Jökull - 01.12.2006, Qupperneq 63

Pleistocene rhyolitic volcanism at Torfajökull, Iceland
SAMPLE SELECTION
Samples for Ar-Ar dating were selected to address
four objectives:
1. To test the hypothesis that the “ring fracture rhyo-
lites” were erupted during one eruptive event.
2. To establish whether all tuyas were erupted during
the same glacial period.
3. To date some of the older rhyolites at Torfajökull.
4. To evaluate whether rhyolite compositions change
systematically with time, as previously suggested
by McGarvie et al. (1990), by combining Ar-Ar
geochronology with geochemical data.
An important constraint in sample selection is that
rocks which have been hydrothermally altered are
unlikely to yield reliable Ar-Ar ages (Hawkesworth
et al., 2004). However Torfajökull’s oldest rhyolitic
rocks lie within a suspected caldera (Sæmundsson
and Friðleifsson, 2001; and Figure 2), and have suf-
fered considerable hydrothermal alteration (Friðleif-
sson and Sæmundsson, 2001). Consequently, these
rhyolites had to be excluded from this study and hence
the age of the oldest rhyolitic rocks at Torfajökull re-
main unknown.
Ar-Ar DATING
The samples selected for Ar-Ar dating comprise fresh
obsidian from the aphyric rhyolites (TJ97-9 and TJ97-
14), and feldspar phenocrysts from the porphyritic
rhyolites (TJ97-18, TJ97-22, and TJ97-29). For the
obsidian analyses sub-millimetre fragments of glass
matrix were hand picked under a binocular micro-
scope to avoid any of the rare microphenocrysts
that may have experienced a different magmatic his-
tory. For the porphyritic rhyolites, the largest and
most euhedral phenocrysts were hand picked under
a binocular microscope. Each sample was ultrasoni-
cally cleaned in methanol and de-ionised water before
being irradiated. Samples were irradiated for 0.5 hr
in the CLICIT facility of the TRIGA reactor at Ore-
gon State University. Neutron interference corrections
were determined from CaF2 and K2SO4 included in
the irradiation and are within error of values previ-
ously determined for the OSU reactor (Nomade et al.,
2005).
Following irradiation, samples were placed into
3 mm diameter wells drilled into an Al disk and placed
into a laser port connected to the MS1 mass spec-
trometer extraction system. Samples were evacuated
to UHV and heated to 100◦C for 48 hours using an
infra-red lamp to remove any adsorbed atmospheric
gases. Argon was released from the samples by heat-
ing using a Nd-YAG continuous wave laser with a spot
size of 3 mm. Argon was released using between 5–7
heating steps each of 3–5 minutes duration, using a
laser output power of up to 21 W in TEM00 mode.
The poor absorption of feldspar at the laser wave-
length of 1064 nm steps resulted in incomplete melt-
ing of the samples at the end of the experiments. Ar
gas was purified using two Zr-Al getters, one at 250◦C
and the other at room temperature, and then admitted
directly to the spectrometer. Argon isotopes and con-
taminant masses (m/z 35 and 41) were determined us-
ing a Faraday detector. Blanks were determined at the
start of every day and after every third sample analy-
sis and gave average values of 28×10−12 cm3 40Ar of
approximately atmospheric isotopic composition. The
blank is typically 1–10% of the 40Ar released during a
sample analysis. Data were corrected for blanks, mass
discrimination, neutron interference and radioactive
decay, however these were relatively small corrections
compared to correction for atmospheric Ar monitored
by the level of 36Ar released from heating samples.
The neutron fluence was determined using the
Fish Canyon sanidine (28.02±0.16 Ma; Renne et al.,
1998). Three Fish Canyon samples irradiated in close
proximity to the Torfajökull samples showed <0.5%
variation that are well-within uncertainties, thus a
weighted average J value of 0.000174±0.000005 was
used to calculate Ar-Ar ages. All uncertainties are
quoted at the 1σ level of uncertainty.
Data were plotted on isotope correlation diagrams
(Figure 3) from which the sample age and trapped
40Ar/36Ar composition were determined. For all sam-
ples the trapped 40Ar/36Ar value was indistinguish-
able, within 2σ uncertainty, from the atmospheric ra-
JÖKULL No. 56 61