Introduction

The Karoo mantle plume has been invoked to explain the origin of the Karoo, Ferrar, and Chon-Aike large igneous provinces (LIPs), as well as volcanic sequences of the Malvinas/Falklands Islands and Ellsworth-Whitmore Terrane1,2,3,4,5,6,7,8,9,10,11,12. The Karoo LIP is essentially widespread in South Africa, East Antarctica, and Malvinas/Falklands Islands, whereas the Ferrar LIP is mainly located in East Antarctica, and the Chon-Aike LIP extends throughout Patagonia, Antarctic Peninsula, and Ellsworth-Whitmore Terrane.

Mantle xenoliths derived from the subcontinental lithospheric mantle (SCLM) of Patagonia have demonstrated mineralogical and compositional heterogeneity related to its complex geodynamic history13,14,15,16,17,18,19,20,21,22. These processes include the lithospheric mantle formation by variable degrees of partial melt extraction mainly during the Paleo- to Mesoproterozoic (2.1–1.0 Ga), the fragmentation of the supercontinent Rodinia, and the subsequent collision of continental blocks assembled at southwestern Gondwana13. The similar compositions and TRD model ages based on Os isotopes between the SCLM of southern Patagonia and southern South Africa support their common origin and connection since the Mesoproterozoic13. Later, the Mesozoic and Cenozoic geological evolution includes the formation of an active subduction zone at the western margin of South America (Andean Cordillera), where the collisions of the Farallon–Aluk spreading ridge (Paleocene-Eocene) and Chile Ridge (since Miocene) promoted the opening of successive slab-window segments beneath the southern Patagonia.

Noble gas isotopes, particularly helium, are important tools for the definition of mantle reservoirs, however, very limited noble gas isotope data are available for mantle xenoliths from Patagonia. It is well-constrained that, in general, the SCLM is characterized by radiogenic 3He/4He ratios (6 ± 1 RA 17,23) (where RA corresponds to atmospheric ratio of 1.4 × 10−6 24) compared to mid-ocean ridge basalts (MORB = 8 ± 1 RA 25,26) and primordial/undegassed source of ocean island basalts (OIB usually >10 RA27). In Patagonia, the existence of an archetypical SCLM was well constrained by Pali-Aike xenoliths, which have uniform 3He/4He composition (total crushing of 6.91 ± 0.04 RA, n = 6017). Mantle xenoliths from Gobernador Gregores are characterized by typical MORB-like component related to the Chile Ridge subduction (total crushing of 7.34 ± 0.02 RA, n = 3517). Differently, mantle xenoliths from Coyhaique evidence the effective recycling of subduction-related melt/fluids in the Patagonian SCLM (3He/4He = 0.24–1.74 RA20).

In this paper, we report the presence of a deep, undegassed, and pristine mantle component with high-3He/4He isotope ratios (21.65–27.68 RA) preserved by spinel-bearing mantle xenoliths stable in the unusual shallow asthenosphere beneath Patagonia. Also, we discuss the most plausible geodynamic processes responsible for the upwelling of pristine volatiles possibly associated to the Karoo mantle plume and its apparent relation with a long-lived trans-lithospheric structure related to the collision of the Deseado and North Patagonian continental blocks at the southwestern edge of Gondwana.

Results and discussion

Mantle xenoliths were collected from the Central-North Patagonia region (40’46”–44’52” °S), in the domains of the North Patagonian Massif [Cerro del Mojón (PM4); Estancia Alvarez (PM7); and Prahuaniyeu (PM8)], as well as close to an inferred collision zone [Cerro Chenque (PM10) and Cerro de los Chenques (PM12)] (Fig. 1). This trans-lithospheric suture zone would have been generated by the north-easterly collision between the Deseado and the North Patagonian massifs during the Carboniferous-Permian28.

Fig. 1: Elastic thickness (Te) map of southern South America29.
figure 1

The low Te values (<20 km) on the boundary between the North Patagonian Massif and the inferred collision zone reveal a thin and weak lithosphere. Note that the occurrence of mantle xenoliths containing pristine high-3He/4He ratios (21.65–27.68 RA) is restricted to within or near the region with low Te values. Mantle xenolith occurrences are indicated by white diamonds (new helium data) and white squares (helium data from ref. 17). 1 = Cerro del Mojón (PM4), 2 = Estancia Alvarez (PM7), 3 = Prahuaniyeu (PM8), 4 = Cerro Chenque (PM10), 5 = Cerro de los Chenques (PM12), 6 = Gobernador Gregores (PM23), 7 = Laguna Ana (PM14), and 8 = Laguna Timone (PM18). Also are indicated the Chilenia (Ch), Cuyania (C), and Pampia (P) terranes. The North Patagonian (NPM) and Deseado (DM) massifs, as well as the Huincul Fault (HF) are indicated.

Full size image

Most of the lavas hosting mantle xenoliths have Pliocene to Pleistocene K-Ar ages (3.76–0.73 Ma; Cerro del Mojón, Prahuaniyeu, and Cerro de los Chenques) (Supplementary Data 1). A lava flow from Estancia Alvarez yielded Miocene K-Ar age (22.06 ± 1.18 Ma) (Supplementary Data 1). Cerro Chenque is interpreted as an ancient volcanic conduit composed by two distinct dikes. The oldest is formed by a porphyritic strongly weathered rock containing large mantle xenoliths (up to 20 cm). It was not possible to determine the eruption age of this rock using geochronological methods, but it is estimated to be Lower Eocene (~52 Ma) according to its geochemical, textural, and structural similarities with the adjacent volcanoes14. Differently, the younger dike is composed by fresh lava with aphanitic texture, Pliocene K-Ar age (4.36 ± 0.24 Ma) and hosts small mantle xenoliths (up to 5 cm) (Supplementary Data 1). Only mantle xenoliths from the older lava flow could be analyzed in this study.

Helium, neon, and argon isotope ratios and concentrations of mantle xenoliths are presented in Supplementary Data 2. In general, helium isotope ratios (in RA) discussed here are assumed as mantle-derived because they were extracted by both stepwise and single crushing methods under high vacuum, which avoid the cosmogenic or radiogenic contributions from the mineral matrix. However, even analyzed by crushing, due to the lack of high-quality neon data to better constrain the existence of pristine high-3He component, we have decided to disregard doubtable results possible containing minor contribution from the mineral matrix (Supplementary Information and Supplementary Data 2). Previous measurements aiming to estimate the abundance of gas present in studied mantle xenoliths were performed by single heating extraction method. In order to constrain whether the crystal-lattice-hosted component is primordial or cosmogenic in samples PM10-Q77 and PM10-Q109, we analyzed their rock powders (after stepwise crushing) by heating extraction.

The 3He/4He isotopic ratios of truly trapped fluid inclusions extracted by crushing method reveal the existence of intrinsic SCLM beneath Central-North Patagonia (5.33–6.99 RA), MORB-like (7.26–7.60 RA), and primordial (23.02–27.68 RA) components (Fig. 2). The pristine component observed in mantle xenoliths from Cerro Chenque (PM10) is confirmed by the 3He/4He ratio obtained by heating the rock powder after stepwise crushing (21.65 ± 0.32 RA). This value is equivalent (slightly lower) to the previous results obtained by stepwise crushing, indicating negligible effect of cosmic ray irradiation and small addition of a radiogenic 4He from U and Th (Supplementary Fig. 1). Based on these results, we conclude that the high-3He/4He ratio (single crushing = 23.02 ± 6.84 RA) revealed by sample PM10-B2 Ol also reflect a true pristine component. It is justified by its identical 3He/4He ratio compared to the result revealed by the sample PM10-Q77, although the analytical uncertainty is higher. Conversely, the high-3He/4He isotope ratios obtained by heating extraction (10.50–188.08 RA) are essentially cosmogenic (18–140 %) (see Supplementary Information and Supplementary Fig. 2).

Fig. 2: 3He/4He isotopic ratios versus 4He concentration (10−9 cm3 STP/g) of mantle xenoliths from Central-North Patagonia.
figure 2

These measurements were determined by crushing extraction method and represent trapped components. Uncertainties are 1σ. In most cases, the uncertainty is smaller than the symbol size. Typical mid-ocean ridge basalt composition (MORB = 8 ± 1 RA25,26) is represented by yellow while typical subcontinental lithospheric mantle composition (SCLM = 6 ± 1 RA17,23) is represented by light blue. For comparison, Patagonian mantle xenoliths representing intrinsic SCLM (Pali-Aike Volcanic field17) and MORB-like (Gobernador Gregores17), as well as the Chile Ridge basalts36 were plotted. Mineral separates of mantle xenoliths (olivine, pyroxenes, and amphibole) from Antarctic (Ferrar large igneous province) also were plotted38,39. In addition, basaltic lavas from South Atlantic hotspots (Shona, Bouvet, and Discovery), which represent the Karoo mantle plume43,44,45,46, and olivine separates of lavas from Paraná-Etendeka large igneous province were plotted48.

Full size image

Most neon isotopes contain strong contributions of blank (>10 %), as well as of 40Ar++ on 20Ne+ and CO2++ on 22Ne+ (both >10 %), defining insufficient quality to be considered in the discussion. The data containing enough analytical quality do not differ from air composition (20Ne/22Neair = 9.80; 21Ne/22Neair = 0.02924) within 2σ analytical uncertainties. The 40Ar/36Ar ratios between 304 and 2692 vary from a near-atmospheric ratio (29624) toward a small contribution of a mantle endmember exemplified by the increasing isotope ratio values.

Helium isotopes, elastic thickness (T e) and seismic tomographic data

Notably, mantle xenoliths with high-3He/4He isotopic ratios are found in a region characterized by low values of elastic thickness estimations (Te < 20 km, Fig. 1), suggesting the presence of thin and weak (low viscosity) continental lithosphere13,29. The highest 3He/4He isotopic ratios (21.65–27.68 RA) were obtained in mantle xenoliths from Cerro Chenque (PM10), located in the southern area of the North Patagonian Massif (Fig. 1). It is worthy of note that the zone characterized by the lowest values of Te estimations is displaced to the north with respect to the inferred collision zone proposed by ref. 28. (Fig. 1).

Seismic tomographic data (SA201930) reveal a strong low-velocity anomaly in the mantle wedge beneath Central-North Patagonia (~50–180 km), immediately overlying the top of the Nazca oceanic plate31 (Fig. 3). This anomaly possibly reflects thermal evidence of the connection between a ~N-S slab tearing related to the Farallon–Aluk mid-ocean ridge during the Paleocene-Eocene (north of 46°30’S) and the slab window produced by the Miocene to Present collision of the Chile Ridge, which covers a significant part of Santa Cruz province (south of 46°30’S32,33). Thus, this region indicates a lithospheric thinning and an asthenospheric rising. This observation is consistent with the Te values presented in Fig. 1, as well as with the shallow lithosphere-asthenosphere boundary34 observed in the inferred collision zone (~50 km, Fig. 3). In this context, the crust-mantle boundary (Moho) has ca. 40 km, revealing an extremely thin lithospheric mantle with ca. 10 km thick35 (Fig. 3).

Fig. 3: Seismic tomographic results of southern South America (SA201930).
figure 3

A strong low-velocity anomaly (dVs/Vs (%)) is observed in the subcontinental lithospheric mantle (SCLM) between depths of 50 and 180 km beneath Central-North Patagonia. This anomaly probably represents the current thermal influence generated by the connection between a slab tearing widespread along Central-North Patagonia (Paleocene-Eocene Farallon–Aluk mid-ocean ridge north of 46°30’S) with the slab-window associated with the Chile Ridge subduction. The horizontal solid black line in (a) represents the profile of the vertical tomographic slice. The dashed blue line is the spreading Chile Ridge. Sample localities are shown with white triangles and the same numbers as in Fig. 1. In (b), the white dashed line is the crust-mantle boundary (Moho) along the profile35. Black dashed line is the lithosphere-asthenosphere boundary (LAB) along the profile34. The green dashed line is the top of the current oceanic slab obtained from the model presented by ref. 31. In the mantle transition zone (410–660 km), a high-velocity anomaly is identified, probably representing the stagnant Farallon Plate37.

Full size image

Based on these results, we exclude the presence of an active mantle plume (thermal anomaly) in this region due to the absence of continuity of the low-velocity anomaly toward the deeper mantle column (>200 km) (Fig. 3). Additional evidence is given by mantle xenoliths containing pristine helium composition (high-3He/4He), which have low temperature estimates than those necessary to justify the existence of a hot mantle plume (838 °C; sample PM10-Q7714). Also, we discard any influence of the Chile Ridge into the high-3He/4He ratios of Patagonian mantle xenoliths essentially due to its homogeneous and slightly radiogenic MORB composition recorded by basaltic glasses (total heating = 7.84 ± 0.08 RA36, Fig. 2). In the mantle transition zone (410–660 km), there is a high-velocity anomaly that probably represents the stagnant Farallon Plate37 (Fig. 3).

Constraining the upper mantle reservoirs beneath Patagonia

Aiming to confirm the quality of helium data determined by crushing extraction, we evaluated the contribution of atmospheric, post-eruptive radiogenic 4He, and cosmogenic 3He components into the 3He/4He isotope ratios of Patagonian mantle xenoliths (see detailed calculations in the Supplementary Information and Supplementary Data 2). In summary, our results show a negligible contribution of atmospheric (usually <1 %) and post-eruptive radiogenic production of 4He (usually <0.3 %). Conversely, considering the xenoliths from the inferred collision zone (Cerro Chenque) that have truly high-3He/4He ratios, the effect of cosmic ray irradiation reaches 15 %. Although this percentage seems high, its influence on the isotopic ratios is not significant and does not change the interpretation of the data.

Helium isotopes allowed the identification of three reservoirs in the upper mantle at the spinel stability field beneath Central-North Patagonia. Most mantle xenoliths from the North Patagonian Massif preserved their SCLM component (3He/4He = 5.33–6.45 RA, most >6.0 RA), which was partially overprinted by a MORB-like component (3He/4He = 7.31–7.60 RA) (Fig. 2). Contrary, mantle xenoliths outcropping through the inferred collision zone are characterized by the mixing between primordial component (3He/4He = 21.65–27.68 RA) with subordinated SCLM and MORB-like components (3He/4He = 6.89–7.26 RA) (Fig. 2).

Notably, our samples have strong compositional similarity with mantle xenoliths from Ferrar LIP (Antarctica38,39), which mainly preserved the SCLM reservoir (3He/4He = 5.86–6.98 RA) superimposed by MORB-type (3He/4He = 7.00–9.05 RA) and OIB-like (3He/4He = 10.60–26.80 RA) components (Fig. 2). It is important to observe that although the high-3He/4He identified beneath the Ferrar LIP was not discussed as primordial by the authors mentioned above, there is no reason to rule out this source. For example, ref. 38 have decided to ignore 3He/4He ratios higher than 10 RA (up to 26.80 RA) due to the mass fractionation revealed by neon and argon results. In this case, we agree that samples within the Solar and/or mass fractionation lines coupled with 38Ar/36Ar ratios lower than the atmosphere (0.18824) must be disregarded (see Fig. 4 of ref. 38). However, considering the analytical uncertainties (2σ), five samples show neon isotopes compatible with the atmospheric, MORB, and OIB components (Supplementary Fig. 340,41), plotting within the mantle array in the 20Ne/22Ne versus 38Ar/36Ar (see Fig. 4 of ref. 38). Remarkably, ref. 39 found a high-3He/4He ratio (20.18 RA) accompanied by atmospheric neon isotopic ratios, which coincides with the result obtained for Cerro Chenque (total crushing of 26.28 RA) as well as with the highest ratio reported by ref. 38. (26.80 RA) (Fig. 2).

Fig. 4: Geodynamic model showing the complex evolution involved in the generation of the upper mantle beneath Central-North Patagonia since the Pangea consolidation (not on scale).
figure 4

a, b The final stages of the collision between the Deseado Massif and the North Patagonian Massif during the Carboniferous-Permian allowed the generation of a weak suture zone, the so-called inferred collision zone28. At this moment, occurred the break-off of the oceanic plate that favored the development of a slab-window and the asthenospheric upwelling beneath the North Patagonian Massif. c Lithospheric hybridization promoted by the edge-driving convection mechanism (lithospheric erosion) triggered by the opening of the Golfo San Jorge Rift Basin during the Early Jurassic52. This event strongly contributed to the consolidation of the inferred collision zone in Central-North Patagonia. d The flat-slab subduction favored the continuity of the lithospheric thinning and asthenosphere upwelling beneath the Patagonian back-arc8. e The geometry of the oceanic plate drastically changed from flat to steep while its detached fragments sank. The rearrangement of the oceanic plate triggered the westward influx of the Karoo mantle plume (KMP) by advection8. The arrows indicate the direction of movement between the massifs and tectonic plates. DM Deseado Massif, NPM North Patagonian Massif, CLM Continental Lithospheric Mantle, SAP South American Plate.

Full size image

The South Atlantic hotspots Shona (beneath the Weddell Sea), Bouvet (beneath the Antarctica), and Discovery (beneath the Karoo province) are assumed as the expression of the Jurassic Karoo-Ferrar mantle plume12,42. These hotspot lavas are characterized by pristine and high-3He/4He isotopic ratios (12–15 RA43,44,45,46) (Fig. 2). Regarding the lower isotopic ratios recorded by these samples compared to our results, it can be explained by the less frequent occurrence of hotspots with high 3He/4He ratios in the Atlantic large low-shear velocity province (LLSVP47). According to these authors, Shona and Bouvet hotspots are connected to (supplied by) a mantle plume (in this case, Karoo mantle plume) while Discovery is defined as disconnected. Alternatively, the lower 3He/4He ratios observed in these hotspot lavas can be a consequence of the plume-ridge interaction in which a deeper and undegassed component (OIB) interacted with a shallower and degassed (MORB) mantle source (Fig. 2). Similarly, olivine-rich magmas from the Paraná-Etendeka LIP (Namibia) show similar high-3He/4He ratios (18.80–25.85 RA48) compared to the Karoo mantle plume. These results revealed a complex mixing between different sources, such as radiogenic SCLM, MORB-type, and plume material (Fig. 2).

Based on the information above, we conclude that the primordial helium (>20 RA) found in the upper mantle beneath Central-North Patagonia represents remnants of a deep and common component related to the initial stages of the Gondwana breakup. In addition, the helium mixed-source observed beneath Patagonia, South Atlantic hotspots, and Antarctica is likely to be the norm rather than the exception in which the deep-seated material interacted with the shallow SCLM and MORB components.

Geodynamic model for the upwelling of pristine 3He from the deep mantle

The high-3He/4He of mantle xenoliths from Central-North Patagonia indicates the existence of a pristine and deep helium component beneath southern South America. The rising of this component was most likely channelized throughout an inferred collision zone, which is a weakness zone related to a trans-lithospheric suture formed since the establishment of Pangea (~320–300 Ma) (Fig. 4).

Considering the geodynamic evolution of Patagonia, we infer the combination of two mechanisms to explain the rejuvenation and thinning of the base of the continental lithosphere beneath this collision zone (Fig. 4). The development of a flat subduction during the Carboniferous-Triassic continental collision promoted the dehydration of the oceanic slab and the subsequent hydration of the lithospheric mantle, decreasing its rheological strength13,49,50,51 (Fig. 4a). Posteriorly, the subduction angle changes from shallow to steep, inducing inflow of the asthenosphere through the opening of a slab-window as a response to the slab break-off13 (Fig. 4b). Thus, this event promoted the lithosphere erosion beneath the southern edge of the North Patagonian Massif13. The inferred collision zone seems to be related to the region with lowest Te (Fig. 1), as well as with a shallow lithosphere-asthenosphere boundary (~50 km; Fig. 3).

Thereafter, the continental surface of this collision zone was covered by the thick Late Triassic-Jurassic sedimentary successions of the Golfo San Jorge Rift Basin during the continental subsidence52 (Fig. 4c). The dominant extensional tectonic regime comprised by deep normal faults that enhanced the lithospheric thinning is associated to the opening of this rift basin (Fig. 4c). This event is coeval with the early stages of the fragmentation of southwestern Gondwana, which was triggered by the Karoo mantle plume4,5,8. In this context, during the Late Triassic to Lower Jurassic, the consolidation of another flat subduction was followed by a slab steepening-induced (rollback) mantle flow that resulted in a southwestward asthenospheric advection4,5,8 (Fig. 4d, e). Thus, the mantle flow may have exerted a suction effect of a component enriched in primordial high-3He from the Karoo mantle plume towards Patagonia (Fig. 4e). In addition, the influence of hotspots from the South Atlantic Ocean (Shona, Bouvet, and Discovery) has been invoked to explain the compositional heterogeneity evidenced by lavas from the Patagonian back-arc53 and Chile Ridge54. Therefore, the development of flat-slabs and the thermal erosion promoted by a mantle plume replaced the base of weak lithosphere by uprising asthenosphere reducing its viscosity and thickness. Consequently, we conclude that the mantle xenoliths from Cerro Chenque represent asthenospheric fragments while the mantle xenoliths from other Patagonian localities are derived from the SCLM.

Expanding the perspective beyond the Patagonian back-arc, we propose that the complexity involved in the long-lived tectonic evolution of intraplate continental settings favors the conformation of upper mantle reservoirs with strong geochemical and isotopic compositional heterogeneity, that can include deep and pristine asthenospheric mantle.

Methods

40K/40Ar chronology

K-Ar ages of mantle xenolith host lavas were determined using the unspiked sensitivity method (Supplementary Data 1). Ar analyses were performed using a noble gas mass spectrometer MS-III (modified VG5400) at the Geochemical Research Center, Graduate School of Science, University of Tokyo. The crushed and sieved whole-rock samples (~0.5 g) were wrapped in aluminum foil 10-μm thick and loaded into a sample holder made of Pyrex-glass, which is connected to a vacuum line at an extraction oven were the sample is completely melted at 1700 °C. The gas released was purified and separated before its introduction in the mass spectrometer. The Ar isotope analyses were made on a relatively small amount of Ar gas (<2 × 10−7 cm3 STP). When the amount of Ar gas extracted from the sample exceeded this limit, it was reduced using the purification line. Uncertainties on 40Ar sensitivity and 40Ar/36Ar ratio are estimated to be 5 and 0.2 %, respectively, based on repeated measurements of the atmospheric standard containing 1.5 × 10−7 cm3 STP of 40Ar. The K concentration of each sample was determined for an aliquot of the crushed and sieved whole-rock fractions used for Ar analysis by the X-ray fluorescence (XRF) method (Phillips PW2400). The analytical precision was verified by analysis of the rock standard JB-1a, issued by Geological Survey of Japan.

Noble gas isotopes

He-Ne-Ar isotopic ratios of mantle xenoliths from Central-North Patagonia were performed at the Research Center for Advanced Science and Technology, University of Tokyo (Supplementary Data 2). Here we applied different extraction methods to obtain the noble gas isotopic ratios. In order to release the noble gases trapped in fluid inclusions from whole-rock (WR), olivine (Ol), clinopyroxene (Cpx), and olivine plus orthopyroxene (Opx) samples, in-vacuum stepwise and single crushing measurements were performed. Twelve samples (0.21–1.33 g) were crushed in a stainless-steel tube with a single (2000 strokes) or sequential number of strokes (usually applying 100, 500, 1000, and 2000 strokes) from a nickel piston driven from outside the vacuum by a solenoid magnet55, implying 25 measurements. Single step heating extraction method was previously performed (before crushing measurements) for 20 disaggregated minerals representing whole-rock samples (0.52–0.68 g). Aiming to verify whether the high-3He/4He ratios displayed by samples PM10-Q77 WR and PM10-Q109 WR have cosmogenic contribution or whether they represent a pristine component, additionally single step heating measurements were carried out using their rock powders.

Based on the reproducibility of measurements of a Japanese helium standard (HESJ) and those of a calibrated air standard for other noble gases, the estimated experimental uncertainties for noble gas concentrations were 5 % for helium and argon, and 10 % for neon. Uncertainties assigned to the observed isotopic ratios were one standard deviation (1σ), including uncertainties of blank and mass discrimination corrections determined by the measurements of HESJ (3He/4He = 20.415 ± 0.029 RA56) and atmospheric gas. Helium concentrations [4He] corrected to the air range from 0.39 to 53.39 × 10−9 cm3 STP/g for crushing measurements, and from 1.38 to 265.04 × 10−9 cm3 STP/g for heating. Blanks were run by using the same procedure as that used for the samples.

The blank contribution for 4He in crushing measurements is low (<2.5 %; n = 23 of 25), except for samples PM7-B1 WR (6.1 % in stepwise 10 strokes) and PM10-B2 Ol (4.1 % in single 2000 strokes). Comparatively, the blank contribution for 4He during heating extraction is high, with 10 of 22 results containing >2.0 % (up to 16.62 %).

Neon results obtained by crushing and heating extraction methods were corrected for 40Ar++ on 20Ne+ and CO2++ on 22Ne+ following the method described by ref. 57. They vary, respectively, from 0.1 to 29 % and 0.8 to 74 %. Considering both extraction methods, the blank contribution for 20Ne vary from 0.39 to 198.00 %, where 38 of 48 measurements have less than 15 %. However, considering 2σ uncertainty and the contributions of 40Ar++ on 20Ne+ and CO2++ on 22Ne+ as less than 10 %, no neon data has enough quality to be discussed.

For 40Ar, the blank contribution is <21 % considering crushing measurements, but <10 % eliminating the 100 strokes of Cerro Chenque (PM10-Q77 and PM10-Q109). Regarding heating measurements, the blank contribution for 40Ar usually is lower than 15 % (n = 16 of 22), reaching up to 56.82 %.

Inductively coupled plasma mass spectrometer (ICP-MS)

The thorium (Th) and uranium (U) concentrations (in ppm) of whole-rock mantle xenoliths were measured at the LMTG (Observatoire Midi-Pyrénées) of the University Paul Sabatier (Toulouse III). These results were used to calculate the 4He radiogenic contribution into Patagonian mantle xenoliths (Supplementary Information). The measurements were carried out using a quadrupole inductively coupled plasma mass spectrometer (ICP-MS) Elan 6000 PerkinElmer. Calibrations, internal standard, and interferences corrections were done following the procedure described in ref. 58. Data quality was controlled by running BCR-2 standard. Relative standard deviations are generally ≤5 %.