Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust

TL;DR: In this paper, the average chemical compositions of the continental crust and the oceanic crust (represented by MORB), normalized to primitive mantle values and plotted as functions of the apparent bulk partition coefficient of each element, form surprisingly simple, complementary concentration patterns.

About: This article is published in Earth and Planetary Science Letters.The article was published on 1988-11-18 and is currently open access. It has received 3062 citations till now. The article focuses on the topics: Oceanic crust & Continental crust.

Summary

1. Introduction

• In order to test and extend the inferences made from isotopic studies of very few trace elements.
• The result is a surprisingly simple and consistent pattern of complementary enrichment and depletion for nearly all lithophile elements for which there are analytical data.
• This pattern is adequately explained by a simple two-stage extraction model of continental and oceanic crust from an initially primitive mantle.
• This global crust-mantle differentiation was not, however, the primary process that generated the present-day, intra-oceanic mantle heterogeneities.

2. Element abundance patterns

• Klein and Langmuir [1] have shown that Na20 concentrations in MORB are well correlated with the depths of the axial valley floors of the spreading ridges.
• This indicates that the average composition of their sample set is within 10% of the true MORB average.
• These differences are small enough, so that they may be neglected for the purpose of the present paper.

2.1. Compatibility sequence and general properties of the MORB pattern

• (4) The standard deviations of the average concentrations are small for major elements and increase systematically toward the left hand side of the diagram, as the elements become more incompatible.
• (A similar increase toward the right, affecting mostly Ni and Cr, is known to be caused by variable fractionation of olivine and chromite in the basalts and is not further considered here.).

3. Extraction model

• The highest concentration that can be attained by such a stage-2 melt is the same as the maximum attained during stage 1.
• This means in general that the residual mantle is incapable of producing second-stage melts with element concentrations exceeding the maximum concentrations found in the continental crust, unless there is another enrichment process, which either pre-enriches the source or "post-enriches" the melt (for example by crystal fractionation).
• It seems likely that the very high trace element enrichments found in many alkali basalts of oceanic islands require such a pre-enrichment.

A quantitative model for the composition of MORB.

• In the simple model discussed so far, it has been assumed that the bulk partition coefficient is the same for each element during stages 1 and 2, respectively.
• For the major and the moderately incompatible elements, the fit of the MORB data in Fig. 5 is surprisingly tight.
• The scatter is greater for the highly incompatible elements, but this is expected because of their intrinsically greater variability, the uncertainties of the partition coefficients, and the fact that other effects such as retention of melt during stage 1 or fractional crystallization during stage 2 have all been neglected.
• This may be compared with experimental data as a check for the internal consistency of the model.
• Thus a residual mantle assemblage containing 10% clinopyroxene and no other REE-bearing minerals would have a bulk partition coefficient between 0.03 and 0.1.

3.2. Constraints on melt fraction during MORB production

• If the two-stage melting model is replaced by multiple stages, the final melt fractions must become smaller, if the same abundance maximum is to be maintained in the final MORB production stage.
• If the authors relax the constraint that the depletion is produced by previous production of the continental crust, and retain only the constraint of the maximum enrichment factor Cz* = 9 in MORB, the upper limit for the melt fraction producing MORB is still only F 2 = 0.11.
• Slightly larger melt fractions are obtained for more efficient extraction mechanisms such as fractional melting and the continuous extraction discussed by Ribe [22] .
• These more detailed considerations are beyond the scope of this paper.

3.4. Trace element enrichment of MORB through fractional crystallization

• The fractionation model used is that of O'Hara [23] for a steady-state magma chamber which is periodically refilled, tapped and fractionated.
• The steady-state fractionated melt shows a concentration curve which overlaps the stage-1 melt at D values between 0.1 and 0.5.
• (A significant decrease in partition coefficients during stage 2 was another way discussed earlier.).
• This increases the concentrations of the most highly incompatible elements by a factor of three (i.e. much higher than inferred from the small negative Eu anomaly actually found in average MORB).
• This purposely exaggerated example illustrates how fractional crystallization can, in principle, cause the observed crossover of MORB and continental abundances.

SOblrCt2

• These elements constitute a portion of the right-hand descending limb of the MORB abundance curve (Fig. 1 ), and their abundances are substantially lower than the maximum of that curve.
• In these cases the degree of melting is indeed high enough so that clinopyroxene has been consumed by the melt.
• Observations on incompatible element concentrations and ratios are biased toward the lower part of the column, where melt fractions are smaller.
• Most authors (e.g. [5, 28] ) believe that the basalts richest in MgO, or highest in Mg/Fe, represent the most "primitive" melts and are therefore also the best representatives of the parental liquids of the more evolved, less magnesian basalts.
• By avoiding the bias toward high melt fractions, one must accept another: the "garden variety" MORB have almost certainly evolved to some degree by fractional crystallization.

4.2. Variation of the element abundances in MORB

• These considerations lead to the expectation that the bulk partition coefficients of the elements should be negatively correlated with the standard deviations of the concentrations in MORB, and this is indeed observed in Fig. 2 .
• The most noticeable exceptions in this general trend are the anomalously low standard deviations of the Pb, Sr, and Na concentrations.
• Part of this effect is caused by variable fractionational crystallization involving plagioclase, which will cause the REE and other incompatible element concentrations to vary, while at the same time buffering Na, Sr, and possibly also Pb because of its high partition coefficient for these elements.

4.3. Alternative compatibility sequence based on oceanic basalts

• Fig. 8 shows the standard deviations of the MORB concentration averages plotted in the compatibility sequence given in Fig. 7 .
• The decrease in variability with increasing compatibility is now much more nearly monotonic than in Fig. 2 .
• The slight negative anomalies for Pb, Sr, and Eu are probably caused by buffering of these elements during fractional crystallization (see previous section).
• All other anomalies (except for Cu, which remains unexplained), may be ascribed to analytical uncertainties or uncertainties in the exact element sequence.

5.1. Mineralogical causes of the compatibility switch

• In summary, the specific mechanisms causing the special behavior of Nb, Ta, and Pb are not yet well understood.
• It seems clear that different mineral phases are responsible for the anomalous behavior of Nb and Ta on the one hand, and of Pb on the other.
• It is certainly not surprising that chemical elements exist which behave very differently in (presumably hydrous and relatively deepseated) subduction processes than in (compara-tively dry and shallow) ocean ridge processes.
• To this author, it is much more surprising that all the other chemical elements conform to the simple and complementary relationship displayed by Figs. 1 and 7.

6. Conclusions

• The batch melting equations used to construct the model are not meant to imply that the continental or oceanic crusts were actually produced by simple batch melting.
• Rather, they demonstrate that a satisfactory global mass balance exists for the chemistries of these crusts, the residual mantle, and the initial primitive mantle.
• Such a balance had previously been shown to exist only for the combined Rb-Sr and Sm-Nd abundances and isotopic compositions, and the results had been in apparent conflict with the isotopic data for Pb.
• This conflict is resolved by the dual partitioning behavior of Pb.