1 Introduction
The helium isotope systematics give constraints on the mantle structure and its evolution [7,16,17,19]. Helium has two isotopes; 4He is radiogenic and mainly derives from the radioactive decay of uranium and thorium, whereas 3He is primordial and was inherited during the Earth's accretion. Because helium can leave the atmosphere to space, there is almost no helium in air or in seawater, allowing little atmospheric contamination of samples, and especially of submarine samples. The 4He/3He ratios (or the R/Ra ratio where R is 3He/4He measured in the sample and Ra is the atmospheric ratio of 1.4×10−6) are different in Mid Oceanic Ridge Basalts (MORB) and in Oceanic Island Basalts (OIB) [18,19]. The MORB source appears to be homogeneous, the mean MORB ratio being (R/Ra∼8±1), [2]. This reflects either a well mixed reservoir, due to low viscosity, or high melting rates, allowing melting of a large part of the mantle. On the other hand, the OIB show a much larger range of the 4He/3He ratios, from 15 000 (R/Ra∼50) in the Iceland plume [35] to 200 000 (R/Ra∼3.6) in Sao Miguel island [27]. The low 4He/3He ratios suggest low time integrated (U+Th)/3He ratios in the source. However, the fact that the OIB sources show, in general, higher uranium and thorium contents than the MORB source implies higher 3He content in the OIB source. This suggests the existence of a relatively little degassed reservoir, probably located in the lower mantle [1], and sampled by mantle plumes that start either at the 670 km or 2900 km boundary layers. The high 4He/3He ratios can be interpreted in two ways. The first proposes that high U/3He material is located in the plume source. This can be recycled oceanic crust (e.g., Tubuai or St Helena) or sediments [10,11]. Another interpretation is shallow level contamination or post degassing radioactive production in the magma chamber [14,37].
The Macdonald seamount is the recent expression of the Austral chain volcanism (Fig. 1). This is one of the most active submarine volcanoes in the Pacific [34]. At least two volcanic events have been identified since 1987 [34]. The Macdonald seamount has a circular base of about 45 km in diameter at a depth of 3900 m and a narrow summit less than 40 m in diameter. The summit is presently about 40 m under water, allowing us to think that an island will emerge soon. The aim of this paper is to estimate the isotopic composition of the helium of the Macdonald seamount and to determine the nature of the Macdonald hotspot.

Map showing the location of three active submarine volcanoes in Polynesia (Mehetia, Pitcairn and Macdonald). The present study concerns the Macdonald seamount, located at the extreme east of the Australes chain.
Figure montrant la localisation de trois monts sous-marins actifs en Polynésie (Mehetia, Pitcairn et Macdonald). Cette étude concerne le mont sous-marin de Macdonald, situé à l'Extrême-Est de la chaı̂ne des Australes.
2 Sample location, analytical procedure and results
Sample locations are given in Fig. 2. Most of the samples have basaltic compositions [13]. All the samples come from depths lower than 2000 m.

Sample location in a simplified bathymetric map of Macdonald seamount.
Localisation des échantillons (carte bathymétrique simplifiée du mont sous-marin de Macdonald).
Helium was obtained by crushing under vacuum and analyzed with our glass mass spectrometer ARESIBO I, equipped with a Faraday cup and an ion counting system. The 4He blank was 1.9±0.1×10−8 ccSTP with R/Ra ratio of 0.7±0.3(1σ). Standards were analyzed daily. The helium standard is a gas from a Reunion island (Cilaos spring), that has a 4He/3He ratio of 56 980 (R/Ra=12.68). Results are given in Table 1. With the exception of samples SO47 DS2 and SO47 60GTVa3, the helium isotopic ratio is around 60 000 (R/Ra∼11.5). Sample SO47 DS2 shows a different helium ratio (45 800; R/Ra=15.8±0.3) whereas sample SO47 60GTVa3 shows a radiogenic ratio of 203 640 (R/Ra=3.6±0.5), associated with a helium content of 1.5×10−8 ccSTP/g (the lowest). Most of the samples have more primitive ratio than the mean MORB ratio of 90 000 (R/Ra∼8) [2]. The lowest helium ratio is similar to those found in some Iceland or Hawaii samples [20–22].
Helium isotopic ratio and Helium content (in μccSTP/g) of the Macdonald seamount. Note that sample TH28-07 was already analyzed by Marty and Dauphas [23]. They obtained R/Ra=11.2±0.1, similar to our measurement of 11.3±0.1. Marty and Dauphas [23] analyzed also sample TH30-03 which gives [4He]=0.08 μccSTP/g and R/Ra=5.62±0.10. Helium concentrations are in ccSTP/g. Ra is the 3He/4He ratio of air (1.4×10−6)
Rapport isotopique et teneur en hélium (μccSTP/g) du mont sous-matin de Macdonald. À noter que l'échantillon TH28-07 a déjà été analysé par Marty et Dauphas [23]. Ces auteurs ont obtenu R/Ra=11,2±0,1, similaire à notre propre résultat de 11,3±0,1. Marty et Dauphas [23] ont aussi analysé l'échantillon TH30-03, qui donne [4He]=0,08 μccSTP/g et R/Ra=5,62±0,10. Les concentrations en He sont fournies en ccSTP/g. Ra est le rapport 3He/4He de l'air (1,4×10−6)
Samples | Weight (g) | 4He/3He | ± | R/Ra | ± | [4He] |
SO47 55Ds1J | 0.16 | 60 100 | 790 | 12.0 | 0.2 | 1.1 × 10−5 |
SO47 64DS2 | 0.23 | 45 830 | 920 | 15.8 | 0.3 | 2.3 × 10−7 |
SO47 60GTVa3 | 0.70 | 203 640 | 26 400 | 3.6 | 0.5 | 1.5 × 10−8 |
SO47 57DS7 | 0.21 | 72 840 | 1270 | 9.9 | 0.2 | 3.6 × 10−7 |
SO47 68DS2 | 0.71 | 61 100 | 375 | 11.8 | 0.1 | 2.9 × 10−6 |
CY1011 TH28-03 | 1.77 | 61 790 | 485 | 11.7 | 0.1 | 6.4 × 10−7 |
CY1011 TH28-07 | 1.03 | 64 192 | 535 | 11.3 | 0.1 | 2.9 × 10−7 |
CY1011 TH28-10 | 0.70 | 62 320 | 535 | 11.6 | 0.1 | 3.7 × 10−7 |
CY1011 TH28-01 (olivines) | 0.59 | 62 460 | 550 | 11.6 | 0.1 | 1.2 × 10−7 |
3 Discussion
The Macdonald seamount lava clearly show a ‘high 3He’ signature (4He/3He<46 000; R/Ra>15.8) that may reflect the surface expression of a mantle plume that derives from a ‘primitive’ reservoir. These low 4He/3He ratios are associated with relatively high helium contents (up to 10−5 ccSTP/g) that are still minimum values, since degassing and vesicle bursting occurred during magma eruption or in the lab. Such helium contents are typical of OIB glasses (e.g. the highest Loihi helium content is 4 μccSTP/g). High helium contents have been measured in Iceland (10 μccSTP/g in some sub-glacial gassy samples) [12,21,29] or at Mehetia, Societies (30 μccSTP/g) [33]. These values are still lower than MORB concentrations (100 μccSTP/g) [5,24,28,30]. Considering the shallow eruption (<2000 m), it is legitimate to think that a helium loss occurred by degassing. However, because argon was not measured on these samples, the He/Ar ratio cannot be used to constrain the degassing rate [5,15,24,31].
Moreover, the fact that most samples have a 4He/3He ratio close to 60 000 (except for samples 55DS1J and 60GTVa3) may indicate that either the Macdonald seamount source is homogeneous or that there is a sampling bias of the Macdonald lava, i.e. the surface is very young. Similar observations can be made for other isotopic ratios (87Sr/86Sr=0.7037±0.00008, 143Nd/144Nd=0.51282±0.00003, 206Pb/204Pb=19.41±0.06; 207Pb/204Pb=15.62±0.03 and 208Pb/204Pb=39.19±0.09) [13]. This variation is small, considering the variation for the whole Austral chain.
Fig. 3 shows the R/Ra and 4He/3He ratios as a function of the 4He content (in μccSTP/g) in the samples for different submarine volcanoes of the Pacific island chains (Mehetia, Rocard, Teahitia: Societies; Bounty and Adams: Gambier and Macdonald: Austral chain). This figure shows clearly the effect of the degassing and the radioactive decay of U and Th on the helium isotopic ratio. Degassing decreases the mantle-derived helium content and makes the in situ radiogenic helium contribution more important. Only samples with more than 3×10−7 ccSTP/g preserve their mantle signature (Fig. 3). This process can be seen on Macdonald lava due to their high uranium and thorium contents ([U]=1 ppm [13]). For example, for sample SO47 60GTVa3, which has the most radiogenic helium isotopic ratio of the Macdonald samples, it is possible to write the radioactive production equation:

R/Ra and 4He/3He ratios as a function of the helium concentration. Other recent submarine volcanoes from Polynesia are shown for comparison. Only Mehetia and Macdonald show clear ‘primitive’ helium signature. The grey area is the mean MORB ratio. Data below this MORB trend clearly reflect radiogenic 4He addition in degassed magma chamber. Error bars of Macdonald samples are inside the symbols.
Rapports R/Ra et 4He/3He en fonction de la concentration d'hélium. Pour comparaison, d'autres données obtenues sur des monts sous-marins du Pacifique sont montrées. Seuls Mehétia et Macdonald montrent une signature « primitive » claire. La zone grisée montre la moyenne des MORB. Les données situées en dessous de cette valeur moyenne des MORB montrent clairement l'addition d'hélium radiogénique dans une chambre magmatique dégazée. Les barres d'erreur des échantillons de Macdonald sont dans les symboles.
The mean uranium content of Macdonald samples is U=1.0±0.4 ppm with Th/U=3.3±0.8 [13]. With these values, we get
For sample SO47 60GTVa3, 3He content is 7.5×10−14 ccSTP/g. Starting from (4He/3He)0=45 000, the time necessary to obtain 4He/3He=203 640 is 57 000 years after degassing. This is a realistic time for magma residence in a magma chamber. If the initial 4He/3He ratio is 60 000 (as in most samples), the residence time becomes 51 000 years, almost identical. Therefore, there is no need for another explanation (source, crust assimilation) to explain the radiogenic character of helium in some samples.
4 Constraints on the origin of the superswell
The Macdonald seamount shows a relatively primitive helium signature with a ratio of (R/Ra∼15.8). Davaille [8] and Davaille et al. [9] have suggested that the volcanism from the superswell is the expression at the surface of plumes deriving from the top of a dome rising from the deep mantle. Such a model is consistent with the helium data from active volcanoes from this area. A dome from the lower mantle, containing less-degassed material, produces plumes at its top that result in hotspot volcanism at the surface. On the other hand it has also been suggested that this volcanism just reflects shallow features, with no need of mantle plumes. In this case, the primitive helium ratio either reflects a ‘stored’ ancient mantle helium in the lithosphere [3] or the preferential melting of ‘primitive’ veins. We exclude the lithospheric hypothesis with the following argument (Fig. 4). Assuming the lithosphere has an age of 100 Ma under the Macdonald seamount, the Anderson's model suggests that the mantle source of the lithosphere had a helium isotopic ratio of 45 000 100 Ma years ago (the lowest measured helium ratio of the Macdonald seamount). To get the present day MORB value of 90 000 [2], the U/3He ratio of the depleted mantle that is necessary is 2.2×105 (mol/mol). Using this ratio and calculating the helium isotopic ratio 500 Ma ago, we get a negative helium isotopic ratio for the depleted mantle (Fig. 4)! Therefore, this U/He ratio is not realistic and this model of stored helium in lithosphere is not correct. Fertile veins with primitive helium that preferentially melt under the thick pacific lithosphere to give hotspot volcanism is a possible model. These veins have to be larger than the characteristic diffusion length of the helium at mantle temperatures, otherwise the helium isotopic ratio would be similar to the MORB ratio (which, in such a model, is a mixture of primitive helium veins and a more radiogenic matrix). With a diffusion coefficient D=10−8 cm2/s [36] and for t=100 Ma, we get a characteristic length of ∼50 m (for 1 Ga, we get ∼200 m). Taking D=10−7 cm2/s, the characteristic length is ∼200 m and with D=10−9 cm2/s, it is ∼20 m in 100 Ma. This does not change the following interpretations. These veins could be either un-melted parts of the mantle (really primordial helium), pieces of lower mantle (with primordial helium) entrained when the plume forms or ancient subducted material with low U/3He [6]. From helium measurements only on the Macdonald seamount, it is hard to distinguish between the ‘canonical’ model and the primitive vein model. However, considering the fact that the samples are relatively rich in helium, the subducted material (residual mantle) can be excluded because such a residual mantle does not contain significant helium [4,25] and is probably not a fertile material that can melt preferentially. Therefore, a model of veins of undegassed material, either due to lower mantle entrainment or to un-melted primitive peridotites, is possible, with the unproven condition that this material melts preferentially to pyroxenites. If such a case is possible, the high 3He signature observed on some parts of the mid oceanic ridges [26,32] needs to be explained. In such a geodynamic context, the melting rate is certainly higher and therefore, the previous model does not hold.

The perisphere model of Anderson [3]. In this model, the primitive helium ratios observed in some OIB reflect a lithospheric helium rather than a lower mantle derived helium. This is due to an assumed U/3He ratio of zero in the residual material, allowing freezing an ancient and low 4He/3He ratio. However, the required U/3He ratio to increase the isotopic ratio from a ratio of 45 000 to an actual MORB ratio of 90 000 in 100 Ma is 40 times too high compared to the estimate of this ratio in the depleted mantle [1]. This gives a negative helium isotopic ratio 300 Ma years ago in the mantle, which is not possible.
Le modèle de périsphère de Anderson [3]. Dans ce modèle, le rapport primitif de l'hélium qui est observé dans les points chauds intraplaques, est issu de la lithosphère océanique plutôt que du manteau inférieur. En effet, dans ce modèle, le rapport U/3He est supposé égal à zéro dans la lithosphère, ce qui permet de « geler » le rapport isotopique du manteau lors de la formation de la lithosphère. Toutefois, le rapport U/3He qui est nécessaire pour augmenter le rapport isotopique de l'hélium de 45 000 à 90 000 (le rapport moyen des MORB) en 100 Ma est 40 fois trop grand par rapport aux estimations que l'on peut faire de ce rapport [1]. De plus, cela donnerait un rapport isotopique négatif dans le manteau il y a 300 Ma, ce qui n'est pas possible.
Therefore, the simplest explanation for Macdonald helium is the presence of a plume derived either from the top of a dome rising from the deep mantle and containing primordial material or a plume derived from either the 670- or 2900-km boundary. In any case, this plume entrains less-degassed material and is mixed with MORB-like material.
5 Conclusion
We have presented new helium isotopic data for the Macdonald seamount (Austral chain). Most of the samples have a 4He/3He ratio close to 63 000 (R/Ra∼11.5). One sample shows a very low ratio (45 000: R/Ra=15.8). Another sample has a radiogenic ratio that can be explained by radioactive decay of a degassed uranium rich magma. A residence time of 50 000y is necessary to produce the helium ratio. The simplest explanation of the Macdonald seamount helium signature is the presence of a mantle plume derived from either a dome rising from the lower mantle or from a thermal boundary in the mantle and entraining less-degassed material (∼1–5%).
Acknowledgements
Marine Ravetto is thanked for her great help during the analyses. L. Dosso and P. Burnard are thanked for their constructive reviews. This is IPGP contribution number 1990.