Genesis of Hawaiian lavas by crystallization of picritic magma in the deep mantle

The petrogenesis of Hawaiian lavas is debated, with models required to match both major and trace element characteristics. Considering a lithospheric thickness of ~100–110 km beneath Hawaii, the study evaluates existing hypotheses using high-pressure (3–4 GPa) experimental data compared with Hawaiian lava compositions. Trace element signatures (incompatible element enrichment and garnet-compatible element depletion) indicate low-degree (~5 wt%) melting of garnet lherzolite, but such melts are too low in FeO and SiO2 and too high in CaO and Al2O3 to be parental to Hawaiian lavas. Re-equilibration with harzburgite can raise SiO2 but does not reconcile FeO, Al2O3, and CaO. Mixing melts from garnet lherzolite and silica-saturated pyroxenite produced by recycled oceanic crust explains high SiO2 and NiO and low CaO but not Al2O3 and FeOT. Many past studies assumed primary magma compositions remain unchanged during mantle-to-crust transit, which may be incorrect. Motivated by pyroxenite xenoliths and experiments on Kilauea-like compositions, the authors hypothesize deep-lithosphere fractionation followed by lithospheric re-equilibration as key steps in generating Hawaiian lavas.

• Low-degree partial melting of garnet lherzolite (~5 wt%) fits trace element patterns (incompatible enrichment, garnet signature) but fails to match major elements (too low FeO, SiO2; too high CaO, Al2O3).
• Source heterogeneity from recycled oceanic crust is widely accepted; mixing peridotite melts with melts of silica-saturated pyroxenite (formed by reaction with recycled crust) explains high SiO2, NiO, and low CaO but still misfits Al2O3 and FeOT.
• Prior assumptions of unchanged primary magma during ascent are challenged by evidence for deep lithospheric fractionation and lack of equilibrium with garnet lherzolite on the liquidus; parental magmas are co-saturated with olivine and orthopyroxene at ~1.4 GPa and 1425 °C in earlier studies.
• Pyroxenite xenoliths and high-pressure phase equilibria suggest fractionation in deep mantle chambers and subsequent melt-rock reaction as plausible processes for Hawaiian magma evolution.

High-pressure crystallization experiments were conducted to test the proposed three-step model. A synthetic starting glass (P19), compositionally similar to a partial melt of peridotite at 3.0 GPa and 1540 °C (Walter 1998 Run #30.14), was prepared from oxide and carbonate mixes (in natural basalt BCR-2) and preconditioned (melted at 1650 °C, quenched, ground to <5 μm; CCO-buffered at 900 °C; final drying). FTIR indicates ~0.33 wt% H2O in P19, comparable to primary Hawaiian magmas. Experiments used a 1000-ton Walker-type multi-anvil press with an 18/12 assembly. Starting material was encapsulated in graphite capsules nested in Pt capsules; glassy carbon spheres at the capsule top facilitated melt extraction. Capsules were vacuum-dried and sealed. Type C thermocouples monitored temperature; expected gradients were <25 °C and <0.1 GPa. Quenches were by power shutoff, and samples were sectioned, mounted, and polished. Microstructures were examined by SEM; major elements of minerals and melts were measured by EPMA (JEOL JXA-8100) with standard ZAF corrections and appropriate standards. Analytical conditions included 15 kV accelerating voltage, beam sizes of 1–5 μm for minerals and 10–20 μm for quenched melts, and currents of 20 nA (minerals) and 10 nA (melts). Low-temperature runs used quenched glass in carbon-sphere pores; totals were low due to overlap with carbon, so melt compositions were normalized to 100% for comparison. Equilibrium assessment included: (1) uniform mineral compositions across charges, (2) low residuals in mass-balance (0.02–0.40), and (3) olivine–melt Fe–Mg exchange coefficients between 0.31 and 0.41, consistent with equilibrium. Fe-loss was minimized by the Pt–graphite double capsule; mass-balance residuals and KD values suggest negligible Fe loss. Modeling: Melt evolution paths included two averaged mixtures representing deep fractionation: AM-40 (melt fraction MF 85–9%, total MF ~40%) and AM-28 (MF 55–9%, total MF ~28%). Orthopyroxene assimilation at low pressure (<~2 GPa; depths shallower than ~60 km) was modeled with contemporaneous olivine precipitation and an assimilation/crystallization ratio of ~1.2–1.4 (opx assimilated to olivine crystallized). Ni contents were calculated using phase proportions/compositions, literature partition coefficients (for garnet/pyroxene; Ni in olivine from Matzen et al. 2017), assuming 2000 ppm Ni in source peridotite and 722 ppm Ni in lithospheric harzburgite orthopyroxene. REE patterns were modeled assuming a 75:25 primitive mantle:depleted mantle source and using high-pressure partition coefficients (3.0 GPa) for peridotite and Fe-rich picrite systems, including a scenario with 10 wt% decomposed garnet to release HREE during ascent.

• Experimental phase relations: Olivine is present in all runs (~3 wt%) except at 1350 °C (#R1244). Orthopyroxene disappears at lower temperatures, while clinopyroxene and garnet stabilize and increase with decreasing temperature; ~2 wt% phlogopite appears at 1200 °C. With decreasing temperature, melts increase in FeO, TiO2, alkalis; decrease in MgO and SiO2; CaO and Al2O3 rise initially (due to olivine and opx crystallization) then decrease slightly (with garnet and clinopyroxene formation).
• Three-step petrogenetic model validated: (1) Primary magmas form by partial melting of mantle peridotite at 3–4 GPa. (2) At the lithosphere base (~90–100 km), heat exchange causes substantial clinopyroxene and garnet crystallization in a deep magma chamber, producing FeOT-rich, SiO2-poor derivatives (AM-40, AM-28). (3) During ascent through shallower lithosphere (~60–10 km; <~2 GPa), melts assimilate orthopyroxene and precipitate olivine, driving compositions toward Hawaiian lava trends. Mixing of magmas with varying deep-fractionation degrees is expected at ~60 km.
• Compositional matching: AM-40/AM-28 derivatives have lower SiO2 and higher FeO, TiO2, CaO, Al2O3 than Hawaiian lavas; assimilation of ~0–30 wt% orthopyroxene shifts them onto observed Hawaiian tholeiitic trends (SiO2–MgO, TiO2–MgO, Al2O3–MgO, FeO–MgO, CaO–MgO, Na2O–MgO). The role of olivine addition/subtraction explains deviations from trends where crystals are not efficiently removed.
• Ni systematics: Peridotite partial melts alone have too low Ni at a given MgO relative to Hawaiian lavas. The model shows that deep clinopyroxene+garnet fractionation and subsequent opx assimilation reduce MgO while keeping Ni ~constant, producing parental melts with ~700–730 ppm Ni and MgO of 14–20 wt%, consistent with Hawaiian whole-rock trends. Modeled low-pressure olivine crystallized from these parents reproduces high Ni in Ni-rich olivine phenocrysts.
• REE patterns: Modeled parental melts (AM-40/AM-28 and their opx-assimilated derivatives) match Hawaiian lava REE patterns within ~20% error, with better fits to LREE than HREE. Slight HREE deficits can be reconciled if ~10 wt% of precipitated garnet decomposes during ascent, releasing HREE and improving the match.
• Alkaline lavas: Major and trace element data of post-shield alkaline lavas can be explained as melts from the same peridotite-derived system after deep fractionation, without requiring recycled oceanic crust, consistent with experimental constraints on early-crystallized olivine Mn and Ni.
• Physical migration: Primary melts likely pond at the lithosphere–asthenosphere boundary (~90–100 km) due to low lithospheric porosity, forming deep chambers where fractionation occurs. Ascent through the lithosphere likely proceeds via discontinuous conduits or fracturing; explosive ascent initiates at ~60 km. Reactive porous flow is enhanced by opx assimilation and olivine precipitation, increasing melt volume and porosity, with dunite formation being a potential byproduct. Entrained cpx and garnet are not preserved in erupted lavas due to reaction/decomposition at shallow depths; olivine is the dominant xenocryst/phenocryst.

The study resolves long-standing inconsistencies between inferred Hawaiian primary magmas and direct partial melts of peridotite or peridotite–pyroxenite hybrids by introducing deep-lithosphere fractionation and shallow lithospheric re-equilibration. High-pressure experiments demonstrate that clinopyroxene and garnet fractionation at ~90–100 km shifts primary magma compositions toward FeOT-rich, SiO2-poor melts. Subsequent orthopyroxene assimilation with coeval olivine precipitation at <~2 GPa increases SiO2 and adjusts major elements to match Hawaiian tholeiitic trends, while preserving near-constant Ni contents. The model explains whole-rock Ni–MgO trends and high-Ni olivine phenocrysts without invoking large contributions from recycled oceanic crust. REE patterns of modeled parent melts reproduce observed LREE and, with modest garnet breakdown, HREE as well. The process-based framework links physical melt migration (ponding, fractionation, fracture-controlled ascent, reactive porous flow) with geochemical evolution, offering a coherent account for both shield tholeiites and post-shield alkaline lavas originating from a common peridotitic source modified during ascent.

Hawaiian lavas can be generated from peridotite-derived primary magmas via a three-step process: (1) partial melting at 3–4 GPa, (2) extensive clinopyroxene and garnet fractionation in deep lithospheric magma chambers at ~90–100 km, and (3) orthopyroxene assimilation coupled with olivine precipitation during ascent through the shallower lithosphere (~60–10 km). This high-pressure crystallization and re-equilibration model reproduces major-element trends, Ni systematics, REE patterns, and mineralogical observations of Hawaiian shield tholeiites and post-shield alkaline lavas without requiring dominant recycled oceanic crust in the source. The findings highlight the importance of deep fractionation and lithospheric melt–rock interactions in shaping ocean-island basalt compositions. Future work should refine partitioning parameters at relevant P–T–X conditions, quantify kinetics of opx assimilation and garnet breakdown during ascent, integrate isotopic constraints, and explore variability across different Hawaiian volcanoes and eruption stages.

• The model relies on assumptions about source composition (75:25 primitive mantle:depleted mantle), partition coefficients (Ni, REE) at 3.0 GPa, and low-pressure Ni partitioning between olivine and melt; uncertainties in these inputs affect quantitative outcomes.
• Experimental runs are at a fixed pressure (3.0 GPa) and a discrete set of temperatures; natural systems experience pressure–temperature–composition paths that may vary spatially and temporally.
• The degree of orthopyroxene assimilation (0–30 wt%) and the opx-assimilated/olivine-precipitated ratio (1.2–1.4) are based on literature and modeling; direct constraints in the Hawaiian lithosphere are limited.
• REE model fits require assuming partial decomposition (~10 wt%) of precipitated garnet during ascent; the extent and timing of garnet breakdown in nature remain uncertain.
• Magma transport mechanisms in the lithosphere (fracturing vs. porous flow) are inferred from seismic and theoretical considerations, with limited direct observations.
• Isotopic signatures are not explicitly modeled; while major/trace elements can be explained without recycled oceanic crust, isotopic data often imply its presence, indicating additional source heterogeneity not addressed here.