Understanding the evolution of the continental crust is a challenge due to the diversity of geological environments where it forms and to the variety of reworking processes it may have undergone throughout the geological time. Chelogenic cycles1, terrane accretion2, or continental collision are among fundamental processes that allow the preservation of the archives of crustal evolution3,4,5,6. Particularly, terrane accretion is one of the main processes for lateral continental growth through Earth’s history6,7,8,9.

The formation processes for the early Archean tonalite–trondhjemite–granodiorite (TTG) associations is incompatible with the Phanerozoic-style of subduction10. This initial TTG generation was through partial melting of hydrated low-Mg basaltic rocks within the base of a thickened basaltic crust11,15. However, the 3.2 Ga Mesoarchean to 2.3 Ga Paleoproterozoic continental crust may represent a transition period from an early non-plate tectonic mode to modern-style plate tectonics by accreted oceanic arcs and oceanic plateaus, mainly through ultrahigh-temperature processes12,13,14,15. Therefore, the preservation of Meso- to Neoarchean felsic continents may represent the initiation of plate tectonics in some form15,16. In this debate, the application of a geodynamic unifying model or the reconciliation of different models for the ancient continents generation is still in dispute15,16,17. However, it appears that there was a shift from the Archean continental crust produced by accretion and lithospheric peeling processes to Proterozoic continental crust generated by magmatic arcs18,19,20,21. At the center of this debate is the mechanical behavior of subsiding crust during the Archean and its lifetime, and how the transition to continental arcs and Phanerozoic-style subduction took place18,21. Some studies suggest long time scales (3.2 to 2.5 Ga) for a profound change in average crustal chemistry22. Gradual decrease in the rate of crust generation may be explained by the secular cooling of the mantle23, and the decline in crustal reworking may be explained by the “cratonization” of continental crust4.

Compositional diversity and complex evolution of the accretionary orogens are related to the plate boundary parallel migration, and orthogonal accretion of juvenile and reworked crustal segments9. In this context, Sm-Nd isotopes may provide a mean for determining (1) the crustal residence time24,25, (2) crustal reworking processes26, and (3) mantle mixing27. Therefore, Nd isotopes allow the characterization of protolith sources as a way to describe the geometry and direction of continental crust growth24,28.

In this study, we show evidence of continental growth via terrane accretion within the Campo Grande Block of the Borborema Province, NE Brazil. Using petrographic mapping, and spatial distribution of coupled U-Pb zircon ages and Sm-Nd isotopic data, we show that repetitive accretion of crustal terranes occurred within this area from the late Archean to the Neoproterozoic.

Regional Geology

The Borborema Province is a Precambrian shield29,30,31,32,33 within the north-northeastern part of the South American continent30,31 (Fig. 1A). It is formed of discontinuous remnants of Archean crust, Paleoproterozoic migmatitic gneiss complexes, and Meso- to Neoproterozoic supracrustal rocks29,31,32. The Paleoproterozoic complexes comprise the 2.2-2.0 Ga gneiss-migmatite basement of Neoproterozoic supracrustal sequences and granite intrusions34,36. These high-grade gneisses and anatectic domes may be related to the 2.25-1.98 Ga Eburnian Orogeny30,35,36. The final configuration of the Borborema Province resulted from the diachronic convergence of the West African, Amazonian and São Francisco-Congo cratons during the Neoproterozoic Brasiliano/Pan-African orogeny33,35 (Fig. 1A).

Figure 1
figure 1

Regional geological setting. (A) Localization map of the Borborema Province in West Gondwana. (B) Geological map of the central portion of the Rio Grande do Norte domain. (C) U-Pb zircon age distribution and (D) ternary gamma-spectrometric map of the Caicó-São Vicente, Lajes, Antônio Martins and Campo Grande-Itajá regions in which the Rio Grande do Norte basement is exposed29,31,36,44. Note that the Campo Grande-Itajá area represents the unique basement dome in the Rio Grande do Norte domain (D). Legend: RPC - Rio de La Plata Craton, SFC - São Francisco Craton, SLC – São Luiz Craton, TC - Tanzania Craton. PoL - Portalegre Lineament, PJCSZ – Picuí-João Câmara shear zone, PaL - Patos Lineament, ADL - Adamaoua Lineament. JD - Jaguaribe domain, RGND - Rio Grande do Norte domain, and SJCM - São José do Campestre massif.

The Rio Grande do Norte domain (RGND; Fig. 1B), the northeastern portion of the Borborema Province, is limited westwards by the NE-trending rectilinear Portalegre dextral strike-slip shear zone and by the Patos-Adamaoua EW-trending shear zone at the southern boundary29,31,34. Several shear zones represent local adjustments within each terrain, as well as divide the RGND into four high-grade migmatite-gneiss blocks (e.g., Caicó, Lajes, Antônio Martins and Campo Grande-Itajá; Fig. 1C, D). Zircon U-Pb ages indicate that Rhyacian (2.25 to 2.15 Ga) metamorphic high-K calc-alkaline magmatic rocks37 and supracrustal rocks form the basement of the Neoproterozoic Seridó Group32.

Result and Discussion


The Campo Grande Block is a small crustal fragment, 1,500 km2 in area, with dome to ellipsoidal geometry, SSW-NNE axis, exposed in the central portion of the Rio Grande do Norte domain, around Campo Grande town (Fig. 1B, C). The CGB consists of an Archean tonalitic to granitic migmatite complex and mafic-ultramafic rocks in the core, rimmed by Paleoproterozoic alkaline orthogneisses, surrounded by an outer rim of Neoproterozoic K-feldspar-rich granite intrusions (e.g., Caraúbas granite). The block shows intense deformation, with coaxial refolding, pervasive foliation, and north-northeast trending shear zone systems38,39. The Campo Grande-Itajá region represents a unique basement dome in the Rio Grande do Norte domain (Fig. 1D). The migmatites in the central area display higher Th and K concentrations (Fig. 1D), followed by an abrupt reduction of these elements in the inner rim orthogneiss, and again high contents in the outer rim granite, reflecting distinct geological compartments from west to east. In addition, based on integrated analysis of structural pattern, ternary gamma-spectrometric map (Fig. 1D) and thorium anomaly map, we suggest that shear zone systems define major terrane boundaries. The Portalegre Lineament corresponds to a 20–40 km wide shear zone that separates the Rio Grande do Norte and Jaguaribe domains (Fig. 1C). The Paraú Lineament divides the west part of the Rio Grande do Norte domain into the distinct eastern Itajá and western Campo Grande blocks.

The Campo Grande Block consists of migmatitic gneisses that display multiple phases of partial melting38. These migmatites comprise Archean tonalitic gneisses that contain granitic Proterozoic leucosomes and alkali granite dikes. The mafic-ultramafic rocks comprise amphibolites and pyroxenites that are present as boudinaged bodies within the Archean migmatitic complex, which are further oriented parallel to the leucosomal layers of the host migmatites39. The overall outcrop pattern suggests that these mafic-ultramafic rocks were originally emplaced as dykes, intruding the host migmatitic gneisses. The ultramafic pyroxenites show relict cumulate texture, and re-equilibration to cummingtonite-grunerite-rich rocks, with varying proportions of chlorite, serpentine and magnetite. Amphibolites comprise massive poikiloblastic garnet and granoblastic amphibole with variable proportions of plagioclase + clinopyroxene in symplectitic texture, typical of retrograded high-pressure rocks39. The Itajá Block is composed of Paleoproterozoic K-feldspar-rich orthogneiss, and wehrlite intrusions that occur as elongated boudins (<100 m) in the host orthogneiss; minor amphibolite and supracrustal rocks also appear. Neoproterozoic pegmatite and alkaline granite intrusions make up almost 20–30% of both blocks.

Spatial Pattern of Ages based on the Nd Evidence for Diachronous Crustal Accretion

The evolution of the Campo Grande Block involves at least seven thermal-tectonic events (Supplementary Table 1). The first magmatic event remains recorded in 2.98 to 2.91 Ga old tonalitic paleosome (Fig. 2A), which constitutes the central core of the block. All zircon crystals from tonalite samples are prismatic (100 to 300 μm), with Th/U ratios from 0.125 to 0.583 and internal zonation (Fig. 2A), all typical features of magmatic crystals40. The 2.9 Ga calc-alkaline magma represents a rare record of this age41, particularly in West Gondwana42,43. Inherited zircon cores of 3311 ± 52 Ma suggest a Paleoarchean crust as protolith source for the 2.9 Ga magmatism. The second partial melting event is represented by 2.65 Ga alkaline leucosome (e.g., ADE-23 sample) with thick K-feldspar-rich layers from the central portion of the strongly migmatized Archean core. Forty-four prismatic zircon crystals from this sample yield a Neoarchean Discordia age of 2651 ± 19 Ma. The 2.0 Ga and ca. 600 Ma zircon cores and rims are recorded in the migmatites. For example, the ADE-12 granitic migmatite sample yielded only 2.0 Ga prismatic zircon grains, while most of ca. 600 Ma Neoproterozoic ages are obtained in the overgrowth rims from the Archean migmatite zircon cores.

Figure 2
figure 2

(AF) Histograms of U-Pb zircon ages of the Campo Grande Block separated by different rock type with cathodoluminescence images of representative zircon grains (Data from supplementary Table). (G,H) Histograms of U-Pb zircon ages of the Itajá Block separated by different rock type with cathodoluminescence images of representative zircon grains (Data from supplementary Table 2).

The clinopyroxene-garnet amphibolite lenses show the same 2.69-2.65 Ga age range, interpreted as the crystallization age of the protolith, also based on internal zonation, morphology and high Th/U ratio in zircon cores (Fig. 2B). The well-rounded (50 to 100 μm) zircon grains from amphibolite samples show zonation from core to rim, with well-defined rims, showing low to very bright luminescence (Fig. 2B), therefore indicating subsequent resorption and recrystallization40. The amphibolite samples also have 2.0 Ga well-rounded zircon crystals with zoned cores followed by outermost CL-bright overgrowths (Fig. 2B) possibly due to a subsequent event. The ca. 600 Ma Neoproterozoic homogeneous zircon grains, without internal zonation (Fig. 2B), are also recorded in the amphibolites. This confirms that the 2.65 Ga tholeiitic intrusions were subsequently torn apart during 2.0 Ga and 600 Ma tectonothermal events and are now present as isolated boudins. The 2.9 Ga inherited zircon grains were captured or assimilated by 2.65 Ga tholeiitic magma during its ascent and emplacement in the 2.9 Ga host tonalite basement39. Furthermore, as there are no fine-grained felsic veins intruding the amphibolite, it is unlikely that the 2.65 Ga zircon grains obtained in the mafic lenses came from other sources39.

The ultramafic pyroxenite lenses contain 2.7-2.6 Ga Archean and 2.3 Ga Paleoproterozoic zircon crystals with distinct morphological features like oscillatory zoning (Fig. 2C), typical of magmatic zircon39. Furthermore, pyroxenite samples present 2.0 Ga and ca. 600 Ma zircon grains that show varying degrees of rounding or absorbed borders (Fig. 2C), similar to features described in zircon grains from ultramafic rocks in high-grade metamorphic terrains44,45. The supracrustal rocks, such as garnet-biotite gneiss (AT-23 sample), bear 2.7-2.6 Ga zircon cores, suggesting that Neoarchean tonalite and tholeiitic rocks were the main provenance (Fig. 2D). The majority of these Archean zircon cores from the supracrustal sample display 2.2 to 2.0 Ga overgrowth rims (Fig. 2D). Besides, a few 2.46 to 2.44 Ga old zircon grains obtained in the Archean tonalite and supracrustal rocks suggest restricted Siderian magmatism.

In the eastern portion of the Campo Grande Block, the K-feldspar-rich alkali granite magmatism of 2.23-2.18 Ga Rhyacian age generated a large volume of magmatic rocks (Fig. 2E). Lastly, the K-feldspar-rich (20–30%) granitic plutons make up the western limit of the study area. The feldspar crystals develop a strong foliation parallel to the transcurrent shear zone. Neoproterozoic granites emplaced along the Portalegre shear zone (ADE-13 sample) have elongated prismatic zircon grains (3:1) that yield a crystallization age of 604 ± 12 Ma (Fig. 2F). On the other hand, granitic intrusions sampled in the central portion of the Campo Grande Block (e.g., AT-23 sample) show prismatic zircon crystals (2:1) crystallized at 566 Ma (Supplementary Table 1; Fig. 2F).

The TDM model ages and εNd(t) values of migmatite samples support a complex history for the study area (Table 1, and Fig. 3A to H). The 2.9 Ga tonalitic migmatite displays positive and negative εNd(t) values of -3.9 to +4.8 with TDM model ages between 3.3 and 2.7 Ga, suggesting juvenile sources and crustal reworking at 2.9 Ga (Fig. 3B). All these Archean rocks are concentrated in the core of the structural dome of the Campo Grande Block. The 2.65 Ga and 2.0 Ga old alkaline granitic migmatites have negative εNd(t) values (−5.47 to −2.74) and younger TDM model ages between 2.8 and 2.4 Ga. The 2.65 Ga old amphibolites display negative εNd(t) values (−1.03 to −7.97) with older TDM model ages (3.7 to 3.3 Ga) and positive εNd(t) values (+1.97 to +8.17) with younger TDM model ages of 2.0 to 2.65 Ga, supporting a Neoarchean juvenile source (Fig. 3A) and contamination of crustal material.

Table 1 Nd isotope data and U-Pb zircon age for the Campo Grande and Itajá blocks.
Figure 3
figure 3

(AF) εNd(t) versus U-Pb zircon age from the major rock-types for the Campo Grande and Itajá blocks (Data from Table 1), Northeast Brazil. Gray ellipse - magmatic age, green ellipse - metamorphic age. (G) Histogram of TDM model age for the Campo Grande and Itajá blocks. (H) Schematic model of continental accretion for the Campo Grande and Itajá blocks.

The pyroxenites display heterogeneous Nd isotopic data (Table 1). The 2.6 Ga old pyroxenite samples display positive and negative εNd(t) values with TDM model ages between 2.6 and 3.2 Ga, whereas 2.3 Ga old pyroxenites show positive εNd(t) values with restrict TDM model ages of 2.29-2.37 Ga (Fig. 3E). The younger TDM model ages of 1.4 and 2.0 Ga with strongly negative εNd(t) values may suggest metamorphic alteration in the Sm-Nd isotopic system during Proterozoic times (Figs. 3E, 2C). Furthermore, we suggest that the negative values of εNd(t) and older TDM for the 2.65 Ga ultramafic rocks may reflect enriched sources or crustal assimilation. The supracrustal protoliths have TDM model ages of 3.6 to 2.6 Ga with positive to negative εNd(t) values for the 2.65 Ga crystallization age and negative εNd(t) values (Fig. 3D) during Paleoproterozoic events. The 2.2 Ga K-feldspar-bearing augen orthogneisses display a Nd isotopic signature characterized by negative (-8.0) to positive (+5.0) εNd(t) values and TDM between 2.3 and 3.1 Ga (Fig. 3C), indicating a Rhyacian calc-alkaline magmatism with crustal reworking and juvenile sources contributions. Lastly, the Neoproterozoic granites present strongly negative εNd(t) values (-20.57 and -14.25) with relatively younger TDM model ages of 2.10 and 2.39 Ga (Table 1).

TDM model ages and εNd(t) values support a complex history for the Campo Grande Block (Fig. 3A–H). Whole-rock Nd isotope results indicate that the isotope system preserved the protolith source signature despite of crustal reworking and high-grade metamorphic events that affected the Archean core. Paleoproterozoic ages appear in the 1.95 Ga granitic leucosome generation and 2.0 Ga metamorphic overgrowth zircon rims on Neoarchean zircon cores from the ultramafic and supracrustal protolith rocks inside the Archean core. The Rhyacian orthogneisses from the eastern portion and 2.0 Ga granitic leucosome from the Archean central portions display similar TDM model ages and εNd(t), meaning that both K-feldspar-rich alkaline magmatism and crustal anatexis have similar sources. Nevertheless, crustal reworking was intense in the eastern block area, practically obliterating the Archean protolith record. A second high-grade metamorphic event - the seventh recorded event – is indicated by 614-593 Ma old zircon grains and rims around the Archean zircon cores from the amphibolite samples39. Moreover, 604 Ma old K-feldspar-rich granitic intrusions and 566 Ma pegmatite veins suggest a more restricted Neoproterozoic partial melting when compared to the large volume of neosome generated during the Rhyacian. The Neoproterozoic granite intrusions and alkaline leucosome samples have strongly negative εNd(t) values (−20.57 and −14.25) and relatively younger TDM ages of 2.10 and 2.39 Ga. These Nd isotope results suggest that the Paleoproterozoic crust is the main protolith source for the Neoproterozoic alkali granitic magmatism. That is, on the outermost overgrowths of the Archean dome the reworking process is dominant when compared to the core (Fig. 3G). The progressive decrease in TDM model ages from the core (3.7 Ga) towards the margins (2.1 Ga) of the block, integrated with structural, thorium anomaly map, and U-Pb zircon age patterns suggest accretionary processes for the continental growth (Fig. 3H). Thus, Nd isotope evolution reflects the crustal growth from the Archean core protolith, following extensive Paleoprotezoic juvenile accretion and reworking, as well as Neoproterozoic crustal magmatism at the outer rim.

In contrast, the Itajá Block only records two events of magma generation (Supplementary Table 2). The first event is represented by orthogneisses that were formed at 2.23 Ga (Fig. 2H), displaying negative to weakly positive εNd(t) values (Fig. 3F) and TDM model ages between 2.2 and 2.7 Ga (Table 1). Clinopyroxenites and wehrlites, crystallized at 2.19 Ga (Fig. 2G), with positive εNd(t) values (Fig. 3F), intruded these orthogneisses, indicating juvenile tholeiitic magmatism. Therefore, alkali granitic and ultramafic magmatism took place in a short time interval of ~40 Ma (2.23 to 2.19 Ga), similar to the reported events in the Lajes Block44, which is exposed 40 km eastwards, separated from the Itajá Block by the Neoproterozoic Seridó intracontinental fold belt (Fig. 1C,D). Furthermore, inherited zircon grains of Siderian age (ca. 2.32 Ga) are recorded in the host orthogneiss from the Itajá area. The intense Rhyacian reworking obliterated the possible older sources (Fig. 2F,G). Therefore, a genetic correlation with the Archean core of the Campo Grande Block is unclear (Fig. 3G). Nevertheless, it is indisputable that the protolith sources are dominantly Neoarchean, as suggested for the Lajes Block45.

Crustal Reworking and Terrain Docking

The integration of all Nd isotope and U-Pb zircon age patterns allowed the establishment of limits and genetic correlations between the crustal fragments that form the Campo Grande and Itajá blocks (Fig. 4A–D). Our results support that 2.9 Ga and 2.7-2.6 Ga Archean crustal reworking and minor 2.2 Ga Paleoproterozoic juvenile mantle were the primary sources for the continental growth through accretionary mechanisms5,15,16,19,45,46. The first rim around the Archean core seems to engulf the core migmatites in a circular shape (Fig. 4A–D). This geometry is feasible via a 2.9 Ga domal fashion of tonalitic magmatism that engulfed the Archean core. However, the subsequent events may have occurred due to terrane accretionary mechanisms. Therefore, our results may indicate a change in the mechanism of continental evolution, namely dome formation at 2.9 Ga to terrane accretion starting at 2.7 Ga.

Figure 4
figure 4

(A) Simplified geological and (B) Thorium anomaly map of the Campo Grande and Itajá blocks and adjacent areas. (C) U-Pb zircon age distribution of the Campo Grande and Itajá blocks. (D) Plot of crystallization ages and TDM model ages of the Campo Grande and Itajá blocks (Data from Table 1). (E,F) Histograms of U-Pb zircon age of Campo Grande and Itajá blocks (Data from supplementary Table 1 and 2), Northeast Brazil.

Based on the frequency histogram of the U-Pb zircon ages and the area mapped, it is suggested that at least 30–40% of the Campo Grande Block was already formed at 2.9 Ga (Fig. 3G, H). After 2.9 Ga, there was an increase in the rate of continental crust growth, probably due to subduction-like processes and peeling-off driven convergent settings12,13,20. Therefore, the accretionary orogenic collage derived from a complex diversity of protolith sources47,48, as described in this study (Fig. 4A–F). That is, the continental evolution is complex and includes several components of different scale, composition, and age10,15,47,48.

Thermal and compositional contrasts between continental and oceanic lithosphere lead to subsidence processes by plate tectonics49,50. The subsidence of oceanic crust allowed the efficient mechanical coupling of the microcontinents and remnant magmatic arcs in the orogenic wedge2,3. In this scenario, magmatic arc formation is probably the most important mechanism to maintain the continental crust reservoir18,19. Paleoproterozoic 2.25-2.18 Ga high-K calc-alkaline magmatism may represent a thermal weakening zone that allowed the reworking and juvenile magmatism11,18. In the Borborema Province, Paleoproterozoic arc magmatism represents a more significant period of crustal growth within the South American continent19, similar to the study area. Thus, terrain accretion and partial melting mainly in the root of the magmatic arc setting from 2.2 Ga promote the differentiation and growth of the continental crust5,15,51.

The preservation of the felsic continental block between 2.9 to 2.2 Ga in the Borborema Province may mark the transition and initiation of plate tectonics, implying a higher consumption of mafic crust during Proterozoic physical mechanisms of accretion compared to late Archean processes. One possibility would be crustal reworking via lower mafic crustal peeling-off (e.g. delamination) during continent-continent convergence15,16. Despite the significant increase in isotopic studies, late Archean reworking and recycling processes remain largely unknown15,16. Therefore, a different style of plate tectonics and subduction possibly occurred during the early Archean, with transitional physical mechanisms between the late Archean and the Phanerozoic-style. However, any model that calls upon fractionation of a single magmatic event or process to produce continental crust is unrealistic51.


Nd isotope data and U-Pb geochronology within the distinct terrains provide constraints for the succession of magmatic and metamorphic phases that resulted in continental accretion of heterogeneous rocks from 2.9 Ga to ca. 566 Ma ago in northeast Brazil. These led to the assembly of the Rio Grande do Norte domain. The Campo Grande Block represents high-grade metamorphic terrains with multiple partial melting, meta-ultramafic, and metamafic lenses that record polyphase metamorphism, magmatism, and intense shearing. Our data bear evidence that the distribution and nature of the continental crust reflect the secondary processes of reworking. The age succession associated with the geochemical patterns of the Precambrian evolution of the Campo Grande Block highlights the importance of the accretionary dynamics for the continental growth. The accretionary process is cyclic and repeated in space and time, allowing the continental growth to start by Mesoarchean to Neoarchean crustal peeling-off driven lithospheric convergence to Proterozoic magmatic arc accretion. When the events ended at the Neoproterozoic (ca. 566 Ma), the Archean to Paleoproterozoic Campo Grande and Rhyacian Itajá complexes amalgamation in the center of West Gondwana was concluded.


Geological Mapping and Petrography

Geological mapping was undertaken in the Campo Grande area with the purpose of investigating the gneiss-migmatite complex. Geological mapping was supported by geochemical, geophysical and petrographic investigations. Systematic thin sections cut relative to foliation were obtained from representative samples from outcrops of migmatite, orthogneiss, ultramafic and supracrustal rocks. The petrography was done at the Microscopy Laboratory of the Institute of Geosciences of Universidade de Brasília (Brazil).

U-Pb isotopes

Zircon grains from samples were separated by conventional procedures and magnetic separator after concentration by hand panning. U-Pb isotopic analyses were performed on zircon grains using a Thermo-Fisher Neptune High Resolution Multicollector Inductively Coupled Plasma Mass Spectrometer (HR-MC-ICP-MS) coupled with a Nd:YAG UP213 New Wave laser ablation system at the Laboratory of Geochronology of Universidade de Brasília. U-Pb analyses on zircon grains were carried out by the standard-sample bracketing method52, using the GJ-1 standard zircon53 in order to quantify the amount of ICP-MS fractionation. The tuned masses were 238, 207, 206, 204 and 202. The integration time was 1 second and the ablation time was 40 seconds. A 30 µm spot size was used and the laser setting was 10 Hz and 2-3 J/cm2. Two to four unknown grains were analyzed between GJ-1 analyses. 206Pb/207Pb and 206Pb/238U ratios were time corrected. The raw data were processed off-line and reduced using an Excel worksheet54. During the analytical sessions, the zircon standard 9150055 was also analyzed as an external standard.

Common 204Pb was monitored using the 202Hg and (204Hg + 204Pb) masses. Common Pb corrections were not done due to very low signals of 204Pb (<30 cps) and high 206Pb/204Pb ratios. Reported errors are propagated by quadratic addition [(2SD2 + 2SE2)1/2] (SD = standard deviation; SE = standard error) of external reproducibility and within-run precision. External reproducibility is represented by the standard deviation obtained from repeated analyses (~1.1% for 207Pb/206Pb and up to ~2% for 206Pb/238U) of the GJ-1 zircon standard during the analytical sessions, and the within-run precision is the standard error calculated for each analysis. Concordia diagrams (2σ error ellipses), probability density plots and weighted average ages were calculated using the Isoplot-3/Ex software56.

Sm-Nd Isotopes

Sm–Nd isotopic analyses followed the method described by Gioia and Pimentel (2000)57 and were also carried out at the Geochronology Laboratory of Universidade de Brasília. Whole-rock powders (~50 mg) of 60 samples were mixed with 149Sm–150Nd spike solution and dissolved in Savillex Digestion Vessels. Sm and Nd extraction of whole-rock samples followed conventional cation exchange chromatography techniques, with Teflon columns containing LN-Spec resin (HDEHP – diethylhexil phosphoric acid supported on PTFE powder). Sm and Nd fractions were loaded on Re evaporation filaments of double filament assemblies, and the isotopic measurements were carried out on a multicollector TRITON thermal ionization mass spectrometer in static mode. Uncertainties of Sm/Nd and 143Nd/144Nd ratios were better than ±0.1% (2 σ standard error) and ±0.0015% (1σ), respectively, according to repeated analyses of the international rock standard BHVO-1. 143Nd/144Nd ratios were normalized to 146Nd/144Nd = 0.7219, and the decay constant used was 6.54 ×10−12. The TDM values were calculated using the DePaolo (1981) model24.