Geol.

655 Isotope Geochemistry

Lecture 4 Spring 2009

THE K-AR-CA SYSTEM

We have now discussed many of the basic aspects of radiogenic isotope geochemistry and we can
now consider how it is applied to solving questions about the Earth. We will begin by discussing geo-
chronology, because this aspect is basic to other aspects of isotope geochemistry. To understand the
more chemical and geological aspects of isotope geochemistry, we must first learn to tell time, so to
speak. We will consider the various decay systems separately. Except for some special aspects of K-Ar
and U-Th-Pb, the principles involved do not differ between these systems. The differences that do exist
arise from the different chemical behavior of the elements involved.
Two aspects of K-Ar-Ca system make it special. First, it is a branched decay: a 40K nucleus (an odd-
odd nuclide) may decay to either a 40Ca by β- or to a 40Ar atom by electron capture (or much more
rarely by positron emission — which is just as well for us). It is impossible to predict how a given 40K
atom will decay, just as it is impossible to predict when it will decay. We can predict quite accurately
what proportion of a large number of 40K atoms will decay to each, however. The ratio of electron cap-
tures to beta decays is called the branching ratio and is defined as:
!e
R= 4.1
!"
where the two lambda's are the decay constants (i.e., the probability of decay) for each mode. The
branching ratio is 0.117, λe = 0.581 × 10-10 yr, λ β = 4.962 × 10-10 yr. The total decay constant for 40K is:
! = ! " + ! e = 5.543 # 10 $10 yr $1 4.2
We need to take account of this branched decay in our equation, because while K decaying to Ca does
not produce radiogenic Ar, it is no longer K and not available for 40Ar production. Thus our equation
for radiogenic daughter production (equation 3.7) becomes:
! e 40
40
Ar* = K(e!t " 1) 4.3
!
where the asterisk indicates radiogenic 40Ar. Note we can write a similar equation for 40Ca* by substi-
tuting λ β for λe.
Most, although not all, of the work on the K-Ca-Ar system has focused on Ar because the 40K/40Ca ra-
tio is usually small. 40K is the least abundant of the K isotopes (0.012%), whereas 40Ca is the most abun-
dant 40Ca isotope (96.92%), and Ca is a more abundant element than K (40Ca is even-even, 40K is odd-
odd). As a result, variations in the 40Ca/42Ca ratio resulting from radioactive decay are quite small and
difficult to measure. Only in very favorable circumstances, such as halide salt deposits, is geochronol-
ogy practical.
As one might expect, particularly in view of the above discussion, one of the most important criteria for a
useful radiometric chronometer is that the variations in the radiogenic isotope be large relative to the precision and
accuracy with which they can be measured. In this respect, a short half-life is advantageous, and K has one
of the shortest half-lives of the long-lived radioactive isotopes. Because of the volatility of Ar, the Earth
either lost much of its Ar during its formation, or never acquired much, giving the Earth a rather high
K/Ar ratio. Furthermore, much of the Ar the Earth retained is now in the atmosphere. As a result,
40
K/40Ar ratios in the solid Earth tend to be quite high. Because of the high 40K/40Ar ratios and the rela-
tively short half-life of 40K, the K-Ar system is often the one of choice when the task at hand is to date very
young events. Meaningful ages (‘meaningful’ means the uncertainty is small relative to the age) of less
than 30,000 years have been determined in favorable circumstances. 40K/40Ar it is the only long-lived
decay system that has produced age information of use in archeology and human evolution.
Much of what is special about K-Ar derives from Ar being a noble gas and its resulting refusal to be
chemically bound in crystal lattices. Ar in rocks and minerals is simply trapped there. It has difficulty

37 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009
escaping because the atoms of the lattice block its escape path, but it is does not form chemical bonds
with other atoms in the lattice. Thus when a mineral crystallizes from a lava, it will generally, although
not always, do so with very little Ar. Pillow basalts formed on the seafloor are one example of excep-
tions to this rule. The combination of relatively high pressure and rapid transition from the liquid to
the solid state can result in trapping of substantial amounts of Ar. Similarly, minerals crystallizing
from plutonic rocks may also retain Ar.
In favorable circumstances, essentially no Ar will be trapped in a mineral crystallizing from lava. The
great advantage of this, from a geochronological viewpoint, is we have only one unknown, namely t,
and we can use equation 4.3 to solve for it by measuring the 40K and 40Ar in one sample. Actually, one
need not assume that no ‘initial’ Ar whatsoever is present. Indeed, in detail, this would seem a poor as-
sumption since a mineral crystallizing in contact with the atmosphere can be expected to absorb a small
but finite amount of atmospheric Ar. For relatively young samples, this atmospheric Ar is readily cor-
rected for since the atmosphere has a uniform ratio 40Ar/36Ar of 295.5. By measuring the amount of
36
Ar present, we can deduce the amount of atmospheric 40Ar initially present. Our age equation (equa-
tion 4.17) becomes simply:
40
Ar ! 40
K !t
= 295.5 + e (e " 1) 4.4
36
Ar ! 36
Ar
If we suspect that the composition of ‘initial’ Ar differs significantly from atmospheric, it is then neces-
sary to employ the isochron approach, measuring K and Ar in a number of cogenetic samples and solv-
ing simultaneously for t and the initial 40Ar/36Ar ratio.
Diffusion, Cooling Rates, and Closure Temperatures
Because Ar is not chemically bound in lattices, the K-Ar clock will generally be reset more readily
than other systems. We concluded earlier that an event that ‘resets’ a radiometric clock is generally a
thermal one. In the case of K-Ar, we might guess that the system would be reset whenever tempera-
tures are high enough to allow Ar to diffuse out of the rock or mineral of interest. It is worth consider-
ing this on a slightly more quantitative level.
It can be shown both theoretically and experimentally that the rate at which a species will diffuse
through a medium is related exponentially to temperature:
! E A / RT
D = D0 e 4.5
where D is the diffusion coefficient, Do is the 'frequency factor’, EA is the activation energy, R is the gas
constant and T is thermodynamic, or absolute, temperature, (i.e., kelvins). The diffusion ‘flux’ is related
to the concentration gradient by Fick’s First Law:
# "C %
J = !D 4.6
$ "x &
where C is the concentration and x is distance. Figure 4.1 shows a plot of experimentally determined
values of D for Ar in biotite plotted against the inverse of temperature. The point to be made here is
that relatively small increases in temperature result in large increases in the diffusion coefficient. For
example, increasing the temperature from 600° C to 700°C results in a two order of magnitude increase
of the diffusion coefficient, and, for a given concentration gradient, of the Ar diffusion flux. Using the
values of EA and D0 given in the figure, we can calculate the diffusion coefficient for temperatures not
shown in the graph. The value of R is 1.987 cal/Kelvin-mole (8.314 joules/Kelvin-mole). For a tem-
perature of 300 K (27°C), D would be 4 × 10-36 cm2/sec. For any reasonable concentration gradient,
there would be no significant Ar loss from biotite, even over extremely long times. At 600 K (327°C),
we obtain D = 6 × 10 -19 cm2/sec, which implies a slow, but significant diffusion rate. At 700° C, how-
ever, loss of Ar would be quite rapid: about 1/3 of the Ar would be lost from biotites of 97µ radius in 2-

38 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009
3 weeks (you can understand then why the ex-
periments were done at these temperatures and
not lower ones).
The following equation (from Crank, 1975) is of
use in interpreting diffusion coefficients. The
equation gives the fraction of the species lost as a
function of time (t), diffusion coefficient, and the
diffusion radius (a).
1/ 2 3/2
4 # Dt % Dt 1 # Dt %
f ! 1/ 2 2 ' 2 ' 1/ 2 2 4.7
" $a & a 3" $ a &
The equation assumes radial diffusion in a cylinder
of infinite length and radius a. We can use this
equation to understand how Ar loss will vary with
temperature*. We assume a value for a of 150 µ,
and use the Do and EA values given in Figure 4.1.
Figure 4.2 shows the results of this calculation per-
formed for various times (and at various tempera-
Figure 4.1. Log of the diffusion coefficient for tures: D is a function of temperature, of course).
argon in biotite against the inverse of thermody- Let's consider the geological implications of this
namic temperature. Circles and squares indicate diagram. Imagine a body of rock, either igneous or
different size fractions of biotite used in 1 kbar metamorphic, cooling from high temperature, a
experiments. Triangles are 14 kbar experiments. temperature high enough so that all Ar is lost.
From Harrison et al. (1985). Let’s pick up the story when the body is still 400°C
and cooling at a rate of 100°C/Ma. At this temperature, the biotite would just be beginning to retain
radiogenic Ar; that is, it is not being lost quite as fast as it is being created. After the first additional
million years, it would have cooled to 300°C, and biotite would be retaining most of its radiogenic Ar
(loss rate of about 10 percent per Ma). If cooling continues at this rate for another million years (in the
real world, it is unlikely cooling rates would be so constant), biotite would be loosing Ar at a rate of
only a tenth of a percent per Ma, a fairly insignificant rate. If the body then cooled completely, and if
we sampled biotite for K-Ar dating some 100 Ma later, assuming the biotite was not reheated, the ‘age’
we would calculate would refer to that 2 Ma period when the biotite cooled from 400°C to 200°C, and
probably closer to the time it passed from 400°C to 300°C. We say the biotite 'closed' at that time, and
we can estimate the closure temperature at between 300°C and 400°C.
Suppose cooling was slower, say 10°/Ma. In this case, 10 Ma would be required to cool from 400°C
to 300°C, and 20 Ma to cool to 200°C. A much smaller fraction of the radiogenic Ar produced while the
biotite was in the 200-400°C range would have been retained. The ‘age’ we would calculate using equa-
tion 4.4 would be younger than in the example above. It would thus seem that under these circum-
stances, the ‘closure temperature’ would depend on the cooling rate. This is indeed the case.
Dodson (1973) derived an equation for ‘closure temperature’ (also sometimes called blocking tem-
perature) as a function of diffusion parameters, grain size and shape, and cooling rate:
EA
Tc = 4.8
# ART 2 D &
R ln % ! 2 c 0 (
$ a E A" '

*Note that this equation implies that diffusional loss will be inversely related to crystal size: larger crys-
tals will lose less.

39 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009

Figure 4.2. Fraction of Ar lost from a 150 µ cylindrical crystal as a function of

temperature for various heating times. All Ar is lost in 10 Ma at 340°C, or in 1
Ma at 380° C.
where τ is the cooling rate, a is the characteristic diffusion dimension (e.g., radius of a spherical grain),
and A is a geometric factor (equal to 55 for a sphere 27 for a cylinder, and 9 for a sheet). Unfortunately,
this is not directly solvable since Tc occurs both in and out of the log, but it can be solved by indirect
methods†.
There are several important notions we can come away with. First, a ‘closure temperature’ is a useful
concept, but a mineral will not suddenly stop loosing Ar, or any other radiogenic component, at its clo-
sure temperature. Closure temperature reflects a trade off between loss and creation of the radiogenic
component. Second, there is some ultimate geological limitations on the meaning of an age of a slowly
cooled rock, such as a large intrusion or regionally metamorphosed body of rock. We might also expect
the age we obtain will depend on the mineral we use for dating (since the diffusion coefficient will
vary), and perhaps on its composition (there is in fact some compositional dependence of the Ar diffu-
sion coefficient on the Fe/Mg ratio in biotite; but apparently none in hornblende). Finally, we get the
sense that it might be rather easy for K-Ar mineral age to be partially reset. This is certainly the case.
We discuss next a technique that can at least identify partially reset minerals, and in favorable cases,
provide a reasonable estimate of the original ‘age’.
40
Ar-39Ar Dating
If you look at the table of isotopes you'll see 39Ar has a half-life of 269 years and does not occur nat-
urally. You might justifiably wonder how it could be used for dating. The so-called 40-39 method is ac-
tually 40K-40Ar dating but employing a somewhat different analytical technique for the potassium. The
key is the production of 39Ar by a nuclear reaction on 39K, the most abundant of potassium’s three iso-
topes:
39
K (n,p) 39Ar
The reaction is produced by irradiating a sample with neutrons in a reactor. It is important to distin-
guish this reaction from simple neutron capture, but we can nevertheless define a reaction cross-sec-

† The Solver, an add-in tool that comes with Microsoft Excel™ can be used to solve problems such as this.

40 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009
tion. The amount of 39Ar produced is then a function
of the amount of 39K present, the reaction cross-
section, the neutron flux, and the irradiation time.
Since the 40K/39K ratio in the Earth is constant (at any
given time), the amount of 40K can be calculated from
39
Ar. In practice, the situation is more complex be-
cause the reaction cross-section is a function of neu-
tron energy and there typically is a spectrum of ener-
gies present. The production of 39Ar from 29K can be
expressed as:
39
Ar= 39K! $ " e# e de 4.9
where e is the neutron energy, φe is the flux of neu-
trons with energy e, and σe is the capture cross section
for that energy, and τ is irradiation time. The
40
Ar*/39Ar is obtained by dividing equ. 4.3 by 4.9:
40
Ar * ! e 40
K(e!t " 1)
Figure 4.3. 40 39
Ar/ Ar age spectrum produced = 4.10
!
# $ % &
39
Ar
39
K de
by step heating of a biotite from a granitic e e

gneiss. From McDougall and Harrision (1988). In practice, the analysis is performed by simul-
taneously irradiating and analyzing a stan-
dard of known age. The flux, capture cross
section, and decay constant terms will be the
same for the standard as for the unknown
sample. We can combine them into a single
term, C, as:
!e 1
C= 4.11
! " % # e$ e de
and equation 4.9 becomes:
40
Ar * 40
K(e!t " 1)
39 =C 39 4.12
Ar K
The value of C can be determined from analy-
sis of the standard, so that 4.12 can be solved
for the age.
Another problem is the production of both
39
Ar and 40Ar by other reactions that include
40
K(n,p)40Ar, 40Ca(n,nα)36Ar, and 42Ca(n,α)39Ar.
These must be corrected for. Generally, 37Ar,
Figure 4.4. Ar release spectra for hornblendes taken produced by Ca(n, α) Ar, is used to make
40 37

from varying distance from a 114 million year old intru- these corrections.
sion. The crystallization age of the samples is 367 Ma. In conventional K-Ar dating, Ar is released
Curves show calculated release spectra expected for from samples by fusing in vacuum. How-
samples that lost 31, 57, and 78% of their argon. From ever, we might guess from our knowledge of
McDougall and Harrison, 1988. diffusion that a sample will begin to lose Ar
before it reaches its melting temperature. If
the ratio of radiogenic 40Ar to K (and therefore to 39Ar) were distributed uniformly throughout the sam-

41 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009
ple, a sample of gas taken before the sample fused would produce the same age as for total fusion. We
might guess, however, that some parts of the crystal have preferentially lost Ar through diffusion dur-
ing the initial cooling of the crystal, or perhaps during some subsequent reheating event. Since the dif-
fusion rate is proportional to the concentration gradient, we can anticipate that diffusion will be faster
from the rims of crystals where the concentration gradient is higher than in the interior of crystals. So
we might expect crystal rims to experience Ar loss at lower temperatures than crystal interiors. The
rims would then record younger ages. As we heat the sample, we would also expect rims to start to
give up there Ar at the lowest temperatures, partly for this reason, and partly because the Ar has less
distance to go to get out. The lower 40Ar/39Ar of the gas in the rim would be seen as a lower age (which
may or may not have significance). As we increased the temperature, the more retentive parts of the
crystal would release their gas, and we could expect the 40Ar/39Ar and the apparent age to increase. If
some parts of the crystals have lost no gas, their 40Ar/39Ar ratios would record the ‘correct’ age, even
though the crystal as a whole has suffered some loss. Figure 4.3 is an Ar release diagram for a biotite
exhibiting this sort of behavior. Conventional K-Ar dating would have produced an age intermediate
between the ‘correct’ age and the apparent young age of those parts of crystal that have suffered loss of
radiogenic 40Ar. Thus the combination of the 40Ar/39Ar method with step-heating provides a means of
retrieving useful geochronological information from samples whose value would have otherwise been
compromised because of diffusional loss. In a certain sense, we are relaxing our requirement that the
system must have remained closed: with 40Ar/39Ar dating, we require only that some part of the system
have remained closed.
Most Ar release spectra are not so simple as that in Figure 4.3. Figure 4.4 shows Ar release spectra for
a series of hornblende samples taken at varying distances from the contact with an intrusive gra-
nodiorite. All show significant Ar loss as a result of heating from the intrusion. None retain, even at
the highest release temperature, the true age of 367 Ma.
Other Complications Affecting 40Ar-39Ar Spectra
In the previous section, we saw two examples of 40Ar/39Ar release spectra: one where there was only
minor loss of Ar from the rims, and another where significant fractions of the total Ar had been lost due
to metamorphic heating. Figure 4.4 showed spectra that almost perfectly match theoretical patterns for
diffusional loss. Such examples are relatively rare; most spectra are more complex. For example, some
samples that have been reheated show false plateaus that correspond to ages intermediate between the
crystallization age and the reheating age. An additional problem in interpreting such spectra is that
samples that have not been subjected to reheating events but cooled slowly originally can show release
spectra that mimic those of reheated samples in Figure 4.4.
Recoil of 39Ar produced by the 39K(n,p)39Ar reaction during irradiation can also produce problems.
The recoil results in loss of 29Ar from sites near the mineral surface. For large grains, this is largely in-
significant, but for small grains, this can lead to significant 39Ar loss, leading in turn to erroneously old
apparent ages.
In most case, the Ar present in a sample will not be pure radiogenically produced Ar. Non-radio-
genic argon is often called excess Ar. 40Ar/39Ar ratios used to calculate ages in release spectra are typi-
cally corrected for the presence of atmospheric Ar by measuring the 40Ar/36Ar ratio. Atmospheric ar-
gon has a constant 40Ar/36Ar ratio of 295.5. Only 40Ar present in excess of this ratio is considered radio-
genic and used to calculate the 40Ar/39Ar ratio. Nevertheless, some samples can have “initial” 40Ar/36Ar
ratios greater than the atmospheric ratio; this will lead to too old an age if not properly accounted for.
It is this “excess” argon that is of greatest concern.
Excess Ar can have two sources. First, it can arise when minerals crystallize under a finite partial
pressure of Ar. For example, mantle-derived basalts have been shown in some cases to have initial
40
Ar/36Ar ratios of up to 40,000. The high 40Ar/36Ar ratio reflects production of 40Ar by decay of 40K
within the mantle. Minerals crystallizing in the presence of this gas will trap some of this 40Ar, which

42 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009
will result an anomalously old age upon analysis. This
is referred to as inherited Ar. Second, during a thermal
event, 40Ar diffusing out of some minerals may be
taken up by other minerals. Since this 40Ar is diffusing
into the mineral grain, its concentration will be highest
in the exterior of grains and thus will tend to be re-
leased at the lowest temperatures. An example is
shown in Figure 4.5.
When excess Ar is held in more than one crystal-
lographic site, for example different minerals in the
analyzed sample, release spectra can reveal a saddle
shape. An example is shown in Figure 4.6. This sam-
ple is a calcic plagioclase from Broken Hill in Austra-
lia. The true metamorphic age is approximately 1600
Ma. Even the minimum values in the bottom of the
saddle are too old. Electron microscopy of the plagio-
clase revealed that it had exsolved into a Ca-rich and
Na-rich plagioclase. The saddle shape results because
Figure 4.5. Ar release spectrum of a horn- Ar in one of the phases diffuses readily and is thus
blende in a Paleozoic gabbro reheated in the released at low temperature, and diffuses more slowly
Cretaceous by the intrusion of a granite. in the other, resulting in release at high temperature.
Anomalously old apparent ages in the lowest
temperature release fraction results from diffu-
sion of radiogenic Ar into the hornblende dur-
ing the Cretaceous reheating. From McDougall
and Harrison (1988).
A new technique, developed only in the last
15 years, involves releasing Ar from small ar-
eas of a sample through laser ablation. This al-
lows release of Ar from areas with diameters
less than a millimeter, and provides the possi-
bility of spatial resolution of Ar diffusional
loss.
40
Ar-39Ar Isochrons
The data from various temperature release
steps are essentially independent observations
of Ar isotopic composition. Because of this,
they can be treated much the same as in con-
ventional isochron treatment. The isochron
equation, written for the K-Ar system is:
40
Ar ! 40
Ar $ 40
K 't
= + (e – 1) 4.13 Figure 4.6. Ar release spectrum from a calcic plagio-
36
Ar #" 36
Ar &% 0 36
Ar clase from Broken Hill, Australia. Low temperature
and high temperature fractions both show errone-
When 40Ar/36Ar data from a series of samples ously old ages. This peculiar saddle shaped pattern,
are plotted against 40K/36Ar, the slope of the re- which is common in samples containing excess Ar,
sulting line will be proportional to age, and the results from the excess Ar being held in two different
intercept gives the initial 40Ar/36Ar ratio. Since lattice sites. From McDougall and Harrison (1988).

43 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009
for all release fractions of a sample the efficiency
of production of 39Ar from 39K is the same and
40
K/39K ratios are constant, we may substitute
C39Ar for 40K:
40
Ar !# 40 Ar $ 39
Ar &t
= + C(e –1) 4.14
36
Ar " 36 Ar % 0 36
Ar
C is a constant that depends on the efficiency of
39
Ar production during irradiation. Thus when
40
Ar/36Ar ratios from a series of release fractions
are plotted against 39Ar/40Ar, the slope of the
resulting line will be proportional to the age of
the sample, as is illustrated in Figure 4.7.
The use of the isochron diagram can help to
identify excess Ar and its nature (e.g., atmos-
Figure 4.7. Hypothetical 40Ar-39Ar isochron dia- pheric, inherited, etc.). It also provides a crucial
gram. The slope is proportional to the age and the test of whether ages obtained in release spectra
intercept gives the initial 40Ar/36Ar ratio, which is are meaningful or not. A drawback of this dia-
commonly atmospheric, as is illustrated here. gram is that 36Ar, which is the denominator in
both the ordinate and abscissa, is often present in
only trace amounts and is difficult to measure precisely. Because it appears in both ordinate and ab-
scissa, errors in its measurements can produce correlations that imitate isochrons.
An alternative is to use a plot of 36Ar/40Ar against 39Ar/40Ar (Figure 4.8), often called an inverse iso-
chron plot. We can think of the Ar in a sample as a mixture of a trapped component and a radiogenic
component. As such, the data for various release fractions should plot as a straight line on such a plot.
The radiogenic component has a 36Ar/40Ar ratio of 0 (because 36Ar is not produced by radioactive de-
cay), whereas the trapped, non-radiogenic
component can be found by extrapolating to
a 39Ar/40Ar ratio of 0 (corresponding to a
39
K/40Ar ratio of 0; since 39K is proportional to
40
K, this also corresponds to a 40K/40Ar ratio
of 0). Thus the age may be computed from
the 39Ar/40Ar ratio obtained by extrapolating
the correlation line to 36Ar/40Ar to 0, and the
composition of the trapped component by ex-
trapolating to 39Ar/40Ar of 0.
Figure 4.9 provides an example of how the
inverse isochron plot may be used to identify
trapped components. The original release
data showed a disturbed pattern and lacked
a plateau (not shown). The inverse isochron
plot (Figure 4.9a) revealed two correlations
suggesting the presence of two distinct
trapped components. The lower intercept
yielded an age of 149.1 Ma. When the data
were corrected for the trapped component
Figure 4.8. Plot of 36Ar/40Ar vs 39Ar/40Ar, also called an
and replotted on a release spectrum, they
inverse isochron diagram. Age is obtained from the
produced a plateau corresponding to the
value of 39Ar/40Ar corresponding to 36Ar/40Ar = 0.
same age as the isochron age.

44 January 28, 2009

 Geol. 655 Isotope Geochemistry
Lecture 4 Spring 2009

Figure 4.9. (a) Inverse isochron diagram revealing the presence of two excess Ar components. (b) Ar re-
lease spectrum for the same sample after correction for non-atmospheric excess Ar.

REFERENCES AND SUGGESTIONS FOR FURTHER READING

Albarède, F., The thermal history of leaky chronometers above their closure temperature, Geophys.
Res. Lett., 30(1), 1015, doi:10.1029/2002GL016484, 2003.
Crank, J., The Mathematics of Diffusion, Oxford Press, 2nd ed., 414 pp., 1975.
Dodson, M. H., Closure temperature in cooling geochronological and petrological systems, Contrib.
Mineral. Petrol., 40: 259-274, 1973.
Dickin, A. 1995. Radiogenic Isotope Geochemistry. Cambridge: Cambridge University Press.
Ehlers, K. E. and R. Powell, An empirical modification of Dodson's equation for closure temperature
in binary systems, Geochim Cosmochim Acta, 58:241-248, 1994.
Harrison, T. M., I. Duncan, and I. McDougall, Diffusion of 40Ar in biotite: temperature, pressure and
compositional effects, Geochim. Cosmochim. Acta, 49, 2461-2468, 1985.
McDougall, I. and T. M. Harrison. 1988. Geochronology and Thermochronology by the 40Ar/39Ar Method.
New York: Oxford Univ. Press.
Onstott, T. C. and M. W. Peacock, Argon retentivity of hornblendes: a field experiment in a slowly
cooled metamorphic terrane, Geochim. Cosmochim. Acta, 51, 2891-2904, 1987.

45 January 28, 2009