Depth of Origin of Hawaii Basalts: Discussion

Gillian Foulger, 12th June, 2011
Dear David, I note that in your AGU 2010 abstract, you claim that the depth of OIB extraction at Hawaii is 1-2.5 GPa. In their recent J. Petrology paper, Presnall & Gudfinnsson (2011) argue that a depth of origin of Hawaii tholeiites of 4-5 GPa can be concluded. They also argue that the alkalic lavas that sandwich (physically and chronologically) the tholeiites (i.e., the pre-tholeitic alkalics, and "post-erosional" alkalics) come from similar depths as well. Where does this major disagreement creep in?

David Green, 7th July, 2011
Dear Gillian, The AGU abstract is based on material published in Green & Falloon (2005) [GF], Falloon et al. (2007a) [FGD], Falloon et al. (2007b) [F et al.] and the papers Green et al. (2010, 2011). To focus the discussion I will refer to Figure 2 of Presnall & Gudfinnsson (2011) [PG] and to Figures 1, 2 & 3 of Green & Falloon (2005) [GF].

We agree on the position of the volatile-free lherzolite solidus, although GF select one from three solidi differing in their "fertility" (mainly (Na+K)/Ca), i.e., Hawaiian pyrolite > MOR pyrolite > Tinaquillo lherzolite. We differ slightly with mantle potential temperature, PG using 1500°C compared with GF 1450°C (Figures 1 & 2) or 1430°C (Figure 3). Although PG illustrate in Figure 2a a geotherm which is a little over 1600°C at 10 GPa, and matches their T_P = 1500°C adiabat beneath Hawaii, they refer to it as a "perturbed oceanic geotherm". The perturbation relative to the adiabat and Hawaiian geotherm is to cooler temperatures at intermediate depths and to higher temperatures at < 4 GPa approx. This has the effect of placing the MOR geotherm through the maximum on the Peridotite-CO₂ solidus giving a “lid” to the asthenosphere at ~ 70 km, which they also indicate as the depth of MORB extraction. I am unclear as to the reasoning to derive the different geotherms in PG’s Figures 1a,b and consequently unclear whether PG accept that near-adiabatic upwelling occurs beneath MORs, rifts, or "hotspots", and that such upwelling leads to increasing melt fraction up to a melt-extraction process at some melt fraction (a permeability/porosity issue for the melt + residue system). The issues of melt fraction, melt extraction process, latent heat of fusion are not addressed in PG, I think.

In GF, these are expressed in the adiabat* intersecting the volatile-free solidus (GF Figure 1) or 2% melting contour for MOR Pyrolite+200 ppm H₂O+100 ppm CO₂ (GF Figure 3a). Continued upwelling results in increased melt fraction and also in cooling below the solid adiabat because of latent heat of melting. Melt segregation from residue and extraction as a picritic liquid is located at 1.5-2 GPa at ~ 15-20% melt fraction, liquids ascending on a liquid adiabat (GF Figure 1) or cooler if they react with wallrocks. GF draw their conductive and boundary layer geotherm at shallower depths (oceanic geotherm) based on the Clark & Ringwood (1964) oceanic and continental geotherms (heat flow, conductivity and heat production arguments) of which the oceanic geotherm passes through the lherzolite+H₂O solidus at pargasite breakdown (i.e. 3 GPa, 1000-1100°C; GF Figures 2 & 3). Also see Green & Liebermann (1976) p 61 for a comparison of this geotherm with oceanic geotherms beneath 100 Ma oceanic crust based on plate cooling models.

(*represents the assumption that as a convective system, the upper mantle has achieved a "steady state" temperature gradient – the adiabatic gradient – and that departures from this leading to upwelling are small, in contrast to the driving element, the cold downwelling slabs.)

 In my reading of their paper, PG draw their conductive and boundary layer geotherms differently for their "mature LVZ" (Hawaii) and MOR settings. They do this because they want to fit two petrological constraints which they believe are sound. They recognise picritic glasses at Hawaii and olivine control on early stages of primitive melt fractionation. They base their location of depths of origin of OIB on picritic parental magmas and inferred residual garnet in melt extraction. Based on the study of the simple system CMAS and extrapolation into the natural system (mainly the effect of Na, Fe, Ti) and, importantly, to ensure that their melts at extraction were equilibrated with garnet lherzolite residue, they place the melt extraction at the garnet lherzolite solidus at 3.5-5 GPa. They do not discuss melt fraction but call the segregated melts "at the solidus" and bring them to the surface without upwelling of lherzolite. So the constraints on their "mature LVZ" geotherm are the intersection of the 1500°C adiabat with the volatile-free lherzolite solidus and their wish to intersect the lherzolite+CO₂ solidus at its sharp inflection at ~2 GPa (due to the reaction Ol+Di+CO₂->Opx+Dolomite eliminating CO₂ at higher pressure in favour of dolomite-bearing lherzolite).

For their MOR settings, PG refer to the geotherm as "perturbed" and again have two constraints to locate it. PG do not accept that there are parental MOR glasses which are saturated with olivine alone at low P and discard examples of magnesian olivine phenocrysts in such glasses as exotic and accidental xenocrysts. They thus reject the olivine-addition calculations of GF, FGD and F et al. which lead to MOR picrite melts and depths of melt segregation of ~2 GPa. Instead PG place parental/primitive MORB as melts extracted at the plagioclase-lherzolite to spinel-lherzolite transition at 1-1.5 GPa. Such liquids would be saturated with Ol+Opx+Cpx+Sp±Plag at ~ 1 GPa and they would segregate very close to the solidus. Thus, their geotherm in Figure 2a passes through ~1 GPa, 1260°C. It also passes through the maximum on the lherzolite+CO₂ solidus, fitting a 60-70 km "lid" to the LVZ near the ridge. Since both MOR and mature geotherms converge on the adiabat at 10 GPa in Figures 2a & 2b there is an unexplained situation in which the mature geotherm is cooled from ~8 to 5.5 GPa and heated from 5.5 GPa to < 0.5 GPa. How this fits into plate tectonics, ridge jumps, ridge migration etc is not explained.

Two major criticisms of PG are:

1. PG completely discount the effect of water on lowering the volatile-free lherzolite solidus and attribute the major role for CO₂ and subsolidus carbonate. Water in the upper mantle is placed into Nominally Anhydrous Minerals (NAMs) as dilute solid solutions having negligible effect on melting or phase stability, particularly pargasite at P < 3 GPa. No account is given to the observation that MORB glasses, including "popping rocks" have H₂O > CO₂. Particularly important are the observations and arguments that in Hawaii H₂O > CO₂ and the primitive picrite glasses used by PG have ~5000 ppm water, lowering their liquidus temperatures by ~ 60°C and arguing for mantle source water contents of 500 ppm (10% melt fraction) or 1000 ppm H₂O (20% melt fraction – see GF p. 227, final para). Although Green et al. (2010) is referenced by PG there is no recognition that this study showed that the limit on water storage in NAMs at 2 to 4 GPa is ~ 200 ppm H₂O, and that at this water content in NAMs, pargasite is stable at 2.5 GPa and forms the major water storage site in the uppermost mantle. At > 3 GPa the vapour-saturated solidus (water-rich vapour) has only 200 ppm water in NAMs at the solidus and any higher water content forms melt or vapour above or below the solidus. PG do not consider the role of pargasite whereas this new study validates GF Figures 1, 2 & 3 extremely well, particularly the roles of small quantities of both water and CO₂, of pargasite, carbonatite and carbonate-bearing silicate melts in the asthenosphere.
2. The second major criticism of PG (and of earlier papers which consider CO₂ in the mantle without adequate attention to the role of H₂O) is consideration of the role of fO₂ in the lherzolite+C+H+O system. This is outlined in Figure 2 of GF where the carbonatite+pargasite+garnet lherzolite field is shown, with carbonatite melt being present (but only below the lherzolite+H₂O solidus) if fO₂ is around iron/wustite (IW)+3,4 log units. If the asthenospheric or deeper mantle is at lower fO₂ (IW+1,2 log units) then graphite/diamond and H₂O-rich fluid is stable below the solidus and the solidus which is relevant is the lherzolite+H₂O solidus. At higher T, the CO₂ is dissolved in the hydrous silicate melt (olivine melilites to olivine basanites) – the incipient melting regime of GF, Figure 2. The types of melts in this region are outlined in GF Figure 3b (circled numbers 1-5), and match the alkalic melts of Figure 2b of PG but are produced at much lower temperature because of their water and CO₂ contents. In GF Figure 2 the carbonatite field "fingers out" in patchy fashion representing inhomogeneity in fO₂ but overall decrease in fO₂ with depth. Many have argued for lowered fO₂ at deeper levels of the upper mantle, e.g., Foley (2011), Stagno & Frost (2010), Rohrbach et al., (2011) and references therein going back to the mid-1980s. At fO₂ ~IW+1,2 log units, the lherzolite+C+H+O is graphite/diamond+H₂O-rich vapour at subsolidus conditions with melting beginning at the lherzolite+H₂O solidus. Melts are olivine melilitite to olivine nephelinite with dissolved (CO₃)^2-

Thus, on the grounds of both the role of water and the mantle oxidation state, I think the PG paper is incorrect. The most recent work on lherzolite+H₂O (+ trace CO₂) to 6 GPa – Green et al. (2010, 2011, including the Addendum), and work currently in preparation for publication, all fit well with a mantle Tp of ~1430-1450°C, and a MOR pyrolite or HZ1 lherzolite source composition for MORB, upwelling from ~ 250-300 km depth beneath MORs to melt segregation (MOR picrites) at ~2 GPa. Intraplate (including rift) basalts segregate at temperatures below the 1430°C adiabat as in Figure 3b of GF, in large part because of the higher H₂O+CO₂ contents of intraplate basalts.

The very important additional feature with respect to "hot spots" is mantle heterogeneity in the upper mantle, probably from suspended or buoyant old subducted slabs, introducing redox contrasts and slab-sourced refertilization of mantle. The source(s) of Hawaiian magmas are not well mixed and reflect mantle heterogeneity in which the geochemical signatures of arc and continental sources/processes, mantle refertilization from eclogite partial melting, and intraplate asthenospheric melting can be discerned. However in the compositions of most primitive, mantle-derived melts, and the P,T conditions at which they were multiply saturated in Ol + Opx±Cpx, Ga, Sp (i.e., inferred depth of magma segregation from residual mantle and after which such magmas moved in dykes or dunite channels to eruption or crustal magma chamber evolution) both MORB and OIB are consistent with T_P ~ 1430°C and depths of magma segregation of 50-70 km.

• Clark, S.P., A.E. Ringwood, (1964), Density distribution and composition of the upper mantle. Reviews of Geophysics, 2, 35-88.
• Falloon, T.J., D.H. Green, L.V. Danushevsky, (2007a), Crystallization temperatures of tholeiite parental liquids: implications for the existence of thermally driven mantle plumes, in ‘The Origins of Melting Anomalies: Plates, Plumes, and Planetary Processes, Eds. G.R. Foulger, D.M. Jurdy, Geological Society of America Special Paper 430, 235-260. [FG]
• Falloon, T.J., L.V. Danushevsky, A. Ariskin, D.H. Green, (2007b), The application of olivine geothermometry to infer crystallization temperatures of parental liquids: implications for the temperature of MORB magmas. Chem. Geol., 241, 207-233. [F et al.]
• Foley, S.F., (2011), A Reappraisal of Redox Melting in the Earth’s Mantle as a Function of Tectonic Setting and Time, J. Petrology, first published online November 3, doi:10.1093/petrology/egq061
• Green, D.H., Liebermann, R.C., (1976), Phase equilibriums and elastic properties of a pyrolite model for the oceanic upper mantle, Tectonophysics, 32, 61-92.
• Green, D., T.J. Falloon (2005), Primary magmas at mid-ocean ridges, "hotspots", and other intraplate settings: constraints on mantle potential temperature, in Plates, Plumes and Paradigms, Eds. G.R. Foulger, J.H. Natland, D.C. Presnall, D.L. Anderson, Geological Society of America Special Paper 388, 217-247. [GF]
• Green, D.H., T.J. Falloon, (2010), Plate tectonics constrained by evidence-based magmatic temperatures and phase relations of fertile lherzolite, Abstract V21A-2320 presented at 2010 Fall Meeting, AGU, San Francisco, Calif., 13-17 Dec.
• Green, D.H., W.O. Hibberson, I. Kovacs, & A. Rosenthal, (2010), Water and its influence on the lithosphere–asthenosphere boundary, Nature, 467, 448-452, doi:10.1038/nature09369. Click here for supplementary material.
• Green, David H., William O. Hibberson, István Kovács & Anja Rosenthal, (2011), Water and its influence on the lithosphere–asthenosphere boundary, Nature, 472, 504, doi:10.1038/nature09948
• Presnall, D., G. Gudmundsson, (2011), Oceanic volcanism from the low-velocity zone without mantle plumes. Journal of Petrology, in press, 10.1093/petrology/egq093. [PG]
• Rohrbach, C., P.U. Ballhaus, U. Golla-Schindler, D. Schonbohm, (2011), Experimental evidence for a reduced metal-saturated upper mantle. Journal of Petrology, 52, 717-731
• Stagno, V., D.J. Frost, (2010), Carbon speciation in the asthenosphere: Experimental measurements of the redox conditions at which carbonate-bearing melts coexist with graphite or diamond in peridotite assemblages, Earth Planet. Sci. Lett., 300, 72-84.