Ultra-depleted and Ultra-enriched Primary Melts in Hawaii Plume: Evidence From Melt Inclusions Study

I. K. Nikogosian Acad. Sci. Vernadsky Institute of Geochemistry, Kosygin str., 19, Moscow 117975 Russia


A. V. Sobolev Acad. Sci. Vernadsky Institute of Geochemistry, Kosygin str., 19, Moscow 117975 Russia

The data on mantle peridotites (Johnson et al., 1990) and ultra-depleted melt inclusions (Sobolev and Shimizu, 1993) both argue for very efficient melt extraction and critical melting scenario for Mid Ocean Ridge (MOR) setting. However similar data for mantle plumes are not yet known. This opens the question: how does mantle melt in mantle plumes? The available answers are: (1), closed system batch melting (Budahn and Schmitt, 1985) or (2), melting in the open system but homogenizing of melts by their percolation through the mantle (Watson and McKenzie; Watson, 1993). This paper presents new data on the nearly primary melts of Hawaii tholeiites unchanged by mixing or later percolation found as inclusions in high-Mg olivine phenocrysts. The samples studied are picritic lavas from Mauna Loa volcano, Hawaii (historic (1868) oceanite H-OC and prehistoric (pre-1868) picrite H-1 (Sobolev and Nikogosian, 1994).

The following steps of this investigation have been used: (1) Massive search (around 2000 electron probe measurements) for the most magnesium and thus earliest olivines phenocrysts which probably contain inclusions of unchanged primary melts. (2). Study of melt inclusions in selected high Mg olivines using high temperature microscopy (to produce homogeneous glass) and electron probe. (3) The inclusions with extreme compositions as well as typical inclusions were measured for trace elements by SIMS on IMS-4f ion probe. (4) The obtained compositions of primary melts were used to model the melting regime of mantle source. The primordial mantle source (Sun and McDonough, 1989) with 3 to 6 wt% of garnet and distribution coefficients for garnet and Cpx from (Johnson, 1994) have been used for calculations.

The following results have been obtained:

1. The ultra-depleted (UDM, La/SmN = 0.3) and ultra-enriched (UEM, La/SmN = 3.2) melts unknown before for mantle plume related tholeiites were found as inclusions trapped in high-Mg olivines (Fo 88-91).

2. The found compositions suggest that melting regime in the mantle plumes may be similar in general to Mid-Ocean Ridges and requires effectively open mantle system with amount of residual melt lower or equal to 1%. Melting requires garnet in source region (melt inclusions have Sm/YbN = 3 - 2). Estimated extend of melting varies from 2 to 8% for instantaneous melts and up to 11-12% for pooled bulk rocks.

3. The natural occurrence of observed melts suggests that at least some primary melts in mantle plumes environment could escape both percolation process and magma mixing.

4. The presence of garnet through entire melting interval suggests that the main differences between melting regimes in Hawaiian plume and Mid-Oceanic Ridges is probably depth of melting and potential temperature, higher for mantle plume.


Budahn, J.R. & Schmitt, R.A., Geochim. Cosmochim. Acta 49, 67-87 (1985).

Johnson, K.T. M., Mineral. Mag. 58A, 454-455 (1994).

Johnson, K.T.M., Dick, H.J. & Shimizu, N. J., Geoph. Res. 95, 2661-2678 (1990).

Sobolev, A.V. & Nikogosian, I.K., Petrology 2, 111-144 (1994).

Sobolev, A.V. & Shimizu, N., Nature 363, 151-154 (1993).

Sun, S.-S. & McDonough, W.F., In Magmatism in ocean basins 313-345 (Geol. Soc. Spec. Publ. 42, 1989).

Watson, S., J. Petrol. 34, 763-783 (1993).

Watson, S. & McKenzie, D., J. Petrol. 32, 501-538 (1991).

Fig. 1: Melt inclusions in olivine from Hawaiian tholeiites, Mauna Loa compared with Hawaiian tholeiitic rocks. All data are normalized to 10 wt.% Al2O3 of primary melt from Sobolev and Nikogosian (1994).