Jökull - 01.01.2019, Page 96

Eyjafjallajökull ankaramites, South Iceland
Regarding the Brattaskjól olivine zonation, the
most plausible explanation for the detected steps and
changes in the slope of the Fe-Mg zoning patterns in
the Fo>85.4 olivine grains (Figure 2b) is a combina-
tion of solid-state diffusion and episodes of olivine
growth from an evolving host liquid. The rim-growth
events in the normally zoned olivine macrocrysts may
have been caused by step-like ascent of the olivine-
carrying host liquid, episodic changes in the host-
liquid composition, infiltration of liquid into the sys-
tem (i.e., magma mixing), or related to movement of
olivine macrocrysts within a magmatic system with
temperature gradients (see Pankhurst et al., 2018).
The high-Fo bands in the rims of the complex re-
verse zoned Brattaskjól olivine macrocrysts indicate
mixing between an olivine rim forming primitive melt
and the more evolved olivine cores with Fo80−84.4.
Below, we refer to these high-Fo bands in the olivine
macrocryst rims as ’mixing plateaus’. Although the
Fo content in the mixing plateaus varies, three olivine
crystals show mixing plateaus with Fo85.4 composi-
tion (Figure 6a, b and d), and the lowest Fo mixing
plateaus are in-fact not compositional plateaus but ap-
pear as sharp high-Fo tips in the analytical traverses
(Figure 6c and e). This, along with the fact that
all Fo<84.4 olivine macrocrysts have complex reverse
zoned rims, implies that all the mixing plateaus near
olivine rims may have had an original composition
of Fo85.4 and the Fo content in some of the mixing
plateaus has later degreased in response to partial dif-
fusive equilibration with the carrier melt.
We suggest that these Fo85.4 mixing plateaus
formed during a magmatic recharge event in which a
melt (with Mg#melt of ∼63 in equilibrium with Fo85.4
olivine; Toplis, 2005) intruded an olivine-bearing
crystal mush. We propose this "magma-recharge
model" as (i) it offers a simple explanation why the
Fo<84.4 olivine grains are consistently reverse zoned,
(ii) the large amount (∼30 vol%) of compositionally
variable macrocrysts in the Brattaskjól ankaramite
suggests a cumulative origin for the macrocrysts, and
(iii) mixing of magmas and crystal mushes is com-
mon in Icelandic magmatic systems (Halldorsson et
al., 2008; Neave et al., 2013; Halldórsson et al.,
2018). In particular, the consistent change from com-
plex reverse to normally zoned olivine at Fo84.4−85.4
(Figure 2a), and the varying offset in Fo content (1–
3 mol%) between mixing plateaus and crystal cores,
support a magma-recharge origin for the complex re-
verse zonation in olivine macrocrysts. Alternatively,
the complex reverse zoned olivine macrocrysts could
have been formed during their movement in a crustal
intrusion with temperature gradients (cf. Pankhurst et
al., 2018). Considering the low number of analysed
crystals, our data cannot disprove, but neither particu-
larly support, this mode of formation.
Assuming that our hypothesis of the magma
recharge origin for the complex reverse zoned crys-
tals is valid, we can utilize diffusion modelling (Costa
et al., 2008; Zhang and Cherniak, 2010) of the chem-
ical re-equilibration in the Fe-Mg profiles (e.g., Fig-
ure 2d) to constrain the time frame within which the
Brattaskjól complex reverse zoned olivines cooled af-
ter the recharge event. We did this by using the fi-
nite difference diffusion code of Kahl et al. (2015)
that follows the procedures outlined in Costa and
Chakraborty (2004) and Costa et al. (2008), with
Fe-Mg inter-diffusion coefficients of Dohmen et al.
(2007) and Dohmen and Chakrabortny (2007). As the
pre-diffusion initial state in our model crystals, we as-
sumed the measured Fo contents in olivine cores, and
a variably thick (9–26µm) Fo85.4 mixing plateau for
the olivine crystals (stippled lines in Figure 6). We
only modelled the time of diffusive re-equilibration
between the mixing plateau and olivine-core, not be-
tween the mixing plateau and the outer crystal rim,
because the outermost rims have likely been formed,
at least partly, by crystallization from cooling and
evolving host-magma after the magma recharge, not
by solid-state diffusive re-equilibriation. Some late-
stage crystallization in the outermost macrocryst rims
is to be expected considering the crystalline ground-
mass (Figure 1b) and as the macrocrysts are hosted by
a comparatively slowly cooled lava.
In diffusion modelling, we used a fO2 of
FMQ+0.5, temperature of 1170◦C and pressure
of 3.0 kbar, corresponding to the mean conditions
of clinopyroxene crystallization in the Brattaskjól
ankaramite (Figure 5). Varying the fO2 and pres-
sure has only a minor effect on the model results (e.g.,
JÖKULL No. 69, 2019 95