Title:
Major element composition of the Hadean crust: constraints from Nd isotope systematics and high-pressure melting experiments

Abstract:
Introduction
The composition of the mantle has changed during formation and recycling of the crust. The behavior and residence time of the crust at surface and in the convecting mantle depend on the density, that in turn depends on the major element composition of the crust. Therefore, the major element composition of the Hadean crust controls the initial condition of the mantle chemical evolution and the mantle dynamics in the Hadean. The major element composition of the Hadean crust also important for the habitable environment, because serpentinization of high-Mg crust yields H for metabolic energy and nutrients such as K and P concentrate in the crust. In order to understand the evolution of the early mantle and the early habitable environment, we should know the major element composition of the Hadean crust.

The only relic of the Hadean crust is the Hadean zircon, and previous studies argued the composition of the Hadean crust from chemical analyses of the Hadean zircon (Harrison 2009; Kemp et al. 2009; Hopkins et al. 2010; Iizuka et al. 2015; Reimink et al. 2016). However, because analyses of the Hadean zircon provide mainly isotopic and trace elemental data, there has been no conclusive constraint on the major element composition of the Hadean crust. Therefore, I adopted a novel and original strategy for the major element composition of the Hadean crust. In this strategy, I used ¹⁴²Nd/¹⁴⁴Nd ratio to determine the melting condition for the formation of the Hadean crust. ¹⁴²Nd is a daughter nuclide of short-lived extinct radionuclide ¹⁴⁶Sm (half-life = 68 Myr). Because Nd concentrates into silicate melts slightly more than Sm, silicate differentiation in the Hadean era, before the extinction of ¹⁴⁶Sm, should have generated low-¹⁴⁶Sm/¹⁴⁴Nd melt (crust) and high-¹⁴⁶Sm/¹⁴⁴Nd residual solid (residual mantle). The extent of differentiation in Sm/Nd ratio depends on melting condition (pressure and melt fraction). Therefore, from difference in ¹⁴²Nd/¹⁴⁴Nd ratio, we can constrain the melting condition for the Sm/Nd differentiation occurred in the Hadean. Then, we can determine the major element composition of the low-¹⁴⁶Sm/¹⁴⁴Nd melt by performing high-pressure melting experiments at obtained melting condition. Difference in ¹⁴²Nd/¹⁴⁴Nd ratio between chondrites and Accessible Silicate Earth (ASE), source mantle of the Mid Ocean Ridge Basalt (MORB) and Ocean Island Basalt (OIB), suggests silicate differentiation in the very early Hadean, > 4.53 Ga, and difference in ¹⁴²Nd/¹⁴⁴Nd ratio between ASE and 3.8 Ga rocks in Isua Supracrustal Belt (ISB) suggests silicate differentiation in the early Hadean, 4.42-4.47 Ga. I investigated what happened during these two silicate differentiations by determining the major element composition of the Hadean crust.

Current research

1. 1. Silicate differentiation >4.53 Ga

Data in previous studies (Boyet & Carlson 2005, Jackson & Carlson 2012) showed that the ASE has higher ¹⁴²Nd/¹⁴⁴Nd ratio than chondrites. When we assume chondritic composition at the formation of the Earth, a low Sm/Nd reservoir (Early Enriched Reservoir, EER) should have been generated >4.53 Ga and have been missing in or out of the Earth, in order to explain the difference in ¹⁴²Nd/¹⁴⁴Nd ratio between chondrites and the ASE. Important questions are, where the EER is and whether the EER formed the Hadean crust or not. I calculated melting condition for the silicate differentiation >4.53 Ga from difference in ⁴²Nd/¹⁴⁴Nd ratio between chondrites and the ASE, by using my novel strategy mentioned above. I found out that near-solidus (melt fraction0%) melting at 7 GPa is required to generate the Sm/Nd differentiation. Then I determined the major element composition of the 7 GPa near-solidus melt with Kawai-type multi-anvil high-pressure apparatus at Institute for Planetary Materials, Okayama University, Misasa, by performing Modified Iterative Sandwich Experiment (MISE, Hirschmann and Dasgupta 2007). The experimental results demonstrated that the major element composition of the 7 GPa near-solidus melt is Fe-, Ti- and alkali-rich komatiite. I calculated density of the 7 GPa near-solidus melt from method of Matsukage et al. (2005), and confirmed that the 7 GPa near-solidus melt has smaller density than mantle peridotite. Therefore, I concluded that the 7 GPa near-solidus melt ascended in the mantle and formed Fe-, Ti-, and alkali-rich komatiitic crust (the EER). A probable situation for the formation of the komatiitic crust (the EER) is near-solidus melting at the base of a thick (~200 km) lithosphere in the stagnant-lid convection proposed by previous studies (Korenaga 2009; Foley et al. 2014), because at the base of the thick lithosphere adiabatically ascending mantle reaches the solidus and starts melting, and then a near-solidus melt segregates immediately after its formation. The komatiitic crust >4.53 Ga (the EER) covered the surface of the early Hadean Earth, and was probably ejected from the Earth by impact erosion before or at the last giant impact (about 4.47 Ga, from Carlson et al. 2014 and Bottke et al. 2015). Thus, the komatiitic crust > 4.53 Ga (the EER) has been lost, and the modern bulk silicate Earth should be depleted than the initial, chondritic composition.

1. 2. Silicate differentiation 4.42-4.47 Ga

Difference in ⁴²Nd/¹⁴⁴Nd ratio between the ASE and 3.8 Ga rocks in Isua Supuracrustal Belt (ISB) suggests silicate differentiation 4.42-4.47 Ga (Carlson et al. 2006; Rizo et al. 2011), and the difference in ¹⁴²Nd^/144Nd ratio has disappeared by 2.7 Ga (Rizo et al. 2013). The decrease and disappearance of difference in ¹⁴²Nd^/144Nd ratio imply subduction and remixing of the Hadean crust, and the mafic and dense crust is easier to subduct than the felsic crust. On the other hand, the existence of the Hadean zircon 4.2-4.4 Ga requires some amount of felsic crust, which is important in supplying nutrients to the surface. I calculated melting condition for the silicate differentiation 4.42-4.47 Ga with the same method as in silicate differentiation >4.53 Ga. The required melting condition was also near-solidus melting at 7 GPa, and therefore, the major element composition of crust is also the Fe-, Ti-, and alkali-rich komatiite. The situation for the formation of the Hadean oceanic crust 4.42-4.47 Ga was the same as that of >4.53Ga, but the Hadean oceanic crust 4.42-4.47 Ga probably subducted and was mixed back with the residual mantle, through intermittent drip-like subduction in the stagnant-lid convection proposed by previous studies (Foley et al. 2014; O'Neill and Debaille 2014). Then, because zircon cannot be crystallized from the komatiite, we need more evolved crust to explain the Hadean zircon. Because existence of surface water has been suggested from the Hadean zircons (Mojzsis et al. 2001), subduction of the komatiitic crust probably caused partial melting under hydrous condition, and generated melt which formed "continental" crust. Therefore, in order to determine the major element composition of the melt, I performed melting experiments of the hydrous Fe-, Ti-, and alkali-rich komatiite at 1-3 GPa, 1000-1300°C, and 1.5 wt% of water content, with Boyd-England type piston-cylinder apparatus at Kyoto University. In these experiments, I carefully controlled and assessed the oxygen fugacity in the capsule. The melt composition was determined to be Ti- and alkali-rich basaltic to picritic. Therefore, the Hadean "continental" crust had mafic composition. From these results, I concluded that the surface environment in the early Hadean was significantly different from the present Earth, because there was no vast felsic crust and was dominantly mafic-ultramafic crust. In my model, zircon-bearing felsic crust was not formed, and other processes are needed.

Plan for next research
From my current research, the unsolved problem is, from when and how much amount the felsic crust has been formed. This problem includes problems that from when and how much amount nutrients have been supplied and how to explain the Hadean zircon. As a solution, I am planning high-pressure melting experiment of the Hadean mafic continental crust to determine the forming condition of the zircon-bearing felsic melt. In the Hadean subduction zone, the Hadean continental crust probably re-melted by the intrusion of anew Ti- and alkali-rich basaltic-picritic melt and zircon-bearing felsic melt could have been generated. Zr saturation in melt can be assessed from Zr content in melt and compositional proxy of Zr saturation (ex. M value = (K+Na+2Ca)/(SiAl), Harrison & Watson 1983; Boehnke et al. 2013). I am panning to synthesize Zr-bearing starting materials of Ti-and alkali-rich basaltic-picritic composition. From melting experiments, we can obtain the major element composition including K and P, value of compositional proxy, and Zr content, of melt, and therefore can constrain the forming condition of the Zr-bearing felsic melt and its K and P content at the same time. Because existing compositional proxy such as M value can be unsuitable for Ti-rich composition, I am also planning revision of the compositional proxy by Zr-saturation experiments in compositional range from Ti-poor to Ti-rich. From this research, I can unite researches about the formation and composition of the Hadean crust and the process to concentrate nutrients from mantle to surface, construct a probable scenario about the Hadean silicate differentiations and its effect to construction of the habitable surface environment.