Archaean continental crust formed from mafic cumulates

The study addresses how Archean tonalite–trondhjemite–granodiorite (TTG) crust formed and what source rocks and processes produced it. TTGs are central to early continental growth and cratonisation, yet their petrogenesis, tectonic setting, and sources remain debated. Traditional models involve hydrous, LILE-enriched mafic sources that melted at varying depths, with high La/Yb, Sr/Y and positive Eu anomalies often interpreted as signatures of deep, garnet-stable melting akin to subduction zones. However, analogies with modern arc granitoids are disputed, and alternative settings include melting of underplated/over-accreted lower crust, thickened plateau-like crust, or material in crustal downwellings. The authors seek empirical constraints on primary TTG melt compositions using HFSE systematics (Nb and Ti anomaly), to disentangle effects of melting, residual phases, and magmatic differentiation, and to reassess TTG sources and geodynamic setting.

Prior work suggested TTGs could be generated by dehydration melting of basaltic crust at high pressure where garnet is stable and plagioclase is absent, implying a subduction-like setting and marking potential onset of modern plate tectonics. Other studies propose intracrustal processes: melting of underplated lower crust, thickened arcs or plateaus, crustal drips, or hybridized mantle–crust sources. Thermodynamic modeling has challenged simple depth proxies (La/Yb, Sr/Y, Eu anomalies), showing high values can occur at lower pressures depending on source composition, redox, and water content. Alternative sources include amphibolites of ambiguous protolith, mafic cumulates (gabbro/gabbro-norite), bimineralic eclogites from lower crust–picrite interaction, and metasomatized mantle peridotites. Observationally, TTGs often show strong crystal fractionation and evidence of mineral segregation and melt loss, complicating reconstruction of primary melt signatures.

The study constrains primary TTG melt compositions using high field-strength element (HFSE) systematics, focusing on Nb concentration and the Ti anomaly (Ti^N/Ti*; N denotes Primitive Mantle normalization; Ti* ≈ sqrt(Zr^N·Gd^N)). Conceptual basis: during melting of mafic sources at 1.0–1.8 GPa and 800–950 °C, titaniferous residual phases (rutile vs. ilmenite) lower Ti^N/Ti* but affect Nb differently due to contrasting partition coefficients (D^Nb_rutile >> D^Nb_ilmenite). Residual rutile yields Nb-poor, low Ti^N/Ti* melts; ilmenite yields more Nb-rich melts at similar Ti depletion. Titanite is not stable at these P–T in mafic systems. They use global granitoid datasets (GEOROC) to build a reference “Granitoid End Member” (GEM) trend and compare TTGs against modern granitoids, orogenic and Andean granites/granodiorites, adakites, etc. They evaluate effects of fractional crystallization and assimilation: modeling amphibole, plagioclase, clinopyroxene, and common mineral fractionation vectors on Nb–Ti^N/Ti* space, identifying that amphibole fractionation most strongly raises Nb and lowers Ti^N/Ti*. They define a fractionation factor F to parameterize cumulative magmatic differentiation independent of initial Ti^N/Ti* or degree of melting: F = 10·(Nb_x − Nb_GEM)/( (Ti^N/Ti*)_x − (Ti^N/Ti*)_GEM ), using GEM values Nb_GEM = 8 ppm and (Ti^N/Ti*)_GEM = 1. Partial melting and fractional crystallization are quantified via standard batch and fractional crystallization equations with mineral/melt partition coefficients compiled for HFSE and REE in relevant phases and lithologies (basaltic and andesitic–dacitic systems). They perform modal and non-modal melt modeling for candidate sources (basalt, eclogite, metagabbro, gabbro-norite, plagioclase-rich mafic cumulates), testing degrees of melting (down to ~10%) and residual assemblages (garnet, clinopyroxene, amphibole, rutile; ilmenite where relevant). They interrogate whether observed REE systematics of low-F TTGs (high La/Sm, Sm/Yb, Sr/Y, Eu/Eu*) can arise solely from melting/fractionation of MORB-like sources versus requiring fractionated source compositions. Data normalization to Primitive Mantle is used for multi-element patterns and Ti anomaly; REE ratios and Eu/Eu* computed conventionally. They assess CaO, Na2O, Sr correlations to test plagioclase accumulation/assimilation hypotheses and use case studies (e.g., Pilbara Craton) to illustrate fractionation bias. Data sources are primarily GEOROC; additional partition coefficients are compiled in Supplementary Data 1–3.

• TTG compositions in Nb–Ti^N/Ti* space are best explained by melting in the presence of residual rutile; ilmenite as a significant residual phase is ruled out because TTGs are Nb-poor with high (Ti^N/Ti*)/Nb at low Nb.
• Primary TTG melts are relatively poor in incompatible elements and characterized by variably high La/Sm, Sm/Yb, Sr/Y, and positive Eu anomalies (Eu/Eu* > 1). Differences in these parameters reflect degree of melting and subsequent fractional crystallization, not melting depth.
• Fractional crystallization of amphibole and plagioclase is the dominant control on TTG differentiation trends: amphibole fractionation increases Nb and lowers Ti^N/Ti*, while plagioclase (and sanidine in evolved stages) lowers Sr/Y and Eu/Eu* and raises Ba/K. Low-F TTGs show low incompatible-element concentrations with large Sr and Pb excesses and the highest La/Sm, Sm/Yb, Sr/Y, Eu/Eu*.
• Low-F TTGs display fractionated REE patterns and positive Eu anomalies that cannot be produced solely by melting of MORB-like basaltic sources at realistic degrees of melting (e.g., 10% melting yields only moderate La/Sm ~3, Sm/Yb ~4, Eu/Eu* ~1.2). Thus, source rocks must themselves have fractionated REE and positive Eu anomalies.
• The only source lithologies that satisfy these constraints are mafic plagioclase-cumulate rocks (gabbro/gabbro-norite), transformed to hydrous eclogitic assemblages (garnet + clinopyroxene + amphibole + rutile) at depth. Hydrous contents (~2.3 wt% H2O on average in gabbros) enable melting at estimated P–T conditions.
• Primary TTG melt major-element ranges reflect degree of melting from a common rutile-bearing GEM-like source: at high degrees of melting, ca. 62 wt% SiO2, ~2 wt% MgO, ~6 wt% CaO; at low degrees, ca. 75 wt% SiO2, <0.2 wt% MgO, ~1 wt% CaO.
• TTG melting occurs at ~1.0–1.8 GPa and 800–950 °C, with residual rutile stable at ≥1.4 GPa indicated by high (Ti^N/Ti*)/Nb. Nb/Ta in low-F TTGs spans ~5–45, consistent with residual rutile and progressive source depletion.
• Apparent secular changes in REE signatures are likely preservation/analytical biases: younger TTGs more frequently preserve low-F, less-fractionated compositions.
• Geodynamic model: Volcanic resurfacing over mafic proto-continents thickens and buries mafic cumulate lower crust, warming the Moho to near 900 °C along ~20 °C/km geotherms, inducing melting of hydrous metagabbros to form TTGs. Extensive melt extraction leaves dense eclogitic residues that delaminate, triggering short-lived mantle-derived sanukitoid magmatism and, upon thermal relaxation, high-K granites (“granite blooms”). This provides an endogenic mechanism for Archean crustal growth without requiring subduction onset or impacts.

The findings resolve key ambiguities in TTG genesis by showing that primary TTG melts carry a residual rutile signature and originate from melting of mafic plagioclase cumulates rather than hydrous basalt or peridotite sources. High La/Sm, Sm/Yb, Sr/Y, and Eu/Eu* in primary TTGs are sourced from fractionated cumulate protoliths and amplified by low degrees of melting; subsequent amphibole–plagioclase fractionation modulates these signatures. Thus, REE ratios commonly used as depth indicators do not uniquely reflect melting depth; many primary TTGs formed at moderate pressures (≥1.4 GPa for rutile stability) with signatures governed by source composition and melting degree. The Nb–Ti anomaly systematics effectively distinguish residual rutile vs. ilmenite, constraining P–T conditions and ruling out ilmenite-dominated residues. Geodynamically, TTG formation is linked to intracrustal processes in proto-continental roots: burial by volcanic resurfacing elevates temperatures at the Moho sufficiently to melt hydrated metagabbroic cumulates. The subsequent delamination of eclogitic residues explains the temporal association of TTGs with sanukitoids and later high-K granites and accounts for observed isotopic transitions (e.g., δ18O and Hf–Sr systems). Collectively, the results argue against a necessary subduction origin for Archean TTGs and support an endogenic, intracrustal differentiation pathway for early continental growth and cratonisation.

This study demonstrates that Archean TTGs predominantly formed by melting of rutile- and garnet-bearing mafic plagioclase cumulates in proto-continental roots at ~1.0–1.8 GPa and 800–950 °C. HFSE (Nb and Ti anomaly) systematics, coupled with REE and major-element trends parameterized by a fractionation factor F, show that amphibole and plagioclase fractionation governs TTG differentiation, while primary high La/Sm, Sm/Yb, Sr/Y, and positive Eu anomalies reflect cumulate-source signatures and melting degree rather than depth. Ilmenite is excluded as a significant residual phase. The petrogenetic chain—TTG melting, residue delamination, sanukitoid magmatism, and subsequent high-K granites—provides an internally driven mechanism for Archean crustal growth and the evolution toward andesitic continental crust without invoking external triggers or the onset of global plate tectonics. Future work could refine constraints on source variability, water contents, and P–T–fO2 conditions via integrated HFSE–REE partitioning experiments, isotopic tracers, and targeted studies of minimally fractionated (low-F) TTGs and their cumulate source remnants.

• Many TTGs are strongly affected by crystal fractionation, mineral segregation, and possible assimilation, complicating reconstruction of primary melt compositions and potentially biasing datasets toward evolved signatures.
• Preservation bias likely favors younger, better-preserved TTGs, increasing the apparent incidence of low-F examples; older records may be underrepresented.
• Distinguishing cumulate-derived from basalt-derived sources can be subtle, and differentiation may obscure primary signatures (e.g., Pilbara case), challenging source identification in individual terranes.
• Modeling relies on compiled partition coefficients and assumptions about modal vs. non-modal melting, mineral assemblages, and reference endmembers (GEM), which introduce uncertainties.
• The geodynamic scenario (volcanic resurfacing, burial, delamination) is inferential and supported by thermodynamic plausibility and analogues rather than direct observation for Archean events.