WHAT CAN WE LEARN FROM ISOTOPES? - Chemical Fundamentals of Geology and Environmental Geoscience - page 184

Box 10.2 40 Ar/ 39 Ar dating: solutions to 40 K- 40 Ar dating problems

Conventional K-Ar geochronology relies on a K-rich min-
eral having remained sealed with respect to 40 K and 40 Ar
migration ever since the event being dated. Experience
shows, however, that 40 Ar may leak out of such minerals
('Ar loss') or indeed may diffuse in, for example from a
hydrothermal fluid percolating through the sample (leav-
ing 'excess Ar' in the crystal). Such deviations from
closed-system behaviour introduce systematic errors in
measured K-Ar ages that are impossible to detect rout-
inely and which severely limit the method's reliability as a
dating tool.
The so-called 40 Ar/ 39 Ar (or just 'Ar-Ar') dating technique
circumvents these and other shortcomings of K-Ar geo-
chronology in two ingenious ways, as outlined in
Table 10.2.1. The same K-bearing minerals (or volcanic
whole-rocks) are used as for K-Ar dating.
These systematic errors do not manifest themselves
uniformly through a mineral crystal, a fortunate fact that
makes them easier to decipher. Argon loss is more marked
near the margins of a crystal (or near to cracks). During
stepwise heating, early heating steps preferentially extract
argon from the 40 Ar-depleted (low 40 Ar/ 39 Ar) rim of a crystal,

giving apparent ages younger than the geological age,
whereas 40 Ar in the more retentive interior will only be
released in higher-temperature steps. The age spectrum
of a sample affected by argon loss therefore exhibits an
initial rise (Figure 10.2.1b), and commonly leads to an age
plateau from which consistent, accurate age values can be
calculated.
'Excess' 40 Ar that has diffused into a sample after crys-
tallization (most common in metamorphic minerals) tends
to reside on subgrain boundaries or in fluid inclusions,
from where it escapes at low temperatures giving anom-
alously old ages from early heating steps (Figures 10.2.1b
and 10.2.2). Later steps are more likely to yield accurate
geological ages, although sometimes older ages emerge
again towards the end of argon release, reflecting 40 Ar
trapped in melt or mineral inclusions (Kelley, 2002a,b).
Whereas K-Ar dating yields absolute ages, Ar-Ar is a
relative dating method that requires calibration by
analysing a standard of known geological age alongside
unknown sample(s). Recent re-calibration has brought
Ar-Ar dates into closer agreement with other geochron-
ometers (Kerr, 2008).

Table 10.2.1 How 40 Ar- 39 Ar dating overcomes K-Ar problems

Problems encountered with conventional 40 K- 40 Ar dating Features of 39 Ar- 40 Ar dating that circumvent these problems

1. K and 40 Ar are determined by different analytical

techniques on separate sample aliquots , introducing
errors and reducing the internal consistency of the
40 K/ 40 Ar ratio.

A single sample aliquot is irradiated with neutrons in a nuclear
reactor to convert 39 K (a stable K isotope) to 39 Ar.* The 39 Ar/ 40 Ar
ratio of the irradiated aliquot is readily determined by mass
spectrometry (Box 10.3), from which a precise 40 K/ 40 Ar
parent:daughter ratio for the aliquot can be calculated.

2. The K-Ar technique is prone to systematic errors that

alter the daughter/parent ratio:
● 40 Ar loss from the sample, giving low age values relative
to true age.
● 'Excess Ar' (including 40 Ar) diffusing into the sample, giv-
ing high age values.

An irradiated sample aliquot is heated under vacuum in a series of
temperature steps § until all the Ar has been extracted. The 39 Ar/ 40 Ar
ratio is measured by mass spectrometry - and the apparent age
calculated - separately for each heating step (Figure 10.2.1b). The
resulting 'age spectrum' commonly reveals Ar loss or excess Ar in
the earlier heating steps. Higher-temperature steps usually define a
consistent age 'plateau' from which a reliable mean age for the
sample - free of systematic error - can be determined. Data from a
sample giving no age plateau can be disregarded.

3.

40 Ar is extracted for mass spectrometric analysis by
completely melting the sample aliquot under vacuum.
This provides only one 40 Ar determination (and thus only
one 40 K/ 40 Ar ratio) that offers no direct indication of
systematic errors (Figure 10.2.1a).

* 39 Ar is radioactive with a half-life of 269 years; being short-lived, it is not shown as a box in Figure 10.1.1. Nonetheless, on a
laboratory timescale, it can be treated arithmetically as a stable isotope.
§ In a furnace or using a laser.