Successive episodes of reactive liquid flow through a layered intrusion (Unit 9, Rum Eastern Layered Intrusion, Scotland)

Abstract

We present a detailed microstructural and geochemical study of reactive liquid flow in Unit 9 of the Rum Eastern Layered Intrusion, Scotland. Unit 9 comprises an underlying lens-like body of peridotite overlain by a sequence of troctolite and gabbro (termed allivalite), with some local and minor anorthosite. The troctolite is separated from the overlying gabbro by a distinct, sub-horizontal, undulose horizon (the ‘major wavy horizon’). Higher in the stratigraphy is another, similar, horizon (the ‘minor wavy horizon’) that separates relatively clinopyroxene-poor gabbro from an overlying gabbro. To the north of the peridotite lens, both troctolite and gabbro grade into poikilitic gabbro. Clinopyroxene habit in the allivalite varies from thin rims around olivine in troctolite to equigranular crystals in gabbro and to oikocrysts in poikilitic gabbro. The poikilitic gabbros contain multiple generations of clinopyroxene, with Cr-rich (~1.1 wt% Cr₂O₃) anhedral cores with moderate REE concentrations (core1) overgrown by an anhedral REE-depleted second generation with moderate Cr (~0.7 wt% Cr₂O₃) (core2). These composite cores are rimmed by Cr-poor (~0.2 wt% Cr₂O₃) and REE-poor to -moderate clinopyroxene. We interpret these microstructures as a consequence of two separate episodes of partial melting triggered by the intrusion of hot olivine-phyric picrite to form the discontinuous lenses that comprise the Unit 9 peridotite. Loss of clinopyroxene-saturated partial melt from the lower part of the allivalite immediately following the early stages of sill intrusion resulted in the formation of clinopyroxene-poor gabbro. The spatial extent of clinopyroxene loss is marked by the minor wavy horizon. A second partial melting event stripped out almost all clinopyroxene from the lowest allivalite to form a troctolite, with the major wavy horizon marking the extent of melting during this episode. The poikilitic gabbro formed from clinopyroxene-saturated melt moving upwards and laterally through the remobilized cumulate pile and precipitating clinopyroxene en route. This process, called reactive liquid flow, is potentially important in open magma chambers.

Appendix 1: Quantification of successive reactive liquid flow episodes in a layered intrusion (Unit 9, Rum Eastern Layered Intrusion, Scotland)

Despite the fact that our model accounts for every observation so far, we are aware that the proposed quantitative model is not necessarily unique and different proportions of gabbro partial melting might also explain the cumulate evolution. Quantification is made especially difficult because of the superimposed successive partial melting events. Also the mineral saturation temperature and proportion rely on MELTS calculations, considering the Rum picrite liquid line of descent, but without knowing the exact pressure, water content and oxygen fugacity. The accuracy of MELTS in calculating crystallisation paths for these liquids is also unknown and requires some experimental verification.

Timing of picrite sills intrusions and solidification

Using the Θ _cpp variation as a function of calculated crystallization times (Holness et al. 2012b), we can estimate the solidification timescale varying from 200 years in the basal clinopyroxene-poor gabbro, to 100 years in the gabbro body (4.5–11.8 m and the lower troctolite (−9.15 to −4.75 m), to less than 10 years in the central troctolite (−3.8 to −1.15 m).

Estimation of the protolith volume

We have demonstrated that the hybrid melt, generated by mixing of partial melt with invading reactive liquid was expelled from the heated allivalite. We have approximated the thickness of the original gabbro layer, prior to the two partial melting events, using equation 1.

$$ T_{ 2} = \left( { 100 - F} \right) \cdot T_{ 1} + L \cdot F \cdot T_{ 1} $$

(1)

With T ₁ : the original thickness (m), T ₂ : the residue thickness (m), F: the partial melting fraction (%) and L: the crystallized trapped interstitial melt fraction (%).

The 10-m-thick troctolite (T ₂ ) corresponds to the residue after 72 % partial melting (F) of a ~24-m-thick (T ₁ ) clinopyroxene-poor gabbro in which ~20 % hybrid interstitial melt has crystallized (L). Also, the clinopyroxene-poor gabbro (~26 m, T ₂ ′) was produced by crystallization of ~20 % trapped liquid (L′) in a partially molten (62 %, F′) ~ 52-m-thick gabbro column (T ₁ ′). The original gabbro column was ~67 m thick, but only 15 m remain today. The expelled liquid most likely was expelled into the overlying bulk magma in the chamber. Peridotite cumulates have also lost abundant interstitial melt. We have estimated the trapped liquid to have been ~40 vol% in the major peridotite body. Assuming the two subsequent sills had an identical thickness and composition, we propose 20 % olivine-phyric picritic magma emplaced as two 15-m-thick sills.

Thermal modelling

We have calculated the gabbro 1D thermal evolution when intruded by two olivine-phyric picrite sills, using Eqs. 2 and 3 (from Furlong et al. 1991), in a grid of i columns and j lines. For all nodes within the solution domain:

$$ \kappa = \frac{k}{c\rho } $$

(2)

$$ T_{i,j + 1} = \frac{\kappa \Delta t}{{(\Delta x)^{2} }}\left[ {T_{i + 1,j} - 2T_{i,j} + T_{i - 1,j} } \right] + T_{i,j}. $$

(3)

We used the constants of Furlong et al. (1991) for gabbro and peridotite respectively: κ the thermal diffusivity (m² s⁻¹), k the thermal conductivity (2.63–3.81 W m⁻¹°K⁻¹), c the heat capacity (897 and 1,000 m² s⁻²°K⁻¹), ρ the density (2,900–3,300 kg m⁻³), t the time [86,400 s (10 days)], x the distance (1.5 m) and T the temperature (°K). The boundary temperatures (T _i=0 and T _i = max) are maintained constant at the host rock temperature (1,160 °C).

Intrusions of two successive 15-m-thick olivine-phyric picrite at 1,240 °C, at an interval of 200 years, heat the host gabbro to 1,190 °C (i.e. plagioclase saturation) at 4.5 m, 1,180 °C (i.e. clinopyroxene saturation) at 15 m and 1,165 °C (i.e. gabbro with less than 25vol % clinopyroxene) at 84 m (Appendix Figure 1 in ESM). The maximal temperature at 10 m from the contact is reached ~30 years after each intrusion. The discrepancy between this simple model and the above volume estimates is largely covered by the model uncertainties and estimations. This model shows clinopyroxene, and also plagioclase close to the sill intrusion, are unstable, possibly producing a dunite restite. Thus, part of the peridotite cumulate might have been generated by gabbro partial melting and post-compaction crystallization of trapped hybrid liquid. O’Driscoll et al. (2007) suggested unconsolidated crystals could have slumped during central sagging of the Rum layered suite. If picrite sills crystals moved after their emplacement in the Unit 9, the volume of hot magma might be underestimated in our calculations, with important effects on the isotherm locations.