; Kamber et
al., 2002; Martin & Moyen, 2002; Kleinhanns et al., 2003; Moyen & Stevens, 2006; Moyen & Martin,
2012; Martin et al., 2014).

Figure 1.8 – Schematic summary of the most accepted mechanism of TTG production. Hbl – hornblende; Grt –
garnet; Pl – plagioclase; Cpx – clinopyroxene; Ilm – ilmenite; PM – partial melting; FC – fractional
crystallization. (modified after Martin, 1993).

18
 Several geodynamic contexts have been suggested to explain the partial melting of a hydrated
metabasalt at the conditions mentioned above, and the issue is still a major point of contention. These
models are developed around two predominant ideas (Fig. 1.9): (1) the ―hot subduction‖ model, where
a basaltic oceanic subducting slab reaches partial melting conditions before dehydration (Fig. 1.9a and
b; e.g. Martin, 1986, 1987, 1999; Defant & Drummond, 1990; Martin & Moyen, 2002; Rapp et al.,
2003; Martin et al., 2014), or (2) the ―intraplate‖ hypothesis, where the base of an overthickened crust
undergoes partial melting, in response to mantle plume activity (Fig. 1.9d; e.g. Smithies, 2000; Zegers
& van Keken, 2001; Bédard et al., 2003; van Thienen et al., 2004; Bédard, 2006, 2013; Willbold et
al., 2009). Predominant in the literature, the ―hot subduction‖ model essentially hinges on: (1) the
necessity to attain pressures up to ~ 20 kbar (at least to form the high pressure TTGs), which is not
easily reachable within a hotter Archean crust unable to withstand great thicknesses (Rey &
Houseman, 2006), (2) the need to bury hydrous basalts to such depths, (3) the existence of interactions
between TTG magmas (i.e. slab melts) and peridotites (i.e. mantle wedge) to produce sanukitoid
magmas (see next section) (e.g. Shirey & Hanson, 1984; Martin et al., 2009), and (4) the strong
resemblance between TTGs and high-SiO2 adakites, modern arc lavas generated in subduction
environments by partial melting of hydrous basalts (Martin et al., 2005). The question that remains is
whether or not Archean oceanic crust was indeed subductable, which is a matter of balance between
driving (essentially slab pull) and resisting (i.e. resistance to bending, friction and buoyancy) tectonic
forces (e.g. van Hunen & van der Berg, 2008; van Hunen & Moyen, 2012). Geodynamic modelling
does not provide an unequivocal answer (e.g. de Wit & Hart, 1993; Korenaga, 2006; Davies, 2007;
Sizova et al., 2010, 2014; Johnson et al., 2014), nor does the search for petrotectonic indicators of
plate tectonics (see previous section), and many authors continue to plead for alternative models. In
the ―intraplate‖ model, TTGs are formed by partial melting of the lower portion of a thick pile of
basalts, either at the base of an oceanic plateau (i.e. above a mantle plume) or as delamination occurs
at the base of a thickened crust. A major inconsistency with this hypothesis lies in that it fails to
provide a viable mechanism to bring water down to the source of melting (see discussion in Arndt,
2013). As discussed by Moyen & Stevens (2006) in a summary of experiments investigating the
generation of TTG melts, the position of the solidus and the melt fractions are both controlled by water
availability, and anhydrous sources are unsuccessful in generating melts with TTG compositions.

Finally, it is perhaps appropriate to quote Moyen (2011): ―There is not one unique
environment for the growth of Archaean continental crust; rather, it formed in a range of different
settings throughout the Archaean Earth. Consequently, the challenge for Archaean geologists is not so
much to describe “the” site of Archaean crustal growth, but rather to propose a planet-wide model
that can accommodate the contrasting tectonic environments suggested by the diverse geochemistry of
Archaean TTGs.‖

19
Figure 1.9 – Diagrams illustrating the two main ideas for the geodynamic sites of TTG generation. a-c: proposed
subduction models in the (a) early Archean, (b) late Archean, and (c) present-day (Martin, 1986); (d) Intraplate
models: over a downwelling zone (left), rocks from the lower portion of overthickened crust delaminate and sink
into the mantle where they melt; over an upwelling mantle plume, the base of a thick oceanic plateau undergoes
partial melting (Condie & Abbott, 1999). CC – continental crust; OC – oceanic crust; FMP – fluid
metasomatised peridotite.

1.3.1.2 Late Archean granitoids

The Meso- and Neoarchean periods saw a diversification in the nature of crustal granitoids in
all cratons worldwide. This transition is globally observed in the geochemistry of felsic igneous rocks,
changing from sodic (Na2O/K2O ~ 2) to potassic (Na2O/K2O ~ 1) compositions across the Archean-
Paleoproterozoic boundary (Fig. 1.2f; Keller & Schoene, 2012). The wide range of mineralogical and
chemical characteristics of these late Archean, K-rich granitoids was reviewed recently by Laurent et
al. (2014), who attempted to build a petrogenetic classification for these rocks and discuss the
implications in terms of crustal evolution. These K-rich granitoids represent ~ 20 % of the volume of
preserved Archean crust (Condie, 1993), and were generally emplaced during the last magmatic stages
of the Archean history of the cratons. Based on an extensive review, Laurent et al. (2014) recognized
three groups of granitoids: (1) biotite- and two-mica granites, (2) Mg-rich monzodiorites,
monzogabbros and their differentiated products (sanukitoids thereafter), and (3) granodiorites and
granites with hybrid characteristics.

Biotite- and two-mica granites are found in every Archean terrane around the world (e.g.
Sylvester, 1994), and consist of homogeneous, medium-grained granites and leucogranites (Fig.

20
1.10a). They contain K-feldspar, sodic plagioclase and quartz roughly in equal modal proportions (~
30 %), biotite, and sometimes primary muscovite. Accessory mineral phases are ilmenite, magnetite,
apatite, zircon, allanite, and in places pistacite. They are slightly peraluminous (biotite granites) to
peraluminous (two-mica granites), silica-rich (SiO2 ≥ 68 wt.%) materials, with low ferromagnesian
contents (FeOt + MgO + MnO + TiO2 ≤ 4 wt.%), high K2O and low Na2O (resulting in K2O/Na2O
ratios > 0.5), low CaO contents (CaO < 2 wt.%) and moderate Al 2O3. They are typically richer in
incompatible elements (i.e. Rb, Th) and Y-HREE than TTGs, with low High Field Strength Element
(HFSE) and transition element contents (e.g. Zr < 265 ppm; V < 20 ppm). They are characterized by
variously fractionated REE patterns (15 ≤ (La/Yb)N ≤ 65) and significant Eu anomalies (~ 0.5) (Fig.
1.11). These compositional features are attributed mostly to a derivation from partial melting of an
older TTG crust (e.g. Feng & Kerrich, 1992; Sylvester, 1994; Moyen et al., 2001, 2003; Whalen et al.,
2004; Frost et al., 2006; Dey et al., 2012; Almeida et al., 2013). This interpretation partly hinges on
the compositional similarities between biotite granites and experimental melts derived from broadly
tonalitic and granodioritic materials at crustal pressure under fluid-absent conditions (Patiño Douce &
Beard, 1995; Singh & Johannes, 1996; Skjerlie & Johnston, 1993; Watkins et al., 2007). A few
studies, however, point out the highly sodic nature and overall infertility of TTGs (i.e. containing ≤ 10
% modal proportion of hydrous mafic minerals) to generate large volumes of K-rich granitic melts
(e.g. Castro, 2004; Watkins et al., 2007; Farina et al., 2015a). They propose instead the involvement of
more evolved crustal rocks such as immature metasediments (i.e. greywackes) in the petrogenesis of
these granitoids. A metasedimentary contribution is required to explain the more peraluminous two-
mica late Archean granitoids, similar to modern S-type granites (e.g. Breaks & Moore, 1992; Jaguin et
al., 2012).

21
Figure 1.10 – Field photographs of late Archean granitoids. (a) Coarse-grained granodiorite, Mamona batholith,
Quadrilátero Ferrífero, Southeast Brazil; (b) Porphyritic monzogranite, Closepet batholith, South India (Martin et
al., 2009); (c) Mingling between a porphyritic monzogranite and monzodiorites, Matok pluton, Limpopo Belt,
South Africa (Martin et al., 2009); (d) Heterogeneous, coarse-grained and porphyritic monzogranite (hybrid
granite), Mashashane pluton, Limpopo Belt, South Africa (Laurent et al., 2014).

The sanukitoid group includes a wide diversity of rocks, ranging from (monzo)diorites to
(monzo)granites (Fig. 1.10b and c), often defining complete magmatic series (Martin et al., 2005,
2009; Heilimo et al., 2010; Laurent et al., 2014). Microgranular mafic enclaves are common, as are
mafic aggregates of biotite + amphibole ± clino- and orthopyroxene, and large K-feldspar phenocrysts
(Fig. 1.10b). Despite a broad range of field occurrences and compositions, sanukitoids s.l. share
several key geochemical features. They are metaluminous (0.7 ≤ A/CNK ≤ 1.0), calc-alkaline to
alkali-calcic rocks, ranging from mafic (SiO2 ~ 50 wt.% and MgO ~ 8 wt.%) to felsic (SiO2 ~ 75 wt.%
and MgO ~ 0.1 wt.%) end-members. They are variably potassic (1.5 ≤ K2O ≤ 5.0 wt.%), and generally
contain high contents of calcium and ferromagnesian oxides (5 ≤ FeOt + MgO + MnO + TiO2 ≤ 25
wt.%). Trace element patterns of sanukitoids s.l. are generally similar to those of TTGs, but with a
notable enrichment in all incompatible elements (Ba > 1000 ppm; Sr > 400 ppm) and transition
elements (V > 50 ppm; 15 < Ni < 200 ppm; 20 < Cr < 500 ppm) (Fig. 1.11). These rocks show a wide
range of REE patterns (10 < (La/Yb)N < 75) with variable Eu anomalies (0.5 ≤ EuN/Eu* ≤ 1.0). This
enrichment in both compatible and incompatible elements is a characteristic feature of all sanukitoids.

22
These geochemical particularities together with experimental results suggest that sanukitoids are the
products of the interaction between mantle peridotites and a component rich in incompatible elements
at mantle levels (e.g. Stern & Hanson, 1991; Rapp et al., 1999, 2010; Smithies & Champion, 2000;
Moyen et al., 2001, 2003; Martin et al., 2009; Halla et al., 2009; Heilimo et al., 2010; Almeida et al.,
2013; Laurent et al., 2014). Sanukitoid magmas owe their diverse compositions to: (1) different
conditions occurring at the source (variations in P, T, and relative volume of peridotite and
metasomatic agent), (2) different contaminants, being either TTG melts (e.g. Rapp et al., 1999), H2O-
or CO2-rich brines (Stern et al., 1989; Lobach-Zhuchenko et al., 2008), sediment-derived fluids or
melts (King et al., 1998; Halla, 2005; Laurent et al., 2011; Mikkola et al., 2011), alkaline melts
(Heilimo et al., 2010) or carbonatites (Steenfelt et al., 2005), or (3) fractionation of the more mafic
magmas to produce the felsic sanukitoid end-members (Lobach-Zhuchenko et al., 2005, 2008;
Oliveira et al., 2010a; Laurent et al., 2013).

Figure 1.11 – Trace-element geochemistry of late Archean granitoids and TTGs. Symbols are examples from the
Pietersburg block and the Limpopo Belt and grey fields are literature data (Laurent et al., 2014). (a) Multi-
element patterns of average composition for each group, normalized to primitive mantle concentrations
(McDonough & Sun, 1995); (b) Sr/Y vs. Ba/Rb ratios; (c) HFSE (represented by Zr) vs. transition (represented
by V) elements; (d) EuN/Eu* vs. ΣLREE.
23
 Finally, late Archean granitoids with overall intermediate characteristics between TTGs,
biotite- and two-mica granites and sanukitoids were labelled ―hybrid granite‖ by Laurent et al. (2014).
This group encompasses rocks that formed as a result of metasomatism, mixing or mingling between
magmas or sources of all three previous types of rocks. Where they occur, they generally form
composite magmatic complexes, with monzogranites being dominant over the more mafic
granodiorites and diorites, in contrast to the sanukitoids (Fig. 1.10d). They consist mostly of
plagioclase (~ 30 %), K-feldspar (~ 30 %), quartz (~ 25 %) and biotite (< 15 %), with common
magmatic epidote and accessory magnetite, ilmenite, apatite, titanite and allanite. The wide range of
sources and petrogenetic processes involved in the formation of these magmas from one craton to
another preclude the establishment of discriminative geochemical criteria, and their compositions
generally overlap the compositional fields of the three other types of Archean granitoids (Fig. 1.11).
Where such rocks have been described, scientists invoked a mixed origin, which could either be
mixing between a TTG melt and a differentiated sanukitoid magma (Almeida et al., 2010),
contamination by an old crustal component of lower crustal mafic melts (Smithies & Champion, 2000;
Jayananda et al., 2006), hybridization of TTG magmas by enriched mantle (Feng & Kerrich, 1992;
Prabhakar et al., 2009), or an interaction between various processes (Champion & Sheraton, 1997).

All late Archean granitoid types are syn- to post-tectonic, and were emplaced during a
relatively short time-span (0.02-0.15 Ga) following a longer period of TTG magmatism (0.2-0.5 Ga).
Both the nature of the sources (metasomatized mantle, TTG crust or metasediments) and the processes
(burial of crustal components into the mantle, intracrustal differentiation) involved in the production of
the late Archean K-rich granitoids suggest that this period marked the onset of subduction-collision
cycles on a planetary scale (e.g. Laurent et al., 2014; Halla et al., 2017). Such changes however
occurred diachronously, from ~ 3.1 Ga (in the Barberton area of the Kaavaal craton; Kamo & Davis,
1994; Yearron, 2003; Clemens et al., 2010) and ~ 2.95 Ga (in the Pilbara craton; Champion &
Sheraton, 1997; Champion & Smithies, 1999, 2007) to ~ 2.5 Ga (e.g. in the North China and Dharwar
cratons; Jahn et al., 1988; Moyen et al., 2001, 2003; Liu et al., 2004; Jayananda et al., 2006; Yang et
al., 2008; Dey et al., 2012, 2014).

1.3.2 Greenstone belts

Greenstone belts are metamorphosed volcano-sedimentary sequences, representing only ~ 30

% of the volume of preserved Archean crust (Condie, 1993), yet they have gathered a significant
amount of scientific interest for various reasons: (1) they hold the potential to uncover information
about the suitability of Earth‘s surface to host life, or to reveal signs of life itself (e.g. Whitehouse &
Fedo, 2007; Van Kranendonk, 2007), (2) they provide valuable information on surface (i.e. continent
preservation, relief, erosion), igneous and metamorphic processes which can be used in geodynamic
reconstructions (e.g. Smithies et al., 2007; Stevens & Moyen, 2007), and (3) they are hosts to many
ore-bearing lithologies (e.g. Cu-Zn Volcanogenic Massive Sulfide (VMS) deposits, komatiite-
24
associated Ni deposits, syn- to post-orogenic Au deposits, rare-metal pegmatites) (e.g. Groves &
Barley, 1994).

One of the best preserved, and hence most studied greenstone sequence is undoubtedly that
cropping out in the Barberton region of South Africa (Fig. 1.12; e.g. Viljoen & Viljoen, 1969;
Eriksson, 1980; de Wit, 1982; Anhaeusser, 1983; de Wit et al., 1987; Heubeck & Lowe, 1994a, b;
Lowe & Byerly, 1999). The layout and preserved thicknesses of Archean greenstone belt sequences
vary from one craton to another, but that of Barberton provides a good illustration for the general
greenstone belt stratigraphic variations. It is broadly subdivided into three main units, from the base to
the top: the Onverwacht (~ 3.55-3.25 Ga), the Fig Tree (~ 3.25-3.22 Ga) and the Moodies Group
(3.23-? Ga) (Lowe & Byerly, 1999; Brandl et al., 2006). The basal Onverwacht Group (the Sandspruit,
Theespruit and Komati Formations) consists mostly of komatiites, overlain by pillowed and massive
tholeiitic basalts interlayered with felsic lavas and tuffs, siliciclastic units, cherts and banded iron
formations (the Hooggenoeg, Kromberg and Mendon Formations). The Onverwacht Group is
interpreted as a portion of oceanic crust, locally hydrothermally altered. The overlying Fig Tree Group
consists of siliciclastic units (shales, siltstones) associated with calc-alkaline felsic lavas and cherts,
coarsening upwards into chert conglomerates and coarse-grained sandstones and conglomerates of the
Moodies Group. These units indicate that deposition took place in progressively shallower
environments (i.e. shallow to moderately deep subaqueous, alluvial, fan-delta), with mature sediment
input.

25
Figure 1.12 – General stratigraphic column of the Swaziland Supergroup, Barberton greenstone belt, South
Africa. (Van Kranendonk et al., 2009).

1.3.3 Architecture of Archean granite-greenstone terranes

One of the most distinctive features of Archean crust which does not have a modern
equivalent is the dome-and-basin structures. Especially remarkable examples can be found in the
Pilbara (Fig. 1.13; Hickman, 2004; Van Kranendonk et al., 2007) and Dharwar (Bouhallier et al.,
1995) cratons. These structures are defined by large (up to 100 km in diametre) ellipsoidal granitoid-
dominated antiforms, separated by greenstone belts occurring as deep and elongated synclinal
structures (e.g. MacGregor, 1951; Anhaeusser et al., 1969; Allen & Chamberlain, 1989; Marshak et
al., 1992, 1997; Choukroune et al., 1995; Chardon et al., 1996, 2002, 2009; Van Kranendonk et al.,
2002, 2004). These features are generally interpreted as gravitational phenomena (also referred to as

26
―vertical tectonics‖, ―sagduction‖, or ―partial convective overturn‖), whereby the overlying dense
(ultra)mafic supracrustals founder into the more buoyant granitic-gneissic basement due to inverse
density gradients, in a manner similar to (inverse) diapirism (Fig. 1.14; e.g. Bouhallier et al., 1995;
Chardon et al., 1996, 1998; de Bremond d‘Ars et al., 1999; Bédard et al., 2003; Van Kranendonk et
al., 2004; Bédard, 2006; Bodorkos & Sandiford, 2006). The interplay between diapirism and regional
shortening and folding has also been invoked by several authors (e.g. Dalstra et al., 1998; Lin, 2005;
Parmenter et al., 2006; Erickson, 2010). Other proposed mechanisms include: (1) folding during
regional shortening (Snowden & Bickle, 1976; Blewett et al., 2004, 2010), (2) core complex formation
during extension (Williams & Whitaker, 1993; Kloppenburg et al., 2001; Zegers et al., 2001, Kisters
et al., 2003), (3) channel flow, in which a weak viscous crustal layer flows laterally between rigid
crustal slabs as a result of horizontal pressure gradients (Cagnard et al., 2006; Parmenter et al., 2006;
Harris et al., 2012), (4) simple intrusion (Hofmann et al., 2003), and (5) a combination of diapirism
and orogenic collapse/core complex (Marshak et al., 1997; Tinkham & Marshak, 2004). These
architectural features are rare in post-Archean terranes, and the processes that shaped them are hard to
reconcile within the framework of modern-day plate tectonics.

Figure 1.13 – Schematic structural map of the East Pilbara craton. (Van Kranendonk et al., 2002).

27
Figure 1.14 – Schematic model of partial convective overturn for the east Pilbara craton. (a) generation of a
gravitationally stable, layered crust through emplacement of a sheeted TTG sill complex into greenstones.
Domes are initiated at the eruptive centres; (b) eruption of a 5–8 km thick basaltic pile instigates an inverted
crustal density profile and thermally incubates the TTG sill complex, initiating overturn; (c) extensive partial
melting of the crust permits partial convective overturn of the crust, driven by sinking greenstones. (Van
Kranendonk et al., 2004).

1.4 CONCLUSIONS

The continental crust records more than 4 Ga of the evolution of the Earth and as such, it is a
unique archive to study the processes that shaped it. The volume of preserved continental crust,
however, exponentially decreases back in time, and the nature of crustal lithologies themselves has
changed from the early Earth to the modern era. Most variables indicate that these changes likely took
place gradually during a period centred around the Neoarchean. In particular, the mineralogy and
chemical composition of igneous rocks have shifted from early magmas of tonalite, trondhjemite and
granodiorite compositions to modern calc-alkaline granites. These modifications are best explained by
variations of partial melting conditions in subduction zones, within the framework of a cooling planet.

28
However, great debate remains surrounding the geodynamic sites of formation of the continental crust
and the conditions at the surface of the Earth during Archean times.

The present thesis is focused on the southernmost portion of the São Francisco craton (SE
Brazil), and more specifically on the Quadrilátero Ferrífero district, built and consolidated between the
Mesoarchean and Paleoproterozoic. The geodynamic evolution of this crustal fragment is relatively
poorly understood, and hence it represents an object of choice to investigate the secular evolution of
the continental crust during this period. In particular, does the crust in the Quadrilátero Ferrífero
record the important changes observed on a planetary scale? If so, what is their significance in terms
of geodynamics? What were the timing, sites and mechanisms of crustal growth? How and when did
the craton stabilize?

The following chapter presents a comprehensive review of the geology of the Quadrilátero
Ferrífero, in the form of an article published in Journal of South American Earth Sciences in 2016. It
is important to note that: (1) it aims to provide an up-to-date summary of the scientific knowledge
surrounding the Archean and Paleoproterozoic evolution of the region, but intentionally omits certain
geological aspects (i.e. the origin of the numerous ore deposits are not considered), and (2) this review
postdates the publication of (and hence include) the first results obtained within the framework of this
thesis, which are reported in the third chapter of this volume.

29
30
 CHAPTER 2
THE ARCHEAN-PALEOPROTEROZOIC EVOLUTION OF THE
QUADRILÁTERO FERRÍFERO (BRAZIL): CURRENT MODELS AND
OPEN QUESTIONS
2

ABSTRACT

The Quadrilátero Ferrífero is a metallogenic district (Au, Fe, Mn) located at the southernmost
end of the São Francisco craton in eastern Brazil. In this region, a supracrustal assemblage composed
of Archean greenstone and overlying Neoarchean-Paleoproterozoic sedimentary rocks occurs in
elongated keels bordering domal bodies of Archean gneisses and granites. The tectonomagmatic
evolution of the Quadrilátero Ferrífero began in the Paleoarchean with the formation of continental
crust between 3500 and 3200 Ma. Although this crust is today poorly preserved, its existence is
attested to by the occurrence of detrital zircon crystals with Paleoarchean age in the supracrustal rocks.
Most of the crystalline basement, which is composed of banded gneisses intruded by leucogranitic
dikes and weakly foliated granites, formed during three major magmatic events: Rio das Velhas I
(2920-2850 Ma), Rio das Velhas II (2800-2760 Ma) and Mamona (2760-2680 Ma). The Rio das
Velhas II and Mamona events represent a subduction-collision cycle, probably marking the
appearance of a modern-style plate tectonic regime in the Quadrilátero Ferrífero. Granitic rocks
emplaced during the Rio das Velhas I and II events formed by mixing between a magma generated by
partial melting of metamafic rocks with an end member derived by recycling gneissic rocks of older
continental crust. After deformation and regional metamorphism at ca. 2770 Ma, a change in the
composition of the granitic magmas occurred and large volumes of high-K granitoids were generated.
The ca. 6000 m-thick Minas Supergroup tracks the opening and closure of a basin during the
Neoarchean-Paleoproterozoic, between 2600 and 2000 Ma. The basal sequence involves continental to
marine sediments deposited in a passive margin basin and contain as a marker bed the Lake Superior-
type Cauê Banded Iron Formation. The overlying sediments of the Sabará Group mark the inversion of
the basin during the Rhyacian Minas accretionary orogeny. This orogeny results from the collision
between the nuclei of the present-day São Francisco and Congo cratons, generating the fold-and thrust
belt structure of the Quadrilátero Ferrífero. Afterwards, the post- orogenic collapse resulted in the
deposition of the Itacolomi Group and in the genesis of the dome-and-keel structure. In this paper, we
review current knowledge about the 1500 Ma long-lasting tectonomagmatic and structural evolution of
the Quadrilátero Ferrífero, identifying the most compelling open questions and future challenges.

F. Farina*, C. Albert, C. Martínez Dopico, C. Aguilar Gil, H. Moreira, J. P. Hippertt, K. Cutts, F. F. Alkmim, C. Lana

Departamento de Geologia, Universidade Federal de Ouro Preto, Morro do Cruzeiro, 35400-000, Ouro Preto, MG, Brazil
* Corresponding author: Federico Farina. E-mail: [email protected]
Article published in Journal of South American Earth Sciences 68 (2016), 4-21
Keywords: Quadrilátero Ferrífero; Archean; Paleoproterozoic; Tectonomagmatic evolution; São Francisco craton

31
2.1 INTRODUCTION

In the mid-twentieth century, the ca. 7000 km2 portion of the Brazilian highlands south of the
city of Belo Horizonte (Fig. 2.1) became well-known under the name of Quadrilátero Ferrífero (―Iron
Quadrangle‖). This region, which has been an important mining site since the XVII century, represents
the most intensively studied region of Brazil. Its valuable iron deposits together with its puzzling
geological complexity have attracted the attention of many scientists over the last century. The goal of
the present paper is to provide an up-to-date state of the art about the geology of the Archean
crystalline basement and Archean-Paleoproterozoic metasedimentary sequences forming the
Quadrilátero Ferrífero. This review, focussed on the Archean and Paleoproterozoic evolution of this
portion of the southern São Francisco craton, is not aiming to be exhaustive. Some important topics
will not be addressed (e.g. the origin of ore deposits) while others will be only briefly presented (e.g.
the structural evolution). The goal of this contribution is to discuss the present-day petrologic and
geodynamic models proposed for the Quadrilátero, drawing attention to the main open problems and
limitations in an attempt to highlight future challenges and favourable research directions.

2.2 HISTORICAL BACKGROUND AND GEOGRAPHIC BOUNDARIES

In the sixteenth century, the discovery of large gold and silver deposits in the Spanish colonial
possessions in South America triggered the Portuguese empire to embark on the exploration of inland
Brazil seeking precious metals. For ca. 100 years the hunt for payable precious metals proved futile
and only in 1646 the first discovery of small alluvial gold deposit in Brazil was made in the southern
state of Paraná (Figueiredo, 2011). In 1693, the first gold placer deposit was found in southeastern
Brazil (state of Minas Gerais), close to the city of Ouro Preto (―black gold‖ in Portuguese). The
discovery of many rich gold deposits in the region that is today known as the Quadrilátero Ferrífero,
revitalized Brazil's economy causing such a stir that, by the end of the century a considerable
proportion of Brazil‘s population had rushed to the site of the discovery (Costa et al., 2003). More
significantly, as news of the discovery spread to the mother country, thousands of Portuguese
adventurers moved to Brazil at the turn of the eighteenth century: the first great gold rush had begun.
Between 1693 and 1720, the population of the gold-rich province grew exponentially. Such was the
growth that half of Brazil's entire population was residing in Minas Gerais in 1725. Gold production
increased as the eighteenth century advanced, peaking around mid-century and then, due to the
rudimental mining technique implemented (e.g. mostly panning and sluicing in shallow water stream),
slowly declined (Costa et al., 2003).

32
Figure 2.1 – Geological map of the Quadrilátero Ferrífero modified after Alkmim & Marshak (1998). Batholith
and pluton abbreviations: C – Caeté; F – Florestal; M – Mamona; P – Pequi; As – Samambaia; SN – Souza
Noschese. Inset: tectonic sketch of the São Francisco craton showing the location of the bordering Brasiliano
orogenic belts as well as of the Paleoproterozoic Mineiro Belt. Abbreviations: G, J, IS and S are the Gavião,
Jequié, Itabuna-Salvador-Curaçá and Serrinha blocks, respectively.

In 1810, the mineralogist and geologist Baron Wilhelm Ludwig von Eschwege arrived in
Brazil entrusted by King Dom João VI to study the ores and the mining activity with the goal of
implementing new mining techniques to increase gold production. The baron made a significant step
forward in the description and understanding of the geology of the Quadrilátero Ferrífero. Eschwege
published various books (e.g. Pluto brasiliensis (1833)) in which he outlined the main geological
features of the Precambrian terranes of central Brazil proposing a stratigraphic subdivision according
to Werner's Neptunism theory.

In the early nineteenth century, the gold-rich region around Ouro Preto returned to the
spotlight because of its iron and manganese deposits. One century after Eschwege arrived in Brazil,
the American geologist Derby published a work titled ―The iron ores of Brazil‖ that attracted the
interest of the mining community to the iron and manganese deposits of Minas Gerais. In the mid-
twentieth century, an agreement was set between the Brazilian National Department of Mineral
Production (DNPM) and the U.S. Geological Survey to undertake the first detailed geological study of
the region where the main iron and manganese deposits were located (Machado, 2009). This
agreement and the two decades of work that followed marked the most important step forward in the

33
understanding of the geology of the Quadrilátero Ferrífero. The outcome of this joint project was a set
of geological maps at the 1:25.000 scale followed by a stratigraphic column and a report written by the
head of the team, John Van N. Dorr II, in which the main geological features of the studied region
were defined. Dorr and collaborators coined the term Quadrilátero Ferrífero in 1952 (Machado, 2009)
to give credit to the abundance of high-grade iron ore deposits in the region. These authors defined the
geographic boundaries of the Quadrilátero producing a detailed 1:150.000 geological map. Oddly, as
shown in Fig. 2.1, the ―Iron Quadrangle‖ of Dorr and collaborators is not a quadrangle but rather a
complex polygon, including the Itabira mining district. The reason for the use of the term Quadrilátero
proposed by Dorr is historical. In fact, the term was introduced in 1923 by the Brazilian geologist
Flores de Moraes Rego to define the four-sided area comprised between the cities of Belo Horizonte,
Santa Bárbara, Congonhas and Mariana (Machado, 2009). Clear geographic features do not delimit the
Quadrilátero Ferrífero as defined and mapped by Dorr and collaborators and thus, the term has been
used loosely to describe the geology of areas that are not part of the original map of the Quadrilátero.
As shown in Fig. 2.1, we will also extend the limits of the Quadrilátero Ferrífero to include the
Bonfim and the Belo Horizonte domes in their entirety.

2.3 MAIN GEOLOGICAL FEATURES

The São Francisco Craton in eastern Brazil is one of the major and probably the best-exposed
shield area forming the South American platform. The craton is subdivided into several Archean to
Paleoproterozoic blocks bounded by major ca. 2100 Ma old sutures zones (Teixeira & Figuereido,
1991; Barbosa & Sabaté, 2004) and surrounded by Neoproterozic fold-belts (e.g. the Araçuaí belt;
Pedrosa-Soares et al., 2001). Four crustal segments form the northern part of the craton (Barbosa &
Sabaté, 2004): the Gavião, Jequié, Serrinha and Itabuna-Salvador-Curaçá blocks (Fig. 2.1). To the
south, the São Francisco Craton comprises various granitoid-gneiss complexes (i.e. Campo Belo,
Passa Tempo, Bonfim, Belo Horizonte, Bação, Caeté) and Archean-Paleoproterozoic supracrustal
sequences and hosts the Quadrilátero Ferrífero mining district. The Quadrilátero Ferrífero occupies the
eastern part of the Southern São Francisco Craton and is fringed by the Late Neoproterozoic-Cambrian
Araçuaí belt to the east and by the Paleoproterozoic Mineiro Belt to the south (Fig. 2.1). This district
exhibits NNW-verging folds and thrusts and has a metamorphic overprint at about 2100-2000 Ma,
which was originally known as the ―Minas diastrophism‖ (e.g. Cordani et al., 1980). The distinctive
structural architecture of the Quadrilátero is its dome-and keel geometry, in which belts of low-grade
Paleoproterozoic supracrustal rocks surround medium to high-grade granitoid-gneiss Archean
complexes (Marshak et al., 1997). The Quadrilátero Ferrífero can be subdivided into four Archean-
Paleoproterozoic lithostratigraphic units (Fig. 2.2):

(i) Archean metamorphic complexes composed of gneisses, migmatites, and granitoids;

(ii) the Archean Rio das Velhas Supergroup, formed of low- to medium-grade
metavolcanic and metasedimentary rocks;
34
 (iii) the Neoarchean-Paleoproterozoic Minas Supergroup, consisting of low- to medium-
grade metasedimentary rocks;
(iv) the Paleoproterozoic Itacolomi Group composed of metasandstones and
conglomerates.

In addition, the Quadrilátero Ferrífero includes small granite bodies and pegmatite veins
locally cutting the youngest strata of the Minas Supergroup as well as different generations of mafic
dikes showing contrasting metamorphic grade, composition and trending direction. A stratigraphic
section of the Quadrilátero Ferrífero is presented in Fig. 2.2 and the different lithostratigraphic units
are described in the following paragraphs.

Figure 2.2 – Stratigraphic column of the supracrustal sequences in Quadrilátero Ferrífero. Abbreviations: RVI
and RVII are Rio das Velhas I and II events, SB is the Santa Bárbara magmatic event. Modified after Dorr
(1969) and Alkmim & Marshak (1998).

35
2.4 THE ARCHEAN METAMORPHIC COMPLEXES

2.4.1 Field relationships and petrography

In the field, the rocks of the basement can be subdivided into three main groups:

(i) fine-grained banded orthogneisses intruded by (ii) and (iii);

(ii) leucogranites and aplitic/pegmatitic veins and dikes;
(iii) medium- to coarse-grained, mostly weakly foliated granites (sensu lato).

The gneisses are characterized by the alternation between leucocratic and mesocratic, or more
rarely, melanocratic bands, varying in width from 2 mm up to 10 cm and defining a penetrative
amphibolite-facies foliation (Fig. 2.3a and b). The mesocratic bands are rich in plagioclase and biotite
and display lepidoblastic textures, whereas the leucocratic ones, containing predominantly plagioclase,
quartz and minor microcline, show granoblastic textures (Lana et al., 2013). Alkali-feldspars are
generally interstitial, but they occasionally form cm-scale phenocrysts. A subset of gneiss samples
show complex polydeformed structural patterns exhibiting disharmonic folding of the banding
associated with high volumes (up to 20 vol.%) of non-foliated leucogranitic material forming folded
sheets as well as decimetric pods. Locally, some outcrops display an even more chaotic aspect with
boudinaged and dismembered gneiss fragments isolated in a granitic matrix. Farina et al. (2015a)
interpret these rocks as migmatites.

The gneisses are typically intruded by multiple, metre- to cm-scale leucogranitic sheets
subparallel to the gneissosity as well as by crosscutting younger felsic and/or pegmatitic dikes (Lana et
al., 2013). Foliated or massive leucogranitic sheets are oriented subparallel to the banding and have
thicknesses of up to 60 cm (Fig. 2.3c and d). A younger generation of leucogranitic, pegmatitic and
aplitic dikes that crosscut both the gneissic banding and the leucogranitic sheets have widths reaching
a maximum of about 2 m (Fig. 2.3c). These dikes are only occasionally slightly folded or boudinaged,
but in general appear little stretched, nor shortened.

Granitic rocks form texturally and compositionally composite large batholiths (e.g. the
Mamona batholith; Fig. 2.1) as well as decimetre- to metre-scale domains and stocks intruded in or
closely associated with the banded gneisses (Farina et al., 2015a). Although typically weakly foliated,
granitoids may also locally develop a prolate L > S fabric. The granitic rocks are medium- to coarse-
grained and exhibit either equigranular or porphyritic textures ranging in composition from tonalite to
syenogranite. Based on their ferromagnesian minerals, three rock types are recognised: biotite-bearing
granodiorites and granites (Fig. 2.3e), two-mica granites and biotite- and amphibole-bearing tonalites
and granodiorites. The first group represents the most abundant granite-type while muscovite- and
amphibole-bearing granitoids are rare, mostly cropping out in the Bonfim complex. The relationship
between banded gneisses and granitoids is observable in some key outcrops in the Bação and Bonfim

36
complexes. The intrusion of granites into banded gneisses produces different features. From gneiss- to
granite-dominated outcrops, we observe:

(i) dikes of medium-grained granites cutting both the gneissic banding and the
leucogranite sheets;
(ii) complex intermingled gneiss-granite structures (Fig. 2.3f);
(iii) metre- to decametre-scale domains of granites intruding the banded gneisses;
(iv) xenoliths of banded gneisses hosted within granites.

Figure 2.3 – Field photographs of the basement. (a) Banded gneiss, Bação complex; (b) Banded gneiss hosting
cm-scale leucogranitic sheets following the gneissosity, Bação complex; (c) Leucogranitic sheets in a gneiss
37
crosscut by a late fine-grained leucocratic dike; (d) Leucogranitic dike parallel to foliation in the banded gneiss,
Bação complex; (e) Medium-grained granodiorite containing cm-sized K-feldspar and rare mafic enclaves,
Mamona batholith, Bonfim complex; (f) Domains of medium- to coarse-grained granitoid intruding banded
gneisses, Bonfim complex. Hammer for scale is 30 cm long in a, b, c, f and 35 in d. The coin in e is 2 cm in
diametre.

2.4.2 Geochemistry: medium- and high-K granitoids

Whole rock major and trace element data for granites and gneisses in the Quadrilátero
Ferrífero is scarce, with different contributions generally investigating the compositional variability of
individual complexes without attempting any regional-scale correlation (Carneiro & Teixeira, 1992;
Noce et al., 1997). An exception to this approach is represented by the work of Farina et al. (2015a, b)
in which a large database of chemical compositions for the granites and gneisses of the Bação, Bonfim
and Belo Horizonte complexes was produced. In Fig. 2.4, 2.5 and 2.6, we plotted the major and trace
element composition of gneisses and granitoids from the Bação, Bonfim, Belo Horizonte and Caeté
complexes as well as the composition of dacites from the Rio das Velhas greenstone belt. A complete
dataset of whole rock compositions is presented in Appendix A1.

Gneisses, granitoids and leucogranites of the Quadrilátero Ferrífero are silica-rich (i.e. ca. 70
% of samples have silica content higher than 72 wt.%) and all oxides are broadly negatively correlated
with SiO2 (Fig. 2.5a and b), except K2O which displays a positive correlation.

In the normative feldspar classification diagram for granitoids (An-Ab-Or; O'Connor, 1965),
most of the gneisses and granites plot either in the trondhjemite or in the granite fields (Fig. 2.4), with
only the biotite- and hornblende-bearing Samambaia pluton described by Carneiro (1992) plotting in
the tonalite field. The two groups can be distinguished in the K 2O vs. SiO2 diagram where most of the
trondhjemites of Fig. 2.4 plot in the medium-K field defined by Gill (1981) while granitic rocks plot in
the high-K field (Fig. 2.5b). The occurrence of medium- and high-K gneisses and granitic rocks in the
basement of the Quadrilátero Ferrífero is evident once the K 2O/Na2O of the rocks is plotted in a
frequency histogram (Fig. 2.5c): the distribution is bimodal showing well-defined peaks at 0.4-0.6 and
1.2-1.6 K2O/Na2O. Systematic compositional differences exist between medium- and high-K rocks.
Medium-K gneisses and granitoids have higher Al2O3, Na2O, CaO and Sr as well as lower silica and
Rb than high-K rocks (Fig. 2.5). In addition, medium-K granitoids have relatively high contents of
Light Rare Earth Elements (LREE; i.e. average La-Sm = 120 ppm) and low Heavy Rare Earth Element
contents (HREE; i.e. average Gd-Lu = 8.2 ppm), resulting in steep REE patterns (La/Yb up to 70)
while high-K rocks exhibit less fractionated REE patterns with HREE contents that are significantly
higher (average Gd-Lu = 38 ppm) than those of medium-K rocks (Fig. 2.6). The two rock types
display different Eu anomalies with no Eu anomaly or slightly negative in the medium-K rocks and a
well pronounced Eu anomaly for high-K granitoids and gneisses. Overall, medium-K2O rocks share
similar chemical features with rocks of the Archean tonalite-trondhjemite-granodiorite (TTG) series

38
while high-K granitoids are similar to high-silica I-type granites. The latter plot consistently in the
field of biotite- and two-mica granites recently defined by Laurent et al. (2014) for late-Archean
granites (Farina et al., 2015a).

Figure 2.4 – Normative An-Ab-Or triangle (O'Connor, 1965) showing the composition of gneisses, granites and
leucogranites as well as the composition of Rio das Velhas dacites. The field for Archean TTGs is from Moyen
& Martin (2012). Data are from Gomes (1985), Carneiro et al. (1992), Noce et al. (1997), Da Silva et al. (2000)
and Farina et al. (2015a).

Finally, leucogranitic sheets and dikes, representing ca. 10 % of the bulk crystalline basement
exposed in the Quadrilátero Ferrífero, display characteristic low to very low MgO + Fe 2O3tot contents
(0.2-1.1 wt.%) and scattered major element composition with K2O ranging from 3.8 to 8.9 wt.% and
silica between 71 and 76 wt.% (Fig. 2.6). In the SiO2 vs. Al2O3 diagram, they generate a quasi-linear
negative trend and for equivalent silica contents they are slightly enriched in Al 2O3 and depleted in
MgO and Fe2O3tot than both granites and gneisses (Fig. 2.5). The leucogranitic rocks exhibit Eu
anomalies varying from slightly negative to strongly positive (Eu/Eu* = 0.9-2.8), low REE contents
and flat REE patterns resulting in low La/Yb ratios.

39
Figure 2.5 – Harker diagrams for the igneous rocks of the Quadrilátero Ferrífero: (a) SiO2 vs. Al2O3; (b) SiO2 vs.
K2O. Grey dots are TTGs from Moyen (2011). In (c), histogram showing the frequency of K 2O/Na2O values
exhibited by the rocks of the Quadrilátero Ferrífero. The bin width used is 0.2. The height of the bars represents
the number of samples with the corresponding K2O/Na2O value.

40
Figure 2.6 – Average chondrite-normalized REE patterns for high-K granites and medium-K gneisses and
granitoids. The trace element pattern for high-K granites is obtained averaging the composition of eleven
samples, patterns for medium-K gneisses and granitoids are from fifteen and sixteen samples, respectively. Data
are from Farina et al. (2015a). The TTG field is based on the low-to high-pressure TTGs after Moyen (2011).
Normalization values are from McDonough & Sun (1995).

2.4.3 Geochronology

A review of published U-Pb ages from the basement of the Quadrilátero Ferrífero (Fig. 2.7
and Appendix A2) allowed for the identification of four main magmatic events (Farina et al., 2015a, b;
Lana et al., 2013; Romano et al., 2013). These periods of magmatic activity, described as the Santa
Bárbara (SB), Rio das Velhas I (RVI), Rio das Velhas II (RVII) and Mamona, embody a significant
part of the protracted tectonomagmatic Archean history of the Quadrilátero Ferrífero, spanning from
3220 to 2680 Ma. The first magmatic pulse, poorly preserved in the north-south elongated Santa
Bárbara complex (Fig. 2.1), ranges from 3220 to 3200 Ma, with these rocks representing the only
Paleoarchean crust dated in the Quadrilátero Ferrífero so far. Most of the gneisses in the Quadrilátero
Ferrífero formed during the following periods of magma production, the Rio das Velhas I and II
events (Lana et al., 2013). The Rio das Velhas I period, which was originally supposed to have ages
spanning between 2930 and 2900 Ma, has been recently expanded down to 2850 Ma by Farina et al.
(2015a). Similarly, new geochronological data of Farina et al. (2015a) redefine the time of the Rio das
Velhas II period between 2800 and 2760 Ma. It is worth noting that two weakly deformed plutons
intruded the orthogneisses during the Rio das Velhas II event at about 2770 Ma (the Caeté and
Samambaia plutons; Machado et al., 1992). The age of these granites suggest that the last regional
metamorphic event in the Quadrilátero Ferrífero occurred in the early Neoarchean, during the Rio das
Velhas II event. The occurrence of this metamorphic event is supported by the observation that many
magmatic zircon grains from the banded gneisses of the Rio das Velhas I period are overgrown by
41
metamorphic rims yielding Rio das Velhas II ages (Lana et al., 2013). Just after the regional
metamorphic event, the Quadrilátero Ferrífero underwent a new period of widespread magmatism
between 2750 and 2680 Ma (Romano et al., 2013) that has been named by Farina et al. (2015a): the
Mamona event. During this event, slightly- to non-foliated rocks were emplaced as large batholiths
and small leucogranitic veins and dikes into the pre-existing deformed crust. Finally, a volumetrically
minor event of granite production accounting for less than 1 % of the continental crust in the
Quadrilátero occurred at ca. 2612 Ma (Romano et al., 2013).

Two leucogranitic-pegmatitic dikes analysed by Machado et al. (1992) in the southern part of
the Bação complex gave Paleoproterozoic U-Pb monazite ages (i.e. 2130 and 2122 Ma; Appendix
A3). These ages match with U-Pb data obtained from metamorphic titanites from amphibolitic dikes
(ca. 2050 Ma) and gneisses in the Bação, Bonfim and Belo Horizonte complexes (Machado et al.,
1992; Noce et al., 1998; Aguilar et al., 2017) as well as with the crystallization age of the Alto
Maranhão suite (Noce et al., 1998; Seixas et al., 2013). The Alto Maranhão tonalitic-dioritic batholith,
located on the southern edge of the Quadrilátero Ferrífero, is part of the Mineiro Belt, formed during
the Rhyacian orogenesis of the South American platform (Seixas et al., 2013).

Figure 2.7 – Timeline showing 207Pb/206Pb zircon ages for intrusive and volcanic rocks of the Quadrilátero
Ferrífero. Circles, squares, diamonds and triangles indicate the Bação, Bonfim, Belo Horizonte and Santa
Bárbara complexes, respectively. White and dark grey symbols are for high- and medium-K rocks, respectively.
In black, mafic and intermediate dikes and metamorphic zircon ages. The vertical light grey fields indicate the
different magmatic events: SB – Santa Bárbara; RdV I – Rio das Velhas I; RdV II – Rio das Velhas II; Mam –
Mamona. Data are from Romano et al. (2013); Lana et al. (2013); Machado & Carneiro (1992); Machado et al.
(1992); Noce et al. (1998); Noce et al. (1997); Chemale et al. (1993a); Noce et al. (2005).

42
2.5 THE RIO DAS VELHAS SUPERGROUP

2.5.1 Stratigraphy

The metavolcanic and metasedimentary rocks of the Rio das Velhas Supergroup form a typical
Archean greenstone belt sequence characterized by the association between mafic and ultramafic rocks
(komatiite-basalt; Fig. 2.8a and b), evolved volcanic (dacites) and volcaniclastic rocks and immature
clastic sediments (Dorr, 1969). These rocks are metamorphosed at greenschist to lower amphibolite-
facies conditions and are commonly affected by hydrothermal alteration (Ladeira et al., 1983;
Zucchetti et al., 2000). Many different stratigraphic subdivisions were proposed for the Rio das Velhas
Supergroup (Baltazar & Zucchetti, 2007 and references therein). A first level of classification was
introduced by Dorr (1969), which subdivided the greenstone belt into the Nova Lima and Maquiné
Groups, the former occurring at the base of the sequence and hosting the major gold deposits of the
Quadrilátero Ferrífero (Fig. 2.2). Recently, Baltazar & Zucchetti (2007), following the approach of
Eriksson et al. (1994) subdivided the Nova Lima Group into six sedimentary lithofacies associations
forming four sedimentary cycles. From bottom to top these are: (i) mafic-ultramafic volcanic; (ii)
volcano-chemical-sedimentary; (iii) clastic-chemical-sedimentary, (iv) volcaniclastic; (v)
resedimented and (vi) coastal. The basal lithofacies of the Nova Lima Group is made of mafic and
ultramafic lavas intercalated with minor intrusions of gabbro, anorthosite and peridotite as well as with
banded iron formation, ferruginous cherts, chemical carbonaceous sediments and rare felsic
volcaniclastic rocks. The ultramafic lavas are massive and pillowed komatiites characterized by
spinifex textures (Schorscher, 1978; Sichel, 1983), with layers of cumulus olivine/intercumulus
orthopyroxene and a level of lahar-type breccia (Baltazar & Zucchetti, 2007). This event of mafic
volcanism and carbonate precipitation was followed by submarine deposition of pelites, greywackes
and quartzites with intercalated rocks of banded iron formation, quartz-dolomites, quartz-ankerites
(hosting the world-class Morro Velho gold deposit; Lobato et al., 2001), conglomerates and
carbonaceous phyllites (lithofacies (ii) and (iii)). Volcaniclastic and resedimented volcaniclastic rocks
as well as minor lava flows of dacitic composition and turbiditic graywakes form the overlying
lithofacies (lithofacies (iv) and (v)). It is worth noting that the minor dacitic lava flows have a TTG
signature, exhibiting high Na2O and Al2O3 contents (Fig. 2.5) and strongly fractionated Rare Earth
Element patterns that are consistent with partial melting of garnet amphibolite sources (Da Silva et al.,
2000). Finally, the lithofacies at the top of the sequence (lithofacies (vi)) is formed of sandstones with
herringbone cross-bedding, ripple marks and large-scale cross-bedding. This lithofacies is restricted to
a small area northwest of the Bação complex and was probably deposited in a shallow marine
environment.

43
Figure 2.8 – Field photographs of the supracrustal rocks. (a) Komatiite showing spinifex structure, basal portion
of the Nova Lima Group; (b) Metabasalt with deformed pillow structure, basal portion of the Nova Lima Group;
(c) Polymictic conglomerate of the Maquiné Group containing stretched clasts; (d) Cauê Itabirite showing the
typical intercalation of hematite and quartz rich bands; (e) Conglomerate of the Sabará Group containing clasts
of quartzite, gneisses, granites and banded iron formation embedded in a chlorite-rich matrix; (f) Quartz
metasandstone of the Itacolomi Group showing crossbedding. Crossbedding sets are marked by concentration of
heavy minerals, especially iron oxides. The pens in b, c, d and e are 14 cm-long.

Overlying the Nova Lima Group, the Maquiné Group represents a 2000 m-thick clastic
association comprising conglomerates and sandstones (Fig. 2.8c) that was described by Dorr (1969) as
a flysch to molasse-type sequence, consisting of a coarsening upward succession of sandstones getting
more quartz-rich and conglomeratic toward the top. The Maquiné Group was divided into two
44
formations: the basal Palmital (O'Rourke, 1957) and the upper Casa Forte (Gair, 1962). The Palmital
Formation was deposited in a marine environment as proximal turbidites while the Casa Forte
Formation is interpreted as a non-marine alluvial fan-braided river deposit. The contact between the
Nova Lima Group and the overlying sandstones and conglomerates of the Maquiné Group is either
gradational or locally unconformable and marked by fault zones (Dorr, 1969). The lack of a clear
discordance between the Palmital Formation and the top of the Nova Lima Group as well as the
marine environment of deposition for the Palmital Formation led Baltazar & Zucchetti (2007) to
associate this formation to the coastal association defined in the Nova Lima Group.

2.5.2 Geochronology

U-Pb ages of detrital zircons from the Rio das Velhas greenstone belt were determined by
Machado et al. (1992, 1996), Noce et al. (2005) and Hartmann et al. (2006) using different analytical
techniques such as SHRIMP, LA-ICP-MS and ID-TIMS (Appendix A4). Eight of the nine samples
analysed are greywackes from the Nova Lima group while only five zircon grains were analysed for
one sample collected by Machado et al. (1996) from the Maquiné Group. Three volcaniclastic
greywackes in the Nova Lima Group gave maximum deposition ages of 2792 ± 11, 2773 ± 7 and 2751
± 9 Ma, indicating ca. 40 Ma of felsic volcanism in the Quadrilátero Ferrífero (Machado et al., 1992,
1996; Noce et al., 2005). The occurrence of a Neoarchean felsic volcanic event is supported by a
zircon 207Pb/206Pb crystallization age of 2772 ± 6 Ma determined by Machado et al. (1992) for a dacitic
flow intercalated within the sequence of mafic to ultramafic volcanic rocks in the greenstone belt.
Detrital zircons from two sandstones from the top of the Nova Lima Group dated by SHRIMP by
Hartmann et al. (2006) gave a maximum depositional age of 2749 ± 7 Ma.

U-Pb age data from detrital zircon grains in the Nova Lima Group are plotted in the frequency
diagram of Fig. 2.9. In this diagram, ages that are more than 10 % discordant were excluded together
with spot analyses with Th/U < 0.1. Detrital zircon grains from the volcaniclastic greywackes of the
Nova Lima Group define a polymodal age spectrum with ages ranging from 2700 to 3450 Ma. The
occurrence of a large number of detrital zircon grains suggests that these rocks were not formed in an
intra-oceanic arc environment, but more likely in an intra-continental or continental-margin tectonic
setting (Noce et al., 2005). When the age distribution of detrital zircons in the Nova Lima Group is
compared with the age of the main magmatic events defined by Lana et al. (2013), Romano et al.
(2013) and Farina et al. (2015a), the following observations stand out:

- The main magmatic event preserved in the metasedimentary rock record is the 2800-2760
Ma Rio das Velhas II event;
- The second highest frequency peak is at ca. 2860 Ma matching the youngest limit of the
Rio das Velhas I event (2920-2860 Ma);
- There is a peak at ca. 3200 Ma matching the age of the Santa Bárbara event;

45
 - The ca. 3000 Ma peak suggests an event of continental crust production that took place
between the Rio das Velhas I and Santa Bárbara magmatic events;
- Small peaks at ca. 3450 and 3550 Ma were found by Noce et al. (2005) and Machado et
al. (1996), respectively. The ages of these peaks do not match with any of the magmatic
event preserved in the basement, suggesting the existence of pre-3200 Ma (older than the
Santa Bárbara event) continental crust.

Figure 2.9 – Frequency histogram showing the age distribution of detrital zircon grains in the rocks of the Rio
das Velhas Supergroup. The probability curve is produced considering 109 U-Pb analyses from Machado et al.
(1992); Machado et al. (1996); Noce et al. (2005); Hartmann et al. (2006). Maximum discordance accepted 10
%. Analyses with Th/U < 0.1 were excluded. Vertical bands mark the age of the main magmatic events in the
basement.

The limited available U-Pb age data for the Maquiné Group suggest that a continental block
ranging in age from 3260 to 2877 Ma was the main source for the sandstones and conglomerates of
this group (Machado et al., 1996). A recent study has associated this sedimentary sequence with the
genesis of an inferred arc formed during the Rio das Velhas II event (Lana et al., 2013). Moreira et al.
(2016) produced geochronological U-Pb data of ca. 1500 zircon grains from different units of the
Maquiné Group. The new data indicate that the main sources of the basin are rocks formed between
2760 and 2800 Ma (i.e. Rio das Velhas II event) and that the maximum depositional age for the Casa
Forte Formation is 2730 Ma, thus suggesting that the basin closed after the Mamona magmatic event.

46
 Finally, Schrank & Machado (1996a, b) dated monazite grains from the Nova Lima and upper
Maquiné Groups obtaining Paleoproterozoic metamorphic ages spanning between 2080 Ma and 1989
Ma (Appendix A3). These data are consistent with the ages obtained for metamorphic monazite and
titanite grains from the basement (Noce et al., 1998; Aguilar et al., 2017) as well as with U-Pb zircon
magmatic ages from two leucocratic dikes analysed by Machado et al. (1992) in the southern part of
the Bação complex.

2.6 THE MINAS SUPERGROUP

2.6.1 Main stratigraphic features

The Paleoproterozoic Minas Supergroup (Dorr, 1969; Babinski et al., 1991; Renger et al.,
1995) is a ca. 6000 m-thick package of clastic and chemical rocks (Alkmim & Marshak, 1998) lying
unconformably on the Archean greenstone belt (Fig. 2.2). Following Alkmim & Martins-Neto (2012),
the Minas Supergroup can be subdivided into two sequences separated by a regional unconformity.
The basal sequence, involving continental to marine sediments (Dorr, 1969; Renger et al., 1995)
represents the development stage of a passive margin basin (Schorscher, 1992; Canuto, 2010). The
overlying sequence, consisting of turbidites of the Sabará Group was interpreted as a submarine fan
deposit marking the inversion of the passive margin (Alkmim & Marshak, 1998).

The basal continental to marine sedimentary sequence has been further subdivided into:

- A ca. 600 m-thick package of gold-uranium-bearing alluvial to marine sandstones,

conglomerates and subordinate offshore pelites comprising the Tamanduá and Caraça
groups (Simmons & Maxwell, 1961; Dorr, 1969). These rocks represent the rift and
transitional phases of the passive margin development (Renger et al., 1995; Alkmim &
Marshak, 1998);
- A 400 m-thick package of marine sediments that includes banded iron formations and
carbonates of the Itabira Group (Dorr, 1969), recording the thermal subsidence stage of
the continental passive margin (Alkmim & Marshak, 1998);
- A ca. 450 m-thick pile of deltaic to shallow marine to deep water sediments (Piracicaba
Group; Renger et al., 1995), consisting mainly of siliciclastic rocks with minor carbonates.

In the next sections, we will present the main stratigraphic and geochronological features of
each group. A summary of ages for the Minas Supergroup is presented in Appendix A5.

2.6.2 Tamanduá and Caraça Groups

The beginning of the rifting stage is recorded by the deposition of the Tamanduá Group and by
the overlying basal unit of the Caraça Group: the Moeda Formation (Dorr, 1969). The Tamanduá
Group (Simmons & Maxwell, 1961; Dorr, 1969) comprises mostly quartzites, quartzose and
argillaceous schists – also referred to as Cambotas Quartzite – and minor phyllitic and dolomitic

47
itabirites at the top. The unit crops out in small and discontinuous areas, lying in contact with the
Maquiné Group by an erosional unconformity (Dorr, 1969). The absence of well-preserved
sedimentary structures or marker beds hinders possible correlations between this group and other
similar deposits within the Quadrilátero Ferrífero.

The Caraça Group (Dorr, 1959) is a blanket deposit, having great lateral extent compared with
its average thickness. It encompasses metasedimentary quartzitic rocks (Moeda Formation) and
phyllites (Batatal Formation) that conformably underlie chemical sediments of the Itabira Group, and
commonly overlie with angular and erosional discordance rocks of the Archean greenstone belt. The
Moeda Formation comprises mostly quartzites and minor basal pyritiferous and locally auriferous
metaconglomerates, cropping out predominantly in the western part of the Quadrilátero Ferrífero
(Dorr, 1969; Vilaça, 1981; Renger et al., 1988; Koglin et al., 2012). The average thickness of the
formation is ca. 300 m, locally reaching up to 1000 m. This sequence represents a braided river system
locally alternating with deltaic to beach deposits as well as with thin deposits formed during marine
transgression events (Vilaça, 1981; Canuto, 2010). The Moeda Formation exhibits significant
mineralogical and grain size lateral variation. In several localities, the basal conglomerates exhibit
lenticular shaped pebbles and cobbles of phyllites probably from the Nova Lima Group as well as
rounded quartz and quartzite cobbles (Dorr, 1969; Koglin et al., 2012). Some of the basal fluvial
metaconglomerates exhibit heavy mineral layers of detrital pyrite with economic gold concentrations
(Renger et al., 1988). The Caraça cycle ends with the deposition of the Batatal Formation (Maxwell,
1958; Dorr, 1969) that conformably overlies the Moeda Formation. The contact between the two
formations is commonly sharp, but locally the units can also be intergradational (Wallace, 1965). The
Batatal formation consists of bluish-grey phyllites with minor metacherts, iron formation and graphitic
phyllites (Simmons, 1968) cropping out in the western and central areas of the Serra do Curral where
it has a thickness ranging from 30 to 200 m (Dorr, 1969). The sedimentation of the Batatal Formation
reflects the transition towards a marine to coastal marine environment (Moraes, 1985), registering the
passage from the rift-opening stage to the passive margin stage (Alkmim & Marshak, 1998).

Zircon U-Pb detrital age data for the quartzites of the Caraça and Tamanduá sediments show
two main age populations: 2.85-2.90 and 2.68-2.75 Ga (Machado et al., 1996; Hartmann et al., 2006;
Koglin et al., 2014; Fig. 2.10). The age of these population peaks matches with the Rio das Velhas I
and Mamona events for the genesis of the gneisses and granites of the basement. The youngest
concordant detrital zircon age for a quartzite sample of the Tamanduá Group indicates a maximum age
of sedimentation of 2676 ± 23 Ma (Koglin et al., 2014). The deposition age for the Moeda Formation
is a matter of debate. A maximum depositional age as young as 2580 ± 7 Ma was proposed by
Hartmann et al. (2006) based on U-Pb SHRIMP data from a zircon grain derived from a quartzite at
the top of the Serra da Moeda. This maximum age of sedimentation is slightly younger than the
deposition ages determined by Machado et al. (1996) for three quartzites collected near the city of

48
Ouro Preto, south of the Serra do Gandarela and in western flank of the Serra da Moeda (2606 ± 47
Ma, 2649 ± 16 Ma and 2651 ± 33 Ma, respectively). Recently, U-Pb analyses carried out on zircon
grains from the southernmost tip of the Serra do Gandarela led Koglin et al. (2014) to re-establish a
maximum depositional age for the upper part of the Moeda Formation of 2623 ± 14 Ma. In addition,
Koglin et al. (2014) determined the Lu-Hf isotope composition of detrital zircon grains from
metaconglomerates of the Moeda Formation. The zircon grains show mainly subchondritic initial εHf
and large εHf variations for all the population peaks described by Koglin et al. (2014).

2.6.3 Itabira Group

The sedimentation of the Caraça shallow-water pelites (Batatal Formation) is interdigited with
rocks formed during a major marine transgression recording a period of iron-rich chemical
sedimentation. Together, these sediments are known as the Itabira Group (Dorr, 1969). This event led
to the accumulation of a more than 350 m-thick Lake Superior-type banded-iron deposit called the
Cauê Formation (Fig. 2.8d), and to the subsequent deposition of ca. 600 m of stromatolite-rich
carbonates of the Gandarela Formation (Fig. 2.3; Dorr, 1969; Babinski et al., 1995; Machado et al.,
1996). The Cauê Formation (Dorr, 1969; Klein & Ladeira, 2000) or Cauê Itabirite – as
metamorphosed rocks of the banded iron-formations are known in Brazil – is nowadays the
economically most important unit of the Quadrilátero Ferrífero, hosting world-class hematite-rich iron
ore deposits producing more than 180 Mt per year (Rosière et al., 2008). Three compositionally
different lithofacies of the metamorphosed iron formations exist: quartzitic itabirite, dolomitic itabirite
and the volumetrically minor amphibolitic itabirite. Rosière et al. (2008) reviewed the main features of
the iron ores in the Quadrilátero Ferrífero, presenting the petrogenetic and metallogenic models
proposed for the origin of these rocks. The youngest unit of the Itabira Group is the Gandarela
Formation (Dorr, 1969; Babinski et al., 1993). This formation is predominantly composed of
dolomites, limestones, carbonaceous phyllites and dolomitic iron-rich formation in which stromatolitic
structures are preserved (Souza & Müller, 1984). This formation crops out in the Serra da Moeda, in
the central part of the Serra do Curral and in the Gandarela syncline where it reaches its maximum
thickness (750 m; Dorr, 1969). Its basal contact with the Cauê Formation consists of a transitional
zone (up to ten metres thick) in which the dolostone is associated with the dolomitic itabirite. Babinski
et al. (1995) provided an isochron Pb-Pb age of 2419 ± 19 Ma for a stromatolithic limestone from an
intermediate member of the formation sampled from the Gandarela syncline. This age, due to the
preservation of organic structures and the absence of deformation in the rocks of the Ganderela
Formation, is considered to represent the sedimentation age of the carbonates.

The age of deposition of the Cauê Formation was conservatively bracketed between 2620 and
2420 Ma; i.e. between the maximum age of deposition of the top of the underlying Moeda Formation
(Koglin et al., 2014) and the age of the overlying Gandarela Formation (Babinski et al., 1995).
Recently U-Pb dating of zircon grains from a metavolcanic layer sampled within the Itabira iron
49
formation led Cabral et al. (2012) to propose a considerably earlier (2650 Ma) deposition age for this
unit (Fig. 2.10). The interpretation of this age is controversial, since it contradicts most of the data
produced so far for the underlying Caraça Group. Recently, LA-ICPMS U-Pb dating of detrital zircon
grains from a quartzite lens hosted within the Cauê Formation showed a quasi-unimodal distribution
with a peak at 2795 Ma (Cassino et al., 2014; Martínez Dopico et al., 2017). The youngest concordant
zircon grain in this lens yielded an age of 2453 ± 18 Ma.

2.6.4 Piracicaba Group

The carbonates of the Gandarela Formation are in contact with deep marine sandstones and
pelites of the Piracicaba Group (Dorr et al., 1957; Dorr, 1969). This group is composed of ca. 1300 m-
thick metasediments consisting of quartz-rich sandstones that gradually fines and thins upward to
mudstone and graphitic mudstone. The Piracicaba Group is subdivided into four formations that are
known as the Cercadinho, Fêcho do Funil, Taboões and Barreiro formations (Fig. 2.3). The lowest
sequence (Cercadinho Formation) comprises coarse to fine grained hematite-rich quartzites, quartzites,
silvery sheen phyllites, Fe-rich phyllites as well as dolomites (Simmons, 1968; Dorr, 1969). A
conglomerate containing pebbles derived from the underlying Cauê and Gandarela Formations, from
which the Cercadinho Formation is separated by an erosional unconformity, marks the base of the
Piracicaba Group.

Detrital zircon grains from quartzites and poorly sorted conglomeratic quartzites at the base of
the Cercadinho Formation, in the western part of Serra do Curral, show an Archean contribution
during the initial stages of deposition of the Piracicaba Group (Mendes et al., 2014). The detrital
zircon ages from these samples range between 2750 and 2900 Ma with two distinguishable peaks at
2793 Ma and 2859 Ma (Fig. 2.10), which correspond to the Rio das Velhas II and Rio das Velhas I
events. The deposition age for the Piracicaba Group is still poorly constrained between 2420 Ma (the
age of the Gandarela Formation; Babinski et al., 1995) and ca. 2100 Ma (Babinski et al., 1993). The
lowest age of deposition for the Piracicaba Group (2100 Ma) was obtained by Babinski et al. (1993)
dating deformed dolomitic carbonates from the Fecho do Funil Formation. This Pb-Pb isochron age is
interpreted as the metamorphic age of the rock, thus defining the minimum age of deposition for the
carbonates.

2.6.5 Sabará Group

The Sabará Group is the youngest and thickest unit of the Minas Supergroup (Fig. 2.2),
comprising up to 3.5 km-thick pile of coarsening upwards sequences of metapelites, greywackes, lithic
conglomerates (Fig. 2.8e), and diamictites (Dorr, 1969; Barbosa, 1979; Renger et al., 1995; Reis et al.,
2002). This formation is interpreted as representing a turbiditic, submarine fan deposit formed during
the inversion of the Minas Supergroup passive margin (Alkmim & Martins-Neto, 2012). The unit is

50
well exposed and can be mapped continuously for more than 60 km along the Serra do Curral (Fig.
2.1).

Detrital zircon ages obtained by Machado et al. (1996) for a greywacke from the Serra do
Curral suggest a distinctive age distribution pattern showing a few Proterozoic ages spreading between
2100 and 2500 Ma and a well-defined peak at 2850-2900 Ma (Fig. 2.10). The U-Pb age distribution of
detrital zircon grains in this sample is remarkably different from the age distribution of zircon grains
from a felsic schist belonging to the Sabará Group, collected close to Ouro Preto. The schist, analysed
by Hartmann et al. (2006) by SHRIMP, shows no Proterozoic ages and two Archean peaks at 2720
and 2900 Ma (Fig. 2.10). Although Hartmann et al. (2006) reported no Proterozoic ages, a Rhyacian
maximum age of deposition for the Sabará Group is confirmed by two zircon ID-TIMS U-Pb ages
yielding 2125 ± 4 Ma (Machado et al., 1992) and 2131 ± 5 Ma (Machado et al., 1996). These
Rhyacian ages match well with the ages obtained by Noce et al. (1998) and Seixas et al. (2013) from
the Alto Maranhão suite (ca. 2130 Ma) and associated granitoids of the Mineiro Belt (2100-2200 Ma)
that bounds the Quadrilátero Ferrífero to the south. In addition, the deposition of the Sabará turbidites
marks a major change in the source of Minas sediments. Paleogeographic studies (Dorr, 1969; Renger
et al., 1995; Machado et al., 1996) indicate that during the genesis of the passive margin the Archean
sources were located to the north. On the other hand, a non-cratonic source, located to the south and
southeast, is suggested to account for the occurrence of 2100 Ma old granitoid clasts and zircons in the
conglomerates of the Sabará Group (Alkmim & Martins-Neto, 2012). The turbiditic pelites,
greywackes, lithic conglomerates and diamictites of the Sabará Group were interpreted as representing
syn-orogenic sediments (Dorr, 1969; Renger et al., 1995; Reis et al., 2002) shed from a colliding
magmatic arc and spread over an evolving foreland basin onto the São Francisco craton margin during
the Rhyacian orogeny (Alkmim & Martins-Neto, 2012).

51
52
Figure 2.10 – Frequency histogram for the Minas Supergroup. Histograms and Probability Density plots for the
available U-Pb zircon ages of the Minas Supergroup and Itacolomi Group. Age display software (Sircombe,
2004) was used to build the graphs and evaluate the data. All weighted zircon data was 95 % concordant
(excepting data from Machado et al., 1996) and treated as sigma-1 errors.

2.7 THE ITACOLOMI GROUP

The Itacolomi Group (Dorr, 1969), the youngest unit in the Quadrilátero Ferrífero supracrustal
sequence, comprises an up to 2 km-thick section of medium- to coarse-grained impure
metasandstones, metaconglomerates and minor phyllites separated from the underlying Minas
Supergroup by a low-dipping regional unconformity (Dorr, 1969; Alkmim & Martins-Neto, 2012).
The occurrence of a major erosional event at the base of the unit is manifested by the polymictic
nature of the metaconglomerates, in which pebbles of quartzite, itabirite and granitic rocks were
recognized (Dorr, 1969). The group has a restricted areal distribution, limited to the southernmost
region of the Quadrilátero Ferrífero south of Ouro Preto (Fig. 2.1). These rocks preserve primary
sedimentary structures such as ripple marks and crossbedding and exhibit abrupt lateral changes of
sedimentary facies (Fig. 2.8f). These features suggest that the Itacolomi Group represents
interfingering marine and continental deposits. The only studies of sedimentary provenance that used
U-Pb dating of detrital zircons for the Itacolomi Group are those published by Machado et al. (1996)
and more recently, Hartmann et al. (2006). A summary of zircon U-Pb ages in the Itacolomi Group is
presented in Appendix A5. The two datasets show significant differences that are difficult to reconcile
as the datasets are hardly comparable. Machado et al. (1996) determined by laser ablation the U-Pb
age of many detrital grains, however the age uncertainties are variable and typically large (1σ = 15-
200 Ma) and the degree of concordance is not provided, therefore casting doubt about the reliability of
these ages. On the other hand, Hartmann et al. (2006) used a more accurate technique (SHRIMP) and
provided information about the degree of concordancy as well as on accuracy and precision. However,
the database of Hartmann et al. (2006) is very limited and likely not representative as these authors
analysed only seven concordant zircons. Hartmann et al. (2006) determined a maximum depositional
age of 2143 ± 16 Ma for the Itacolomi Group in its type locality while Machado et al. (1996)
determined depositional ages that are ca. 100 Ma younger (i.e. 2039 ± 42 and 2059 ± 58 Ma; Fig.
2.10). Another difference between the two datasets is that Machado et al. (1996) found many grains
yielding 207Pb/206Pb ages in the range 2200-2500 Ma (Fig. 2.10) which were not found by Hartmann et
al. (2006). Finally, in contrast to what is observed for the underlying Sabará Group, only a subset of
zircon grains from the sandstone and conglomerates of the Itacolomi Group yielded Archean ages. The
large percentage of Rhyacian detrital zircons (i.e. between 2300 and 2050 Ma) in the conglomerates of
the Itacolomi basins support a non-cratonic source (Machado et al., 1996; Alkmim & Marshak, 1998)
suggesting that most of the sediments were derived from terrains generated during the Rhyacian
orogeny. The source was probably located to the south and southeast, with the Alto Maranhão Suite
(ca. 2130 Ma; Seixas et al., 2013), Juiz de Fora (2041 ± 7 and 2137 ± 19 Ma; Noce et al., 2007) and
53
Mantiqueira complexes (2119 ± 16 to 2084 ± 13 Ma; Noce et al., 2007) being the most suitable
candidates. The Itacolomi Group is interpreted as an intermontane molasse deposit developed along
the margin of the Archean nucleus of the São Francisco craton during the collapse phase of the
Rhyacian orogeny (Marshak et al., 1992; Alkmim & Marshak, 1998; Alkmim & Martins-Neto, 2012).

2.8 MAFIC AND INTERMEDIATE DIKES, BOUDINS AND LENSES

Mafic and intermediate amphibolites in the Quadrilátero Ferrífero crop out as metre-scale
blocks, boudins and lenses hosted within the gneisses as well as dikes crosscutting granites and
supracrustal rocks. Carneiro (1992) and Carneiro et al. (1998) defined four groups of mafic-
intermediate amphibolites in the Bonfim complex, based on field relationships, trending directions as
well as on textural and compositional data. Two swarms of dikes crosscut the basement and give
Neoarchean Sm-Nd model ages (Carneiro et al., 1998) while the other two groups are younger as
suggested by the fact that they crosscut Paleoproterozoic metasediments and exhibit well-preserved
igneous textures. In the Belo Horizonte complex, unmetamorphosed tholeiitic dikes crosscutting the
basement and the supracrustal sequences yielded K-Ar feldspar ages of ca. 1000 Ma (Chaves, 1997).
These dikes, related to the separation between the São Francisco and the Congo cratonic blocks in the
late Mesoproterozic (De Min et al., 2005), will not be described any further as this late evolution of
the Quadrilátero is beyond the scope of this review.

The oldest population of amphibolites in the Quadrilátero is composed of mafic and

intermediate rocks that intruded as dikes and were later strongly rotated and disrupted, appearing now
as boudins and lenses within the enclosing gneisses. These boudins or lenses (hereafter ―rafts‖) are
fine- to medium-grained, with textures varying from nematoblastic to granoblastic and geochemistry
of within-plate tholeiitic basalts. Recently, one of these rafts from the Bação complex was dated by
zircon U-Pb geochronology by Lana et al. (2013) yielding a well-defined Concordia age of 2778 ± 8
Ma. Titanite U-Pb ages from the same mafic raft yield Paleoproterozoic (ca. 2100 Ma) ages suggesting
resetting of the basement in the southern Quadrilátero during the Rhyacian (Aguilar et al., 2017).
Farina et al. (2015a) dated a dike crosscutting the ca. 2770 Ma old Samambaia tonalite in the Bonfim
207
dome. Zircon grains from this rock gave a weighted mean Pb/206Pb age of 2719 ± 14 Ma, with this
age matching well with the age of the oldest mafic dikes in the Uauá block in the northern São
Francisco craton (2726.2 ± 3.2 Ma; Oliveira et al., 2013). These geochronological data seem to
confirm the occurrence of two Neoarchean systems of dikes. The first swarm that emplaced at ca.
2780 Ma was successively metamorphosed and deformed together with the hosting gneisses. The
second population of dikes crosscuts granitoids formed during the 2760-2680 Ma Mamona event.
These dike systems are useful relative time markers allowing bracketing of the metamorphism and
deformation between 2780 Ma and 2720 Ma. The tectonic significance of these Archean dikes is
currently unknown as no systematic geochemical and isotopic studies were attempted.

54
2.9 MAIN STRUCTURES AND STRUCTURAL EVOLUTION

The Quadrilátero Ferrífero experienced a polyphase tectonic history that produced complex
regional patterns of rock deformation (e.g. Dorr, 1969; Drake & Morgan, 1980; Endo, 1997; Alkmim
& Marshak, 1998; Chemale et al., 1994; Chauvet et al., 1994). This complexity, together with the lack
of absolute ages dating the tectonic structures, gave rise to different and in many cases conflicting
interpretations for the deformation history of the Quadrilátero. Three main sets of structures
characterize the Quadrilátero Ferrífero; from the oldest to the youngest, these are:

- Northwest-verging and north-northeast-trending regional-scale folds and thrusts

generating large asymmetric synclines (the Moeda, Dom Bosco, Pitangui-Peti, Mateus
Leme and Souzas folds) and the steeply dipping to overturned Serra do Curral homocline
(e.g. Pomerene, 1964; Dorr, 1969; Pires, 1979; Marshak et al., 1992; Alkmim & Marshak,
1998; Fig. 2.1);
- Extensional structures related to the formation of a typical dome-and-keel province, in
which troughs of deformed and metamorphosed Paleoproterozoic supracrustal rocks
surround domes of Archean basement (e.g. Belo Horizonte, Caeté, Bação complexes;
Chemale et al., 1991, 1994; Alkmim & Marshak, 1998);
- West-verging thrust folds reactivated and overprinted pre-existent structures in the region
east of a north-trending line that follows the west edge of the Moeda syncline (Fig. 2.1;
Chemale et al., 1994).

According to Alkmim & Marshak (1998), these structures correspond to three kinematic
phases that affect the rocks of both the Rio das Velhas and Minas Supergroups. The maximum age of
formation of the oldest structures is constrained by the deposition age of the top of the Minas
Supergroup: the Sabará Group. This 3.5 km-thick flysch sequence that has a maximum deposition age
of 2130 Ma (Machado et al., 1996) is involved in northwest-verging folding. This age constraint led
Alkmim & Marshak (1998) to interpret the northwest-verging folds and thrusts as a fold and thrust belt
formed shortly after the deposition of the Sabará Group in the foreland of a Rhyacian collisional
orogeny. Regional lower- to high-grade metamorphism affecting the Minas Supracrustal rocks is
associated with the northwest-verging folds. The metamorphic grade increases eastward from the
Bonfim to the Bação complexes, reaching granulite-facies in the easternmost part of Quadrilátero
Ferrífero (Herz, 1978; Gomes, 1986; Gomes & Muller, 1987). This event, however, did not generate a
strong foliation.

The second structures relate to the formation of the dome-and-keel geometry of the
Quadrilátero. The contacts between granitic-gneissic complexes and the supracrustal sequences are
tectonic, marked by thrusts and/or normal faults, and by ductile shear zones showing variable sense of
displacement (Hippertt et al., 1992; Machado et al., 1996; Alkmim & Marshak, 1998). For instance,

55
the enveloping shear zones of the Bonfim and Belo Horizonte domes show normal sense kinematic
indicators (Hippertt et al., 1992), while the shear sense observed in the contact zone of the Bação and
Caeté domes varies between reverse, reverse-oblique and strikeslip (Marshak & Alkmim, 1989;
Hippertt, 1994; Endo, 1997). Kinematic indicators in supracrustal rocks clearly indicate supracrustal-
side-down displacement (Marshak et al., 1997) supporting an extensional event (Hippertt et al., 1992).
Metamorphic aureoles in the supracrustal rocks at the contact with the domes have different thickness
(Pomerene, 1964; Herz, 1978; Carneiro & Teixeira, 1992; Jordt-Evangelista et al., 1992; Marshak et
al., 1992, 1997). The dome-border shear zones and their related metamorphic aureoles overprint the
foliation of Archean gneisses as well as the early regional lower greenschist-facies metamorphism and
the foliation, folds and faults in the supracrustal sequence. This evidence indicates that the doming
event post-dated the northwest-verging thrusts and associated syncline folds (Marshak et al., 1992;
Alkmim & Marshak, 1998). Syn-kinematic garnets formed in a dome-border shear zone in rock of the
Minas Supergroup gave a Sm-Nd age of 2095 ± 65 Ma (Marshak et al., 1997). This age, that is
consistent with titanite and monazite U-Pb metamorphic ages obtained in the basement (Machado et
al., 1992; Schrank & Machado, 1996a, b; Aguilar et al., 2017), suggests that dome emplacement
occurred during the extensional collapse of the Rhyacian orogen.

Finally, the collisional and extensional structures described were overprinted and reactivated
by a series of west-verging thrust faults and associated structures, attributed to the Neoproterozoic to
early Ordovician Brasiliano event (650-480 Ma; e.g. Endo and Fonseca, 1992; Chemale et al., 1994;
Alkmim & Marshak, 1998).

2.10 DISCUSSION: MODELS AND OPEN PROBLEMS

This section is subdivided into five sub-sections; in the first four we discuss the evolution of
the Quadrilátero Ferrífero from the genesis of the continental crust in the Paleoarchean to the
Paleoproterozoic Minas orogeny. In the last sub-section, we present a series of unresolved questions
related to the genesis and evolution of the Quadrilátero Ferrífero.

2.10.1 Building up the continental crust: the Paleo- to Mesoarchean rock record

The oldest preserved rocks in the Quadrilátero Ferrífero are banded gneisses from the Santa
Bárbara complex that formed at the Paleo-Mesoarchean boundary (i.e. 3200 Ma; Lana et al., 2013).
Unfortunately, very little study has been conducted on these rocks so the Paleo-Mesoarchean rock
record of the Quadrilátero Ferrífero remains enigmatic. However, the presence of these gneisses,
despite being extremely rare in the rock record, provides various lines of evidence to suggest that a
significant volume of continental crust formed during the Paleoarchean and in the late Mesoarchean
(Lana et al., 2013). Firstly, zircon ages yielding 3000-3400 Ma ages occur as inherited components in
a limited number of gneisses dated by Lana et al. (2013) and represent a significant subset in the
detrital spectra of the greenstone belt succession and the Minas Supergroup (Fig. 2.9 and 2.10). In

56
particular, minor peaks at ca. 3200 Ma occur in the Tamanduá, Caraça, Itabira and Piracicaba Groups
(Fig. 2.10) as well as in the greenstone belt in both the Nova Lima (Fig. 2.9) and Maquiné Groups
(Moreira et al., 2016). Secondly, Sm-Nd model ages for Neoarchean gneisses and granites in the
Bonfim complex gave values up to 3300 Ma (Teixeira et al., 1996). Finally, most of the detrital
zircons from the Moeda, Cauê and Piracaba formations have subchondritic initial εHf values resulting
in calculated depleted mantle model ages varying from 3000 to 3800 Ma (Koglin et al., 2014;
Martínez Dopico et al., 2017). These model ages are in good agreement with those calculated by
Albert et al. (2016) for magmatic zircons in gneisses and granitoids from the Bonfim complex. Taken
together these geochronological and isotopic data advocate for the existence in the Quadrilátero
Ferrífero of a large segment of Paleoarchean continental crust that was reworked during successive
pulses of magmatic activity and eroded during tectonic denudation.

Recently, Farina et al. (2015a) compared the composition of gneisses and granitoids from the
Quadrilátero Ferrífero with TTGs from other cratons and experimental melts produced by fluid-absent
melting of tonalites. Medium-K rocks are significantly more silica- and K2O-rich and less Na2O- and
Al2O3-rich than typical TTGs (Fig. 2.5) and show intermediate compositions between TTGs and
experimental melt obtained through fluid-absent partial melting of TTG sources such those used as
starting material by Watkins et al. (2007). Based on whole-rock chemical arguments, Farina et al.
(2015a) suggested that medium-K banded gneisses and granitoids formed by mixing between a TTG-
like melt produced by partial melting of basaltic oceanic crust and a melt derived by reworking of the
continental crust. The occurrence of a recycled crustal component in the genesis of the medium-K
rocks formed during the Rio das Velhas I and II magmatic events is supported by the negative εHf
composition displayed by both detrital zircons in the Moeda and Piracicaba formations (Koglin et al.,
2014; Martínez Dopico et al., 2017) and magmatic zircon crystals in gneisses and granitoids (Albert et
al., 2016). A subset of detrital zircons in the Moeda Formation (< 10 % of the grains) and magmatic
zircons from a few gneisses sampled in the Bação complex show superchondritic Hf isotopic
compositions suggesting that juvenile crust was also formed during the Meso- and Neoarchean Rio das
Velhas I and II magmatic events (Koglin et al., 2014; Albert et al., 2016).

2.10.2 The record of plate tectonics: the Meso-Neoarchean

Two significant changes occurred in the Quadrilátero Ferrífero during the early Neoarchean,
between ca. 2800 and 2700 Ma. Firstly, high-K granites similar to post-Archean high-silica I-type
granites were produced, largely replacing medium-K granitoids and gneisses; the latter showing
chemical affinity similar to TTGs and representing the volumetrically dominant rocks produced during
the Rio das Velhas I and II events (ca. 2920-2760 Ma). Secondly, mature sedimentary sequences
emerged and were deposited in the Rio das Velhas greenstone belt forming the 2000 m-thick
association of conglomerates and sandstones of the Maquiné Group. We argue that a model involving
subduction of oceanic crust and subsequent continental collision between two continental blocks
57
account for these two major changes, explaining coherently most of the Neoarchean evolution of the
Quadrilátero Ferrífero. The model that is illustrated in Fig. 2.11 has been proposed to explain the late-
Archean geodynamic evolution of many other terranes worldwide (e.g. Percival et al., 2006; Laurent
et al., 2014).

In the Quadrilátero, medium-K rocks were produced during the Rio das Velhas I and II events
(2920-2850 and 2800-2760 Ma). These rocks have chemical composition suggesting mixing between
magmas derived by partial melting of oceanic crust and magmas derived by crustal reworking. These
compositions can be generated by significant interaction between melts derived by melting a
subducting oceanic slab and an upper plate formed of a Meso-Paleoarchean continental nucleus. In
this scenario, clastic sediments from the eroded upper-plate were deposited in a back arc basin (i.e the
Maquiné Group) and volcanic rocks with TTG-affinity erupted forming dacitic flows in the greenstone
belt.

The granitoids formed during the Rio das Velhas II event were deformed and metamorphosed
between 2780 and 2730 Ma as indicated by the age of the earliest granites emplaced in the Bação,
Bonfim and Belo Horizonte complexes (Machado et al., 1992; Romano et al., 2013; Farina et al.,
2015a). During this ca. 50 Ma of metamorphism, the medium-K granitoids were deformed and
transformed into banded gneisses and leucogranitic sheets and mafic/intermediate dikes were
emplaced and rotated into parallelism with the banding in the gneisses. The pressure and temperature
conditions attained during the metamorphic event are not constrained, but the occurrence of
migmatites formed by fluid-present melting of the gneisses suggests that the basement locally reached
upper amphibolite-facies conditions (Farina et al., 2015b). We suggest that this high-grade
metamorphic event records the collision between two continental blocks. During crustal thickening,
parts of the metasedimentary pile were stacked, deformed and buried. Then, thermal relaxation and
extension following lithospheric delamination triggered the upwelling of the asthenosphere, heating up
the continental crust and inducing partial melting. High-K granites and crosscutting mantle-derived
dikes formed in this syn- to late-collisional geodynamic environment. Based on whole-rock chemical
arguments, Farina et al. (2015a) suggested that high-K granites in the Quadrilátero were produced by
low degree of melting of metasedimentary sources deposited during the Rio das Velhas II event. In
this scenario, the transition between medium-K and high-K granitoids reflects a change in the sources
undergoing melting.

It is worth noting that during the Neoarchean the northern (the Belo Horizonte complex) and
southern (the Bação and Bonfim complexes) portions of the Quadrilátero Ferrífero experienced
different magmatic evolutions. In the southern portion, medium-K granitoids emplaced for ca. 20 Ma
(2790-2770 Ma) after the metamorphic peak and were then replaced by high-K monzogranite and
syenogranite with composition of late Archean biotite and two-mica granites (Farina et al., 2015a, b).
This rather sharp compositional change was not recorded in the Belo Horizonte complex where the ca.
58
2750-2720 Ma massive granitic batholiths (Pequi and Florestal; Fig. 2.1) have TTG affinity. In the
Belo Horizonte complex, minor bodies of high-K granites (e.g. the Santa Luzia granite) were only
emplaced at ca. 2700 Ma (Noce et al., 1997).

The emplacement of high-K granites during the Mamona magmatic event marks the final
cratonization of the Quadrilátero Ferrífero, which remained stable from ca. 2700 Ma onwards. This
probably occurred because the emplacement of high-K magmas at shallow level in the crust
concentrated heat-producing elements (e.g. K, Th, U) in the upper crust, leaving the middle crust
thermally stable (e.g. Sandiford & McLaren, 2002). In addition, partial melting of the continental crust
left the lower crust dry (refractory) and therefore more resistant to future episodes of partial melting
(Romano et al., 2013).

2.10.3 Evolution of the Minas Basin

The rocks of the Minas Supergroup represents a passive-margin to syn-orogenic sedimentary

package tracking the operation of a Wilson cycle between ca. 2.6 and 2.0 Ga (Schorscher, 1992;
Alkmim & Marshak, 1998; Canuto, 2010). The development of the Minas passive margin took place
in the time interval between approximately 2600 and 2400 Ma (i.e. between the deposition of the
Moeda and Gandarela formations; Babinski et al., 1995; Koglin et al., 2014) along the borders of the
São Francisco and Congo cratons. The clastic metasedimentary rocks, which form the lower part of
the Minas Supergroup (the Tamanduá and Caraça groups), were deposited during the early subsidence
of the precursor Archean passive margin basin. The intermediate Itabira Group that is composed of the
Cauê Banded Iron Formation, hosting the valuable iron ore deposits of the Quadrilátero Ferrífero, and
of limestones of the Gandarela Formations, mark the thermally subsiding stage of the Minas Basin
(Alkmim & Martins-Neto, 2012). Finally, clastic immature and flysch-like metasedimentary rocks of
the uppermost Sabará Group, marks the inversion of the basin representing syn-orogenic sediments
shed from a colliding magmatic arc (Dorr, 1969; Barbosa, 1979; Renger et al., 1995). The Sabará
Group contains detrital zircon grains as young as 2125 ± 4 Ma (Machado et al., 1992); i.e. 300 Ma
younger than the age of the underlying units of the Minas Supergroup. This marks a significant change
in both the depositional setting and sediment source compared to the older Minas units that were
mainly derived from Archean material. The Sabará turbidites are a submarine fan deposit interpreted
as having formed either as a syn-orogenic deposit scraped off from an active magmatic arc (Hartmann
et al., 2006) or as a foreland basin (Machado et al., 1996).

2.10.4 The Minas accretionary orogeny

The regional setting of the Quadrilátero Ferrífero is dominated by two set of structures: i)
northwest-verging and north-northeast-trending regional-scale folds and thrusts with associated low-
to medium-grade metamorphism affecting the supracrustal rocks, and ii) gneiss-granitic domes
surrounded by elongate keels of polydeformed supracrustal rocks (dome-and-keel structures). These

59
structures formed in the Paleoproterozoic as indicated by the depositional age of the Sabará Group
(2125 Ma; Machado et al., 1996), which is involved in the folding and thrusting, as well as by the age
of the dome-related metamorphic contact aureole (ca. 2095 Ma; Marshak et al., 1997). The fold and
thrust belt formed in the Paleoproterozoic during the collision between the nuclei of the present-day
São Francisco and Congo cratons (Alkmim & Martins-Neto, 2012). In the southern portion of the São
Francisco craton, the continental collision between these two plates generated the Mineiro Belt and the
adjoining Mantiqueira and Juiz de Fora belts (Teixeira et al., 2015). The Mineiro Belt, bordering the
Quadrilátero Ferrífero to the south, is a km-wide corridor of polydeformed Archean gneisses and
greenstone belt remnants intruded by 2350-2000 Ma granitoids (Seixas et al., 2012). This belt was
generated through successive accretion of oceanic and continental arcs that were active during the
Paleoproterozoic and then final collision with the São Francisco craton, representing the foreland of
the belt (Teixeira et al., 1996; Alkmim & Marshak, 1998; Ávila et al., 2010). This event is recorded in
the basement by monazite and titanite crystals yielding 2100-1950 Ma U-Pb metamorphic ages
(Machado et al., 1992; Schrank & Machado, 1996a, b; Aguilar et al., 2017). The tectonomagmatic
event affecting the southern São Francisco craton, which is usually referred to as the Transamazonian
orogeny such as for the Amazonian Craton (e.g. Machado et al., 1996), has been named by Teixeira et
al. (2015) the Minas accretionary orogeny.

In the Orosirian, after the formation of the fold and thrust belt the Quadrilátero underwent
orogenic collapse, resulting in the deposition of alluvial sandstones, conglomerates and pelites of the
Itacolomi Group and development of the dome-and-keel structure (Alkmim & Marshak, 1998). The
large percentage of Rhyacian detrital zircons (i.e. between 2300 and 2050 Ma) in the conglomerates of
the Itacolomi basins support a non-cratonic source (Machado et al., 1996; Alkmim & Marshak, 1998)
suggesting that most of the sediments were derived from terranes generated during the Minas
accretionary orogeny.

On a regional scale, the Minas accretionary orogeny correlates with the tectono-magmatic
events recognized in the Itabuna-Salvador-Curaçá belt in the northern portion of the São Francisco
craton (Teixeira et al., 2015). These events seem to reflect the assembly of a supercontinent during the
Orosirian period, the Atlantica supercontinent (Rogers, 1996), followed by Columbia/Nuna (Rogers &
Santosh, 2004; Zhao et al., 2004), whose reconstructions, though not fully accomplished, have
progressed significantly in the last few years.

60
Figure 2.11 – Sketch of the geodynamic evolution of the Quadrilátero Ferrífero during the Rio das Velhas I, Rio
das Velhas II and Mamona periods. In the 2920-2850 Ma cartoon, we tentatively propose that the continental
crust formed by multiple accretion of island-arcs. During the Rio das Velhas II period, the subduction of an
oceanic crust under a continental block led to the formation of medium-K granitoids by mixing between two
components: melts derived by partial melting of the mafic oceanic crust and melts derived by recycling of older
continental crust. During this event, volcanic rocks erupted above the ultramafic-mafic sequence of the Nova
Lima Group and mantle-derived magmas intruded the basement. Finally, during the Mamona event, two
continental blocks collided. Slivers of metasediments were buried and started melting producing high-K granites
and clastic sediments were deposited forming the Maquiné Group. Based on an illustration from Laurent (2012).

61
 2.10.5 Questions for future research

Recently, new geochronological data helped improve our understanding of the ca. 1.5 Ga of
evolution of the Quadrilátero Ferrífero (Lana et al., 2013; Romano et al., 2013; Mendes et al., 2014;
Farina et al., 2015a). However, a number of questions remain open for further research, as outlined
below:

- What are the metamorphic conditions attained by the basement during the Neoarchean
metamorphic event? Did the basement undergo extensive partial melting between 2780
and 2730 Ma?
- During the Mamona event, TTG-like granitoids emplaced in the Belo Horizonte complex
while high-K granites were generated in the Bação and Bonfim complexes. Why were
magmas with different chemical affinity formed at the same time in the northern and
southern portions of the Quadrilátero Ferrífero? Moreover, what is the chemical
composition of gneisses and granitoids in the Santa Bárbara and Caeté complexes?
- Metaluminous Mg-Fe- and K-rich monzodiorite and granodiorites forming the sanukitoid
series typify the late Archean evolution of many cratons (e.g. Laurent et al., 2014). These
rocks derive from hybridization between mantle peridotite and a component rich in
incompatible elements (Laurent et al., 2014). Magmatic rocks with sanukitoid affinity
were not described in the Quadrilátero Ferrífero. How does the proposed model of
Neoarchean subduction-collision for the Quadrilátero account for the lack of sanukitoids?
- What is the age of the different mafic-intermediate dike swarms crosscutting the basement
and the supracrustal sequences? In particular, what is the chemical composition of the two
populations of Neoarchean dikes and what do they tell us about the geodynamic evolution
of the Quadrilátero Ferrífero?
- What is the age of the mafic and ultramafic rocks in the Rio das Velhas Supergroup? What
is the age of the greenschist facies metamorphism of the Rio das Velhas Supergroup?
What is the tectonic significance of the Maquiné basin?
- Does the deposition of the Minas Supergroup track an entire Wilson cycle or did these
rocks form during the opening and closure of an intracratonic rift basin?
- What is the age and tectonic significance of the unconformity bounded Piracicaba Group,
whose depositional age is still poorly constrained?
- Why are the U-Pb depositional ages and age distributions obtained by Machado et al.
(1996) and Hartmann et al. (2006) for the rocks of the Sabará Group different (Fig. 2.11)?
Does this discrepancy reflect the fact that the samples were collected from different
localities and/or different stratigraphic positions? Ultimately, what is the depositional age
and age distribution of the turbidites of the Sabará Group?

62
 - Monazite and titanite grains from the basement gave Rhyacian U-Pb ages, while no
metamorphic zircon crystals/overgrowths yielded non-Archean ages. What are the
metamorphic conditions attained by the basement during the Minas orogeny? What is the
geographic extent of the Rhyacian metamorphic overprint? Why was no magmatism
produced in the Quadrilátero Ferrífero during the continental collision between the São
Francisco and Congo protocratons?

This paper was supported by CNPq (grant numbers: 401334/2012-0; 402852/2012-5 and 302633/2011-1) and FAPEMIG (grant numbers
APQ03943 and RDP 0067-10) grants. We thank two anonymous reviewers for the constructive comments that helped improving the quality
of this work. Prof. V. Janasi is acknowledged for the editorial guidance offered.

63
64
 CHAPTER 3
THE NEOARCHEAN TRANSITION BETWEEN MEDIUM- AND HIGH-
K GRANITOIDS: CLUES FROM THE SOUTHERN SÃO FRANCISCO
CRATON (BRAZIL)
3

ABSTRACT

Field observations, geochemical data and LA-ICP-MS U–Pb ages for orthogneisses and
granitoids from the Southern São Francisco Craton (SE, Brazil) indicate a major change in the
composition of the continental crust during the Neoarchean. The crystalline basement comprises fine-
grained banded gneisses that are intruded by leucogranitic sheets and dikes, by large granitoid
batholiths and by small granitic bodies closely associated with the host banded gneisses. Granites and
gneisses have high silica contents of 70–76 wt.% and, based on their major and trace element
composition are sub-divided into medium- and high-K granitoids. The distinctive geochemistry
between these rock types reflects melting of different sources. Medium-K rocks have chemical
affinities similar to rocks of the tonalite–trondhjemite–granodiorite series (TTGs): high Na2O, Al2O3
and Sr contents, low heavy rare-earth contents. However, their overall chemical features suggest
mixing between two end-members: a magma generated by partial melting of metamafic rocks and a
component derived by recycling TTG rocks of older continental crust. The composition of high-K
granites is similar to that of late-Archean biotite and two-mica granites. These rocks cannot be
generated by melting of TTGs, but rather require melting of Archean metasediments capable of
producing large volumes of K2O-rich magmas. Medium-K rocks were volumetrically dominant during
two earlier periods of magmatism and associated greenstone belt deposition (2920–2850 and 2800–
2760 Ma), while high-K rocks mark the final cratonization event and consolidation of the granitic
crust of the craton between 2750 and 2720 Ma. The relatively sharp transition between the two rock
types reflects the onset of basin deposition during the Neoarchean followed by melting of the neo-
formed basins at depth.

F. Farina*, C. Albert, C. Lana

Departamento de Geologia, Universidade Federal de Ouro Preto, Morro do Cruzeiro, 35400-000, Ouro Preto, MG, Brazil
* Corresponding author: Federico Farina. E-mail: [email protected]
Article published in Precambrian Research 266 (2015), 375-394
Keywords: TTG series; High-K granites; Continental crust; Neoarchean

65
3.1 INTRODUCTION

Fundamental changes concerning the composition of the continental crust, the stratigraphy of
sedimentary successions, the style of crustal deformation and the metamorphic conditions attained
within the continental crust occurred at the end of the Archean aeon (e.g. Condie, 1989; Taylor &
McLennan, 1985; Cagnard et al., 2011; Keller & Schoene, 2012). During this period, the Earth‘s
continental crust experienced a pronounced compositional change, with the transition from Archean
sodic granitoids forming the so-called tonalite–trondhjemite–granodiorite (hereafter TTG) series (Jahn
et al., 1981) to medium and high potassium granitoid composititions which define the granite-
granodiorite series dominating the late- to post-Archean granitoid record (Martin et al., 2005). Since
the 1970s, when they were first described, TTGs have been the subject of many studies (Moyen &
Martin, 2012 and references therein) because of the potential they hold for providing information on
the geodynamic environments operating on the Earth during the Archean and the links between these
and the production of the first continental crust. TTGs, which represent the main component of the so-
called Paleo- and Mesorchaean ―grey gneiss‖ complexes (Moyen, 2011), are silica-rich (SiO2 is
commonly greater than 70 wt.%) and have higher Na2O contents and correlated lower K2O/Na2O and
CaO/Na2O ratios than post-Archean granitoids (Bickle et al., 1989; Frost et al., 1998; Keller &
Schoene, 2012; Laurent et al., 2014). Moreover, TTGs exhibit a characteristic trace element signature,
with high ratios of light to heavy rare-earth elements coupled with the lack of significant Eu and Sr
anomalies. These chemical features suggest that TTGs formed through partial melting of metaigneous
mafic rocks (Martin et al., 2014), at sufficient pressure to stabilize a significant fraction of garnet in
the residuum (e.g. Foley et al., 2002; Rapp et al., 2003). Much of the current debate on the
geodynamic scenarios relevant to the formation of TTG magmas focuses on how the necessary high
pressures are achieved. Some studies argue for gravity driven foundering of dense mafic crust into hot
mantle (e.g. Bédard et al., 2013), others that the geochemical signature is a consequence of partial
melting of subducted oceanic crust within early subduction zones that were fundamentally different to
those operating on the modern Earth (e.g. Turner et al., 2014; Laurie & Stevens, 2012).

The processes that controlled the late-Archean transition between TTGs and medium- and
high-K granitoid rocks (e.g. Champion & Sheraton, 1997; Moyen et al., 2003; Frost et al., 2006;
Mikkola et al., 2011; Almeida et al., 2013) are also the subject of considerable debate. Recently,
Laurent et al. (2014) reviewed the petrogenesis and global-scale evolution of late Archean granitoids.
These authors pointed out that although the chronology of granitoid emplacement during the Archean
varies from craton to craton, it may be depicted as a two-stage evolution. A first and generally long
period (0.2–0.5 Ga) of Archean magmatic evolution characterized by TTG emplacement is followed
by a shorter period (0.02–0.15 Ga) during which TTGs are associated with granitoids that are either
generated by the interaction between the mantle and a component enriched in incompatible element
(sanukitoids; e.g. Martin et al., 2009) or by partial melting of an older continental crust (e.g. Feng &

66
Kerrich, 1992). On a planetary scale, the relatively short period in which TTGs and high-K granites
form contemporaneously occurs between 2.5 and 3.0 Ga and essentially marks the end of TTG
production as well as the final stabilization of the cratonic lithosphere.

There are still many questions about the transition between TTGs and high-K granites that
need to be answered. Firstly, not all the cratons are well documented, so it not clear if the typical
granitoids associations described by Laurent et al. (2014) occurs everywhere. Secondly,
geochronological data are required to establish the chronology of granitoid emplacement in different
cratons. In particular, these data are crucial to answer the following question: how sharp is the
transition between TTGs and high-K granites? Finally, understanding the origin of the different types
of granitoids during the late Archean holds the key for a better comprehension of the crucial
geodynamic changes affecting the Earth by the end of the Archean (Condie & O‘Neill, 2010).

In this paper, we report field observations, major and trace element compositions and U–Pb
zircon ages for orthogneisses and granites from the Southern São Francisco craton (Brazil). Although
some authors proposed based on mineralogical and textural observations, TTG-affinity for the
orthogneisess and calc-alkaline potassic signature for the granites (e.g. Noce et al., 1997; Lana et al.,
2013), the chemical composition of these rocks has never been investigated in detail. With the dataset
presented in this study, we are able to outline the main chemical features of the basement of a large
segment of the Southern São Francisco craton as well as determine the chronology of the transition
from TTG-like to potassic granitoids. These data allow constraints to be placed on the processes that
led to the formation of this specific segment of Archean continental crust and consequently provide a
better understanding of the evolution of the continental crust in the late Archean.

3.2 REGIONAL GEOLOGY

3.2.1 The Southern São Francisco craton

The São Francisco craton in eastern Brazil, one of the major shield areas forming the
backbone of the South American platform, is subdivided into several Archean to Paleoproterozoic
blocks bounded by major ca. 2.0 Ga old sutures zones (Teixeira & Figueiredo, 1991; Barbosa &
Sabaté, 2004; Oliveira et al., 2010b). The southernmost sector of the craton, known as the Southern
São Francisco craton (Fig. 3.1), is bounded to the east and to the west by Neoproterozoic fold belts
(e.g. the Araçuaí belt; Pedrosa-Soares et al., 2001) and to the south by a km-wide corridor of
polydeformed Archean gneisses and greenstone belt remnants intruded by 2350–2000 Ma granitoids
(Teixeira et al., 1996; Alkmim & Marshak, 1998; Seixas et al., 2012). This corridor, known as the
Mineiro Belt, formed during the Paleoproterozoic through successive accretion of oceanic and
continental arcs and final collision with the São Francisco craton (Teixeira et al., 2015).

The Southern São Francisco craton comprises granitoid-gneiss complexes forming domes that
are tens of km across (e.g. the Bação complex) surrounded by elongated troughs containing
67
polydeformed, low-grade supracrustal sequences of Archean and Paleoproterozoic age (Fig. 3.1; Dorr,
1969). This study is focussed on three granitoid-gneiss complexes, namely the Bação, Bonfim and
Belo Horizonte complexes, cropping out in the easternmost part of the craton (Fig. 3.1). The studied
area hosts an important mining district (the Quadrilátero Ferrífero; Dorr, 1969) known for hosting
world-class orogenic gold deposits (Lobato et al., 2001) and for the great economic relevance of its
reserves of iron (Spier et al., 2007).

Figure 3.1 – Geological map of the study area showing the distribution of the basement, the Rio das Velhas
Supergroup, the Minas Supergroup and the Itacolomi Group. In the basement, the basal metamorphic complexes
shown in the map are Divinópolis, Belo Horizonte, Bonfim, Bação and Santa Bárbara. The squares (numbered
from 1 to 40) indicate the sample localities. Batholith and pluton abbreviations: F – Florestal; M – Mamona; P –
Pequi; Sa – Samambaia; SN – Souza Noschese. Inset: tectonic sketch of the São Francisco craton showing the
location of the bordering Brasiliano orogenic belts. Modified from Alkmim & Marshak (1998).

3.2.2 Gneiss-granitoid complexes and supracrustal sequences

In the Southern São Francisco craton, the Archean basement is composed of banded
orthogneisses intruded by medium- to coarse-grained granites as well as by leucogranites and aplitic-
pegmatitic dikes (Lana et al., 2013). The gneisses range from banded metatonalite to metagranodiorite
(Lana et al., 2013) and are characterized by a penetrative amphibolite-facies foliation and locally by
the development of stromatic migmatites (Alkmim & Marshak, 1998). The gneisses are typically
intruded by multiple, metre- to cm-scale leucogranitic sheets sub-parallel to the gneissosity as well as
68
by crosscutting younger felsic and/or pegmatitic dikes (Lana et al., 2013). Granitoid plutons and
batholiths comprise ca. 30% of the bulk of the exposed Archean crust (Romano et al., 2013). The
older granitoids are marked by a solid-state fabric with regionally consistent trends, indicating a pre-
to syn-tectonic timing of emplacement, while the younger granitic rocks are weakly foliated or
undeformed. The granitic rocks range in composition from hornblende-bearing tonalite to pinkish K-
feldspar-rich syenogranite (Romano et al., 2013).

The supracrustal succession surrounding the granitoid-gneiss metamorphic complexes

includes Archean metavolcanic and metasedimentary rocks of the Rio das Velhas Supergroup as well
as Paleoproterozoic metasedimentary sequences forming the Minas Supergroup and the Itacolomi
Group (Dorr, 1969). The rocks of the Rio das Velhas Supergroup form a typical Archean greenstone
belt sequence characterized by the association between mafic and ultramafic rocks (komatiite–basalt),
evolved volcanic, volcaniclastic rocks and immature clastic sediments (Baltazar & Zucchetti, 2007).
Dorr (1969) proposed subdividing the rocks of the greenstone belt into the Nova Lima and Maquiné
Groups, the former occurring at the base of the sequence and hosting the major gold deposits of the
Quadrilátero Ferrífero, the latter representing a 2000 m-thick clastic association of conglomerates and
sandstones. The Paleoproterozoic Minas Supergroup (Dorr, 1969) is a ca. 6 km-thick package of
clastic and chemical rocks (Alkmim & Marshak, 1998) lying unconformably on the Archean
greenstone belt. These rocks represent a passive-margin to syn-orogenic sedimentary package tracking
the operation of a Wilson cycle between ca. 2.6 and 2.0 Ga (Alkmim & Marshak, 1998). Following
Alkmim & Martins-Neto (2012), the Minas Supergroup can be subdivided in two sequences separated
by a regional unconformity. The basal sequence involves continental to marine sediments (Dorr, 1969;
Renger et al., 1995) representing the development stage of the passive margin basin while the
overlying sequence marks the inversion of the Minas Supergroup passive margin (Renger et al., 1995).
Finally, the Itacolomi Group is the youngest unit in the supracrustal sequence. The Group comprises
an up to 2000 m-thick pile of alluvial sediments that is separated from the Minas Supergroup by a
regional unconformity. The Itacolomi Group is viewed as an intermontane molasse accumulated
during the collapse phase of the Rhyacian orogeny (Alkmim & Martins-Neto, 2012).

3.2.3 Tectono-metamorphic and magmatic evolution

Recently Lana et al. (2013) provided combined LA-ICP-MS and SHRIMP U–Pb ages for the
gneisses of the Southern São Francisco craton identifying three main periods of magmatism spanning
from 3220 to 2770 Ma. The first magmatic pulse, which is poorly preserved in the north-south
elongated Santa Barbara complex east of the Bação dome, ranges from 3220 to 3200 Ma. These ages
match with Sm–Nd model ages determined by Teixeira et al. (1996) from orthogneisses sampled in
the western part of the craton (i.e. the Campo Belo complex) as well as with the age of detrital zircons
from the greenstone belt (Machado et al., 1992, 1996) suggesting the existence of various fragments of
Paleoarchean crust. The following periods of magmatic production, named Rio das Velhas I and II by
69
Lana et al. (2013), occurred at 2930-2900 Ma and 2800–2770 Ma respectively and account for most of
the gneiss occurrences in the craton. Whole-rock Sm-Nd model ages determined by Teixeira et al.
(1996) from gneisses crystallized during these events support the involvement of older sialic crust in
their genesis indicating crustal reworking. Lana et al. (2013) have found that many magmatic zircons
from the gneisses of the Rio das Velhas I period (ca. 2900 Ma) record metamorphic ages that overlap
within error with the magmatic ages of the Rio das Velhas II event (2800–2770 Ma). Moreover, a few
plutons intruded during the 2800–2770 Ma event crop out as homogeneous weakly to non-deformed
bodies (i.e. the Caeté trondhjemite and the Samambaia tonalite; Carneiro, 1992; Machado et al., 1992).
These lines of evidence suggest that the entire continental crust of the Southern São Francisco craton
was affected by magmatism and amphibolite-facies metamorphism between 2800 and 2770 Ma. In
addition, during the Rio das Velhas II magmatic-metamorphic event, felsic volcanic rocks were
erupted within the Rio das Velhas greenstone belt. SHRIMP and ID-TIMS U–Pb zircon dating of three
volcanoclastic greywacke samples dated the eruptive event at 2792 ± 11, 2773 ± 7 and 2751 ± 9 Ma,
indicating a range of about 40 Ma for the felsic magmatism (Machado et al., 1992, 1996; Noce et al.,
2005). Zircon U–Pb crystallization age for subordinate dacitic flows intercalated within a sequence of
mafic- to ultramafic volcanic rocks in the greenstone belt yielded an age of 2772 ± 6 Ma (Machado et
al., 1992).

Following the emplacement of the gneisses during the Rio das Velhas II event, the Southern
São Francisco craton underwent a last period of widespread granite magmatism between 2755 and
2700 Ma (Romano et al., 2013). This produced slightly- to medium-foliated rocks, emplaced as large
batholiths, extensive horizontal sheets or smaller veins and dikes into the pre-existing crust. Finally, a
volumetrically minor event of granite production (accounting for less than 1% of the Archean crust)
emplaced at ca. 2612 Ma (Romano et al., 2013).

Paleoproterozoic U–Pb TIMS ages were obtained by Machado et al. (1992) in the Bação
complex from monazite crystals (2020–2030 Ma) from two pegmatites cutting the basement close to
the contact with the supracrustal rocks as well as for titanite from an amphibolite enclave in the
gneisses (2059 Ma). This evidence indicates that partial melting producing granitic pegmatites and
amphibole-facies metamorphism affected the Archean basement during the Transamazonian orogeny.
However, Paleoproterozoic deformation of the Archean crust seems restricted to narrow zones at the
tectonic contact between the basement and the supracrustal sequences, not being regionally
penetrative. No Paleoproterozoic ages were obtained from zircon crystals from the gneisses and
granitoids of the Southern São Francisco craton. In the central part of the Bação and Bonfim
complexes, zircon U–Pb ages determined for leucocratic dikes both concordant with and crosscutting
the foliation of banded gneisses yielded ages ranging from 2703 to 2774 Ma (Carneiro, 1992; Romano
et al., 2013).

70
3.3 ANALYTICAL TECHNIQUES

Fifty-nine rock samples were collected from three domes in the Southern São Francisco
craton, namely the Bação, Bonfim and Belo Horizonte complexes (Fig. 3.2) and were analyzed for
major and trace elements at the Central Analytical Facility, University of Stellenbosch, South Africa.
Major element compositions were analyzed by X-ray fluorescence spectrometry (XRF) on glass beads
prepared with La-free flux. Trace element compositions have been obtained from the same fused
beads used for major element determination by applying the method described by Eggins (2003) and
analyzed using an Agilent 7500 ICP-MS coupled with a Nd:YAG 213 nm New Wave laser ablation
system. A detailed description of individual rock samples as well as a full report on the analytical
techniques used for major and trace element determination are provided in Appendix A6. Selected
whole-rock major and trace element analyses are reported in Table 3.1, the complete dataset is
presented in Appendix A7.

We determined the U–Pb zircon age of sixteen granitoid and gneiss samples as well as the age
of an amphibolitic and a pegmatitic dike intruding into the Bonfim and Bação domes, respectively.
From these samples, zircon grains were separated from ca. 5 kg of rock by means of standard crushing
techniques (jaw crusher, disc mill), manual panning and magnetic separation. Approximately 200
grains from each sample were hand-picked under a binocular microscope, mounted in epoxy resin, and
then ground down and polished to expose their interior. Prior to analytical work, all grains were
characterized by cathodoluminescence (CL) and back-scattered electron imaging using a scanning
electron microscope (JEOL JSM-6490) at Stellenbosch University. Zircon U–Pb isotope analyses were
undertaken at the Departamento de Geologia of the Universidade Federal de Ouro Preto (Brazil) in
two analytical sessions: the first one in June–July 2013, the second in May 2014. The analyses of the
first session were performed using an Agilent 7700 Q-ICP-MS coupled to a 213 nm New Wave laser
while the analyses produced one year later used a Thermo Scientific Element 2 sectorfield (SF)-ICP-
MS coupled to a CETAC LSX-213 G2+ laser system. The two methodologies are described in more
detail in Appendix A6.

In order to test the validity of the applied methods and the accuracy and external
reproducibility of the obtained age data, multiple analyses of Plešovice reference zircon (Sláma et al.,
2008) were performed during each analytical session. In addition, multiple analyses of reference
zircon M127 (Klotzli et al., 2009) and of the in-house reference zircon BB were performed during the
second analytical session. The Plešovice secondary standard gave Concordia ages of 339.6 ± 0.8 Ma
(2SD, n = 50; MSWD = 1.4) and 337.0 ± 0.6 Ma (2SD, n = 98; MSWD = 0.96) for analytical session
one and two, respectively. The calculated age is consistent, within uncertainty, with the ID-TIMS
value reported by Sláma et al. (2008). The M127 secondary standard gave a Concordia age of 523.3 ±
1.0 Ma (2SD, n = 63; MSWD = 0.70). This age is consistent with the ID-TIMS age of 524 ± 2 Ma
reported by Klotzli et al. (2009). The in-house BB reference material gave a Concordia age of 565.1 ±
71
1.8 Ma (2SD, n = 25; MSWD = 0.70). This is in agreement with the LA-ICP-MS long-term average of
BB (562 Ma; Santos et al., 2014) reference zircon at the Universidade Federal de Ouro Preto. The
results of the LA-ICP-MS analyses for unknowns and secondary standards are presented in Appendix
A8. The LA-ICP-MS data were reduced using the software Glitter (Van Achterbergh et al., 2001) with
ages calculated and plotted on Concordia diagrams using the IsoplotEx 4 program (Ludwig, 2003).
Uncertainties given for individual analyses (ratios and ages) are at the 1 sigma level.

Eleven of the fifty-nine rock samples for which the whole-rock major and trace element
composition was determined, were collected in the same location of the samples dated by Romano et
al. (2013) and Lana et al. (2013). Thus, for twenty-eight samples both U–Pb zircon ages and
compositional data are available. A compilation of zircon U–Pb age data for the Bação, Bonfim and
Belo Horizonte domes is presented in Table 3.2.

3.4 RESULTS

3.4.1 Field relationships

The Bação, Bonfim and Belo Horizonte complexes (Fig. 3.2) are poorly exposed with fresh
outcrops limited to recent road cuts, quarries and riverbeds. In the field, the rocks of the basement can
be subdivided into four main groups:

(i) fine-grained banded gneisses intruded by (ii) and (iii);

(ii) leucogranites and aplitic/pegmatitic veins and dikes;
(iii) medium- to coarse-grained, mostly weakly foliated granites (sensu lato);
(iv) migmatitic banded gneisses intruded by (ii).

The fine-grained gneisses are characterized by alternating bands of light-coloured and

mesocratic, or very rarely, melanocratic material which vary in width from 2 mm to up to 10 cm, with
most of the bands having widths of ca. 1 cm (Fig. 3.2a and b). The deformation is mostly recorded by
the orientation of biotite and feldspar grains in the mesocratic bands while leucocratic bands are
typically coarser-grained and more isotropic. The banded gneisses contain locally abundant
leucogranitic, pegmatitic and aplitic veins that are both parallel to and crosscut the gneissosity (rocks
of group (ii)). Felsic bodies oriented sub-parallel to the banding (i.e. hereafter leucogranitic sheets)
have widths ranging from a few centimetres to ca. 60 cm and may either be foliated or undeformed.
Their abundance in the study area is highly variable, reaching a maximum volume of up to ca. 15 % of
the whole outcrop (Fig. 3.2a). The leucogranitic sheets together with the main foliation are often
folded forming tight to isoclinal folds. The folds are commonly restricted to single leucogranitic sheets
not affecting neighboring sheets of identical appearance. A younger generation of leucogranitic,
pegmatitic and aplitic dikes crosscuts both the gneissic banding and the leucogranitic sheets and also
intrudes the medium- to coarse-grained granites of group (iii). Pegmatitic and aplitic dikes have a
maximum width of about 2 m and commonly show variable orientation and heterogeneous thickness
72
along their length. The dikes are only occasionally slightly folded or boudinaged, but in general appear
neither stretched nor shortened.

Figure 3.2 – Field photographs. (a) Leucogranitic sheets and dikes forming disharmonic folds hosted within a
banded gneiss, Bação complex; (b) Banded gneiss hosting cm-scale leucogranitic sheets following the
gneissosity, Bação complex; (c) Medium-grained granite domain crosscutting the gneiss banding, Bonfim
complex. The red arrows in the upper left part the picture indicate the granite intruding parallel to the
gneissosity; (d) Metre-scale xenolith of banded gneiss hosted within medium-granite granites, Bação complex;
(e) Coarse-grained granodiorite, Mamona batholith, Bonfim complex; (f) Contact between biotite-rich and
plagioclase–quartz–biotite granites crosscut by a leucogranitic dike.

73
 Medium- to coarse-grained granitic rocks form texturally and compositionally composite large
batholiths (e.g. the Mamona batholith; Fig. 3.2) and plutons as well as relatively small-scale domains
and stocks intruded in or closely associated with banded gneisses. Although typically weakly foliated,
they may also locally develop strong solid-state fabrics such as prolate L > S fabric defined by aligned
crystals of biotite and stretched rods of quartz and plagioclase. Moreover, coarse-grained highly
foliated biotite-bearing syenogranites locally developing augen gneiss structure were observed in the
north-eastern part of the Pequi batholith in the Belo Horizonte complex (Fig. 3.1).

The relationship between banded gneisses and granitoids is observable in some key outcrops
in the Bação and Bonfim complexes. The intrusion of granitic rocks into banded gneisses produces
different features depending on the relative proportion of the two rock-types. From gneiss- to granite-
dominated outcrops, we observe:

(i) small dikes of medium-grained granites cutting both the gneissic banding and the
leucogranite sheets;
(ii) complex intermingled gneiss-granite structures (Fig. 3.2c);
(iii) metre-to decametre-scale domains of granites intruding the banded gneisses;
(iv) xenoliths of banded fine-grained gneisses hosted within granites (Fig. 3.2d).

Five banded gneiss samples show complex polydeformed structural patterns exhibiting
disharmonic folding of the banding associated with high volumes (up to 20 vol.%) of non-foliated
leucogranitic material forming folded sheets as well as decametric pods. Locally, some outcrops
display an even more chaotic aspect with boudinaged and dismembered gneiss fragments isolated in a
granitic matrix. These rocks are interpreted as migmatite. This interpretation is highly controversial as
no clear metamorphic reactions indicating in situ partial melting were recognized in the field. The
apparent lack of peritectic phases probably suggests congruent melting of the orthogneisses.

Finally, it is worth noting that although the three complexes display similar field relationship
and rock associations, minor but significant differences exist between them. In particular, large
granitic batholiths characterize the outer part of the Bonfim and Belo Horizonte domes, at the contact
with the supracrustal rocks, while in the Bação complex, granitic rocks mostly crop out in the central
part of the dome. Moreover, in the Bação complex, gneisses and granites are often closely spatially
associated, with medium-grained granitic stocks and domains intruding the gneisses.

3.4.2 Petrography

The rock-forming minerals in the fine-grained banded gneisses are predominantly plagioclase
(25–50 vol.%), quartz (25–35 vol.%), biotite (5–10 vol.%) and K-feldspar (5–20 vol.%). Biotite is
ubiquitous and commonly the only mafic mineral in the rock, but hornblende may also occur.
Magnetite, ilmenite, zircon, apatite, titanite and more rarely monazite are the most common accessory
phases. The plagioclase over alkali feldspar ratio in the gneisses is strongly variable reflecting
74
composition ranging from trondhjemite/tonalite to granodiorite and monzogranite. Alkali-feldspars are
generally interstitial, but they occasionally form cm-scale phenocrysts. We used this characteristic to
further subdivide the gneisses in two groups. The first and most common group of rocks is made of K-
feldspar poor gneisses (K-feldspar < 10 vol.%) in which K-feldspars form anhedral interstitial crystals.
The second group is composed of rocks having subhedral phenocrysts of K-feldspar reaching up to 4
cm in length.

Leucogranitic sheets and leucogranitic, pegmatitic and aplitic dikes have similar modal
composition consisting of quartz (30–50 vol.%), K-feldspar (20–40 vol.%), plagioclase (10–20 vol.%)
and minor biotite (< 5 vol.%). In the pegmatites, the modal abundance of K-feldspar may exceed 50
vol.%, tourmaline is common and muscovite is also observed.

The granitic rocks in the Bação, Bonfim and Belo Horizonte complexes are medium- to-
coarse-grained and exhibit either equigranular or porphyritic textures, with the latter given by the
occurrence of cm-sized K-feldspar crystals set in a medium-grained groundmass. Granitic rocks range
in composition from tonalite to syenogranite and most of the rocks have biotite as the only mafic
mineral. Granitic rocks from the Bonfim complex show the greatest mineralogical variability. In the
northern and northeastern part of this complex, Carneiro (1992) described a foliated two-mica granite
in which muscovite is locally more abundant than biotite (the Souza Noschese pluton; Fig. 3.1) as well
as a medium-grained amphibole- and epidote-bearing tonalite (the Samambaia pluton; Fig. 3.1). Based
on the relative abundance of K-feldspar and plagioclase, we subdivided the granitoids into two groups.
The first one is composed of plagioclase-dominated rocks (i.e. tonalite and granodiorite) in which K-
feldspar is mostly anhedral, and the second one of monzogranites and syenogranites in which euhedral
and subhedral K-feldspars are visible at hand sample scale. The distribution of the two rock types is
not the same in the different complexes: in the Belo Horizonte complex plagioclase-dominated rocks
prevail while monzogranites are dominant in the Bonfim complex.

Associating field and textural observations, six rock types were recognized:

(i) fine-grained banded gneisses having minor and interstitial K-feldspar (banded
gneisses);
(ii) banded gneisses characterized by the occurrence of cm-sized K-feldspar (Kfs-bearing
gneisses);
(iii) Plagioclase-dominated granitic rocks (Plg-rich granitoids);
(iv) K-feldspar-dominated granitic rocks (Kfs-rich granites);
(v) migmatites;
(vi) leucogranitic sheets and leucogranitic, aplitic and pegmatitic dikes (leucogranites).

In the study area, the different rock types are not equally represented. The banded gneisses
represent the most abundant and widespread rock type followed by Kfs-rich granites in the Bonfim

75
and Bação complexes and by Plg-rich granitoids in the Belo Horizonte complex. Leucogranites,
aplites and pegmatites are ubiquitous in the three complexes while migmatites and Kfs-bearing
gneisses are rare.

3.4.3 Whole-rock geochemistry

The normative feldspar classification diagram for granitoids (An–Ab–Or; O‘Connor, 1965)
shows that the Bação, Bonfim and Belo Horizonte complexes are formed of rocks having relatively
low calcium contents (i.e. low An) and variable potassium over sodium ratios (Fig. 3.3). The great
majority of samples (ca. 90 %) plot either in the trondhjemite or granite fields: banded gneisses plot in
the trondhjemite field, Kfs-rich granites and leucogranites in the field of granites while Plg-rich
granitoids straddle the trondhjemite-granite boundary. Most of the rocks analyzed have high to very
high silica contents (SiO2 > 72 wt.%), peraluminous affinity, low amounts of ferromagnesian oxides
(Fe2O3tot + MgO < 4 wt.%) and low CaO concentrations (1 < CaO > 3 wt.%; Fig. 3.4). Banded
gneisses and Plg-rich granitoids share similar chemical features exhibiting higher Al 2O3, CaO and
Na2O as well as lower silica and K2O than Kfs-rich granites and leucogranites (Fig. 3.4). Moreover,
Kfs-rich granites have mg# (Mg/(Mg + Fetot)) that is significantly lower (mg# 0.2–0.3) than the mg#
values of both banded gneisses and Plg-rich granitoids (mg# 0.3–0.4).

Figure 3.3 – Normative An-Ab-Or triangle (O‘Connor, 1965). The yellow field indicates Archean TTGs, from
Moyen & Martin (2012).

76
 The major and trace element composition of Plg-rich granites is highly variable. About two-
thirds of the samples from this group (i.e. mostly those from the Belo Horizonte complex) have
chemical composition perfectly matching the composition of banded gneisses while the rest of the
samples exhibit chemical features that are transitional (e.g. 3 < K 2O < 4.5 wt.%; Fig. 3.4) between
banded gneisses and Kfs-rich granites. Banded gneisses and Plg-rich granitoids are characterized by
values in Rb and Y that are lower and Sr contents that are higher than those exhibited by Kfs-rich
granites (Fig. 3.5). In Fig. 3.6, we compared the average rare earth element patterns for the different
groups of granites and gneisses. Banded gneisses and Plg-rich granitoids rocks have patterns similar to
TTGs, showing relatively high contents of Light Rare Earth Elements (LREE), low Heavy Rare Earth
Element (HREE) contents and Eu anomalies ranging from positive (Eu/Eu* = 1–1.3) to negative
(Eu/Eu* = 0.5–1). The relatively steep REE patterns of these rocks result in La/Yb values that,
although highly variable, are generally higher than 10. On the contrary, Kfs-rich granites have higher
Rare Earth Elements (REE) contents, showing relatively enriched HREE contents and pronounced Eu
anomalies (Eu/Eu* = 0.2–0.5). In the La/Yb vs. Yb and Sr/Y vs. Y diagrams, which are commonly
used to describe the TTG series, most of the banded gneisses and Plg-rich granitoids (ca. 90 %) plot in
the compositional field of TTGs (Fig. 3.5).

Migmatites and Kfs-bearing gneisses have silica and Al2O3 contents similar to banded
gneisses and Plg-rich granitoids but display large variations in both K2O (2.5 < K2O < 5.5 wt.%),
Na2O (3.2 < Na2O < 5.0 wt.%), Sr and Rb (Fig. 3.5). The five migmatitic samples have high La/Yb
and Sr/Y ratios with REE patterns that are steeper than those observed for banded gneisses (Fig. 3.6).
On the other hand, Kfs-bearing gneisses display REE patterns that are transitional between those of
banded gneisses and Kfs-rich granites. Finally, leucogranitic sheets and leucogranitic dikes have quite
scattered major element composition with K2O ranging from 3.8 to 8.9 wt.%, low to very low MgO +
Fe2O3tot (0.2–1.1 wt.%) and silica between 71 and 76 wt.% (Fig. 3.3). In the SiO2 vs. Al2O3 diagram
they generate a quasi-linear negative trend (r2 = 0.90) and, for equivalent silica contents, they are
slightly enriched in Al2O3 with respect to both granites and gneisses (Fig. 3.4). The leucogranitic rocks
exhibit Eu anomalies varying from slightly negative to strongly positive (Eu/Eu* = 0.9–2.8), low REE
contents and flat REE patterns resulting in low La/Yb ratios (Fig. 3.5 and 3.6).

77
Figure 3.4 – Harker diagrams showing the major element features of gneisses, granites and leucogranites. Grey
dots are TTGs from Moyen (2011). Black, grey and white stars indicate the average composition for high-,
medium- and low-pressure TTGs, respectively (Moyen & Martin, 2012). Discriminant boundaries in the K 2O vs.
SiO2 diagram are from Gill (1981). In the same diagram, the field of post-Archean I-type granitoids is from
Clemens et al. (2011). The two grey fields are from Carneiro (1992) and indicate the composition of the Souza
Noschese two mica granite (SN. granite) and Samambaia tonalite (S. tonalite). Arrows A and B indicate the
compositional trend resulting from fractionation of 10 wt.% of hornblende. Arrow C indicates the compositional
trend resulting from fractionation of 10 wt.% of plagioclase. The composition of hornblende and plagioclase
used in the fractional crystallization modelling are from Watkins et al. (2007). Arrows D, E and F result from
adding 30 wt.% of a high-K magma having the composition of the Souza Noschese (sample FQ49) pluton to the
average composition of high- and medium-pressure TTGs.

78
Figure 3.5 – Harker diagrams showing the trace element composition of gneisses, granites and leucogranites.
Grey dots are TTGs from Moyen (2011). The two grey fields in the Rb vs. Sr diagram are from Carneiro (1992)
and indicate the composition of the Souza Noschese two mica granite (SN. granite) and Samambaia tonalite (S.
tonalite).

Figure 3.6 – Average chondrite-normalized REE patterns for Kfs- and Plg-rich granitoids, migmatites, banded
gneisses and Kfs-bearing gneisses. The field of TTGs is drawn using the composition of high-, medium- and
low-pressure TTGs (Moyen, 2011). Normalization values are from McDonough & Sun (1995).

79
Table 3.1 – Major and trace elements composition of orthogneisses and granites from the Bação, Bonfim and
Belo Horizonte complexes. Major oxide composition were normalized to 100 wt.%. The sample locality is
shown in Fig. 3.1.
Sample FQ5 FQ6 FQ8 FQ13 FQ17 FQ21 FQ22 FQ28 FQ29 FQ32
Magmatic
Bação Bação Bação Bação Bação Bação Bação Bonfim Bonfim Bonfim
dome
Locality 12 12 11 5 10 8 7 20 19 17
Field Granite Gneiss Gneiss Gneiss Gneiss Leucogr. Gneiss Granite Granite Granite
Kfs- Kfs- Kfs-
Group Plg-rich Banded Banded Leucogr. Banded Kfs-rich Plg-rich
bearing bearing rich
SiO2 74.31 70.79 73.56 72.09 71.43 72.98 72.53 75.25 67.99 74.18
TiO2 0.17 0.49 0.36 0.37 0.30 0.03 0.36 0.20 0.61 0.26
Al2O3 14.32 14.99 13.90 14.69 15.51 15.32 15.02 13.17 15.30 13.22
Fe2O3tot 1.59 3.14 2.23 3.03 2.05 0.37 2.24 1.90 4.48 2.09
MnO 0.02 0.03 0.05 0.04 0.03 0.01 0.03 0.03 0.07 0.03
MgO 0.30 0.78 0.47 0.57 0.80 0.07 0.58 0.28 1.52 0.37
CaO 1.44 2.45 1.03 2.60 2.74 1.12 2.29 0.99 3.86 0.89
Na2O 4.27 3.58 3.56 4.91 4.83 3.97 5.10 3.38 3.43 3.29
K2O 3.54 3.61 4.73 1.56 2.21 6.12 1.75 4.73 2.55 5.60
P2O5 0.04 0.15 0.11 0.13 0.10 0.01 0.10 0.07 0.18 0.06
LOI 0.75 1.05 0.69 0.70 0.42 0.51 0.32 0.36 0.74 0.72
K2O/Na2O 0.83 1.01 1.33 0.32 0.46 1.54 0.34 1.40 0.74 1.70
Sc 5.8 7.0 7.0 5.5 4.9 2.7 3.7 5.2 10.4 5.2
V 16.1 38.8 25.8 28.1 32.8 13.1 31.0 19.4 75.3 21.6
Cr 14.3 23.2 21.4 24.3 23.6 10.5 18.5 13.0 31.0 18.3
Ni 7.7 9.7 7.6 7.7 9.7 8.1 6.5 9.3 16.2 5.0
Rb 100 152 343 61 73 220 84 236 117 233
Sr 126 154 59 602 564 205 307 62 315 45
Y 22.6 24.5 58.7 15.3 4.2 9.4 13.3 59.3 34.9 42.1
Zr 127 311 178 268 130 43 229 148 211 245
Nb 18.9 21.1 24.7 6.8 3.5 3.5 8.7 28.6 14.0 12.6
Mo 0.6 1.6 0.9 0.7 0.3 0.4 0.3 0.7 2.8 1.7
Cs 2.7 6.2 17.3 3.9 3.7 6.6 6.2 4.4 3.0 5.9
Ba 776 940 296 608 1087 440 937 340 522 472
La 38.90 66.77 42.32 69.63 27.60 4.63 45.56 50.05 65.64 53.59
Ce 78.9 152.2 93.6 132.3 47.6 8.5 82.6 115.5 110.5 101.3
Pr 8.28 14.56 10.96 13.99 4.77 0.92 8.12 11.64 12.16 10.64
Nd 29.04 52.01 42.56 51.02 17.51 3.24 26.38 39.12 42.49 36.79
Sm 6.03 8.39 10.04 8.31 2.63 1.20 4.13 8.87 7.24 7.43
Eu 1.08 1.55 0.43 1.85 0.64 0.38 1.17 0.57 1.32 0.60
Gd 5.12 5.97 10.21 5.37 1.66 1.30 2.89 8.83 6.34 7.04
Tb 0.79 0.85 1.76 0.66 0.21 0.27 0.40 1.57 1.01 1.12
Dy 4.52 4.66 10.63 3.64 0.96 1.71 2.31 10.30 5.90 7.85
Ho 0.94 0.87 2.13 0.59 0.15 0.40 0.43 2.11 1.19 1.51
Er 2.29 2.48 6.05 1.49 0.47 1.03 1.28 6.10 3.47 4.40
Tm 0.36 0.33 0.92 0.18 0.06 0.16 0.18 0.85 0.53 0.67
Yb 2.32 2.19 6.23 1.30 0.38 1.00 1.32 5.77 3.42 4.45
Lu 0.32 0.35 0.87 0.17 0.05 0.13 0.16 0.77 0.51 0.60
Hf 4.28 8.38 5.78 7.13 3.58 1.67 5.95 4.70 6.10 8.30
Ta 1.54 1.55 3.04 0.98 0.90 1.24 1.48 3.66 2.02 2.08
Pb 35.3 31.8 35.0 15.1 15.9 32.2 20.0 50.2 15.1 30.6
Th 18.2 20.3 35.9 13.1 6.4 9.3 18.0 33.3 17.5 13.1
U 2.8 2.7 1.1 1.8 0.7 17.5 4.7 11.4 3.0 4.7

80
 Sample FQ41 FQ44 FQ49 FQ53 FQ60 FQ65 FQ70 FQ77
Magmatic Belo Belo Belo Belo
Bonfim Bonfim Bonfim Bonfim
dome Horizonte Horizonte Horizonte Horizonte
Locality 21 22 28 27 35 32 37 30
Field Gneiss Leucogr. Granite Gneiss Granite Granite Gneiss Gneiss

Group Banded Leucogr. Kfs-rich Banded Plg-rich Plg-rich Banded Kfs-rich

SiO2 73.84 76.08 75.75 74.42 72.79 72.03 73.93 75.80

TiO2 0.15 0.02 0.10 0.19 0.19 0.33 0.16 0.17
Al2O3 14.92 13.81 13.61 13.91 15.05 14.71 14.68 12.80
Fe2O3tot 1.19 0.16 1.00 1.40 1.80 2.72 1.34 2.00
MnO 0.03 0.01 0.01 0.03 0.04 0.05 0.04 0.02
MgO 0.39 0.05 0.13 0.46 0.48 0.67 0.37 0.24
CaO 1.52 0.68 0.89 1.74 1.93 2.38 1.73 0.69
Na2O 5.05 3.72 3.80 4.29 4.86 4.55 4.71 3.54
K2O 2.85 5.44 4.65 3.46 2.80 2.46 3.00 4.70
P2O5 0.05 0.03 0.05 0.08 0.07 0.10 0.04 0.03
LOI 0.48 0.37 0.49 0.65 0.86 0.64 0.28 0.68
K2O/Na2O 0.56 1.46 1.22 0.80 0.58 0.54 0.64 1.33
Sc 4.0 2.3 3.5 3.7 4.3 7.2 3.7 3.7
V 21.6 11.8 12.6 25.8 27.9 33.6 25.0 27.5
Cr 13.2 10.7 35.3 24.8 25.7 18.3 20.4 10.0
Ni 7.6 5.7 12.5 8.8 14.3 10.4 7.3 5.1
Rb 109 172 250 93 82 110 104 323
Sr 334 195 39 337 381 205 342 46
Y 5.7 4.5 32.5 3.7 7.3 11.2 6.0 93.3
Zr 73 15 67 96 95 210 111 254
Nb 6.8 1.3 15.8 3.5 4.1 9.3 4.7 34.8
Mo 0.4 0.5 0.3 0.5 0.8 0.4 0.5 0.4
Cs 4.0 1.0 9.8 1.1 4.5 3.6 2.4 1.0
Ba 420 713 263 1125 677 431 790 335
La 9.27 9.55 18.42 18.86 13.55 53.74 21.25 107.67
Ce 19.9 16.9 43.7 32.2 27.4 89.6 41.9 217.8
Pr 2.00 1.72 4.58 3.49 2.83 9.44 4.04 24.03
Nd 7.53 5.62 16.45 11.18 10.20 32.51 13.18 82.41
Sm 1.51 1.16 4.36 2.07 2.01 5.78 2.32 17.78
Eu 0.37 0.40 0.63 0.65 0.51 0.68 0.55 0.52
Gd 1.41 0.97 4.59 1.51 1.61 4.62 1.95 15.16
Tb 0.21 0.18 0.97 0.21 0.26 0.58 0.24 2.61
Dy 1.07 0.99 5.77 0.96 1.27 2.89 1.29 16.28
Ho 0.24 0.18 1.25 0.15 0.26 0.47 0.21 3.45
Er 0.60 0.46 3.42 0.41 0.76 1.03 0.62 10.33
Tm 0.09 0.06 0.49 0.05 0.11 0.14 0.10 1.71
Yb 0.60 0.49 3.06 0.31 0.72 0.86 0.58 12.06
Lu 0.10 0.06 0.42 0.06 0.12 0.12 0.09 1.81
Hf 3.08 0.74 2.88 3.24 2.43 5.98 3.28 10.02
Ta 1.23 0.82 2.07 0.63 0.58 0.65 0.63 1.99
Pb 33.8 49.5 50.2 26.2 17.2 25.8 26.3 41.3
Th 4.7 4.0 10.2 5.6 4.0 22.8 11.4 47.9
U 3.8 2.2 10.5 1.4 1.2 3.6 2.0 4.2

81
 3.4.4 Geochronology: cathodoluminescence images and U–Pb ages

Zircon LA-ICP-MS U–Pb age data were obtained from eighteen rocks from the Belo
Horizonte, Bonfim and Bação complexes. The dated samples are from five of the six rock types
recognized as no U–Pb ages were obtained for migmatites. In addition, an amphibolite dike
crosscutting a foliated granite in the Bonfim complex was also dated. In Appendix A6, detailed
description of zircon size, morphology, internal structure as well as U–Pb data is provided. In Table
3.2, we present a summary of U–Pb ages for the rocks of the three complexes. Concordia and
207
Pb/206Pb weighted average diagrams are presented in Fig. 3.8.

Figure 3.7 – Cathodoluminescence images of zircon grains. Light grey circles indicate the position of laser spots
and the 207Pb/206Pb ages.

In all samples, zircon crystals are typically prismatic or stubby, with sizes that mostly vary
between 75 and 400 μm and length:width ratios up to c. 5:1 (Fig. 3.7). Cathodoluminescence images
allow three main groups of samples to be distinguished. For most of the samples,

82
cathodoluminescence images reveal a main population of zircon grains showing fine-scale oscillatory
zoning patterns of inferred magmatic origin and a subset (< 10 %) of dark and feature-less zircon
crystals yielding highly discordant U–Pb ages. A second group of rocks (five samples) is characterized
by zircon crystals that show complex zoning, displaying typical core–rim relationships with
homogeneous light-coloured cores surrounded either by oscillatory zoned or, more rarely, by
structureless rims (e.g. FQ17; Fig. 3.7). Finally, four rocks collected in the Belo Horizonte complex
are characterized by homogeneously dark and structureless zircon grains (e.g. FQ 70 and FQ74; Fig.
3.7). These grains are interpreted to be formed by recrystallization of the original magmatic zircon
either during a late magmatic stage or during a subsequent and likely fluid-assisted metamorphic
event. In individual samples, the Th/U of the zircon grains is highly variable (0.70–0.02) with rims
having similar Th/U ratios to the cores. Overall, ca. 60 % of grains have Th/U ranging between 0.2
and 0.5 and only 10 % of them have Th/U < 0.1. For just one of the samples (FQ 2) with zircon grains
displaying core–rim relationship there is a systematic difference in Th/U between cores and rims, with
rims having lower Th/U ratios.

We determined the zircon U–Pb age for nine gneisses. Eight of them are fine-grained banded
orthogneisses while one, sampled in the Belo Horizonte complex, is a K-feldspar rich augen gneiss.

Three banded orthogneisses, yielding a crystallization age of ca. 2860 Ma record a

metamorphic overprint occurring between 2670 and 2705 Ma (e.g. sample FQ2; Fig. 3.8). Three
gneisses gave a younger crystallization age of ca. 2780 Ma (e.g. sample FQ17; Fig. 3.8). The
occurrence of inherited cores with ages of ca. 2900 Ma in these younger gneisses suggest recycling of
older crust in the genesis of the ca. 2780 Ma rocks. For three gneisses, no crystallization ages were
obtained. Dark, structureless grains from these rocks gave metamorphic ages between 2727 and 2638
Ma (samples FQ70 and FQ52; Fig. 3.10), with the youngest metamorphic age determined from the K-
feldspar rich augen gneiss (sample FQ74; Fig. 3.8). We determined the zircon U–Pb age for five
granites. In the Bação complex, six analyses on fine-scale oscillatory zoned zircon grains from a
granitic stock gave a well-defined Concordia age of 2790 ± 3 Ma (MSWD = 0.35), which is
interpreted as the crystallization age of the rock. This granitic stock, crosscutting a banded gneiss,
represents the oldest granitoid in the studied area (Table 3.2; sample FQ11). Four other granites gave
crystallization ages between 2770 and 2680 Ma showing no inheritance and no metamorphic
overprint. In particular, ten concordant analyses (< 1 % discordant) from the foliated medium-grained
Samambaia tonalite in the Bonfim complex (sample FQ29), yield a Concordia age of 2773 ± 2 Ma
(MSWD = 1.3; Fig. 3.8). This age is identical to the 2778 ± 3 Ma age determined by TIMS by
Machado & Carneiro (1992) for the same rock. The last granite analyzed was collected in the Belo
Horizonte complex and is characterized by fractured structureless dark or light grey zircon grains.
Nine concordant (< 1 % discordant) spot analyses gave a Concordia age of 2644 ± 4 Ma (95 % c.l.;

83
MSWD = 2.3; Table 3.2) which is interpreted to reflect either the timing of the latest high-grade
metamorphic event in the complex or the interaction with late-magmatic fluids.

Table 3.2 – Summary of U–Pb ages for gneisses, granitoids and leucogranites of the Bação, Bonfim and Belo
Horizonte complexes. Data are from Romano et al. (2013); Lana et al. (2013); Machado & Carneiro (1992);
Machado et al. (1992); Noce et al. (1997, 1998); Chemale et al. (1993b); Noce et al. (2005). All errors are
quoted at 2σ level. Samples dated in previous studies for which whole-rock chemical analyses were performed in
this study are marked with an asterisk.
Classification Age (Ma)
Sample Field Group Technique Inherited Magmatic Metamorphic References
Bação dome
MR011 Granite SHRIMP 2717 ± 5 [1]
FQ1 Granite Kfs-rich LA-Q-ICP-MS 2711 ± 3 This study
FQ2 Gneiss Kfs-poor LA-Q-ICP-MS 2868 ± 10 2705 ± 18 This study
FQ23 Gneiss Kfs-poor LA-Q-ICP-MS 2898 ± 12 2783 ± 18 This study
D07A-FQ20* Gneiss Kfs-poor SHRIMP 2918 ± 10 2775 ± 39 [2]
D07B-FQ21* Leucogranite Leucogranite SHRIMP 2931 ± 12 2774 ± 11 [2]
D06 Gneiss SHRIMP 2925 ± 8 2794 ± 15 [2]
MP1-FQ22* Gneiss Kfs-poor LA-ICP-MS 2902 ± 12 2770 ± 29 [2]
D04-FQ10* Gneiss Kfs-poor SHRIMP 2795 ± 7 [2]
FQ17 Gneiss Kfs-poor LA-Q-ICP-MS 2778 ± 2 2732 ± 10 This study
FQ11 Granite Plg-rich LA-Q-ICP-MS 2905 ± 17 2790 ± 3 2719 ± 14 This study
FQ13 Gneiss Kfs-poor LA-Q-ICP-MS ca. 2900 2790 ± 3 This study
SG-2-FQ16* Leucogranite Leucogranite LA-ICP-MS 2730 ± 7 [2]
SG-1 Amphibolite LA-ICP-MS ca. 2900 2778 ± 8 [2]
MR11-FQ5* Granite Plg-rich SHRIMP 2744 ± 10 [1]
D12-FQ6* Gneiss Kfs-bearing SHRIMP 2764 ± 10 [2]
OPU 4094 Pegmatite LA-SF-ICP-MS 2693 ± 13 This study

Bonfim dome
FQ29 Granite Plg-rich LA-SF-ICP-MS 2773 ± 2 This study
625-656 Granite TIMS 2778 ± 3 [3]
FQ37 Amphibolite LA-SF-ICP-MS 2879 ± 14 2719 ± 14 This study
FQ41 Gneiss Kfs-poor LA-SF-ICP-MS 2852 ± 16 This study
FQ40 Gneiss Kfs-bearing LA-SF-ICP-MS 2854 ± 18 2670 ± 15 This study
FQ52 Gneiss Kfs-poor LA-SF-ICP-MS 2727 ± 11 This study
FQ51 Granite Plg-rich LA-SF-ICP-MS 2678 ± 10 This study
FQ44 Leucogranite Leucogranite LA-SF-ICP-MS 2782 ± 17 2679 ± 37 This study
11 Gneiss TIMS ca. 2920 2772 ± 6 [3]
12 Leucogranite TIMS 2703 ± 20 [3]
M-88-11 Granite Kfs-rich TIMS 2721 ± 3 [4]
MR14A-FQ32* Granite Kfs-rich LA-ICP-MS 2730 ± 7 [1]
MR70A Leucogranite LA-ICP-MS 2723 ± 7 [1]
MR87A-FQ50* Granite Kfs-rich LA-ICP-MS 2613 ± 6 [1]
MR22A-FQ28* Granite Kfs-rich LA-ICP-MS 2730 ± 8 [1]
MR148A Granite LA-ICP-MS 2722 ± 9 [1]
MR31A Leucogranite LA-ICP-MS 2700 ± 8 [1]
MR10C Granite SHRIMP 2729 ± 9 [1]
MR10A Leucogranite SHRIMP 2719 ± 5 [1]
MR137A Leucogranite LA-ICP-MS 2708 ± 7 [1]
D11-FQ38* Gneiss Migmatite SHRIMP 2895 ± 13 2749 ± 10 [2]

Belo Horizonte dome

84
 MR234A-FQ62* Granite Plg-rich LA-ICP-MS 2755 ± 8 [1]
FQ60 Granite Plg-rich LA-SF-ICP-MS 2728 ± 16 This study
MR231A Granite LA-ICP-MS 2750 ± 13 [1]
MR259A Granite LA-ICP-MS 2722 ± 7 [1]
MR257A Leucogranite LA-ICP-MS 2706 ± 7 [1]
Classification Age (Ma)
Sample Field Group Technique Inherited Magmatic Metamorphic References
MR51A Granite LA-ICP-MS 2700 ± 8 [1] and [6]
CO-1 Gneiss LA-ICP-MS 3219 ± 13 2898 ± 7 [2]
D14 Gneiss SHRIMP 2783 ± 23 [2]
FQ74 Gneiss Kfs-rich LA-SF-ICP-MS 2638 ± 14 This study
FQ70 Gneiss Kfs-poor LA-SF-ICP-MS 2713 ± 3 This study
FQ65 Granite Plg-rich LA-SF-ICP-MS 2644 ± 4 This study
N-33 Gneiss TIMS 2860 ± 14 [5]
Santa Luzia Granite Kfs-rich TIMS 2712 ± 5 [5] and [6]
Ibirité Granite Kfs-rich TIMS 2698 ± 18 [6] and [7]

Volcanic event in the Rio das Velhas greensone belt

FR38 TIMS 2883 ± 6 2776 ± 26 [3]
SHRIMP > 3000 2792 ± 11 [8]
MZ-53 TIMS 2998 ± 14 2751 ± 11 [8]
SS-144 TIMS 3035 ± 5 2777 ± 7 [8]
References: [1] - Romano et al. (2013); [2] - Lana et al. (2013); [3] - Machado & Carneiro (1992); [4] - Machado et al.
(1992); [5] - Noce et al. (1998); [6] - Noce et al. (1997); [7] - Chemale et al. (1993b); [8] - Noce et al. (2005)

Zircon grains from a leucogranitic sheet crosscutting a banded gneiss in the Bonfim complex
gave a poorly defined age of 2679 ± 37 Ma. Spot analyses on five concordant and subconcordant cores
from the leucogranite gave a weighted 207Pb/206Pb mean age of 2782 ± 17 Ma that is interpreted as the
age of the hosting gneiss. Zircon grains from a pegmatitic dike in the Bação complex gave a magmatic
age of 2693 ± 13 Ma. Finally, zircon grains separated from an amphibolite dike crosscutting a foliated
granite in the Bonfim complex show sub-rounded bright cores surrounded by homogeneous dark or
oscillatory-zoned rims (sample FQ37; Fig. 3.7). Six concordant analyses of the cores (< 1 %
discordant) yield a weighted age of 2879 ± 14 Ma (MSWD = 0.68). Nine analyses of the rims passed
207
the < 5 % discordancy filter test giving a weighted mean Pb/206Pb age of 2719 ± 14 Ma (95 % c.l.;
MSWD = 0.62), which is interpreted as the crystallization age of the dike.

The U–Pb zircon ages determined in this work are in good agreement with previously
published zircon U–Pb age data for the Belo Horizonte, Bonfim and Bação complexes (Machado &
Carneiro, 1992; Machado et al., 1992; Chemale et al., 1993b; Noce et al., 1998; Lana et al., 2013;
Romano et al., 2013).

85
Figure 3.8 – Representative Concordia diagrams for the rocks of the Bação, Bonfim and Belo Horizonte
complexes. The insets are either 207Pb/206Pb weighted average age plots or Concordia age plots. Error ellipses
and error bars are at 1σ.

86
3.5 Discussion

3.5.1 Medium- and high-K granites in the Southern São Francisco craton

The major and trace element composition of granitic rocks and gneisses from the Bação,
Bonfim and Belo Horizonte complexes is compared with the composition of Archean grey gneisses
and granitoids from other cratons. In Figs. 3.4 and 3.5, the compositions of the analyzed samples are
plotted together with a large database of TTG compositions compiled by Moyen (2011). In Fig. 3.9,
gneisses and granitoids from the basement of the Southern São Francisco craton are plotted in the
ternary diagram recently proposed by Laurent et al. (2014) to classify late-Archean granitoids.

Figure 3.9 – Ternary classification diagram for late Archean granitoids. The diagram and the fields for TTG,
sanukitoids, biotite and two-mica granites as well as hybrid granites are from Laurent et al. (2014). The triangle
apexes are: 2 * A/CNK (molar Al2O3/(CaO + K2O + Na2O) ratio), Na2O/K2O and 2 * (FeOtot + MgO) * (Sr + Ba)
wt.% (= FMSB).

Banded gneisses and Plg-rich granitoids share similar chemical features with TTGs; i.e. they
have high Na2O, Al2O3 and Sr associated with relatively low K2O and HREE contents (Fig. 3.4 and
3.5) and exhibit REE patterns similar to those characteristics of rocks from the TTG series (Fig. 3.6).
However, these rocks are not typical TTGs as they are significantly more silica-, K2O- and Rb-rich

87
and less aluminiferous than most of the TTGs of Moyen (2011). It is worth noting that just five out of
thirty-one of the banded gneisses and Plg-rich granitoids match the required criteria recently
established by Moyen & Martin (2012) which a rock should fulfill to be defined as a TTG. In the
ternary diagram of Laurent et al. (2014), most of the banded gneisses and Plg-rich granitoids plot in
the lower part of the field of TTGs, in the area where this field overlaps with the field of hybrid
granitoids (Fig. 3.9). Noteworthy, four Plg-rich granitoid samples collected from small granitic stocks
closely associated with banded gneisses plot in the field of biotite and two-mica granites. These
samples show chemical features that are transitional between those exhibited by banded gneisses and
the chemistry of Kfs-rich granites. In Fig. 3.9, Kfs-rich granites and leucogranites plot in the field of
biotite and two-mica granites. These rocks have chemical features that strongly differ from those of
TTG as well as those exhibited by banded gneisses and Plg-granitoids. In particular, Kfs-rich granites
have higher SiO2 and K2O and lower Al2O3 and Na2O than TTGs and banded gneisses (Fig. 3.4).
Moreover, Kfs-rich granites exhibit higher contents of incompatible elements (e.g. Rb), lower Sr
contents and higher concentrations of Y and HREE than TTGs (Fig. 3.5 and 3.6). Overall, the
composition of Kfs-rich granites resembles the composition of high-silica I-type granites (Clemens et
al., 2011).

Whole-rock major and trace element data for the rocks of the Bação, Bonfim and Belo
Horizonte complexes suggest the existence of two main groups of rocks forming ca. 80–90 % of the
basement: banded gneisses and Plg-rich granitoids constitute the first type, Kfs-rich granites the other.
It is worth noting that in the silica vs. K2O diagram of Gill (1981) banded gneisses and Plg-rich
granitoids plot in the field of medium-K rocks while Kfs-rich granites in the field of high-K rocks
(Fig. 3.4). Hereafter, we will use the term medium-K to refer to rocks having the overall chemical
characteristics of banded gneisses and Plg-rich granitoids. The term high-K granites will be used as
synonym of Kfs-rich granites.

3.5.2 Petrogenesis of gneisses and granitoids

3.5.2.1 Are medium- and high-K magmas cogenetic?

The chemical transition between medium- and high-K can be explained following two
different approaches. The first one considers the two rock types to be cogenetic (i.e. derived through
the same process or combination of processes), while the other assumes a different origin for medium-
and high-K granitoids arguing for a change in the rock-forming processes and/or in the sources
involved in their genesis. In this section, we discuss the hypothesis that rocks with TTG-like affinity
and high-K granites are cogenetic. The following models are considered:

(i) fractional crystallization from a precursor magma having chemical composition

similar to TTGs;
(ii) binary mixing between TTGs and high-K granites.

88
 The studied rocks are silica rich and exhibit, as dominant geochemical features, a positive
correlation between silica and K2O and a negative correlation between silica and Na2O. These
correlations may suggest derivation of granites and gneisses by fractional crystallization of a TTG-like
parental magma, leaving a cumulate made of hornblende ± biotite ± plagioclase. Biotite fractionation
can be easily excluded, as separation of this mineral from the parental magma would decrease the K 2O
of the residual melt generating a negative correlation between silica and K 2O. Contrarily, removal of
about 15 wt.% of early crystallized hornblende from an hypothetical high-pressure TTG magma may
account for the increase in SiO2 as well as for a decrease in CaO and MgO + Fe2O3tot (Fig. 3.4).
However, this process would not significantly increase the K 2O of the magma. The coupled
mechanical removal of hornblende and sodic plagioclase would slightly increase the K2O content (and
the K2O/Na2O) of the residual magma. This process, however, is not able to account for the chemical
variability exhibited by the inferred medium- to high-K magmatic series (Fig. 3.4). The results of our
simple fractional crystallization model match with the observation that the K2O/Na2O of adakites and
TTGs worldwide remains low throughout differentiation, i.e. their K 2O/Na2O ratio does not correlate
with differentiation indicators such as SiO2 (Moyen & Martin, 2012).

The alternative process involves mixing between TTG-like magmas and a K2O-rich granite
(model (ii)). The inferred granitic end-member is considered to have chemical composition similar to
the Souza Noschese batholith (sample FQ49; Table 3.1). Two TTG-like end-members are tested; one
has the average composition of high-pressure TTGs, the other the composition of medium-pressure
TTGs. Mixing trends obtained using the average medium-pressure component fail to account for the
chemical variability of medium-K granitoids in the silica vs. K2O, Al2O3 and Na2O diagrams (Fig.
3.4). Therefore, this end-member is not considered any further. On the other hand, mixing trends
generated using a high-pressure TTG end-member reproduce the main major element features
exhibited by the rocks having TTG-like chemical affinity. However, these trends do not account for
the overall large compositional scatter nor for the lack of correlation between silica and K2O and
between Rb and Sr exhibited by the high-K granites (Figs. 3.4 and 3.5). Moreover, even in the case
where the mixing model could describe the major and trace element composition of the analysed
rocks, this process would only provide an apparent solution to the conundrum of the transition
between medium- and high-K granitoids. In particular, what would be the origin of the inferred high-
K2O end-member? Another line of chemical evidence argues against a cogenetic origin for the two
groups of rocks. In fact, intermediate compositions are underrepresented (Fig. 3.4), resulting in a
puzzling gap in the possible crystal fractionation or mixing trends. We therefore conclude that
medium-K gneisses and granitoids and high-K granites are produced through melting of contrasting
sources. The petrogenesis of the different rock types will be discussed in the next two sections and a
synthetic sketch of the model proposed is presented in Fig. 3.10.

3.5.2.2 Crustal recycling and TTGs magmas: origin of medium-K granitoids

89
 We propose that medium-K granitoids formed by mixing between a TTG-like melt produced
by partial melting of basaltic oceanic crust and a melt derived by reworking of continental crust (Fig.
3.10). The involvement of recycled crust is supported by the Sm–Nd model ages of the gneisses,
suggesting reworking of continental crust with components up to 3380 Ma old (Teixeira et al., 1996),
as well as by the occurrence of inherited zircon cores in the rocks of the Rio das Velhas I and II events
(Lana et al., 2013). Moreover, the large number of Paleoarchean detrital zircon grains (3600–3000
Ma) in the rocks of the greenstone belt (Hartmann et al., 2006) indicates the existence of Paleoarchean
continental crust today preserved as small crust fragments in the Santa Bárbara complex and in the
Mineiro Belt (Lana et al., 2013). Numerous studies have investigated the composition of melts derived
by partial melting of natural and synthetic tonalitic-granodioritic starting materials (Skjerlie &
Johnston, 1993; Gardien et al., 1995, 2000; Patiño Douce & Beard, 1995; Castro, 2004; Patiño Douce,
2005; Watkins et al., 2007). The compositions of the starting materials as well as the composition of
melts produced in these experiments are plotted in Fig. 3.11 together with the chemistry of the studied
gneisses and granites. The melts providing the best match with the composition of medium-K
granitoids are those produced by Watkins et al. (2007) by fluid-absent partial melting of both a
hornblende-bearing and a biotite-bearing tonalite at pressures of 0.6–1.2 GPa. Melts produced from
other experiments are either too K2O-rich (i.e. high K2O/Na2O; Skjerlie & Johnston, 1993; Castro,
2004), too Al2O3-rich and poor in mafic components (Patiño Douce, 2005) or too CaO-rich (Patiño
Douce & Beard, 1995). Although providing the best match, the melts produced by Watkins et al.
(2007) have K2O/(Na2O + CaO) values that are significantly higher than those exhibited by medium-K
rocks (Fig. 3.11). In fact, only 20 % of samples from the study area plot in the field of the melts
produced by Watkins et al. (2007). This chemical discrepancy advocates for the involvement of melts
generated by partial melting of metabasalts (i.e. ―true‖ TTG melts) in the genesis of medium-K
granitoids. The REE composition of medium-K granitoids supports this mixing scenario. In fact, in the
residual assemblages produced in the fluid-absent partial melting experiments of Watkins et al.
(2007), plagioclase is abundant and garnet scarce. Therefore, melts formed by partial melting of
typical TTG rocks are expected to have significant negative Eu anomalies. Otherwise, medium-K
granitoids have ill-defined negative Eu anomalies supporting contribution from TTG magmas in their
genesis (Fig. 3.5). Fractional crystallization of plagioclase and hornblende may also contribute to
shape the chemical composition of medium-K granitoids but this process cannot account for the
overall chemical variability exhibited by the medium-K rocks nor can it explain the chemical diversity
observed between the rocks studied and TTGs. A simple argument can be used to rule out fractional
crystallization as a major rock forming process. The chemical variability in K 2O and silica exhibited
by medium-K granitoids may result from removal of about 30 wt.% of sodic plagioclase from a TTG
end-member. However, such a percentage of plagioclase removed from the composition of a TTG
melt has also the effect of decreasing the Al 2O3 of the residual magma down to ca. 12 wt.%, a value
that is significantly lower than the Al2O3 content of medium-K granitoids (Al2O3 > 14 wt.%).
90
 Our model, advocating the involvement of melts derived by recycling of earlier TTG crust in
the petrogenesis of medium-K granitoids, is in good agreement with the findings of other studies on
Archean grey gneisses (e.g. Friend & Nutman, 2005; Champion & Smithies, 2007). These studies
show that sizeable portions of grey gneiss complexes have old model ages, pointing to a long-lived
continental history.

Figure 3.10 – Sketch of the model proposed for the petrogenesis of gneisses and granitoids in the Bação and
Bonfim complexes. Abbreviations: FC, fractional crystallization; ACC, accumulation. The emplacement of
Neoarchean amphibolitic dikes (2778–2719 Ma; Fig. 3.12) suggests partial melting of the mantle.

3.5.2.3 Melting Archean metasediments: origin of high-K granites

In the ternary diagram of Laurent et al. (2014), high-K granites plot in the field of biotite and
two-mica granites, this group representing, according to Sylvester (1994), the second most abundant
lithology in late-Archean terranes. Biotite and two-mica granites are considered to represent purely
crustal-derived granites formed by partial melting of an older TTG crust (e.g. Almeida et al., 2013;
Dey et al., 2012) and/or from melting of metagreywackes and metapelites (e.g. Breaks & Moore,
1992). The composition of high-K granites is not well matched by any of the experimental melts
produced by partial melting of TTG sources (Fig. 3.11): melts produced by Patiño Douce (2005),
Skjerlie & Johnston (1993) and Castro (2004) are too Al2O3-rich while those from Watkins et al.
(2007) are too low in K2O. In spite of using similar starting material, pressure ranges and experimental
setup, the experimental work of Watkins et al. (2007) and Castro (2004) produced contrasting results.
91
Watkins et al. (2007) generated a large amount of leucogranodioritic melts (ca. 30 % of melt at 900
°C, with ca. 10 wt.% of hydrous phases in the starting material) whose composition was not as K2O-
rich as the composition of late-Archean high-K granites. On the contrary, small amounts of strongly
potassic melt (melt < 10 % at T > 900 °C) are produced in the experiments of Castro (2004). The
overall infertility of the source material, rather than the composition of the melt, led Castro (2004) to
rule out the possibility that late-Archean high-K granites formed by partial melting of TTG rocks.
Taken together all these arguments advocate in favour of genesis of high-K granites by partial melting
of Archean metasediments (Fig. 3.10). The relatively small number of samples collected from
individual plutons and batholiths exposed in a vast area (ca. 2000 km2) preclude ascertaining the
nature of the different sources involved in the genesis of high-K granites. However, their mildly
peraluminous character and the absence of aluminiferous rock-forming minerals such as garnet
suggest that metapelites are not the main source for the high-K granites, with these rocks probably
formed by biotite fluid-absent melting of metagreywackes of intermediate composition. Moreover, the
high-silica content of the granitoids associated with their overall restricted chemical variability suggest
that the high-K granites are produced by a low degree of partial melting of the inferred source.

3.5.2.4 Other rock types: leucogranites, Kfs-bearing gneisses and migmatites

Leucogranites and Kfs-bearing gneisses comprise approximately 10–20 % of the bulk

crystalline basement exposed in the Bação, Bonfim and Belo Horizonte complexes. In the field, these
rock-types are typically closely associated wih the banded gneisses; the leucogranites form dikes
either parallel to or crosscutting the gneissosity, while the Kfs-bearing gneisses were emplaced as
minor bodies within the gneisses. The chemical composition of the four Kfs-bearing gneisses analysed
is highly variable. However, these rocks show an overall major and trace element composition that is
transitional between the compositions of medium- and high-K granitoids. In the (Na + Ca) – K – (Fe +
Mg + Ti) triangular diagram of Fig. 3.11, they plot in the field of the experimental melts of Watkins et
al. (2007), suggesting that Kfs-bearing gneisses might represent melt produced by partial melting of
either TTGs or medium-K banded gneisses. Alternatively, they could also be produced by a small
degree of fractional crystallization of a plagioclase-dominated assemblage from a magma having the
composition of the medium-K banded gneisses.

92
Figure 3.11 – Ternary plots comparing the composition of gneisses, granitic rocks and leucogranites of the
Bação, Bonfim and Belo Horizonte complexes with the composition of experimental glasses generated by fluid-

93
absent partial melting of granitic protoliths. In the insets, the compositions of Kfs-rich granites, Plg-rich
granitoids and banded gneisses (grey fields) are compared with the compositions of the starting materials for the
different experiments and with the composition of TTG rocks from Moyen (2011). The average composition of
Archean continental crust is from Taylor & McLennan (1985) and Condie (1997). In the main triangles, the
different fields represent the chemical variability of glasses from the experiments considered. Abbreviations:
SJ93, Skjerlie & Johnston (1993); PD&B95, Patiño Douce & Beard (1995); G95, Gardien et al. (1995); PD05,
Patiño Douce (2005); C04, Castro (2004); W07, Watkins et al. (2007).

The very low ferromagnesian content displayed by the leucogranites supports two contrasting
interpretations for their genesis. These rocks might be produced through high degree of fractional
crystallization of a hornblende-dominated assemblage from a granitic magma, and/or might represent
melts formed by fluid-present melting of the gneisses. The Eu/Eu* for these rocks varies from 0.9 to
2.8 (Fig. 3.5) indicating that in some dikes plagioclase was accumulated while extreme K 2O contents
support K-feldspar accumulation. The compositional scatter observed for the leucogranites together
with the occurrence of different populations of felsic magmas (sheets and dikes) suggest that various
processes contributed to shape their chemistry. The petrogenesis of leucogranites with similar
chemical composition from the Kianta Complex (Karelian Province, Finland) was recently discussed
in detail by Mikkola et al. (2012).

Migmatites sampled from the three complexes are polydeformed exhibiting disharmonic
folding of the banding and high volumes of leucogranitic material forming sheets, dikes and pods. In
these outcrops, it is difficult to distinguish the inferred leucosomes from the leucogranite dikes as well
as the thin leucogranitic veins from the light-coloured bands formed during sub-solidus segregation.
Therefore, the major and trace element composition of the analysed migmatites is probably
unrepresentative of the composition of the original gneiss. For this reason, no petrogenetic
interpretation of these rocks was attempted.

3.5.3 Chemical evolution of the continental crust

The whole rock chemical and U–Pb geochronological data collected for the basement
complexes of the Southern São Francisco craton allow tracking of the compositional evolution of the
continental crust during the Meso- and Neoarchean. During the Rio das Velhas I (2920–2850 Ma) and
Rio das Velhas II (2800–2760 Ma) events, the Bação, Bonfim and Belo Horizonte complexes
experienced a similar igneous and metamorphic evolution (Fig. 3.12). These periods were dominated
by the emplacement of medium-K granitoids formed by mixing between melts produced by partial
melting of metabasalts and melts derived by reworking of Paleo- and Mesoarchean TTGs. At the end
of both the Rio das Velhas I and II events, minor bodies of Kfs-bearing granitoids were also formed
(Fig. 3.12), with these intrusions most likely representing magmas produced by pure recycling of the
continental crust without any contribution from the mafic oceanic crust. At the end of the Rio das
Velhas II event, the Southern São Francisco craton was affected by amphibolite-facies regional
metamorphism deforming the granitoid basement to produce banded gneisses. The pressure and

94
temperature conditions attained during the metamorphic event are not constrained, but the occurrence
of migmatites suggests that the basement reached locally granulite-facies conditions (Alkmim &
Marshak, 1998). After the metamorphic peak, the northern (i.e. the Belo Horizonte complex) and
southern (the Bação and Bonfim complexes) portions of the craton had different magmatic evolutions.
In the southern part of the study area, Plg-rich granitoids were emplaced over ca. 20 Myrs (2790–2770
Ma) forming small bodies closely associated with the banded gneisses in the Bação complex and a ca.
40 km2 pluton (i.e. the Samambaia tonalite) in the Bonfim complex. After the production of these
magmas, a crucial change took place in the Bação and Bonfim complexes during the Mamona event
(2760–2680 Ma; Fig. 3.12): medium-K granitoids were not produced anymore, replaced by high-K
monzogranite and syenogranite (i.e. the Kfs-rich granites) with compositions of late Archean biotite
and two-mica granites. This sharp compositional change was not recorded in the Belo Horizonte
complex where, contrarily to what was inferred by Romano et al. (2013), the ca. 2750–2720 Ma
granitic batholiths (Pequi and Florestal; Fig. 3.1) have TTG-affinity. In the Belo Horizonte complex,
minor bodies of high-K granites (e.g. the Santa Luzia granite) were only emplaced at ca. 2700 Ma
(Noce et al., 1997).

Figure 3.12 – Timeline showing 207Pb/206Pb zircon ages for intrusive and volcanic rocks of the Southern São
Francisco craton. Circles, squares and diamonds indicate the Bação, Bonfim and Belo Horizonte complexes,
respectively. Abbreviations: SB, Santa Bárbara magmatic event; RdVI, Rio das Velhas I magmatic event; RdVII,
Rio das Velhas II magmatic event. Data are from Table 3.2. The age of the Santa Bárbara event is from Lana et
al. (2013).

3.5.4 The emergence of sedimentary basins in the Neoarchean

During the ca. 50 Ma separating the peak of production of medium-K magmas (ca. 2780 Ma)
and the emplacement of large volumes of high-K granites (ca. 2730 Ma), the sources involved in the
genesis of the continental crust in the Bação and Bonfim complexes changed, reflecting crustal

95
maturation. Most of the continental crust formed during the Rio das Velhas II event was produced by
recycling older orthogneisses and by partial melting of the mafic oceanic crust while, in the Mamona
event, the primary melt-forming sources were metasedimentary rocks. The main conundrum is why
metasedimentary basins were either not melting or were not producing significant volumes of magmas
during the Rio das Velhas I and II events, and why they became the main magma source during the
Mamona event. We argue that thick, sizeable sedimentary basins were missing from the rock record of
the Southern São Francisco craton before the Neoarchean and thus the scarcity of high-K granites
during the Rio das Velhas I and II events reflects the absence of fertile metasedimentary rocks at
depth. Two simple conditions have to occur to produce siliciclastic sedimentary basins: a significant
volume of continental crust must exist and this continental crust must be exposed to weathering and
erosion. The second condition is critical as the high heat flux characterizing the early evolution of the
Earth produce high temperatures that, according to various authors (Rey & Coltice, 2008; Flament et
al., 2008), strongly limit the volume of continental crust exposed to weathering and erosion. In fact,
the temperature of the mantle controls the thickness of the oceanic crust as well as the rheology of the
continental lithosphere and thus its ability to sustain topography. According to Flament et al. (2013),
the coupled effect of a lower topography and higher sea levels reduce the area of emerged land in the
Archean to less than ∼ 4 % of Earth‘s surface. Moreover, another factor limiting weathering and
erosion processes in the Neoarchean is the occurrence of thick subaqueous continental flood basalts
covering the continental crust and thus isolating it from the atmosphere and from the oceans (Arndt,
1999). The absence of sedimentary basins account for the absence of large volumes of high-K granites
during the early Archean, but does not explain why, in the Southern São Francisco craton and in many
other cratons worldwide, we witness the outburst of metasediment-derived high-K granitoids between
3.0 and 2.5 Ga. Recently, Flament et al. (2013) provided a possible explanation to this conundrum.
These authors, based on a numerical model estimating the area of emerged land as a function of
mantle temperature, continental growth and distribution of surface elevations, proposed that large
areas of the continental crust rose above the sea level during the Neoarchean.

The formation of fertile sedimentary basins is necessary but not sufficient to generate K-rich
granitoids as the sediments need to be heated up to amenable temperatures for partial melting to occur
(ca. 750–850 °C). Lana et al. (2013) proposed that felsic volcanic units in the Rio das Velhas
greenstone belt as well as granitoids formed during the Rio das Velhas II event were produced during
subduction of oceanic crust below the margin of a 3220–2900 Ma continental block. In this scenario,
the regional metamorphism affecting the continental crust at the end of the Rio das Velhas II event
was produced during continental collision triggering high-grade metamorphism and partial melting of
the lower crust to produce high-K granites. It is worth noting that sanukitoids seem to be lacking from
the rock record of the Bação, Bonfim and Belo Horizonte complexes implicating the absence of
magmas derived by metasomatized mantel peridotite and then, casting doubts on the occurrence of

96
continental subduction in this area. Although, subduction cannot be conclusively excluded, the lack of
magmas with sanukitoid-affinity in the three complexes investigated represents a peculiar feature of
the southern portion of the São Francisco craton.

3.6 CONCLUSIONS

A comprehensive dataset of whole rock major and trace element compositions from the
basement of the Southern São Francisco craton allows subdivision of the granitic rocks of the Bação,
Bonfim and Belo Horizonte complexes into two main groups: medium-K gneisses and granitoids and
high-K granites. Medium-K rocks were formed by mixing between an end-member derived by partial
melting of metamafic rocks and a component resulting from recycling older TTG rocks. On the other
hand, the chemical composition of high-K granites suggests that these rocks were produced by partial
melting of Archean metasediments. Zircon U–Pb geochronological data show that medium-K rocks
are the main component of the continental crust generated during the Rio das Velhas I and II events,
with these events taking place at 2920–2850 Ma and 2800–2760 Ma, respectively. In the Bação and
Bonfim complexes, high-K granites are the dominant rock-type that formed during the Mamona event
(2760–2680 Ma). On the contrary, granitoids formed in the Belo Horizonte complex during the 2750–
2700 Ma interval have TTG-like affinity. The chemical diversity between the northern and southern
portions of the studied area suggests a different Neoarchean magmatic evolution for the two areas. We
propose that the Neoarchean sharp transition between medium- and high-K granitoids in the Bação
and Bonfim complexes reflect the emergence of clastic sedimentary basins marking the beginning of
modern-type sedimentary cycles on Earth.

This paper was supported by CNPq (grant numbers: 401334/2012-0, 402852/2012-5 and 302633/2011-1) and FAPEMIG (grant numbers:
APQ03943 and RDP 0067-10) grants. We thank Hugo Moreira, Leonardo Alkmim and João Pedro Hippertt for the assistance provided in the
field. We thank Wilson Teixeira and Oscar Laurent for valuable and constructive comments that significantly improved the quality of this
work. Randall Parrish is acknowledged for the editorial guidance offered.

97
98
 CHAPTER 4
ARCHEAN CRUSTAL EVOLUTION IN THE SOUTHERN SÃO
FRANSISCO CRATON, BRAZIL: CONSTRAINTS FROM U-PB, LU-HF
AND O ISOTOPE ANALYSES
4

ABSTRACT

In this study we present U–Pb and Hf isotope data combined with O isotopes in zircon from
Neoarchean granitoids and gneisses of the Southern São Francisco craton in Brazil. The basement
rocks record three distinct magmatic events: Rio das Velhas I (2920–2850 Ma), Rio das Velhas II
(2800–2760 Ma) and Mamona (2750–2680 Ma). The three sampled metamorphic complexes (Bação,
Bonfim and Belo Horizonte) have distinct εHf vs. time arrays, indicating that they grew as separate
terranes. Paleoarchean crust is identified as a source which has been incorporated into younger
magmatic rocks via melting and mixing with younger juvenile material, assimilation and/or source
contamination processes. The continental crust in the Southern São Francisco craton underwent a
change in magmatic composition from medium- to high-K granitoids in the latest stages, indicating a
progressive HFSE enrichment of the sources that underwent anatexis in the different stages and
possibly shallowing of the melting depth. Oxygen isotope data shows a secular trend towards high
δ18O (up to 7.79 ‰) indicating the involvement of metasediments in the petrogenesis of the high
potassium granitoids during the Mamona event. In addition, low δ18O values (down to 2.50 ‰)
throughout the Meso- and Neoarchean emphasize the importance of meteoritic fluids in intra-crustal
magmatism. We used hafnium isotope modelling from a compilation of detrital zircon compositions to
constrain crustal growth rates and geodynamics from 3.50 to 2.65 Ga. The modelling points to a
change in geodynamic process in the Southern São Francisco craton at 2.9 Ga, from a regime
dominated by net crustal growth in the Paleoarchean to a Neoarchean regime marked by crustal
reworking. The reworking processes account for the wide variety of granitoid magmatism and are
attributed to the onset of continental collision.

C. Albert*,a, F. Farinaa,b, C. Lanaa, G. Stevensc, C. Storeyd, A. Gerdese, C. Martínez Dopicoa,f

a
Departamento de Geologia, Universidade Federal de Ouro Preto, Morro do Cruzeiro, 35400-000, Ouro Preto, MG, Brazil
b
University of Geneva, Section of Earth and Environmental Sciences, 13 Rue des Maraichers, 1205 Geneva, Switzerland
c
Department of Earth Sciences, Centre for Crustal Petrology, Stellenbosch University, Private Bag X1, Stellenbosch, Matieland 7602, South
Africa
d
School of Earth and Environmental Sciences, University of Portsmouth, Burnaby Building, Burnaby Rd, Portsmouth PO1 3QL, UK
e
Institut für Geowissenschaften, Goethe-Universität Frankfurt, Altenhöferallee 1, D-60438 Frankfurt am Main, Germany
f
Instituto de Geocronología y Geología Isotópica, CONICET, Universidad de Buenos Aires, Avenida Int. Güiraldes s/n, Ciudad
Universitaria, Buenos Aires, Argentina
* Corresponding author: Capucine Albert. E-mail: [email protected]
Article published in Lithos 266-267 (2016), 64-86
Keywords: Crustal evolution; Zircon; U-Pb dating; Lu-Hf isotopes; O isotopes; São Francisco craton

99
4.1 INTRODUCTION

Zircon has been extensively used for geochronological and geochemical studies for the past
decades. This is primarily due to its abundance in crustal rocks, its resistance to weathering, its ability
to retain complex growth zoning and its ability to be precisely dated. Combined U-Pb, Lu-Hf and O
isotope studies on zircons have been widely used to trace the evolution of the continental crust,
particularly in the Archean (e.g. Dhuime et al., 2012; Kemp et al., 2009a; Naeraa et al., 2012;
Pietranik et al., 2008; Zeh et al., 2009, 2014).

The Southern São Francisco craton (SSFC) encompasses a typical segment of

Archean/Paleoproterozoic continental crust. The basement consists of granitoid-gneiss complexes
exposed within domelike structures, in contact with Neoarchean and Paleoproterozoic supracrustal
sequences. The geology of the SSFC has been investigated for over three centuries, primarily for its
significant economic resources, and is a key location for studying Archean crustal evolution in South
America (e.g. Carneiro et al., 1997; Farina et al., 2015a, 2016; Hartmann et al., 2006; Koglin et al.,
2014; Lana et al., 2013; Machado & Carneiro, 1992; Machado et al., 1992, 1996; Moreira et al., 2016;
Noce et al., 1998, 2005; Romano et al., 2013; Teixeira & Figueiredo, 1991; Teixeira et al., 1996).
Although a substantial number of these studies have aimed to unravel the Archean magmatic history
of the SSFC, researchers are only recently starting to use a more regional approach.

Geochronological studies have shown that several Meso- and Neoarchean magmatic pulses
have led to the construction of this crustal segment, followed by an episode of crustal reworking
leading to the emplacement of large K-rich granitoids and the subsequent stabilization of the SSFC
(Lana et al., 2013; Romano et al., 2013). Farina et al. (2015a) assembled a large geochemical database
on rocks from the basement and showed that in the Neoarchean there was a pronounced change in the
composition of the crust, with a transition from medium- to high-K magmatism. These authors
proposed that this transition reflected the onset of basin deposition followed by the reworking of these
rocks in the lower crust. Previous studies, however, lack the support of isotopic data. In order to place
better constraints on the late-Archean evolution of the SSFC, we present the first set of U-Pb, Lu-Hf
and O isotope data on single zircon grains from the basement rocks. This type of combined dataset has
proven a powerful tool for deciphering geodynamic processes, particularly from times and places from
which other geological records are scarce or absent.

In this paper we present a comprehensive dataset of zircon Lu-Hf and O isotope analyses for
30 gneisses, granitoids and amphibolitic dikes from the SSFC. This allows us to address several
questions with implications for the Neoarchean crustal evolution of the Southern São Francisco craton:

(i) What were the mechanisms of crust formation?

(ii) What was the nature and importance of crustal recycling/reworking and crust-mantle
interactions?

100
4.2 GEOLOGICAL SETTING

The São Francisco craton represents one of the oldest segments of continental crust exposed in
South America. It is composed of several Archean to Paleoproterozoic blocks, thought to have
amalgamated during the 2.2–1.9 Ga Transamazonian orogenic event, and is bounded on all sides by
Neoproterozoic orogenic belts (Almeida et al., 1981; Barbosa & Sabaté, 2004; Teixeira & Figueiredo,
1991). The southern edge of the craton exposes a section of Archean and Paleoproterozoic crust,
including the Quadrilátero Ferrífero mining district that hosts world-class iron and gold deposits (Dorr,
1969; Lobato et al., 2001).

The SSFC was formed during a succession of magmatic pulses spanning from 3200 to 2600
Ma, in part concomitant with the deposition of a greenstone belt sequence, the Rio das Velhas
Supergroup. The latest magmatic event saw the emergence of a stable continental platform, enabling
the deposition of a thick Paleoproterozoic succession of volcanic, sedimentary and chemical strata,
including the Minas Supergroup (Carneiro, 1992; Lana et al., 2013; Machado et al., 1992; Romano et
al., 2013; Teixeira et al., 1996) (Fig. 4.1). The Archean basement in the SSFC is exposed within
dome-like bodies reaching several tens of kilometres across (e.g. the Bação and Bonfim complexes),
separated by elongate troughs containing polydeformed, low-grade supracrustal sequences. Together,
these complexes have a typical Archean dome-and-basin geometry (Alkmim & Marshak, 1998;
Marshak et al., 1992, 1997).

The Archean basement in the SSFC consists mainly of banded orthogneisses, intruded by
several generations of granitoid bodies, leucogranitic sheets and dikes and late pegmatitic and aplitic
veins. The gneisses display a complex pattern of amphibolite facies foliation, and locally exhibit
stromatic migmatitic features (Lana et al., 2013). Lana et al. (2013) identified three main periods of
magmatism in the SSFC, spanning between 3220 and 2770 Ma. Firstly, two gneisses cropping out in
the south of the Santa Barbara complex were dated at 3212–3210 Ma, defining the eponymous Santa
Barbara magmatic event. These ages are further supported by Sm-Nd TDM model ages obtained by
Teixeira et al. (1996) from medium- to high-grade gneisses located to the west of the SSFC (Campo
Belo complex), and by a subset of detrital zircons from the Rio das Velhas and Minas Supergroups,
suggesting the existence of fragments of Paleoarchean crust in the SSFC (Hartmann et al., 2006;
Koglin et al., 2014; Machado et al., 1992, 1996; Moreira et al., 2016). The Rio das Velhas I and II
(RVI and RVII) events are represented by gneisses and granitoids from the three main complexes
(Bação, Bonfim and Belo Horizonte) yielding ages of 2920–2850 and 2800–2760 Ma respectively
(Farina et al., 2015a and references therein). The widespread distribution of RVI and RVII rocks in
and around the SSFC, and the presence in the detrital record of a large number of ca. 2800 and 2900
Ma zircon grains (Hartmann et al., 2006; Koglin et al., 2014; Moreira et al., 2016) emphasize the
importance of the RVI and RVII as crust forming events.

101
Figure 4.1 – Geological map of the Southern São Francisco craton with sample locations. Abbreviations: F –
Florestal; M – Mamona; P – Pequi; Sa – Samambaia; SN – Souza Noschese. Inset: simplified map of the São
Francisco craton, showing the location of the exposed Archean provinces and the bordering Neoproterozoic
orogenic belts. The box indicates the Quadrilátero Ferrífero. Modified from Alkmim & Marshak (1998).

The presence in the SSFC of several mostly undeformed granitoid batholiths dated at 2780–
2770 Ma (namely the Samambaia and Caeté batholiths; Machado & Carneiro, 1992; Machado et al.,
1992) indicates that the whole crust in the SSFC experienced regional metamorphism at the end of the
RVII event. This crustal segment was later affected by one final magmatic event (namely the Mamona
event; Farina et al., 2015a), responsible for the production of voluminous granitoid batholiths, smaller
granitic domains and leucogranitic sheets and dikes, intruded in and around the older gneissic-
greenstone crust between 2750 and 2700 Ma (Machado et al., 1992; Noce et al., 1998; Romano et al.,
2013). These large granitoid batholiths represent ~ 30 % of the exposed surface of the SSFC (Romano
et al., 2013). They are typically weakly foliated bodies, mostly encountered on the topographically
higher outskirts of the Bonfim and Belo Horizonte complexes. In contrast, the older gneisses are found
in the more eroded centre of the domes (Fig. 4.1). Noce et al. (1998) and Romano et al. (2013) also
reported the occurrence in the south of the Bonfim complex of small granitoid bodies dated at 2613–
2612 Ma. Romano et al. (2013) infer that these bodies account for less than 1 % of the granitoid crust
of the SSFC. In a recent contribution, Farina et al. (2015a) provided the first detailed database of

102
major and trace element compositions for gneisses and granitoids of the SSFC. These authors
observed that the crust in the SSFC experienced a major compositional change during the Mamona
event, shifting from medium- to high-K magmatism. As opposed to what several studies had
previously assumed, Farina et al. (2015a) demonstrated that the composition of the majority of the
pre-Mamona intrusives does not match perfectly that of ―true‖ Archean TTGs as described by Moyen
(2011) (―true‖ TTGs refer to magmas that are formed via partial melting of an oceanic mafic crust,
without any involvement from the continental crust). Instead, Farina et al. (2015a) highlighted the
―hybrid‖ nature of the medium-K rocks, which display whole rock major and trace element
compositions that are intermediate between those of ―true‖ TTGs and experimental melts derived from
partial melting of TTGs, leading these authors to suggest an origin via mixing between an end-member
derived by partial melting of metamafic crust and a component resulting from the reworking of older
TTGs. On the other hand, the major and trace element compositions of the high-K granitoids argue in
favour of derivation from low-degree partial melting of immature metasediments (e.g.
metagreywacke). This transition was made possible by the continental emergence and appearance of
sizeable clastic sedimentary basins in the Neoarchean (the Rio das Velhas greenstone belt),
subsequently buried to provide the more fertile metasedimentary protoliths required for the
petrogenesis of the high-K magmas. It is important to note that the crust of the Belo Horizonte
complex does not record this transition from medium- to high-K magmatism. Indeed, the voluminous
Pequi and Florestal batholiths intruded in the west during the Mamona event have compositional
characteristics very similar to those of TTGs, testifying to a somewhat different evolution for that
complex.

4.3 ANALYTICAL TECHNIQUES

Zircon grains from 30 samples were analysed for Lu-Hf isotopes. For 9 samples, Lu-Hf was
analysed on zircons previously dated by Farina et al. (2015a). The remaining 21 samples were
analysed for U-Pb during this study. Of these, 6 new samples were analysed, 8 samples previously
studied by Romano et al. (2013) (labelled MR-) were re-dated using the same zircon mounts, and for
the remaining 7 samples (previously dated by Farina et al. (2015a)), further U-Pb analyses were done
on additional zircons picked from the same zircon separate and mounted separately for O isotope
analyses. These additional zircons were mounted together with the zircon standards Temora (Valley,
2003) and 91500 (Wiedenbeck et al., 2004) and care was taken during polishing in order to minimize
the effects of sample geometry and topography (Kita et al., 2009). For this mount, the O isotope
analyses were done prior to U-Pb and Lu-Hf analyses to ensure that the O data were not compromised
by the laser pits.

U-Pb analyses were first conducted at the Universidade Federal de Ouro Preto (UFOP) during
two analytical sessions: the first one in June-July 2013 using an Agilent 7700 Quadrupole (Q)-ICP-MS
coupled to a New Wave UP213 (λ = 213 nm) Nd:YAG laser, while the second session was done
103
between May 2014 and July 2015 using a Thermo-Scientific Element 2 Sector Field (SF) ICP-MS
coupled to a CETAC LSX-213 G2+ (λ = 213 nm) Nd:YAG laser. U-Pb data reduction was done using
the GLITTER® software package (Van Achterbergh et al., 2001). Concordia diagrams were generated
using the Isoplot/Ex 4 program (Ludwig, 2003). Following each session of U-Pb dating, Lu-Hf isotope
analyses were carried out during two analytical sessions, following the methods of Gerdes & Zeh
(2006, 2009). The first one was performed in August 2013 using a multi-collector (MC)-ICP-MS
Thermo-Finnigan Neptune system coupled to a Resonetics RESOlution M-50 193 nm Excimer laser at
Goethe Universität Frankfurt (GUF) (Germany). The second session was conducted at UFOP between
September 2014 and July 2015, using a multi-collector (MC)-ICP-MS Thermo-Scientific Neptune
Plus system coupled to a Photon Machines 193 (λ = 193 nm) ArF Excimer laser ablation system.
Oxygen isotopic compositions were determined by Secondary Ion Mass Spectrometry (SIMS) using a
Cameca® IMS1270 multi-collector SIMS at the Edinburgh Materials and Micro-Analysis Centre
(EMMAC, UK), following the methods described by Kemp et al. (2006, 2007). Laser spots for Lu-Hf
analyses were drilled ―on top‖ of the U-Pb and O spots or immediately beside, but always within the
same zircon domain characterized by CL imaging.

The instrumental parameters used for U-Pb and Lu-Hf isotope analyses during both analytical
sessions are shown in Table 4.1. A complete description of the analytical techniques is given in
Appendix A9.

Table 4.1 – Operating conditions and instrumental settings for the U-Th-Pb and Lu-Hf analyses.
U-Th-Pb Session 1 U-Th-Pb Session 2 Lu-Hf Session 1 Lu-Hf Session 2
Thermo-Scientific
Instrument Agilent 7700 Element 2 Thermo-Finnigan Neptune
Neptune+
Scan mode E-Scan E-Scan Static Static
Scanned 204, 206, 207, 208, 232, 202, 204, 206, 207, 208, 172, 173, 175, 176, 177, 172, 173, 175, 176, 177,
masses 238 232, 238 178, 179, 180 178, 179, 180
Mass
300 300 300 300
resolution
10 (204), 15 (206), 40 4 (202), 4 (204), 14 (206),
Dwell time (207), 10 (208), 10 (232), 20 (207), 10 (208), 10 0.52 s 0.52 s
15 (238) ms (232), 14 (238) ms
Integration
0.9 s 0.9 s 1s 1s
time
Background 20 s 19 s - -
Ablation time 40 s 30 s 36 s 36 s
1.0 L/min He (+ 1.0 L/min 0.5 L/min He (+ 1.0 L/min 0.6 L/min He (+ 0.9 L/min 1.3 L/min He (+ 1.0
Carrier gas
Ar) Ar) Ar, 0.007 L/min N) L/min Ar, 0.15 L/min N)
Resonetics RESOlution M-
Laser 213 nm New Wave CETAC LSX-213 G2+ Photon Machines 193
50
Spot size 30 µm 20 µm 40 µm 40-50 µm
Laser settings 10 Hz, 6-8 J/cm2 10 Hz, 3.5 J/cm2 6 Hz, 4 J/cm2 5 Hz, 3 J/cm2
Cell volume Low (teardrop) Low (teardrop) Low (Two volume cell) Low (Two volume cell)

104
4.4 RESULTS

A summary of all zircon U-Pb, Lu-Hf and O isotope data from this study is presented in Table
4.2 and 4.3 and Fig. 4.2 and 4.3. Tables with the complete results are given in Appendices A10, A11
and A12, along with Concordia diagrams, zircon CL images and a full description of the
geochronological data reduction applied for each sample in Appendix A9. All uncertainties on the U-
Pb analyses cited below are quoted at the 1ζ level. All errors on the Lu-Hf analyses are quoted at the
2ζ level.

Zircon textures together with U-Pb analyses indicate that ca. 50 % of the studied samples are
characterized by magmatic zircon crystals that do not show metamorphic overgrowth. Zircon grains
from these samples contain fine-grained oscillatory zoning patterns and yield well-defined U-Pb
Concordia or weighted mean ages. Moreover, all zircons, including those affected by different degrees
176
of Pb-loss within individual samples, have identical Hf/177Hft ratios, leading to nearly horizontal
arrays when plotted versus time (Fig. 4.2). Zircons from ca. 30 % of the samples show more complex
structures in CL images, typically displaying core-rim relationships with bright zoned or homogeneous
cores surrounded by darker rims either showing oscillatory zoning or, more rarely, structureless. The
rims display similar to slightly higher 176Hf/177Hft ratios with respect to the related cores (Fig. 4.2), and
were generally interpreted as related to a subsequent metamorphic event. The cores were either
interpreted as reflecting the crystallization age of the sample (e.g. FQ2), or as inherited zircon
xenocrysts either from the source (e.g. FQ14) or by assimilation during intrusion (e.g. FQ37). Finally,
the remaining ca. 20 % of the samples are characterized by homogeneously dark and/or structureless
zircons in CL images. A subset of zircons displays some locally recrystallized domains and/or
disruption of concentric oscillatory zoning. Ages obtained from these grains were interpreted to date a
later metamorphic event. Within individual samples, the Th/U ratios of the zircons can be highly
variable, and ca. 75 % of the grains have ratios between 0.2 and 0.7. Moreover, except for 2 samples
(FQ2 and FQ23) which display a systematic difference between high-Th/U cores and low-Th/U rims,
cores and rims have similar Th/U ratios. Based on these considerations, this ratio was not used as
criteria to distinguish different domains within a single grain nor as an argument to interpret an age.
When possible, the α-decay damage accumulation of the zircons was calculated following Murakami
et al. (1991), and used as an indicator of the degree of metamictization of the grains. Zircons showing
α-decay doses > 8 × 1015 α/mg are considered highly damaged and those analyses were discarded.
Overall, 9 out of 202 spot analyses were discarded. The α-decay doses of the remaining analyses range
from 0.29 to 7.62 × 1015 α/mg, with an average of 2.84 × 1015 α/mg, testifying to the general good
preservation of the zircon magmatic/metamorphic ages.

105
Table 4.2 – Selection of representative analyses of U-Pb, Lu-Hf and O isotopes.
Sample spot
α-decay 204 Pb/Pb
Grain (Hg Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U
238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description 15 + Pb) (%) (‰)
10 α/mg) (Ma)
Bação complex
FQ1
14_1 14 p m os 4.81 0.46 21 2.031 0.015 0.1874 0.0017 2719 15 97.1 0.00159 11 0.280988 28 -1.7 1.0
16_1 16 p m os 3.94 0.74 27 1.909 0.015 0.1869 0.0018 2715 16 100.0 0.00158 10 0.281087 39 1.8 1.4
22_1 22 p m os 3.80 0.82 46 2.075 0.019 0.1854 0.0021 2702 19 96.5 0.00167 11 0.280995 22 -1.8 0.8
39_1 39 p m os 5.08 0.36 31 1.952 0.015 0.1865 0.0021 2711 18 99.1 0.00173 11 0.281028 27 -0.4 1.0
58b_1 58 p m os 2.96 0.34 9 1.917 0.017 0.1860 0.0024 2707 21 100.0 0.00088 6 0.280981 36 -2.2 1.3
75_1 75 p m os 3.15 1.13 5 1.903 0.016 0.1872 0.0024 2718 21 100.1 0.00156 10 0.281024 26 -0.4 0.9
12a_2 15 8 p m os 0.28 0 2.073 0.015 0.1853 0.0021 2701 19 96.4 0.00111 8 0.281004 38 -1.5 1.4 7.00 0.19
15a_2 18 7 p m os 0.24 474 2.331 0.016 0.1786 0.0024 2640 22 92.0 0.00187 11 0.281007 14 -2.9 0.5 6.42 0.18
15b_2 18 7 p m os 0.24 474 2.331 0.016 0.1786 0.0024 2640 22 92.0 0.00187 11 0.281013 20 -2.6 0.7 6.42 0.18
17a_2 19 6 p m os 0.80 568 3.431 0.025 0.1665 0.0019 2522 19 74.4 0.00144 16 0.280960 24 -7.3 0.8 4.11 0.30
16a_2 21 p m os 0.66 239 2.329 0.017 0.1816 0.0020 2667 18 91.4 0.00124 9 0.280951 16 -4.2 0.6
16b_2 22 p m os 0.60 0 3.258 0.024 0.1661 0.0019 2519 19 77.5 0.00109 7 0.280966 33 -7.1 1.2
18a_2 29 p m os 0.59 529 2.136 0.018 0.1863 0.0028 2710 25 94.7 0.00175 11 0.280950 16 -3.3 0.6
18b_2 29 p m os 0.59 529 2.136 0.018 0.1863 0.0028 2710 25 94.7 0.00121 8 0.280927 19 -4.1 0.7
20a_2 30 5 p m os 0.44 150 2.816 0.020 0.1671 0.0018 2528 18 85.1 0.00094 6 0.280998 17 -5.8 0.6 6.24 0.21
21a_2 32 3 p m os 0.56 292 7.198 0.058 0.1311 0.0017 2112 23 48.0 0.00151 11 0.281022 25 -14.6 0.9 4.76 0.21
23b_2 34 4 p m os 0.04 60 2.916 0.024 0.1687 0.0023 2544 22 82.8 0.00182 13 0.280963 25 -6.7 0.9 5.28 0.22
22a_2 37 1 p m os 0.17 494 4.503 0.031 0.1248 0.0015 2026 21 76.4 0.00272 19 0.281014 17 -16.9 0.6 7.34 0.19
26b_2 43 11 p m os 0.18 581 6.552 0.050 0.1389 0.0019 2213 23 48.6 0.00205 13 0.281023 18 -12.2 0.6 5.75 0.26
24b_2 47 12 p m os 0.71 0 2.778 0.026 0.1696 0.0040 2554 38 85.2 0.00135 8 0.281014 22 -4.6 0.8 4.65 0.27
25b_2 47 12 p e os 0.71 0 2.778 0.026 0.1696 0.0040 2554 38 85.2 0.00128 8 0.281016 17 -4.5 0.6 4.65 0.27
13a_2 49 p e os 0.58 287 4.967 0.036 0.1451 0.0017 2289 20 61.4 0.00122 7 0.280998 20 -11.4 0.7

FQ2
24_1 24 p m os 2.95 0.51 13 1.7792 0.014 0.2059 0.0022 2874 17 100.0 0.00418 35 0.281020 28 3.1 1.0
27a_1 27 p m os 0.98 0.51 38 1.7739 0.017 0.2042 0.0026 2860 21 100.4 0.00167 11 0.281016 31 2.6 1.1
27b_1 27 p m os 0.98 0.51 38 1.7739 0.017 0.2042 0.0026 2860 21 100.4 0.00079 5 0.280991 19 1.7 0.7
43_1 43 p m os 0.80 0.34 28 1.7636 0.016 0.2090 0.0025 2898 19 100.0 0.00157 10 0.281006 19 3.2 0.7
45_1 45 p m os 2.56 0.45 0 1.7983 0.014 0.2030 0.0020 2850 16 100.0 0.00241 15 0.281060 29 4.0 1.0
53_1 53 p e rx 2.09 0.32 24 1.9291 0.020 0.1848 0.0032 2696 29 99.9 0.00134 10 0.281040 31 -0.4 1.1

106
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
56b_1 56 p m os 2.37 0.52 3 1.7658 0.014 0.2083 0.0024 2892 18 100.0 0.00302 18 0.281050 23 4.6 0.8
57_1 57 p m os 2.24 0.47 21 1.7874 0.015 0.2049 0.0022 2866 17 100.0 0.00245 15 0.280995 25 2.0 0.9
61_1 61 p m os 2.39 0.44 0 1.7941 0.014 0.2033 0.0022 2853 18 100.1 0.00225 14 0.281037 18 3.2 0.6
62_1 62 p m os 2.85 0.50 33 1.7864 0.015 0.2048 0.0022 2865 17 100.0 0.00196 12 0.281048 22 3.9 0.8
67_1 67 p m os 2.34 0.59 41 1.7683 0.016 0.2078 0.0027 2888 21 100.0 0.00523 34 0.281106 25 6.5 0.9
69_1 69 p m os 2.67 0.46 14 1.7902 0.014 0.2044 0.0020 2861 16 100.0 0.00315 34 0.280978 35 1.3 1.3
70_1 70 an e os 2.87 0.03 12 1.9301 0.018 0.1843 0.0031 2692 27 100.0 0.00246 44 0.281018 34 -1.3 1.2
71_1 71 p e os 4.79 0.02 0 2.8981 0.027 0.1718 0.0023 2575 22 84.9 0.00269 21 0.281035 25 -3.4 0.9
73_1 73 ov m h 2.50 0.47 9 1.7879 0.015 0.2045 0.0027 2862 21 100.0 0.00125 8 0.281018 17 2.7 0.6
74_1 74 p m os 2.35 0.49 8 1.7894 0.014 0.2049 0.0023 2866 18 99.9 0.00294 18 0.281002 24 2.3 0.8
75_1 75 p e rx 3.88 0.07 0 1.9289 0.016 0.1854 0.0021 2702 18 99.8 0.00262 24 0.281025 25 -0.8 0.9
76_1 76 peh 3.48 0.21 0 2.6047 0.022 0.1867 0.0027 2713 24 86.4 0.00242 20 0.281012 23 -1.0 0.8
32_2 15 10 p m os 0.21 163 3.0019 0.024 0.1747 0.0021 2603 20 79.5 0.00191 12 0.280985 17 -4.5 0.6 3.93 0.22
40_2 19 ov m os 0.28 53 2.1265 0.016 0.1902 0.0022 2744 19 94.2 0.00115 7 0.281038 16 0.7 0.6
41_2 26 11 p e rx 0.04 50 3.2286 0.023 0.1598 0.0018 2454 19 79.9 0.00109 7 0.281030 23 -9.4 0.8 4.80 0.20
42_2 28 12 p m os 0.36 36 2.3211 0.017 0.1880 0.0021 2724 18 90.2 0.00257 23 0.281018 23 -0.5 0.8 4.64 0.14
39_2 32 1 p e rx 0.03 186 7.0636 0.057 0.0910 0.0013 1446 28 78.7 0.00175 12 0.281060 22 -28.6 0.8 3.11 0.19
37_2 39 2 p m os 0.42 0 1.9508 0.018 0.2033 0.0033 2853 26 96.0 0.00233 16 0.281034 22 3.1 0.8 3.06 0.14
36_2 40 p m os 0.24 81 2.3268 0.018 0.1807 0.0023 2659 21 91.6 0.00267 17 0.280989 33 -3.1 1.2
35_2 44 6 p e rx 0.02 0 3.0307 0.022 0.1621 0.0019 2478 20 82.6 0.00112 7 0.281076 20 -4.2 0.7 3.77 0.21
27_2 46 p m os 0.43 0 1.9391 0.017 0.2036 0.0032 2855 25 96.3 0.00189 14 0.281019 24 2.6 0.9
29_2 48 8 p e os 0.04 95 2.7267 0.020 0.1649 0.0020 2507 21 87.4 0.00170 10 0.281019 41 -5.6 1.4 4.62 0.14

FQ13
23 23 pch 5.80 0.32 10 1.7194 0.013 0.2093 0.0017 2900 13 101.1 0.00181 11 0.280883 26 -1.2 0.9
26 26 pmh 1.33 0.49 0 1.8431 0.018 0.1965 0.0026 2797 22 100.0 0.00127 8 0.281030 27 1.6 1.0
28 28 p m os 1.98 0.67 22 1.8597 0.016 0.1957 0.0018 2790 15 99.7 0.00133 10 0.281029 33 1.4 1.2
33 33 p m os 1.30 0.56 13 1.9248 0.019 0.1854 0.0020 2702 18 99.9 0.00127 8 0.281036 30 -0.4 1.1
34 34 pmh 0.90 0.57 22 1.8415 0.016 0.1953 0.0021 2788 17 100.2 0.00088 6 0.281010 22 0.7 0.8
38 38 p m os 1.49 0.68 7 1.8400 0.015 0.1964 0.0018 2796 15 100.0 0.00143 9 0.281059 29 2.7 1.0
39 39 p m os 0.87 0.52 9 2.0158 0.019 0.1832 0.0022 2682 20 98.2 0.00100 8 0.281046 25 -0.5 0.9
49 49 p m os 2.24 0.34 5 1.8515 0.015 0.1947 0.0018 2783 15 100.0 0.00082 8 0.281022 23 1.0 0.8
50 50 p m os 5.44 0.41 27 1.8452 0.016 0.1947 0.0016 2783 14 100.2 0.00079 10 0.281027 21 1.2 0.8
57 57 pmh 3.00 0.11 3 1.8375 0.015 0.1969 0.0016 2800 14 100.0 0.00081 7 0.280945 15 -1.3 0.6

107
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
58 58 pmh 3.83 0.09 22 1.8238 0.015 0.1968 0.0016 2800 14 100.4 0.00084 7 0.280909 21 -2.6 0.7
59 59 p m os 1.11 0.58 26 1.8363 0.016 0.1938 0.0020 2775 17 100.6 0.00118 8 0.281044 30 1.6 1.1
61 61 p m os 1.49 0.69 19 1.8431 0.015 0.1958 0.0020 2791 16 100.1 0.00189 13 0.281075 31 3.1 1.1
68 68 pmh 1.67 0.67 9 1.8433 0.015 0.1959 0.0018 2792 15 100.0 0.00129 9 0.281049 27 2.2 1.0
72 72 p m os 0.65 0.39 0 1.8428 0.020 0.1958 0.0027 2791 23 100.1 0.00140 14 0.281025 34 1.3 1.2
83 83 p m os 1.25 0.56 7 1.7996 0.015 0.1964 0.0020 2796 17 101.1 0.00132 8 0.281052 17 2.4 0.6
84 84 p m os 1.33 0.64 14 1.8645 0.016 0.1930 0.0019 2768 16 100.0 0.00100 8 0.281032 23 1.0 0.8

FQ17
10aI_1 10_a ov m os 5.40 0.06 23 1.9114 0.017 0.1883 0.0018 2727 16 99.7 0.00125 9 0.281062 30 1.1 1.1
10bI_1 10_a ov m os 5.40 0.06 23 1.9114 0.017 0.1883 0.0018 2727 16 99.7 0.00102 9 0.281056 28 0.9 1.0
11I_1 11_a eq m os 5.76 0.05 5 1.9041 0.015 0.1893 0.0019 2736 17 99.7 0.00082 5 0.281060 27 1.3 1.0
14I_1 14_a p m os 5.13 0.05 9 1.9076 0.015 0.1874 0.0018 2719 16 99.9 0.00067 4 0.281015 22 -0.7 0.8
15I_1 15_a p m os 0.46 0.33 0 1.9055 0.030 0.1884 0.0043 2728 37 99.8 0.00044 3 0.280999 23 -1.1 0.8
19II_1 19_b p e os 4.00 0.07 12 1.9006 0.015 0.1881 0.0018 2726 16 100.0 0.00036 2 0.281035 18 0.2 0.7
23I_1 23_a p m os 6.40 0.43 16 1.8891 0.015 0.1896 0.0017 2739 15 100.0 0.00061 4 0.280987 19 -1.2 0.7
30I_1 30_a p m os 5.40 0.35 9 1.8539 0.014 0.1941 0.0018 2778 15 100.1 0.00119 9 0.281014 22 0.6 0.8
32I_1 32_a p m os 1.48 0.06 6 1.8577 0.016 0.1944 0.0021 2780 17 99.9 0.00091 10 0.280963 17 -1.1 0.6
33I_1 33_a p m os 4.21 1.09 9 1.8560 0.015 0.1942 0.0019 2778 16 100.0 0.00110 11 0.280989 25 -0.3 0.9
33II_1 33_b p m os 4.55 0.20 20 1.9454 0.018 0.1850 0.0025 2698 22 99.5 0.00085 6 0.281018 29 -1.1 1.0
35II_1 35_b pch 7.34 0.41 48 1.7840 0.015 0.2055 0.0022 2871 17 100.0 0.00166 14 0.280913 24 -0.8 0.9
42I_1 42_a pmh 3.64 0.25 59 1.8551 0.014 0.1946 0.0018 2782 15 99.9 0.00055 4 0.280979 20 -0.5 0.7
43aII_1 43_b pmh 3.10 0.16 25 1.9035 0.017 0.1874 0.0024 2720 21 100.0 0.00059 4 0.280999 21 -1.3 0.7
43bII_1 43_b pmh 3.10 0.16 25 1.9035 0.017 0.1874 0.0024 2720 21 100.0 0.00081 6 0.281006 15 -1.0 0.6
50II_1 50_b p c os 4.54 0.46 58 1.8170 0.015 0.1998 0.0024 2825 19 100.0 0.00094 6 0.280934 21 -1.1 0.7
51II_1 51_b peh 7.38 0.54 24 1.8810 0.016 0.1906 0.0024 2747 20 100.0 0.00057 4 0.281000 18 -0.6 0.6
52II_1 52_b p m os 3.66 0.29 43 1.8892 0.016 0.1900 0.0021 2742 18 99.9 0.00076 5 0.281012 21 -0.3 0.7
56II_1 56_b p m os 4.33 0.36 3 1.8983 0.017 0.1910 0.0023 2751 19 99.5 0.00148 10 0.281045 22 1.1 0.8
61II_1 61_b pch 2.28 0.33 46 1.7882 0.019 0.2039 0.0039 2858 31 100.1 0.00121 8 0.281032 19 3.2 0.7
48_2 12 8 p m os 0.32 38 1.9276 0.015 0.1847 0.0020 2695 18 100.0 0.00071 6 0.281042 29 -0.3 1.0 5.76 0.15
47_2 14 7 pmh 0.51 310 3.3735 0.024 0.1841 0.0020 2690 18 70.0 0.00175 21 0.280988 18 -2.4 0.6 4.00 0.22
45_2 15 5 p e os 0.24 0 2.0449 0.016 0.1859 0.0030 2706 26 97.0 0.00113 7 0.281015 18 -1.0 0.6 6.29 0.19
46_2 16 6 p m os 0.39 160 2.4549 0.018 0.1755 0.0019 2611 18 90.1 0.00072 6 0.280991 16 -4.1 0.6 5.46 0.21

108
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
44_2 17 p m os 0.63 83 1.9270 0.014 0.1865 0.0021 2711 18 99.7 0.00143 11 0.280985 18 -2.0 0.7
43_2 18 p e os 0.04 157 2.0333 0.014 0.1868 0.0021 2714 18 97.0 0.00137 10 0.280998 18 -1.4 0.6
52_2 19 p m os 0.36 153 1.8932 0.014 0.1870 0.0020 2716 17 100.4 0.00123 11 0.281003 23 -1.2 0.8
50_2 21 4 p m os 0.25 378 1.9737 0.016 0.1871 0.0034 2717 30 98.4 0.00068 4 0.281013 19 -0.8 0.7 5.56 0.15
51_2 21 4 p e os 0.25 378 1.9737 0.016 0.1871 0.0034 2717 30 98.4 0.00146 10 0.281007 21 -1.1 0.8 5.56 0.15
54_2 30 p m os 0.24 441 1.9255 0.015 0.1869 0.0027 2715 24 99.6 0.00104 7 0.280999 20 -1.4 0.7
56_2 35 3 p m os 0.13 619 2.0030 0.016 0.1961 0.0024 2794 20 96.0 0.00196 12 0.280988 21 0.1 0.7 4.88 0.20
55_2 36 p m os 0.35 193 2.0845 0.017 0.1859 0.0028 2706 24 96.0 0.00086 6 0.280983 34 -2.2 1.2
57_2 38 p m os 0.41 190 1.8980 0.014 0.1945 0.0023 2781 19 98.9 0.00098 6 0.281012 25 0.6 0.9
58_2 39 p e os 0.03 62 1.9543 0.014 0.1861 0.0021 2708 18 99.1 0.00066 4 0.281019 25 -0.8 0.9
59_2 41 10 p m os 0.29 78 2.1471 0.016 0.1979 0.0024 2809 20 92.2 0.00124 10 0.280919 19 -2.0 0.7 5.98 0.15
63_2 42 11 p m os 0.25 10 1.9787 0.016 0.1869 0.0026 2715 23 98.4 0.00089 7 0.280985 18 -1.9 0.7 6.10 0.19
65_2 49 pmh 0.21 358 2.7349 0.020 0.1865 0.0027 2712 23 81.7 0.00252 16 0.280950 19 -3.1 0.7
64_2 50 2 p e os 0.06 6 2.5948 0.019 0.1778 0.0024 2632 22 86.6 0.00090 6 0.280981 19 -4.0 0.7 5.01 0.23
66_2 52 1 p m os 0.04 46 2.7391 0.020 0.1844 0.0022 2692 20 82.1 0.00075 5 0.280995 15 -2.1 0.5 5.66 0.18
67_2 53 p m os 0.22 571 1.9676 0.015 0.1978 0.0028 2808 23 96.6 0.00294 18 0.281002 22 0.9 0.8
68_2 55 p m os 0.17 0 2.0438 0.015 0.1857 0.0022 2705 20 97.0 0.00080 7 0.280994 25 -1.8 0.9
69_2 58 12 pmh 0.11 690 2.4740 0.022 0.1811 0.0035 2663 31 88.4 0.00183 12 0.281011 13 -2.2 0.5 4.49 0.18
70_2 65 pmh 0.68 128 2.3516 0.017 0.2004 0.0025 2829 20 86.9 0.00066 5 0.280970 16 0.3 0.6
71_2 66 p e os 0.15 0 2.5148 0.018 0.1778 0.0024 2632 22 88.3 0.00087 6 0.280974 18 -4.2 0.6
60_2 68 9 p m os 0.21 80 3.4464 0.025 0.1700 0.0022 2558 21 73.0 0.00079 7 0.280985 21 -5.6 0.8 5.59 0.19
61_2 71 p m os 0.32 377 1.9187 0.015 0.1968 0.0028 2800 23 98.0 0.00097 8 0.280999 17 0.6 0.6

Bonfim Complex
FQ29
20_2a 15 p m os 0.31 24 2.0304 0.016 0.1926 0.0021 2764 18 96.1 0.00089 7 0.280883 15 -4.3 0.6
19_2a 16 pmh 0.23 46 1.8707 0.015 0.1935 0.0021 2772 17 99.8 0.00072 5 0.280881 15 -4.2 0.5
21_2a 17 p n os 0.21 0 1.8834 0.014 0.1949 0.0025 2784 20 99.2 0.00121 8 0.280917 21 -2.7 0.8
23_2a 21 p m os 0.25 71 1.7063 0.013 0.1967 0.0025 2799 20 103.6 0.00134 9 0.280889 12 -3.3 0.4
24_2a 21 p m os 0.25 71 1.7063 0.013 0.1967 0.0025 2799 20 103.6 0.00084 5 0.280895 16 -3.1 0.6
49_2a 31 p m os 0.42 10 2.1532 0.016 0.1943 0.0025 2779 21 93.2 0.00153 11 0.280902 18 -3.3 0.6
25_2a 32 p m os 0.30 80 1.8770 0.015 0.1917 0.0022 2757 18 99.9 0.00058 4 0.280907 13 -3.7 0.5
26_2a 33 p m os 0.21 75 2.1377 0.016 0.1895 0.0024 2738 20 94.4 0.00137 9 0.280894 17 -4.6 0.6
27_2a 33 p m os 0.21 75 2.1377 0.016 0.1895 0.0024 2738 20 94.4 0.00169 15 0.280902 19 -4.3 0.7

109
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
29_2a 34 p m os 0.27 7 1.8559 0.015 0.1945 0.0021 2781 18 100.0 0.00129 8 0.280900 16 -3.4 0.6
28_2a 35 p m os 0.22 0 1.8534 0.014 0.1947 0.0022 2782 18 100.0 0.00173 15 0.280901 17 -3.3 0.6
30_2a 36 p m os 0.35 49 1.8528 0.015 0.1922 0.0021 2761 17 100.4 0.00138 8 0.280888 17 -4.3 0.6
31_2a 48 p m os 0.56 36 1.8794 0.014 0.1929 0.0024 2767 20 99.7 0.00165 12 0.280902 22 -3.6 0.8
36_2a 50 p m os 0.46 112 1.8526 0.016 0.1959 0.0027 2792 23 99.8 0.00111 8 0.280919 17 -2.4 0.6
34_2a 52 p e os 0.26 0 2.1081 0.018 0.1938 0.0028 2774 23 94.2 0.00078 5 0.280902 14 -3.5 0.5
38_2a 69 p m os 0.42 32 2.5144 0.023 0.1939 0.0039 2776 33 86.6 0.00124 10 0.280901 15 -3.5 0.5
39_2a 70 p m os 0.28 0 2.0041 0.017 0.1926 0.0029 2765 24 96.7 0.00136 9 0.280897 16 -3.8 0.6
72_2b 11 p e os 0.19 181 2.3455 0.014 0.1907 0.0020 2748 17 89.0 0.00130 8 0.280922 24 -3.4 0.8
73_2b 16 p e os 0.17 481 2.3599 0.014 0.1940 0.0021 2776 18 88.0 0.00082 6 0.280922 18 -2.7 0.6
74_2b 17 p m os 0.19 132 1.8802 0.013 0.1967 0.0025 2799 21 99.0 0.00142 9 0.280921 22 -2.2 0.8
78_2b 19 p m os 0.19 220 2.0579 0.013 0.1953 0.0020 2787 17 94.8 0.00133 8 0.280917 27 -2.6 0.9
77_2b 20 p m os 0.20 0 1.9227 0.012 0.1954 0.0020 2788 16 98.1 0.00169 10 0.280943 26 -1.7 0.9
76_2b 21 p m os 0.21 106 2.4086 0.015 0.1903 0.0019 2745 16 87.7 0.00164 11 0.280915 22 -3.7 0.8
75_2b 26 p m os 0.17 357 1.8789 0.014 0.1958 0.0023 2792 19 99.1 0.00149 9 0.280928 20 -2.1 0.7
12_2b 32 p m os 0.21 0 2.0933 0.013 0.1977 0.0021 2808 17 93.5 0.00224 17 0.280949 22 -1.0 0.8
11_2b 34 p e os 0.18 365 2.6075 0.016 0.1910 0.0020 2750 17 83.2 0.00159 10 0.280908 17 -3.8 0.6
80_2b 35 8 p m os 0.26 289 2.5051 0.016 0.1915 0.0020 2755 17 85.3 0.00112 7 0.280910 16 -3.6 0.6 4.95 0.23
16_2b 47 7 p e os 0.15 0 2.0930 0.013 0.1948 0.0020 2783 17 94.1 0.00145 9 0.280893 19 -3.6 0.7 5.29 0.25
15_2b 49 p m os 0.26 133 2.6635 0.017 0.1904 0.0021 2746 18 82.2 0.00147 9 0.280906 16 -4.0 0.6
13_2b 51 9 p e os 0.19 312 2.1952 0.014 0.1981 0.0022 2811 18 91.0 0.00143 9 0.280905 16 -2.5 0.6 3.46 0.15
14_2b 53 10 p m os 0.23 226 2.0130 0.014 0.1947 0.0021 2782 18 95.9 0.00111 8 0.280886 17 -3.8 0.6 4.90 0.14
21_2b 59 11 p m os 0.15 324 1.9404 0.013 0.1966 0.0021 2798 17 97.5 0.00118 9 0.280915 18 -2.4 0.6 6.03 0.21
20_2b 62 14 p m os 0.12 177 1.9064 0.013 0.1966 0.0021 2799 17 98.3 0.00060 8 0.280972 16 -0.4 0.6 4.73 0.18
19_2b 64 p m os 0.19 0 2.1561 0.014 0.1987 0.0024 2815 19 91.8 0.00152 11 0.280924 23 -1.7 0.8
18_2b 66 3 p m os 0.15 0 2.1031 0.014 0.1959 0.0021 2792 18 93.6 0.00124 11 0.280923 17 -2.3 0.6 5.23 0.26
27_2b 69 5 p m os 0.24 103 1.8554 0.013 0.1949 0.0022 2784 18 100.0 0.00142 9 0.280900 20 -3.3 0.7 4.86 0.15
28_2b 69 5 p m os 0.24 103 1.8554 0.013 0.1949 0.0022 2784 18 100.0 0.00160 12 0.280946 16 -1.7 0.6 4.86 0.15
25_2b 71 6 p m os 0.20 354 2.8988 0.020 0.1810 0.0020 2662 18 79.8 0.00125 8 0.280909 20 -5.8 0.7 5.80 0.14
26_2b 71 6 p m os 0.20 354 2.8988 0.020 0.1810 0.0020 2662 18 79.8 0.00100 6 0.280894 20 -6.4 0.7 5.80 0.14
24_2b 74 13 p m os 0.20 0 2.1463 0.014 0.1943 0.0021 2779 17 92.9 0.00118 8 0.280922 17 -2.6 0.6 5.04 0.22
22_2b 76 p m os 0.14 158 2.1085 0.014 0.1967 0.0022 2799 18 93.3 0.00134 8 0.280915 18 -2.4 0.6
23_2b 77 12 p m os 0.18 144 1.7882 0.012 0.1990 0.0022 2818 18 101.0 0.00114 9 0.280907 19 -2.2 0.7 5.11 0.18

110
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U
238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
FQ41
14 14 p m os 0.25 134 2.0479 0.017 0.1988 0.0025 2816 19 94.7 0.00128 8 0.280883 18 -3.1 0.6
13 15 p m os 0.32 51 1.8742 0.013 0.2029 0.0023 2850 18 98.1 0.00099 6 0.280890 21 -2.1 0.7
15 17 p e os 0.29 133 2.4565 0.021 0.2009 0.0024 2834 19 86.5 0.00138 10 0.280907 20 -1.9 0.7
16 17 p e os 0.29 133 2.4565 0.021 0.2009 0.0024 2834 19 86.5 0.00116 7 0.280912 16 -1.7 0.6
12 18 p m os 0.39 63 2.1878 0.017 0.2034 0.0026 2853 20 91.0 0.00134 10 0.280908 19 -1.4 0.7
19 21 p m os 0.31 58 1.8235 0.014 0.2029 0.0023 2850 18 99.4 0.00111 7 0.280903 17 -1.6 0.6
17 23 p m os 0.34 138 2.0642 0.017 0.1957 0.0024 2791 19 94.8 0.00124 8 0.280885 25 -3.7 0.9
18 24 p m os 0.38 53 2.2154 0.017 0.1981 0.0021 2810 17 91.3 0.00137 8 0.280919 19 -2.0 0.7
20 25 p m os 0.27 241 2.4523 0.021 0.2039 0.0025 2857 20 86.1 0.00191 14 0.280899 14 -1.6 0.5
23 32 p m os 0.38 17 2.4668 0.019 0.1955 0.0021 2789 18 87.2 0.00124 12 0.280908 27 -2.9 0.9
22 39 p m os 0.29 43 1.9341 0.015 0.2029 0.0023 2850 18 96.6 0.00102 6 0.280909 25 -1.4 0.9
25 58 p m os 0.39 55 1.8561 0.014 0.2024 0.0024 2846 18 98.6 0.00151 11 0.280894 21 -2.1 0.7
26 62 p m os 0.25 88 1.8318 0.014 0.1970 0.0028 2802 22 100.1 0.00150 11 0.280886 22 -3.4 0.8
27 62 p m os 0.25 88 1.8318 0.014 0.1970 0.0028 2802 22 100.1 0.00106 11 0.280906 18 -2.7 0.6
28 63 eq m os 0.40 51 2.0871 0.016 0.2020 0.0024 2842 19 93.3 0.00120 8 0.280907 21 -1.7 0.7
29 66 p e os 0.31 46 2.0283 0.015 0.2024 0.0029 2846 23 94.5 0.00105 7 0.280890 19 -2.2 0.7
30 67 eq e os 0.24 10 1.8288 0.016 0.2055 0.0026 2870 21 98.8 0.00124 8 0.280867 17 -2.4 0.6

FQ52
11_2a 13_7 pmh 0.26 0 1.8873 0.017 0.1903 0.0030 2745 26 99.9 0.00097 7 0.280995 19 -0.8 0.7
12_2a 13_7 pmh 0.26 0 1.8873 0.017 0.1903 0.0030 2745 26 99.9 0.00102 10 0.280981 16 -1.3 0.6
13_2a 16_23 p m os 0.35 47 2.0083 0.015 0.1895 0.0023 2738 19 97.2 0.00120 7 0.281002 19 -0.8 0.7
15_2a 17_7 pmh 0.22 118 2.0043 0.021 0.1908 0.0032 2749 27 97.0 0.00060 4 0.280986 20 -1.1 0.7
16_2a 19_23 p m os 0.21 5 1.8867 0.014 0.1905 0.0021 2747 18 99.9 0.00070 7 0.280983 18 -1.3 0.6
17_2a 21_23 p m os 0.16 71 1.8939 0.015 0.1856 0.0020 2704 18 100.6 0.00079 5 0.280968 18 -2.7 0.6
19_2a 24_23 p m os 0.26 0 1.9875 0.015 0.1887 0.0021 2731 18 97.8 0.00069 5 0.280966 20 -2.2 0.7
26_2a 44_23 p e os 0.17 28 1.9433 0.017 0.1904 0.0024 2746 20 98.5 0.00083 5 0.280977 23 -1.4 0.8
20_2a 58_7 p m os 0.26 41 1.9175 0.016 0.1876 0.0033 2721 28 99.7 0.00061 4 0.280996 20 -1.4 0.7
21_2a 61_7 p m os 0.35 0 1.9482 0.018 0.1892 0.0036 2735 31 98.6 0.00073 5 0.281001 24 -0.8 0.8
22_2a 64_7 pmh 0.17 7 1.8905 0.023 0.1900 0.0040 2742 34 99.9 0.00032 2 0.280970 19 -1.8 0.7
27_2a 83_7 p m os 0.24 27 1.9622 0.020 0.1891 0.0041 2735 35 98.3 0.00086 10 0.281005 19 -0.7 0.7
52_2b 18 6 p m os 0.67 193 2.0731 0.017 0.1873 0.0020 2718 17 96.0 0.00186 17 0.281015 29 -0.7 1.0 10.88 0.35
53_2b 18 6 p m os 0.67 193 2.0731 0.017 0.1873 0.0020 2718 17 96.0 0.00172 11 0.281031 21 -0.2 0.7 10.88 0.35

111
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
54_2b 20 p m os 0.74 133 2.1038 0.019 0.1872 0.0026 2718 23 95.4 0.00118 10 0.281018 31 -0.6 1.1
56_2b 27 3 p m os 0.48 62 2.3487 0.018 0.1725 0.0019 2582 18 93.0 0.00176 11 0.281018 21 -3.8 0.7 4.48 0.29
55_2b 29 4 pmh 0.48 235 6.8733 0.054 0.1062 0.0012 1735 20 67.7 0.00150 24 0.281050 40 -22.3 1.4 2.88 0.29
58_2b 35 9 p m os 0.40 190 4.2757 0.035 0.1574 0.0017 2428 18 64.8 0.00142 12 0.281025 20 -7.2 0.7 4.17 0.31
59_2b 48 8 p m os 0.44 52 2.1764 0.019 0.1819 0.0023 2670 21 94.7 0.00066 4 0.280998 14 -2.5 0.5 6.73 0.34
60_2b 48 8 p m os 0.44 52 2.1764 0.019 0.1819 0.0023 2670 21 94.7 0.00121 10 0.281002 20 -2.3 0.7 6.73 0.34
61_2b 50 p m os 0.61 232 3.3519 0.026 0.1556 0.0018 2408 19 79.3 0.00165 14 0.281024 22 -7.7 0.8
63_2b 52 p m rx 0.42 0 2.3087 0.021 0.1782 0.0024 2636 22 92.6 0.00134 9 0.281008 19 -2.9 0.7
64_2b 53 p m rx 0.18 192 3.4598 0.028 0.1554 0.0020 2406 21 77.6 0.00106 10 0.281012 24 -8.2 0.9
62_2b 54 14 p m os 0.62 265 4.4823 0.036 0.1464 0.0017 2305 19 66.6 0.00095 8 0.281085 22 -7.9 0.8 4.33 0.40
65_2b 57 12 p m os 0.56 457 2.9284 0.025 0.1696 0.0025 2553 24 82.4 0.00166 22 0.281042 24 -3.6 0.9 4.35 0.36
66_2b 62 p e os 0.76 503 3.1377 0.025 0.1666 0.0019 2524 19 79.4 0.00189 12 0.281000 20 -5.8 0.7
48_2b 73 2 p m os 0.47 0 2.0209 0.017 0.1777 0.0025 2631 23 99.1 0.00177 15 0.281009 18 -3.0 0.6 6.33 0.44
49_2b 73 2 p m os 0.47 0 2.0209 0.017 0.1777 0.0025 2631 23 99.1 0.00070 5 0.280985 17 -3.9 0.6 6.33 0.44
51_2b 75 1 p m os 0.49 110 4.7192 0.039 0.1382 0.0020 2205 25 67.5 0.00207 20 0.281037 24 -11.9 0.9 4.46 0.39

MR14A
48 48 pmh 1.28 0.29 76 1.9897 0.017 0.1865 0.0018 2711 16 98.2 0.00048 3 0.280937 24 -3.7 0.8
49 49 fch 1.47 0.43 120 2.0806 0.018 0.2027 0.0021 2848 17 93.3 0.00054 4 0.280952 26 0.1 0.9
54 54 f e os 2.82 0.28 54 1.9062 0.015 0.1869 0.0017 2715 15 100.1 0.00049 3 0.280912 21 -4.5 0.7
56 56 fmh 3.54 0.29 34 2.0197 0.016 0.1860 0.0017 2707 15 97.6 0.00067 5 0.280975 25 -2.4 0.9
58 58 f m os 1.89 0.29 40 1.9022 0.015 0.1867 0.0017 2714 15 100.2 0.00052 3 0.280956 33 -3.0 1.2
70 70 p m os 3.49 0.40 100 1.9424 0.020 0.1888 0.0029 2731 25 98.9 0.00067 6 0.280970 29 -2.1 1.0
71 71 f c os 2.02 0.42 184 1.9489 0.018 0.2248 0.0027 3016 19 93.0 0.00111 7 0.280910 24 2.5 0.9
74 74 bmh 1.87 0.37 44 1.9268 0.017 0.1859 0.0029 2706 25 99.8 0.00123 9 0.280959 30 -3.0 1.1
75 75 eq m h 1.26 0.36 36 1.9015 0.016 0.1881 0.0020 2726 18 100.0 0.00056 4 0.280975 33 -2.0 1.2

Belo Horizonte complex

FQ60
15_2a 145 pmh 0.55 118 1.9164 0.015 0.1866 0.0022 2713 19 99.9 0.00070 5 0.280967 18 -2.6 0.7
19_2a 153 pmh 0.31 312 2.4612 0.020 0.1832 0.0026 2682 23 89.4 0.00071 4 0.280968 20 -3.2 0.7
20_2a 159 pmh 0.29 20 2.2232 0.017 0.1857 0.0023 2705 20 93.3 0.00084 5 0.280973 14 -2.5 0.5
21_2a 159 pmh 0.29 20 2.2232 0.017 0.1857 0.0023 2705 20 93.3 0.00082 6 0.280948 18 -3.4 0.7

112
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
22_2a 160 pmh 0.43 0 3.3584 0.029 0.1676 0.0021 2534 20 80.2 0.00056 4 0.280968 28 -6.7 1.0
24_2a 165 p m os 0.33 154 1.9724 0.016 0.1735 0.0026 2592 23 101.1 0.00083 5 0.281000 29 -4.2 1.0
25_2a 166 p m os 0.10 198 2.8405 0.023 0.1668 0.0023 2526 23 86.7 0.00084 7 0.280952 22 -7.5 0.8
16_2a 172 pmh 0.54 0 1.9085 0.016 0.1883 0.0026 2728 23 99.8 0.00051 4 0.280964 18 -2.3 0.6
71_2b 13 13 p m os 0.18 0 6.2996 0.049 0.1380 0.0016 2202 19 51.3 0.00140 14 0.281013 17 -12.9 0.6 2.86 0.33
69_2b 15 peh 0.56 135 2.1648 0.017 0.1877 0.0022 2722 19 93.8 0.00073 5 0.281096 33 2.2 1.2
72_2b 20 p m os 0.34 29 1.9434 0.016 0.1858 0.0023 2705 20 99.4 0.00100 6 0.280932 20 -4.0 0.7
73_2b 30 1 f m os 0.22 337 2.8275 0.025 0.1884 0.0030 2728 26 79.3 0.00155 16 0.281011 22 -0.7 0.8 3.12 0.27
74_2b 34 5 p m os 0.30 577 4.3016 0.034 0.1864 0.0021 2710 18 53.7 0.00084 6 0.280957 25 -3.0 0.9 3.09 0.33
77_2b 43 14 f m os 0.30 144 1.9481 0.016 0.1859 0.0021 2706 19 99.3 0.00042 3 0.280921 20 -4.4 0.7 7.04 0.33
76_2b 44 15 f m os 0.29 22 1.9203 0.016 0.1855 0.0022 2703 19 100.0 0.00054 3 0.280924 22 -4.3 0.8 6.60 0.29
78_2b 45 7 ov c h 0.12 91 1.8973 0.016 0.1961 0.0024 2794 20 98.6 0.00025 3 0.280997 19 0.4 0.7 5.41 0.40
79_2b 46 8 ov c h 0.20 299 1.8430 0.016 0.1940 0.0026 2776 22 100.4 0.00028 2 0.280981 17 -0.6 0.6 5.34 0.37
11_2b 48 p m os 0.25 731 2.6782 0.022 0.1891 0.0024 2734 21 82.2 0.00118 7 0.280920 18 -3.7 0.6
80_2b 50 4 p m os 0.25 310 3.8220 0.031 0.1723 0.0020 2580 19 66.1 0.00125 8 0.281000 25 -4.5 0.9 5.01 0.29
14_2b 54 9 p m os 0.22 266 5.0446 0.040 0.1547 0.0019 2399 20 56.0 0.00196 15 0.280985 17 -9.3 0.6 3.84 0.37

FQ74
18_2a 16 pmh 0.53 0 2.2873 0.018 0.1784 0.0022 2638 21 93.4 0.00066 5 0.280994 19 -3.4 0.7
23_2a 17 bmh 0.69 197 2.4164 0.018 0.1620 0.0020 2477 18 94.5 0.00132 16 0.280942 15 -9.0 0.5
21_2a 18 bmh 0.86 112 1.9859 0.016 0.1806 0.0021 2658 19 99.4 0.00077 11 0.280961 16 -4.1 0.6
22_2a 18 bmh 0.86 112 1.9859 0.016 0.1806 0.0021 2658 19 99.4 0.00071 8 0.280957 18 -4.2 0.6
25_2a 20 pmh 0.67 35 2.0394 0.017 0.1797 0.0022 2650 20 98.3 0.00100 7 0.280974 18 -3.8 0.6
26_2a 20 peh 0.67 35 2.0394 0.017 0.1797 0.0022 2650 20 98.3 0.00070 5 0.280997 16 -3.0 0.6
19_2a 34 pmh 0.72 7 2.1548 0.018 0.1777 0.0022 2632 21 96.2 0.00122 13 0.280975 22 -4.2 0.8
20_2a 34 pmh 0.72 7 2.1548 0.018 0.1777 0.0022 2632 21 96.2 0.00086 8 0.280963 20 -4.6 0.7
16_2a 35 b m os 0.75 7 1.9880 0.015 0.1781 0.0022 2636 20 99.8 0.00120 9 0.281008 26 -2.9 0.9
24_2a 39 p e os 0.73 4 2.1173 0.017 0.1791 0.0022 2644 20 96.7 0.00099 17 0.280961 23 -4.4 0.8
27_2a 43 beh 0.63 138 2.0353 0.016 0.1752 0.0020 2608 18 99.3 0.00075 6 0.280969 20 -5.0 0.7
28_2a 43 beh 0.63 138 2.0353 0.016 0.1752 0.0020 2608 18 99.3 0.00155 16 0.280966 22 -5.1 0.8
29_2a 55 pmh 0.88 0 2.1397 0.024 0.1772 0.0035 2627 33 96.6 0.00141 20 0.280994 24 -3.6 0.9
31_2a 55 pmh 0.88 0 2.1397 0.024 0.1772 0.0035 2627 33 96.6 0.00111 13 0.280960 18 -4.8 0.6
32_2a 74 pmh 0.61 68 2.1565 0.020 0.1790 0.0031 2644 29 95.9 0.00119 9 0.280956 22 -4.6 0.8
33_2a 74 pmh 0.61 68 2.1565 0.020 0.1790 0.0031 2644 29 95.9 0.00141 13 0.280954 20 -4.7 0.7

113
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U 238
U/206Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description Pb) (%) (‰)
1015 α/mg) (Ma)
38_2a 96 bmh 0.59 23 1.9935 0.016 0.1794 0.0026 2647 23 99.4 0.00115 14 0.280964 21 -4.2 0.7
39_2a 96 bmh 0.59 23 1.9935 0.016 0.1794 0.0026 2647 23 99.4 0.00089 9 0.280959 20 -4.4 0.7
33_2b 12 bmh 0.56 18 2.0267 0.022 0.1764 0.0030 2619 28 99.3 0.00098 7 0.280976 24 -4.5 0.9
32_2b 14 b e os 0.32 66 3.0582 0.024 0.1597 0.0017 2452 18 82.9 0.00316 22 0.281004 22 -7.4 0.8
31_2b 16 p m os 0.40 55 2.4930 0.020 0.1754 0.0020 2610 19 89.3 0.00134 9 0.280947 20 -5.7 0.7
40_2b 19 14 p m os 0.16 407 4.1825 0.036 0.1869 0.0024 2715 21 55.4 0.00203 13 0.280980 15 -2.1 0.6 5.46 0.36
39_2b 26 p e os 0.51 361 2.6748 0.021 0.1709 0.0019 2566 18 86.8 0.00179 21 0.280968 19 -6.0 0.7
38_2b 27 p m rx 0.41 0 2.0689 0.016 0.1752 0.0019 2608 18 98.6 0.00091 6 0.281022 11 -3.1 0.4
37_2b 30 p m rx 1.08 0 2.6683 0.022 0.1757 0.0020 2613 18 85.7 0.00220 13 0.280991 13 -4.1 0.5
36_2b 33 1 p m os 0.65 245 4.0358 0.034 0.1625 0.0019 2482 20 66.3 0.00136 9 0.280976 16 -7.7 0.6 4.45 0.32
45_2b 35 p m os 0.78 0 2.8783 0.025 0.1749 0.0023 2605 22 81.8 0.00223 14 0.280958 17 -5.4 0.6
44_2b 41 3 bmh 0.38 89 2.0282 0.017 0.1760 0.0023 2615 22 99.4 0.00293 20 0.281006 18 -3.5 0.6 6.46 0.37
43_2b 45 p m os 0.52 41 2.0034 0.017 0.1781 0.0021 2635 20 99.5 0.00098 8 0.280950 18 -5.0 0.7
42_2b 47 12 p m os 0.54 262 2.0430 0.017 0.1742 0.0021 2598 20 99.4 0.00142 9 0.280991 15 -4.4 0.5 7.79 0.41
47_2b 49 11 p m os 0.61 0 2.5936 0.021 0.1775 0.0020 2630 19 86.7 0.00106 6 0.280988 17 -3.8 0.6 7.29 0.37
46_2b 60 9 pmh 0.64 0 6.5574 0.055 0.1248 0.0016 2026 22 57.1 0.00307 19 0.281070 21 -14.9 0.8 2.50 0.34
50_2b 64 p m os 0.70 372 2.8111 0.026 0.1771 0.0033 2625 30 82.6 0.00095 6 0.280982 19 -4.1 0.7
48_2b 69 10 p e os 0.44 0 2.4799 0.020 0.1707 0.0020 2565 20 90.7 0.00150 10 0.281008 17 -4.6 0.6 6.02 0.36
51_2b 70 p e os 0.63 31 2.9303 0.024 0.1639 0.0020 2497 20 83.9 0.00069 4 0.280967 23 -7.6 0.8
52_2b 78 6 b m os 0.65 40 2.0243 0.017 0.1782 0.0023 2636 21 99.0 0.00084 5 0.280974 19 -4.1 0.7 4.50 0.33
53_2b 78 8 b m os 0.65 40 2.0243 0.017 0.1782 0.0023 2636 21 99.0 0.00095 6 0.281002 16 -3.1 0.6 3.87 0.27
54_2b 80 5 p e os 0.27 785 5.5334 0.048 0.1843 0.0029 2692 26 36.9 0.00109 7 0.280968 16 -3.0 0.6 5.87 0.42
55_2b 81 p e os 0.83 28 2.0064 0.017 0.1798 0.0023 2651 21 99.1 0.00151 10 0.280954 17 -4.5 0.6

MR257A
22 22 eq m os 2.33 0.46 37 1.9011 0.015 0.1860 0.0019 2707 16 100.4 0.00071 4 0.280967 33 -2.7 1.2
34 34 p m os 3.40 0.67 40 1.9156 0.014 0.1874 0.0018 2719 16 99.8 0.00087 6 0.280990 33 -1.6 1.2
37 37 p m os 3.41 0.37 95 1.9304 0.016 0.1895 0.0017 2738 15 99.0 0.00082 5 0.280989 29 -1.2 1.0
42 42 p m os 1.39 0.40 30 1.9217 0.017 0.1882 0.0021 2727 18 99.5 0.00057 5 0.280934 22 -3.4 0.8
48 48 eq m os 2.19 0.39 52 1.9057 0.017 0.1876 0.0022 2721 19 100.0 0.00060 4 0.280945 27 -3.2 1.0
51 51 p m os 1.54 0.35 39 1.9273 0.015 0.1878 0.0017 2723 15 99.4 0.00058 4 0.280953 26 -2.8 0.9
54 54 p m os 1.64 0.38 25 1.9204 0.015 0.1870 0.0017 2716 15 99.7 0.00064 4 0.280929 28 -3.9 1.0
55 55 eq m os 3.08 0.32 71 1.9088 0.017 0.1879 0.0019 2724 17 99.8 0.00054 3 0.280935 33 -3.5 1.2

114
 Sample spot
α-decay 204 Pb/Pb
Grain (Hg + 238 206 Conc. δ18OVSMOW
Lu-Hf U-Pb a O b events/mg (x Th/U U/ Pb 1ζ 207
Pb/206Pb 1ζ age 1ζ 176
Lu/177Hf 2ζ 176
Hf/177Hft 2ζ εHft 2ζ 2ζ
description 15 Pb) (%) (‰)
10 α/mg) (Ma)
60 60 f m os 0.78 0.57 30 1.9359 0.016 0.1877 0.0019 2722 17 99.2 0.00087 5 0.280948 37 -3.1 1.3
61 61 eq m os 2.32 0.32 40 1.9107 0.015 0.1875 0.0018 2720 15 99.9 0.00093 10 0.280968 32 -2.4 1.1
62 62 p m os 3.00 0.42 78 1.9206 0.016 0.1870 0.0018 2716 16 99.7 0.00153 15 0.280963 24 -2.7 0.9
64 64 p m os 4.32 0.45 54 1.9046 0.018 0.1875 0.0023 2720 20 100.0 0.00080 7 0.280957 28 -2.8 1.0
66 66 p m os 2.24 0.27 109 1.9410 0.017 0.1882 0.0024 2726 21 99.0 0.00037 3 0.280976 27 -1.9 0.9
a
Numbers in italic refer to geochronological data published by Farina et al. (2015a)
b
Grain description: Habit: p - prismatic, f - fragment, eq - equant, b - blocky, ov - oval, an - anhedral; Analysis site: c - core, m - middle, e - edge; Zonation: os - oscillatory, h - homogeneous, rx
- recrystallized

115
 4.4.1 Geochronological results

The majority of the samples analysed during this study have already been dated by several
authors, (see Table 4.3). For the seven samples previously dated by Farina et al. (2015a), and for
which we have analysed additional zircons, we used the intrusion/metamorphic ages provided by these
authors, because these were generally more concordant and more representative of the samples,
because they were obtained from a larger number of grains. This is the case for samples FQ1 (2711 ±
3 Ma), FQ2 (2868 ± 10 Ma), FQ17 (2778 ± 2 Ma), FQ29 (2773 ± 2 Ma), FQ52 (metamorphic age:
2727 ± 11 Ma), FQ60 (2728 ± 16 Ma) and FQ74 (metamorphic age: 2638 ± 14 Ma). For five samples,
the ages obtained during this study are identical, within error, to those from the literature. This is the
case for samples FQ5 (2761 ± 11 Ma), MR31A (2716 ± 14 Ma), MR70G (2716 ± 6 Ma), MR259A
(2721 ± 9 Ma) and MR51A (2708 ± 10 Ma) (Table 4.3). Five additional samples yielded slightly
different ages than those obtained during previous studies (< 1.5 % age difference): FQ6 (2779 ± 4
Ma), MR22A (2715 ± 2 Ma), MR87A (2646 ± 9 Ma), MR14A (2715 ± 3 Ma) and MR257A (2723 ± 8
Ma) (Table 4.3). For one sample (FQ20), we were not able to reproduce the published
geochronological data. This sample corresponds to sample D07A from Lana et al. (2013), who
obtained a SHRIMP magmatic age of 2918 ± 9 Ma, and a poorly defined metamorphic age of 2775
Ma. FQ20 yielded a majority of dark and structureless zircons, with a Concordia age of 2723 ± 3 Ma
which we interpreted as a metamorphic age of the rock.

In addition, we provide new LA-ICP-MS U-Pb ages for three, so far undated rocks. These
consist of a banded gneiss crosscut by numerous leucogranitic sheets and dikes (FQ8), and a
migmatitic gneiss (FQ14) from the Bação complex, as well as a foliated granitoid from the Caeté
dome (FQ81). Zircons from FQ8 mostly display dark structureless centres, yielding a Concordia age
of 2612 ± 10 Ma, interpreted as a metamorphic age. A subset of zircon grains contains bright banded-
or oscillatory-zoned inherited cores. Four discordant analyses on these cores gave older apparent ages
that plot on a regression line yielding an age of ca. 2770 Ma for the protolith of the gneiss. CL images
from zircons from FQ14 revealed a majority of dark and featureless grains. Ca. 40% of the zircons
contain bright inherited cores, surrounded by dark homogeneous rims. The cores yielded concordant
ages ranging from 2925 to 3472 Ma, clustered in four main populations at 2933, 3202, 3358 and 3465
Ma, while the rims gave a Concordia age of 2692 ± 4 Ma. Sample FQ81 was collected from a foliated
granitoid exposed in the Caeté dome (FQ81), where Machado & Carneiro (1992) had obtained a
zircon U-Pb TIMS crystallization age of 2776 +7/−6 Ma from an outcrop located several kilometres
away. In CL images, zircons from FQ81 are mostly dark and structureless, and yielded a younger age
of 2671 ± 10 Ma, interpreted as the age of a late metamorphic event.

Several samples analysed during this study yielded metamorphic ages significantly younger
than the Mamona event in the SSFC (< 2680 Ma). Several authors have previously reported similarly

116
young metamorphic ages in the Passa Tempo complex, located south of the Bonfim complex (2622
Ma; Campos et al., 2003), as well as in the Belo Horizonte complex (2670–2638 Ma; Farina et al.,
2015a). It is important to note that some of these ages are contemporaneous with the magmatic ages
obtained from small granitoid intrusions from the south of the SSFC (2612–2613 Ma; Noce et al.,
1998; Romano et al., 2013). However, the significance of this spread of metamorphic ages remains
elusive.

4.4.2 Lu-Hf isotopes

The RVI event is represented by four banded trondhjemitic gneisses, collected from the
eastern Bação complex and the central Bonfim complex. The gneisses from the Bação complex yield
176
Hf/177Hft of 0.28090–0.28110 (± 0.00002) corresponding to superchondritic εHft values between
176
+0.7 and +5. By contrast, the rocks from the Bonfim complex have significantly lower Hf/177Hft
ratios (Fig. 4.2a). Sample FQ40 contains several zircon cores with Hf isotope ratios ( 176Hf/177Hft =
0.28087 ± 0.00002) identical to those of the gneiss FQ41 cropping nearby ( 176Hf/177Hft = 0.28090 ±
0.00002), which corresponds to subchondritic εHft values between −1.1 and −3.4.

The RVII event is represented by six samples, five of them were collected within the Bação
complex. They consist of two biotite-bearing banded gneisses, a fine-grained plagioclase-rich gneiss
and two weakly deformed Kfs-bearing granitoids collected from small domains within the gneisses.
The last sample was collected in the eastern Bonfim complex within the Samambaia tonalite pluton,
and consists of a medium-grained amphibole- and epidote-bearing granitoid. As for the RVI event,
176
samples from the Bação complex yield systematically higher Hf/177Hft ratios (0.28095–0.28107 ±
0.00002) than those from the Bonfim complex (0.28090 ± 0.00002) (Fig. 4.2b). It is worth noting that
most of the rocks of the Bação complex exhibit superchondritic but slightly less radiogenic εHf t values
than the RVI gneisses. Three samples from the Bação complex display clear core/rim relationships,
with rims dating a regional metamorphic event affecting the complex during the Mamona event, and
yielding average 176
Hf/177Hft of about 0.28102, corresponding to εHft ~ −1. Four samples from the
Bação complex contain inherited zircon cores dated between 2825 to 2920 Ma, with 176Hf/177Hft ratios
overlapping those of the RVI gneisses, indicating the involvement of RVI rocks in the petrogenesis of
RVII magmas.

117
Table 4.3 – Summary of ages and isotopic compositions for samples analysed in this study. Data are from Farina et al. (2015a), Lana et al. (2013), and Romano et al. (2013).
Average Average
UTM UTM Crystallization Metamorphic Published
Sample Classificationa Typec N1d Inherited (Ma) 176
Hf/177Hfint ± εHfint ± N2f δ18O ± 2SDg Referenceh
Lat Long age ± 2ζ (Ma)b (Ma) age
2ζe 2SDe
Bação complex
FQ1 643228 7746959 Kfs-rich 2711 ± 3 conc 6 0.280989 ± 23 -1.8 ± 2.3 22 of 22 5.5 ± 2.1 [1] - This study
FQ2 644685 7749318 Banded 2868 ± 10 wm 13 2705 ± 18 0.281009 ± 24 2.6 ± 2.1 19 of 19 3.9 ± 1.6 [1] - This study
FQ5 (= MR11) 625597 7762152 Plag-rich 2761 ± 11 wm 9 1 core at 3292 0.281002 ± 22 -0.2 ± 1.8 18 of 18 2744 ± 10 [2] - This study
FQ6 (= D12) 625629 7762157 Kfs-bearing 2779 ± 4 conc 3 2891 ± 3 2705 ± 7 0.280970 ± 23 -0.9 ± 2.2 3 of 3 2764 ± 10 [3] - This study
FQ8 628585 7760332 Kfs-bearing - wm 2770 ± 42 2612 ± 10 This study
FQ11 628157 7751740 Plag-rich 2790 ± 3 conc 6 2905 ± 17 2719 ± 14 0.281019 ± 18 1.1 ± 2.0 5 of 5 [1]
FQ13 628731 7748442 Banded 2790 ± 3 conc 17 ca. 2900 0.281030 ± 26 1.5 ± 1.2 12 of 14 [1]
from ca. 2930
FQ14 637895 7750487 Migmatite 2692 ± 4 conc 6 0.281020 ± 29 -1.2 ± 2.3 4 of 4 This study
to ca. 3470
FQ17 625643 7755180 Banded 2778 ± 2 conc 8 2862 ± 2 2732 ± 10 0.280982 ± 21 -0.2 ± 1.3 8 of 9 5.4 ± 1.6 [1] - This study
2 cores at 2829
FQ20 (= D07A) 620091 7749557 Banded - conc 2723 ± 3 2918 ± 10 [3] - This study
and 2772
FQ23 644680 7749530 Banded 2898 ± 12 wm 8 2783 ± 18 0.280977 ± 21 2.1 ± 3.3 8 of 8 [1]

Bonfim complex
FQ29 603409 7765331 Plag-rich 2773 ± 2 conc 10 0.280910 ± 18 -3.2 ± 1.3 42 of 42 4.9 ± 1.8 [1] - This study
FQ37 603293 7759954 Amphibolite 2719 ± 14 wm 9 2879 ± 14 0.280956 ± 24 -2.8 ± 2.9 9 of 9 [1]
FQ40 593263 7767241 Kfs-bearing 2854 ± 18 ui 4 2670 ± 15 0.280871 ± 20 -2.7 ± 1.1 4 of 4 [1]
FQ41 595299 7765480 Banded 2852 ± 16 wm 5 0.280898 ± 20 -1.8 ± 0.9 17 of 17 [1]
FQ51 603378 7741084 Plag-rich 2678 ± 10 wm 14 0.281026 ± 22 -1.3 ± 1.3 28 of 28 [1]
FQ52 582109 7761295 Banded - wm 2727 ± 11 4.6 ± 2.1 [1] - This study
MR31A 594650 7775356 2716 ± 14 wm 13 0.280958 ± 28 -2.8 ± 0.7 12 of 12 2700 ± 8 [2] - This study

3 cores at 3515,
MR22A (= FQ28) 602440 7769392 Kfs-rich 2715 ± 2 conc 22 0.280930 ± 26 -3.8 ± 1.8 7 of 7 2730 ± 8 [2] - This study
3141 and 2835

MR70G 602011 7727899 2716 ± 6 wm 33 0.280948 ± 28 -3.2 ± 1.5 20 of 20 2723 ± 7 [2] - This study
MR87A (= FQ50) 608581 7738759 Kfs-rich 2646 ± 9 wm 22 2734 ± 8 0.280960 ± 26 -4.4 ± 1.6 21 of 21 2613 ± 6 [2] - This study
MR14A (=FQ32) 608654 7756333 Kfs-rich 2715 ± 3 conc 9 0.280955 ± 28 -3.0 ± 1.6 7 of 7 2730 ± 7 [2] - This study

118
 Average Average
UTM UTM Crystallization Inherited Metamorphic δ18O ± Published
Sample Classificationa Typec N1d 176
Hf/177Hfint ± εHfint ± N2f Referenceh
Lat Long age ± 2ζ (Ma)b (Ma) (Ma) 2SDg age
2ζe 2SDe
Belo Horizonte complex
FQ60 548149 7801529 Plag-rich 2728 ± 16 wm 6 2786 ± 29 0.280962 ± 21 -2.4 ± 2.0 17 of 18 4.6 ± 2.9 [1] - This study
FQ65 (= MR259A) 537885 7835546 Plag-rich - wm 2645 ± 8 [1]
FQ70 579027 7790240 Banded - conc 2713 ± 3 [1]
FQ74 555202 7842055 Kfs-rich - wm 2638 ± 14 5.4 ± 3.2 [1] - This study
FQ81 642637 7796689 - conc 2671 ± 10 This study
MR257A 537893 7835513 2723 ± 8 wm 18 0.280958 ± 30 -2.7 ± 1.4 13 of 13 2706 ± 7 [2] - This study
MR259A 539550 7834072 Plag-rich 2721 ± 9 wm 15 0.280949 ± 29 -3.0 ± 2.2 9 of 9 2722 ± 7 [2] - This study
MR51A 621958 7803440 2708 ± 10 wm 10 0.280950 ± 27 -3.3 ± 1.3 18 of 19 2700 ± 8 [2] - This study
a
According to the classification of Farina et al. (2015a);
b
Intrusion age. Ages in bold indicate ages obtained during this study, regular font indicates ages already published by Farina et al. (2015a);
c
Type: conc - concordia, wm - weighted mean, ui - upper intercept;
d
Number of individual U-Pb spots used in age calculations;
e 176
Hf/177Hfint and εHfint calculated using the intrusion age of each sample (excluding inherited cores or metamorphic domains);
f
Number of individual Lu-Hf spots used in the calculation of the average 176Hf/177Hfint and εHfint;
g
δ18O calculated using the intrusion age of each sample, except for FQ52 and FQ74 where we used the metamorphic age
h
References: [1] - Farina et al. (2015a), [2] - Romano et al. (2013), [3] - Lana et al. (2013).

119
Figure 4.2 – 176Hf/177Hft vs. apparent 207Pb/206Pb age diagrams for each magmatic event. Filled symbols
represent analyses with > 95 % concordance, open symbols are analyses with < 95 % concordance. Grey
horizontal arrows indicate Pb-loss trends within individual samples. The Depleted Mantle (DM) (full line)
composition is from Guitreau (2012), while the dashed line represents the most radiogenic DM composition
from Blichert-Toft & Puchtel (2010) (see text for explanation).

The Mamona event is documented by twenty samples collected from the Bação, Bonfim and
Belo Horizonte/Caeté complexes (Fig. 4.2). The rocks from the Bação complex consist of a small
weakly-foliated biotite-poor granitic body intruded at the edge of the complex, two banded gneisses
crosscut by numerous leucogranitic sheets and dikes and a migmatitic gneiss. Zircons from banded
176
gneisses are mostly dark and structureless, interpreted as metamorphic. They yield Hf/177Hft ratios
ranging from 0.28099 ± 0.00002 to 0.28102 ± 0.00003. These values are similar to those obtained for
120
the biotite-poor granite, and correspond to εHft values from −1 to −6 (Fig. 4.2c). Many zircons from
the migmatitic gneiss (FQ14) contain bright inherited cores, surrounded by dark homogeneous rims.
176
The cores yielded variable Hf/177Hft ratios ranging from 0.28101 to 0.28085, corresponding to
highly scattered superchondritic to subchondritic εHft values (Fig. 4.2c). The rims exhibit 176
Hf/177Hft
ratios of 0.28102 ± 0.00003, indistinguishable from those obtained for the banded gneisses and the
granite.

The Mamona event in the Bonfim complex is represented by eight samples: two small
leucogranitic bodies, two weakly foliated aerially extensive granitoid phases from the Mamona and
Souza Noschese batholiths, two small foliated granitoid bodies, a banded trondhjemitic gneiss
cropping out in the centre of the dome and an amphibolitic dike. Zircons from all the granitoids show
176
simple magmatic zoning patterns. All but one sample (FQ51, with Hf/177Hft = 0.28103 ± 0.00002)
have identical 176
Hf/177Hft ratios of 0.28095 ± 0.00003 which corresponds to εHft ~ −3.8 (Fig. 4.2d).
Zircons from the gneiss yielded comparatively higher Hf ratios of ~ 0.28101, and were interpreted by
Farina et al. (2015a) to date a metamorphic event. Zircons from the amphibolite sample display
core/rim relationships. Hf ratios for the rims average 0.28096 which is identical to the values obtained
for the majority of the granitoids from the complex, while those of the cores range from 0.28087 to
0.28096, which is similar to the values obtained for the RVI gneisses from the Bonfim complex.

The Mamona event in the Belo Horizonte complex is documented by seven samples: four
plagioclase-rich granitoids collected from the Pequi and Florestal batholiths, one fine-grained banded
gneiss exposed in the southern part of the complex, one Kfs-rich augen gneiss and a medium-grained
leucogranite. One additional sample was collected from a foliated granitoid exposed in the Caeté dome
176
(FQ81). These eight samples have Hf/177Hft ratios comprised between 0.28092 ± 0.00002 and
0.28098 ± 0.00002, which corresponds to εHft values comprised between −6.2 and −1.7 (Fig. 4.2e).

121
Figure 4.3 – δ18O vs. intrusion age diagram. Filled symbols represent O analyses on zircons with > 90 %
concordance, open symbols are zircons with < 90 % concordance. Fields for Hadean and Archean ―supracrustal
zircon‖ and Archean ―low δ18O zircon‖ are from Cavosie et al. (2005) and Hiess et al. (2011) respectively. The
―mantle zircon‖ field is 5.3 ± 0.3 ‰ as defined in Valley et al. (1998). Inset: δ18O values of dated zircons from
this study, plotted against a compilation from Valley et al. (2005) and Zeh et al. (2014). Zircons from the
Greenland (Hiess et al., 2011) and Northern Australian (Hollis et al., 2014) cratons where low δ18O values were
reported are plotted for comparison.

4.4.3 O isotopes

Oxygen isotopic ratios were measured for 75 zircon grains from seven samples representative
of the three main magmatic events, for which isotopic compositions (U-Pb and Lu-Hf) had already
been obtained. Ca. 20 % of the analyses fall within the range of δ18O values in equilibrium with
mantle-derived rocks (δ18O = 5.3 ± 0.3 ‰) (Fig. 4.3). Ca. 30 % analyses record slightly higher, more
evolved compositions (> 5.6 ‰), with only 6 spots falling in the range of ―supracrustal zircon‖ (6.5–
7.5 ‰) as defined by Cavosie et al. (2005) (Fig. 4.3). This field indicates a range of magmatic δ18O
values that are elevated with respect to those from mantle zircons, thus requiring input from rocks
whose compositions have been shifted towards higher δ18O values by low-temperature processes at
Earth's surface. The upper limit of the field is based on the absence in the Hadean and Archean zircon
record of values with δ18O > 7.5 ‰. One exception is a grain fromFQ52 yielding an anomalously high
δ18O value of 10.88 ± 0.35 ‰. This grain is euhedral and shows oscillatory growth zoning under CL,
and has an apparent 207Pb/206Pb age of 2718 ± 17 Ma (96 % concordant). This is much higher than the

122
sample average, and higher than δ18O reported in Archean zircons (< 7.5 ‰). Although there was no
apparent crack or fracture at the surface of the grain, we suggest that the heavy O isotope value might
originate from post crystallization processes, as the gneiss has undergone metamorphism during the
Mamona event. The remaining 50 % of the analyses yielded δ18O values below the range of mantle
values (< 5.0 ‰). These low δ18O values make up a significant proportion of the zircons analysed
during the three magmatic events, with 15 spots falling in the compositional field of Archean ―low
δ18O zircon‖ (2–4 ‰) from Hiess et al. (2011) (Fig. 4.3). Low δ18O values in zircon indicate that the
grains either crystallized from, or diffusively exchanged O with a low δ18O melt or fluid (e.g.
Bindeman et al., 2008; Gilliam & Valley, 1997; Hiess et al., 2011; Valley et al., 2005). Two gneisses
from the Bação complex contain metamorphic rims surrounding oscillatory zoned grain interiors
interpreted as crystallization features. For both samples, the δ18O values obtained for the rims are
identical to those obtained for the magmatic domains, indicating either that there was no change in O
isotope between the magmatic and metamorphic magmas/fluids, or that the O isotope system was not
disturbed during the metamorphic event. Additionally, the two following trends can be observed
within this dataset: (1) the δ18O values for individual samples increase with time from an average of
3.9 ± 0.2 ‰ at 2868 Ma (FQ2) to 5.4 ± 0.4 ‰ at 2638 Ma (FQ74), in concert with (2) a notable
increase in the range of δ18O composition per sample, from ± 1.6 ‰ for FQ2 to ± 3.2 ‰ for FQ74
(2SD) (Fig. 4.3).

4.5 DISCUSSION

4.5.1 Origin of the Meso- and Neoarchean granitoids

4.5.1.1 Interpretation of the Hf vs. time arrays

The crustal evolution of the three basement complexes of the SSFC will be treated separately
in the following discussion. The linear Hf array (Fig. 4.4), the presence in RVII rocks of RVI inherited
zircon cores, as well as the presence in RVI and RVII rocks of metamorphic rims formed during the
Mamona event indicate that the Bação complex underwent progressive crustal reworking throughout
the Neoarchean time. These lines of evidence can be correlated with (1) the close relationships
observed in the field between gneisses and small granitic intrusions (e.g. xenoliths of gneisses within
the granites), and (2) the geochemistry of the gneisses suggesting a derivation from mixing between a
crustal component derived by melting of an older TTG and a more mafic juvenile end-member (Farina
et al., 2015a). Indeed, these authors observed that these rocks have intermediate compositions between
those of ―true‖ TTGs and those of experimental melts derived from fluid-absent partial melting of
TTGs. The RVI gneisses in the Bação complex were mostly generated through melting of a mafic
crust, with only a limited proportion of reworked felsic crust. This is supported by: (1) overall
geochemical compositions that are comparable to those of ―true‖ TTGs, (2) the scarcity of inherited
zircons, and (3) superchondritic Hf isotope compositions. Patchett & Arndt (1986) demonstrated that

123
because of its higher Nd content compared to that of the mantle, the amount of continental crust
needed to reduce the εNd composition of a Proterozoic granite from +5 down to 0 is small (< 10 %).
Although this example was made using Nd isotopes, the same logic applies to the Hf system.
Therefore we suggest that the volume of reworked component involved in the generation of the RVI
gneisses from the Bação complex was probably < 10 %.

Figure 4.4 – εHft vs. age diagrams for: (a) igneous and metamorphic zircons (this study). The diagonal grey
arrows indicate crustal evolution trends derived from the Hf data of this study, using an average 176Lu/177Hf of
0.0113 for the average continental crust. The arrays 1 and 2 are fitted through the average compositions of the

124
granitoids and gneisses of the (1) Bação and (2) Bonfim + Belo Horizonte complexes respectively, indicating
that the complexes evolved as different terranes or portions of the crust, and (b) detrital zircons (data from
Koglin et al., 2014; Moreira et al., 2016; Martínez Dopico et al., 2017) of the SSFC. The Depleted Mantle (DM)
evolution is as in Fig. 4.2.

It is interesting to note the large range in initial Hf isotopic compositions displayed by the RVI
gneisses (εHf between 0.0 and +6.5). This is rather uncommon in typical I-type granitoids, and could
be explained in several ways, which include incomplete mixing between two isotopically different
magmas (in this case, a crustal component and a more juvenile, mafic one), or as inherited from the
source, as a result of incomplete homogenization of the magma during dissolution and crystallization
of isotopically heterogeneous zircons (Farina et al., 2014; Villaros et al., 2012). The subsequent
reworking of this crust during RVII, with some minor contribution from juvenile material produced
176
granitoid magmas with lower (superchondritic to chondritic) Hf/177Hf isotope ratios and
compositions that are ―transitional‖ between those of a TTG-like end-member and a high-K one.
Further reworking of this ―transitional‖ crust during the Mamona event generated high-K granitoids
with subchondritic Hf isotope compositions. One gneiss sample from the Bação complex (FQ14)
displaying migmatitic structures with garnet-rich leucosomes provides evidence of inheritance of
distinctly older crust (concordant zircons of up to ~ 3.47 Ga). The occurrence of inherited cores with
ages clustering between 2933 and 3465 Ma suggest that the protolith for FQ14 was a
(meta)sedimentary rock that underwent partial melting during the Mamona event.

In the Bonfim complex, the RVI event is represented by two gneiss samples (Alberto Flores
gneiss), with low εHft = −1 to −4. Samples from the Alberto Flores gneiss show highly variable
compositions which differ from those of ―true TTGs‖ (e.g. higher Si, K, Rb and lower Al contents).
This suggests the involvement of reworked continental crust in their petrogenesis (Farina et al.,
2015a), a hypothesis which can be reconciled with their subchondritic Hf isotope data. During RVII,
the Bonfim complex is intruded in the east by the Samambaia tonalite. This magma has an evolved Hf
isotope composition (εHft = −2 to −4) that falls within the εHf array defined for the Bonfim complex.
However, a derivation from an older felsic crust formed during RVI for example can be ruled out, as it
cannot account for the intermediate compositions of the Samabaia tonalite (i.e. SiO2 ranging from 64
to 72 wt.%) (Carneiro, 1992; Farina et al., 2015a). Instead, its enriched Hf isotopic signature probably
originates from assimilation of older felsic crust by a more mafic magma. The Mamona event is
mostly represented by large high-K batholiths emplaced in the eastern and northern borders of the
dome, namely the Mamona and Souza Noschese batholiths. These granites all have subchondritic εHft
values ranging from −1 to −6 that plot on an array defined by the gneisses of the Bonfim complex
(Array 2; Fig. 4.4). This array suggests that high-K granitoids formed by reworking of the previously
formed crust. However, the K and LILE contents of these rocks are too high to be explained by a
derivation from partial melting of the orthogneisses. Farina et al. (2015a) therefore argued that the
large high-K batholiths emplaced during the Mamona event require a source that is more enriched in K
125
and LILE, and more fertile than the average continental (medium-K) crust present in the SSFC at the
time, suggesting their derivation by melting of metasedimentary protoliths such as metagreywackes.

In the Belo Horizonte complex, the Mamona event is marked by the coeval intrusion of
granitoids with very different field and geochemical features: (1) in the east, two voluminous
batholiths (Pequi and Florestal) are characterized, as opposed to those emplaced in the Bonfim
complex, by homogeneous sodic compositions presenting some similarities with TTGs (e.g. typically
high Al2O3, CaO, Na2O, Sr, LREE and low K2O, Rb, Y, HREE), and (2) smaller granitoid
domains/dikes and leucogranites, showing a wider range of more enriched compositions. The
petrology and geochemistry of the former suggest a derivation from a hydrous mafic rock. Assuming
this is the case, their low Hf isotope signatures can be explained in two ways: (1) partial melting of an
old mafic crust. This however implies that this mafic precursor was extracted from the depleted mantle
176
long before remelting to produce these magmas. In particular, considering an average Lu/177Hf =
0.022 for a mafic crust (Nebel et al., 2007), this corresponds to > 600 m.y. of crustal residence time
(Fig. 4.4), or (2) partial melting of a mafic crust but with assimilated portions of much older crust.

The existence of several Archean basement complexes with distinct εHf vs. time arrays (Fig.
4.4) suggests that these terranes underwent partly/largely different histories. In this scenario, it is
possible to imagine that the basement presently exposed in the SSFC consists of a collage of different
micro-continental blocks amalgamated during a late collisional stage.

4.5.1.2 Secular trend: from sodic to potassic magmatism

We have combined the Hf isotope data presented here with geochemical results obtained by
Farina et al. (2015a) for sixteen igneous samples. The results are presented in Fig. 4.5, where εHft for
each sample is plotted against some key geochemical features (K 2O/Na2O, Th and Sr/Y),
representative of the trends observed here. Magmas produced during RVI share similar features with
TTGs, such as high Na2O, (La/Yb)N, low K2O, Th, U contents, and high Sr/Y, usually interpreted to
reflect the depth at which those TTG magmas are produced. However, it is important to note that small
differences do occur between ―true‖ TTG magmas and those of the SSFC (e.g. lower Sr/Y; Fig. 4.5).
During RVII, the magmas display a wider range of more evolved compositions (e.g. K 2O/Na2O up to
1, Th up to 20 ppm), however still overlapping the compositional field of TTGs. The magmas
produced during the Mamona event all have subchondritic Hf isotope signatures but heterogeneous
trace element compositions. In particular, we observe systematic differences between magmas forming
large batholiths in the Belo Horizonte complex which are similar to TTGs, and those emplaced as the
large batholiths of the Bonfim complex, which present the highest K/Na and lowest Sr/Y compositions
of this dataset. Between these two end-members, we observe a range of (much less voluminous)
magmas with intermediate compositions and different petrological features (from sodic Plag-rich
granites to Kfs-rich granitoids; Farina et al., 2015a).

126
Figure 4.5 – Plots of εHft vs. geochemical parameters, showing general trends of enrichment with decreasing
age and εHft of the magmas (grey arrows). The grey fields represent TTG compositional fields from Moyen
(2011).

127
 Overall, the most striking feature is the relatively continuous evolution trend with decreasing
age (and decreasing εHft), from compositions close to those of TTGs to more enriched granitoids. In
particular, we observe a progressive enrichment in High Field Strength Elements (HFSE) with time
(e.g. Th from 6 to 30 ppm). These trends cannot be explained as a result of different degrees of
fractionation from a similar source, as these magmas almost all have very similar high SiO2 contents
(> 70 wt.%). HFSE are highly immobile elements, and are usually considered as good tracers of
source enrichment. We infer that these trends reflect crustal maturation and differentiation via
progressive reworking at shallower crustal levels. In addition, it is important to note that the wide
range of geochemical compositions characteristic of Mamona granitoids mirrors the diversity of
processes and sources involved in the generation of these magmas, which cannot be appreciated using
Hf isotopes alone. The trends documented here are not exclusive to the SSFC, and similar ones are
well described in the Yilgarn and Pilbara cratons of Western Australia for example (Griffin et al.,
2004; Ivanic et al., 2012). There, nearly continuous crustal melting over a protracted period of time
has recorded a transition from sodic (TTG) to potassic magmatism, which is interpreted to reflect a
similar scenario of progressive reworking.

4.5.1.3 Insights from O isotopes

Do these δ18O values represent primary magmatic features?

The interpretation of O isotope data depends critically on whether the measured O isotope
compositions reflect that of the magmas from which the zircons crystallized, or if they are product of
secondary alteration processes and isotopic exchange. In CL images, zircons generally display
euhedral habits and oscillatory growth zonation. Th/U are between 0.02 and 0.8, and only 7 out of 75
spots analysed have Th/U < 0.1. However, a significant amount of zircons have low levels of U-Pb
concordance (ca. 30 % of zircons have concordance < 80 %), which can cast doubts on the
interpretation of the isotope compositions as primary values. Valley et al. (1994) observed that Pb-loss
(indicated by higher levels of discordance) is always associated with resetting of O isotope ratios,
shifting δ18O values up to 2 ‰ lower than their primary values. However, these authors observed that
there is no apparent proportional relationship between the degree of concordance and the shift in O
isotope ratio (Valley et al., 1994; their Fig. 4). In other words, elevated levels of discordance are not
necessarily coupled with highly disturbed O isotope ratios. For most samples in this dataset, no
correlation is observed between δ18O and Th/U, 204(Hg + Pb) or the degree of concordance (%), which
suggests that δ18O values and radiation doses U-Pb systematics cannot be directly associated (Fig.
4.6). Overall, we argue that most of the zircons from the São Francisco craton analysed in this study
have preserved their primary isotopic signatures and that these reflect that of the magmas they
crystallized from.

128
Figure 4.6 – δ18O vs. U-Th-Pb systematics diagrams. The lack of apparent correlations indicates that the δ 18O
values are not products of secondary alteration and can instead be considered as primary features (see text).

129
 We note however that two samples (FQ52 and FQ60) show a slight positive correlation
between δ18O and the level of concordance of the zircons (Fig. 4.4c). Zircons from FQ52 sometimes
display faint oscillatory zoning disrupted inwards from the grain boundary by recrystallized domains.
The edges of the grains commonly contain inclusions and/or fractures suggesting that the zircons were
affected by fluid-dominated recrystallization (as described by Hoskin & Black, 2000), probably during
the metamorphic event dated by this sample at 2727 ± 11 Ma. By contrast, the zircons analysed from
FQ60 are mostly euhedral and display clear oscillatory zoning patterns, with no sign of pervasive
alteration in CL or BSE images. In this case, it is not clear what process caused the Pb-loss and the
disruption of the Hf system. Together, this suggests that FQ52 and maybe FQ60 underwent some
degree of post-magmatic alteration causing a shift towards some of the low δ18O values observed in
the most discordant zircons.

Secular trend of increasing δ18O values: evidence for supracrustal reworking

If they can fingerprint involvement of crustal material, Hf isotopes alone do not allow
determination of whether contamination occurred via (1) source mixing, whereby the recycling of
subducted materials into the source reservoir influences the isotopic signature of the resulting magmas,
or (2) crustal contamination, that is contamination of the magma during ascent and/or emplacement by
interaction (assimilation) with the crust it intrudes. In modern arcs, stable (e.g. O) and radiogenic (e.g.
Nd, Hf, Sr) isotopes have been used widely to discriminate between these contamination processes
(Appleby et al., 2010; Kemp et al., 2006; Peck et al., 2000).

The first trend in the O isotope data is the steadily increasing maximum δ 18O throughout the
Neoarchean, from ~ 5.3 ‰ at 2.87 Ga to ~ 7.8 ‰ at 2.64 Ga (Fig. 4.3). This follows the general trend
defined by a global O isotope dataset (Roberts & Spencer, 2014; Valley et al., 2005). In the global
dataset, this secular rise of magmatic O isotope ratios is explained by a combination of modifications
in sediment composition, availability, weathering and burial, originating from a change in tectonic
styles at the Archean-Proterozoic boundary (Valley et al., 2005). In this study, the appearance of high
δ18O(Zrc) (> 6.5 ‰) in rocks at ~ 2.7 Ga indicates reworking of supracrustal lithologies during the last
Mamona event, this either being sedimentary rocks (10 to 40 ‰) or altered volcanics (20 ‰) (Eiler,
2001). Archean sediments, dominated by greenstone belt assemblages made of volcaniclastics,
pyroclastics and less mature sediments record lower δ18O(WR) ratios compared to their modern
counterparts (Longstaffe & Schwarcz, 1977; Veizer & Mackenzie, 2003). Using an average of 15 ‰
for Archean sediments (average of Archean sandstones and shales from Valley et al., 2005), we
estimate that 10–20 % of sedimentary contaminant would be required to increase the δ18O of a normal
igneous rock and zircons by 1–2 ‰, the amount required to reach the range of δ18O measured here.
The question remains to determine whether these high δ18O result from upper-crustal contamination or
via direct sediment reworking. While we cannot completely rule out the possibility that these high

130
δ18O(Zrc) values originate from assimilation of uppercrustal lithologies, the involvement of
metasediments is in agreement with the model proposed earlier for the petrogenesis of these rocks.

The origin of the low δ18O magmas

About 50 % of all O isotope analyses have values below the range of mantle zircon
compositions (< 5.3 ± 0.3 ‰; Fig. 4.3). Low δ18O values indicate that the zircons either crystallized
from, or exchanged O via diffusion with a low δ18O melt or fluid. The only low δ18O materials are
meteoric and seawater and materials that have undergone alteration with these fluids at high
temperatures. Eiler (2001) indicated that the lower gabbroic portion of oceanic crust preserves a low
δ18O(WR) (0–5 ‰) acquired during high-temperature interaction with percolating water at the ridge. We
propose two different possible interpretations to account for the systematic presence of low δ 18O
zircons in all three magmatic events.

The first model stems from the fact that all these rocks derived from a diversity of rather well-
characterized sources and processes (Farina et al., 2015a; this study), and that perhaps a single unique
mechanism is unlikely to account for the low δ18O values observed in all of them. We therefore
consider them separately. The magmas emplaced during the RVI event (FQ2) have δ18O values that
range from ~ 3 to 5 ‰. This rock is inferred to result from mixing between a juvenile material and a
TTG-derived melt (Farina et al., 2015a). The superchondritic εHft values displayed by the zircons in
FQ2 indicate only a minor component of evolved crust within this magma (< 10 %). These first-order
observations suggest that > 90 % of the source of this rock was a juvenile rock, and that it had to
account for the O isotope compositions observed. The low δ18O(Zrc) values of FQ2 can be explained as
a result of partial melting of a lower and relatively young δ18O-depleted gabbroic oceanic crust. The
resulting melts will then mix with a small proportion of a continental crustal component, possibly
originating by local wall-rock melting of the surrounding TTG crust. The RVII magmatic zircon
population, documented by sample FQ29 and a few inherited cores displaying mantle-like values, is
centred around mantle-like δ18O values. It is worth noting that zircons from FQ29 display a small
spread towards low δ18O values, with two grains plotting in the field of ―low δ18O zircons‖, and
relatively clustered evolved Hf compositions (εHft = −1.0 to −4.2). Derivation of these rocks by
reworking of RVI continental crust to explain its evolved Hf composition, as may be suggested by the
Hf linear array, can be ruled out on the basis that this existing crust had SiO 2 contents greater than 70
wt.% and was therefore incapable of giving rise to some of the more mafic compositions (SiO 2 = 65–
70 wt.%) recorded in the Samambaia samples (Carneiro, 1992; Farina et al., 2015a). Moreover, field
and geochemical observations suggest the involvement of a mantle source component for the
Samambaia tonalite (Farina, personal comment) which is in agreement with the observed mantle-like
δ18O compositions. In this context, the unradiogenic Hf isotope composition of sample FQ29 could be
explained either by assimilation or source contamination by an evolved and slightly δ18O-depleted
crustal component. It is interesting to speculate on the origin of the low δ18O magmatic zircons
131
observed in rocks from the Mamona event. Samples emplaced at that time all record δ 18O values
above those of the mantle, attributed to the reworking of δ18O-enriched metasediments. However, this
process does not account for the δ18O-depleted zircons. Firstly, we should address the question of the
nature of the fluid responsible for the δ18O-depleted compositions. If the O composition of seawater is
assumed to be constant and equivalent to its present-day value (δ18O ~ 0 ‰; Gregory, 1991), then high
temperature exchange with seawater would require very large volumes of seawater to explain the low
δ18O observed. Instead, meteoric fluids have compositions between 0 and −55 ‰ (Bindeman, 2011), a
range that is more plausible to produce the δ18O-depleted compositions. In recent magmatic systems
reporting low δ18O magmas, the latter are interpreted to originate within shallow sub-volcanic magma
chambers where assimilation of hydrothermally altered wall rock has occurred (e.g. Bindeman &
Valley, 2001; Monani & Valley, 2001; Wotzlaw et al., 2012). Where these low δ18O zircons occur in
the Archean, similar settings (shallow-level geothermal systems) have been proposed (Hiess et al.,
2011; Hollis et al., 2014). In such environments, the emplacement of new granitoid magmas
effectively drives groundwater into the crust through fracture networks where these meteoric fluids
heterogeneously interact with wall rocks, lowering their O isotope compositions (Taylor, 1977). The
isotopic heterogeneity observed in some samples (e.g. FQ74) is consistent with such an environment.
These systems generally require that magmas are emplaced at relatively shallow levels, to have access
to meteoric water. In general, low δ18O values are scarce, particularly amongst Archean rocks (Valley
et al., 2005), where they have only been reported in southwest Greenland (Hiess et al., 2011) and
northern Australia (Hollis et al., 2014) (see inset in Fig. 4.3), testifying of either their rarity, or a lack
of preservation of such materials. Therefore, this model may be hampered by the fact that it requires
low δ18O material to be generated at three consecutive times in the same area, which may be
considered unlikely for this and has not been documented elsewhere.

Alternatively, the systematic presence of these low δ18O values could be interpreted as
resulting from a single hydrothermal alteration event, producing a significant amount of δ 18O-depleted
material whose O isotope signature is then carried through the subsequent crustal reworking events. In
this scenario, we propose that the source of the RVI gneisses, essentially a juvenile mafic rock, most
probably oceanic crust, was contaminated by a significant amount of meteoric water during formation
at a submarine rift zone, generating dynamic hydrothermal systems (see Eiler, 2001). Moreover, this is
in agreement with the consensus that TTGs commonly derive from a partially hydrated mafic crust
(e.g. Rapp & Watson, 1995). The systematic presence of these values in rocks from all three magmatic
events highlights both the lack of efficient magma mixing, as well as the fact that reworking is not
swamped by addition of new crust through the subsequent important magmatic events. In this
hypothesis, the low δ18O signatures act as an effective tracer for crustal reworking in the SSFC, much
in agreement with the Hf isotope data.

132
 Overall, although both models propose slightly different interpretations of these low δ 18O
values, the data definitely indicates the interaction between meteoric water and crustal rocks at least
once in the Mesoarchean and possibly also during the Neoarchean Mamona event in the SSFC. Given
the proposed importance of oceanic crust in the formation of TTG magmas, it is perhaps surprising
that low δ18O TTGs have not been more commonly reported.

4.5.2 The growth of the continental crust in the SSFC

4.5.2.1 Archean depleted mantle evolution

Model ages are commonly used to produce estimates of crustal growth events (e.g. Bennett &
De Paolo, 1987). This is based on the premise that model ages reflect the age of extraction from a
depleted mantle reservoir, with zircon crystallization occurring only later on. Current models suggest
that the depletion of the upper mantle from a chondritic reservoir started very early in Earth's history
176
and was extrapolated more or less continuously, resulting in a linear Hf/177Hf array to reach its
current composition, constrained by present-day MORB. Although some rocks show local evidence
for early (Hadean) mantle depletion (e.g. suprachondritic εHft values recorded in Eoarchean rocks
from west Greenland, Pilbara and Barberton; Amelin et al., 2000), there is little data supporting
derivation of early continental crust from a long-term Hf-depleted mantle reservoir. In fact, zircons
from felsic rocks with ages > 3.5 Ga from a number of Archean terranes have maximum Hf isotope
compositions similar to chondritic values (Amelin et al., 2000; Kemp et al., 2009b; Lancaster et al.,
2014, 2015; Naeraa et al., 2012; Satkoski et al., 2013; Zeh et al., 2009). Similarly in the SSFC, there is
a striking lack of zircon data with superchondritic Hf compositions, particularly observed within the
detrital record (Fig. 4.4b). Koglin et al. (2014) and Moreira et al. (2016) argued that the subchondritic
εHft compositions of zircons from the SSFC reflect intense reworking episodes of the crust, starting as
early as ~ 3.4–3.5 Ma. An alternative view would be that the zircon record does not necessarily reflect
crustal reworking, but instead argues against the presence of a strongly depleted mantle beneath the
continental crust. Using a compilation of ~ 13,000 Lu-Hf analyses on zircons, Guitreau (2012)
proposed a model of evolution for the depleted mantle that fits the maximum Hf compositions of
zircons through time. This model is defined by a period of only minor increase of εHf t for the first 1
Gy, followed by a period from ~ 3.5 to ~ 2.5 Ga of enhanced differentiation of the depleted mantle,
followed after ~ 2.5 Ga by a constant increase, although slightly less steep, until a present-day value of
εHftoday = +18. This model corresponds broadly to the ―two-stage‖ evolution model discussed by Zeh
et al. (2009) (their Fig. 12). It implies that for the first half of the history of the Earth, the source of the
continental crust is not as depleted as proposed in the more commonly used models (e.g. that of
Blichert-Toft & Puchtel, 2010). Applied to the SSFC, the model of Guitreau (2012) fits exceptionally
well with the detrital zircon data (Fig. 4.4b). In addition, this model reduces the crustal residence time
of zircons > 2.9 Ga from an average of 400 down to 200 m.y. Different tectonic settings in the
Archean – with higher radiogenic heat production resulting in a large number of small unstable
133
microplates and more dynamic tectonics – are more easily reconciled with shorter crustal residence
times. For the next section, we tentatively use this depleted mantle model from Guitreau (2012) to
discuss the crustal evolution of the SSFC.

4.5.2.2 Crustal growth

In order to discuss continent formation, understanding the growth of the continental crust
involves evaluating the ratio between new magmatic additions directly extracted from the mantle
(juvenile) and ones that originate from remelting of older crust (reworking). The main challenge when
trying to determine the distribution of crustal growth using detrital zircon is to correct for the data that
represents crustal reworking in order to calculate a crustal generation curve. Temporal Hf isotopic
trends have been used in several regional studies (i.e. Boekhout et al., 2013; Kemp et al., 2009a) in
order to estimate the varying degrees of mantle input in the generation of granitic magmas. In this
study, the regional magmatic evolution of the SSFC was explored using U-Pb and Lu-Hf data from
detrital zircon from recent publications focused on the Neoarchean supracrustal successions of the
Maquiné Group (Moreira et al., 2016) and Moeda Formation (Koglin et al., 2014; Martínez Dopico et
al., 2017). Modelling of the degree of juvenile growth was largely based on the calculation techniques
applied by Belousova et al. (2010). Five zircons with εHft > 2 εHfDM were rejected. In total, 1178
analyses were used for modelling. All the crustal T DM2 ages for these zircons were recalculated
according to the depleted mantle evolution model of Guitreau (2012) for which we have graphically
determined the equation for this age interval (4.0–2.5 Ga) (Fig. 4.7b). For all the other parameters (Lu
176
decay constant, CHUR parameters and average Lu/177Hf of the crust), we have adopted the same
values as indicated in Appendix A9. The results of the modelling are summarised in Fig. 4.7.

The distribution of U-Pb detrital zircon ages along with an integral curve are plotted in Fig.
4.7a. The three major peaks observed at ca. 2.88, 2.79 and 2.72 Ga correspond to the RVI, RVII and
Mamona events, respectively. Their presence reflects the good preservation of Neoarchean rocks in
the SSFC. The existence of continental crust older than 2.9 Ga, although it is absent in the field, is
evidenced by the Hf model ages of the detrital zircons. The dark blue line in Fig. 4.7a represents the
cumulative curve of crustal model ages obtained for this dataset. It suggests that if the generation of
new crust was predominant during the Paleoarchean, the importance of crustal reworking increased
after 2.9 Ga. However, the Hf model ages on their own do not address the possibility that some of
these zircons were produced from a mixture between a radiogenic component and an older crustal
material with a lower Hf isotope composition. Payne et al. (2016) estimated that only a small
proportion (14 %) of Hf model ages actually provide a meaningful indicator of the timing of crustal
growth, the rest of the model ages likely resulting from mixtures of melt derived from multiple mantle
and crustal sources. Recent studies have used different methods in an attempt to remove this ―mixed
signal‖ and thereby accurately evaluate the proportion of juvenile material added to the crust at each
step of its evolution (Belousova et al., 2010; Dhuime et al., 2012; Kemp et al., 2007). Zircons with
134
juvenile Hf compositions are defined as falling in a range of ± 2ε units (or ± 0.75 %) around the
depleted mantle composition (Griffin et al., 2014). Using this simple definition, we have calculated the
proportion of juvenile material produced at each time step, and obtained a cumulative curve (the green
line in Fig. 4.7a) that is comparable to the one obtained from model ages, only slightly shifted towards
older ages. Although debatable, these two simplistic approaches to estimate the growth of the
continental crust in the SSFC indicate that most of it (ca. 95 %) was already generated by 2.9 Ga (Fig.
4.7a), and that crustal reworking was dominating over net juvenile additions after 2.9 Ga. This
conclusion is consistent with the geochemical data from granitoids of the SSFC, as discussed in
Section 3.5.1.

Assuming that the generation of Archean granitoids involved interactions between juvenile
and crustal sources, the average proportion of new crust through time can be estimated using two-
components mixing calculation. This requires some assumptions to be made about the geochemical
and isotopic compositions of the mantle and crustal end-members. Here we estimated the isotope
composition of the crustal end-member from an ―integral crust‖ calculation as described by Belousova
et al. (2010). This integral crust is calculated so that the average signature of new crust is added
successively to that of an older and more evolved crust. This method is similar to the calculation of
crustal model ages but with an interpolation projected forward instead of backwards. This provides a
more realistic estimate for the average crustal composition than the use of a single value. Ages
younger than 2.60 Ga and older than 3.45 Ga are largely under-represented in this dataset, therefore
the calculations are statistically more subject to bias and the results obtained for these periods will not
be discussed further. The juvenile fraction between a depleted mantle (that of Guitreau, 2012) and a
crustal component (given by the integral curve) for bins of 20 Ma is calculated using mean Hf
concentrations of 2.31 (Kelemen et al., 2003) and 4.0 ppm (average of TTG; Laurent et al., 2014)
respectively. If we consider a truly depleted mantle with [Hf] DM = 0.157 ppm (Workman & Hart,
2005), the calculated juvenile contribution shifts towards higher proportions. However, as discussed in
the previous section, the presence of a true depleted mantle beneath the SSFC in the Archean is
questionable. Moreover, even if the quantitative proportions of juvenile material depend on this
assumption, the temporal trends will remain the same. The results shown in Fig. 4.7b (red line)
indicate that 20–70 % of the melts generated at all times were juvenile. The following trends can be
observed: (1) there is a general decrease in the juvenile contribution to the magmas over time (red box
in Fig. 4.7b), dropping from an average of 55 % during the Paleo- and Mesoarchean to ~ 40 % at 2.9
Ga, and down to ~ 20 % at around 2.7 Ga. We suggest that the reason for this general trend is a change
in geodynamics at 2.9 Ga, with a transition from island arc to continental arc (see discussion in the
next section). Additionally, this drop of mantle input coincides with the increase of sediment input to
the generation of crustal granitoids as suggested by the O isotope data, (2) we observe some
significant variations in the juvenile input during the Paleoarchean, with rapid drops of up to 30 %

135
immediately followed by rapid increases. Although similar variations have been interpreted to reflect
tectonic switching from regimes dominated by compression punctuated by extensional events (e.g.
Collins, 2002; Collins et al., 2011; Kemp et al., 2009a), it is likely that some of these apparent
variations of mantle input can simply be an artefact related to the scarcity of the Paleoarchean record,
largely underrepresented in this dataset.

Figure 4.7 – (a) U-Pb age distribution for detrital zircons from the SSFC (grey bars; right scale) (data from
Koglin et al., 2014; Moreira et al., 2016; Martínez Dopico et al., 2017). Cumulative/integral curves of U-Pb ages
(light blue) and of crustal TDM ages (dark blue). The green line represents the integrated curve obtained from
zircons defined as ―juvenile‖ (with εHft > 0.75 ∗ εHfDM). (b) εHft vs. age diagram. The integral crust (grey line)
is calculated following the method of Belousova et al. (2010). The proportion of juvenile component (red curve;
right scale) is calculated in reference to the depleted mantle evolution from Guitreau (2012) and the integral crust
(see text for explanation).

136
 Overall, the detrital zircon record indicates a change of geodynamic processes at ca. 2.9 Ga.
During the Paleoarchean, processes of renewed juvenile crust generation were operating, and although
we have little direct evidence, it was certainly characterized by relatively high mantle contribution to
magmatic episodes. At ca. 2.9 Ga there is a net increase in the extent of crustal reworking (further
supported by geochemical arguments; Farina et al., 2015a) in the transition to a regime with balanced
growth and destruction.

4.5.3 A geodynamic model for the SSFC

The combined dataset of magmatic and detrital zircons indicates that the SSFC was
continuously affected by magmatic activity from ca. 3.50 to 2.65 Ga. Any geodynamic model chosen
to represent the evolution of this portion of the crust must account for all of the features within this
combined U-Pb-Hf-O dataset, as well as the field, petrographic and geochemical evidence collected by
Farina et al. (2015a). Below we discuss the magmatic and geodynamic evolution of the SSFC.

4.5.3.1 Pre-2.9 Ga

The early evolution of the SSFC is represented solely by detrital zircons, displaying
continuous and homogeneous subchondritic εHft values between ca. 3.5 Ga and 2.9 Ga. The lack of
Paleoarchean rocks in the field precludes any precise interpretation on the nature of the crust that was
formed at the time. However, similar isotopic values have been described in TTGs from other Archean
terranes (Moyen & Martin, 2012). Additionally, the modelling discussed in Section 3.5.2 indicates
that: (1) the majority of the crust (~ 90 %) was originally formed before 2.9 Ga, and (2) the juvenile
proportion of newly formed crust averages ~ 55 % over that period, before decreasing notably after 2.9
Ga. Evidence is missing however in the SSFC to further discuss Paleoarchean tectonics and to explain
the long-lived production of juvenile continental crust in the SSFC.

4.5.3.2 Post-2.9 Ga

The proposed Neoarchean evolution model comprises two stages, the first one coincides with
the RVI and RVII events, and the second one corresponds to the Mamona event.

During stage I (Fig. 4.8a), the model must account for: (1) reworking of Paleoarchean felsic
crust evidenced by older model ages in post-2.9 Ga magmas, the erosion and deposition of this crust
into Neoarchean and Paleoproterozoic sedimentary basins, (2) the general geochemical trends
indicating a progressive HFSE enrichment of the source of these magmas, (3) lesser juvenile
contribution to the magmatism, and (4) the systematic differences (field, geochemical and isotopic)
observed between the three complexes. These data can be reconciled with a continental collision
model, during which the accretion of various proto-continents leads to a tectonically thickened crust
which undergoes progressive reworking. This scenario also accounts for the metamorphic event
between 2.78 and 2.73 Ga recorded in the SSFC (Farina et al., 2015a; Lana et al., 2013).

137
Figure 4.8 – Sketch illustrating the evolution of the Archean continental crust exposed in the SSFC. (a) From 2.9
Ga, the accretion of several proto-continents progressively leads to a tectonically thickened continental nucleus
that undergoes differentiation, producing the medium-K magmas characteristic for the RVI and RVII events. The
felsic volcanism associated with RVII is recorded in the greenstone belt sequence; (b) During the post-collisional
Mamona event (< 2.75 Ga), the induced inflow of asthenospheric mantle beneath the orogenic prism triggers
further melting of lower crustal and upper mantle lithologies.

Following this episode of crustal thickening, we suggest that stage II (Fig. 4.8b) marks a
modification of the tectonic regime into extensional or non-compressional settings, during which slab
break-off or retreat, lithospheric delamination or late- to post-orogenic gravitational collapse can

138
trigger further melting of lower crustal lithologies (e.g. Duretz & Gerya, 2013; van Hunen & Allen,
2011). In this scenario, the compression of local geotherms at the base of the crust promotes renewed
partial melting and regional metamorphism of the lower crust (Sandiford & Powell, 1986) and upper
mantle. This is further supported by the existence of mafic-intermediate amphibolitic dikes emplaced
during this time. We suggest that the magmatism produced during this event generates three types of
granitoids. First, the local remelting of older (RVI and RVII) gneisses (the medium-K magmas of
Farina et al., 2015a) generates small volumes of magmas with a wide range of compositions reflecting
the compositional and isotopic heterogeneity of the source(s). These rocks are currently exposed in all
three complexes as decimetre- to metre-scale granitic veins and dikes, small plutons (General
Carneiro, Santa Luzia, Ibirité, Brumadinho) and domains closely associated with the gneisses as well
as leucocratic veins and dikes. Secondly, in the Bonfim complex, the reworking at depth of
metasediments produced large volumes of biotite and two-mica granitoids (Mamona and Souza
Noschese batholiths). And finally in the Belo Horizonte complex, the generation of the Pequi and
Florestal batholiths primarily reflects the remelting of a metabasaltic source, generating magmas that
later assimilated some older crust. This final event is synchronous with the complete depletion of the
lower crust in heat-producing elements and the subsequent stabilization of the SSFC (Lana et al.,
2013; Romano et al., 2013; Sandiford & McLaren, 2002).

Such a scenario has been proposed for a number of analogous Archean terranes such as for
example the Yilgarn and Greenland cratons, interpreted as collages of several crustal blocks formed
during arc magmatism, terrane accretion and collisional orogeny (e.g. Czarnota et al., 2010; Windley
& Garde, 2009). Additionally, Laurent et al. (2014) recently reviewed the temporal evolution of
several well-characterized Archean terranes, identifying a two-stage sequence. First, a long-lived
period of TTG magmatism, followed by the generation of a range of more enriched granitoids
emplaced during a shorter event. These authors tentatively proposed that this pattern reflects a global
geodynamic model of subduction and subsequent continental collision, taking place on a planetary
scale between 3.0 and 2.5 Ga, as a result of the progressive cooling of the Earth. The zircon dataset
presented in this study can easily be reconciled with such scenario, although it does not provide direct
evidence for it. 2.9 Ga marks a clear transition in the SSFC into a period where crustal reworking has
dominated over net juvenile magmatic additions. By analogy, we infer that this change relates to the
onset of accretionary and collisional events in the SSFC, leading to oceanic closure and progressive
amalgamation of Paleoarchean proto-continental blocks.

4.6 CONCLUSION

The conclusions that arise from this U-Pb-Hf-O isotope study on Neoarchean granitoids and
gneisses from the SSFC are as follows:

139
 - Samples from the three complexes (Bação, Bonfim and Belo Horizonte) plot on distinct
crustal evolution arrays, suggesting the involvement of Paleoarchean crust in their
generation, which occurs via different processes (remelting, mixing, assimilation and/or
source contamination). Combined with field data, this suggests that these complexes
represent terranes with different identities (different proto-continents and/or levels of the
crust) accreted together during a late collisional stage.
- The present Hf dataset combined with whole-rock geochemistry indicates: (1) a
continuous differentiation trend from sodic (medium-K) to potassic (high-K) magmatism
with age, reflecting progressive enrichment of the source in HFSE and probable
shallowing of the depth of melting, and (2) that Hf isotopes alone do not represent the
diversity of rocks and processes evidenced here.
- Isotope results indicate: (1) a secular trend towards high δ18O values confirming the
involvement of metasediments in the petrogenesis of Neoarchean high-K granitoids, (2)
the presence of upper-crustal level hydrothermal systems during the Meso- and
Neoarchean magmatism in the SSFC.
- Isotopic modelling, based on available Hf from detrital zircons record complements this
dataset, indicating a major change in the evolution of the SSFC at 2.9 Ga. This change
marks the transition between a Paleoarchean regime possibly dominated by TTG
production and net crustal growth, and a Neoarchean regime that is dominated by
reworking processes, producing a wide variety of granitoid magmas with highly scattered
radiogenic isotope compositions. We attribute this transition to the onset of continental
collision in the SSFC.

This paper was supported by CNPq (grant numbers: 401334/2012-0, 402852/2012-5 and 302633/2011-1) and FAPEMIG (grant numbers:
APQ03943 and RDP 0067-10) grants. We thank Rafael Romano for providing his sample mounts. Thanks to Ana Maria Saldarriago for
picking and CL imaging grains at Portsmouth and for O isotope analysis at Edinburgh. Elena Belousova and an anonymous reviewer are
thanked for their valuable and constructive comments which improved the manuscript. Nelson Eby is acknowledged for his editorial
handling.

140
 CHAPTER 5
BORON ISOTOPE COMPOSITION (LA MC-ICP-MS) OF
TOURMALINE FROM THE QUADRILÁTERO FERRÍFERO
DISTRICT, SE BRAZIL: CONSTRAINTS ON ARCHEAN MAGMATIC-
HYDROTHERMAL EVOLUTION
5

ABSTRACT

In the Archean Quadrilátero Ferrífero district (SE Brazil), tourmaline occurs as a major
constituent in a leucogranitic intrusion, pegmatitic bodies, numerous plagioclase- and quartz-veins, as
well as disseminated in the surrounding greenstone belt rocks. The chemical and boron isotope
composition of these tourmalines was determined by electron microprobe and LA MC-ICP-MS to
investigate the hydrothermal evolution of the region. The tourmalines vary from schorl in the
leucogranite to dravite compositions in the metasediments, consistent with a decrease in Fe/(Fe+Mg)
ratio. A net increase in Cr contents from magmatic to hydrothermal tourmalines illustrates that
tourmaline major-element variations are mostly controlled by the host rock compositions. The full
range of tourmaline δ11B is from -27.1 to -10.7 ‰, with a major cluster around -12 to -18 ‰. There is
a general trend of decreasing δ11B values from tourmaline hosted in leucogranite and coarse-grained
plagioclase veins (~ -13.5 ‰) towards those present in the schists and quartz veins (~ -18 ‰). This
trend of isotope fractionation is consistent with a model of tourmaline growth from late-stage exsolved
magmatic fluids. Evidence for a separate fluid component is given by the presence in some
metasediments and crosscutting quartz veins of tourmalines with δ11B as low as -27 ‰, far beyond the
reach of fractionation due to fluid exsolution. Quartz vein-hosted tourmalines are chemically and
isotopically zoned. Two coexisting types of grains are distinguished, with opposite core-to-rim
variations: one (dominant) has an isotopically heavy (-18 ‰) Al-rich blue core, whereas the other has
a relatively light (-21 ‰), Mg-Ca-Ti-rich dark brown core. Their respective rims show chemical and
isotopic compositions intermediate between the two. We suggest that the dark brown cores formed
within the schists from an isotopically light fluid released during contact metamorphism, and the blue
cores crystallized directly from the late-stage magmatic fluids. The opposite geochemical trends in
both types of tourmaline indicate that the fluids were then partially mixed during hydrothermal
alteration of the country rocks.

C. Albert*,a, C. Lanaa, A. Gerdesb, M. Schannora, F. Narduzzia, G. Queirogaa

a
Departamento de Geologia, Universidade Federal de Ouro Preto, Morro do Cruzeiro, 35400-000, Ouro Preto, MG, Brazil
b
Institut für Geowissenschaften, Goethe-Universität Frankfurt, Altenhöferallee 1, D-60438 Frankfurt am Main, Germany
* Corresponding author: Capucine Albert. E-mail: [email protected]
Article submitted to Geochimica et Cosmochimica Acta
Keywords: Boron isotopes; Tourmaline; LA MC-ICP-MS; magmatic-hydrothermal system; Quadrilátero Ferrífero

141
5.1 INTRODUCTION

Studies of boron isotope variations in silicates have been shown to be a powerful tool to
resolve fluid-related processes, including studies of ore deposits (Krienitz et al., 2008; Xavier et al.,
2008; Pal et al., 2010; Tornos et al., 2012; Lambert-Smith et al., 2016), magmatic-hydrothermal
systems (Trumbull & Chaussidon, 1999; Kaliwoda et al., 2011; Siegel et al., 2016), metasomatic
processes (Marschall et al., 2006; Trumbull et al., 2009; MacGregor et al., 2013; Bast et al., 2014) and
geochemical cycling in the lithosphere (Ishikawa & Nakamura, 1994; Nakano & Nakamura, 2001; Ota
et al., 2008; Tonarini et al., 2011; Scambelluri & Tonarini, 2012; Konrad-Schmolke & Halama, 2014).
As the most common borosilicate in the continental crust and an exceptionally stable mineral,
tourmaline is an ideally suited candidate to study B isotope variations and the origin and evolution of
fluids it crystallized from (Dutrow & Henry, 2011; van Hinsberg et al., 2011).

In this paper, we investigate the isotope and chemical composition of tourmalines from the
Quadrilátero Ferrífero mining district of the southern São Francisco craton (SE Brazil). The
Quadrilátero Ferrífero exposes domes of Meso- and Neoarchean gneisses and granitoids in contact
with belts of Neoarchean and Paleoproterozoic volcano-sedimentary units including the Rio das
Velhas greenstone sequence. As with most Archean terranes, the region underwent a complex tectono-
metamorphic-hydrothermal evolution. Following crustal growth and craton stabilization in the Meso-
and Neoarchean, the region was affected – to various extents – by two orogenic events: the
Transamazonian orogeny in the Paleoproterozoic and the Brasiliano orogeny during the
Neoproterozoic (e.g. Alkmim & Marshak, 1998). One illustration of this polymetamorphic history is
the presence of various signs of hydrothermal activity and alteration in and around the Quadrilátero
Ferrífero. These include the presence of numerous barren pegmatitic/aplitic veins and dikes and
ubiquitous quartz veins, as well as abundant hydrothermal ore deposits, among which are some world-
class gold deposits (e.g. Lobato et al., 2001). However, the timing of hydrothermal activity, whether is
it related to mineralization or not, and the origin of the fluids are still debated.

This study focuses on barren veins and dikes in and around the Archean basement complexes
of the Quadrilátero Ferrífero, where tourmaline occurs within leucogranite and pegmatite bodies and
quartz veins, as well as disseminated crystals in the adjacent metasediments, in an attempt to
reconstruct part of its magmatic and metamorphic-hydrothermal history.

5.2 REGIONAL GEOLOGY

The Quadrilátero Ferrífero mining district (QF) is located at the southernmost tip of the São
Francisco craton in SE Brazil. In this area, the Archean-Paleoproterozoic basement includes broad
magmatic-metamorphic complexes (e.g. the Bação complex), surrounded by narrow belts of
supracrustal rocks, forming distinctive dome-and-keel geometries. The basement complexes consist of
gneisses and migmatites, intruded by several generations of granitoids, leucogranitic sheets and dikes

142
and late pegmatites and aplites. This basement was built during four main magmatic pulses: the Santa
Barbara (SB, 3220-3200 Ma), Rio das Velhas I (RVI, 2920-2850 Ma), Rio das Velhas II (RVII, 2800-
2760 Ma) and Mamona (2750-2680 Ma) events (Lana et al., 2013; Romano et al., 2013; Farina et al.,
2015a). During the Mamona event, Farina et al. (2015a) described a change in magma composition
from medium- to high-K, which was interpreted as the result of the involvement of a metasedimentary
component in the source undergoing melting. This interpretation was later supported by Hf and O
isotopes in zircon (Albert et al., 2016). The supracrustal units are divided between the Archean Rio
das Velhas Supergroup and the Paleoproterozoic Minas Supergroup and Itacolomi Group. The
metavolcanic and metasedimentary rocks of the Rio das Velhas Supergroup make up a typical
Archean greenstone belt sequence. It has been further subdivided into the Nova Lima Group, host to
world-class gold deposits, consisting of a basal mafic-ultramafic volcanic and volcano-chemical unit
overlain by volcaniclastic and resedimented associations, and the overlying Maquiné Group
representing a thick clastic association of conglomerates and sandstones (Baltazar and Zucchetti,
2007; Moreira et al., 2016). The Minas Supergroup represents a continental margin sequence
unconformably overlying the greenstone belt, and consists of a ca. 6 km-thick package of clastic and
chemical sedimentary associations (Dorr, 1969; Martínez Dopico et al., 2017). Finally the Itacolomi
Group represents a volumetrically minor unit made of metasandtones and metaconglomerates (Dorr,
1969).

Both Archean and Paleoproterozoic units display a complex deformation and metamorphic
history. In the Neoarchean, the Rio das Velhas greenstone belt was metamorphosed under greenschist
facies conditions, locally reaching lower amphibolite facies (Herz, 1978). Prior to the opening of the
Minas basin (< 2.65 Ga), the greenstone belt was affected by a deformation event which formed large
NNE-SSW verging folds and thrusts. In addition to its Archean magmatic-metamorphic history, the
Quadrilátero Ferrífero (QF) was subsequently affected by two orogenic events in the Proterozoic. The
first one occurred during the 2100 Ma Transamazonian orogeny, affecting the eastern and southeastern
margin of the QF. This orogenic event is thought to be responsible for northwest-verging folds and
thrusts affecting the supracrustal sequences (Dorr, 1969; Alkmim & Marshak, 1998) and for the
prominent dome-and-keel geometry of the QF. The dome and keel formation has been attributed by
Marshak et al. (1992, 1997) and Alkmim & Marshak (1998) to the extensional collapse phase of the
Transamazonian orogeny. Marshak et al. (1997) obtained a Sm-Nd age of 2095 ± 65 Ma from the
metamorphic aureole of the dome-border shear zone, which is in agreement with U-Pb monazite and
titanite metamorphic ages obtained from the basement (Machado et al., 1992; Aguilar et al., 2017).
The final deformation event that affected the QF formed west-verging thrust folds, reactivating and
overprinting pre-existing structures, and was attributed to the Neoproterozoic Brasiliano orogeny (650-
480 Ma) (Chemale et al., 1994; Alkmim & Marshak, 1998).

143
5.3 SAMPLE LOCATIONS AND TOURMALINE OCCURRENCES

Samples from the QF were collected from two locations (Fig. 5.1): one at the southeastern
margin of the Bação complex and the second at its northeastern margin. Table 5.1 presents a summary
of the tourmaline samples investigated in this study.

Figure 5.1 – (a) Simplified map of the São Francisco craton highlighting the location of the exposed Archean
provinces and the bordering Neoproterozoic orogenic belts; (b) Geological map of the Quadrilátero Ferrífero
mining district; (c) Map of the Bação complex and surrounding country rocks, showing the location of the
samples studied. The star shows the location of a pegmatite sample dated by Farina et al. (2015a) (see text).

144
 At the southeastern margin of the Bação complex, the basement gneisses are intruded by (1) a
small biotite-poor medium- to coarse-grained granitic intrusion dated at 2716 ± 5 Ma (Romano et al.,
2013), (2) a small tourmaline-bearing medium-grained leucogranite intrusion, and (3) numerous
pegmatitic and aplitic dikes and veins, also found within the Nova Lima schists close to the border of
the dome, both crosscutting and parallel to the schistosity. The first sample was taken from the
tourmaline-bearing leucogranite, a few tens of metres from the contact with the supracrustals (Fig.
5.2a). It consists of quartz, plagioclase (sericitized and saussuritized), K-feldspar and tourmaline as the
only mafic phase, with minor garnet and secondary muscovite. Tourmaline forms anhedral crystals up
to 2 mm in diametre, it is moderately pleochroic, locally with blueish centres and displays light to dark
green optical zoning (Fig. 5.3a). It often contains inclusions of quartz, plagioclase and K-feldspar. In
Back Scatter Electron (BSE) images tourmaline displays zoned magmatic cores, surrounded by
homogeneous domains with curviplanar boundaries embayed from the grain rim to its interior (Fig.
5.4a). Some of the most anhedral grains are entirely homogeneous in BSE contrast. The tourmaline-
leucogranite is in tectonic contact with the schists of the Nova Lima Group (Fig. 5.2a), from which
samples PM-1J/N and PM-6M/P were collected. In this area, the Nova Lima schists mostly consist of
biotite, plagioclase and quartz, with variable amounts of retrograde chlorite (from ~ 3 vol.% in PM-1N
to ~ 20 vol.% in PM-1J) and sericite. Around PM-6, the schists locally also contain garnet and/or
staurolite. Secondary muscovite is present in all four samples as fine interstitial needles, in variable
but small proportions. In PM-1N, the schist is intruded by a coarse-grained plagioclase-tourmaline
vein, occurring parallel to the schistosity (Fig. 5.3b). Tourmaline forms porphyroblastic crystals in the
schist where its abundance increases with proximity to the vein, and petrographic observation shows
that tourmaline appears identical in both the vein and the schist, with euhedral habits, brown with local
patchy blueish centres. In PM-1J and PM-6M/P, tourmaline forms euhedral to subhedral crystals of 20
(in PM-6P) to 350 μm in diametre (in PM-6M) that occur parallel to but overgrowing the biotite +
chlorite assemblages defining the foliation (Fig. 5.3c). In all schist samples, tourmaline commonly
contains inclusions of quartz and feldspars. In BSE images, these tourmalines display a layering
texture defined by the alternation of darker and lighter bands, that is continuous to that of the
surrounding biotite and chlorite assemblage, indicating that tourmaline grew by replacement of the
micas (Fig. 5.4b).

145
Figure 5.2 – (a) Tectonic contact between the tourmaline-bearing leucogranite (southeastern Bação complex)
and the surrounding schists of the greenstone belt (Nova Lima group); (b) Photograph illustrating the typical
field relationships around the Bação complex between the hydrothermal veins and pegmatite dikes and the
country rocks. The view shows a pegmatitic vein intruding the Nova Lima schists (sample H1) at the
northeastern border of the Bação complex. Inset: pluricentimetric pocket of Tur (+ Qz?), deformed parallel to the
schistosity. The dotted lines represent the schistosity. Fds – Feldspars; Ms – Muscovite; Tur – Tourmaline.

At the northeastern border of the complex, the sampling site is situated within the schists of
the Nova Lima Group, a few tens of metres away from the contact with the basement gneisses. The
schistosity becomes progressively shallower away from the contact. The schist is intruded by: (1)
feldspar- and muscovite-rich decimetric pegmatoids, (2) abundant centimetric to decimetric quartz-
tourmaline veins, and (3) centimetric to decimetric felsic apophyses. Mostly, these veins and dikes are
found crosscutting the host schist, but they also appear in minor proportions deformed together with
the schistosity (Fig. 5.2b). No zircon ages were obtained from this outcrop, but Farina et al. (2015a)

146
dated a pegmatite body intruded within the basement ~ 15 km further south (Fig. 5.1c) and obtained a
crystallization age of 2693 ± 13 Ma. Four samples were collected from this outcrop. These include a
weathered and pale purple sample of the host schist (H1), a sample from one of the crosscutting felsic
apophyses (IRV02), and two samples from crosscutting quartz-tourmaline veins (FC2A and B, FC2A
being more coarse-grained). The host schist (H1) comprises, by order of decreasing abundance biotite,
plagioclase, quartz, muscovite and chlorite. The rock contains large (up to 1-2 cm) sigmoidal
porphyroclasts of plagioclase + quartz. Tourmaline typically occurs as aggregates parallel to the main
schistosity. Aguilar et al. (2017) dated some monazite grains from this schist and obtained an upper
intercept age of 2617 ± 9 Ma, interpreted as a metamorphic age for this rock. Sample IRV02 was
collected from a small felsic (mostly quartz and feldspars) apophysis intruded in the schist. Both
samples from the schist and the felsic apophysis were too weathered to allow thin sections to be made,
but tourmaline was present in the mineral separates of these rocks (< 250 μm), and was handpicked
and mounted in epoxy for analysis. In BSE images, these crystals exhibit variable internal structures,
including oscillatory zoning, normal zoning (dark core, bright rim), complex zoning (oscillatory-zoned
cores truncated by a bright rim), and homogeneous grains Tourmaline from the quartz-tourmaline
veins forms euhedral, randomly oriented crystals or grain aggregates. In the more coarse-grained
sample (FC2A), tourmaline is heavily fractured and fragmented. In both tourmaline-quartz veins
samples, two types of tourmaline cores were observed: (1) well-defined blue cores mantled by brown
rims, present in FC2A only (Type A; Fig. 5.3d and 5.4c), and (2) dark brown cores surrounded by
lighter brown rims, present in minor proportions in both FC2A and FC2B (Type B; Fig. 5.3e and 5.4d
and e). A special feature of type B cores is the development of sector zoning apparent in BSE images
(Fig. 5.4d). Both types of cores are also distinguished clearly on the basis of their geochemical and
isotopic compositions (see section 4.5). In addition, tourmaline overgrowths (interpreted here as a
second generation) occurs either as replacement veinlets crosscutting the main-stage tourmaline, as
replacement on the edges of quartz-filled cracks, as thin outer rims on zoned crystals and as small
disseminated grains (Fig. 5.3d).

147
Figure 5.3 – Photomicrographs of tourmaline occurrences. (a) Tourmaline-bearing leucogranite (sample FC1)
with assemblage quartz + plagioclase (sericitized) + K-feldspar + tourmaline; (b) Biotite-plagioclase-quartz-
chlorite schist (PM-1N) intruded by plagioclase-tourmaline vein (to the left). Tourmaline is also present in the
schist overgrowing the schistosity; (c) Subhedral tourmaline crystals aligned parallel to and overgrowing the
schistosity (sample PM-6M) (cross polarized light); (d) Type A (well-defined blue core) and (e) type B (dark
brown core) tourmaline from quartz-tourmaline vein (sample FC2A). Late-stage tourmaline occurs along cracks
and on the outer rims of the grains in (d). Black circles mark the ablated spots from LA-ICP-MS analyses. Bt –
Biotite; Chl – Chlorite; Kfs – K-Feldspar; Pl – Plagioclase; Qz – Quartz; Tur – Tourmaline.

148
Figure 5.4 – Back-scattered electron (BSE) images of representative tourmaline. (a) Magmatic tourmaline in
leucogranite sample FC1, showing light to dark oscillatory-zoned centres and lighter surrounding replacement
domains; (b) Tourmaline in a biotite-quartz matrix in schist sample PM-6M. The grains display an alternation of
dark and light bands suggesting they grew at the expense of the micas; (c) Type A tourmaline from quartz-
tourmaline vein (sample FC2A); (d) Tourmaline and quartz intergrowth in fine-grained hydrothermal vein
(sample FC2B). Note the presence of some type B (lighter, sector zoned cores) tourmaline in minor proportions;
(e) Type B tourmaline from coarse-grained quartz-tourmaline vein (sample FC2A). Abbreviations same as
Figure 5.3.

149
Table 5.1 – Summary of tourmaline samples investigated during this study. Mineral abbreviations are after Whitney & Evans (2010); Bt = biotite, Chl = chlorite, Grt = garnet,
Kfs = K-feldspar, Ms = muscovite, Pl = plagioclase, Qz = quartz, St = staurolite, Tur = tourmaline.
Sample UTM long UTM lat Sample description Mineral association Tur optical characteristics Tur diametre
Moderate pleochroism, light to dark
Anhedral Tur (10-15 vol.%) associated with Qz, Pl green optical zoning, locally with small
(sericitized and saussuritized) and Kfs, with minor blue centres. In BSE images, Tur crystals
FC1 643520 7747335 Tur-bearing leucogranite 0.2-2 mm
Grt and secondary Ms. Tur contains inclusions of Qz, have oscillatory zoned centres
Pl and Kfs surrounded by homogeneous rims
indicating dissolution-reprecipitation
Part A: vein containing anhedral Tur (30-40 vol.%)
Tur is brown, locally with blueish
and Pl (sericitized and carbonatized), with secondary 0.5-3 mm
centres, and displays weak pleochroism
Ms. Tur contains abundant Pl inclusions
Coarse-grained Pl-Tur vein (Part A) Part B: Bt (30)-Pl (35)-Qz (20)-Chl (3) schist with
PM-1N 643281 7746557
intruding Bt-Pl-Qz-Chl schist (Part B)opaque phases and secondary Ms. Tur (~ 10 vol.%) is
porphyroblastic, euhedral to subhedral, and contains Tur is brown, locally with blueish
0.5-1.5 mm
Pl inclusions. Tur is overgrowing the schistosity, and centres, and displays weak pleochroism
its abundance decreases away from the contact with
the vein
Euhedral to subhedral disseminated Tur (~ 3 vol.%)
Tur crystals are brown and exhibit
Bt-Pl-Qz-Chl schist, crosscut by late in Bt (10-15)-Pl (30)-Qz (30)-Chl (20) schist with
PM-1J 643281 7746557 moderate pleochroism, with no optical 50-300 μm
Qz veinlets (~ 1 mm wide) minor opaque phases and secondary Ms. Tur crystals
zoning
are overgrowing the schistosity
Euhedral to subhedral Tur (~ 4 vol.%) in Bt (30)-Pl
Tur crystals are brown, locally with
(35)-Qz (25)-Chl (5) schist with opaque phases and
PM-6M 645097 7750235 Bt-Pl-Qz-Chl schist blueish centres and exhibit moderate 80-350 μm
secondary Ms. Tur crystals are lineated parallel to
pleochroism
and overgrowing the schistosity
Euhedral to subhedral Tur (~ 1-2 vol.%) in Bt (15)-Pl
Tur crystals are brown, locally with
Bt-Pl-Qz-Chl schist, intruded by Qz (40)-Qz (35)-Chl (10) schist with minor St, opaque
PM-6P 645097 7750235 blueish centres and exhibit moderate 20-150 μm
veinlets (0.5-1 cm wide) minerals and secondary Ms. As in PM-6M, Tur
pleochroism
occurs parallel to and overgrowing the schistosity
In BSE images, Tur grains expose highly
variable internal structures, including
Bt-Pl-Qz-Ms-Chl schist, intruded in 'normal' zoning (dark core, bright rim),
the field by: (1) Pl- and Ms-rich Bt-Pl-Qz-Ms-Chl schist. Tur forms aggregates complex zoning (oscillatory zoned centre
H1* 638670 7762034
pegmatoids, (2) numerous Tur-Qz parallel to the schistosity and homogeneous rim), oscillatory
veins, and (3) felsic apophyses zoning (alternance of dak and bright
bands along the c axis), and
homogeneous grains

150
 Sample UTM long UTM lat Sample description Mineral association Tur optical characteristics Tur diametre
Moderate pleochroism. 90-95 % of Tur display
well-defined blue cores surrounded by
Randomly oriented, euhedral Tur (~ 80 vol.%) in Tur-Qz
homogeneous brown rims (Type A). 5-10 %
Coarse-grained Tur-Qz vein vein. Tur grains occur as aggregates or disseminated crystals.
FC2A 638670 7762034 exhibit a dark brown core surrounded by a brown 0.5-10 mm
crosscutting the schist Tur is heavily fractured and fragmented, with Qz filling up
rim (Type B). Late-stage Tur occurs as
the cracks
replacement along cracks, thin outer rims and
irregular patches
Moderate pleochroism. Most Tur grains are
Randomly oriented, euhedral Tur (~ 70 vol.%) in Tur-Qz brown, sometimes displaying ill-defined blueish
Tur-Qz vein (more fine-grained than vein. Tur mostly occurs as aggregates, but also as centres. As in FC2A, a minority of Tur have
FC2B 638670 7762034 10-200 μm
FC2A) crosscutting the schist disseminated individual grains or fine needles in the Qz distinctly different dark brown cores (Type B). In
matrix BSE images, Type B cores display well-defined
sector zoning
In BSE images, the Tur either expose 'normal'
IRV02* 638670 7762034 Felsic apophysis intruding the schist
zoning, oscillatory zoning or homogeneous grains
* Tur handpicked from mineral separate and mounted in epoxy. The grain size for these Tur is not indicated as it is not representative of the sample (i.e. the separate only contains grains < 250
μm)

151
5.4 ANALYTICAL METHODS

Electron-Probe Microanalyses (EPMA) of tourmalines (mounted in epoxy and in thin

sections) were performed using a JEOL JXA-8230 at the Universidade Federal de Ouro Preto (UFOP).
Operating conditions were 15 kV accelerating voltage, 20 nA beam current and 10 μm beam diametre.
Appropriate natural and synthetic reference materials were used for calibration. Tourmaline structural
formulae were calculated by normalizing to 15 cations in the tetrahedral and octahedral sites (T + Z +
Y) and assuming 3 boron atoms per formula unit (Henry & Dutrow, 2002) using the Excel spreadsheet
of Tindle et al. (2002).

For B isotope analyses, we used a Thermo-Scientific NeptunePlus multi-collector ICP-MS

coupled to a Photon Machines G2 193 Excimer laser ablation system. The laser was operated with a
fluence of about 6-7 J/cm2 at a repetition rate of 15 Hz with a laser spot diametre of 20 μm. Samples
were ablated in a He atmosphere (0.3 L/min) using a Helix ablation cell. The 10B and 11B ion beam
intensities were measured on the L2 and H2 detectors respectively, in low resolution mode. Isotope
ratios for each measurement were collected in a single block of data consisting of 98 cycles with an
integration time of 0.5 s/cycle.

Data were processed offline using an in-house spreadsheet. The intensity of the gas blank was
on average 0.006 V. Following correction of the measured signal intensities for the gas blank, the
11
measured B/10B ratio of each unknown sample was corrected for instrumental mass fractionation
(IMF) using a standard-sample bracketing approach, with schorl (#112566) tourmaline Reference
Material (RM) (Dyar et al., 2001) as calibrant, in an effort to minimize potential matrix effects
11
(Mikova et al., 2014). The instrumental mass discrimination of the measured B/10B ratio was
11
typically 11 %. The drift-corrected ratios were referenced to the published B/10B value of the RM
and the results are reported as δ11B values relative to NIST SRM 951 boric acid using its certified
11
B/10B value of 4.04362 ± 0.00137 (Catanzaro et al., 1970).

The reproducibility of the data was monitored during the five days of analyses using
tourmaline dravite (#108796, δ11B = -6.60 ‰) and elbaite (#98144, δ11B = -10.40 ‰) (Leeman &
Tonarini, 2001) as well as tourmaline B4 (schorl, δ11B = -8.71 ‰) (Gonfiantini et al., 2003) as
external RMs. Over the five days of analyses, data were treated as seven separate analytical sessions
(i.e. after sample exchange in the ablation cell) to avoid potential effects due to different analytical
conditions. The observed internal precision for individual analyses varies between 0.002 and 0.04 %
(mean of 0.01 %). Following IMF corrections, a systematic offset towards heavier values was noted
between the measured and the certified 11B/10B ratios of all external RMs (Fig. 5.5). This offset was
corrected for each session of analyses using a best-fit estimate through the three external tourmaline
RMs. The applied correction varied from 0.9993 to 0.9988, which corresponds to an offset of 0.7 to
1.2 ‰ (mean of 0.9 ‰). Interestingly, we note that this offset was systematic, and irrespective of the

152
composition of the tourmaline RM: no significant matrix effect was observed, which differs from the
conclusion of Mikova et al. (2014). The origin of this offset is unknown, but it is important to state
that 0.9 ‰ is well within the range of long-term interlaboratory reproducibility of RMs for B isotope
analyses (~ 2.1 ‰ (1SD); Gonfiantini et al., 2003). During the five days of analyses, we found no
time-dependent trend in the measured 11B/10B values of the different RMs. Reported uncertainties were
propagated by quadratic addition of the within-run precision of each analysis (standard error), the
external reproducibility (standard deviation) of the calibrating RM (used for IMF correction), and the
external reproducibility of the three external RMs (used for offset correction). A summary of the B
isotope data obtained for all RM over the five days of analyses is given is Table 5.2.

Overall, the data presented in this study compares favourably with previously published data
on tourmaline using both SIMS (e.g. Trumbull et al. 2008, 2013; Farber et al., 2015) and LA-ICP-MS
techniques (Tiepolo et al., 2006; Mikova et al. 2014; Martin et al., 2015), with analytical uncertainties
and long-term reproducibility similar to or higher than those associated with SIMS analyses.

Figure 5.5 – B isotope compositions obtained for B4, dravite and elbaite Reference Materials (RM) over the five
days of analyses. Vertical dotted lines indicate changes in analytical conditions during sample exchange in the
ablation cell. The light and dark grey spots represent analyses before and after offset corrections respectively
(see text). The corresponding grey fields indicate the average ± standard deviation (1SD) of the data. The red
line indicates the accepted value (PTIMS) of each RM.

153
Table 5.2 – Summary of LA MC-ICP-MS B isotope analyses of reference tourmaline.
11
Date/Analytical session B/10Ba 1SE IMF Offset δ11B (‰) ±1ζ (‰) n
11 10 11
Harvard 108,796 Dravite ( B/ B = 4.017 and δ B = -6.60 ‰)
14.09.2016 4.504 0.003 0.893 0.9993 -6.36 0.67 11
15.09.2016 4.496 0.002 0.895 0.9989 -6.20 0.51 6
4.504 0.001 0.893 0.9988 -6.07 0.30 3
16.09.2016 4.507 0.002 0.892 0.9992 -6.05 0.41 4
4.498 0.001 0.894 0.9991 -6.33 0.21 6
19.09.2016 4.495 0.002 0.895 0.9993 -6.32 0.65 14
20.09.2016 4.532 0.002 0.887 0.9993 -6.37 0.66 10
Mean 4.505 0.002 0.893 0.9991 -6.24 0.49 54
Repeatability (1SD) 0.35

Harvard 98,144 Elbaite (11B/10B = 4.001 and δ11B = -10.40 ‰)

14.09.2016 4.486 0.003 0.892 0.9993 -10.64 0.66 5
15.09.2016 4.476 0.002 0.895 0.9989 -10.73 0.50 9
4.483 0.001 0.893 0.9988 -10.90 0.31 6
16.09.2016 4.486 0.002 0.892 0.9992 -10.82 0.41 7
4.478 0.001 0.894 0.9991 -10.81 0.21 6
19.09.2016 4.476 0.002 0.894 0.9993 -10.58 0.65 14
20.09.2016 4.512 0.002 0.887 0.9993 -10.71 0.66 10
Mean 4.485 0.002 0.893 0.9991 -10.74 0.49 57
Repeatability (1SD) 0.42

IAEA B4 (Schorl) (11B/10B = 4.015 and δ11B = -8.71 ‰)

14.09.2016 4.492 0.003 0.893 0.9993 -9.05 0.67 5
15.09.2016 4.485 0.002 0.895 0.9989 -8.52 0.51 6
4.494 0.001 0.893 0.9988 -8.83 0.31 6
16.09.2016 4.494 0.002 0.893 0.9992 -8.98 0.40 4
4.488 0.001 0.894 0.9991 -8.61 0.20 6
19.09.2016 4.483 0.002 0.895 0.9993 -8.89 0.65 14
20.09.2016 4.521 0.002 0.887 0.9993 -8.78 0.66 10
Mean 4.494 0.002 0.893 0.9991 -8.81 0.49 51
Repeatability (1SD) 0.33
a
Corrected for gas blank

5.5 RESULTS

The complete sets of EPMA and boron isotope results are given in Appendices A13 and A14.

5.5.1 Tourmaline classification and major element variations

Most of the tourmalines fall within the field of tourmalines from Al-rich and Al-poor
metapelites and meta-arenites of Henry & Guidotti (1985) (Fig. 5.6a). A few exceptions include type
B cores of sample FC2, which plot on the Al-poor and Mg-rich side of the schorl-dravite join into the
field of Fe3+-rich metapelite, quartz-tourmaline and calc-silicate rocks, as well as tourmaline from the
leucogranite which partly fall within the field of Li-poor granite, pegmatite and aplite. All the
tourmalines are classified as ―alkali‖ in the nomenclature of Hawthorne & Henry (1999) with low Ca
and moderate vacancy on the X-site (Fig. 5.6b). Tourmalines from the QF belong to the schorl-dravite
series, with Fe/(Fe+Mg) ranging from 0.30 to 0.87 (Fig. 5.7a). Some of the crystal-chemical controls

154
on the observed compositional variations of individual samples are represented on plots of X-site
vacancies vs. Fe and Al per formula unit (p.f.u) (Fig. 5.7b and c). These plots indicate that the increase
in Al content in excess of 6 cations p.f.u (ideal schorl-dravite solid solution) is dominantly controlled
by the magnesiofoitite exchange vector (□,Al)(Na,Mg)-1.

Figure 5.6 – Ternary diagrams showing tourmaline compositions. (a) Al-Fe-Mg diagram after Henry & Guidotti
(1985). Labelled fields are: (1) Li-rich granitoids, pegmatites and aplites, (2) Li-poor granitoids, pegmatites and
aplites, (3) Fe3+-rich quartz-tourmaline rocks (altered granitoids), (4) metapelites and metapsammites with Al-
saturating phase, (5) metapelites and metapsammites lacking Al-saturating phase, (6) Fe3+-rich quartz-tourmaline
rocks, calc-silicate rocks and metapelites, (7) low-Ca metaultramafic rocks and Cr-V-rich metasediments, (8)
metacarbonates and meta-pyroxenites; (b) Classification diagram according to the principal constituent on the X-
site (Hawthorne & Henry, 1999).

155
Figure 5.7 – Chemical compositions of tourmaline expressed in terms of (a) Fe/(Fe+Mg) vs Na/(Na+Ca); (b) Fe
vs X-site vacancy; (c) Altotal vs X-site vacancy and (d) Fe/(Fe+Mg) vs Cr2O3 (wt.%). Symbols are the same as in
Figure 5.6.

Significant differences exist between tourmaline hosted by the leucogranite, the quartz veins
and the schists. In particular, they can be distinguished in terms of their ferromagnesian contents (Fig.
5.7a), with Fe/(Fe+Mg) ratio increasing from the schist-hosted tourmalines (0.30 to 0.59) compared to
those in the quartz veins (0.43 to 0.70) and the leucogranite (0.66 to 0.87). Another distinctive feature
of the tourmalines disseminated in the schists is their higher chromium content (up to 0.75 wt.%; Fig.
5.7d).

Tourmalines from the schists show no significant variations in composition within individual
samples, whereas grains from the leucogranite and the quartz veins have systematic core-to-rim
compositional variations. In the leucogranite (FC1), the replacement domains are characterized by a
depletion in Al and a slight enrichment in Fe, Ti and Ca compared to the oscillatory-zoned cores (not
shown). In the quartz veins (FC2), type A cores are typically characterized by high Al and X-site
vacancies and low Ti and Ca contents, whereas type B cores show opposite compositional trends
including a significant enrichment in Ti (up to 2.5 wt.%), Ca and Mg (Fig. 5.8). Both types of cores
are surrounded by light brown rims whose major element compositions tend towards intermediate
values between both core end-members (Fig. 5.8).

156
Figure 5.8 – Plots of (a) MgO (wt.%) vs CaO (wt.%) and (b) Al2O3 (wt.%) vs TiO2 (wt.%), illustrating the
compositional zoning between type A and B cores of tourmalines from sample FC2 and their respective rims.

5.5.2 Boron isotope compositions

The total range in δ11B isotope compositions for the tourmalines investigated in this study is
from -27.1 to -10.7 ‰ (Fig. 5.9). These values fall within the lower end of the range of B isotope
compositions of tourmalines from granites and granite-related veins worldwide (Fig. 5.9g)
(Chaussidon & Albarède, 1992; Marschall & Jiang, 2011).

The most significant observations from this dataset are as follows:

- There is a general decrease in δ11B values of magmatic tourmaline from the leucogranite
to hydrothermal tourmaline in the schists and in the quartz veins.
- B isotope variations within individual samples range generally between 1.5–2 ‰ in the
schists and 3–4 ‰ in the quartz veins and leucogranite (one grain at ~ -11 ‰ in IRV02
was excluded). Tourmaline from schist sample H1 has the largest variation of δ11B values
ranging from -16.6 to -27.1 ‰, with two main populations at -18.5 and -26.3 ‰ (Fig.
5.9e). Despite this scatter, neither significant nor systematic within-grain isotope
variations were observed, and these isotope results could not be correlated clearly with the
tourmaline internal structures observed in BSE images. Such variations are beyond the

157
 analytical uncertainty of the LA-MC-ICP-MS method, suggesting the involvement of B
fractionation processes and/or multiple B sources during tourmaline growth.
- Tourmalines in the schists show no B isotope zoning, but variations were observed in
tourmaline from the leucogranite and the quartz veins. In the leucogranite (FC1), the
oscillatory-zoned cores are on average 1.1 ‰ heavier than the homogeneous overgrowths
(Fig. 5.9a). However, we observed no isotope variations within the oscillatory-zoned
cores, in contrast to what was expected. In the quartz veins, we observed opposite trends
of isotope zoning, with type A cores being isotopically heavier (-17.7 ‰) and type B cores
being systematically lighter (-20.9 ‰) than their respective rims (-19.2 ‰ on average; Fig.
5.9d).

5.6 DISCUSSION

5.6.1 Tourmaline growth

It is essential to determine the timing of tourmaline growth in the geologic history of the QF in
order to interpret both geochemical and isotopic data.

The first order important observation is that tourmaline occurs in all products of granite
magmatism and associated hydrothermal activity in the QF. Indeed, its presence essentially within and
at the edges of the granitic-gneissic complexes strongly advocates for granite-derived B metasomatism
(Herz, 1978). Where it is present in the schists, tourmaline always occurs parallel to the foliation, yet
clearly replacing the biotite and chlorite, indicating a post-metamorphic timing of formation. The B-
rich hydrothermal fluids are unlikely to be of metamorphic origin (i.e. formed from the progressive
release of B during mineral breakdown) for two main reasons: (1) the first forceful argument is that
tourmaline distribution is not widespread throughout the greenstone belt as would be expected of a
regional metamorphic product, but is instead restricted to the vicinity of the basement domes, and (2)
tourmaline occurrence in the schist sample PM-1N is clearly associated with the crosscutting
plagioclase-tourmaline vein (Fig. 5.3b), therefore suggesting an external influx of B-rich fluid which is
likely to be derived from the adjacent tourmaline-bearing leucogranite.

The last magmatic episode recorded in the QF was the Mamona event at 2750-2680 Ma,
during which high-K granitoids were emplaced at upper crustal levels. We did not obtain any precise
geochronological constraint in our tourmaline-bearing samples because of the absence of datable
minerals, but some indirect age information do exist. Farina et al. (2015a) dated a pegmatite dike in
the centre of the Bação complex and obtained a crystallization age of 2693 ± 13 Ma (Fig. 5.1c).
Aguilar et al. (2017) performed U-Pb dating on separated monazite grains from the schist sample H1
and obtained an age of 2617 ± 9 Ma, which the authors interpreted as a metamorphic age for the rock.
Although it is slightly younger than the Mamona event, this age is similar to that of a small granitoid
intrusion (2612-2613 Ma; Romano et al., 2013) and several metamorphic ages obtained throughout the

158
QF (Farina et al., 2015a; Albert et al., 2016). Thus we tentatively propose that this monazite age of
2617 Ma represents a good estimate for the age of hydrothermal activity and tourmaline growth, at
least in the northern part of the Bação complex.

The presence of B-rich phases in granite and/or granite-derived products is controlled by the
presence and stability of B-rich phases in the protolith, the fractional crystallization of such melts and
the associated generation of hydrothermal fluids in which B is partitioned. Based on geochemical and
isotope evidence, Farina et al. (2015a) and Albert et al. (2016) argued for the involvement of a
metasedimentary protolith for the high-K granitic melts produced during the Mamona event.
Subsequent melt fractionation produced a range of pegmatitic/aplitic melts and hydrothermal fluids
which often contain tourmaline. These late stage magmatic fluids are enriched in B, which as a
strongly incompatible element is preferentially partitioned into the fluid and vapor phase. Tourmaline
then crystallizes in the schists adjacent to the granite intrusions, where B-rich fluids mix with host-
rock derived ferromagnesian components.

It is well-established that in hydrothermally altered rocks, tourmaline major element

composition is mostly controlled by that of the host rock (Henry & Guidotti, 1985; London &
Manning, 1995; Slack & Trumbull, 2011). Indeed, tourmaline from the QF bears evidence for a strong
protolith control. Most of the tourmalines from this study plot in the field of tourmalines from Al-rich
and Al-poor metapelites and meta-arenites (Fig. 5.6a), suggesting that the composition of the fluid had
little influence on the composition of these tourmalines. Particularly, sedimentary-hosted tourmaline
displays systematically higher Cr contents than magmatic and quartz vein-hosted tourmaline (Fig.
5.7d). As a fluid-immobile element, Cr is unlikely to have been transported by the hydrothermal fluid
itself, and hence the higher Cr content in sedimentary-hosted tourmaline is interpreted to be derived
from the host country rocks. In fact, chromian chlorite has been found in a variety of rocks, including
the schists of the Nova Lima Group (Herz, 1978).

5.6.2 Boron isotope variations and fluid evolution

The first order observation is the general decrease towards lighter B isotope compositions
from magmatic to hydrothermal tourmaline. Before attempting to identify the processes that controlled
the observed variations, it is important to review the potential factors that can influence the
fractionation of B in a magmatic-hydrothermal system.

The partitioning of 11B and 10B is mostly controlled by the coordination environment of boron
11 10
between coexisting phases, with B occurring in trigonal B(OH)3 complexes whereas B is
partitioned into tetrahedral B(OH)4 complexes. The fractionation between both isotopes is also
temperature dependent (Palmer et al., 1992; Williams et al., 2001; Hervig et al., 2002; Wunder et al.,
2005; Meyer et al., 2008). Fractionation is expected to be large between phases with different boron
coordination, and small between phases with equally coordinated boron. For example, experiments

159
between synthetic boromuscovite and fluid from Wunder et al. (2005) indicate that mica is enriched in
10
B compared to the fluid, with Δ11Bmica-fluid = -10.9 at 500 °C. This is due to the substitution of boron
for tetrahedrally coordinated aluminium in the mica, whereas boron in near-neutral fluids occurs as
trigonal complexes (Schmidt et al., 2005). On the other hand, tourmaline-fluid boron isotope
fractionation is comparatively small (Δ11Btourmaline-fluid = -2 ‰ at 500 °C; Meyer et al., 2008; or -0.6 ‰
at 500 °C; Kowalski et al., 2013), as would be expected because boron in tourmaline – unlike in most
silicate minerals – occurs almost exclusively as trigonal complexes. In contrast, B partitioning is not
well known in silicate melts, and B partitioning between such melts and tourmaline has, to date, not
been investigated. In silicate melts, B can occur as trigonal or tetrahedral complexes, depending on
factors such as water, Si and B contents and alkali/aluminium ratio (Dingwell et al., 2002 and
references therein). Hervig et al. (2002) experimentally investigated the B fractionation between
silicate melts and aqueous fluids and found that B isotope fractionation is high between both phases
(Δ11Bmelt-fluid ~ -8 ‰ at 650 °C), suggesting that B is mostly tetrahedrally coordinated in the melt.
However, several recent studies have argued that these predictions may be too high, based on
contradicting empirical evidence that a significant proportion of B in granitic melt may be present as
trigonal complexes (Kaliwoda et al., 2011; Trumbull et al., 2013; Siegel et al., 2016). For example,
Tonarini et al. (2003) measured high proportions (> 75 %) of trigonally coordinated B in three natural
rhyolitic glasses. Similarly, Geisinger et al. (1988) reported ca. 60 % of trigonally coordinated B in an
albite-rich glass. Following Trumbull et al. (2013) and Siegel et al. (2016), we will assume a 50:50
proportion of tetrahedral and trigonal B in the melt, which implies a reevaluation of the melt-fluid B
partitioning to half of the value predicted by Hervig et al. (2002), from -8 ‰ to -4 ‰ at an estimated
near-solidus temperature of 650 °C for the leucogranite. Combined with a Δ11Btourmaline-fluid = -1 ‰ at
650 °C (Meyer et al., 2008), B fractionation between melt and tourmaline is predicted to be about -3
‰.

Magmatic tourmaline from the QF leucogranite has a mean δ11B value of -13.5 ± 0.5 ‰
(average of analyses on oscillatory-zoned cores, Fig. 5.9a). Assuming a melt-tourmaline fractionation
factor of -3 ‰ as discussed above, the initial granitic melt would have a bulk δ11B value of -16.5 ‰.
Using melt-fluid and tourmaline-fluid fractionation factors of -4 ‰ (at 650 °C) and -2 to -4 ‰ (over
an estimated temperature range of 500 to 300 °C for hydrothermal tourmaline formation) respectively,
the expected range of B isotope compositions for granite-derived hydrothermal tourmalines would be
between -13.6 and -17.5 ‰ (dotted line in Fig. 5.9). Most of the tourmalines from the QF fall inside
this range of values within error. A few exceptions include some tourmaline grains from samples FC2
and H1, which will be discussed in the next section.

160
161
Figure 5.9 – (a) to (f) Frequency distribution diagrams of tourmaline B isotope composition and average δ 11B (±
1SD) per sample. The black dotted lines show the modelled B isotope fractionation between melt, fluid and
tourmaline (see text for explanation); (g) Compilation of tourmaline B isotope composition in granite and
granite-related veins (data from Marschall & Jiang, 2011).

We suggest that the best explanation for the general trend towards lighter isotope
compositions is late-stage magmatic degassing. Tourmalines in the leucogranite and in the coarse-
grained plagioclase-tourmaline vein (PM-1N) have identical B isotope compositions (means of -13.5
and -13.6 ‰ respectively). This can be expected if tourmaline in the vein crystallizes before magmatic
degassing. No B fractionation would occur, and the isotope composition of the tourmaline in the vein
would then be no different than that in the parental leucogranite. In this scenario, the lower δ11B values
in tourmaline grains from the schist samples PM-6M/P (from -14.7 to -16.2 ‰) and in some grains of
the quartz-tourmaline veins (as low as -15.4 ‰) can be explained if they formed after the exsolution of
a 11B-rich fluid. Additionally, this decreasing trend could potentially reflect decreasing temperatures,
whereby tourmaline in the quartz veins crystallizes at lower temperature than that in the plagioclase
vein. In this case, the lighter isotope composition of quartz vein hosted tourmaline is consistent with
the increasing B isotope fractionation at low temperature. Interestingly, we note that this trend is also
consistent with the increasing distance away from the tourmaline-bearing leucogranite (see Fig. 5.1c).

In the leucogranite, the lobe domains surrounding the magmatic cores (Fig. 5.4a) closely
match the description of replacement textures produced by fluid-mediated dissolution-reprecipitation
as described in monazite grains for example (e.g. Harlov et al., 2011). Stability experiments indicate
that tourmaline becomes unstable in the presence of alkali-rich fluids (Morgan and London, 1989;
London et al., 2002). In order for dissolution-reprecipitation process to produce tourmaline rims with a
slightly lower B isotope composition as observed in this sample (~ 1.1 ‰ on average; Fig. 5.9a), the
reactive fluid should itself be slightly isotopically lighter than the melt from which the magmatic cores
grew. Based on the considerations discussed above, this shift towards lighter δ11B values in the rims
could result from either tourmaline crystallization or removal of an exsolved fluid (both would remove
11
B from the remaining leucogranitic melt or fluid). However the shift towards lower isotope
11
compositions is abrupt, and not progressive as would be expected from continuous B depletion
during tourmaline growth. We therefore suggest that the removal of a small volume of exsolved fluids
caused the small shift in isotope composition between the magmatic cores and the replacement rims in
sample FC1.

The observed isotope trends from magmatic to hydrothermal tourmaline in this study are very
similar to those reported by Trumbull & Chaussidon (1999) in the Archean Sinceni granite-pegmatite
system in Swaziland. These authors observed an overall decrease towards lighter δ11B values from
magmatic to hydrothermal settings (from -12.7 to -23.0 ‰) that they attributed to B fractionation
during fluid exsolution. On the other hand, we also note that these results are at odds with the findings

162
of Tonarini et al. (1998) who studied zoned pegmatitic tourmalines from Elba (Italy) crystallizing
from melt and fluid assemblages over a temperature range of > 300 °C, and found no significant
isotope variations. More detailed studies are required in order to establish a more rigorous
interpretation of the isotope composition of tourmaline from the QF than the reconnaissance level of
this work. In particular, better constraints on crystallization temperatures and relative timing of
tourmaline growth would be useful. Nonetheless, this work provides an important example to
constrain the behaviour of B isotopes within a granitic melt-hydrothermal fluid-tourmaline system.

5.6.3 B isotope constraints on fluid mixing

In the northeastern part of the Bação complex, tourmaline is ubiquitous in the Nova Lima
schists (H1) and in the crosscutting veins and dikes (FC2 and IRV02). Three important B isotope
features of these tourmaline need to be addressed: (1) the overlapping populations at δ11B ~ -18.0 ‰ in
all three samples (illustrated by the grey vertical bars in Fig. 5.9d to f), (2) the large compositional
range observed in schist sample H1 (Fig. 5.9e), and (3) the core-to-rim variations between types A and
B cores within the quartz-tourmaline veins (Fig. 5.9d). The light tourmaline B isotope compositions in
these samples can neither be explained by processes involving a single source of B nor by isotope
fractionation. The presence of a separate, isotopically lighter fluid component is therefore required to
explain the observed δ11B values as low as -27.1 ‰.

One critical matter in understanding the complex variations observed in the quartz veins
tourmalines is identifying the origin of type B cores, and whether both types of cores formed
contemporaneously or one after the other. No petrographic evidence was found which could indicate
that one core formed before or after the other. A detrital origin is considered unlikely for the two
following reasons: (1) the type B cores show well-preserved growth faces (Fig. 5.4d), not rounded or
irregular shapes as would be expected in detrital grains, and (2) B isotope analyses revealed a narrow
range of δ11B values from -19.9 to -21.9 ‰ in type B cores (Fig. 5.9d), whereas one would expect a
population of detrital tourmalines that grew in different environments to display a wider range of
isotope compositions. The main argument in support of a fluid mixing hypothesis is the critical
observation that type A and B tourmalines show clear opposite core-to-rim variation, both in terms of
major element (Fig. 5.8) and isotope compositions (Fig. 5.9d). Assuming that both types of cores are
contemporaneous, this would imply that the tourmalines formed from a geochemically and
isotopically heterogeneous fluid environment which evolved towards a homogeneous intermediate
composition. The higher Ca, Mg and Ti contents in the cores of type B tourmaline (Fig. 5.8) are taken
as indications that the fluid from which they grew was derived from the host rock (Dini et al., 2008;
Trumbull et al., 2008; Buriánek et al., 2011). Apart from tourmaline, the most common reservoirs for
B in metasedimentary rocks are white micas, with concentrations reaching > 1000 ppm (Henry &
Dutrow, 2002). In the schists surrounding the Bação complex, white mica is generally only present as
fine retrograde needles in very small proportions. Macroscopic observation in the field of the schist
163
sample H1 however suggests that at this location, muscovite is more abundant than in samples at the
southeastern border of the complex where chlorite dominates the retrograde assemblages, and that it
10
may therefore represent a viable reservoir for B in those schists. This idea is supported by the
presence of a second population of tourmalines in sample H1, with an average δ11B value of -26.3 ±
0.8 ‰ (Fig. 5.9e). We propose that this peak may be representative of the lighter fluid end-member. In
this case, the apparent spread of δ11B values in sample H1 can be explained by different degrees of
mixing between both fluid end-members.

Overall, we propose a scenario in which granite-derived fluids intrude the country rocks,
crystallizing several quartz-tourmaline veins and disseminated tourmaline grains in all the lithologies
(evidenced by the presence of numerous tourmalines with δ11B ~ -18.0 ‰). Contact metamorphism
induces the release of a small amount of B-rich fluids from the country rocks (represented by the
lowest peak of δ11B values in H1). Both magmatic and external fluids then mix to produce the
observed variations in quartz vein tourmaline.

5.7 CONCLUSIONS

Tourmaline is an important component in differentiated products of granitic magmatism and

the country rocks they intrude in the Neoarchean Quadrilátero Ferrífero province of SE Brazil. It
occurs as a major constituent in a small leucogranitic intrusion and numerous plagioclase- and quartz-
veins, as well as disseminations in the nearby metasediments of the greenstone belt. All tourmalines
from this study fall in the schorl-dravite series. Systematic chemical variations in these tourmalines
indicate a strong protolith control on their mineral chemistry, with notably a decrease in Fe/(Fe+Mg)
ratio and an increase in Cr contents from magmatic to hydrothermal tourmaline present in the
metasediments. Heavily zoned tourmaline grains from quartz veins injected in the schists display two
coexisting types of cores and opposite core-to-rim variations: one is enriched in Al whereas the other
is characterized by high Mg, Ca and Ti contents, and their respective rims have intermediate
compositions between both cores.

The total spectrum of δ11B values in the QF tourmalines ranges from -27.1 to -10.7 ‰. We
observe a general trend of decreasing δ11B values from magmatic tourmaline in the leucogranite and
plagioclase-tourmaline veins (~ -13.5 ‰) to hydrothermal tourmaline in the schists and in quartz veins
(down to ~ -15.5 and -18 ‰), which can be reconciled with a model of isotope fractionation during
magmatic degassing. A small fraction of tourmaline grains display lighter δ 11B values (down to ~ -27
‰) that cannot be explained by melt-fluid fractionation processes only. This is the case for some
metasedimentary-hosted tourmaline, as well as for the Mg-Ca-Ti-rich tourmaline cores in the quartz
veins. We propose that the simplest model to explain these signatures requires the involvement of two
separate fluids: one is a granite-derived hydrothermal fluid and the other isotopically lighter
component is derived from the intruded country rocks by contact metamorphism.

164
 Furthermore, this study demonstrates the potential of the laser ablation MC-ICP-MS technique
in measuring in-situ B isotopes in tourmaline, with a high resolution (20 µm spot diametre) and a
precision of isotope ratio measurements (~ 0.4 ‰) comparable with that of the more widely used
SIMS method.

B isotopes in tourmaline are commonly used to unravel the origin of ore-related fluids,
particularly in orogenic gold deposits including those hosted in Archean provinces, where
mineralization often occurs in association with granite-related veins (e.g. Jiang et al., 2002; Krienitz et
al., 2008; Baksheev et al., 2015; Lambert-Smith et al., 2016; Molnár et al., 2016). In this regard, it is
interesting and perhaps somewhat puzzling to note the absence of mineralization associated with the
granitic melt-hydrothermal system investigated in this study.

Hanna Jordt-Evangelista, Hugo Moreira and Lucas Cassino are thanked for providing some of their samples. EPMA analyses were obtained
from the Laboratório de Microanálises do DEGEO/EM – Laboratório integrante da RMIc, Rede de Microscopia e Microanálises de Minas
Gerais – FAPEMIG. We thank Marco Paulo de Castro for his help during EPMA analyses. This work was supported by CNPq (grant
numbers: 401334/2012-0, 402852/2012-5 and 302633/2011-1) and FAPEMIG (grant numbers: APQ03943 and RDP 0067-10) grants.

165
166
 CHAPTER 6
GENERAL CONCLUSIONS
6 6

This thesis aimed to investigate the processes that led to the formation and stabilization of the
southernmost portion of the São Francisco craton (SE Brazil), via a combination of field and
petrographic observations, whole rock geochemistry and isotope (U-Pb, Lu-Hf, O, B) analyses. The
results and conclusions drawn from this work, presented in chapters 3, 4 and 5 are as follows:

The basement of the Quadrilátero Ferrífero district in the southern São Francisco craton
consists of orthogneisses, intruded by numerous granitic batholiths, pods and veins and abundant
pegmatitic and aplitic dikes. The crust presently exposed in the Quadrilátero Ferrífero was built during
three main magmatic events: the Rio das Velhas I and II events (at 2.92-2.85 and 2.80-2.76 Ga
respectively) and the Mamona event (2.75-2.68 Ga), which represents the last magmatic pulse
recorded in the region. Granitoids and gneisses of the Bação, Bonfim and Belo Horizonte basement
complexes were subdivided into two groups: medium- and high-K rocks. Medium-K gneisses and
granitoids were generated in the Meso- and Neoarchean (i.e. during the Rio das Velhas I and II
events). These rocks have some similarities with Archean TTGs, despite a small enrichment in SiO2,
K2O and depletion in Al2O3, and as such are thought to result from mixing between an end-member
derived by partial melting of a metabasaltic rock and a melt resulting from recycling of older TTG
crust. High-K granites match the compositions of late Archean biotite- and two-mica granites. They
postdate the emplacement of the medium-K rocks (i.e. they were emplaced during the Mamona event),
and are inferred to result from the partial melting of immature metasediments. The involvement of
metasediments in the petrogenesis of the Neoarchean high-K magmas is further supported by: (1) a
trend towards high δ18O(Zrc) values, indicating O isotope equilibration at near-surface conditions, and
(2) the presence of boron-rich magmatic melts and fluids emplaced at ca. 2.70 Ga, likely derived from
a boron-rich metasedimentary protolith. The transition between medium- and high-K magmatism in
the Neoarchean is interpreted to reflect the emergence of clastic sedimentary basins marking the
beginning of modern-type sedimentary cycles on Earth.

Even though the three basement complexes seem to have been similarly affected by the three
main crust-forming magmatic events in the Meso- and Neoarchean, some noticeable differences exist
between them. These include: (1) the emplacement during the Mamona event of voluminous batholiths
similar to TTGs in the Belo Horizonte complex, indicating that on the contrary to the Bação and
Bonfim complexes, this dome did not experience the transition from medium- to high-K magmatism,
and (2) different crustal evolution arrays evidenced by hafnium isotopes, especially evident between
the Bação and Bonfim complexes. We suggest that the three basement domes presently exposed in the

167
Quadrilátero Ferrífero represent separate crustal blocks with different identities, which were accreted
together during a late collisional stage.

Hafnium isotope modelling indicates a major change in the geodynamic evolution of the
southern São Francisco craton in the Mesoarchean (at ~ 2.90 Ga). This period marks the transition
from a regime dominated by net crustal growth and widespread production of juvenile crust, to a
Neoarchean regime dominated by crustal reworking, accounting for the generation of a wide variety of
granitoid magmas, including the medium- and high-K granitoids. Additionally, oxygen isotopes
revealed the interaction of granitic magmas with meteoric fluids in the upper crust throughout the
Meso- and Neoarchean in the Quadrilátero Ferrífero, indicating the emergence of large portions of the
continental crust. This transition at ~ 2.90 Ga is interpreted as the onset of continental collision in the
southern São Francisco craton.

Magmatic and hydrothermal tourmaline present in Neoarchean granitic/pegmatitic intrusions,

quartz veins and in their surrounding country rocks provides evidence for the presence of widespread
magmatic late-stage fluids percolating the crust at 2.70-2.60 Ga and interacting with the nearby
supracrustal rocks. The presence of hydrothermal tourmaline of magmatic origin within the greenstone
belt in close vicinity of the basement complexes suggests that the formation of the dome-and-basin
structures, which put into contact the magmatic basement with the supracrustal sequences, occurred
prior to the last magmatic event (i.e. before 2.70 Ga). This idea however seems at odds with the
current idea that the Quadrilátero Ferrífero acquired its present-day architecture in the
Paleoproterozoic.

168
 References
Aguilar C., Alkmim F. F., Lana C., Farina F. 2017. Palaeoproterozoic assembly of the São Francisco craton, SE
Brazil: New insights from U-Pb titanite and monazite dating. Precambrian Res., 289:95-115.
Albert C., Farina F., Lana C., Stevens G., Storey C., Gerdes A., Martínez Dopico C. 2016. Archean crustal
evolution in the Southern São Francisco craton, Brazil: Constraints from U-Pb, Lu-Hf and O isotope
analyses. Lithos, 266-267:64-86.
Alkmim F. F. & Marshak S. 1998. Transamazonian Orogeny in the Southern São Francisco Craton Region,
Minas Gerais, Brazil: Evidence for Paleoproterozoic collision and collapse in the Quadrilátero Ferrífero.
Precambrian Res., 90:29–58.
Alkmim F. F. & Martins-Neto M. A. 2012. Proterozoic first-order sedimentary sequences of the São Francisco
Craton, eastern Brazil. Mar. Petrol. Geol., 33:127–139.
Allen T. & Chamberlain C. P. 1989. Thermal consequences of mantled gneiss dome emplacement. Earth Planet.
Sci. Lett., 93:392 – 404.
Almeida F. F. M., Hasui Y., Brito Neves B. B., Fuck R. A. 1981. Brazilian structural provinces: an introduction.
Earth-Sci. Rev., 17:1–29.
Almeida J. A. C., Dall'Agnol R., Dias S. B., Althoff F. J. 2010. Origin of the Archean leucogranodiorite–granite
suites: evidence from the Rio Maria terrane and implications for granite magmatism in the Archean. Lithos,
120:235–257.
Almeida J. A. C., Dall'Agnol R., Oliveira M. A., Macambira M. J. B., Pimentel M. M., Rämö O. T., Guimarães
F. V., Leite A. A. S. 2011. Zircon geochronology, geochemistry and origin of the TTG suites of the Rio
Maria granite–greenstone terrane: implications for the growth of the Archean crust of the Carajás Province,
Brazil. Precambrian Res., 187:201–221.
Almeida J. A. C., Dall‘Agnol R., Leite A. A. S. 2013. Geochemistry and zircon geochronology of the Archean
granite suites of the Rio Maria granite–greenstone terrane, Carajás Province, Brazil. J. S. Am. Earth Sci.,
42:103–126.
Amelin Y., Lee D. C., Halliday A. N., 2000. Early-middle Archaean crustal evolution deduced from Lu–Hf and
U–Pb isotopic studies of single zircon grains. Geochim. Cosmochim. Acta, 64:4205–4225.
Anhaeusser C. R. 1983. The geology of the Schapenburg greenstone remnant and surrounding Archaean granitic
terrane, south of Badplaas, Eastern Transvaal. Geological Society of South Africa, Special Publications,
9:3l–44.
Anhaeusser C. R., Mason R., Viljoen M. J., Viljoen R. P. 1969. A reappraisal of some aspects of Precambrian
shield geology. Geol. Soc. Am. Bull., 80(11):2175-2200.
Appleby S. K., Gillespie M. R., Graham C. M., Hinton R. W., Oliver G. J. H., Kelly N. M., E. I. M. F. 2010. Do
S-type granites commonly sample infracrustal sources? New results from an integrated O, U–Pb and Hf
isotope study of zircon. Contrib. Mineral. Petrol., 160:115–132.
Armstrong R. L. 1981. Radiogenic isotopes: The case for crustal recycling on a near-steady-state no-continental-
growth Earth. Phil. Trans. R. Soc. London A, 301:443–472.
Arndt N. T. 1999. Why was flood volcanism on submerged continental platforms so common in the
Precambrian? Precambrian Res., 97:155–164.
Arndt N. T. 2013. Formation and evolution of the continental crust. Geochemical Perspectives, 2(3):405-533.
Arndt N. T. & Chauvel C. 1991. Crust of the Hadean Earth. Bull. Geol. Soc. Den., 39:145-151.
Atherton M. P. & Petford N. 1993. Generation of sodium-rich magmas from newly underplated basaltic crust.
Nature, 362:144–146.
Avila C. A., Teixeira W., Cordani U. G., Moura C. A. V., Pereira R. M. 2010. Rhyacian (2.23-2.20) juvenile
accretion in the souhtern São Francisco craton, Brazil: geochemical and isotopic evidence from the Serrinha
magmatic suite, Mineiro belt. J. S. Am. Earth Sci., 29:143-159.
Babinski M., Chemale Jr. F., Van Schmus W. R. 1991. Pb/Pb geochronology of carbonate rocks of Minas
Supergroup, Quadrilátero Ferrífero, Minas Gerais, Brazil. In: Am. Geophys. Union Fall Meet. Suppl., 72,
531.

169
Babinski M., Chemale Jr. F., Van Schmus W. R. 1993. A idade das formacões ferríferas bandadas do
Supergrupo Minas e sua correlação com aquelas da Africa do Sul e Austrália. In: Soc. Bras. Geol., Núcleo
Bahia/Sergipe, Anais, 152-153.
Babinski M., Chemale Jr. F., Van Schmus W. R. 1995. The Pb/Pb age of the Minas Supergoup carbonate rocks,
Quadrilátero Ferrífero, Brazil. Precambrian Res., 72:235-245.
Baksheev I. A., Prokofiev V. Y., Trumbull R. B., Wiedenbeck M., Yapaskurt V. O. 2015. Geochemical
evolution of tourmaline in the Darasun gold district, Transbaikal region, Russia: evidence from chemical and
boron isotopic compositions. Miner. Depos., 50:125–138.
Baltazar O. F. & Zucchetti M. 2007. Lithofacies associations and structural evolution of the Archean Rio das
Velhas greenstone belt, Quadrilátero Ferrífero, Brazil: A review of the setting of gold deposits. Ore Geol.
Rev., 32:1–2.
Barbosa A. L. M. 1979. Variacões de fácies na Série Minas. Soc. Bras. Geol. Bol., 1:89-100.
Barbosa J. S. F. & Sabaté P. 2004. Archean and Paleoproterozoic crust of the São Francisco craton, Bahia,
Brazil: geodynamic features. Precambrian Res., 133:1–27.
Barker F. & Arth J. G. 1976. Generation of trondhjemitic–tonalitic liquids and Archaean bimodal trondhjemite-
basalt suites. Geology, 4(10):596–600.
Bast R., Scherer E. E., Mezger K., Austrheim H., Ludwig T., Marschall H. R., Putnis A., Löwen K. 2014. Boron
isotopes in tourmaline as a tracer of metasomatic processes in the Bamble sector of Southern Norway.
Contrib. Mineral. Petrol., 168:1-21.
Bédard J. H. 2006. A catalytic delamination-driven model for coupled genesis of Archaean crust and sub-
continental lithospheric mantle. Geochim. Cosmochim. Acta, 70:1188–1214.
Bédard J. H. 2013. How many arcs can dance on the head of a plume? A ‗Comment‘ on: a critical assessment of
Neoarchaean ‗plume only‘ geodynamics: evidence from the Superior Province, by Derek Wyman,
Precambrian Research, 2012. Precambrian Res., 229:189–197.
Bédard J., Brouillette P., Madore L., Berclaz A. 2003. Archaean cratonization and deformation in the northern
Superior Province, Canada: an evaluation of plate tectonic versus vertical tectonic models. Precambrian
Res., 127:61–87.
Bédard J. H., Harris L. B., Thurston P. C. 2013. The hunting of the snArc. Precambrian Res., 229:20–48.
Belousova E. A., Kostitsyn Y. A., Griffin W. L., Begg G. C., O'Reilly S. Y., Pearson N. J. 2010. The growth of
the continental crust: constraints from zircon Hf-isotope data. Lithos, 119:457–466.
Bennett V. C. & De Paolo D. J. 1987. Proterozoic crustal history of the western United States as determined by
neodymium isotopic mapping. Geol. Soc. Am. Bull., 99:674–685.
Bickle M. J., Bettenay L. F., Chapman H. J., Groves D. I., McNaughton N. J., Campbell I. H., de Laeter J. R.
1989. The age and origin of younger granitic plutons of the Shaw Batholith in the Archaean Pilbara Block,
Western Australia. Contrib. Mineral. Petrol., 101:361–376.
Bindeman I. N. 2011. When do we need pan-global freeze to explain 18-O depleted zircons and rocks? Geology,
39:799–800.
Bindeman I. N. & Valley J. W. 2001. Low-δ18O rhyolites from Yellowstone: magmatic evolution based on
analyses of zircons and individual phenocrysts. J. Pet., 42:1491–1517.
Bindeman I. N., Fu B., Kita N., Valley J. W. 2008. Origin and evolution of silicicmagmatism at Yellowstone
based on ion microprobe analysis of isotopically zoned zircons. J. Pet., 49:163–193.
Bleeker W. 2003. The late Archean record: a puzzle in ca. 35 pieces. Lithos, 71:99-134.
Blewett R. S., Shevchenko S., Bell B. 2004. The North Pole Dome: a non-diapiric dome in the Archaean Pilbara
Craton, Western Australia. Precambrian Res., 133:105–120.
Blewett R. S., Czarnota K., Henson P. A. 2010. Structural-event framework for the eastern Yilgarn Craton,
Western Australia, and its implications for orogenic gold. Precambrian Res., 183:203–229.
Blichert-Toft J. & Puchtel I. S. 2010. Depleted mantle sources through time: evidence from Lu–Hf and Sm–Nd
isotope systematics of Archean komatiites. Earth Planet. Sci. Lett., 297(3–4):598–606.

170
Bodorkos S. & Sandiford M. 2006. Thermal and mechanical controls on the evolution of Archean crustal
deformation: examples from Western Australia. In: K. Benn, J. C. Mareschal, K. C. Condie (eds.) Archean
Geodynamics and Environments. American Geophysical Union, 131–147.
Boekhout F., Roberts N. M., Gerdes A., Schaltegger U. 2013. A Hf-isotope perspective on continent formation
in the south Peruvian Andes. In: N. M. W. Roberts, M. Van Kranendonk, S. Parman, S. Shirey, P. D. Clift
(eds.) Continent formation through time. Geological Society, London, Special Publications, 305-321.
Bouhallier H., Chardon D., Choukroune P. 1995. Strain patterns in Archaean dome-and-basin structures: the
Dharwar craton (Karnataka, South India). Earth Planet. Sci. Lett., 135:57–75.
Bowring S. A. & Williams I. S. 1999. Priscoan (4.00–4.03 Ga) orthogneisses from northwestern Canada.
Contrib. Mineral. Petrol., 134:3–16.
Bradley D. C. 2011. Secular trends in the geologic record and the supercontinent cycle. Earth-Sci. Rev., 108:13-
33.
Brandl G., Cloete M., Anhaeusser C. R. 2006. Archean greenstone belts. In: M. R. Johnson, C. R. Anhaeusser,
R. J. Thomas (eds.) The geology of South Africa. Geological Society of South Africa, Pretoria, 9–56.
Breaks F. W. & Moore J. M. 1992. The Ghost Lake batholith, Superior Province of Northwestern Ontario: a
fertile, S-type, peraluminous granite – rare-element pegmatite system. Can. Mineral., 30:835–875.
Brown M. 2006. A duality of thermal regimes is the distinctive characteristic of plate tectonics since the
Neoarchean. Geology, 34:961–964.
Brueckner H. K., Cunningham D., Alkmim F. F., Marshak S. 2000. Tectonic implications of Precambrian Sm-
Nd dates from the southern São Francisco craton and adjacent Araçuai and Ribeira belts, Brazil.
Precambrian Res., 99:255-269.
Buriánek D., Hanžl P., Hrdličková K. 2011. Pegmatite dykes and quartz veins with tourmaline: an example of
partial melting in the contact aureole of the Chandman Massif intrusion, SW Mongolia. J. Geosci., 56:201-
213.
Cabral A. R., Zeh A., Koglin N., Gomes A. A. S., Viana D. J., Lehmann B. 2012. Dating the Itabira iron
formation, Quadrilátero Ferrífero of Minas Gerais, Brazil, at 2.65 Ga: depositional U-Pb age of zircon from
a metavolcanic layer. Precambrian Res., 204:40-45.
Cagnard F., Durrieu N., Gapais D., Brun J. P., Ehlers C. 2006. Crustal thickening and lateral flow during
compression of hot lithospheres, with particular reference to Precambrian times. Terra Nova, 18:72–78.
Cagnard F., Barbey P., Gapais D. 2011. Transition between ―Archaean-type‖ and ―modern-type‖ tectonics:
insights from the Finnish Lapland Granulite Belt. Precambrian Res., 187:127–142.
Campbell I. H. & Allen C. M. 2008. Formation of supercontinents linked to increases in atmospheric oxygen.
Nat. Geosci., 1:554–558.
Campos J. C. S., Carneiro M. A., Basei M. A. S. 2003. U-Pb evidence for Late Neoarchean crustal reworking in
the southern São Francisco Craton (Minas Gerais, Brazil). An. Acad. Bras. Cienc., 75(4):497–511.
Canuto J. R. 2010. Estratigrafia de sequências em bacias sedimentares de diferentes idades e estilos tectônicos.
Rev. Bras. Geocienc., 40(4):537-549.
Carneiro M. A. 1992. O complexo metamórfico Bonfim setentrional (Quadrilátero Ferrífero, Minas Gerais):
Litoestratigrafia e evolução geológica de um segmento de crosta continental do Arqueano. University of
São Paulo, Brazil, PhD Thesis, 233p.
Carneiro M. A. & Teixeira W. 1992. Discordancia de idades radiometricas U-Pb e Rb-Sr no craton do São
Francisco meridional: evidencias a partir do Complexo Metamórfico Bonfim Setentrional, Quadrilátero
Ferrífero, Minas Gerais. In: Congr. Bras. Geol., 37, São Paulo, Anais, 189-190.
Carneiro M. A., Jordt-Evangelista H., Teixeira W. 1997. Eventos magmaticos arqueanos de natureza calcio-
alcalina e tholeiitica no Quadrilátero Ferrífero e suas implicações tectonicas. Rev. Bras. Geocienc., 27:121–
128.
Carneiro M. A., Carvalho Jr. I. M., Teixeira W. 1998. Petrologia, geoquímica e geocronologia dos diques
máficos do Complexo Bonfim Setentrional (Quadrilátero Ferrífero) e suas implicações na evolução crustal
do Craton São Francisco Meridional. Rev. Bras. Geocienc., 28(1):29-44.

171
Cassino L. F. 2014. Distribuição de idades U-Pb de zircões detríticos dos Supergrupos Rio das Velhas e Minas
na Serra de Ouro Preto, Quadrilátero Ferrífero, MG e Implicações para a evolução sedimentar e tectônica.
Departamento de Geologia, Universidade Federal de Ouro Preto, Brazil, MSc Thesis, 52p.
Castro A. 2004. The source of granites: inferences from the Lewisian complex. Scott. J. Geol., 40:49–65.
Catanzaro E. J., Champion C. E., Garner E. L., Marinenko G., Sappenfield K. M., Shields W. R. 1970. Standard
reference materials: boric acid; isotopic and assay standard reference materials. U.S. National Bureau of
Standards, Special Publication No. 260–17, U.S. Government Printing Office, Washington DC, 70p.
Cavosie A. J., Valley J. W., Wilde S. A., E. I. M. F. 2005. Magmatic δ 18O in 4400–3900 Ma detrital zircons: a
record of the alteration and recycling of crust in the Early Archean. Earth Planet. Sci. Lett., 235:663–681.
Cawood P. A., Hawkesworth C. J., Dhuime B. 2013. The continental record and the generation of continental
crust. Geol. Soc. Am. Bull., 215(1-2):14-32.
Champion D. C. & Sheraton J. W. 1997. Geochemistry and Nd isotope systematics of Archaean granites of the
Eastern Goldfields, Yilgarn Craton, Australia: implications for crustal growth processes. Precambrian Res.,
83(1–3):109–132.
Champion D. C. & Smithies R. H. 1999. Archaean granites of the Yilgarn and Pilbara cratons, Western
Australia: secular changes. In: 4th Hutton Symposium on Granites and Related Rocks, Clermont-Ferrand,
France.
Champion D. C. & Smithies R. H. 2007. Geochemistry of Paleoarchean granites of the East Pilbara Terrane,
Pilbara Craton, Western Australia: implications for early Archean crustal growth. In: M. J. Van Kranendonk,
R. H. Smithies, V. C. Bennett (eds.) Earth’s oldest rocks. Developments in Precambrian Geology,
Amsterdam, Elsevier, 369–410.
Chardon D., Choukroune P., Jayananda M. 1996. Strain patterns, décollement and incipient sagducted
greenstone terrains in the Archaean Dharwar craton (South India). J. Struct. Geol., 18:991–1004.
Chardon D., Choukroune P., Jayananda M. 1998. Sinking of the Dharwar Basin (South India): implications for
Archaean tectonics. Precambrian Res., 91:15–39.
Chardon D., Peucat J. J., Jayananda M., Choukroune P., Fanning C. M. 2002. Archean granite-greenstone
tectonics at Kolar (South India): Interplay of diapirism and bulk inhomogeneous contraction during juvenile
magmatic accretion. Tectonics, 21(3):1016.
Chardon D., Gapais D., Cagnard F., 2009. Flow of ultra-hot orogens: a view from Precambrian, clues for the
Phanerozoic. Tectonophysics, 477(3–4):105–118.
Chaussidon M. & Albarède F. 1992. Secular boron isotope variations in the continental crust: an ion microprobe
study. Earth Planet. Sci. Lett., 108:229-241.
Chauvet A., Faurre M., Dossin I., Charvet J. 1994. Three-stage structural evolution of the Quadrilátero Ferrífero:
consequences for Neoproterozoic age and the formation of gold concentrations of the Ouro Preto area,
Minas Gerais, Brazil. Precambrian Res., 68:139-167.
Chaves M. L. S. C. 1997. Geologia e mineralogia do diamante da Serra do Espinhaço em Minas Gerais.
University of São Paulo, Brazil, PhD Thesis, 289p.
Chemale Jr. F., Rosière C. A., Endo I. 1991. Evolução tectônica do Quadrilátero Ferrífero, Minas Gerais: um
modelo. Pesquisas, 18(2):104-127.
Chemale Jr. F., Alkmim F. F., Endo I. 1993a. Late Proterozoic tectonism in the interior of the São Francisco
Craton. In: Gondwana Eight – Assembly, Evolution and Dispersal, Balkema, 29-42.
Chemale F., Babinski M., Van Schmus W. R. 1993b. U/Pb dating of granitic–gneissic rocks from the Belo
Horizonte and Bonfim complexes, Quadrilátero Ferrífero (Brazil). Report for CNPq and NSFEAR Project
on São Francisco Craton Margin Transect Project, 16p.
Chemale F., Rosière C. A., Endo I. 1994. The tectonic evolution of the Quadrilátero Ferrífero, Minas Gerais,
Brazil. Precambrian Res., 65:25–54.
Choukroune P., Bouhallier H., Arndt N. T. 1995. Soft lithosphere during periods of Archean crustal growth or
crustal reworking. In: M. P. Coward & A. C. Ries (eds.) Early Precambrian processes. Geological Society
of London, Special Publications, 67–86.

172
Clemens J. D., Belcher R. W., Kisters A. F. M. 2010. The Heerenveen batholith, Barberton Mountain Land,
South Africa: Mesoarchaean, potassic, felsic magmas formed by melting of an ancient subduction complex.
J. Pet., 51(5):1099–1120.
Clemens J. D., Stevens G., Farina F. 2011. The enigmatic sources of I-type granites: the peritectic connexion.
Lithos, 126:174–181.
Collins W. J. 2002. Hot orogens, tectonic switching, and creation of continental crust. Geology, 30:535–538.
Collins W. J., Belousova E. A., Kemp A. I. S., Murphy J. B. 2011. Two contrasting Phanerozoic orogenic
systems revealed by hafnium isotope data. Nat. Geosci., 4(5):333–337.
Condie K. C. 1989. Geochemical changes in basalts and andesites across the Archean–Proterozoic boundary:
identification and significance. Lithos, 23(1–2):1–18.
Condie K. C. 1993. Chemical composition and evolution of the upper continental crust: Contrasting results from
surface samples and shales. Chem. Geol., 104:1–37
Condie K. C. 1997. Plate tectonics and crustal evolution. Butterworth-Heinemann, Oxford, 288p.
Condie K. C. 1998. Episodic continental growth and supercontinents: a mantle avalanche connection? Earth
Planet. Sci. Lett., 163:97–108.
Condie K. C. 2005. TTGs and adakites: are they both slab melts? Lithos, 80:33-44.
Condie K. C. 2015. Changing tectonic settings through time: indiscriminate use of geochemical discriminant
diagrams. Precambrian Res., 266:587-591.
Condie K. C. & Abbott D. H. 1999. Oceanic plateaus and hotspot islands: identification and role in continental
growth. Lithos, 46:1–4.
Condie K. C. & Kröner A. 2008. When did plate tectonics begin? Evidence from the geologic record. In: K. C.
Condie & V. Pease (eds.) When did plate tectonics begin? Geological Society of America, Special Paper,
281-294.
Condie K. C. & Aster R. C. 2010. Episodic zircon age spectra of orogenic granitoids: The supercontinent
connection and continental growth. Precambrian Res., 180:227–236.
Condie K. C. & O‘Neill C. 2010. The Archean–Proterozoic boundary: 500 my of tectonic transition in Earth
history. Am. J. Sci., 310:775–790.
Condie K. C., Aster R. C., van Hunen J. 2016. A great thermal divergence in the mantle beginning 2.5 Ga:
geochemical constraints from greenstone basalts and komatiites. Geoscience Frontiers, 7:543-553.
Cordani U. G., Kawashita K., Miiller G., Quade H., Reimer V., Roeser H. 1980. Interpretação tectônica e
petrológica de dados geocronológicos do embasamento na borda sudeste do Quadrilátero Ferrífero. An.
Acad. Bras. Cienc., 52:785-799.
Costa L., Rocha M. M., de Sousa R. M. 2003. O ouro do Brasil: transporte e fiscalidade (1720-1764). In:
Congresso Brasileiro de História Econômica e 6ª Conferência Internacional de História de Empresas,
Associação Brasileira de Pesquisadores em História Econômica, Anais.
Czarnota K., Champion D. C., Goscombe B., Blewett R. S., Cassidy K. F., Henson P. A., Groenewald P. B.
2010. Geodynamics of the Eastern Yilgarn Craton. Precambrian Res., 183:175–202.
Dalstra H. J., Bloem E. J. M., Ridley J. R., Groves D. I. 1998. Diapirism synchronous with regional deformation
and gold mineralisation, a new concept for granitoid emplacement in the Southern Cross Province, Western
Australia. Geologie en Mijnbouw, 76:321–338.
Da Silva L. C., Noce C. M., Lobato L. M. 2000. Dacitic volcanism in the course of the Rio das Velhas (2800-
2960 Ma) Orogeny: a Brazilian Archean analogue (TTG) to the modern adakites. Rev. Bras. Geocienc.,
30:384-387.
Davies G. F. 2006. Gravitational depletion of the early Earth‘s upper mantle and the viability of early plate
tectonics. Earth Planet. Sci. Lett., 243:376–382.
Davies G. F. 2007. Dynamics of the Hadean and Archaean mantle. In: M. van Kranendonk, H. Smithies, V.
Bennett (eds.) Earth’s oldest rocks. Amsterdam, Elsevier, 61-69.

173
de Bremond d‘Ars J., Lécuyer C., Reynard B. 1999. Hydrothermalism and diapirism in the Archean:
Gravitational instability constraints. Tectonophysics, 304:29 – 39.
Defant M. J. & Drummond M. S. 1990. Derivation of some modern arc magmas by melting of young subducted
lithosphere. Nature, 367:662–665.
De Min A., Rosset A., Marques L. S., Chaves A., Piccirillo E. M. 2005. Widespread late Mesoproterozoic
tholeiitic magmatism of the São Francisco craton (Brazil): petrology, geochemistry and geotectonic settings.
Russ. Geol. Geophys., 46(9):981-992.
De Oliveira O. B. & Teixeira W. 1990. Evidências de uma tectônica tangencial proterozóica no Quadrilátero
Ferrífero, MG. In: Congr. Bras. Geol., 36, Anais, 6:2589-2604.
de Wit M. J. 1982. Gliding and overthrust nappe tectonics in the Barberton greenstone belt. J. Struct. Geol.,
4(2):117–136.
de Wit M. J., Hart R. A., Hart R. J. 1987. The Jamestown ophiolite complex, Barberton mountain belt: a section
through 3.5 Ga oceanic crust. J. Afr. Earth Sci., 6:681–730.
de Wit M. J. & Hart R. A. 1993. Earth's earliest continental lithosphere, hydrothermal flux and crustal recycling.
Lithos, 30(3–4):309–335.
Dey S., Pandey U. K., Rai A. K., Chaki A. 2012. Geochemical and Nd isotope constraints on petrogenesis of
granitoids from NW part of the eastern Dharwar craton: possible implications for late Archaean crustal
accretion. J. Asian Earth Sci., 45:40–56.
Dey S., Nandy J., Choudhary A. K., Liu Y., Zong K. 2014. Origin and evolution of granitoids associated with
the Kadiri greenstone belt, eastern Dharwar craton: a history of orogenic to anorogenic magmatism.
Precambrian Res., 246:64–90.
Dhuime B., Hawkesworth C. J., Cawood P. A., Storey C. D. 2012. A change in the geodynamics of continental
growth 3 billion years ago. Science, 335:1334–1336.
Dhuime B., Wuestefeld A., Hawkesworth C. J. 2015. Emergence of modern continental crust about 3 billion
years ago. Nat. Geosci., 8:552–555.
Dilek Y. & Polat A. 2008. Suprasubduction zone ophiolites and Archean tectonics. Geology, 36(5):431-432.
Dingwell D. B., Pichavant M., Holtz F. 2002. Experimental studies of Boron in granitic melts. In: E. S. Grew &
L. M. Anovitz (eds.) Boron: Mineralogy, Petrology and Geochemistry. Reviews in Mineralogy and
Geochemistry, Mineralogical Society of America, 709–744.
Dini A., Mazzarini F., Musumeci G., Rocchi, S. 2008. Multiple hydro-fracturing by boron-rich fluids in the Late
Miocene contact aureole of eastern Elba Island (Tuscany, Italy). Terra Nova, 20:318–326.
Dorr J. V. N., Gair J. E., Pomerene J. G., Rynearson G. A. 1957. Revisão da estratigrafia pré-cambriana do
Quadrilátero Ferrífero. DNPM-DFPM, Rio de Janeiro, 31p.
Dorr J. V. N. 1959. Esbôço geológico do Quadrilátero Ferrífero de Minas Gerais, Brasil. Dep. Nac. Prod. Min.,
Publ. Espec. 1, Rio de Janeiro.
Dorr J. V. N. 1969. Physiographic, stratigraphic and structural development of the Quadrilátero Ferrírfero,
Minas Gerais, Brazil. USGS Prof. Paper 641-A, 110p.
Drake A. A. & Morgan B. A. 1980. Precambrian Plate Tectonics in the Brazilian Shield; Evidence from the Pre-
Minas Rocks of the Quadrilátero Ferrífero, Minas Gerais. USGS Prof. Paper 1119, 1-64.
Duretz T. & Gerya T. V. 2013. Slab detachment during continental collision: influence of crustal rheology and
interactions with lithospheric delamination. Tectonophysics, 602:124–140.
Dutrow B. L. & Henry D. J. 2011. Tourmaline: a geologic DVD. Elements, 7:301–306.
Dyar M. D., Wiedenbeck M., Robertson D., Cross L. R., Delaney J. S., Ferguson K., Francis C. A., Grew E. S.,
Guidotti C. V., Hervig R. L., Hughes J. M., Husler J., Leeman W. P., McGuire A. V., Rhede D., Rothe H.,
Paul R. L., Richards I., Yates M. 2001. Reference minerals for the microanalysis of light elements.
Geostand. Newsl., 25:441–463.
Eggins S. 2003. Laser ablation ICP-MS analysis of geological materials prepared as lithium borate glasses.
Geostand. Geoanal. Res., 27:147–162.

174
Eiler J. M. 2001. Oxygen isotope variations of basaltic lavas and upper mantle rocks. In: J. W. Valley & D. R.
Cole (eds.) Stable isotope geochemistry. Reviews in Mineralogy and Geochemistry, Mineralogical Society
of America, 319–364.
Ellam R. M. & Hawkesworth C. J. 1988. Is average continental crust generated at subduction zones? Geology,
16:314–317.
Endo I. & Fonseca M. A. 1992. Sistema de cisalhamento Fundão-Cambotas no Quadrilátero Ferrífero, MG:
Geometria e cinemática. Rev. Esc. Minas, 45:28-31.
Endo I. 1997. Regimes tectônicos do Arqueano e Proterozóico no interior da Placa San franciscana:
Quadrilátero Ferrífero e areas adjacentes, Minas Gerais. University of São Paulo, Brazil, PhD Thesis,
330p.
Erickson E. J. 2010. Structural and kinematic analysis of the Shagawa Lake shear zone, Superior Province,
northern Minnesota: implications for the role of vertical versus horizontal tectonics in the Archean. Can. J.
Earth Sci., 47:1463–1479.
Eriksson K. A. 1980. Transitional sedimentation styles in the Moodies and Fig Tree Groups, Barberton Mountain
Land, South Africa: Evidence favouring an Archean continental margin. Precambrian Res., 12:141–160.
Eriksson K. A., Krapez B., Fralick P. W. 1994. Sedimentology of Archean greenstone belts: signatures of
tectonic evolution. Earth-Sci. Rev., 37:1-88.
Farber K., Dziggel A., Trumbull R. B., Meyer F. M., Wiedenbeck M. 2015. Tourmaline B-isotopes as tracers of
fluid sources in silicified Palaeoarchean oceanic crust of the Mendon Formation, Barberton greenstone belt,
South Africa. Chem. Geol., 417:134-147.
Farina F., Stevens G., Gerdes A., Frei D. 2014. Small-scale Hf isotopic variability in the Peninsula pluton (South
Africa): the processes that control inheritance of 176Hf/177Hf diversity in S-type granites. Contrib. Mineral.
Petrol., 168:1065.
Farina F., Albert C., Lana C. 2015a. The Neoarchean transition between medium and high-K granitoids: clues
from the Southern São Francisco Craton (Brazil). Precambrian Res., 266:375–394.
Farina F., Cutts K., Lana C. 2015b. Cryptic evidence of fluid-present partial melting of Archean banded gneisses
in the Southern São Francisco craton (Brazil): implications for the evolution of the continental crust. In: 8th
Hutton Symposium on Granites and Related Rocks, Florianópolis, Brazil.
Farina F., Albert C., Martínez Dopico C., Aguilar Gil C., Moreira H., Hippertt J. P., Cutts K., Alkmim F. F.,
Lana C. 2016. The Archean-Paleoproterozoic evolution of the Quadrilátero Ferrífero (Brazil): current
models and open questions. J. S. Am. Earth Sci., 68:4–21.
Feng R. & Kerrich R. 1992. Geochemical evolution of granitoids from the Archean Abitibi Southern Volcanic
Zone and the Pontiac subprovince, Superior Province, Canada: implications for tectonic history and source
regions. Chem. Geol., 98:23–70.
Figueiredo L. 2011. Boa ventura. A corrida do ouro no brasil (1697-1810). Editora record Rio de Janeiro, São
Paulo, 368p.
Fischer R. & Gerya T. 2016. Early Earth plume-lid tectonics: A high-resolution 3D numerical modelling
approach. J. Geodyn., 100:198-214.
Flament N., Coltice N., Rey P. F. 2008. A case for late-Archaean continental emergence from thermal evolution
models and hypsometry. Earth Planet. Sci. Lett., 275:326–336.
87
Flament N., Coltice N., Rey P. F. 2013. The evolution of the Sr/86Sr of marine carbonates does not constrain
continental growth. Precambrian Res., 229:177–188.
Foley S. F., Tiepolo M., Vannucci R. 2002. Growth of early continental crust controlled by melting of
amphibolite in subduction zones. Nature, 417:637–640.
Friend C. R. L. & Nutman A. P. 2005. Complex 3670–3500 Ma orogenic episodes superimposed on juvenile
crust accreted between 3850 and 3690 Ma, Itsaq Gneiss Complex, Southern West Greenland. J. Geol.,
113(4):375–397.
Frost C. D., Frost B. R., Chamberlain K. R., Huselbosch T. P. 1998. The Late Archaean history of the Wyoming
province as recorded by granitic magmatism in the Wind River Range, Wyoming. Precambrian Res.,
98:145–173.

175
Frost C. D., Frost B. R., Kirkwood R., Chamberlain K. R. 2006. The tonalite–trondhjemite– granodiorite (TTG)
to granodiorite–granite (GG) transition in the late Archean plutonic rocks of the central Wyoming Province.
Can. J. Earth Sci., 43:1419–1444.
Froude D. O., Ireland T. R., Kinny P. D., Williams I. S., Compston W., Williams I. R., Myers J. S. 1983. Ion
microprobe identification of 4100--4200 Myr--old terrestrial zircons. Nature, 304:616--618.
Fyfe W. S. 1978. The evolution of the earth‘s crust: Modern plate tectonics to ancient hot spot tectonics? Chem.
Geol., 23:89–114.
Gair J. E. 1962. Geology and Ore Deposit of the Nova Lima and Rio Acima Quadrangles. USGS Prof. Paper
341-A, 67p.
Ganne J. & Feng X. 2017. Primary magmas and mantle temperature through time. Geochem. Geophys. Geosyst.,
18:872-888.
Gardien V., Thompson A. B., Grujic D., Ulmer P. 1995. Experimental melting of biotite + plagioclase + quartz ±
muscovite assemblages and implications for crustal melting. J. Geophys. Res., 100:15581–15591.
Gardien V., Thompson A. B., Ulmer P. 2000. Melting of biotite + plagioclase + quartz gneisses: the role of H 2O
in the stability of amphibole. J. Pet., 41:651–666.
Gastil R. G. 1960. The distribution of mineral dates in time and space. Am. J. Sci., 258:1-35.
Geisinger K. L., Oestrike R., Navrotsky A., Turner G. L., Kirkpatrick R.J. 1988. Thermochemistry and structure
of glasses along the join NaAl3SiO8–NaBSi3O8. Geochim. Cosmochim. Acta, 52:2405–2414.
Gerdes A. & Zeh A. 2006. Combined U–Pb and Hf isotope LA-(MC-)ICP-MS analyses of detrital zircons:
comparison with SHRIMP and new constraints for the provenance and age of an Armorican metasediment in
central Germany. Earth Planet. Sci. Lett., 249:47–61.
Gerdes A. & Zeh A. 2009. Zircon formation versus zircon alteration – new insights from combined U–Pb and
Lu–Hf in-situ LA–ICP–MS analyses, and consequences for the interpretation of Archean zircon from the
Central Zone of the Limpopo Belt. Chem. Geol., 261:230–243.
Gerya T. 2014. Precambrian geodynamics: Concepts and models. Gondwana Res., 25:442-463.
Gill J. B. 1981. Orogenic andesites and plate tectonic. Springer, Berlin, 392p.
Gilliam C. E. & Valley J. W. 1997. Low δ18O magma, Isle of Skye, Scotland; evidence from zircons. Geochim.
Cosmochim. Acta, 61:4975–4981.
Gomes N. S. 1985. Petrologisch-geochemische Untersuchugen im Baçao-Komplex Eisernes Viereck, Minas
Gerais, Brasilien. Mathematisch-Naturwissenchaftlichen Fakultat, Technische Universitat Clausthal, PhD
Thesis, 209p.
Gomes N. S. 1986. Determinações Geotermometricas e Geobarometricas em Parageneses Minerais de Alto Grau
Metamorfico no Complexo do Bação, Quadrilátero Ferrífero – Minas Gerais. In: Congr. Bras. Geol., 34,
Anais, 4:1424-1436.
Gomes N. S. & Muller G. 1987. Caracterização Quimica de Parageneses Minerais de Alto Grau Metamórfico no
Complexo Bação, Quadrilátero Ferrífero, Minas Gerais. Rev. Esc. Minas, 40(l):25-36.
Gonfiantini R., Tonarini S., Gröning M., Adorni-Braccesi A., Al-Ammar A. S., Astner M., Bächler S., Barnes R.
M., Bassett R. L., Cocherie A., Deyhle A., Dini A., Ferrara G., Gaillardet J., Grimm J., Guerrot C.,
Krähenbühl U., Layne G., Lemarchand D., Meixner A., Northington D. J., Pennisi M., Reitznerova E.,
Rodushkin I., Sogiura N., Surberg R., Tonn S., Wiedenbeck M., Wunderli S., Xiao Y., Zack T. 2003.
Intercomparison of Boron isotope and concentration measurements. Part II: Evaluation of results. Geostand.
Newsl., 27:41–57.
Goodwin A. M. 1996. Principles of Precambrian Geology. London, Academic Press, 327p.
Gregory R. T. 1991. Oxygen history of seawater revisited: timescales for boundary event changes in the oxygen
isotope composition of seawater. In: H. P. Taylor, J. R. O‘Neil, I. R. Kaplan (eds.) Stable isotope
geochemistry: A tribute to Samuel Epstein. Geochemical Society, Special Publication, 65–76.
Griffin W. L., Belousova E. A., Shee S. R., Pearson N. J., O'Reilly S. Y. 2004. Archean crustal evolution in the
northern Yilgarn Craton: U–Pb and Hf-isotope evidence from detrital zircons. Precambrian Res., 131:231–
282.

176
Griffin W. L., Belousova E. A., O'Neill C., O'Reilly S. Y., Malkovets V., Pearson N. J., Spetsius S., Wilde S. A.
2014. The world turns over: Hadean-Archean crust-mantle evolution. Lithos, 189:2–15.
Groves D. I. & Barley M. E. 1994. Archean mineralizations. In: K. C. Condie (ed.) Archean crustal evolution.
Amsterdam, Elsevier, 461-503.
Guitreau M. 2012. Les isotopes de l'Hafnium dans les TTG et leurs zircons: témoins de la croissance des
premiers continents. Ecole Normale Supérieure de Lyon, France, PhD Thesis, 275p.
Halla J. 2005. Late Archean high-Mg granitoids (sanukitoids) in the southern Karelian domain, eastern Finland:
Pb and Nd isotopic constraints on crust-mantle interactions. Lithos, 79(1-2):161–178.
Halla J., van Hunen J., Heilimo E., Hölttä P. 2009. Geochemical and numerical constraints on Neoarchaean plate
tectonics. Precambrian Res., 174:155–162.
Halla J., Whitehouse M. J., Ahmad T., Bagai Z. 2017. Archaean granitoids: An overview and significance from a
tectonic perspective. In: J. Halla, M. J. Whitehouse, T. Ahmad, Z. Bagai (eds.) Crust-mantle interactions
and granitoid diversification: Insights from Archaean cratons. Geological Society of London, Special
Publications, 1-18.
Hamilton W. B. 2008. Archean magamtism and deformation were not products of plate tectonics. Precambrian
Res., 91:143-179.
Hamilton W. B. 2011. Plate tectoncis began in Neoproterozoic time, and plumes from deep mantle have never
operated. Lithos, 123:1-20.
Harlov D. E., Wirth R., Hetherington C. J. 2011. Fluid-mediated partial alteration in monazite: the role of
coupled dissolution-reprecipitation in element redistribution and mass transfer. Contrib. Mineral. Petrol.,
162:329-348.
Harris L. B., Godin L., Yakymchuk C. 2012. Regional shortening followed by channel flow induced collapse: A
new mechanism for ―dome-and-keel‖ geometries in Neoarchaean granite-greenstone terrains. Precambrian
Res., 212-213:139-154.
Harrison T. M., Schmitt A. K., McCulloch M. T., Lovera O. M. 2008. Early (≥ 4.5 Ga) formation of terrestrial
crust: Lu–Hf, δ18O, and Ti thermometry results for Hadean zircons. Earth Planet. Sci. Lett., 268:476-486.
Hartmann L. A., Endo I., Suita M. T. F., Santos J. O. S., Frantz J. C., Carneiro M. A., Naughton N. J., Barley M.
E. 2006. Provenance and age delimitation of Quadrilátero Ferrífero sandstones based on zircon U–Pb
isotopes. J. S. Am. Earth Sci., 20:273–285.
Hawkesworth C. J., Cawood P. A., Kemp A. I. S., Storey C., Dhuime B. 2009. A matter of preservation. Science,
323:49–50.
Hawkesworth C. J., Dhuime B., Pietranik A. B., Cawood P. A., Kemp A. I. S., Storey C. D. 2010. The
generation and evolution of the continental crust. J. Geol. Soc. London, 167:229–248.
Hawkesworth C. J., Cawood P. A., Dhuime B. 2016. Tectonics and crustal evolution. GSA Today, 26(9):4-11.
Hawthorne F. C. & Henry D. J. 1999. Classification of the minerals of the tourmaline group. Eur. J. Mineral.,
11:201–215.
Heilimo E., Halla J., Hölttä P. 2010. Discrimination and origin of the sanukitoid series: geochemical constraints
from the Neoarchean western Karelian Province (Finland). Lithos, 115:27–39.
Henry D. J. & Dutrow B. L. 2002. Metamorphic tourmaline and its petrologic applications. In: E. S. Grew & L.
M. Anovitz (eds.) Boron: Mineralogy, Petrology and Geochemistry. Reviews in Mineralogy and
Geochemistry, Mineralogical Society of America, 503–557.
Henry D. J. & Guidotti C. V. 1985. Tourmaline as a petrogenetic indicator mineral – an example from the
staurolite-grade metapelites of NW Maine. Am. Mineral., 70:1–15.
Hervig R. L., Moore G. M., Williams L. B., Peacock S. M., Holloway J. R., Roggensack K. 2002. Isotopic and
elemental partitioning of boron between hydrous fluid and silicate melt. Am. Mineral., 87:769–774.
Herz N. 1978. Metamorphic rocks of the Quadrilátero Ferrírfero, Minas Gerais, Brazil. USGS Prof. Paper 641-
C, 91p.
Heubeck C. & Lowe D. R. 1994a. Depositional and tectonic setting of the Archean Moodies Group, Barberton
Greenstone Belt, South Africa. Precambrian Res., 68:257–290.

177
Heubeck C. & Lowe D. R. 1994b. Late syndepositional deformation and detachment tectonics in the Barberton
Greenstone Belt, South Africa. Tectonics, 13:1514 – 1536.
Hickman A. H. 2004. Two contrasting granite-greenstones terranes in the Pilbara Craton, Australia: evidence for
vertical and horizontal tectonic regimes prior to 2900 Ma. Precambrian Res., 131:153–172.
Hiess J., Bennett V. C., Nutman A. P., Williams I. S. 2011. Archaean fluid-assisted crustal cannibalism recorded
by low δ18O and negative εHf(T) isotopic signatures of West Greenland granite zircon. Contrib. Mineral.
Petrol., 161:1027–1050.
Hippertt J. F., Borba R. P., Nalini H. A. 1992. O contato Formação Moeda-Complexo do Bonfim: Uma zona de
cislhamento normal na borda oeste do Quadrilátero Ferrífero, MG. Rev. Esc. Minas, 45:32-34.
Hippertt J. F. 1994. Structure indicative of helicoidal flow in a migmatitic diaper (Bação Complex, southeastern
Brazil). Tectonophysics, 234:169-996.
Hofmann A., Jagoutz O., Kröner A., Dirks P. H. G. M., Jelsma H. A. 2003. The Chirwa dome: granite
emplacement during late Archaean thrusting along the northeastern margin of the Zimbabwe craton. S. Afr.
J. Geol., 105:285–300.
Hollis J. A., Van Kranendonk M. J., Cross A. J., Kirkland C. L., Armstrong R. A., Allen C. M. 2014. Low δ 18O
zircon grains in the Neoarchean Rum Jungle Complex, northern Australia: an indicator of emergent
continental crust. Lithosphere, 6:17–25.
Hopkins M., Harrison T. M., Manning C. E. 2010. Constraints on Hadean geodynamics from mineral inclusions
in > 4 Ga zircons. Earth Planet. Sci. Lett., 298:367–376.
Hoskin P. W. O. & Black L. P. 2000. Metamorphic zircon formation by solid-state recrystallization of protolith
igneous zircon. J. Metamorph. Geol., 18:423–439.
Hurley P., Hughes H., Faure G., Fairbairn H., Pinson W. 1962. Radiogenic strontium-87 model of continent
formation. J. Geophys. Res., 67(13):5315–5334.
Hurley P. M. & Rand J. R. 1969. Pre--drift continental nuclei. Science, 164:1229--1242.
Ishikawa T. & Nakamura E. 1994. Origin of the slab component in arc lavas from across-arc variation of B and
Pb isotopes. Nature, 370:205-208.
Ivanic T. J., Van Kranendonk M. J., Kirkland C. L., Wyche S., Wingate M. T. D., Belousova E. A. 2012. Zircon
Lu-Hf isotopes and granite geochemistry of the Murchison Domain of the Yilgarn Craton: evidence for
reworking of Eoarchean crust during Meso-Neoarchean plume-driven magmatism. Lithos, 148:112–127.
Jaguin J., Gapais D., Poujol M., Boulvais P., Moyen J. F. 2012. The Murchison Greenstone Belt (South Africa):
a general tectonic framework. S. Afr. J. Geol., 115(1):65–76.
Jahn B. M., Glikson A. Y., Peucat J. J., Hickman A. H. 1981. REE geochemistry and isotopic data of Archaean
silicic volcanics and granitoids from the Pilbara Block, western Australia: implications for the early crustal
evolution. Geochim. Cosmochim. Acta, 45:1633–1652.
Jahn B. M., Auvray B., Shen Q. H., Liu D. Y., Zhang Z. Q., Dong Y. J., Ye X. J., Zhang Q. Z., Cornichet J.,
Macé J. 1988. Archaean crustal evolution in China: the Taishan complex, and evidence for juvenile crustal
addition from long-term depleted mantle. Precambrian Res., 38:381–403.
Jayananda M., Chardon D., Peucat J. J., Capdevila R. 2006. 2.61 Ga potassic granites and crustal reworking in
the western Dharwar craton, southern India: Tectonic, geochronologic and geochemical constraints.
Precambrian Res., 150(1–2):1–26.
Jiang S. Y., Palmer M. R., Yeats C. J. 2002. Chemical and boron isotopic compositions of tourmaline from the
Archean Big Bell and Mount Gibson gold deposits, Murchison Province, Yilgarn Craton, Western Australia.
Chem. Geol., 188:229–247.
Johnson T. E., Brown M., Kaus B. J. P., VanTongeren J. A. 2014. Delamination and recycling of Archaean crust
caused by gravitational instabilities. Nat. Geosci., 7:47-52.
Jordt-Evangelista H., Alkmim F. F., Marshak S. 1992. Metamorfismo progressivo e a ocorrência dos 3
polimorfos Al2SiO5 (cianita, andaluzita e sillimanita) na Formacão Sabará em Ibirité, Quadrilátero Ferrífero,
MG. Rev. Esc. Minas, 45:157-160.

178
Kaliwoda M., Marschall H. R., Marks M. A. H., Ludwig T., Altherr R., Markl G. 2011. Boron and boron isotope
systematics in the peralkaline Ilímaussaq intrusion (South Greenland) and its granitic country rocks: a record
of magmatic and hydrothermal processes. Lithos, 125:51–64.
Kamber B. S., Ewart A., Collerson K. D., Bruce M. C., McDonald G. D. 2002. Fluid-mobile trace element
constraints on the role of slab melting and implications for Archaean crustal growth models. Contrib.
Mineral. Petrol., 144:38–56.
Kamo S. L. & Davis D. W. 1994. Reassessment of Archean crustal development in the Barberton Mountain
Land, South Africa, based on U–Pb dating. Tectonics, 13(1):167–192.
Kelemen P., Hanghoj K., Greene A. 2003. One view of the geochemistry of subduction-related magmatic arcs,
with an emphasis on primitive andesite and lower crust. In: H. Holland & K. Turekian (eds.) Treatise on
Geochemistry. Oxford, Elsevier-Pergammon, 1–70.
Keller C. B. & Schoene B. 2012. Statistical geochemistry reveals disruption in secular lithospheric evolution
about 2.5 Gyr ago. Nature, 485:490–495.
Kemp A. I. S., Hawkesworth C. J., Paterson B. A., Kinny P. D. 2006. Episodic growth of the Gondwana
supercontinent from hafnium and oxygen isotopes in zircon. Nature, 439:580–583.
Kemp A. I. S., Hawkesworth C. J., Foster G. L., Paterson B. A., Woodhead J. D., Hergt J. M., Gray C. M.,
Whitehouse M. J. 2007. Magmatic and crustal differentiation history of granitic rocks from Hf–O isotopes in
zircon. Science, 315(5814):980–983.
Kemp A. I. S., Hawkesworth C. J., Collins W. J., Gray C. M., Blevin P. L. 2009a. Isotopic evidence for rapid
continental growth in an extensional accretionary orogen: the Tasmanides, eastern Australia. Earth Planet.
Sci. Lett., 284:455–466.
Kemp A. I. S., Foster G. L., Scherstén A., Whitehouse M. J., Darling J., Storey C. 2009b. Concurrent Pb–Hf
isotope analysis of zircon by laser ablation multi-collector ICPMS, with implications for the crustal
evolution of Greenland and the Himalayas. Chem. Geol., 261:244–260.
King E. M., Valley J. M., Davis D. W., Edwards G. R. 1998. Oxygen isotope ratios of Archean plutonic rocks
from granite-greenstone belts of the Superior Province: indicator of magmatic source. Precambrian Res.,
92:365–387.
Kisters A. F. M., Stevens G., Dziggel A., Armstrong R. A. 2003. Extensional detachment faulting and core-
complex formation in the southern Barberton granite-greenstone terrain, South Africa: evidence for a 3.2 Ga
orogenic collapse. Precambrian Res., 127:355–378.
Kita N. T., Ushikubo T., Fu B., Valley J. W. 2009. High precision SIMS oxygen isotope analysis and the effect
of sample topography. Chem. Geol., 264:43–57.
Klein C. & Ladeira E. A. 2000. Geochemistry and petrology of some Proterozoic banded iron-formations of the
Quadrilátero Ferrífero, Minas Gerais, Brazil. Econ. Geol., 95(2):405-427.
Kleinhanns I. C., Kramers J. D., Kamber B. S. 2003. Importance of water for Archaean granitoid petrology: a
comparative study of TTG and potassic granitoids from Barberton Mountain Land, South Africa. Contrib.
Mineral. Petrol., 145:377–389.
Kloppenburg A., White S. H., Zegers T. E. 2001. Structural evolution of the Warrawoona Greenstone Belt and
adjoining granitoid complexes, Pilbara Craton, Australia: implications for Archaean tectonic processes.
Precambrian Res., 112:107–147.
Klotzli U., Klotzli E., Gunes Z., Kosler J. 2009. Accuracy of laser ablation U–Pb zircon dating: results from a
test using five different reference zircons. Geostand. Geoanal. Res., 33:5–15.
Koglin N., Cabral A. R., Brunetto W. J., Vymazalova A. 2012. Gold-tourmaline assemblage in a Witwatersrand-
like gold deposit, Ouro Fino, Quadrilátero Ferrífero of Minas Gerais, Brazil: the composition of gold and
metallogenic implications. N. Jb. Miner. Abh., J. Mineral. Geochem., 189(3):263-273.
Koglin N., Zeh A., Cabral A. R., Gomes Jr. A. A. S., Corrêa Neto A. V., Brunetto W. J., Galbiatti H. 2014.
Depositional age and sediment source of the auriferous Moeda Formation, Quadrilátero Ferráfero of Minas
Gerais, Brazil: new constraints from U-Pb-Hf isotopes in zircon and xenotime. Precambrian Res., 255:96–
108.
Konrad-Schmolke M. & Halama R. 2014. Combined thermodynamic-geochemical modeling in metamorphic
geology: Boron as a tracer of fluid-rock interaction. Lithos, 208-209:393-414.
179
Korenaga J. 2006. Archean geodynamics and the thermal evolution of Earth. In: K. Benn, J. C. Mareschal, K. C.
Condie (eds.) Archean Geodynamics and Environments. American Geophysical Union, 7–32.
Korenaga J. 2013. Initiation and evolution of plate tectonics on Earth: Theories and observations. Annu. Rev.
Earth Pl. Sc., 41(1):117–151.
Kowalski P. M., Wunder B., Jahn S. 2013. Ab initio prediction of equilibrium boron isotope fractionation
between minerals and aqueous fluids at high P and T. Geochim. Cosmochim. Acta, 101:285–301.
Krienitz M. S., Trumbull R. B., Hellmann A., Kolb J., Meyer F. M., Wiedenbeck M. 2008. Hydrothermal gold
mineralization at the Hira Buddini gold mine, India: constraints on fluid evolution and fluid sources from
boron isotopic compositions of tourmaline. Miner. Depos., 43:421–434.
Ladeira E. A., Roeser H. M. P., Tobschal H. J. 1983. Evolução petrogenética do Cinturão de Rochas Verdes, Rio
das Velhas, Quadrilátero Ferrífero, Minas Gerais. In: Soc. Bras. Geol. Simp. Geol. Minas Gerais, 2, Belo
Horizonte, 149-165.
Lambert-Smith J. S., Rocholl A., Treloar P. J., Lawrence D. M. 2016. Discriminating fluid source regions in
orogenic gold deposits using B-isotopes. Geochim. Cosmochim. Acta, 194:57–76.
Lana C., Alkmim F. F., Armstrong R., Scholz R., Romano R., Nalini Jr. H. A. 2013. The ancestry and magmatic
evolution of Archean TTG rocks of the Quadrilátero Ferrífero province, southeast Brazil. Precambrian Res.,
231:157–173.
Lancaster P. J., Storey C. D., Hawkesworth C. J. 2014. The Eoarchean foundation of the North Atlantic Craton.
In: N. M. W. Roberts, M. Van Kranendonk, S. Parman, S. Shirey, P. D. Clift (eds.) Continent formation
through time. Geological Society of London, Special Publications, 261-279.
Lancaster P. J., Dey S., Storey C. D., Mitra A., Bhunia R. K. 2015. Contrasting crustal evolution processes in the
Dharwar craton: insights from detrital zircon U-Pb and Hf isotopes. Gondwana Res., 28:1361–1372.
Laurent O. 2012. Les changements géodynamiques à la transition Archéen-Protérozoïque: étude des granitoïdes
de la marge Nord du craton du Kaapvaal (Afrique du Sud). University Blaise Pascal, Clermont-Ferrand,
France, PhD Thesis, 512p.
Laurent O., Martin H., Doucelance R., Moyen J. F., Paquette J. L. 2011. Geochemistry and petrogenesis of high-
K ―sanukitoids‖ from the Bulai pluton, Central Limpopo Belt, South Africa: implications for geodynamic
changes at the Archaean–Proterozoic boundary. Lithos, 123:73–91.
Laurent O., Doucelance R., Martin H., Moyen J. F. 2013. Differentiation of the late-Archaean sanukitoid series
and some implications for crustal growth: insights from geochemical modelling on the Bulai pluton, Central
Limpopo Belt, South Africa. Precambrian Res., 227:186–203.
Laurent O., Martin H., Moyen J. F., Doucelance R. 2014. The diversity and evolution of late-Archean granites:
evidence for the onset of a ―modern-style‖ plate tectonics between 3.0 and 2.5 Ga. Lithos, 205:208–235.
Laurie A. & Stevens G. 2012. Water-present eclogite melting to produce Earth‘s early felsic crust. Chem. Geol.,
314:83–95.
Leeman W. P. & Tonarini S. 2001. Boron isotopic analysis of proposed borosilicate mineral reference samples.
J. Geostand. Geoanal., 25:399–403.
Lin S. 2005. Synchronous vertical and horizontal tectonism in the Neoarchean: kinematic evidence from a
synclinal keel in the northwestern Superior Craton, Manitoba. Precambrian Res., 139:181–194.
Liu S. W., Pan Y. M., Xie Q. L., Zhang J., Li Q. 2004. Archean geodynamics in the Central Zone, North China
Craton: constraints from geochemistry of two constrasting series of granitoids in the Fuping and Wutai
complexes. Precambrian Res., 130:229–249.
Lobach-Zhuchenko S. B., Rollinson H. R., Chekulaev V. P., Arestova N. A., Kovalenko A. V., Ivanikov V. V.,
Guseva N. S., Sergeev S. A., Matukov D. I., Jarvis K. E. 2005. The Archaean sanukitoid series of the Baltic
Shield: geological setting, geochemical characteristics and implications for their origin. Lithos, 79:107–128.
Lobach-Zhuchenko S. B., Rollinson H., Chekulaev V. P., Savatenkov V. M., Kovalenko A. V., Martin H.,
Guseva N. S., Arestova N. A. 2008. Petrology of a late Archaean, highly potassic, sanukitoid pluton from
the Baltic Shield: insights into late Archaean mantle metasomatism. J. Pet., 49(3):393–420.

180
Lobato L. M., Ribeiro-Rodrigues L. C., Vieira F. W. R. 2001. Brazil‘s premier gold province. Part II: geology
and genesis of gold deposits in the Archean Rio das Velhas greenstone belt, Quadrilátero Ferrífero. Miner.
Depos., 36:249-277.
London D. & Manning D. A. C. 1995. Chemical variation and significance of tourmaline from southwest
England. Econ. Geol., 90:495-519.
London D., Morgan G. B., Wolf M. B. 2002. Boron in granitic rocks and their contact aureoles. In: E. S. Grew &
L. M. Anovitz (eds.) Boron: Mineralogy, Petrology and Geochemistry. Reviews in Mineralogy and
Geochemistry, Mineralogical Society of America, 503–557.
Longstaffe F. J. & Schwarcz H. P. 1977. 18O/16O of Archean clastic metasedimentary rocks: a petrogenetic
indicator for Archean gneisses? Geochim. Cosmochim. Acta, 41:1303–1312.
Lowe D. R. & Byerly G. R. 1999. Geological evolution of the Barberton Greenstone Belt, South Africa.
Geological Society of America, Special Paper, 319p.
Ludwig K. R. 2003. Isoplot/Ex Version 3.00: A Geochronological Toolkit for MicrosoftExcel. Berkeley
Geochronology Centre, Berkeley, CA.
Maas R., Kinny P. D., Williams I. S., Froude D. O., Compston W. 1992. The Earth‘s oldest known crust: A
geochronological and geochemical study of 3900–4200 Ma old detrital zircons from Mt. Narryer and Jack
Hills, Western Australia. Geochim. Cosmochim. Acta, 56(3):1281–1300.
MacGregor A. M. 1951. Some milestones in the Precambrian of Southern Rhodesia. Geological Society of South
Africa, 54:27–71.
MacGregor J. R., Grew E. S., De Hoog J. C. M., Harley S. L., Kowalski P. M., Yates M. G., Carson C. J. 2013.
Boron isotopic composition of tourmaline, prismatine, and grandidierite from granulite facies paragneisses
in the Larsemann Hills, Prydz Bay, East Antarctica: Evidence for a non-marine evaporate source. Geochim.
Cosmochim. Acta, 123:261-283.
Machado M. M. M. 2009. Construindo a imagem geológica do Quadrilátero Ferrífero: conceitos e
representações. Universidade Federal de Minas Gerais, Brazil, PhD Thesis, 296 p.
Machado N. & Carneiro M. A. 1992. U–Pb evidence of late Archean tectono-thermal activity in the southern São
Francisco shield, Brazil. Can. J. Earth Sci., 29:2341–2346.
Machado N., Noce C. M., Ladeira E. A., Belo de Oliveira O. 1992. U-Pb Geochronology of Archean magmatism
and Proterozoic metamorphism in the Quadrilátero Ferrífero, southern São Francisco craton, Brazil. Geol.
Soc. Am. Bull., 104:1221–1227.
Machado N., Schrank A., Noce C. M., Gauthier G. 1996. Ages of detrital zircon from Archean-Paleoproterozoic
sequences: implications for Greenstone Belt setting evolution of a Transamazonian foreland basin in
Quadrilátero Ferrífero, southeast Brazil. Earth Planet. Sci. Lett., 141:259–276.
Marschall H. R. & Jiang S. Y. 2011. Tourmaline isotopes: no element left behind. Elements, 7:313–319.
Marschall H. R., Ludwig T., Altherr R., Kalt A., Tonarini S. 2006. Syros metasomatic tourmaline: Evidence for
very high-δ11B fluids in subduction zones. J. Pet., 47:1915-1942.
Marshak S. & Alkmim F. F. 1989. Proterozoic contraction/extension tectonics of the southern São Francisco
region, Minas Gerais, Brazil. Tectonics, 8:555-571.
Marshak S., Alkmim F. F., Jordt-Evangelista H. 1992. Proterozoic crustal extension and the generation of the
dome-and-keel structure in an Archean granite – greenstone terrane. Nature, 357:491–493.
Marshak S., Tinkham D., Alkmim F. F., Brueckner H. K., Bornhorst T. 1997. Dome-and-keel provinces formed
during Paleoproterozoic orogenic collapse-Diapir clusters or core complexes? Examples from the
Quadrilátero Ferrífero (Brazil) and the Penokean Orogen (USA). Geology, 25:415–418.
Martin H. 1986. Effect of steeper Archaean geothermal gradient on geochemistry of subduction-zone magmas.
Geology, 14:753–756.
Martin H. 1987. Petrogenesis of Archaean trondhjemites, tonalites and granodiorites from Eastern Finland:
major and trace element geochemistry. J. Pet., 28:921–953.
Martin H. 1993. The mechanisms of petrogenesis of the Archaean continental crust, comparison with modern
processes. Lithos, 30:373–388.

181
Martin H. 1994. The Archaean grey gneisses and the genesis of continental crust. In: K. C. Condie (ed.)
Archaean crustal evolution. Developments in Precambrian Geology, Elservier, Amsterdam, 205–260.
Martin H. 1999. Adakite magmas: modern analogs of Archaean granitoids. Lithos, 46:411–429.
Martin H. & Moyen J. F. 2002. Secular changes in TTG composition as markers of the progressive cooling of
the Earth. Geology, 30:319–322.
Martin H., Smithies R., Rapp R. P., Moyen J. F., Champion D. 2005. An overview of adakite, TTG, and
sanukitoid: relationships and some implications for crustal evolution. Lithos, 79(1–2):1–24.
Martin H., Moyen J. F., Rapp R. P. 2009. The sanukitoid series: magmatism at the Archaean–Proterozoic
transition. Earth Environ. Sci. Trans. R. Soc. Edinburgh, 100(1–2):15–33.

Martin H., Moyen J. F., Guitreau M., Blichert-Toft J., Le Pennec J. L. 2014. Why Archean TTG cannot be
generated by MORB melting in subduction zones. Lithos, 198–199:1–13.
Martin C., Ponzevera E., Harlow G. 2015. In situ lithium and boron isotope determinations in mica, pyroxene,
and serpentine by LA-MC-ICP-MS. Chem. Geol., 412:107-116.
Martínez Dopico C. I., Lana C., Moreira H. S., Cassino L. F., Alkmim F. F. 2017. U-Pb ages and Hf-isotope data
of detrital zircons from the late Neoarchean-Paleoproterozoic Minas Basin, SE Brazil. Precambrian Res.,
291:143-161.
Maruyama S., Liou J. G., Terabayashi M. 1996. Blueschists and eclogites of the world and their exhumation. Int.
Geol. Rev., 38:485–594.
Maxwell C. H. 1958. The batatal formation. Soc. Bras. Geol. Bol., 7(2):60-61.
McDonough W. F. & Sun S. S. 1995. The composition of the Earth. Chem. Geol., 120:223–253.
Mendes M. D. C. O., Lobato L. M., Suckau V., Lana C. 2014. Datação U-Pb in situ por LA-ICPMS em zircões
detríticos da Formação Cercadinho, Supergrupo Minas. Geol. USP. Sér. Cient., 14(1):55-68.
Meyer C., Wunder B., Meixner A., Romer R. L., Heinrich W. 2008. Boron-isotope fractionation between
tourmaline and fluid: an experimental re-investigation. Contrib. Mineral. Petrol., 156:259–267.
Mikkola P., Huhma H., Heilimo E., Whitehouse M. 2011. Archean crustal evolution of the Suomussalmi district
as part of the Kianta Complex, Karelia: constraints from geochemistry and isotopes of granitoids. Lithos,
125:287–307.
Mikkola P., Lauri L. S., Käpyaho A. 2012. Neoarchean leucogranitoids of the Kianta Complex, Karelian
Province, Finland: Source characteristics and processes responsible for the observed heterogeneity.
Precambrian Res., 206–207:72–86.
Mikova J., Košler J., Wiedenbeck M. 2014. Matrix effects during laser ablation MC ICP-MS analysis of boron
isotopes in tourmaline. J. Anal. At. Spectrom., 29:903-914.
Mitchell R. N., Bleeker W., Breemen O. V., Lecheminant T. N., Peng P., Nilsson M. K. M., Evans D. A. D.
2014. Plate tectonics before 2.0 Ga: evidence from paleomagnetism of cratons within supercontinent Nuna.
Am. J. Sci., 314:878-894.
Molnár F., Mänttäri I., O‘Brien H., Lahaye Y., Pakkanen L., Johanson B., Käpyaho A., Sorjonen-Ward P.,
Whitehouse M., Sakellaris G. 2016. Boron, sulphur and copper isotope systematics in the orogenic gold
deposits of the Archaean Hattu schist belt, eastern Finland. Ore Geol. Rev., 77:133-162.
Monani S. & Valley J. W. 2001. Oxygen isotope ratios of zircon: magma genesis of low δ 18O granites from the
British Tertiary igneous province, western Scotland. Earth Planet. Sci. Lett., 184:377–392.
Moraes M. A. S. 1985. Reconhecimento de facies sedimentares em rochas metamórficas da região de Outro
Preto (MG). In: Soc. Bras. Geol. Simp. Geol. Minas Gerais, 3, Belo Horizonte, 84-93.
Moreira H., Lana C., Nalini Jr. H. A. 2016. The detrital zircon record of an Archaean convergent basin in the
Southern São Francisco Craton, Brazil. Precambrian Res., 275:84-99.
Morgan G. B. & London D. 1989. Experimental reactions of amphibolite with boron-bearing aqueous fluids at
200 MPa; implications for tourmaline stability and partial melting in mafic rocks. Contrib. Mineral. Petrol.,
102:281–297.

182
Moyen J. F. 2011. The composite Archaean grey gneisses: petrological significance, and evidence for a non-
unique tectonic setting for Archaean crustal growth. Lithos, 123(1–4):21–36.
Moyen J. F. & Stevens G. 2006. Experimental constraints on TTG petrogenesis: implications for Archean
geodynamics. In: K. Benn, J. C. Mareschal, K. C. Condie (eds.) Archean Geodynamics and Environments.
American Geophysical Union, 149–178.
Moyen J. F. & Martin H. 2012. Forty years of TTG research. Lithos, 148:312–336.
Moyen J. F., Martin H., Jayananda M. 2001. Multi-element geochemical modelling of crust–mantle interactions
during late-Archaean crustal growth: the Closepet granite (South India). Precambrian Res., 112:87–105.
Moyen J. F., Martin H., Jayananda M., Auvray B. 2003. Late Archaean granites: a typology based on the
Dharwar Craton (India). Precambrian Res., 127:103–123.
Moyen J. F., Stevens G., Kisters A. F. M., Belcher R. W. 2007. TTG plutons of the Barberton granitoid-
greenstone terrain, South Africa. In: M. J. Van Kranendonk, R. H. Smithies, V. Bennett (eds.) Earth's oldest
rocks. Developments in Precambrian Geology, Amsterdam, Elsevier, 606–668.
Moyen J. F. & van Hunen J. 2012. Short-term episodicity of Archaean plate tectonics. Geology, 40(5):451–454.
Murakami T., Chakoumakos B. C., Ewing R. C., Lumpkin G. R., Weber W. J. 1991. Alpha-decay event damage
in zircon. Am. Mineral., 76:1510–1532.
Naeraa T., Scherstein A., Rosing M. T., Kemp A. I. S., Hofmann J. E., Koldelt T. F., Whitehouse M. J. 2012.
Hafnium isotope evidence for a transition in the dynamics of continental growth 3.2 Ga ago. Nature,
485:627–630.
Nakano T. & Nakamura E. 2001. Boron isotope geochemistry of metasedimentary rocks and tourmalines in a
subduction zone metamorphic suite. Phys. Earth Planet. Inter., 127:233–252.
Nebel O., Nebel-Jacobsen Y., Mezger K., Berndt J. 2007. Initial Hf isotope compositions in magmatic zircon
from early Proterozoic rocks from the Gawler Craton, Australia: a test for zircon model ages. Chem. Geol.,
241:27–37.
Nebel-Jacobsen Y. J., Münker C., Nebel O., Gerdes A., Mezger K., Nelson D. R. 2010. Reworking of Earth‘s
first crust: Constraints from Hf isotopes in Archean zircons from Mt. Narryer. Precambrian Res.,
182(3):175-186.
Nicoli G., Mojen J. F., Stevens G. 2016. Diversity of burial rates in convergent settings decreased as Earth aged.
Scientific Reports, 6:26359.
Noce C. M., Teixeira W., Machado N. 1997. Geoquímica dos gnaisses TTG e granitoides neoarqueanos do
Complexo Belo Horizonte, Quadrilátero Ferrífero, Minas Gerais. Rev. Bras. Geocienc., 27:25–32.
Noce C. M., Machado N., Teixeira W. 1998. U–Pb geochronology of gneisses and granitoids in the Quadrilátero
Ferrífero (southern São Francisco craton): age constraints for Archean and Paleoproterozoic magmatism and
metamorphism. Rev. Bras. Geocienc., 28:95–102.
Noce C. M., Zuccheti M., Baltazar O. F., Armstrong R., Dantas E., Renger F. E., Lobato L. M. 2005. Age of
felsic volcanism and the role of ancient continental crust in the evolution of the Neoarchean Rio das Velhas
greenstone belt (Quadrilátero Ferrífero, Brazil): U–Pb zircon dating of volcaniclastic greywackes.
Precambrian Res., 141:67–82.
Noce C. M., Tassinari C., Lobato L. M. 2007. Geochronological framework of the Quadrilátero Ferrífero, with
emphasis on the age of gold mineralization hosted in Archean greenstone belts. Ore Geol. Rev., 32:500-510.
O‘Connor J. T. 1965. A classification for quartz-rich igneous rocks based on feldspar ratios. USGS Prof. Paper
525–B, 79–84.
Oliveira M. A., Dall'Agnol R., Scaillet B. 2010a. Petrological constraints on crystallization conditions of
Mesoarchean sanukitoid rocks, southeastern Amazonian craton, Brazil. J. Pet., 51:2121–2148.
Oliveira E. P., McNaughton N. J., Armstrong R. 2010b. Mesoarchaean to Palaeoproterozoic growth of the
northern segment of the Itabuna-Salvador-Curaçá orogen, São Francisco craton, Brazil. In: T. M. Kusky, M.
G. Zhai, W. Xiao (eds.) The evolving continents: understanding processes of continental growth. Geological
Society of London, Special Publications, 263–286.

183
Oliveira E. P., Silveira E. M., Söderlund U., Ernst R. E. 2013. U-Pb ages and geochemistry of mafic dyke
swarms from the Uauá Block, São Francisco Craton, Brazil: LIPs remnants relevant for Late Archaean
break-up of a supercraton. Lithos, 174:308-322.
O‘Neil J., Carlson R. W., Francis D., Stevenson R. K. 2008. Neodymium-142 evidence for Hadean mafic crust.
Science, 321:1828-1831.
O‘Neill C., Lenardic A., Moresi L., Torsvik T. H., Lee C. T. A. 2007. Episodic Precambrian subduction. Earth
Planet. Sci. Lett., 262:552-562.
O'Rourke J. E. 1957. The stratigraphy of metamorphic rocks of the Rio de Pedras and Gandarela quadrangles,
Minas Gerais, Brazil. University of Wisconsin, Wisconsin, USA, PhD Thesis, 106p.
Ota T., Kobayashi K., Kunihiro T., Nakamura E. 2008. Boron cycling by subducted lithosphere; insights from
diamondiferous tourmaline from the Kokchetav ultrahigh-pressure metamorphic belt. Geochim. Cosmochim.
Acta, 72:3531-3541.
Pal D. C., Trumbull R. B., Wiedenbeck M. 2010. Chemical and boron isotope compositions of tourmaline from
the Jaduguda U (-Cu-Fe) deposit, Singhbhum shear zone, India: Implications for the sources and evolution
of mineralizing fluids. Chem. Geol., 277:245-260.
Palmer M. R., London D., Morgan G., Babb H. 1992. Experimental determination of fractionation of 11B/10B
between tourmaline and aqueous vapour: a temperature and pressure-dependent isotopic system. Chem.
Geol., 101:123–129.
Parman S. W. 2007. Helium isotopic evidence for episodic mantle melting and crustal growth. Nature, 446:900–
903.
Parmenter A. C., Lin S., Corkery M. T. 2006. Structural evolution of the Cross Lake greenstone belt in the
northwestern Superior Province, Manitoba: implications for the relationship between vertical and horizontal
tectonism. Can. J. Earth Sci., 43:767–787.
Patchett P. J. & Arndt N. T. 1986. Nd isotopes and tectonics of 1.9–1.7 Ga crustal genesis. Earth Planet. Sci.
Lett., 78:329–338.
Patiño Douce A. E. & Beard J. S. 1995. Dehydration melting of biotite gneiss and quartz amphibolite from 3 to
15 kbar. J. Pet., 36:707–738.
Patiño Douce A. E. 2005. Vapor-absent melting of tonalite at 15–32 kbar. J. Pet., 46:275–290.
Payne J. L., McInerney D. J., Barovich K. M., Kirkland C. L., Pearson N. J., Hand M. 2016. Strengths and
limitations of zircon Lu-Hf and O isotopes in modelling crustal growth. Lithos, 248-251:175–192.
Pearson D. G., Parman S. W., Nowell G. M. 2007. A link between large mantle melting events and continent
growth seen in osmium isotopes. Nature, 449:202-205.
Peck W. H., King E. M., Valley J. V. 2000. Oxygen isotope perspective on Precambrian crustal growth and
maturation. Geology, 28:363–366.
Pedrosa-Soares A. C., Noce C. M., Wiedemann C. M., Pinto C. P. 2001. The Araçuaí-West Congo orogen in
Brazil: An overview of a confined orogen formed during Gondwanaland assembly. Precambrian Res.,
110:307–323.
Percival J. A., McNicoll V., Bailes A. H. 2006. Strike-slip juxtaposition of ca. 2.72 Ga juvenile arc and >2.98 Ga
continent margin sequences and its implications for Archean terrane accretion, western Superior Province,
Canada. Can. J. Earth Sci., 43(7):895-927.
Pietranik A. B., Hawkesworth C. J., Storey C. D., Kemp A. I. S., Sircombe K. N., Whitehouse M. J., Bleeker W.
2008. Episodic mafic crust formation from 4.5–2.8 Ga: new evidence from detrital zircons, Slave Craton,
Canada. Geology, 36:875–878.
Pires F. R. M. 1979. Structural Geology and Stratigraphy at the Junction of the Serra Do Curral Anticline and
the Moeda Syncline, Quadrilátero Ferrífero, Minas Gerais, Brazil. University of Michigan, Ann Arbor,
USA, PhD Thesis.
Pomerene J. B. 1964. The Geology and Ore Deposits of the Belo Horizonte, Ibirité and Macacos Quadrangles,
Minas Gerais, Brazil. USGS Prof. Paper 341-D.
Poupinet G. & Shapiro N. M. 2009. Worldwide distribution of ages of the continental lithosphere derived from a
global seismic tomographic model. Lithos, 109:125–130.
184
Prabhakar B. C., Jayananda M., Shareef M., Kano T. 2009. Petrology and geochemistry of Late Archean
granitoids in the northern part of Eastern Dharwar, southern India: implications for transitional geodynamic
setting. J. Geol. Soc. India, 74:299–317.
Rapp R. P. & Watson E. B. 1995. Dehydration melting of metabasalt at 8-32 kbar: implications for continental
growth and crust-mantle recycling. J. Pet., 36:891–931.
Rapp R. P.,Watson E. B., Miller C. F. 1991. Partial melting of amphibolite/eclogite and the origin of Archaean
trondhjemites and tonalites. Precambrian Res., 51:1–25.
Rapp R. P., Shimizu N., Norman M. D., Applegate G. S. 1999. Reaction between slab-derived melts and
peridotite in the mantle wedge: experimental constraints at 3.8 GPa. Chem. Geol., 160:335–356.
Rapp R. P., Shimizu N., Norman M. D. 2003. Growth of early continental crust by partial melting of eclogite.
Nature, 425:605–609.
Rapp R. P., Norman M. D., Laporte D., Yaxley G. M., Martin H., Foley S. F. 2010. Continent formation in the
Archean and chemical evolution of the cratonic lithosphere: melt–rock reaction experiments at 3–4 GPa and
petrogenesis of Archean Mg-diorites (sanukitoids). J. Pet., 51(6):1237–1266.
Reis L. A., Martins-Neto M. A., Gomes N. S., Endo I. 2002. A bacia de antepaís paleoproterzóica Sabará,
Quadrilátero Ferrífero, MG. Rev. Bras. Geocienc., 32:43-58.
Renger F. E., Silva R. M. P., Suckau V. E. 1988. Ouro nos conglomerados da Formação Moeda, Sinclinal de
Gandarela, Quadrilátero Ferrífero, Minas Gerais. In: Congr. Bras. Geol., 35, Anais, 44-57.
Renger F. E., Noce C. M., Romano A. W., Machado N. 1995. Evolução sedimentary do Supergrupo Minas: 500
Ma de registro geológico no Quadrilátero Ferrífero, Minas Gerais, Brasil. Geonomos, 2(1):1–11.
Rey P. & Houseman G. 2006. Lithospheric scale gravitational flow: the impact of body forces on orogenic
processes from Archaean to Phanerozoic. In: S. J. H. Buiter & G. Schreurs (eds.) Analogue and Numerical
Modelling of Crustal-scale Processes. Geological Society of London, Special Publication, 153-167.
Rey P. F. & Coltice N. 2008. Neoarchean strengthening of the lithosphere and the coupling of the Earth‘s
geochemical reservoirs. Geology, 36:635–638.
Reymer A. & Schubert G. 1984. Phanerozoic addition rates to the continental crust and crustal growth.
Tectonics, 3:63–77.
Reymer A. & Schubert G. 1986. Rapid growth of some major segments of continental crust. Geology, 14:299-
302.
Rino S., Komiya T., Windley B. F., Katayama I., Motoki A., Hirata T. 2004, Major episodic increases of
continental crustal growth determined from zircon ages of river sands; implications for mantle overturns in
the Early Precambrian. Phys. Earth Planet. Inter., 146:369–394.
Roberts N. M. W. & Spencer C. J. 2014. The zircon archive of continent formation through time. In: N. M. W.
Roberts, M. Van Kranendonk, S. Parman, S. Shirey, P. D. Clift (eds.) Continent formation through time.
Geological Society of London, Special Publications, 197-225.
Rogers J. J. 1996. A history of continents in the past three billion years. J. Geol., 91-107.
Rogers J. J. & Santosh M. 2004. Continents and Supercontinents. Oxford University Press.
Romano R., Lana C., Alkmim F. F., Stevens G., Armstrong R. 2013. Stabilization of the southern portion of the
São Francisco Craton, SE Brazil, though a long-lived period of potassic magmatism. Precambrian Res.,
224:143–159.
Rosière C. A., Spier C. A., Rios F. J., Suckau V. E. 2008. The itabirites of the Quadrilátero Ferrífero and related
high-grade iron ore deposits: an overview. Rev. Econ. Geol., 15:223-254.
Sandiford M. & McLaren S. 2002. Tectonic feedback and the ordering of heat producing elements within the
continental lithosphere. Earth Planet. Sci. Lett., 204:133–150.
Sandiford M. & Powell R. 1986. Deep crustal metamorphism during continental extension: ancient and modern
examples. Earth Planet. Sci. Lett., 79:151–158.
Santos M. M., Lana C. C., Cipriano R. A. S. 2014. Development of zircon standards for U–Pb geochronology by
laser ablation. In: South-American Symposium on Isotope Geology, 9, São Paulo, 296p.

185
Satkoski A. M., Bickford M. E., Samson S. D., Bauer R. L., Mueller P. A., Kamenov G. D. 2013. Geochemical
and Hf–Nd isotopic constraints on the crustal evolution of Archean rocks from the Minnesota River Valley,
USA. Precambrian Res., 224:36–50.
Scambelluri M. & Tonarini S. 2012. Boron isotope evidence for shallow fluid transfer across subduction zones
by serpentinized mantle. Geology, 40:907–910.
Schmidt C., Thomas R., Heinrich W. 2005. Boron speciation in aqueous fluids at 22 to 600C and 0.1 MPa to 2
GPa. Geochim. Cosmochim. Acta, 69:275–281.
Schorscher H. D. 1978. Komatiítos na estrutura ―Greenstone Belt‖ Série Rio das Velhas, Quadrilátero Ferrífero,
Minas Gerais, Brasil. In: Congr. Bras. Geol., Anais, 292-293.
Schorscher H. D. 1992. Arcabouço petrográfico e evolução crustal dos terrenos precambrianos do sudeste de
Minas Gerais: Quadrilátero Ferrífero, Espinhaço Meridional, e domos granitoides-gnaissicos adjacentes.
Universidade de São Paulo, Brazil, Tese de Livre-Docência, 393p.
Schrank A. & Machado N., 1996a. Idades U-Pb em monazitas e zircões das Minas de Morro Velho e Passagem
de Mariana, Quadrilátero Ferrífero, MG. In: Congr. Bras. Geol., 39, Anais, 6:470-472.
Schrank A. & Machado N. 1996b. Idades U-Pb em monazitas e zircões do distrito aurífero de Caeté, da Mina de
Cuiabá e do Depósito de Carrapato, Quadrilátero Ferrífero (MG). In: Congr. Bras. Geol., 39, Anais, 6:473-
475.
Seixas L. A. R., David J., Stevenson R. 2012. Geochemistry, Nd isotopes and U-Pb geochronology of a 2350 Ma
TTG suite, Minas Gerais, Brazil: implications for the crustal evolution of the southern São Francisco craton.
Precambrian Res., 196:61-80.
Seixas L. A. R., Bardintzeff J. M., Stevenson R., Bonin B. 2013. Petrology of the high-Mg tonalites and dioritic
enclaves of the ca. 2130 Ma Alto Maranhão suite: evidence for a major juvenile crustal addition event during
the Rhyacian orogenesis, Mineiro Belt, southeast Brazil. Precambrian Res., 238:18-41.
Shirey S. B. & Hanson G. N. 1984. Mantle-derived Archaean monzodiorites and trachyandesites. Nature,
310:222–224.
Shirey S. B. & Richardson S. H. 2011. Start of the Wilson cycle at 3 Ga shown by diamonds from subcontinental
mantle. Science, 333:434–436.
Siegel K., Wagner T., Trumbull R. B., Jonsson E., Matalin G., Wälle M., Heinrich C. A. 2016. Stable isotope (B,
H, O) and mineral-chemistry constraints on the magmatic to hydrothermal evolution of the Varuträsk rare-
element pegmatite (Northern Sweden). Chem. Geol., 421:1-16.
Simmons G. C. 1968. Geology and Iron Deposits of the Western Serra Do Curral, Minas Gerais. USGS Prof.
Paper 341-G, 968p.
Simmons G. C. & Maxwell C. H. 1961. Grupo Tamanduá da Serie Rio das Velhas: Brazil. Dep. Nac. Prod.
Min., 28p.
Singh J. & Johannes W. 1996. Dehydration melting of tonalites. Part II. Composition of melts and solids.
Contrib. Mineral. Petrol., 125:26–44.
Sizova E., Gerya T., Brown M., Perchuk L. L. 2010. Subduction styles in the Precambrian: insight from
numerical experiments. Lithos, 116:209–229.
Sizova E., Gerya T., Brown M. 2014. Contrasting styles of Phanerozoic and Precambrian continental collision.
Gondwana Res., 25:522–545.
Skjerlie K. P. & Johnston A. D. 1993. Fluid-absent melting behaviour of a F-rich tonalitic gneiss at mid-crustal
pressures: implications for the generation ofanorogenic granites. J. Pet., 34:785–815.
Slack J. F. & Trumbull R. B. 2011. Tourmaline as a recorder of ore-forming processes. Elements, 7:321–326.
Sláma J., Košler J., Condon D. J., Crowley J. L., Gerdes A., Hanchar J. M., Horstwood M. S. A., Morris G. A.,
Nasdala L., Norberg N., Schaltegger U., Schoene B., Tubrett M. N., Whitehouse M. J. 2008. Plešovice
zircon – a new natural reference material for U–Pb and Hf isotopic microanalysis. Chem. Geol., 249:1–35.
Smithies R. H. 2000. The Archaean tonalite–trondhjemite–granodiorite (TTG) series is not an analogue of
Cenozoic adakite. Earth Planet. Sci. Lett., 182:115–125.

186
Smithies R. H. & Champion D. C. 2000. The Archaean high-Mg diorite suite: links to tonalite–trondhjemite–
granodiorite magmatism and implications for early Archaean crustal growth. J. Pet., 41(12):1653–1671.
Smithies R. H., Champion D. C., Sun S. S. 2004. Evidence for early LREE-enriched mantle source regions:
Diverse magmas from the c. 3.0 Ga Mallina Basin, Pilbara Craton, NW Australia. J. Pet., 45(8):1515–1537.
Smithies R. H., Champion D. C., Van Kranendonk M. J. 2007. The oldest well-preserved felsic volcanic rocks
on Earth: Geochemical clues to the early evolution of the Pilbara Supergroup and implications for the
growth of Paleoarchean protocontinent. In: M. J. Van Kranendonk, R. H. Smithies, V. Bennett (eds.) Earth's
oldest rocks. Developments in Precambrian Geology, Amsterdam, Elsevier, 339-367.
Spencer C. J., Cawood P. A., Hawkesworth C. J., Raub T. D., Prave A. R., Roberts N. M. W. 2014. Proterozoic
onset of crustal reworking and collisional tectonics: Reappraisal of the zircon oxygen isotope record.
Geology, 42(5):451–454.
Spier C. A., de Oliveira S. M. B., Sial A. N., Rios F. J. 2007. Geochemistry and genesis of the banded iron
formations of the Cauê Formation, Quadrilátero Ferrífero, Minas Gerais, Brazil. Precambrian Res.,
152:170–206.
Seixas L. A. R., David J., Stevenson R. 2012. Geochemistry, Nd isotopes and U–Pb geochronology of a 2350
Ma TTG suite, Minas Gerais, Brazil: implications for the crustal evolution of the southern São Francisco
craton. Precambrian Res., 196:61–80.
Sichel S. E. 1983. Geologia das rochas Pré-Cambrianas da região de Barão de Cocais e geoquímica preliminar
dos komatiítos do Supergrupo Rio das Velhas, Quadrilátero Ferrífero, MG. Universidade Federal do Rio de
Janeiro, Brazil, MSc Thesis, 232p.
Sircombe K. N. 2004. AgeDisplay: an EXCEL workbook to evaluate and display univariate geochronological
data using binned frequency histograms and probability density distributions. Comput. Geosci., 30:21-31.
Sleep N. H. 2007. Plate tectonics through time. In: G. Schubert (ed.) Treatise on Geophysics. Oxford, Elsevier-
Pergammon, 101–117.
Snowden P. A. & Bickle M. J. 1976. The Chinamora Batholith: diapiric intrusion or interference fold?
Geological Society of London, 132:131–137.
Souza P. C. & Miller G. 1984. Primeiras estruturas algais comprovadas na Formação Gandarela, Quadrilátero
Ferrífero. Rev. Esc. Minas, 2:161-198.
Steenfelt A., Garde A. A., Moyen J. F. 2005. Mantle wedge involvement in the petrogenesis of Archaean grey
gneisses in West Greenland. Lithos, 79:207–228.
Stein M. & Hofmann A. W. 1994. Mantle plumes and episodic crustal growth. Nature, 372:63-68.
Stern R. J. 2005. Evidence from ophiolites, blueschists, and ultrahigh-pressure metamorphic terrains that the
modern episode of subduction tectonics began in Neoproterozoic time. Geology, 33:557–560.
Stern R. A. 2008. Modern-style plate tectonics began in Neoproterozoic time: An alternative interpretation of
Earth‘s tectonic history. In: K. Condie & V. Pease (eds.) When did plate tectonics begin? Geological Society
of America, Special Paper, 265-280.
Stern R. A. & Hanson G. N. 1991. Archaean high-Mg granodiorites: a derivative of light rare earth enriched
monzodiorites of mantle origin. J. Pet., 32:201–238.
Stern R. A., Hanson G. N., Shirey S. B. 1989. Petrogenenesis of mantle-derived, LILE-enriched Archean
monzodiorites and trachyandesites (sanukitoids) in southwestern Superior Province. Can. J. Earth Sci.,
26:1688–1712.
Stern R. J., Tsujimori T., Harlow G., Groat L. A. 2013. Plate tectonic gemstones. Geology, 41(7):723-726.
Stevens G. & Moyen J. F. 2007. Metamorphism in the Barberton granite greenstone terrain: A record of
Paleoarchean accretion. In: M. J. Van Kranendonk, R. H. Smithies, V. Bennett (eds.) Earth's oldest rocks.
Developments in Precambrian Geology, Amsterdam, Elsevier, 669-698.
Sylvester P. J. 1994. Archaean granite plutons. In: K. C. Condie (ed.) Archaean crustal evolution. Developments
in Precambrian Geology, Amsterdam, Elsevier, 261–314.
Taylor S. R. 1977. Island arc models and the composition of the continental crust. In: M. Talwani & W. C.
Pitman III (eds.) Island arcs, deep sea trenches, and back-arc basins. Maurice Ewing Ser. 1. AGU,
Washington DC, 325–335.
187
Taylor S. R. & McLennan S. M. 1985. The continental crust: Its composition and evolution. Blackwell, Oxford,
312p.
Taylor S. R. & McLennan S. M. 1995. The geochemical evolution of the continental crust. Rev. Geophys.,
33:241--265.
Teixeira W. & Figueiredo M. C. H. 1991. An outline of Early Proterozoic crustal evolution in the São Francisco
Craton, Brazil: a review. Precambrian Res., 53:1–22.
Teixeira W., Carneiro M. A., Noce C. A., Machado N., Sato K., Taylor P. N. 1996. Pb, Sr and Nd isotope
constraints on the Archean evolution of gneissic granitoid complexes in the southern São Francisco Craton,
Brazil. Precambrian Res., 78:151–164.
Teixeira W., Ávila C. A., Dussin I. A., Corrêa Neto A. V., Bongiolo E. M., Santos J. O., Barbosa N. S. 2015. A
juvenile accretion episode (2.35–2.32 Ga) in the Mineiro belt and its role to the Minas accretionary orogeny:
Zircon U–Pb–Hf and geochemical evidences. Precambrian Res., 256:148–169.
Tiepolo M., Bouman C., Vannucci R., Schwieters J. 2006. Laser ablation multicollector ICPMS determination of
δ11B in geological samples. Appl. Geochem., 21:788-801.
Tindle A. G., Breaks F. W., Selway J. B. 2002. Tourmaline in petalite-subtype granitic pegmatites: Evidence of
fractionation and contamination from the Pakeagama Lake and Separation Lake areas of northwestern
Ontario, Canada. Can. Mineral., 40:753-788.
Tinkham D. K. & Marshak S. 2004. Precambrian dome-and-keel structure in the Penokean Orogenic Belt of
Northern Michigan, USA. Geological Society of America, Special Papers, 321–338.
Tonarini S., Dini A., Pezzotta F., Leeman W. P. 1998. Boron isotopic composition of zoned (schorl–elbaite)
tourmalines, Mt. Capanne Li–Cs pegmatites, Elba (Italy). Eur. J. Mineral., 11:941–952.
Tonarini S., Forte C., Petrini R., Ferrara G. 2003. Melt/biotite 11B/10B isotopic fractionation and the boron local
environment in the structure of volcanic glasses. Geochim. Cosmochim. Acta, 67:1863–1873.
Tonarini S., Leeman W. P., Leat P. T. 2011. Subduction erosion of forearc mantle wedge implicated in the
genesis of the South Sandwich Island (SSI) arc: Evidence from boron isotope systematics. Earth Planet. Sci.
Lett., 301:275-284.
Tornos F., Wiedenbeck M., Velasco F. 2012. The boron isotope geochemistry of tourmaline-rich alteration in the
IOCG systems of northern Chile: implications for a magmatic-hydrothermal origin. Miner. Depos., 47:483-
499.
Trumbull R. B. & Chaussidon M. 1999. Chemical and boron isotopic composition of magmatic and
hydrothermal tourmalines from the Sinceni granite-pegmatite system in Swaziland. Chem. Geol., 153:125-
137.
Trumbull R. B., Krienitz M. S., Gottesmann B., Wiedenbeck M. 2008. Chemical and boron-isotope variations in
tourmalines from an S-type granite and its source rocks: the Erongo granite and tourmalinites in the Damara
Belt, Namibia. Contrib. Mineral. Petrol., 155:1-18.
Trumbull R. B., Krienitz M. S., Grundmann G., Wiedenbeck M. 2009. Tourmaline geochemistry and δ11B
variations as a guide to fluid-rock interaction in the Habachtal emerald deposit, Tauern Window, Austria.
Contrib. Mineral. Petrol., 157:411-427.
Trumbull R. B., Beurlen H., Wiedenbeck M., Soares D. R. 2013. The diversity of B-isotope variations in
tourmaline from rare-element pegmatites in the Borborema Province of Brazil. Chem. Geol., 352:47-62.
Turner S., Rushmer T., Reagan M., Moyen J. F. 2014. Heading down early on? Start of subduction on Earth.
Geology, 42(1):139–142.
Valley J. W. 2003. Oxygen isotopes in zircon. In: J. M. Hanchar & P. W. O. Hoskin (eds.) Zircon. Reviews in
Mineralogy and Geochemistry, Mineralogical Society of America, 343–385.
Valley J. W., Chiarenzelli J. R., McLelland J. M. 1994. Oxygen isotope geochemistry of zircon. Earth Planet.
Sci. Lett., 126:187–206.
Valley J. W., Kinny P. D., Schulze D. J., Spicuzza M. J. 1998. Zircon megacrysts from kimberlite: oxygen
isotope variability amongmantle melts. Contrib. Mineral. Petrol., 133:1–11.

188
Valley J. W., Lackey J. S., Cavosie A. J., Clechenko C. C., Spicuzza M. J., Basei M. A. S., Bindeman I. N.,
Ferreira V. P., Sial A. N., King E. M., Peck W. H., Sinha A. K., Wei C. S. 2005. 4.4 billion years of crustal
maturation: oxygen isotope ratios of magmatic zircon. Contrib. Mineral. Petrol., 150:561–580.
Van Achterbergh E., Ryan C. G., Jackson S. E., Griffin W. 2001. Data reduction software for LA-ICP-MS. In: P.
Sylvester (ed.) Laser Ablation ICPMS in the Earth Science. Mineralogical Association of Canada, 239–243.
van Hinsberg V. J., Henry D. J., Marschall H. R. 2011. Tourmaline: an ideal indicator of its host environment.
Can. Mineral., 49:1–16.
van Hunen J. & van den Berg A. P. 2008. Plate tectonics on the early Earth limitations imposed by strength and
buoyancy of subducted lithosphere. Lithos, 103:217-235.
van Hunen J. & Allen M. B. 2011. Continental collision and slab break-off: a comparison of 3-D numerical
models with observations. Earth Planet. Sci. Lett., 302(1–2):27–37.
Van Kranendonk M. J. 2007. A review of the evidence for putative Paleoarchean life in the Pilbara craton,
western Australia. In: M. J. Van Kranendonk, R. H. Smithies, V. C. Bennett (eds.) Earth’s oldest rocks.
Developments in Precambrian Geology, Amsterdam, Elsevier, 855-877.
Van Kranendonk M. J. 2011. Onset of plate tectonics. Science, 333:413-414.
Van Kranendonk M. J., Hickman A. H., Smithies R. H., Nelson D. N., Pike G. 2002. Geology and tectonic
evolution of the Archaean North Pilbara terrain, Pilbara Craton, Western Australia. Econ. Geol., 97:695–
732.
Van Kranendonk M. J., Collins W. J., Hickman A. H., Pawley M. J. 2004. Critical tests of vertical vs horizontal
tectonic models for the Archaean East Pilbara Granite-Greenstone Terrane, Pilbara Craton, Western
Australia. Precambrian Res., 131:173–211.
Van Kranendonk M. J., Smithies R. H., Hickman A. H., Champion D. C. 2007. Review: Secular tectonic
evolution of Archean continental crust: Interplay between horizontal and vertical processes in the formation
of the Pilbara Craton, Australia. Terra Nova, 19(1):1-38.
Van Kranendonk M. J., Kröner A., Hegner E., Connelly J. 2009. Age, lithology and structural evolution of the c.
3.53 Ga Theespruit Formation in the Tjakastad area, southwestern Barberton Greenstone Belt, South Africa,
with implications for Archaean tectonics. Chem. Geol., 261:114-138.
van Thienen P., van den Berg A. P., Vlaar N. J. 2004. On the formation of continental silicic melts in
thermochemical mantle convection models: implications for early Earth. Tectonophysics, 394:111-124.
Veizer J. & Jansen S. L. 1979. Basement and sedimentary recycling and continental evolution. J. Geol., 87:341–
370.
Veizer J. & Mackenzie F. T. 2003. Evolution of sedimentary rocks. In: H. Holland & K. Turekian (eds.) Treatise
on Geochemistry. Oxford, Elsevier-Pergammon, 369–407.
Viljoen M. J. & Viljoen R. P. 1969. An introduction to the geology of the Barberton granite–greenstone terrain.
Geological Society of South Africa, Special Publications, 2:9–28.
Villaça J. N. 1981. Alguns aspectos sedimentares da Formação Moeda. In: Soc. Bras. Geol., Núcleo MG, Anais,
2:92-137.
Villaros A., Buick I. S., Stevens G. 2012. Isotopic variations in S-type granites: an inheritance from a
heterogeneous source? Contrib. Mineral. Petrol., 163:243–257.
Voice P. J., Kowalewski M., Eriksson K. A. 2011. Quantifying the timing and rate of crustal evolution: Global
compilation of radiometrically dated detrital zircon grains. J. Geol., 119(2):109–126.
Wallace R. M. 1965. Geology and mineral resources of the Pico de Itabirito district. USGS Prof. Paper 341-F,
68p.
Wang Y., Zhang Y., Zhao G., Fan W., Xia X., Zhang F., Zhang A. 2009. Zircon U–Pb geochronological and
geochemical constraints on the petrogenesis of Taishan sanukitoids (Shandong): Implications for
Neoarchean subduction in the Eastern Block, North China Craton. Precambrian Res., 174:273–286.
Watkins J. M., Clemens J. D., Treloar P. J. 2007. Archaean TTGs as sources of younger granitic magmas:
melting of sodic metatonalites at 0.6–1.2 GPa. Contrib. Mineral. Petrol., 154:91–110.

189
Whalen J. B., Percival J. A., McNicoll V. J., Longstaffe F. J. 2004. Geochemical and isotopic (Nd–O) evidence
bearing on the origin of late- to post-orogenic high-K granitoid rocks in the western Superior Province:
implications for late-Archaean tectonomagmatic processes. Precambrian Res., 132:303–326.
Whitehouse M. J. & Fedo C. M. 2007. Searching for Earth‘s earliest life in southern west Greenland – History,
current status, and future prospects. In: M. J. Van Kranendonk, R. H. Smithies, V. Bennett (eds.) Earth's
oldest rocks. Developments in Precambrian Geology, Amsterdam, Elsevier, 841-853.
Whitney D. L. & Evans B. W. 2010. Abbreviations for names of rock-forming minerals. Am. Mineral., 95:185-
187.
Wiedenbeck M., Hanchar J. M., Peck W. H., Sylvester P., Valley J., Whitehouse M., Kronz A., Morishita Y.,
Nasdala L., Fiebig J., Franchi I., Girard J. P., Greenwood R. C., Hinton R., Kita N., Mason P. R. D., Norman
M., Ogasawara M., Piccoli P. M., Rhede D., Satoh H., Schulz-Dobrick B., Skar O., Spicuzza M. J., Terada
K., Tindle A., Togashi S., Vennemann T., Xie Q., Zheng Y. F. 2004. Further characterization of the 91500
zircon crystal. Geostand. Geoanal. Res., 28:9–39.
Wilde S. A., Valley J. W., Peck W. H., Graham C. M. 2001. Evidence from detrital zircons for the existence of
continental crust and oceans on the Earth 4.4 Gyr ago. Nature, 409:175--178.
Willbold M., Hegner E., Stracke A., Rocholl A. 2009. Continental geochemical signatures in dacites from
Iceland and implications for models of early Archaean crust formation. Earth Planet. Sci. Lett., 279:44–52.
Williams P. R. & Whitaker A. J. 1993. Gneiss domes and extensional deformation in the highly mineralised
Archaean Eastern Goldfields Province, Western Australia. Ore Geol. Rev., 8:141–162.
Williams L. B., Hervig R. L., Holloway J. R., Hutcheon I. 2001. Boron isotope geochemistry during diagenesis.
Part 1. Experimental determination of fractionation during illitization of smectite. Geochim. Cosmochim.
Acta, 65:1769–1782.
Windley B. F. & Garde A. A. 2009. Arc-generated blocks with crustal sections in the North Atlantic craton of
West Greenland: crustal growth in the Archean with modern analogues. Earth-Sci. Rev., 93:1–30.
Workman R. K. & Hart S. R. 2005. Major and trace element composition of the depleted MORB mantle
(DMM). Earth Planet. Sci. Lett., 231(1–2):53–72.
Wotslaw J. F., Bindeman I. N., Schaltegger U., Brooks C. K., Naslund H. R. 2012. High-resolution insights into
episodes of crystallization, hydrothermal alteration and remelting in the Skaergaard intrusive complex. Earth
Planet. Sci. Lett., 355-356:199–212.
Wunder B., Meixner A., Romer R. L., Wirth R., Heinrich W. 2005. The geochemical cycle of boron: constraints
from boron isotope partitioning experiments between mica and fluid. Lithos, 84:206–216.
Xavier R. P., Wiedenbeck M., Trumbull R. B., Dreher A. M., Monteiro L. V. S., Rhede D., Araújo C. E. G.,
Torresi I. 2008. Tourmaline B-isotopes fingerprint marine evaporates as the source of high-salinity ore fluids
in iron oxide copper-gold deposits, Carajás Mineral Province (Brazil). Geology, 36:743-746.
Yang J. H., Wu F. Y., Wilde S. A., Zhao G. 2008. Petrogenesis and geodynamics of Late-Archaean magmatism
in eastern Hebei, eastern North China Craton: geochronological, geochemical and Nd–Hf isotopic evidence.
Precambrian Res., 167:125–149.
Yearron L. M. 2003. Archaean granite petrogenesis and implications for the evolution of the Barberton
Mountain Land, South Africa. Kingston University, London, UK, PhD Thesis, 315p.
Zegers T. E. & van Keken P. E. 2001. Middle Archean continent formation by crustal delamination. Geology,
29(12):1083–1086.
Zegers T. E., Nelson D. R., Wijbrans J. R., White S. H. 2001. SHRIMP U–Pb zircon dating of Archean core
complex formation and pancratonic strike-slip deformation in the East Pilbara granite-greenstone terrain.
Tectonics, 20:883–908.
Zeh A., Gerdes A., Barton Jr. J. M. 2009. Archean accretion and crustal evolution of the Kalahari Craton – the
zircon age and Hf isotope record of granitic rocks from Barberton/Swaziland to the Francistown Arc. J. Pet.,
50:933–966.
Zeh A., Stern R. A., Gerdes A. 2014. The oldest zircons of Africa – their U-Pb-Hf-O isotope and trace element
systematics, and implications for Hadean to Archean crust-mantle evolution. Precambrian Res., 241:203–
230.

190
Zhao G., Sun M., Wilde S. A., Li S. 2004. A Paleo-Mesoproterozoic supercontinent: assembly, growth and
breakup. Earth Sci. Rev., 67(1):91-123.
Zucchetti M., Lobato L. M., Baltazar O. F. 2000. Volcanic and volcaniclastic features in Archean rocks and their
tectonic environment, Rio das Velhas Greenstone Belt, Quadrilátero Ferrífero, MG, Brazil. Rev. Bras.
Geocienc., 30:388-392.

191
192
 Appendices
Appendix A1 – Major and trace element composition of igneus rocks from the Quadrilátero
Ferrífero. (Electronic appendix)

193
Appendix A2 – Summary of U–Pb zircon ages for the basement of the Quadrilátero Ferrífero.
(Electronic appendix)

194
Appendix A3 – Summary of magmatic and metamorphic Rhyacian ages in the Quadrilátero
Ferrífero. (Electronic appendix)

195
Appendix A4 – Summary of U–Pb zircon ages for the Rio das Velhas greenstone belt. (Electronic
appendix)

196
Appendix A5 – Summary of detrital, magmatic and metamorphic ages for the Minas
Supergroup. (Electronic appendix)

197
Appendix A6 – Analytical techniques, description of rock samples, zircon grains and U-Pb ages.

1. Analytical techniques

1.1. Whole-rock geochemistry

Samples of approximately 10 kg were collected and ca. 5 kg of material was crushed and
analysed for major and trace elements at the Central Analytical Facility, University of Stellenbosch,
South Africa. Loss on ignition was calculated by placing powdered samples in an oven at 1000 °C for
1 hour. Major element compositions were analysed by X-ray fluorescence spectrometry (XRF) on
glass beads prepared with La-free flux. Internal standards were basalt BHVO-1 and granite NIM-G.
For the granite, calculated uncertainties (twice the measured deviations for the granite standard
expressed in wt.%) are: 0.35 for SiO2, 0.02 for TiO2, 0.12 for Al2O3, 0.15 for FeOtot, 0.01 for MnO,
0.02 for MgO, 0.07 for CaO, 0.10 for Na2O, 0.02 for K2O and 0.01 for P2O5. Results are plotted after
normalization to 100 wt.% volatile-free. Trace element compositions have been obtained from the
same fused beads used for major element determination by applying the method described by Eggins
(2003) and analysed using an Agilent 7500 ICP-MS coupled with a Nd-YAG 223 nm New Wave
LASER ablation (LA) system operating at a 12 Hz frequency with a mixed He-Ar carrier gas. Three
analyses (each comprising a 30 s blank followed by data collection for 60 s) on each whole rock fused
disc were obtained using a 100 μm diametre aperture, and the results averaged. After every three
samples (i.e. every 10th analysis), NIST612 (Pearce et al., 1997) glass bead was analysed as a
calibration standard, in addition to fused discs of NIM-G (granite) and BHVO-1 (basalt) that were
analysed as secondary standards. Data were collected in time-resolved mode and were reduced using
the SiO2 content measured by XRF as the internal standard. For each element, the reproducibility of
replicate analyses of the samples, and deviation from the certified values of the secondary standards
are less than 10 %, and mostly less than 5 % relative.

1.2. LA-ICP-MS U–Th–Pb dating

Zircon grains were separated from ca. 5 kg rock samples by means of standard crushing
techniques (Jaw crusher, Disc mill), manual panning and magnetic separation. Approximately 200
grains from each sample were hand-picked under a binocular microscope mounted in epoxy resins
ground down and polished to expose their interior. Prior to analytical work, all grains were
characterized by cathodoluminescence (CL) and back-scattered electron imaging using a scanning
electron microscope (JEOL JSM-6490) at Stellenbosch University. Zircon U–Pb isotope analyses were
undertaken at the Departamento de Geologia of the Universidade Federal de Ouro Preto (Brazil) in
two analytical sessions: the first one in June-July 2013, the second in May 2014. The analyses of the
first session were performed using an Agilent 7700 Q-ICP-MS coupled to a 213 nm New Wave laser
while those produced one year later were performed using a ThermoScientific Element 2 sector field
198
(SF) ICP-MS coupled to a CETAC LSX-213 G2+ laser. The LA-ICP-MS data were reduced using the
software Glitter (Van Achterbergh et al., 2001) with ages calculated and plotted on Concordia
diagrams using the IsoplotEx 4 program. Uncertainties given for individual analyses (ratios and ages)
are at the 1 sigma level. The results of the LA-ICP-MS analyses on samples and secondary standards
are presented in the Appendix A8.

1.2.1. First analytical session: June-July 2013

The analyses were performed using an Agilent 7700 Q-ICP-MS and a 213 nm New Wave
laser. Operating conditions were optimized to provide maximum sensitivity for the high masses ( 207Pb
and 238U) while inhibiting oxide formation (ThO+/Th+ < 1.0 %). Standard and unknown zircons were
ablated in a small volume (teardrop shaped) sample cell, with an insert holding a 25 mm-diametre
sample mount and a 7 mm-diametre standard mount. Acquisitions consisted of 20 s measurement of
the gas blank, followed by 40 s measurement of U, Th and Pb signals during ablation, and 30 s
washout (e.g. Buick et al., 2011). Samples, standards and sample holders were acid-washed before
being analysed to remove possible surface Pb contamination. Laser ablations were performed at 40 µm
spot size, ~ 6–8 J/cm2 fluence and 10 Hz repetition rate. Ablations occurred in a He carrier gas, and
the resulting aerosol was mixed with Ar prior to introduction into the ICP-MS via 4 mm Tygontubing
206 207
(pre-cleaned with 1 % ultra-pure nitric acid). Integration times were 15 ms for Pb; 40 ms for Pb
208 204 232 238
and 10 ms for Pb; (Pb + Hg); Th, U. The relevant isotopic ratios (207Pb/206Pb, 208
Pb/206Pb,
208
Pb/232Th, 206
Pb/238U and 207
Pb/235U, where 235
U was calculated from 238
U counts via the natural
235 238
abundance ratio U= U/137.88) have been calculated using the data reduction software Glitter
(Van Achterbergh et al., 2001). Individual isotopic ratios were displayed in time-resolved mode. For
our laser system, isotopic ratios generated during the first 5 s of each analysis were discarded. The
integration window for of each analysis was chosen to exclude signal segments that were related to
204
zones of Pb loss (e.g. fractures), high common Pb (as recognized by Pb + Hg counts markedly
204
higher than the high background caused by Hg contamination) or Pb inheritance. Those analyses
204
that contained high content of common Pb (higher intensity on mass (Hg + Pb) than the average
background value from the carrier gas) were eliminated. Instrumental drift was corrected against the
zircon standard using linear interpolative fits. Calibrations were based on six or more analyses of the
204
standard (6–8 analyses of unknowns bracketed between 2 and 3 analyses of standards). Pb-based
common Pb corrections were not applied. Standard zircon GJ-1 was used as a primary reference
material. In order to test the validity of the applied methods and the accuracy and external
reproducibility of the obtained age data, multiple analyses of Plešovice reference zircon (Sláma et al.,
2008) were performed. The Plešovice secondary standard gave concordia ages of 339.6 ± 0.8 Ma
(2SD; n = 50; MSWD = 1.4). The calculated age is consistent, within uncertainty, with the ID-TIMS
value reported by Sláma et al. (2008). The results for samples and secondary standards are presented
in the Appendix A8.

199
 1.2.2. Second analytical session: May 2014

Uranium, thorium and lead isotopes were analysed using a ThermoScientific Element 2 sector
field ICP-MS coupled to a CETAC LSX-213 G2+ laser system following the method described in
Gerdes & Zeh (2006, 2009) and Frei & Gerdes (2009). Data were acquired in time resolved, peak
jumping, pulse counting mode over 810 mass scans during 19 s background measurement followed by
30 s sample ablation. The laser was operated at a repetition rate of 10 Hz and a nominal energy output
of 30 %, corresponding to a laser energy of ∼ 0.025 mJ/pulse and a laser fluence of ∼ 3.5 J/cm2. Laser
spot-size was 20 µm with a typical penetration depth of ∼ 15–20 µm. Signal was tuned for maximum
254
sensitivity for Pb and U while keeping oxide production, monitored as UO/238U, well below 1 %.
With a depth penetration of ∼ 0.6 µm/s1 and a 0.9 s integration time (= 15 mass scans = 1 ratio) any
significant variation of the Pb/Pb and U/Pb in the µm scale is detectable. Raw data were corrected
offline for background signal, common Pb, laser-induced elemental fractionation, instrumental mass
discrimination, and time-dependent elemental fractionation of Pb/U using Glitter (Van Achterbergh et
al., 2001).

For about 20 % of the analyses, no common Pb correction has been applied as the
204
interference- and background-corrected Pb signal was below the detection limit of about 20 counts
per second (cps). This threshold is defined by the amount of Hg in the argon carrier gas and the
202 204
accuracy to which Hg and thus the interfering Hg can be monitored during analysis. The
204 204
interference of Hg on the mass 204 was estimated using a Hg/202Hg of 0.2299 and the measured
202
Hg. Common Pb correction, which was necessary for about 80 % of the analyses, was done using
204
the interference and background-corrected Pb signal in combination with a model Pb composition
(Stacey & Kramers, 1975). Laser-induced elemental fractionation and instrumental mass
discrimination were corrected by normalization to the reference zircon GJ-1 (Jackson et al., 2004) of
each analytical session. The total offset of the measured Pb/238U ratio from the ―true‖ ID TIMS
206

value of the analysed GJ-1 grain was typically around 3–15 %. Reported uncertainties (2ζ) of the
206
Pb/238U were propagated by quadratic addition of the external reproducibility (2SD) obtained from
the standard zircon GJ-1 during the analytical session and the within-run precision of each analysis
(2SE; standard error). In order to test the validity of the applied method and the reproducibility of the
obtained age data, multiple analyses of the Plešovice and M127 reference zircons (Sláma et al., 2008;
Klotzli et al., 2009) were performed during each analytical session. The Plešovice secondary standard
gave a concordia age of 337.0 ± 0.6 Ma (2SD; n = 98; MSWD = 0.96) while the M127 secondary
standard gave a concordia age of 523.3 ± 1.0 Ma (2SD; n = 63; MSWD = 0.70). The calculated ages
are consistent, within uncertainty, with the ID-TIMS values reported for Plešovice and M127 by
Sláma et al. (2008) and Klotzli et al. (2009), respectively. In addition, multiple analyses of the in-
house reference zircon BB were performed. The in-house BB reference material gave concordia age of
565.1 ± 1.8 Ma (2SD; n = 25; MSWD = 0.70). This is in agreement with the LA-ICP-MS long-term

200
average of BB (562 Ma; Santos et al., 2014) reference zircon at the Universidade Federal de Ouro
Preto. The results of the LA-SF-ICP-MS analyses for samples and secondary standards are presented
in the Appendix A8.

2. Rock sample description, zircon features and U-Pb zircon ages

2.1. Bação complex

 Sample FQ1 (UTM: 643228/7746959) is a medium- to coarse-grained, biotite-poor granite

intruded in the southern margin of the of the Bação dome. The granite is in tectonic contact
with the surrounding supracrustal rocks of the Minas Supergroup and preserves a weak
foliation that parallels with the SE-dipping bedding of the Moeda Formation. Zircons
extracted from this sample were yellow and purple to pale-brown. CL images shows well-
defined oscillatory zoning and no recognizable core–rim relationship. Twenty-two analyses on
both centres and rims plot along a regression line that intercepts the upper Concordia at 2732
± 28 Ma. Six concordant (discordancy < 1 %) spot analyses yield Concordia age of 2711 ± 3
Ma (MSWD = 0.79), which is considered the age of crystallization of the rock. This sample
corresponds to sample MR1 of Romano et al. (2013), yielding a SHRIMP age of 2716 ± 5 Ma.
The two ages are within error identical.

 Sample FQ2 (UTM: 644685/7749318) is from a fine-grained banded gneiss characterized by

the alternation of cm-wide leucocratic and mesocratic bands. The mesocratic bands are rich in
plagioclase and biotite, whereas the leucocratic bands contain predominantly plagioclase,
quartz and minor alkali-feldspar. The gneiss hosts leucogranitic and tourmaline-bearing
pegmatitic dikes intruded subparallel to the gneiss vertical banding. The pegmatites are often
foliated, with biotite defining planar orientation parallel to the direction of foliation in the host
gneiss. Massive pegmatitic and aplitic stocks crosscut the gneiss. Zircons from this sample are
prismatic and stubby reaching up to 300 µm in length. In CL, the grains are either concentric
oscillatory zoned or characterized by a bright homogeneous centres surrounded by a zoned
overgrowth. A subset of dark and structureless grains also occurs. Most of the spot analyses
for both rims and centres gave highly discordant ages. Thirteen concordant spot analyses
(discordancy < 1 %) on both homogeneous centres and zoned crystals yield a weighted mean
age of 2868 ± 10 Ma (95 % c.l.; MSWD = 0.66). This age represents the magmatic age of the
sample. Five subconcordant spot analyses (discordancy < 5 %) on featureless grains yield an
average age of 2705 ± 18 Ma (95 % c.l.; MSWD = 0.17). This structureless grains have Th/U
that is lower (average 0.25) than the average ratio for zoned crystals and homogeneous cores
(Th/U = 0.48). We interpret 2705 Ma as the metamorphic age of the sample.

 Sample FQ3 (UTM: 643228/7746959) is from a massive quartz-rich coarse-grained

pegmatitic body crosscutting FQ2.

201
 Sample FQ4 (UTM: 638297/7749548) is from a fine-grained banded gneiss characterized by
the alternation of cm-wide leucocratic and mesocratic bands.

 Sample FQ5 (UTM: 625597/7762152) is a medium to coarse-grained grey granitoid collected

in the northern margin of the Bação dome. The granite is intruded into banded TTG gneiss and
records a weak lineation defined by aligned biotite and elongated plagioclase–quartz rods. It
mostly consists of biotite, plagioclase, quartz and microcline. This sample corresponds to
sample MR11 of Romano et al. (2013). Zircon grains from sample MR11 gave a
crystallization age of 2744 ± 10 Ma.

 Sample FQ6 (UTM: 625629/7762157) is the fine-grained banded gneiss intruded by FQ5 and
correspond to sample D12 described by Lana et al. (2013). Nine SHRIMP analyses performed
by these authors on the zoned cores yielded concordant to subconcordant results that defined a
Pb-Discordia with an upper intercept age of 2764 ± 10 Ma.

 Sample FQ7 (UTM: 626562/7759887) is a coarse-grained foliated biotite-bearing

leucogranitic stocks intruding the banded gneiss. The NNE-SSW steeply dipping foliation in
the stock (ca. 80°) is parallel to that in the host gneiss.

 Sample FQ8 (UTM: 628585/7760332) is from a fine- to medium-grained gneiss cut by two
systems of cm- to dm-thick leucogranitic dikes. The first system is steeply dipping (70-90°)
and subparallel to the foliation in the host granite (striking N45 to EW). The second cut the
first one dipping ca. 45° south. Metre-thick tourmaline-bearing pegmatitic dikes intruded the
gneiss cutting both dike systems.

 Sample FQ10 (UTM: 637825/7750333) was collected from a quarry in the south-eastern
sector of the dome, a few kilometres from the town of Cachoeira do Campo. It represents a
banded trondhjemitic gneiss, characterized by vertical, mm- to cm-wide leucocratic and
mesocratic bands. The mesocratic bands are rich in plagioclase and biotite, whereas the
leucocratic bands contain predominantly plagioclase, quartz and minor microcline. The gneiss
is locally crosscut by subvertical leucogranitic sheets intruded subparallel to the gneiss vertical
banding. This sample corresponds to sample D04 of Lana et al. (2013). U-Pb dating on zircon
207
from D04 yielded a Pb/206Pb weighted mean age of 2795 ± 7 Ma that is interpreted as the
crystallization age of the sample.

 Sample FQ11 and FQ12 (UTM: 628157/7751740) were collected from a waterfall. FQ11 is a
weakly-foliated biotite-poor granite exhibiting a porphyric texture defined by the occurrence
of large K-feldspar crystals, reaching up to 6 cm in length, hosted in a medium-grained
groundmass. The large K-feldspars host biotite inclusions and are euhedral to subhedral
sometimes showing partially resorbed rims. The abundance of K-feldspar is variable but a
maximum of ca. 20 crystals/m2 is observed. FQ12 is a fine- to medium-grained banded gneiss
202
 showing the alternation of subvertical, mm- to cm-wide leucocratic sheets and grey bands
containing a small amount of K-feldspar reaching up to 2 cm in length. Bands of the medium-
grained unfoliated K-feldspar rich granite locally crosscut the leucocratic sheets. Moreover,
irregular domains of the massive porphyritic granitoid occasionally intrude the gneiss.

Zircon from sample FQ11 is mostly prismatic, ranging in size from 350 to 70 µm and with
maximum length: width ratios up to c. 5:1. Most of the grains in CL exhibit fine-scale
oscillatory zoning while a subset of crystals is dark and homogeneous. A few grains showed
complex, structureless cores surrounded by oscillatory rims. Three spot analyses on cores
207
were concordant (discordancy < 1 %) yielding a Pb/206Pb age of 2905 ± 17 Ma (95 % c.l.;
MSWD = 0.72). Six analyses on the zoned grains gave a well-defined concordia age of 2790 ±
3 Ma (MSWD = 0.35) which is interpreted as the magmatic age of the rock. Finally, seven
concordant and subconcordant analyses on featureless grain and outer rims in zoned crystals
gave a weighted 207Pb/206Pb age of 2719 ± 14 Ma (MSWD = 0.22).

 Sample FQ13 (UTM: 628731/7748442) is from a spotted medium-grained gneiss mostly

composed by plagioclase, quartz, biotite and minor K-feldspar and epidote. The occurrence of
dark biotite and epidote rich cores surrounded by quartz-feldspar mantles (flecks hereafter)
gives a spotted appearance to the rock. The flecks are prolate spheroid reaching 5 mm across
and with a core-to-mantle relative abundance varying from 3:1 to 1:5. Although most of the
flecks are isolated in the rock, occasionally up to three flecks may merge forming complex
structures. Opaque minerals rather than biotite-dominated aggregates constitute some of the
cores in the flecks. Leucogranitic dikes intruded the gneiss and are cut by pegmatitic dikes up
to 60 cm-thick. Zircons from FQ13 are prismatic to stubby reaching 250 µm in length.
Cathodoluminescence images reveal internal fine-scale oscillatory zoning. Few crystals are
homogeneous or weakly zoned. Twenty-eight spot analyses gave concordant to discordant
207
ages that plot along a Pb-loss line with upper intercept Pb/206Pb age of 2786 ± 4 Ma
(MSWD = 3.5). Nineteen concordant analyses (< 1 % discordant) yield a weighted age of
2788 ± 7 Ma (MSWD = 0.42). Seventeen spot analyses yield a concordant age of 2790 ± 3 Ma
(MSWD = 1.7), which is considered the magmatic age of the rock. Five concordant and
subconcordant analyses gave weighted mean apparent age of 2701 ± 15 Ma (MSWD = 0.27).
These spot analyses are not clearly associated with homogeneous grains or crystal rims and
therefore the interpretation of this younger age is controversial. It can either be an artefact due
207
to Pb-loss or dating the metamorphic event. Two concordant cores gave a Pb/206Pb age of
2900 Ma and are interpreted as inherited.

 Sample FQ16 and FQ17 (UTM: 625643/7755180) are from a quarry exposing a fine-grained
biotite-bearing banded gneiss crosscut by numerous leucogranite sheets that vary in width
from several millimetres to a few metres. On the upper part of the quarry wall-pit, the
203
 leucogranitic sheets made up more than 30 % of the rock volume forming disharmonic folds
and generating complex, chaotic structures. Sample FQ16 is from a leucogranitic sheet while
sample FQ17 represents the hosting gneiss. Zircon from FQ17 ranges in size from 350 to 75
µm is stubby to prismatic with maximum elongation of 5:1. CL images exposed well-defined
oscillatory zoning and clear core–overgrowth relationships. The centres display fine-scale
oscillatory zoning while the overgrowths are homogeneous or weakly zoned. Analyses of rims
gave concordant to highly discordant ages plotting along a Pb-loss line with upper intercept
age of 2744 ± 18 Ma. This age is within error identical to a weighted mean age of 2732 ± 10
Ma (95 % c.l.; MSWD = 0.45) obtained from fourteen concordant spot analyses. This age
representing the metamorphic age of the rock. Eight concordant analyses (< 1 % discordant)
yield Concordia age of 2778 ± 2 Ma (MSWD = 1.13), which is interpreted as the
crystallization age of the sample. Finally four concordant core analyses gave a weighted
207
average apparent Pb/206Pb age of 2862 ± 2 Ma. These are considered inherited cores. Lana
et al. (2013) have dated a leucogranitic sheet from this quarry (sample SG-2) obtaining a Pb-
loss regression line intercepting the Concordia at 609 ± 130 Ma and 2730 ± 7 Ma. Lana et al.
(2013) interpreted the upper intercept as the crystallization age of the leucogranites.

 Sample FQ18 (UTM: 624437/7754958) is a fine-grained banded gneiss. Leucogranitic sheets

oriented parallel or subparallel to the banding intrude the gneiss. The leucogranitic sheets and
the main foliation are folded forming tight to isoclinal folds and cut by a second generation of
leucogranitic bodies. The folds are commonly restricted to single leucogranitic sheets not
affecting neighbouring sheets of identical appearance. The leucogranites contain small garnet
crystals.

 Samples FQ19, FQ20 and FQ21 (UTM: 620091/7749557) were collected in a quarry in the
south-western portion of the Bação dome. The quarry exposes strongly banded, fine- to
medium-grained gneiss recording a vertical compositional banding characterized by the
alternation between light-coloured plagioclase-dominated and biotite-rich bands. Although the
bands generally exhibit thickness from few millimetres to up to 2 centimetres, sporadically 50
cm-thick biotite-rich bands also occur. Numerous leucogranite sheets intrude both biotite and
plagioclase-rich bands. Sample FQ19 was extracted from a 50 cm wide medium-grained
biotite-rich band, while sample FQ20 from the banded gneiss. Sample FQ21 is from a
leucogranitic sheet. Samples FQ20 and FQ21 correspond to samples D07A and D07B of Lana
et al. (2013). Zircon grains from sample D07A gave a crystallization age of 2918 ± 10 Ma and
a poorly constrained metamorphic age of 2775 Ma. Analyses of oscillatory rims from the
207
zircon grains from D07B gave a SHRIMP weighted Pb/206Pb mean age of 2774 ± 11 Ma
interpreted by Lana et al. (2013) as the crystallization age of the leucogranites.

204
 Sample FQ22 (UTM: 622543/7747051) comes from a trondhjemitic-banded gneiss cropping
out in the south-western periphery of the dome. This rock records a strong vertical gneissosity
defined by oriented biotite and cm-wide leucocratic bands. Undeformed leucogranite sheets
intrude the gneiss with direction parallel to the banding. This sample was collected in the same
locality of sample MP-1 of Lana et al. (2013). LA-ICP-MS analyses performed by Lana et al.
(2013) on zircon cores and rims gave ages of 2903 ± 12 Ma and 2770 ± 29 Ma, which are
interpreted by these authors as the crystallization and metamorphic ages of the rock.

 Sample FQ23 (UTM: 644680/7749530) was collected a few hundreds of metres from sample
FQ2. This rock displays less pronounced banding representing a more homogeneous domain
of the gneiss. Zircons from this sample are stubby to prismatic, reaching up to 300 µm in
length. CL images show bright rounded and partially resorbed cores overgrown by dark
weakly zoned rims as well as crystals displaying concentric oscillatory zoning. The cores are
either homogeneous or zoned with few grains presenting convolute zoning. Thirteen analyses
of cores are co-linear and define a chord with upper Concordia intercepts at 2893 ± 18 Ma.
This age is identical to a weighted 207Pb/206Pb mean age of 2898 ± 12 Ma (95 % c.l.; MSWD =
0.61) obtained from eight concordant (discordancy < 1 %) analyses from seven cores. We
interpret these cores as inherited. Most of the rims are strongly discordant. Four concordant
grains (discordancy < 1 %) yield a weighted mean age of 2783 ± 18 Ma (95 % c.l.; MSWD =
0.47) which is interpreted as the magmatic age of the gneiss.

 Sample FQ26 (UTM: 644963/7748500) is an aplitic stock cropping out as a 15-metres high
wall following the railway for approximately 100 m. The aplite exhibits a characteristic
spotted appearance given by the widespread occurrence of tourmaline spherical clots reaching
up to 1 cm across.

 Sample OPU4094 (UTM: 636749/7748459) comes from a one metre-thick pegmatite dike.
Zircon separated from this sample ranges in size from 75 to 500 µm, with most of them
ranging between 150 and 250 µm. The crystals are prismatic and in CL they display large dark
or white non-zoned centres surrounded by thin zoned or dark homogeneous rims. An upper
concordia intercept age of 2720 ± 46 Ma is obtained considering all the analyses. Eight
concordant to subconcordant analyses (< 5 % discordant) on non-zoned centres yield an age of
2693 ± 13 Ma (95 % c.l.; MSWD = 0.27) which is considered the crystallization age of the
pegmatite dike.

2.2. Bonfim complex

 Sample FQ28 (UTM: 602350/7768933) is a weakly-foliated medium-grained granitoid

composed by plagioclase, quartz, microcline and aligned biotite clots. This sample was
collected in the same locality of sample MR22A of Romano et al. (2013). Analyses on both

205
 oscillatory zoning and structureless grains plot along Pb-loss chord with an upper intercept
age of 2727 ± 6 Ma. This age, which is identical to a Concordia age of 2730 ± 8 Ma defined
by few concordant points, has been interpreted by Romano et al. (2013) as the crystallization
age of the rock.

 Sample FQ29 (UTM: 603409/7765331) is a medium-grained grey granitoid rock composed of

plagioclase, quartz, microcline, biotite and epidote. The rock exhibits penetrative foliation
mostly defined by the orientation of biotite and stretching lineation defined by the alignment
of feldspar rods. Fine-grained leucogranitic dikes crosscut the granitoid. Sample FQ30 is from
a 20 cm-thick leucogranitic dike. Zircon separated from FQ29 ranges in size from 300 to 100
µm, is stubby to prismatic with length: width ratios up to c. 5:1. Cathodoluminescence images
reveal internal fine-scale oscillatory zoning. A few grains show structureless either CL-dark or
CL-bright centres surrounded by oscillatory rims. Analyses of both rims and centres gave
concordant to highly discordant ages plotting along a Pb-loss line with upper intercept age of
2785 ± 10 Ma (MSWD = 4.3). Ten concordant analyses (< 1 % discordant) yield Concordia
age of 2773 ± 2 Ma (MSWD = 1.3). This age, interpreted as the crystallization age of the
sample, matches with a TIMS age of 2778 ± 3 Ma determined by Machado & Carneiro
(1992).

 Sample FQ32 (UTM: 608714/7756369) is a weakly foliated, very coarse-grained granitoid

mostly consisting of plagioclase, microcline, quartz, and biotite clots. The clots range in size
from a few millimetres to several centimetres wide. This sample was collected in the same
locality of sample MR14A dated by Romano et al. (2013) at 2730 ± 7 Ma.

 Sample FQ36 (UTM: 603293/7759954) is a medium-grained biotite-bearing foliated granitoid

crosscut by numerous foliation-parallel leucogranitic dikes.

 Sample FQ37 (UTM: 603293/7759954) is a fine-grained foliated amphibolitic dike striking

ca. N30. No major and trace element data are presented for this rock. Zircon separated from
the amphibolite dike ranges in size from 250 to 100 µm, is stubby to prismatic with length:
width ratios up to c. 4:1. Cathodoluminescence images reveal the common occurrence of
subrounded bright cores surrounded by either homogeneous dark or oscillatory zoned rims.
Some cores contain internal zoning truncated by the zonation of the oscillatory rims. LA-ICP-
MS analyses of the cores gave concordant to highly discordant ages that plot along a Pb-loss
line with upper intercept age of 2879 ± 17 Ma. Six concordant analyses (< 1 % discordant)
yield a weighted age of 2879 ± 14 Ma (MSWD = 0.68). Nine analyses of the rims passed the <
5 % discordancy filter test giving a weighted mean 207Pb/206Pb age of 2719 ± 14 Ma (95 % c.l.;
MSWD = 0.62). The amphibolite dike intrudes the medium-grained biotite-bearing granite

206
 FQ36. We interpret the crystallization age of the amphibolite to be 2719 ± 14 Ma. The 2879
Ma old inherited cores are interpreted as resorbed crystal assimilated by the dike.

 Sample FQ38 (UTM: 588137/7770374) is a banded gneiss exposed along a road cut in the
northern sector of the complex. The gneiss records strong subvertical foliation and is intruded
by multiple foliation-parallel leucocratic sheets. The gneiss banding is defined by the
alternation between plagioclase- and quartz-rich layers displaying granoblastic texture and
biotite-rich layers. This sample was collected in the same locality of sample D11 of Lana et al.
(2013). Zircon grains analysed from sample D11 gave a crystallization age of 2895 ± 13 Ma
and a metamorphic age of 2749 ± 10 Ma.

 Samples FQ39 and FQ40 (UTM: 593263/7767241) are from a biotite-bearing banded gneiss
crosscut by numerous foliation-parallel leucogranite sheets. Zircon separated from FQ40 is
prismatic with length: width ratios up to c. 4:1 and ranges in size from 300 to 75 µm. In CL,
the crystals show fine to coarse oscillatory zoning with a lesser population of unzoned black
grains. Eleven concordant analyses (< 1 % discordant) yield a weighted mean of 2670 ± 15
Ma (95 % c.l.; MSWD = 1.3) which is within error identical to an upper intercept age of 2690
± 21 Ma obtained from seventeen discordant analyses plus the eleven concordant zircon
analyses. Four subconcordant analyses from the core of two crystals gave an upper intercept
age of 2854 ± 18 Ma. This age is identical within error to the magmatic age obtained for the
gneiss FQ41 cropping out close by. We suggest that this rock crystallized at ca. 2850 Ma and
experienced a metamorphic event at 2670 ± 15 Ma.

 Samples FQ41 (UTM: 595299/7765480) is a biotite-bearing banded gneiss crosscut by

numerous foliation-parallel leucogranite sheets. The rock displays a regular alternation of
leucocratic and biotite-rich bands having thickness generally variable between 0.5 to 5
centimetres. The gneissosity has overall direction N60 80 SE. Zircon separated from FQ41
range in size from 350 to 100 µm, are equant to prismatic with length: width ratios up to c. 4:1
and show internally consistent oscillatory zoning with no obvious core structures. LA-ICP-MS
analyses of the grains gave subconcordant to discordant ages defining a Concordia with an
upper intercept age of 2849 ± 22 Ma. Seven analyses passed the < 5 % discordancy filter test
207
giving a weighted mean Pb/206Pb age of 2840 ± 26 Ma (95 % c.l.; MSWD = 2.1). Two of
these analyses have significantly younger age (ca. 2800 Ma) than the others; if these analyses
are not considered the weighted mean of the remaining five analyses gave an age of 2852 ± 16
Ma (95 % c.l.; MSWD = 0.24) which is interpreted as the crystallization age of the sample.

 Samples FQ44, FQ45 and FQ46 (UTM: 591726/7762759) were collected from the riverbeds
and banks of the Paraopeba river. The outcrop exposes a biotite-bearing banded gneiss
crosscut by numerous foliation-parallel fine- to medium-grained leucogranite sheets that are

207
 cut by both pegmatitic dike and non-foliated medium-grained granitoid domains. The hosting
gneiss contains sporadic cm-scale crystals of K-feldspar. Sample FQ44 is from a one-metre
thick medium-grained leucogranitic sheet, FQ45 and FQ46 are from the banded gneiss.
Zircons from FQ44 are prismatic to stubby reaching 250 µm in length. CL images show bright
rounded cores rimmed by dark overgrowths. The cores are either white and unzoned or
presenting fine oscillatory zoning commonly truncated by the rim. A large number of spot
analyses yielded highly discordant U-Pb ratios. Four concordant and subconcordant rims
yielded a weighted 207Pb/206Pb mean age of 2679 ± 37 Ma (95 % c.l.; MSWD = 0.47) which is
interpreted to be the age of zircon crystallization in the leucogranitic sheet. Spot analyses on
207
five concordant and subconcordant cores gave a weighted Pb/206Pb mean age of 2782 ± 17
Ma (95 % c.l.; MSWD = 0.47). We interpret these cores as inherited, most likely from the host
gneiss.

 Samples FQ47 and FQ48 (UTM: 585221/7768994) were collected from an inactive quarry.
Sample FQ48 is a biotite-bearing banded gneiss crosscut by numerous foliation-parallel
leucogranite sheets. Gently folded equigranular fine-grained leucogranites and pegmatitic
dikes cut the foliation of the hosting gneiss as well as leucogranitic sheets. Sample FQ47 is
from a pegmatitic dike.

 Sample FQ49 (UTM: 579263/7769220) is collected from an inactive quarry 5 km southwest

of the city of Brumadinho. The rock is a grey, medium-grained granitoid with foliation
defined by the orientation of muscovite and biotite.

 Sample FQ50 (UTM: 608712/7738785) is a foliated, fine-grained dark grey granitoid intruded
in the eastern part of the Mamona batholith. The rock consists of recrystallized quartz,
plagioclase, microcline and aligned cm-wide clots of biotite. Crosscutting pegmatites are also
present. This rock was collected in the same location of sample MR87A of Romano et al.
(2013). Spot analyses on eleven prismatic zircon grains performed by Romano et al. (2013)
207
were concordant yielding a Pb/206Pb Concordia age of 2613 ± 6 Ma (MSWD = 0.4). This
age is interpreted as the crystallization age of the rock.

 Sample FQ51 (UTM: 603378/7741084) is a foliated light grey fine- to medium-grained

granitoid consisting of quartz, plagioclase, K-feldspar and biotite. Numerous pegmatitic dikes
cut the granitoid. Zircons from FQ51 are prismatic to stubby with round terminations, ranging
in length from 100 to 300 µm. In CL, the crystals show fine to coarse oscillatory zoning.
Bright centres rimmed by structureless dark rims characterize many grains. LA-ICP-MS
analyses on both centres and rims passing the 5 % discordancy filter gave apparent 207Pb/206Pb
ages between 2588 and 2700 Ma plotting along a Pb-loss Discordia with an upper intercept

208
 age of 2689 ± 12 Ma. Fourteen concordant spot analyses yield a weighted average age of 2678
± 10 Ma (95 % c.l.; MSWD = 1.0) which is considered the crystallization age of the sample.

 Sample FQ52 (UTM: 582109/7761295) is from a banded trondhjemitic gneiss, characterized

by the alternation of vertical, cm-wide leucocratic and mesocratic bands. The mesocratic
bands are rich in biotite and plagioclase whereas the leucocratic bands contain predominantly
plagioclase, quartz and minor microcline. Zircon from FQ52 are prismatic to stubby with
round terminations, ranging in length from 100 to 300 µm. CL images reveal grains showing
either structureless or oscillatory zoning centres surrounded by dark rims. A few grains show
structureless CL-dark or CL-bright centres surrounded by oscillatory rims. Sixteen LA-ICP-
MS analyses on both centres and rims passing the 5% discordancy filter gave apparent
207
Pb/206Pb ages between 2678 and 2749 Ma, resulting in a weighted mean age of 2727 ± 11
Ma (95 % c.l.; MSWD = 0.8). This age is interpreted as the metamorphic age of the gneiss as
field and geochronological data indicate that the Bonfim complex was affected by regional
metamorphism during the Rio das Velhas II period (2760-2800).

 Samples FQ53 and FQ54 (UTM: 580353/7757719) were collected from a road-cut. Sample
FQ53, representing the main lithology in the outcrop, is a fine-grained banded gneiss
characterized by the alternation of mm- to cm-wide melanocratic and leucocratic bands.
Metres-wide irregular domains of a medium- to coarse-grained massive granite intrude the
hosting gneiss; sample FQ54 is from one of these domains.

 Sample FQ55 (UTM: 605269/7754213) is a medium-grained slightly foliated biotite-bearing

granite. The rock has a porphyritic appearance given by the occurrence of large K-feldspar
crystals reaching up to 6 cm in length. The abundance of large K-feldspar in the rock is of
about 5-10 crystals/m2. The granite host small (< 5 cm across) stretched fine-grained
xenoliths.

 Sample FQ56 (UTM: 606037/7754076) is a medium-grained slightly foliated biotite-bearing

granite. As for FQ55, the rock has a porphyritic appearance given by the occurrence of large
K-feldspar crystals reaching up to 10 cm in length. The granite host fine-grained stretched
mafic enclaves.

 Sample FQ57 (UTM: 607230/7753663) is a fine- to medium-grained biotite-bearing granite.

Large K-feldspar crystals in this rock also occur but are less abundant and smaller (up to 4 cm
in length) than in FQ55 and FQ56. Pegmatitic dikes up to 60 cm wide intruded the granite.
The dikes are cut by a system of quartz-tourmaline veins exhibiting maximum width of ca. 5
cm and direction of N100-120.

2.3. Belo Horizonte complex

209
 Sample FQ60 (UTM: 548149/7801529) is a medium-grained, slightly foliated biotite-bearing
granite cut by leucogranitic dikes up to 40 cm thick. Many subvertical quartz veins with
thickness variable from 1 to 20 cm strike ca. N30 cutting both the granite and the
leucogranites. Zircon from FQ60 ranges in size from 350 to 75 µm is stubby to prismatic with
length: width ratios up to c. 5:1. CL images reveal that most of the grains are dark and
structureless (ca. 70 %). Other grains display light grey sometimes-zoned centres surrounded
by dark rims. A large number of spot analyses yielded highly discordant U-Pb ratios. Six
concordant and subconcordant spot analyses from four grains yielded a weighted mean age of
2728 ± 16 Ma (95 % c.l.; MSWD = 2.3). This age is interpreted as the crystallization age of
the sample.

 Sample FQ62 (UTM: 550252/7808324) was collected from a medium-grained, slightly

foliated granite locally characterized by cm-wide microcline crystals in a matrix of
plagioclase, biotite and quartz. This rock was sampled in the same location of sample
MR234A of Romano et al. (2013). U-Pb dating of zircon grains from sample MR234A
yielded a crystallization age of 2755 ± 8 Ma.

 Sample FQ63 (UTM: 538231/7828551) was collected from a medium-grained, slightly

foliated biotite-bearing granite at the contact with the metasedimentary rocks of the
supracrustal sequence. Leucogranitic dikes and quartz veins cut the granite.

 Samples FQ65 and FQ66 (UTM: 537885/7835546) were collected from a large pavement in
the central part of the Pequi batholith. In the outcrop, two co-magmatic slightly foliated
granites alternate forming a layered rock. The layering is made of the alternation between a
medium-grained biotite-bearing granite (FQ65) and a coarse-grained biotite-poor granite
(FQ66). Pegmatitic dikes intrude the granites. Zircons separated from FQ65 are prismatic
reaching up to 300 µm in length. CL images reveal mostly fractured structureless either dark
or light grey grains. A subset of grains show oscillatory zoning centres surrounded by
featureless dark rims. Several attempts to obtain ages for the cores have failed due to the high
level of discordance. Nine concordant (< 1 % discordant) spot analyses from six featureless
crystals gave a Concordia age of 2644 ± 4 Ma (95 % c.l.; MSWD = 2.3). This age is within
error identical to a weighted mean age obtained considering twenty-two analyses (performed
on eight structureless grains) passing the < 5 % discordancy filter test. This weighted mean
age of 2645 ± 8 Ma (MSWD = 0.37) is tentatively interpreted as the metamorphic age for the
rock. Romano et al. (2013) presented zircon U-Pb data from one of the pegmatitic dikes
cutting the granite. Twelve spot analyses on the oscillatory overgrowths of the zircon grains
yielded a Concordia age of 2706 ± 7 Ma (MSWD = 0.66), which they interpreted as the
crystallization age of the dikes.

210
 Samples FQ67 and FQ68 (UTM: 550376/7807134 and 548266/7798859, respectively) are
medium-grained, slightly- to non-foliated biotite-bearing granites cut by rare leucogranitic
and/or pegmatitic bodies.

 Sample FQ70 (UTM: 579027/7790240) is a fine-grained gneiss characterized by the

alternation of mm-scale leucocratic and mesocratic bands and by leucogranitic sheets intruded
parallel to the gneissosity. Locally, the bands and the sheets are folded forming tight to
isoclinal folds. Tourmaline- and muscovite-bearing pegmatitic dikes as well as medium-
grained leucogranitic dikes crosscut the gneissosity of the rock. Zircons separated from FQ70
are prismatic reaching up to 300 µm in length. CL images reveal mostly featureless dark
grains with just a few crystals having dark rims surrounding light grey centres. A large
number of spot analyses yielded highly discordant U-Pb ratios. Twenty analyses plot along a
Pb-loss line with upper intercept age of 2715 ± 6 Ma. This age is identical to a Concordia age
obtained from seven concordant analyses (< 1 % discordant) yielding 2713 ± 3 Ma (MSWD =
0.93) which is considered the metamorphic age of the sample.

 Sample FQ71 (UTM: 580927/7790580) was collected 2 km to the east of FQ70. FQ71 is a
medium-grained biotite-bearing 70 cm-thick leucogranitic dike crosscutting the gneissosity of
a banded gneiss similar to sample FQ70.

 Sample FQ74 (UTM: 555202/7842055) is a medium-grained biotite-bearing reddish gneiss

locally displaying microcline crystals of up to 8 cm in size. Zircons separated from FQ74 are
blocky to prismatic ranging in size from 75 to 400 µm. In CL most of the crystal are dark and
structureless with only a subset of the grains (< 10 %) displaying ill-defined oscillatory
zoning. About 20 % of the zircon crystals contain inclusions of U-Th rich minerals whose
abundance is variable from crystals to crystals to up to 60 % of the hosting zircon. Fifteen
analyses plot along a Pb-loss line with upper intercept age of 2647 ± 8 Ma. This age is within
error identical to a weighted age of 2638 ± 14 Ma (95 % c.l.; MSWD = 0.58) calculated
considering the nine analyses passing the < 5 % discordancy filter test. This age can represent
either the age of the interaction between the granite and late-magmatic fluids or the timing of
the latest high-grade metamorphic event in the complex.

 Sample FQ77 (UTM: 557685/7840864) is a reddish biotite-quartz- feldspar augen gneiss. The
rock is characterized by thin (< 1 cm) fine-grained biotite-rich melanocratic layers embedding
augen. The latter could be either made of a granular mosaic of coarse-grained K-feldspar +
quartz or single cm-sized crystals of microcline. A fine- to medium-grained massive grey
granite intrudes the gneiss. Late quartz-alkali feldspar veins crosscut the two rocks.

 Sample FQ78 (UTM: 565401/7841374) was collected from a road-cut. This rock is a fine-
grained banded gneiss crosscut by foliation-parallel fine-grained leucogranite sheets that are
211
 cut by pegmatitic dikes. The gneissosity displays marked dip variations; i.e. from ca. 70° to
subhorizontal. Locally, weakly-foliated irregular domains interrupt the banding. The gneiss
contains low volume abundance (ca. 1 vol.%) of randomly distributed euhedral to subeuhedral
crystals of alkali feldspar reaching up to 3 cm in length.

 Sample FQ80 (UTM: 616041/7803192) is a fine-grained gneiss characterized by the

alternation of mm-scale leucocratic and mesocratic bands and by leucogranitic sheets intruded
parallel to the gneissosity. The gneissosity strikes NS-N30 with dipping varying from
subvertical to ca. 45°. As for sample FQ78, locally less deformed domains interrupt the
banding.

 Sample FQ82 (UTM: 605268/7825091) was collected from a road-cut. This rock is a biotite-
bearing fine-grained gneiss characterized by the alternation of mm-scale leucocratic and
mesocratic bands and by leucogranitic sheets intruded parallel to the gneissosity.

212
Appendix A7 – Major and trace element composition for samples of the Bação, Bonfim and Belo
Horizonte complexes. (Electronic appendix)

213
Appendix A8 – U-Pb isotope results for samples and standards for both analytical sessions.
(Electronic appendix)

214
Appendix A9 – Analytical techniques, sample and zircon description, detailed isotopic results for
each sample and for zircon standards, Concordia diagrams and zircon CL images.

1. Analytical techniques

Zircon crystals were separated from ca. 5 kg of rock sample using standard crushing
techniques, manual panning and magnetic separation. The grains were subsequently hand-picked
under a binocular microscope (approximately 200 grains per sample), mounted in epoxy resin and
ground down to expose their centres. Prior to analytical work, the zircons were imaged by Scanning
Electron Microscopy (SEM) using a JEOL 6510 equipped with a Centaurus CL detector at the
Departamento de Geologia of the Universidade Federal de Ouro Preto (UFOP) (Brazil), to obtain
information on their internal structures.

1.1. LA-ICP-MS U-Pb dating

A detailed description of the data acquisition procedure used for both U-Pb analytical sessions
was outlined in more details by Farina et al. (2015a).

Standards and unknown zircons were analysed using a teardrop-shaped, low-volume laser cell
during time-resolved data acquisition. Details on data acquisition for both analytical sessions are given
in Table 1. The raw data were corrected offline for background signal, common Pb (see below), laser-
induced elemental fractionation, instrument mass discrimination and time-dependant elemental
fractionation using Glitter (Van Achterbergh et al., 2001). Careful monitoring of the time-resolved
signal was observed to exclude domains with distinct Pb/U ratios (e.g. the epoxy resin, mineral
inclusions in the zircons or fractures and other zones possibly affected by Pb-loss). Reported
uncertainties were propagated by quadratic addition of the external reproducibility (standard
deviation) from the primary standard zircon GJ-1 (Morel et al., 2008) during individual analytical
sessions and the within-run precision of each analysis (standard error). In order to test the validity of
the applied methods and the reproducibility of the data, multiple analyses of the reference zircon
Plešovice (Sláma et al., 2008) were performed. Plešovice gave a concordia age of 338.9 ± 0.6 Ma (n =
51; MSWD Conc. + Equiv. = 1.4). This is in agreement, within uncertainty, with the accepted ID-
TIMS age reported for Plešovice (337.3 ± 0.4 Ma (2SD); Sláma et al., 2008). In addition, several
analyses of the in-house reference zircon BB (Santos et al., 2014) were conducted. BB yielded a
concordia age of 560.0 ± 1.1 Ma (n = 71; MSWD Conc. + Equiv. = 1.3), which is consistent with the
reported ID-TIMS age of this reference material (562.6 ± 0.3 Ma (2SD); Santos et al., 2014).
204
For ca. 90 % of analyses, no Pb-based common Pb corrections were applied, however,
analyses that contained high amounts of common Pb (distinguished by 204(Pb + Hg) counts noticeably
higher than the average background caused by 204Hg contamination in the carrier gas) were discarded.
For the remaining 10 % of analyses, we followed the methods described by Gerdes & Zeh (2006,
215
 204
2009) and Frei & Gerdes (2009). The Hg interference on mass 204 was estimated using the
202 204
measured Hg and a Hg/202Hg ratio of 0.2299. When the interference- and back-ground-corrected
204
Pb signal was below the detection limit of 20 counts per second (value defined by the amount of Hg
202 204
present in the gas and the error on the measurement of Hg and thus the interfering Hg), no
common-Pb correction was applied. Where it was necessary, the common-Pb correction was
204
calculated based on the interference- and background-corrected Pb signal along with a model Pb
composition (Stacey & Kramers, 1975).

The results of the LA-ICP-MS U-Pb dating of the samples and secondary standards are
presented in Appendix A10. All reported uncertainties of unknowns and standards (ratios and ages)
are at the 1 sigma level.

1.2. LA-ICP-MS Lu-Hf isotope analyses

Lu-Hf isotope analyses were carried out by laser ablation-inductively coupled plasma-mass
spectrometry (LA-ICP-MS), following the methods described by Gerdes & Zeh (2006, 2009).
Whenever possible, the Lu-Hf 40-50 μm laser spot was drilled « on top » of the 20-30 μm U-Pb laser
spot. Where this was not the case, care was taken to drill within the same zircon domain (core, rim)
previously analysed for U-Pb, characterized by CL imaging. Information on the data acquisition
methods for both analytical sessions is available in Table 1. Nitrogen was introduced into the Ar
sample carrier gas via an Aridus nebulisation system, inhibiting oxide formation in the plasma and
enhancing the signal sensitivity. As for the U-Th-Pb analyses, the integration window was processed
172 173 175
offline to verify the homogeneity of the ablated zircons. The isotopes Yb, Yb and Lu were
measured during each analysis to allow corrections of isobaric interferences of Lu and Yb isotopes on
mass 176. 176Lu and 176Yb were calculated using a 176Lu/175Lu ratio of 0.02656 and a 176Yb/173Yb ratio
of 0.795015 (both JWG in-house values). Instrumental biases were corrected using an exponential law
and 172Yb/173Yb and 179Hf/177Hf ratios of 1.35351 (mean of Chu et al., 2002 and Segal et al., 2003) and
0.7325 (Patchett et al., 1981) respectively. The mass bias of Yb (βYb) isotopes typically differs from
that of the Hafnium (βHf) with an offset of βHf/βYb, calculated for each analytical session by averaging
the βHf/βYb of multiple analyses of the Temora reference material. The mass bias behaviour of Lu was
assumed to follow that of Yb. For Yb-rich zircons, the βHf/βYb offset factor was determined from the
mean βYb of each analysis. For Yb-poor zircon crystals, the applied correction was calculated using the
βHf of each integration step divided by the average βHf/βYb offset factor of the whole analytical session.
It is important to note that any under- or over-correction for Yb and Lu interference on mass 176
would result in an apparent correlation of the 176Hf/177Hf and 176Yb/177Yb ratios. The dataset presented
176
in this study has been examined by plotting Hf/177Hf against 176
Yb/177Yb (Fig. A1), and the results
show insignificant interference by Yb, testifying of the good quality of the data.

216
 Quoted uncertainties are quadratic additions of the within-run precision and the reproducibility
of the reference zircons GJ-1 (Morel et al., 2008), Plešovice (Sláma et al., 2008), Temora (Wu et al.,
2006), 91500 (Blichert-Toft, 2008), Mud Tank (Woodhead & Hergt, 2005) and the in-house zircon BB
(Santos et al., 2014). During the first analytical session, three reference materials were used before and
during runs: GJ-1, Temora and BB. Multiple analyses of the reference zircon GJ-1 yielded a
176
Hf/177Hf = 0.282014 ± 29 (2SD; n = 36). Temora yielded an average 176
Hf/177Hf = 0.282694 ± 45
176
(2SD; n = 29). 57 analyses on the in-house zircon BB yielded a Hf/177Hf = 0.281673 ± 34 (2SD).
During the second analytical session, five reference materials were analysed before and during runs:
GJ-1 (176Hf/177Hf = 0.282010 ± 24 (2SD); n = 51), Temora (176Hf/177Hf = 0.282661 ± 28 (2SD); n =
75), Plešovice (176Hf/177Hf = 0.282472 ± 19 (2SD); n = 41), Mud Tank (176Hf/177Hf = 0.282498 ± 19
(2SD); n = 27) and BB (176Hf/177Hf = 0.281659 ± 5 (2SD); n = 5). These values agree within error with
the recommended values for GJ-1 (176Hf/177Hf = 0.282000 ± 5 (2ζ); Morel et al., 2008), Temora
(176Hf/177Hf = 0.282680 ± 31 (2ζ); Wu et al., 2006), Plešovice (176Hf/177Hf = 0.282482 ± 13 (2ζ);
Sláma et al., 2008), Mud Tank (176Hf/177Hf = 0.282504 ± 44 (2ζ); Woodhead & Hergt, 2005) and BB
(176Hf/177Hf = 0.281674 ± 18 (2ζ); Santos et al., 2014) (Fig. A2).

Figure A1 – 176Hf/177Hf versus 176Yb/177Hf for the GJ-1, Temora, Plešovice, Mud Tank and BB standards.
Analyses from the first and second sessions are represented in black and grey respectively. The apparent lack of
correlation for the Temora analyses suggests that accurate Yb interference corrections were applied.

217
Figure A2 – 176Hf/177Hf isotope ratios obtained from the GJ-1, Temora, Plešovice, Mud Tank and BB standard
zircons during both analytical sessions. The vertical dotted lines indicate when the system was retuned. The grey
fields represent the accepted range of error of the standards (2ζ).

Epsilon Hf (εHf) values and Hf model ages were calculated using Chondritic Uniform
Reservoir (CHUR) 176Lu/177Hf and 176Hf/177Hf values of 0.0336 and 0.282785 respectively (Bouvier et
al., 2008), ―maximum‖ Depleted Mantle (DM) values of 176
Lu/177Hf = 0.03933 and 176
Hf/177Hf =
0.283294 (Blichert-Toft & Puchtel, 2010), a 176Lu decay constant of λ = 1.867 x 10-11 a-1 (Söderlund et
176
al., 2004) and a crustal source composition with a Lu/177Hf ratio of 0.0113 (mean of average
continental crust as suggested by Taylor & McLennan (1985) and Wedepohl (1995)).

218
 Analyses performed during the first session were filtered using the calculated alpha dose for
each U-Pb-Th spot, following Murakami et al. (1991). Zircons with calculated alpha doses > 8 x 1015
alpha events per milligram are classified as metamict grains, and were therefore excluded from further
consideration. This was the case for only 8 zircon grains, testifying of the overall good preservation of
the zircons analysed in this study.

The results of the LA-MC-ICP-MS Lu-Hf analyses of the samples and reference zircon
standards are presented in Appendix A11.

1.3. SIMS O isotope analyses

Oxygen isotopes were determined from seven samples using secondary ion mass spectrometry
over a 2 days analytical session, following the methods described by Kemp et al. (2006, 2007). The
oxygen isotopic compositions of standards and samples are reported with the standard δ18O notation
(in ‰) relative to the Vienna Standard Mean Ocean Water (VSMOW). All data were normalized to
the reference zircon 91500 (δ18O = 10.07 ± 0.03 ‰ VSMOW; Valley, 2003). Instrumental drift was
monitored each day by bracketing 10-15 unknowns between 5-8 measurements of the 91500 zircon.
133
Analytical conditions were set with a primary beam of Cs+ with a 5.0 nA current focused to an
elliptical 15 x 20 μm pit 2-3 μm deep. Ions were extracted using a 10 kV secondary beam, and
simultaneously collected by dual Faraday cups. Count rates for 16O and 18O respectively average 1.9 x
109 cps and 3.7 x 106 cps for the first day of analyses. The sensitivity increased slightly during the
second day, with count rates averaging 3.8 x 109 cps for 16O and 7.6 x 106 cps for 18O. As a result, the
internal precision on the standards analyses improved from 0.18-0.44 ‰ (2SE) during the first day to
0.14-0.22 ‰ (2SE) during the second day. To monitor the quality of the data, several analyses of the
Temora 2 zircon standard were interposed between 91500 and unknown zircons. Temora 2 yielded
mean δ18O values of 8.38 ± 0.57 ‰ (1SD; n = 7) and 8.33 ± 0.34 ‰ (1SD; n = 5) for the first and
second day respectively. Both averages are in agreement with the accepted value for Temora 2 (δ 18O =
8.2 ± 0.01 ‰ VSMOW; Valley, 2003; Black et al., 2004).

Care was taken to avoid cracks and fractures, the presence of which can greatly enhance the
rate of O isotope exchange with other minerals or fluids and therefore alter the primary magmatic
signature of the grains. Analyses that were inadvertently drilled along cracks were discarded (this was
the case for 4 spots). A subset of analyses was duplicated, and the average precision of these analyses
was 0.38‰ (1SD; n = 9), which falls within the analytical precision of ion probe measurements (± 0.5
‰).

The results of the SIMS O isotope analyses of the samples and reference zircon standards are
listed in Appendix A12.

2. Sample and zircon description and isotopic results

219
 During this study, zircon grains from 29 samples of granitoids and gneisses and 1 amphibolitic
dike from the QF were investigated. Sixteen samples were previously dated by Farina et al. (2015a).
Additional grains were selected from the same zircon separate and mounted separately for seven of
these samples, in order to perform O isotope analyses. Eight samples were previously dated by
Romano et al. (2013) and simply re-dated in this study using the same mounts, in order to perform Lu-
Hf analyses on the same spots.

A brief description of the investigated samples is given below. For more detailed information
on the rocks (sample location, field, petrographic and geochemical information), the reader is referred
to the two articles cited above. Concordia diagrams and CL images of representative grains can be
found in Figures A3 and A4 respectively.

2.1. Bação complex

 FQ1 is from a small domain of medium- to coarse-grained, weakly foliated granite exposed
along the contact with the supracrustal sequences at the southern margin of the dome. Zircons
are yellow and purple to pale-brown. In CL, the grains display well-defined oscillatory zoning,
but no core-rim relationships could be identified. Farina et al. (2015a) obtained a concordia
age of 2711 ± 3 Ma that was interpreted as the crystallization age of the granite. During this
study, 19 U-Pb analyses from slightly to highly discordant zircons plot on the same regression
line than that obtained by Farina et al. (2015a) (Fig. A3a). We therefore retain 2711 ± 3 Ma as
the crystallization age of FQ1. This age is identical to the SHRIMP age of 2716 ± 5 Ma
obtained by Romano et al. (2013) on MR01, collected from the same granitic domain. Th/U
ratios are highly variable, ranging from 0.04 to 1.13 (Table 2). 176Hf/177Hft from 22 spots range
from 0.280927 ± 19 to 0.281087 ± 39, with an average of 0.280997 ± 24 corresponding to a
mean εHf2711 Ma of -1.8 ± 0.8. δ18O from 12 analyses range from 4.11 ± 0.30 to 7.34 ± 0.19 ‰,
with a weighted mean of 5.50 ± 0.24 ‰.

 FQ2 and FQ23 are from a fine-grained banded gneiss outcrop (FQ23 is from a slightly more
homogeneous domain of the gneiss than FQ2 which displays a more pronounced banding).
The gneiss is intruded by numerous massive pegmatitic and aplitic dikes. The zircons from
both FQ2 and FQ23 are stubby to prismatic, reaching up to 300 µm in length. CL images
reveal a majority of zircons characterized by fine-scale oscillatory zoning, sometimes
disturbed. About 20 % of them display bright homogeneous centres surrounded by dark
overgrowths sometimes showing some convolute zoning, and a subset of dark and
structureless grains also occur. For FQ2, Farina et al. (2015a) performed analyses on both the
zoned domains and homogeneous cores and obtained a magmatic age of 2868 ± 10 Ma.
Several spot analyses on the dark structureless zircons yielded an age of 2705 ± 18 Ma
interpreted as the age of a late metamorphic event. For FQ23, these authors mostly targeted

220
 the bright cores as well as several zoned domains and obtained an average age of 2898 ± 12
Ma from eight concordant points, interpreted as inherited. Four concordant grains yielded a
younger age of 2783 ± 18 Ma which was interpreted as the magmatic age of the rock.
However, this interpretation can be debated, as FQ23 was collected only a few hundred metres
away from FQ2, whose well-defined magmatic age of 2868 Ma is comparable within error to
2898 Ma. Moreover, the presence of such a large number of concordant inherited zircon cores
in these rocks is a rather uncommon feature (Romano et al., 2013; Lana et al., 2013). We
tentatively propose that 2898 Ma represents the crystallization age of the gneiss FQ23, and
2783 Ma a metamorphic age. Furthermore, the latter is identical to the metamorphic age of ca.
2.77-2.79 Ga obtained by Lana et al. (2013) on several gneisses from the Bação complex. For
FQ2, 27 U-Pb analyses from subconcordant to highly discordant zircons were performed
during this study. Out of this population, ten were further associated with Hf and/or O
analyses. Six U-Pb spots performed on the bright cores provided subconcordant to discordant
206
Pb/207Pb ages ranging from 2603 ± 20 (79.5 % conc.) to 2855 ± 25 (96.3 % conc.) Ma,
interpreted as magmatic features. Th/U ratios are moderate, ranging from 0.21 to 0.43. Four
analyses on the darker rims are characterized by notably lower Th/U ratios (< 0.05) and
yielded younger discordant ages, interpreted as metamorphic features, in agreement with the
176
interpretation of Farina et al. (2015a). Hf/177Hft from 19 spots from the bright cores and
zoned domains (dated by Farina et al. (2015a) and in this study) range from 0.280978 ± 35 to
0.281106 ± 25 with a weighted mean of 0.281009 ± 24. Hf analyses performed on the darker
rims and on the dark structureless zircons yielded a similar range of values with a weighted
mean of 0.281035 ± 27. O analyses performed on the bright cores (mean of 3.9 ± 0.2 ‰) are
similar to those performed on the dark rims (mean of 4.2 ± 0.2 ‰) (Fig. A3b). The identical
(within error) Hf and O isotope values between the magmatic and the metamorphic domains
possibly indicate that these isotopic systems remained undisturbed during the metamorphic
event affecting FQ2 at ca. 2705 Ma. For FQ23, eight Lu-Hf analyses were performed on the
176
bright magmatic cores and yielded Hf/177Hft from 0.280908 ± 23 to 0.281047 ± 23, with an
176
average of 0.280987 ± 21. Hf/177Hft from three analyses on the darker rims ranged from
0.280990 ± 23 to 0.281060 ± 17. The range of magmatic Hf compositions obtained from
FQ23 is similar to the one from FQ2, and corresponds to εHfint of 2.1 ± 3.3 and 2.6 ± 2.1
(2SD).

 FQ5 and FQ6 were collected at the northern margin of the dome. FQ6 is a fine-grained, Kfs-
bearing banded gneiss (corresponds to sample D12 from Lana et al., 2013) intruded by FQ5, a
medium- to coarse-grained Plg-rich grey granitoid (corresponds to sample MR11 from
Romano et al., 2013). Lana et al. (2013) performed nine SHRIMP analyses on the zoned
zircon centres of D12 and obtained an upper intercept age of 2764 ± 10 Ma interpreted as the

221
crystallization age of the rock. Zircons from sample MR11 yielded an upper intercept age at
2744 ± 10 Ma interpreted by Romano et al. (2013) as the magmatic age of the granitoid.
Zircons from FQ5 are prismatic to stubby with round terminations, reaching up to 400 µm in
length. In CL, the grains are dark and either homogeneous or with fine-scale oscillatory
zoning. LA-ICP-MS analyses on zoned grains and a subset of homogeneous ones gave
concordant to slightly discordant apparent ages ranging between 2721 and 2805 Ma. Th/U
ratios are moderate, comprised between 0.16 and 0.59 (Table 2). Nine analyses passing the 2
% discordancy filter gave a weighted average age of 2761 ± 11 Ma (95 % c.l.; MSWD =
0.99), interpreted as the crystallization age of the granitoid (Fig. A3c). This age is identical
within error to the one obtained by Romano et al. (2013) from MR11. One analysis in the core
of a small rounded grain yielded a concordant age of 3292 ± 15 Ma, interpreted as inherited.
Initial εHft from 18 analyses on concordant and subconcordant zircons range from -1.5 ± 0.7
to 1.5 ± 0.6 with an average of εHf2761 Ma = -0.2 ± 1.8 (2SD). The majority of zircons from
FQ6 are long and prismatic, reaching in length up to 250 µm, with width:length ratios between
1:3 and 1:6. In CL, these zircons are dark and homogeneous, or weakly zoned. A subset of
zircons is stubby with round terminations, sometimes equant, and is characterized by
width:length ratios of ca. 1:2. CL images reveal the presence of bright rounded cores that
sometimes show internal oscillatory zoning, truncated by dark homogeneous rims. U-Pb
analyses on these cores yielded two populations of zircons (Fig. A3d). Seven spots have
206
concordant to slightly discordant apparent Pb/207Pb ages ranging from 2878 to 2912 Ma. Of
these, four concordant points define a concordia age of 2891 ± 3 Ma. A second population of
analyses performed on the bright cores yielded a significantly younger concordia age of 2779
± 4 Ma (95 % c.l.; MSWD = 0.92). Based on CL images, no explanation could be found to
explain this difference. For their sample D12, Lana et al. (2013) obtained a crystallization age
of 2764 ± 10 Ma, which is very close to the one given by our second population of zircons.
We therefore tentatively propose that 2779 Ma represents the crystallization age of the gneiss,
and the 2891 Ma cores population as inherited. LA-ICP-MS analyses on the dark prismatic
zircons yielded concordant to highly discordant apparent ages ranging from 2198 ± 18 to 2727
± 17 Ma. Eleven concordant spots yielded a concordia age of 2705 ± 7 Ma (MSWD = 2.2),
which we interpret as the age of a late magmatic-metamorphic event. There is little variation
of the Th/U ratio within these three populations of zircons, generally comprised between 0.15
and 0.45. However we observe a slight decrease of this ratio with time, from an average of
0.33 at 2891 Ma to 0.25 at 2705 Ma. εHft from the four inherited zircons range from 2.7 ± 1.0
to 6.2 ± 0.8. The three concordant zircons interpreted as magmatic have chondritic εHft values
between -2 ± 0.8 and -0.1 ± 0.9, with a mean εHf2779 Ma = -0.9 ± 2.2 (2SD). The younger
population of zircons has εHft ranging from -2.5 ± 0.8 to 0.7 ± 0.7.

222
 FQ8 is a Kfs-rich fine-grained banded gneiss collected ~5 km southeast from FQ6. It is cut by
several systems of leucogranitic dikes and a later tourmaline-bearing pegmatitic dike. Zircons
from FQ8 are prismatic, ranging from 100 to 250 µm in length. CL images reveal a majority
of dark and structureless zircons (ca. 60 %) while the rest displays some light grey centres
surrounded by dark structureless rims. A handful of zircons exhibit bright cores displaying
banded or oscillatory zoning, surrounded by dark rims. 34 U-Pb analyses gave concordant to
very discordant ages. 15 concordant points define a weighted mean age of 2612 ± 10 Ma (95
% c.l.; MSWD = 1.5). We interpret this age as that of a metamorphic event affecting the
gneiss. Four discordant points from the bright cores gave older apparent ages, plotting on a
regression line that intercepts the upper concordia at 2770 ± 42 Ma (Fig. A3e). We tentatively
propose that this poorly defined age represents the age of inherited zircons from the protolith
of the gneiss, which has been largely reworked at 2612 Ma. Lu-Hf analyses on concordant and
subconcordant grains yielded 176Hf/177Hft values from 0.280934 ± 16 to 0.281081 ± 11. If two
176
points with notably higher Hf/177Hft (0.281060 and 0.281081) are discarded, the 16
remaining analyses give an average 176Hf/177Hft of 0.280979 ± 15 corresponding to an average
εHf2612 Ma = -4.4 ± 0.5. The cores have similar Hf/177Hft values, which corresponds to εHft
176

comprised between -1.4 ± 0.6 and 2.3 ± 0.6.

 FQ11 is a weakly foliated, medium-grained biotite-poor granite, characterized by the

occurrence of large K-feldspar crystals (reaching up to 6 cm). Zircons from this sample are
mostly prismatic, reaching up to 350 µm. In CL, most of the grains display fine-scale
oscillatory zoning, while a handful of zircons are dark and homogeneous. A subset of grains
exhibit complex structureless cores with oscillatory zoned rims. Farina et al. (2015a) obtained
an inherited age of 2905 ± 17 Ma from 3 concordant spots on the cores. Analyses on the zoned
domains yielded a well-defined magmatic age of 2790 ± 3 Ma for the granitoid. Finally these
authors obtained a younger age at 2719 ± 14 Ma from the featureless grains and zoned rims
interpreted as metamorphic. The magmatic and metamorphic domains have similar 176Hf/177Hft
values, averaging respectively 0.281024 ± 18 and 0.281014 ± 22, which corresponds to εHf2790
Ma = 1.1 ± 0.6 and εHf2719 Ma = -1.1 ± 0.8. Initial εHft for 1 core gave a superchondritic value of
4.9 ± 0.7.

 FQ13 is from a medium-grained gneiss showing some fleck structures, intruded by

leucogranitic and pegmatitic dikes. Zircons are prismatic to stubby reaching up to 250 µm in
length. Most grains exhibit fine-scale oscillatory zoning in cathodoluminescence, while a few
crystals are homogeneous or weakly zoned. Farina et al. (2015a) obtained a well-defined
concordia age of 2790 ± 3 Ma interpreted as the magmatic age of the gneiss, as well as a
younger age of 2701 ± 15 Ma which might either result from Pb-loss or represent the
176
metamorphic event. 14 Lu-Hf analyses on the magmatic crystals gave Hf/177Hft values

223
 comprised between 0.280909 ± 23 and 0.281075 ± 31. Excluding two analyses with
176
Hf/177Hft ratios of 0.280909 and 0.280945, this population has an average Hf composition of
Hf/177Hft = 0.281038 ± 26, corresponding to εHf2790 Ma = 1.5 ± 0.9. 2 analyses on younger
176

zircons gave initial εHft values of -0.4 ± 1.1 and -0.5 ± 0.9. One core at 2900 Ma has a εHft
value of -1.2 ± 0.9.

 FQ14 was collected from a migmatitic gneiss in the south of the dome. The rock is
characterized by the occurrence of Qtz-Kfs-Plag leucosomes containing garnet poikiloblasts
reaching up to 6 cm across. The gneiss is cut by decimetre- and metre-scale pegmatitic dikes.
Zircons are mostly prismatic, ranging in size from 90 to 400 µm, with width:length ratios
varying from 1:3 to 1:6. Cathodoluminescence images revealed mostly dark and structureless
zircons. Ca. 40 % of the grains display clear core-rim relationships, with bright and
homogeneous, rarely zoned cores, while the rims are dark and homogeneous. LA-ICP-MS
analyses on the cores gave a wide range of concordant (< 1 % disc.) to discordant apparent
ages ranging from 2897 ± 17 to 3472 ± 19 Ma (Fig. A3f). In particular, four main populations
are observed at 2933, 3202, 3358 and 3465 Ma. These ages are interpreted as inherited from
the protolith of the gneiss. Six concordant analyses on the dark and featureless crystals define
a concordia age of 2692 ± 4 Ma (95 % c.l.; MSWD = 1.3). The occurrence of clusters of
inherited ages between 2925 and 3472 Ma suggest that the protolith for FQ14 was a
(meta)sedimentary rock that underwent partial melting at 2692 Ma. Lu-Hf analyses on
concordant cores revealed a wide range of initial εHft values ranging from of -4.1 ± 0.8 (at
3358 Ma) to 11.1 ± 0.8 (at 3472 Ma). Initial εHft values from the younger zircon population
are comprised between -2.0 ± 0.9 to 0.4 ± 1.2, with an average of -1.2 ± 1.0.

 FQ17 is from a quarry exposing a fine-grained banded gneiss crosscut by a large number of
leucogranite sheets that range from milimetre- to metre-scale. FQ17 was collected from the
host gneiss. Zircons are stubby to prismatic, and range in size from 75 to 350 µm, with a
maximum width:length ratio of 1:5. In CL, the grains are characterized by well-defined
oscillatory zoning, with clear core-rim relationships. The centres exhibit fine-scale zoning,
while the overgrowths are rather homogeneous or weakly zoned. Farina et al. (2015) obtained
a metamorphic age of 2732 ± 10 Ma from analyses on the rims. Additionally, these authors
obtained a crystallization age of 2778 ± 2 Ma, as well as a small population of inherited cores
at ca. 2862 Ma. In this study, 39 analyses gave concordant to highly discordant apparent ages
ranging from 2281 ± 20 to 2829 ± 20 Ma (Fig. A3g). Nine analyses on the zoned centres of
206
the grains yielded older Pb/207Pb ages from 2781 to 2829 Ma, interpreted as recording the
crystallization of the rock. The remaining 30 analyses performed on homogeneous and zoned
crystals gave concordant to discordant ages from 2281 ± 20 to 2719 ± 19 Ma. These points are
interpreted to date the metamorphism affecting FQ17 during the Mamona event. Th/U ratios

224
 are mostly comprised between 0.03 and 0.68, with an outlier at 1.09. Th/U of four cores at ca.
2860 Ma average 0.47. Th/U ratios for the magmatic and metamorphic zircons are similar and
average ca. 0.25. However, we observe a significant number of zircons with Th/U < 0.10 (nine
analyses) within the younger population. 176Hf/177Hft from the four cores range from 0.280913
± 24 to 0.281034 ± 21, which corresponds to εHft from -1.1 ± 0.7 to 3.2 ± 0.7. The magmatic
176
and metamorphic grains have similar Hf/177Hft values averaging 0.280993 ± 21 (excluding
one analysis at 0.280919 ± 19) and 0.281006 ± 21 respectively, which corresponds to εHf 2778
Ma = -0.2 ± 1.3 (2SD) and εHf2732 Ma = -0.8 ± 1.7 (2SD). O isotope analyses from two
subconcordant magmatic zircons gave values close to those of the mantle (4.88 ± 0.20 and
5.98 ± 0.15 ‰). The remaining 10 points from metamorphic zircons yielded δ18O values from
4.00 ± 0.20 to 6.29 ± 0.19 ‰.

 FQ20 were collected at the southwestern margin of the dome. It consists of a strongly banded
fine- to medium-grained gneiss characterized by the alternance between plagioclase-rich and
biotite-rich bands, intruded by numerous leucogranite sheets. It corresponds to sample D07A
of Lana et al. (2013), who obtained a crystallization age of 2918 ± 10 Ma for the gneiss.
Additionally, these authors dated a crosscutting leucogranite sheet, and obtained a SHRIMP
weighted mean age of 2774 ± 11 Ma interpreted as the crystallization age of the leucogranite.
Zircons from FQ20 are stubby to prismatic with round terminations, reaching up to 200 µm in
length. CL images reveal a majority of dark and structureless zircons, while a subset of
crystals exhibit bright rounded cores, sometimes zoned, surrounded by dark featureless rims.
31 LA-ICP-MS U-Pb analyses on homogeneous zircons gave concordant to highly discordant
ages, plotting on a Pb-loss trend with an upper intercept age of 2726 ± 16 Ma (Fig. A3h). This
age is identical to the concordia age of 2723 ± 3 Ma (95 % c.l., MSWD = 0.66) obtained from
five concordant (< 1 % disc.) points, and considered as the metamorphic age of the gneiss.
206
Two analyses yielded older concordant ages Pb/207Pb at 2772 ± 14 and 2829 ± 16 Ma. The
former was analysed from a bright core and is interpreted as inherited, but the latter was
performed on a dark and totally homogeneous zircon grain, casting doubts on the origin of this
older age. Initial εHft values from five concordant homogeneous grains range between -1.2 ±
0.8 to 1.0 ± 1.2, with an average of εHf2723 Ma = -0.3 ± 1.1. The core at 2772 Ma yielded a
superchondritic εHft value of 2.5 ± 1.2.

2.2. Bonfim complex

 FQ29 is from a medium-grained foliated grey granitoid rock exposed in the Samambaia pluton
in the eastern edge of the complex. It is composed of plagioclase, quartz, microcline, biotite
and epidote. The granitoid is intruded by fine-grained leucogranitic dikes. Zircons are stubby
to prismatic, and range in size from 100 to 300 µm. In CL, the grains display well-defined
fine-scale oscillatory zoning. A few zircons exhibit homogeneous centres surrounded by
225
 oscillatory rims. Farina et al. (2015a) obtained a well-defined crystallization age of 2773 ± 2
Ma for FQ29. During this study, 43 analyses were performed on the zoned grains, yielding
concordant to highly discordant ages plotting on the same regression line as that obtained by
176
Farina et al. (2015a) (Fig. A3i). Hf/177Hft from 42 spots range from 0.280881 ± 15 to
0.280972 ± 16, with an average of 0.280910 ± 18 corresponding to a mean εHf 2773 Ma of -3.2 ±
0.6. δ18O from 15 analyses range from 2.68 ± 0.24 to 6.23 ± 0.20 ‰, with a weighted mean of
4.90 ± 0.20 ‰.

 FQ37 is an amphibolitic dike intruded in the northeastern part of the dome. Zircons are stubby
to prismatic, reaching up to 250 µm in length. In CL, they commonly exhibit core-rim
relationships, with bright cores, sometimes zoned, surrounded by darker rims, either
homogeneous or oscillatory zoned. Farina et al. (2015a) obtained an inherited age of 2879 ±
14 Ma from analyses performed on the cores, and a younger age of 2719 ± 14 Ma from the
dark rims. The latter was interpreted by these authors as the age of crystallization of the
amphibolite, while 2879 Ma old cores were interpreted as resorbed crystals assimilated by the
dike. Th/U show a systematic difference between the two populations of zircons, averaging
0.46 for the older grains and 0.24 for the grains at 2719 Ma. On the other hand, these two
176
populations of zircons have similar Hf/177Hft values, ranging from 0.280863 ± 17 to
0.280998 ± 21 (after exclusion of one outlier at 0.280689 ± 21), with averages of 0.280916 ±
23 (2719 Ma) and 0.280956 ± 24 (2879 Ma). These values correspond respectively to εHf 2719
Ma = -2.8 ± 2.9 (2SD) and εHf2879 Ma = -0.5 ± 2.7 (2SD).

 FQ40 is from a biotite-bearing banded gneiss exposed in the central part of the dome, intruded
by numerous leucogranitic sheets parallel to the foliation. Zircons are prismatic, with
width:length ratios up to 1:4. CL images reveal fine-scale oscillatory zoning, with a subset of
small black homogeneous grains. Ca. 20 % of the grains exhibit core-rim relationships, with
CL-bright or CL-dark rounded cores, homogeneous or zoned, surrounded by dark structureless
rims. Farina et al. (2015a) obtained an inherited age of 2854 ± 18 Ma from the cores, and a
younger weighted mean age of 2670 ± 15 Ma. These authors inferred that FQ40 crystallized at
2854 Ma (this age is similar to the magmatic age of the gneiss FQ41 cropping out close by),
176
and underwent a metamorphic event at 2760 Ma. The cores show lower Hf/177Hft values
compared to the metamorphic rims, with averages of 0.280871 ± 20 and 0.280940 ± 22
respectively. These values correspond to εHf2854 Ma = -2.7 ± 1.1 (2SD) and εHf2670 Ma = -4.6 ±
2.8 (2SD).

 FQ41 is from a banded gneiss outcrop just a few kilometres away from FQ40. The gneiss is
intruded by numerous foliation-parallel leucogranite sheets. Zircons extracted from FQ41 are
equant to prismatic, ranging from 100 to 350 µm. In CL, the grains exhibit well-defined
oscillatory zoning interiors with no obvious core structures. Farina et al. (2015a) obtained a
226
 magmatic age of 2852 ± 16 Ma, which is identical to the age obtained from several cores from
sample FQ40 a few kilometres away. Initial εHft range from -1.4 ± 0.7 to -3.7 ± 0.9, with an
average εHf2852 Ma = -1.8 ± 1.9 (2SD), which is similar to the Hf compositions obtained from
the cores of FQ40.

 FQ51 is a foliated fine- to medium-grained granitoid, crosscut by numerous pegmatitic dikes.

Zircons are prismatic to stubby with round terminations, reaching up to 300 µm. CL images
reveal fine to coarse oscillatory zoning. Ca. 20 % of the grains have bright centres surrounded
by dark homogeneous rims. Farina et al. (2015a) obtained a crystallization age of 2678 ± 10
Ma for this sample. Lu-Hf analyses on 28 concordant and subconcordant (< 5 % disc.) grains
176
yielded Hf/177Hft values comprised between 0.281001 ± 20 and 0.281064 ± 24, with an
average of 0.281026 ± 22, which corresponds to near-chondritic εHf2678 Ma = -1.3 ± 1.3 (2SD).

 FQ52 was collected from a trondhjemitic gneiss in the central part of the complex. It is
characterized by a vertical banding made up of an alternance of leucocratic (plagioclase,
quartz, minor microcline) and mesocratic (biotite, plagioclase). Zircons are prismatic to stubby
with round terminations, and are characterized in CL by dark and structureless or weakly
zoned grains. A subset of crystals show CL-bright or CL-dark homogeneous cores surrounded
by dark homogeneous or weakly zoned rims. Farina et al. (2015a) obtained a weighted mean
age of 2727 ± 11 Ma, which these authors interpreted as the metamorphic age of the rock. In
this study, 19 U-Pb analyses yielded apparent ages ranging from 1533 ± 20 to 2718 ± 17 Ma
(Fig. A3j). However, these points reach high discordance levels (up to 40.3 % disc.), therefore
we retain the weighted mean age obtained by Farina et al. (2015a) from 16 concordant
analyses as the metamorphic age of the rock. Th/U ratios are highly variable, ranging from
176
0.16 to 0.76. Initial Hf/177Hft values are comprised between 0.280966 ± 20 and 0.281085 ±
176
22. If we exclude one spot with a significantly higher Hf/177Hft ratio (0.281085 ± 22), we
obtain an average value of 0.281004 ± 21 from 28 analyses, which corresponds to a mean
εHf2727 Ma = -1.1 ± 1.3 (2SD). δ18O from 13 analyses range from 2.88 ± 0.29 to 10.88 ± 0.35
‰. However, 10.88 ‰ is an unusually high value for an Archean zircon (commonly < 7.5 ‰).
This crystal is prismatic, exhibits fine-scale oscillatory zoning and is clear of any fractures or
inclusions that could have indicated some contamination. Moreover, this zircon gave a
subconcordant apparent age of 2718 Ma. Nonetheless, we consider this value as an outlier,
resulting from post-magmatic alteration processes, as it is much higher than the rest of the
sample‘s average (4.58 ± 0.35 ‰).

 MR31A is a sample collected by Romano et al. (2013) from a non foliated leucogranite body
intruded in the northern part of the Souza Noschese batholith. Zircons are long and prismatic
with round terminations, reaching up to 300 µm in length. CL images expose well-defined
oscillatory-zoned crystals, sometimes surrounded by resorbed rims. Romano et al. (2013)
227
 obtained a crystallization age of 2700 ± 8 Ma for this sample. In this study, analyses of the
centres of the zoned grains gave concordant to slightly discordant apparent ages plotting along
a line with an upper intercept age of 2719 ± 14 Ma (MSWD = 6.2) (Fig. A3k). Th/U ratios are
moderate to high, comprised between 0.24 and 0.94. 13 concordant points (< 1 % disc.) define
a weighted mean age of 2716 ± 14 Ma (95 % c.l.; MSWD = 1.6) which is identical to the
intercept age, and that we consider as the magmatic age of the leucogranite. This age is
identical within error to the one obtained by Romano et al. (2013). Initial εHft values range
between -4.7 ± 1.1 to -0.9 ± 0.9, with an average of εHf2716 Ma = -2.8 ± 1.0.

 MR22A consists of a medium-grained weakly-foliated grey granodiorite intrusion in the

Souza Noschese batholith. It was collected by Romano et al. (2013), who obtained a
magmatic age of 2730 ± 8 Ma for this rock. Zircons are prismatic to equant, and display fine-
scale oscillatory zoning in CL. In this study, 41 analyses of the centres of the grains yielded
concordant to discordant ages, almost all of which plot on a Pb-loss line that intercepts the
upper concordia at 2721 ± 7 Ma (MSWD = 5.0). 22 points that passed the 1 % discordancy
filter gave a concordia age of 2715 ± 2 Ma (95 % c.l.; MSWD = 0.097), that is identical within
error to the upper intercept age, and that we interpret as the crystallization age of the granitoid.
This age is slightly younger than that obtained by Romano et al. (2013), but it is still within
the analytical precision of the method (~ 1 % of the age). Three spots gave notably older
concordant or subconcordant apparent ages from 2835 ± 18 to 3515 ± 19 Ma (Fig. A3l). These
ages are not directly associated with apparent cores in cathodoluminescence images, therefore
the interpretation of these ages as inherited may be controversial. Seven Lu-Hf analyses on
concordant magmatic crystals yielded an average εHf2715 Ma = -3.8 ± 1.8 (2SD). Two of the
―older grains‖ have εHft superchondritic values of 2.1 ± 1.0 (at 2835 Ma) and 6.9 ± 1.3 (at
3141 Ma).

 MR70G was collected by Romano et al. (2013) from a non-foliated coarse-grained

leucogranite in the southern extremity of the Mamona batholith. The sample yielded short
prismatic zircons up to 250 µm in length, characterized by well-defined oscillatory zoning
interiors. A subset of grains displays some dark resorbed rims, while others show some
resorbed U-rich cores. Romano et al. (2013) obtained a well-defined magmatic age of 2723 ±
7 Ma, but could not obtain a coherent age for the cores. In this study, we performed 38 spot
analyses, 33 of which passed the 1 % discordancy filter and defined a weighted mean age of
2716 ± 6 Ma (95 % c.l.; MSWD = 0.85), interpreted as the crystallization age of MR70G (Fig.
A3m). This age is identical within error to the one obtained by Romano et al. (2013). Initial
εHft values from 20 concordant points range from -1.6 ± 1.0 to -5.0 ± 0.9, with a mean εHf2716
Ma = -3.2 ± 1.0.

228
 MR87A is from a foliated fine- to medium-grained granodiorite in the eastern part of the
Mamona batholith collected by Romano et al. (2013). Zircons are prismatic to stubby, varying
in length from 150 to > 350 µm. CL images show a majority of grains characterized by well-
defined oscillatory zoning. A subset of grains displays some dark and structureless resorbed
cores. Romano et al. (2013) analysed prismatic zoned crystals and obtained a crystallization
age of 2613 ± 6 Ma for this sample. In this study, we performed 38 U-Pb analyses that yielded
concordant to slightly discordant apparent ages from 2602 ± 21 to 2736 ± 18 Ma. Th/U ratios
are high, ranging from 0.48 to 1.14. 22 concordant (< 1 % disc.) spots yielded a wide range of
ages comprised between 2612 and 2678 Ma, defining a weighted mean age of 2645 ± 9 Ma
(95 % c.l.; MSWD = 1.3), interpreted as the crystallization age of the sample (Fig. A3n). This
age is older than the one obtained by Romano et al. (2013). However, it is important to note
that these authors also observed a handful of zircons with apparent ages slightly older (by ~ 20
Ma) than 2613 Ma. Eight discordant to slightly reversely discordant points plot on a
regression line with an upper intercept age of 2734 ± 8 Ma that is slightly older than the rest of
the zircons. These points are not associated with any core features in particular, hence this
older and more poorly defined age is difficult to interpret. 2734 Ma is identical to the age of
emplacement of the Mamona batholith (2730 Ma; Romano et al., 2013) that MR87A intrudes.
Hence these zircons perhaps represent a few assimilated grains from the Mamona batholith
during ascent or emplacement of this sample. Lu-Hf analyses on 21 concordant zircons
yielded initial εHft values from -2.8 ± 1.2 to -5.7 ± 0.7, with a mean εHf2645 Ma = -4.4 ± 0.9.

 MR14A is from Romano et al. (2013), and consists of a non-foliated coarse-grained granitoid,
which represents the most voluminous phase of the Mamona batholith. Only a handful of
zircons from MR14A were suitable for dating. These zircons are mostly prismatic, reaching
up to 300 µm in length. CL images reveal grains with homogeneous centres surrounded by
oscillatory-zoned rims. Romano et al. (2013) obtained a magmatic age of 2730 ± 7 Ma for this
granitoid. In this study, 23 concordant to discordant points plot on a Pb-loss line that
intercepts the upper concordia at 2725 ± 11 Ma. This age is identical within error to the
concordia age of 2715 ± 3 Ma defined by nine concordant points, interpreted as the magmatic
age of MR14A (Fig. A3o). This age is slightly younger than the one obtained from Romano et
al. (2013), but is still comprised with analytical error of the procedure. Two slightly
discordant analyses yielded older apparent ages of 2848 and 3016 Ma. Romano et al. (2013)
also obtained some discordant inherited cores, suggesting the incorporation of some older
material in the petrogenesis of this rock, however the authors could not be more precise. Here,
these ages were obtained from zircon fragments, from where information on the internal
structures of the crystals is missing. Therefore will not attempt any interpretation on these

229
 ages. Hf compositions from seven concordant points are clustered around an average εHf2715 Ma
value of -3.0 ± 1.6 (2SD).

2.3. Belo Horizonte (+ Caeté) complex

 FQ60 was collected from the western border of the Florestal batholith. It consists of a weakly-
foliated medium-grained granite, intruded by leucogranitic dikes up to 40 cm thick. Multiple
quartz veins cut both the granite and the leucogranitic dikes. Zircons are stubby to prismatic,
with width:length ratios up to 1:5, and range in size from 75 to 350 µm. In CL, zircons are
either dark and structureless or weakly zoned. Some grains exhibit some light grey centres,
sometimes zoned, surrounded by dark rims. Farina et al. (2015a) obtained a weighted mean
crystallization age of 2728 ± 16 Ma from four concordant grains. In this study, 13 spots on
weakly-zoned grains gave concordant to highly discordant ages plotting along the regression
line defined by Farina et al. (2015a) (Fig. A3p). Additionally, two analyses were performed on
the bright core of a subrounded zircon, yielding concordant ages of 2794 ± 20 and 2776 ± 22,
interpreted as inherited. Lu-Hf analyses on 20 concordant and discordant grains yielded
176
Hf/177Hft values comprised between 0.280920 ± 18 and 0.281096 ± 33. If we exclude one
176
spot at 0.281096, the average Hf/177Hft of the sample is 0.280965 ± 21, which corresponds
to an average εHf2728 Ma = -2.4 ± 2.0 (2SD). δ18O from 10 analyses range from 2.86 ± 0.33 to
7.04 ± 0.33 ‰, with a weighted mean of 4.60 ± 0.32 ‰. Two analyses on the concordant core
yielded mantle-like δ18O values of 5.34 ± 0.37 and 5.41 ± 0.40.
 FQ65 is a medium-grained granitoid collected from the Pequi batholith, and corresponds to
sample MR259A of Romano et al. (2013). Zircons are prismatic, reaching up to 300 µm in
length. In CL, the grains are mostly dark to light grey and structureless. A minority of grains
show zoned centres surrounded by dark rims. Farina et al. (2015a) obtained an age of 2645 ± 8
Ma from several structureless grains, interpreted as the metamorphic age of the rock. Initial
εHft values range from -3.1 ± 0.7 to -7.1 ± 0.9, with a mean εHf2645 Ma = -5.3 ± 0.7.
 FQ70 was collected in the southwest of the Belo Horizonte complex, and consists of a fine-
grained banded gneiss, intruded by leucogranitic sheets parallel to the gneissosity. Pegmatitic
and medium-grained leucogranitic dikes also intrude the gneiss, crosscutting the foliation.
Zircons extracted from FQ70 are prismatic, ranging in length from 100 to 300 µm. CL images
reveal a majority of fractured homogeneous and structureless grains, with a handful of zircons
showing light grey homogeneous centres surrounded by dark structureless rims. Farina et al.
(2015a) obtained a concordia age of 2713 ± 3 Ma, which these authors interpreted as the
metamorphic age of the gneiss. Initial εHft values from 11 analyses range from -1.5 ± 0.8 to -
3.7 ± 0.7, with a mean εHf2713 Ma = -2.2 ± 0.7.
 FQ74 consists of a medium-grained biotite-bearing augen gneiss, locally exhibiting large
microcline crystals of up to 8 cm. Zircons separated from FQ74 are blocky to prismatic,

230
 ranging from 75 to 400 µm. CL images reveal highly fractured grains, sometimes containing
numerous U-Th-rich inclusions. Zircons are often dark and homogeneous, but a subset of
crystals display complex structureless, sometimes zoned) centres surrounded by either
homogeneous or weakly zoned rims. Farina et al. (2015a) obtained an age of 2638 ± 14 Ma
from nine concordant to subconcordant analyses. The authors interpreted this age as dating
either an episode of interaction between the granite and late-magmatic fluids or the latest
metamorphic event in the Belo Horizonte dome. In this study, 39 U-Pb analyses gave apparent
ages ranging from 2026 ± 22 to 2715 ± 21 Ma, that plot along the regression line obtained by
Farina et al. (2015a) (Fig. A3q). Th/U ratios are highly variable, ranging from 0.16 to 1.08.
176
Hf/177Hft values from 39 analyses range from 0.280942 ± 15 to 0.281070 ± 21, with a mean
of 0.280976 ± 19, which corresponds to an average εHf2638 Ma of -4.0 ± 1.5 (2SD). FQ74 shows
a wide range of δ18O values, ranging from 2.50 ± 0.34 to 7.79 ± 0.41 ‰.
 FQ81 was collected from the western Caeté dome, in the east of the Belo Horizonte complex.
It consists of a strongly foliated, fine- to medium-grained banded gneiss. Zircons from FQ81
are blocky, ranging in size from 80 to 300 µm. CL images reveal a majority of dark and
structureless zircons, while a few grains show light grey homogeneous centres surrounded by
dark rims. 23 U-Pb analyses yielded concordant to discordant apparent ages ranging from
2531 ± 29 to 2710 ± 29 Ma, which plot on a Pb-loss trend with an upper intercept age of 2674
± 18 Ma (Fig. A3r). 12 analyses that pass the 1 % discordancy filter yielded an identical
weighted mean age of 2674 ± 13 Ma (95 % c.l.; MSWD = 0.97). Machado et al. (1992)
+7
sampled the gneiss a few kilometres away and obtained a TIMS magmatic age of 2776 /-6
Ma for this rock. We suggest that the rock crystallized at 2776 Ma and underwent
metamorphism at 2674 Ma. 11 Lu-Hf analyses on concordant and subconcordant grains gave
initial εHft between -3.3 ± 0.8 and -6.2 ± 0.7, with a mean εHf2674 Ma = -5.1 ± 0.7.
 MR257A and MR259A were collected by Romano et al. (2013) from a large pavement in the
central part of the Pequi batholith. MR259A represents a medium-grained granitoid
crosscutting a grey granodioritic phase, and itself intruded by leucogranitic phases. MR257A
is a coarse-grained granitoid, representing one of this leucogranitic phases crosscutting
MR259A, either as vertical sheets or irregular veins. Romano et al. (2013) obtained intrusion
ages of 2722 ± 7 Ma and 2706 ± 7 Ma respectively for MR259A and MR257A. Zircons from
MR259A are prismatic to equant and blocky, reaching up to 300 µm in length. In CL, the
grains displayed mostly structureless interiors, while some grains show a poorly-defined
zoning. 26 U-Pb analyses gave concordant to discordant apparent ages, plotting on a Pb-loss
trend that intercepts the upper concordia at 2724 ± 5 Ma (Fig. A3t). That age is identical to the
weighted average age of 2721 ± 8 (95 % c.l.; MSWD = 0.2) obtained from 15 concordant (< 1
% disc.) points. The latter is interpreted as the magmatic age of the granitoid, and is identical
to the one obtained by Romano et al. (2013) for the same sample. Nine Lu-Hf analyses on
231
 concordant zircons yielded initial εHft values ranging between -0.9 ± 1.4 and -4.3 ± 0.8, with
an average of εHf2721 Ma = -3.0 ± 1.0. Zircons from MR257A are short prismatic to stubby and
equant, with slightly round terminations. A subset of crystals displays core-rim relationships,
with rounded cores surrounded by oscillatory-zoned rims. 35 analyses on the zoned domains
gave concordant to slightly discordant ages that plot on a regression line with upper intercept
age of 2727 ± 10 Ma (Fig. A3s). 18 spots that passed the 1 % discordancy filter defined a
weighted mean age of 2723 ± 8 Ma (95 % c.l.; MSWD = 0.57) considered as the age of
crystallization of the rock. This age is slightly older than that of Romano et al. (2013), but is
still considered within analytical error of the method. Initial εHf t values range from -1.6 ± 1.2
to -3.9 ± 1.0, with an average of εHf2723 Ma = -2.7 ± 1.0.
 MR51A is from a small granitoid intrusion (General Carneiro Granitoid) in the south of the
Belo Horizonte complex collected by Romano et al. (2013). It consists of a medium-grained
milk-white leucogranite, containing pale yellow to light pink clear zircons. The grains are
prismatic, ranging from 100 to 300 µm, and display well-defined oscillatory zoning in CL. No
clear core structures were observed. Romano et al. (2013) obtained a magmatic age of 2700 ±
8 Ma for this sample. In this study, we targeted mostly the centres of the grains, and obtained
206
24 concordant to slightly discordant Pb/207Pb ages (Fig. A3u). Ten of these points passed
the 1% discordancy filter and gave a weighted mean age of 2708 ± 10 Ma (95 % c.l.; MSWD
= 0.68), which is considered as the age of intrusion of MR51A. This age is identical to the one
176
obtained from Romano et al. (2013). Hf/177Hft from 19 spots range from 0.280913 ± 23 to
0.281104 ± 31, which corresponds to εHft values from -4.8 ± 0.8 to 1.8 ± 1.1. If we exclude
one analysis with a significantly higher Hf/177Hft ratio, the average εHf2708 Ma of this sample
176

becomes -3.3 ± 0.9.

232
Figure A3 – Concordia diagrams from samples from the Bação, Bonfim and Belo Horizonte complexes (only
for samples that were (re-)dated during this study). Error ellipses and error bars are at 1ζ. The data plotted in
grey are from Farina et al. (2015a), the data plotted in blue are from this study.

233
Figure A3 – (suite)

234
Figure A3 – (suite)

235
Figure A3 – (suite)

236
Figure A4 – Cathodoluminscence images of representative zircon grains.

237
Figure A4 – (suite)

238
Appendix A10 – U-Pb isotope results for samples and standards for both analytical sessions.
(Electronic appendix)

239
Appendix A11 – Lu-Hf isotope results. (Electronic appendix)

240
Appendix A12 – O isotope results. (Electronic appendix)

241
Appendix A13 – Electron microprobe tourmaline results. (Electronic appendix)

242
Appendix A14 – B isotope results. (Electronic appendix)

243
References for all appendices

Babinski M., Chemale Jr. F., Van Schmus W. R. 1995. The Pb/Pb age of the Minas Supergoup carbonate rocks,
Quadrilátero Ferrífero, Brazil. Precambrian Res., 72:235-245.
Black L. P., Kamo S. L., Allen C. M., Davis D. W., Aleinikoff J. N., Valley J. W., Mundil R., Campbell I. H.,
Korsch R. J., Williams I. S., Foudoulis C. 2004. Improved 206Pb/238U microprobe geochronology by the
monitoring of a trace-element-related matrix effect; SHRIMP, ID-TIMS, & LA-ICP-MS and oxygen isotope
documentation for a series of zircon standards. Chem. Geol., 205:115–140.
Blichert-Toft J. 2008. The Hf isotopic composition of zircon reference material 91500. Chem. Geol., 253:252–
257.
Blichert-Toft J. & Puchtel I. S. 2010. Depleted mantle sources through time: evidence from Lu–Hf and Sm–Nd
isotope systematics of Archean komatiites. Earth Planet. Sci. Lett., 297(3–4):598–606.
Bouvier A., Vervoort J. D., Patchett P. J. 2008. The Lu-Hf and Sm-Nd isotopic composition of CHUR:
constraints from unequilibrated chondrites and implications for the bulk composition of terrestrial planets.
Earth Planet. Sci. Lett., 273:48-57.
Brueckner H. K., Cunningham D., Alkmim F. F., Marshak S. 2000. Tectonic implications of Precambrian Sm-
Nd dates from the southern São Francisco craton and adjacent Araçuai and Ribeira belts, Brazil.
Precambrian Res., 99:255-269.
Buick I. S., Lana C., Gregory C. 2011. A LA-ICP-MS and SHRIMP U/Pb age constraint on the timing of REE
mineralisation associated with Bushveld granites. S. Afr. J. Geol., 144:1–14.
Cabral A. R., Zeh A., Koglin N., Gomes A. A. S., Viana D. J., Lehmann B. 2012. Dating the Itabira iron
formation, Quadrilátero Ferrífero of Minas Gerais, Brazil, at 2.65 Ga: depositional U-Pb age of zircon from
a metavolcanic layer. Precambrian Res., 204:40-45.
Carneiro M. A. 1992. O complexo metamórfico Bonfim setentrional (Quadrilátero Ferrífero, Minas Gerais):
Litoestratigrafia e evolução geológica de um segmento de crosta continental do Arqueano. University of
São Paulo, Brazil, PhD Thesis, 233p.
Cassino L. F. 2014. Distribuição de idades U-Pb de zircões detríticos dos Supergrupos Rio das Velhas e Minas
na Serra de Ouro Preto, Quadrilátero Ferrífero, MG e Implicações para a evolução sedimentar e tectônica.
Departamento de Geologia, Universidade Federal de Ouro Preto, Brazil, MSc Thesis, 52p.
Chemale F., Babinski M., Van Schmus W. R. 1993. U/Pb dating of granitic–gneissic rocks from the Belo
Horizonte and Bonfim complexes, Quadrilátero Ferrífero (Brazil). Report for CNPq and NSFEAR Project
on São Francisco Craton Margin Transect Project, 16p.
Chu N., Taylor R. N., Chavagnac V., Nesbitt R. W., Boella R. M., Milton J. A., German C. R., Bayon G., Burton
K. 2002. Hf isotope ratio analysis using multi-collector inductively coupled plasma mass spectrometry: an
evaluation of isobaric interference corrections. J. Anal. At. Spectrom., 17:1567–1574.
Da Silva L. C., Noce C. M., Lobato L. M. 2000. Dacitic volcanism in the course of the Rio das Velhas (2800-
2960 Ma) Orogeny: a Brazilian Archean analogue (TTG) to the modern adakites. Rev. Bras. Geocienc.,
30:384-387.
De Oliveira O. B. & Teixeira W. 1990. Evidências de uma tectônica tangencial proterozóica no Quadrilátero
Ferrífero, MG. In: Congr. Bras. Geol., 36, Anais, 6:2589-2604.
Farina F., Albert C., Lana C. 2015a. The Neoarchean transition between medium- and high-K granitoids: Clues
from the Southern São Francisco Craton (Brazil). Precambrian Res., 266:375-394.
Frei D. & Gerdes A. 2009. Precise and accurate in situ U-Pb dating of zircon with high sample throughput by
automated LA-SF-ICP-MS. Chem. Geol., 261:261-270.
Gerdes A. & Zeh A. 2006. Combined U-Pb and Hf isotope LA-(MC)-ICP-MS analyses of detrital zircons:
Comparison with SHRIMP and new constraints for the provenance and age of an Armorican metasediment
in Central Germany. Earth Planet. Sci. Lett., 249:47-61.
Gerdes A. & Zeh A. 2009. Zircon formation versus zircon alteration — new insights from combined U–Pb and
Lu–Hf in-situ LA-ICP-MS analyses, and consequences for the interpretation of Archean zircon from the
Central Zone of the Limpopo Belt. Chem. Geol., 261:230–243.

244
Gomes N. S. 1985. Petrologisch-geochemische Untersuchugen im Baçao-Komplex Eisernes Viereck, Minas
Gerais, Brasilien. Mathematisch-Naturwissenchaftlichen Fakultat, Technische Universitat Clausthal, PhD
Thesis, 209p.
Hartmann L. A., Endo I., Suita M. T. F., Santos J. O. S., Frantz J. C., Carneiro M. A., Naughton N. J., Barley M.
E. 2006. Provenance and age delimitation of Quadrilátero Ferrífero sandstones based on zircon U–Pb
isotopes. J. S. Am. Earth Sci., 20:273–285.
Henry D. J. & Dutrow B. L. 2002. Metamorphic tourmaline and its petrologic applications. In: E. S. Grew & L.
M. Anovitz (eds.) Boron: Mineralogy, Petrology and Geochemistry. Reviews in Mineralogy and
Geochemistry, Mineralogical Society of America, 503–557.
Jackson S. E., Pearson N. J., Griffin W. L., Belousova E. A. 2004. The application of laser ablation-inductively
coupled plasma-mass spectrometry to in situ U–Pb zircon geochronology. Chem. Geol., 211:47–69.
Kemp A. I. S., Hawkesworth C. J., Paterson B. A., Kinny P. D. 2006. Episodic growth of the Gondwana
supercontinent from hafnium and oxygen isotopes in zircon. Nature, 439:580–583.
Kemp A. I. S., Hawkesworth C. J., Foster G. L., Paterson B. A., Woodhead J. D., Hergt J. M., Gray C. M.,
Whitehouse M. J. 2007. Magmatic and crustal differentiation history of granitic rocks from Hf–O isotopes in
zircon. Science, 315(5814):980–983.
Klotzli U., Klotzli E., Gunes Z., Kosler J. 2009. Accuracy of laser ablation U–Pb zircon dating: results from a
test using five different reference zircons. Geostand. Geoanal. Res., 33:5–15.
Koglin N., Zeh A., Cabral A. R., Gomes Jr. A. A. S., Corrêa Neto A. V., Brunetto W. J., Galbiatti H. 2014.
Depositional age and sediment source of the auriferous Moeda Formation, Quadrilátero Ferráfero of Minas
Gerais, Brazil: new constraints from U-Pb-Hf isotopes in zircon and xenotime. Precambrian Res., 255:96–
108.
Lana C., Alkmim F. F., Armonstrong R., Scholz R., Romano R., Nalini Jr. H. R. 2013. The ancestry and
magmatic evolution of Archaean TTG rocks of the Quadrilátero Ferrífero province, Southeast Brazil.
Precambrian Res., 231:157–173.
Machado N. & Carneiro M. A. 1992. U–Pb evidence of late Archean tectonothermal activity in the southern São
Francisco shield, Brazil. Can. J. Earth Sci., 29:2341–2346.
Machado N., Noce C. M., Ladeira E. A., Belo de Oliveira O. 1992. U-Pb Geochronology of Archean magmatism
and Proterozoic metamorphism in the Quadrilátero Ferrífero, southern São Francisco craton, Brazil. Geol.
Soc. Am. Bull., 104:1221–1227.
Machado N., Schrank A., Noce C. M., Gauthier G. 1996. Ages of detrital zircon from Archean-Paleoproterozoic
sequences: implications for Greenstone Belt setting evolution of a Transamazonian foreland basin in
Quadrilátero Ferrífero, southeast Brazil. Earth Planet. Sci. Lett., 141:259–276.
Marshak S., Tinkham D., Alkmim F. F., Brueckner H. K., Bornhorst T. 1997. Dome-and-keel provinces formed
during Paleoproterozoic orogenic collapse-Diapir clusters or core complexes? Examples from the
Quadrilátero Ferrífero (Brazil) and the Penokean Orogen (USA). Geology, 25:415–418.
Mendes M. D. C. O., Lobato L. M., Suckau V., Lana C. 2014. Datação U-Pb in situ por LA-ICPMS em zircões
detríticos da Formação Cercadinho, Supergrupo Minas. Geol. USP. Sér. Cient., 14(1):55-68.
Morel M. L. A., Nebel O., Nebel-Jacobsen Y. J., Miller J. S., Vroon P. Z. 2008. Hafnium isotope
characterization of the GJ-1 zircon reference material by solution and la-ser-ablation MC-ICPMS. Chem.
Geol., 255:231-235.
Murakami T., Chakoumakos B. C., Ewing R. C., Lumpkin G. R., Weber W. J. 1991. Alpha-decay event damage
in zircon. Am. Mineral., 76:1510–1532.
Noce C. M., Teixeira W., Machado N. 1997. Geoquímica dos gnaisses TTG e granitoides neoarqueanos do
Complexo Belo Horizonte, Quadrilátero Ferrífero, Minas Gerais. Rev. Bras. Geocienc., 27:25–32.
Noce C. M., Machado N., Teixeira W. 1998. U–Pb geochronology of gneisses and granitoids in the Quadrilátero
Ferrífero (southern São Francisco craton): age constraints for Archean and Paleoproterozoic magmatism and
metamorphism. Rev. Bras. Geocienc., 28:95–102.
Noce C. M., Zuccheti M., Baltazar O. F., Armstrong R., Dantas E., Renger F. E., Lobato L. M. 2005. Age of
felsic volcanism and the role of ancient continental crust in the evolution of the Neoarchean Rio das Velhas

245
 greenstone belt (Quadrilátero Ferrífero, Brazil): U–Pb zircon dating of volcaniclastic greywackes.
Precambrian Res., 141:67–82.
Patchett P. J., Kouvo O., Hedge C. E., Tatsumoto M. 1981. Evolution of continental crust and mantle
heterogeneity: evidence from Hf isotopes. Contrib. Mineral. Petrol., 78:279-297.
Renger F. E., Noce C. M., Romano A. W., Machado N. 1995. Evolução sedimentary do Supergrupo Minas: 500
Ma de registro geológico no Quadrilátero Ferrífero, Minas Gerais, Brasil. Geonomos, 2(1):1–11.
Romano R., Lana C., Alkmim F. F., Stevens G., Armstrong R. 2013. Stabilization of the southern portion of the
São Francisco Craton, SE Brazil, though a long-lived period of potassic magmatism. Precambrian Res.,
224:143–159.
Santos M. M., Lana C. C., Cipriano R. A. S. 2014. Development of zircon standards for U–Pb geochronology by
laser ablation. In: South-American Symposium on Isotope Geology, 9, São Paulo, 296p.
Schrank A. & Machado N., 1996a. Idades U-Pb em monazitas e zircões das Minas de Morro Velho e Passagem
de Mariana, Quadrilátero Ferrífero, MG. In: Congr. Bras. Geol., 39, Anais, 6:470-472.
Schrank A. & Machado N. 1996b. Idades U-Pb em monazitas e zircões do distrito aurífero de Caeté, da Mina de
Cuiabá e do Depósito de Carrapato, Quadrilátero Ferrífero (MG). In: Congr. Bras. Geol., 39, Anais, 6:473-
475.
Segal I., Halicz L., Platzner I. T. 2003. Accurate isotope ratio measurements of ytterbium by multiple collection
inductively coupled plasma mass spectrometry applying erbium and hafnium in an improved double external
normalization procedure. J. Anal. At. Spectrom., 18:1217–1223.
Sláma J., Košler J., Condon D. J., Crowley J. L., Gerdes A., Hanchar J. M., Horstwood M. S. A., Morris G. A.,
Nasdala L., Norberg N., Schaltegger U., Schoene B., Tubrett M. N., Whitehouse M. J. 2008. Plešovice
zircon – a new natural reference material for U–Pb and Hf isotopic microanalysis. Chem. Geol., 249:1–35.
Söderlund U., Patchett J. P., Vervoort J. D., Isachsen C. E. 2004. The 176Lu decay constant determined by Lu-Hf
and U-Pb isotope systematics of Precambrian mafic intrusions. Earth Planet. Sci. Lett., 219:311-324.
Stacey J. S. & Kramers J. D. 1975. Approximation of terrestrial lead isotope evolution by a two-stage model.
Earth Planet. Sci. Lett., 26:207-221.
Taylor S. R. & McLennan S. M. 1985. The continental crust: Its composition and evolution. Blackwell, Oxford,
312p.
Valley J. W. 2003. Oxygen isotopes in zircon. In: J. M. Hanchar & P. W. O. Hoskin (eds.) Zircon. Reviews in
Mineralogy and Geochemistry, Mineralogical Society of America, 343–385.
Van Achterbergh E., Ryan C. G., Jackson S. E., Griffin W. 2001. Data reduction software for LA-ICP-MS. In: P.
Sylvester (ed.) Laser Ablation ICPMS in the Earth Science. Mineralogical Association of Canada, 239–243.
Wedepohl K. H. 1995. The composition of the continental crust. Geochim. Cosmochim. Acta, 59:1217–1232.
Woodhead J. D. & Hergt J. M. 2005. A preliminary appraisal of seven natural zircon reference materials for in
situ Hf isotope determination. Geostand. Geoanal. Res., 29:183–195.
Wu F. Y., Yang Y. H., Xie L. W., Yang J. H., Xu P. 2006. Hf isotopic compositions of the standard zircons and
baddeleyites used in U–Pb geochronology. Chem. Geol., 234:105–126.

246
247