Thermobarometry

Solid solutions
Rx ky+chl=st+q+H
2
O can be broken in two
41Al
2
SiO
5
+4Fe
5
Al
2
Si
3
O
10
(OH)
8
=10Fe
2
Al
9
Si
4
O
23
(OH)
+13SiO
2
+11H
2
O &
41Al
2
SiO
5
+4Mg
5
Al
2
Si
3
O
10
(OH)
8
=10Mg
2
Al
9
Si
4
O
23
(OH)
+13SiO
2
+11H
2
O
If 2
nd
Rx is subtracted from 1
st
10Mg St+4Fe Chl= 10Fe St+4Mg Chl
Fe-Mg exchange Rx

Since G= for i components sum n
i

i
and at equilibrium
10
MgSt
+ 4
FeCh
= 10
FeSt
+ 4
MgCh
or
10(
FeSt

MgSt
) 4(
MgCh

FeCh
) = 0 and the chemical
potential of each can be written in activity term
e.g.
FeSt
= *
FeSt
+ RT ln a
FeSt
(*
FeSt
is the chem potenial of
pure FeSt at given P-T)

10(*
FeSt
- *
MgSt
)- 4(*
MgCh
- *
FeCh
) = -10RT ln a
FeSt

a
MgSt

+ 4RT ln a
FeCh

a
MgCh


Since * = G the LHS is simply the free energy change of the
Rx

G
exch
= -RT ln a
10
FeSt
a
4
MgCh

a
10
MgSt
a
4
FeCh

This free energy is known as Fe-Mg exchange potential;
Activity needs to be related to mole fraction

Assuming ideal mixing of component i on n equivalent
sites in a mineral activity is a
i
= X
i
n
Since St has two moles
of Fe and Ch has 5 moles of Fe a
FeSt
= X
2
FeSt
and a
FeCh
=
X
5
FeCh
and thus above eq. becomes

G
exch
= -RT ln X
20
FeSt
X
20
MgCh

X
20
MgSt
X
20
FeCh
-G
exch
= 20RT ln X
FeSt
X
MgCh

X
MgSt
X
FeCh
(ratio of mole fraction is
known as distribution coefficient)

K
D
and in this case may be written as K
DFeMg
StCh
It is a
measurable property of a rock
Last eq relates K
D
to the free energy change of a an
exchange Rx between pure end members,
thermodynamic data for which is available
G
exch
= H
o
exch
-T S
o
exch
+(P-10
5
) V
o
exch
(P measured in pascal)

H
o
exch
-T S
o
exch
+(P-10
5
) V
o
exch
= -mRT lnK
D

lnK
D
= - H
o
exch
/mRT + S
o
exch
/mR (P-10
5
) V
o
exch
/mRT
(m= coefficient based on activity model and the coefficients in the Rx)

Possible use of an exchange Rx as thermometer or
barometer depends upon the magnitude of each term in
the above eq


lnK
D
= - H
o
exch
/mRT + S
o
exch
/mR (P-10
5
) V
o
exch
/mRT

For most Fe-Mg exchanges V
o
exch
~ 0 so this eq reduces
to that of a straight line (y=mx+b) (y=lnK
D
; m (slope)=
H
o
exch
/mR; x=1/T and b=intercept on the y axis)

It can be used as a thermometer if 1) either thermodynamic
data is available for end members 2) or the Rx can be
calibrated through exp studies or with other thermometers
When solid solution results in impure phases the
activity of each phase is reduced
Use the same form as for gases (RT ln P or ln f)
Instead of fugacity, we use activity
Ideal solution: a
i
= X
i
n = # of sites in the phase on
which solution takes place
Non-ideal: a
i
=
i
X
i

where
i
is the activity coefficient
Example: orthopyroxenes (Fe, Mg)SiO
3

Real vs. Ideal Solution Models
Activity-composition relationships for the enstatite-ferrosilite mixture in orthopyroxene
at 600oC and 800oC. Circles are data from Saxena and Ghose (1971); curves are model
for sites as simple mixtures (from Saxena, 1973)
Consider the reaction:
2CaMgSi
2
O
6
(cpx) + Fe
2
SiO
4
(ol)
2CaFeSi
2
O
6
(cpx) + Mg
2
SiO
4
(ol)
We can calculate:
A
r
G = 2A
f
G
CaFeSi
2
O
6
+ A
f
G
Mg
2
SiO
4

- 2A
f
G
CaMgSi
2
O
6
+ A
f
G
Fe
2
SiO
4


|
|
.
|

\
|

= = A
ol
SiO Fe
cpx
O CaMgSi
ol
SiO Mg
cpx
O CaFeSi
o
r
a a
a a
RT K RT G
4 2
2
6 2
4 2
2
6 2
ln ln
This equation tells us that ln K is a function of
pressure and temperature. Thus, if we had a rock
with clinopyroxene and olivine which we can assume
were in equilibrium, we could calculate ln K from a
microprobe analysis and an activity model. Given an
independent estimate of P, we could calculate T, or
vice versa.
If the above equation was relatively independent of
pressure, we could use it as a geothermometer even
if we did not have an independent estimate of
pressure. Similarly, if the reaction is relatively T-
independent, we might have a geobarometer.
K RT V P T G
o
s
o
r
ln ) 1 ( ) bar 1 , ( = A + A
CRITERIA FOR GEOTHERMOMETERS
AND GEOBAROMETERS
The ideal geothermometer would depend strongly on
T and weakly on P.
The ideal geobarometer would depend strongly on P
and weakly on T.
The temperature dependence is given by:



and the pressure dependence is given by:
( )
2
ln
RT
H
T
K
o
r
P
A
=
|
.
|

\
|
c
c
( )
RT
V
P
K
o
r
T
A
=
|
.
|

\
|
c
c ln
Thus, a good geothermometer is one in which the
absolute value of A
r
H is relatively high, and A
r
V is
near zero.
Conversely, a good geobarometer is one in which
A
r
H is nearly zero and the absolute value of A
r
V is
relatively high.
P
T
P
T
K = 10 5 2
K = 2
K = 5
K = 10
Potential geothermometer Potential geobarometer
OTHER CRITERIA
1) Reasonably accurate standard-state thermodynamic
data are either known or can be estimated from simple-
system experiments.
2) The relationships between activities of the relevant
components in complex phases and the compositions
of the phases should be well-defined. In practice,
reactions involving major constituents are generally
more useful than those involving trace components (as
concentration increases,
i
0).
EXCHANGE REACTIONS
Consider the reaction we wrote previously:
2CaMgSi
2
O
6
(cpx) + Fe
2
SiO
4
(ol)
2CaFeSi
2
O
6
(cpx) + Mg
2
SiO
4
(ol)
This is an exchange reaction in that Fe
2+
and Mg
2+
simply exchange places between clinopyroxene and
olivine. It can be written in terms of two exchange
components, i.e.,
FeMg
-1
ol
FeMg
-1
cpx

Exchange reactions usually make good geothermo-
meters because they typically have small A
r
V.
GARB
This is based on the reaction:
Fe
3
Al
2
Si
3
O
12
(grt) + KMg
3
AlSi
3
O
10
(OH)
2
(bt)
Mg
3
Al
2
Si
3
O
12
(grt) + KFe
3
AlSi
3
O
10
(OH)
2
(bt)
Now the following must be true at equilibrium:



As a general rule, as temperature increases, K
begins to approach zero. This is because at
sufficiently high temperature, the energetic distinction
between different elements decreases.
|
|
.
|

\
|
= = A
bt
ann
grt
alm
bt
phlog
grt
pyr o
r
a a
a a
RT K RT P T G ln ln ) , (
GARB
KMg
3
AlSi
3
O
10
(OH)
2
+Fe
3
Al
2
Si
3
O
12
=
KFe
3
AlSi
3
O
10
(OH)
2
+Mg
3
Al
2
Si
3
O
12
K
DFeMg
BtGr
= X
FeBt
/X
MgBt
X
FeGr
/X
MgGr


Relationship of distribution coefficient with T-P can be given by-

lnK
D
= - (H
o
exch
/3R)1/T+ S
o
exch
/3R [(P-10
5
) V
o
exch
/3RT] (3 is
for the three sites involved in the exchange)

Term in the square bracket can be dropped
Remaining expression can be plotted as a linear function of 1/T
An experimental formulation of GARB
lnK
D
= - (52108 /3R)1/T+ 19.51 /3R [(P-10
5
) x 0.238 x 10
-5

/3RT] (Ferry and Spear 1978)

V
o
exch
is very small in above so that P does not affect T
estimation; But the above is for ideal mixing of Fe-Mg
e.g. sources of errors and deviation are- Ca and Mn in
grt; Ti substitution in bt and problems if large Fe+3 and F
or Cl are present (e.g. in granulite);
All these affect activities of Fe and Mg in grt and bt

When solid solution results in impure phases the
activity of each phase is reduced
Use the same form as for gases (RT ln P or ln f)
Instead of fugacity, we use activity
Ideal solution: a
i
= X
i
n = # of sites in the phase on
which solution takes place
Non-ideal: a
i
=
i
X
i

where
i
is the activity coefficient
Example: orthopyroxenes (Fe, Mg)SiO
3

Real vs. Ideal Solution Models
Activity-composition relationships for the enstatite-ferrosilite mixture in orthopyroxene
at 600oC and 800oC. Circles are data from Saxena and Ghose (1971); curves are model
for sites as simple mixtures (from Saxena, 1973)
Most garnets in pelitic rocks contain Fe, Mg, Mn and
Ca, and most biotites contain Fe and Mg. Thus,
assuming ideal ionic mixing the activities may be
written:
3
3
) (
|
|
.
|

\
|
+ + +
= =
Ca Mn Mg Fe
Fe
X a
Fe alm
3
3
) (
|
|
.
|

\
|
+ + +
= =
Ca Mn Mg Fe
Mg
X a
Mg pyr
3
3
) (
|
|
.
|

\
|
+
= =
Mg Fe
Fe
X a
Fe ann
3
3
) (
|
|
.
|

\
|
+
= =
Mg Fe
Mg
X a
Mg phlog
Thus, the equilibrium constant can be written:


When dealing with element partitioning, it is common
to define a quantity called the distribution coefficient:
3
|
|
.
|

\
|
=
Bt
Mg
Grt
Fe
Bt
Fe
Grt
Mg
X X
X X
K
( )
( )
3
1
K
Fe Mg
Fe Mg
X X
X X
K
Bt
Grt
Bt
Mg
Grt
Fe
Bt
Fe
Grt
Mg
D
= =
|
|
.
|

\
|
=
Plots showing the disposition of garnet-biotite tie lines at different
temperatures on two different diagrams.
Three ways of showing the partitioning
of Fe and Mg between two phases such
as garnet and biotite. In (b), K
D
is the
slope and in (c) ln K
D
is the intercept at
ln (Mg/Fe)
Bt
= 0.
FERRY & SPEAR (1978) CALIBRATION
From an experimental study, Ferry & Spear (1978)
derived the relationship:
ln K
D
= -2109/T(K) + 0.782
However, it is also possible to derive an expression
using basic thermodynamic properties. If we make the
assumptions that A
r
C
P
~ 0, and A
r
V is independent of
pressure and temperature we can write:
0 ln ) 1 ( ) bar 1 , K 298 ( ) bar 1 , K 298 ( = + A + A A K RT P V S T H
r r r
Plot of ln K
D
vs. 1/T for experimentally equilibrated garnet-biotite
pairs. P = 2.07 kbar. From Ferry and Spear (1978).
The assumption that A
r
C
P
~ 0 is pretty good for an
exchange thermometer because there is substantial
cancellation of products and reactants. Also, if A
r
C
P

were substantially different from zero, the plot of ln K
D

vs. 1/T would be curved.
Ideal mixing can only be assumed when Fe and Mg
are the only major components in garnet. Ca-Mg
mixing is non-ideal and a Margules approach would
be required.
How do we apply the geothermometer when dealing
with zoned garnets?
P-T diagram in which lines of constant K
D
for the garnet-biotite
thermometer have been plotted.
OTHER EXCHANGE
GEOTHERMOMETERS
Fe-Mg exchange between garnet and cordierite.
Fe-Mg exchange between garnet and clinopyroxene.
Fe-Mg exchange between garnet and orthopyroxene.
Fe-Mg exchange between garnet and hornblende.
Fe-Mg exchange between garnet and chlorite.
Fe-Mg exchange between garnet and olivine.
Fe-Mg exchange between biotite and tourmaline.
Fe-Mg exchange between garnet and phengite.
Fe-Mn exchange between garnet and ilmenite.
Stable isotope exchange.
SOLVUS GEOTHERMOMETERS
As name implies, these are based on a solvus
between two immiscible phases. Examples include:
1) Two-pyroxene geothermometry - distribution of Ca
and Mg between cpx and opx.
2) Calcite-dolomite geothermometry - distribution of
Ca and Mg between calcite and dolomite.
3) Two-feldspar geothermometry - distribution of K
and Na between alkali feldspar and plagioclase.
4) Muscovite-paragonite - distribution of K and Na
between muscovite and paragonite.
NET TRANSFER REACTIONS
Many geobarometers are based on net-transfer
reactions, i.e., reactions that cause the production and
consumption of phases. Such reactions often result in
large volume changes, so the equilibrium constant is
pressure-sensitive.
An example (garnet-plagioclase-olivine):
3Fe
2
SiO
4
+ 3CaAl
2
Si
2
O
8

Ca
3
Al
2
Si
3
O
12
+ 2 Fe
3
Al
2
Si
3
O
12
THE GASP GEOBAROMETER
GASP = Garnet-AluminoSilicate-Plagioclase-quartz
This is a net transfer geobarometer based on the reaction:
3CaAl
2
Si
2
O
8
Ca
3
Al
2
Si
3
O
12
+ 2AlSi
2
O
5
+ SiO
2

Solid solution of grossular with other garnet components,
mainly, almandine and pyrope, lowers the pressure at
which the right-hand side assemblage is stable.
Solid solution of anorthite with albite in plagioclase tends
to stabilize the left-hand side to higher pressures.
Because grossular often quite dilute, the pressure-
lowering effect prevails.
Garnet + plagioclase + Al
2
SiO
5
+ quartz is a common
assemblage in the crust.
THERMODYNAMICS OF GASP
Newton & Haselton (1981)
This geothermometer is based on the thermodynamic
equation:



Experimental determinations of the P-T curve for the
end-member reaction (with kyanite) have been
expressed in the form:
P = a + bT
0 ln ) 1 ( ) (
3
4 3 2
=
|
|
.
|

\
|
+ A + A
plag
an
grt
SiO CaAl
r
o
r
a
a
RT V P T G
From Goldsmith (1990) we get:
P(kbars) = -2.10 + 0.0232T(C)
and the error is expected to be on the order of 1 kbar.
The analogous expression for the GASP reaction with
sillimanite is:
P(kbars) = -0.6 + 0.0236T(C)

Once you have P(T), then, A
r
G(T) = -P(T)A
r
V,
where P(T) is the breakdown pressure of anorthite at
a given temperature for the end-member reaction, and
A
r
V = -66.2 cm
3
mol
-1
at 298 K and 1 bar (the molar
volume change for the end-member reaction).
end

KOZIOL & NEWTON (1988)
CALIBRATION
P(kbar) = -1.093 + 0.0228T(C)

This calibration can be written in the alternative form:
0 = -48,357 + 150.66 T(K) + (P-1)(-6.608) + RT ln K
Thus, for this calibration we get:
A
r
H(298 K,1 bar) = -48,357 J mol
-1

A
r
S(298 K,1 bar) = -150.66 J K
-1
mol
-1

A
r
V(298 K,1 bar) = -6.608 J bar
-1
mol
-1

and it is assumed that A
r
V is constant and A
r
C
p
= 0.
Margules Parameters For Garnet
Hensen et al. (1975) obtained the expression:
W
Ca-Mg
= 7460 - 4.3T(K)
in calories per MAl
2/3
SiO
4
. Which goes into the equation
Note the strong temperature dependence of non-
ideality.


The above expression assumes a regular or
symmetrical solution model, which is usually only
strictly valid over a narrow range of composition. In the
work of Hensen et al. (1975) the mole fraction of
grossular only ranged from 0.1 to 0.22.

2
4 3 2 4 3 2
ln
SiO MgAl Mg Ca
grt
SiO CaAl
X W RT

=
Newton et al. (1977) found the relation:
W
Ca-Mg
= 3300 - 1.5T(K) (calories per MAl
2/3
SiO
4
)

Cressey et al. (1978) studied Ca-Fe mixing in garnet.
They found that W
Ca-Fe
can be assumed to be near zero
for most natural garnet compositions.
Additional experimental data suggest that W
Fe-Mg
can be
neglected relative to W
Ca-Mg
.
Experimental data suggest that W
Ca-Mn
is probably not
negligible, so the solution model employed here should
only be employed for low-Mn garnets.

For a ternary garnet solid-solution, the activity of the grossular
component is given by:



But based on the previous discussion, this simplifies to:




Activity Model for Plagioclase
Based on Kerrick & Darkens (1975) Al-avoidance model,
Newton et al. (1980) gave the following equation:

alm pyr Mg Fe Mg Ca Fe Ca pyr Mg Ca alm Fe Ca
grt
gross
X X W W W X W X W RT ) ( ln
2 2

+ + + =
( )
alm pyr pyr Mg Ca
grt
gross
X X X W RT + ~

2
ln
)
`

+
+
= ) 9392 2050 (
) 1 (
exp
4
) 1 (
2 2
an
an an an
plag
an
X
RT
X X X
a
PARTIAL MOLAR VOLUMES OF
GROSSULAR
In garnets where grossular is a minor component, the partial
molal volume of grossular is considerably different from the molal
volume.
The partial molar volume of grossular in solid solution with pyrope
can be expressed as:

where



and A = 125.25; B = -11.205; C = -0.512; D = -0.418; E = 0.94; F
= 0.083. For the grossular-almandine join we get: A = 125.24; B =
-8.293; C = -1.482; D = -0.480; E = 0.914; F = 0.066.
|
|
.
|

\
|
|
|
.
|

\
|
+ + =
2
exp 1
2
2
Z
F
ZX
D CX A V
pyr
pyr gross
F
E X
Z
pyr

=
Partial molal volume curve of Ca
3
Al
2
Si
3
O
12
in pyrope-grossular and
almandine-grossular garnets, as constructed by Cressey et al. (1978).
Calculated pressures of well-documented parageneses of garnet +
plagioclase + Al
2
SiO
5
+ quartz. Open symbols are sillimanite
occurrences and filled symbols are kyanite occurrences.
P-T diagram contoured for values of K
eq
for the GASP geobarometer
reaction.
Geothermobarometry
P-T-t Paths
The results of applying the garnet-biotite geothermometer of Hodges and Spear (1982) and the GASP geobarometer of Koziol (1988, in
Spear 1993) to the core, interior, and rim composition data of St-Onge (1987). The three intersection points yield P-T estimates which
define a P-T-t path for the growing minerals showing near-isothermal decompression. After Spear (1993).
Geothermobarometry
P-T-t Paths
SOME OTHER NET-
TRANSFER EQUILIBRIA
GRIPS = Garnet-Rutile-Ilmenite-
Plagioclase-quartz.
Ca
3
Al
2
Si
3
O
12
+ 2Fe
3
Al
2
Si
3
O
12
+ 6TiO
2

3CaAl
2
Si
2
O
8
+ 6FeTiO
3
+ 3SiO
2


GRAIL = Garnet-Rutile-Aluminosilicate-
ILmenite-quartz.
Fe
3
Al
2
Si
3
O
12
+ 3TiO
2
3FeTiO
3
+
Al
2
SiO
5
+ 2SiO
2

OTHER GEOBAROMETERS
Phengite barometry - phengite content of white
mica.
Sphalerite barometry - Fe content of ZnS coexisting
with pyrrhotite and pyrite.
Magnetite-ilmenite thermometry - Based on two
reactions:
4Fe
3
O
4
+ O
2
6Fe
2
O
3
Fe
2
TiO
4
+ Fe
2
O
3
Fe
3
O
4
+ FeTiO
3

Hornblende barometry - Based on the Al content of
hornblende in certain igneous assemblages.

SOURCES OF ERROR
1) Accuracy of experimental calibrations.
2) Error in measured A
r
V (usually determined
independently of (1)).
3) Analytical imprecision in microprobe analysis.
4) Uncertainty of electron microprobe standard
compositions and correction factors.
5) Cross correlations between errors in temperature
estimates (for barometers) and errors in pressure (for
thermometers).
6) Uncertainties in mineral activity models.
7) Geological uncertainty arising from compositional
heterogeneities.
P-T diagram showing results from garnet-biotite thermometry (steep
lines) and GASP barometry (gentle lines). All curves are calculated
from a single set of mineral analyses from Mt. Moosilauke, NH, USA.
P-T diagram showing the results of thermobarometry calculations on
sample SC-160 from Coolen (1980) using a consistent set of
thermobarometers.