Experimental constraints on the formation of mafic enclaves vs. mafic layers: Why are some silicic plutons layered and others not?

Mafic sheets and enclaves preserved in silicic intrusions may record the injection, stirring, and breakup of dikes entering a convecting magma chamber. Understanding the physical processes that govern enclave formation is a central issue in volcanology and can explain one mechanism by which magma chambers grow and differentiate. Field evidence of such injections can take many forms and highlight the potential complexity of the mafic-silicic interactions (e.g. mafic enclaves vs. mafic layers). Using analog experiments in which buoyant particle-fluid mixtures are injected into a shear flow, we identify the conditions in which mafic injections will pond as layers or breakup as enclaves. Qualitatively, breakup can occur either through the growth of a Rayleigh-Taylor instability or by disaggregation of the mixture (yielding) under tension. Quantitatively, these regimes are governed by two parameters: The ratio S of the timescale for settling through the fluid layer to the timescale for the growth of a R-T instability, and the ratio Y of the timescales for shearing and yielding in tension. If Y is less than a critical value Yc, the injection is in the Tension regime and breaks up in axisymmetric blobs with a diameter smaller than the injection size. Alternatively, if Y>Yc the injection is in the R-T regime and breaks up with lengthscales similar to the size of the injection. At high S and Y conditions, the injections pond at the floor and undergo negligible deformation. Scaling theories for the length scale of deformation in the Tension and R-T regimes predict that enclave sizes in natural settings will either be comparable to, or much smaller than the injection size. Alternatively, observations of the size distribution of natural enclaves can be applied to constrain aspects of the chamber dynamics such as the convective stirring rate as well as the magma rheology. Our results have implications for the dynamics of magma mixing: Dike intrusions that break up in the Tension regime are likely to be more readily stirred in to a convecting magma chamber than dike intrusions in the R-T or ponding regimes. This result leads to predictions for why mafic-silicic layered intrusions might form and also for why some batholiths are relatively homogeneous.