Jökull - 01.01.2019, Page 93

Nikkola et al.
DISCUSSION
Oxygen fugacity, pressure and temperature esti-
mates
We used olivine-spinel pairs to determine oxygen
fugacity (fO2) during olivine crystallization using
the most recent calibration (Nikolaev et al., 2016)
of the Ballhaus-Berry-Green olivine-orthopyroxene-
spinel oxybarometer (Ballhaus et al., 1991; Beattie,
1993). Depending on the size of the spinel grains,
up to three EPMA analyses were performed, with
two to three analyses from the host olivine at least
∼75µm from the spinel and along the same growth
zone of the olivine grain. Assuming spinel-olivine
co-crystallization at 3.0 kbar and 1230◦C, these cal-
culations yielded ∆logfO2(FMQ) values of 0.1±0.5
and 0.5±0.5 for Hvammsmúli and Brattaskjól, re-
spectively (Figure 5a). This is in line with earlier es-
timates of the magma oxidation state at Vestmanna-
eyjar (Steinthorsson 1972; Gerlach, 1980; Schipper
and Moussallam, 2017) and in the Eastern Volcanic
Zone of Iceland (Hartley et al., 2017).
We utilized clinopyroxene-liquid thermobarom-
etry to estimate the temperature and pressure of
clinopyroxene crystallization. This was done us-
ing the thermometer of Putirka (2008), coupled with
the newly calibrated barometer of Neave and Putirka
(2017). The thermobarometric calculations were car-
ried out with an Excel workbook of Neave et al.
(2019a) and by pairing the average clinopyroxene
macrocryst core compositions with basaltic Eyja-
fjallajökull melt compositions from this study (Ta-
ble 1) and the datasets of Loughlin (1995) and
Moune et al. (2012). First, putative melt composi-
tions for individual clinopyroxene crystals were se-
lected on the basis of Fe-Mg equilibrium, assuming
KdMg−Fe
Cpx−Liq = 0.27±0.6, 1 wt% H2O, and Fe2+/
∑
Fe
= 0.82 in accordance with fO2 = FMQ+0.3 (Kress
and Carmichael, 1991). Crystallization pressures and
temperatures for these putative clinopyroxene-melt
pairs were solved using the Eq. 1 barometer of Neave
and Putirka (2017), coupled with Eq. 33 thermome-
ter of Putirka (2008) and by iteratively using an out-
put of one model as an input to another. Then, we
filtered these "pseudo" P-T results by only accept-
ing the clinopyroxene-melts pairs that are in multi-
component equilibrium following a method resem-
bling that of Neave et al. (2019a). Firstly, suitable
clinopyroxene-melt pairs had to be within ±10% Fe-
Mg equilibrium in accordance to Eq. 35 in Putirka
(2008). Secondly, the measured and predicted CaTs-,
EnFs- and DiHd-components in clinopyroxene had to
agree within the 1SEE precision, ±0.03 for CaTs,
±0.05 for EnFs and ±0.06 for DiHd (mol fractions),
of modelling equilibrium clinopyroxene components
from the paired melts (Putirka, 1999; Mollo et al.,
2013). Thirdly, we only accepted P-T estimates from
the clinopyroxene-melt pairs that deviated less than
±40% from Ti equilibrium according to the model
of Hill et al. (2011). If multiple melt compositions
were in equilibrium with a single clinopyroxene after
this extensive equilibrium filtering, the P-T estimates
were most often within the methods 1SEE precision
(±1.4 kbar and ±28◦C; Neave and Putirka, 2017; see
Figure 5), and a mean of the calculated P and T was
assigned for the clinopyroxene crystal. Suitable equi-
librium liquids were found for nearly all clinopyrox-
ene grains with the exception of the most magne-
sian (Mg#cpx 89.7–89.8) clinopyroxene macrocrysts
indentified from Brattaskjól.
Clinopyroxene thermobarometry reveals similar
crystallization temperatures and pressures for the
Brattaskjól and Hvammsmúli ankaramites (Figure 5b
and c). Model temperatures correlate with Mg#,
such that the most primitive clinopyroxene macrocryst
with Mg#cpx>86 record 1190◦C and the most evolved
Mg#cpx<81 clinopyroxene crystal records 1120◦C
(Figure 5b). All P estimates (1.7–4.2 kbar) are within
the methods precision (±1.4 kbar, Neave and Putirka,
2017) and average at 3.0 kbar, although there is a trend
of increasing P as a function of Mg#cpx (Figure 5b).
This P and Mg#cpx correlation is partly related to,
but not fully explained by, the T dependency of the
barometer. The Brattaskjól olivine-hosted melt in-
clusion with the lowest Mg#melt (Ol14_mi1_2, see
Table 1) was found to be a suitable equilibrium liq-
uid for thirteen Mg#cpx 84.9–86.9 clinopyroxenes,
implying crystallization conditions of 1178◦C and
3.7 kbar for these crystals. These are similar to the
mean crystallization T and P of 1182◦C and 3.5 kbar
derived for similar Mg#cpx 84.9–86.9 clinopyroxene
92 JÖKULL No. 69, 2019