Oxygen Isotope Trajectories of Crystallizing Arc Magmas

Abstract:

Oxygen isotopes are essential to quantify mantle-derived versus ‘recycled’ crustal contributions to arc magmas. High δ¹⁸O values in igneous rocks (i.e., δ¹⁸O_SMOW > ~5.7) are generally used to identify supra-crustal inputs, but a melt can also become enriched in ¹⁸O due to magmatic differentiation [1,2]. To assess magmatic δ¹⁸O values of plutonic rocks, δ¹⁸O_zircon values, which are resilient to secondary alteration, are often used. Thus, to disentangle the effects of assimilation versus fractionation, both the absolute increase in melt δ¹⁸O due to differentiation and ∆¹⁸O(WR-zircon) must be determined. However, existing constraints on the effect of magmatic fractionation on melt δ¹⁸O are model-based [2] and calculated relationships between WR SiO₂, δ¹⁸O_zircon, and δ¹⁸O_melt do not incorporate complex melt SiO₂, H₂O, and temperature (T) relationships [3]. To build upon these initial constraints, we combine the first high-precision δ¹⁸O data set on natural samples documenting changes in δ¹⁸O melt values with increasing extent of differentiation and modeling which incorporates experimentally constrained melt SiO₂, H₂O, and T relationships. We analyzed 55 mineral separates with infrared laser-fluorination [4] across large fractionation intervals of two well-studied cumulate sequences: (I) a relatively dry (~1 wt.% H₂O initial) tholeiitic sequence (analyzed minerals include plag, opx, cpx, & Fe-rich ol) from the Bushveld Complex and (II) a hydrous high-K sequence (analyzed minerals include ol, cpx, bt, fsp, & qtz) from the Dariv paleoarc in Mongolia. Our results indicate that multiple per mil increases in melt δ¹⁸O can occur during magmatic fractionation that in detail depend strongly on melt composition and T. Calculated relationships between WR SiO₂ and δ¹⁸O_zircon for experimental melt compositions show that wet, ‘cool’ and dry, ‘hot’ melts are characterized by larger and smaller ∆¹⁸O (melt-zircon) fractionations, respectively. Applying our results to the O isotope record of the Sierra Nevadas [3], we show that some trends in δ¹⁸O_zircon from individual intrusive suites can be explained by fractional crystallization alone and do not require temporal variations in crustal contamination.

[1] Eiler 2001 [2] Bindeman et al. 2004 [3] Lackey et al. 2008 [4] Wang et al. 2011