Rifting of the oceanic Azores Plateau with episodic volcanic activity

Oceanic plateaus are extensive bathymetric swells with thickened oceanic crust and complex magmatic histories. The Azores Plateau, formed largely 10–4 Ma ago near the Mid-Atlantic Ridge, is transected by the ultraslow Terceira Rift, which has been variably interpreted to have initiated extension in the Miocene or as recently as 1–2 Ma. Deep rift basins such as the Hirondelle Basin expose thick volcanic successions along faulted escarpments, offering rare access to submarine stratigraphy. The study aims to determine the timing and nature of volcanism and rifting on the Azores Plateau’s Hirondelle Basin flank, test whether extension has been magmatically or tectonically dominated, and evaluate how mantle source characteristics and melting conditions evolved through time. Understanding these processes informs broader models of rift evolution and oceanic plateau development.

Previous ocean drilling and geophysical studies show oceanic plateaus can form via multiple volcanic phases over tens of millions of years (e.g., Ontong-Java with episodic activity between 121–37 Ma; Shatsky Plateau with sustained activity 144–129 Ma). The main plateau-building stage is often linked to plume head arrival, but post-emplacement evolution is less constrained due to limited stratigraphic access offshore. The Azores Plateau exhibits rifting features and young alkaline volcanism (<1.5 Ma) along the Terceira Rift. Prior work provides conflicting initiation ages for rifting: seismic interpretations suggest long-lived extension since 25–20 Ma, while tectonic studies on islands indicate initiation at 1–2 Ma. Magnetic anomalies parallel to young volcanic edifices have been interpreted as evidence for spreading, yet seismic profiles reveal faulted crust without clear signs of magmatic accretion. The Azores mantle is heterogeneous with distinct mantle sources feeding nearby volcanoes (Terceira, Sete Cidades, D. João de Castro), and geochemical data indicate variable degrees of partial melting and source components across the archipelago.

Sampling and field setting: A ~1.2 km vertical submarine stratigraphic profile (2510–1308 m below sea level) on the northern wall of the Hirondelle Basin was sampled using the MARUM ROV 'Quest 4000' during RV Meteor cruise M128 (2016). All samples are from submarine pillow lava flows, representing eruptive units; lavas and dikes dominate deeper sections, with volcaniclastic rocks and pelagic sediments becoming more common shallower than ~1690 mbsl.

Geochronology (40Ar/39Ar): Groundmass and plagioclase phenocrysts from four representative lava samples were dated at the Oregon State University Argon Geochronology Laboratory. Samples (150–300 µm) were washed, dried, density-separated, and acid-leached (HCl, HNO3, HF as appropriate). Irradiation was for 6 hours in the OSU TRIGA reactor with FCT sanidine flux monitor; J-values were determined via parabolic extrapolation with 0.1–0.2% uncertainties. Incremental heating used a defocused 25 W CO2 laser with gas purification via SAES getters. Ages were measured on two multicollector ARGUS-VI mass spectrometers. Ages were interpreted primarily from inverse isochrons (YORK2 least-squares; ArArCALC v2.7.0), applying robustness criteria (≥50% 39Ar release, S-factor >5%, acceptable MSWD, 40Ar/36Ar intercept ≥295.5±0.7). Where no resolvable isochron existed but Ar was highly radiogenic, plateau model ages were calculated assuming atmospheric trapped Ar.

Whole-rock major and trace elements: One glass and multiple whole-rocks were analyzed at GeoZentrum Nordbayern (FAU). Weathered surfaces and fillings were removed; samples were washed, crushed, and powdered in agate. Major elements were measured by electron microprobe (JEOL JXA-8200), and trace elements by established protocols; international standards (BHVO-2, BE-N, BR, GA) were repeatedly measured for quality control.

Radiogenic isotopes: Sr and Nd isotope ratios were measured by TIMS (Thermo Scientific Triton), Pb isotopes by double-spike MC-ICP-MS (Thermo Neptune Plus), and Hf isotopes by MC-ICP-MS after ion-exchange separation (modified Ln-Spec procedures with additional Ti/Zr removal). Reagent blanks were low (Pb ~30 pg; Sr ~200 pg; Nd ~80 pg). Standards yielded values consistent with published references (e.g., NBS981 for Pb; NBS987 for Sr; in-house Nd equivalent to La Jolla; AMES Hf standard).

Data analysis: Stratigraphic-geochemical correlations were made using depth versus elemental ratios (e.g., Nb/Zr) and isotopes (Nd, Hf, Pb) to define compositional units. Comparisons were made with lavas from Terceira, Sete Cidades (São Miguel), and D. João de Castro to infer mantle source variations and melting conditions. Petrogenetic interpretations leveraged REE systematics (Ce/Yb, Dy/Yb) and major element trends (SiO2 vs MgO) to discriminate degrees and depths of partial melting.

• The Hirondelle Basin northern flank preserves a 1.2 km-thick volcanic section of alkali basalts to basanites, subdivided into two geochemically distinct eruptive units defined by Nb/Zr.
• Lower unit (2510–1438 mbsl; Nb/Zr < 0.2): Volumetrically dominant (~1060 m of profile), erupted between ~2.020 ± 0.010 Ma and ~1.657 ± 0.004 Ma (groundmass plateau ages). Compositions show lower TiO2 at given MgO, higher 176Hf/177Hf, higher 143Nd/144Nd, and generally higher 206Pb/204Pb (up to ~19.77), indicating a more depleted mantle source akin to Terceira.
• Upper unit (1390–1308 mbsl; Nb/Zr > 0.2): Thin (~uppermost ~30 m), formed over ~100 kyr, with a youngest groundmass age of 1.558 ± 0.005 Ma. Features lower 143Nd/144Nd, lower 176Hf/177Hf, and higher 87Sr/86Sr relative to the lower unit, trending toward source characteristics of Sete Cidades and D. João de Castro.
• Age-depth profile indicates rapid construction: the lower ~1060 m formed within ~350–400 kyr; the uppermost ~30 m formed within ~90–100 kyr. The youngest in-situ lava at ~1.56 Ma constrains that tectonic opening of the ~35 km-wide Hirondelle Basin postdates this time.
• REE systematics: Lavas are LREE-enriched with chondrite-normalized Ce/Yb ~7–11; high (Dy/Yb)N and low SiO2 at given MgO imply melts generated at high pressure in garnet peridotite, consistent with low degrees of partial melting (<5%).
• Temporal geochemical trends: With time, mantle source signatures become less radiogenic in Nd, Hf, and Pb, and primitive melts show higher SiO2 and lower (Dy/Yb)N in the younger D. João de Castro lavas, consistent with increasing degrees and shallower depths of melting.
• Tectono-magmatic implications: The data support episodic volcanism along the Terceira Rift, alternating tectonically dominated rifting (post-1.56 Ma basin opening along normal faults) with later localized magmatism (D. João de Castro). There is no evidence for sustained magmatic spreading producing new oceanic crust in the basin during this interval.

The stratigraphic, geochronological, and geochemical data collectively address the timing and style of extension on the Azores Plateau. The alkaline, basanitic to alkali basaltic composition of lavas forming the upper >1 km of crust indicates deep, low-degree melting beneath thick lithosphere, contrasting with tholeiitic plateau-building or MORB-like magmatic accretion. Rapid construction of the lower unit (~350–400 kyr) followed by thinner upper flows and a shift in source isotopic signatures suggests a reduction in melt flux and a compositional transition of the mantle source around ~1.6 Ma. The youngest in-situ lava age (~1.56 Ma) implies subsequent tectonic rifting formed the Hirondelle Basin through normal faulting, after which magmatism resumed and focused at D. João de Castro. Comparisons of Nd-Hf-Pb isotopes and Nb/Zr with nearby volcanic centers indicate a temporal shift from a source similar to Terceira toward one resembling Sete Cidades/D. João de Castro, evidencing rapid mantle heterogeneity replacement in the melting zone. The evolution mirrors processes at ultraslow-spreading ridges, where extensional regimes alternate between amagmatic and magmatic phases, although here under a thicker lithosphere with alkaline magmas. Crucially, the absence of tholeiitic magmas and systematic magnetic anomalies tied to along-rift seafloor spreading, plus seismic indications of faulted but not accreting crust, argue against long-lived magmatic spreading. Instead, the Terceira Rift exhibits episodic, localized magmatism that weakens the lithosphere and promotes strain localization, analogous to the evolution of narrow continental rifts.

The upper ~1.2 km of crust exposed on the northern flank of the Hirondelle Basin formed rapidly (~500 kyr total), with the lower ~1.0–1.06 km erupting within ~350–400 kyr and the uppermost tens of meters within ~90–100 kyr. Lavas are dominantly basanitic to alkali basaltic, formed by low-degree melting in the garnet stability field beneath thick lithosphere. Geochronology constrains the opening of the Hirondelle Basin to after ~1.56 Ma, via tectonic normal faulting, followed by renewed and focused alkaline volcanism at D. João de Castro. Radiogenic isotopes (Nd-Hf-Pb) and Nb/Zr reveal a temporal shift from a more depleted mantle source (Terceira-like) to a more enriched/less radiogenic source akin to Sete Cidades/D. João de Castro. Slight increases in SiO2 and decreases in (Dy/Yb)N in younger lavas suggest increasing degrees and shallower depths of melting, plausibly due to rifting-induced lithospheric thinning. Overall, the Terceira Rift underwent alternating tectonically and magmatically dominated phases, resembling ultraslow-spreading and narrow continental rift systems rather than continuous magmatic seafloor spreading. Potential future research: Better constrain the onset and duration of post-rift volcanism at D. João de Castro through expanded geochronology; extend stratigraphic sampling to deeper sections to capture earlier plateau-building stages; and integrate high-resolution geophysics with geochemistry to map spatial variations in mantle source and melting conditions along the rift.

Only four 40Ar/39Ar ages were obtained to bracket a 1.2 km-thick sequence, limiting temporal resolution within units. Some samples showed minor excess Ar (addressed via isochron corrections). The onset age of volcanism at D. João de Castro seamount remains unknown. Stratigraphic access is confined to exposed rift walls; deeper or older portions of the plateau were not sampled. Geophysical evidence for or against magmatic spreading remains inconclusive in parts of the rift due to complex magnetic signals.