Origin of kimberlite from the base of the upper mantle

: 2 3 Kimberlite is characterized by explosive eruption powered by excess carbon dioxides 4 (CO 2 ) 1 and water 2 . Given that diamond is the dominant stable phase of carbon in the 5 upper mantle 3 , it is obscure where does the excess CO 2 in kimberlite has come from. 6 Here we show that ferric iron oxidizes diamond at 1900K, 20GPa and 2000K, 25GPa, 7 forming CO 2 . The lower mantle is dominated by bridgmanite, which is rich in ferric 8 iron 4 . Bridgmanite decomposes once it is brought to the upper mantle, releasing extra 9 ferric iron. Therefore, the oxidation of diamond may have been popularly occurring 10 at the base of the upper mantle, forming CO 2 -rich carbonated domains that are the 11 main source of kimberlite. The rising kimberlitic magma reaches the lithosphere 12 mantle of thick cratons before it crosses the solidus line of mantle peridotite, and thus 13 keeps its volatile-rich nature that drives explosive eruptions. When the lithospheric 14 mantle is thinner than ~140 km, kimberlite changes into much less explosive magmas 15 due to partial melting of mantle peridotite, and, consequently, entrained diamond is 16 mostly oxidized during the magma’s slower ascension. 17

However, cratons are old (e.g., >2.5 billion years), whereas kimberlites are young (mostly 23 < 0.6 billion years), suggesting that kimberlite is not necessarily coupled with cratons. Plate 24 reconstruction shows that most of the kimberlite were located above one of the two large 25 low shear wave velocity provinces in the lower mantle, suggesting their genetic 26 connections with mantle plumes 7 . Kimberlite has several percent of CO2 and also H2O, 27 which are the main components that make kimberlite explosive 1,2,8,9 . However, cratons 28 have been highly devolatized, such that the roots of ancient cratons are generally dry 10-13 , 29 which is essential to maintain cratons stable for billions of years. Moreover, carbonate 30 cannot be stored in the asthenospheric mantle at depths shallower than 300 km (Figure 2), 31 because the solidus line of carbonated peridotite is lower than the mantle geotherm 14 . 32 Diamond, rather than CO2, is the dominant carbon species in the upper mantle [15][16][17] . 33 Therefore, conventional models have difficulties to explain the source of large quantities 34 of excess CO2 in kimberlites. Previous studies showed that oxidation of diamond 18 through 35 redox melting is required to form CO2 in mid-ocean ridge basalts 3,19 . Under normal mantle 36 condition, redox melting occurs at depths between 150 to 120 km. For mid-ocean ridge 37 basalts, 30 ppm of C in average should have been oxidized to carbonate at the expense of 38 3% of Fe 3+ /Fe Total in the upper mantle 3 . However, such redox melting has complications to 39 explain the high abundance of carbonate and excess CO2 in kimberlite.

40
First of all, formation of kimberlite through redox melting will require ferric iron 41 abundance more than 100 times higher than the estimated value of the upper mantle.

42
Second, the oxygen fugacity decreases with increasing depths 20,21 , and diamond is the 43 stable carbon species deeper than 170 km in the upper mantle. Therefore, carbonate-44 induced melting may occur but only at shallow depths of 150-120 km 3 , whereas kimberlite 45 forms mostly deeper than 140 kilometers 3,5 . In addition, both the reductant (diamond) and 46 the oxidizing agent (ferric iron) are solid or hosted in solid minerals that are not readily 47 mobile. It is difficult to concentrate the scarce diamond into kimberlite during dispersed 48 small degree redox melting. More importantly, the key point of redox melting is that 49 elemental carbon (diamond or graphite) is oxidized within upwelling. Therefore, the degree 50 of redox melting increases with increasing upwelling distance. In this case, the thicker the 51 lithosphere, the less the redox melting, i.e., melts underneath cratons should have less CO2.

52
This is exactly the opposite of the distribution of kimberlite.

53
The best place to oxidize diamond is the base of the upper mantle. The lower mantle 54 is rich in ferric iron because of ferrous iron disproportionation that forms native and ferric 55 iron 4 . Native iron is much denser and thus sinks towards the core. Ferric iron is hosted in

72
The reaction equation is as follows: According to Equation (1), the total amount of cation decreases while the total amount 75 of anion increases during the reaction, resulting in excess CO2.  here after). All the kimberlites plot in the mixing field among these endmembers.

85
Considering that the cratonic mantle is highly depleted in terms of Sr-Nd isotopes, the Sr-

86
Nd isotopes suggest that a large proportion of kimberlites were not originated from the 87 cratonic mantle.

88
The continental crust endmember may be plausibly explained by recycling of 89 subducted sedimentary carbonates, whereas the depleted mantle endmember is attributed 90 to assimilation of the depleted mantle by kimberlites during the upwelling.

119
The upwelling of hot kimberlite melt may cause partial melting of mantle peridotite at