Department o F Geophysics and Planetary Physics, School o F Physics, The University, Newcastle NE1 7R U, England

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

EARTH AND PLANETARY SCIENCE LEITERS 11 (1971) 1-6.

NORTH-HOLLAND PUBLISHING COMPANY

THE ORTHOPYROXENE GEOBAROMETER

B. J. WOOD and R. G. J. STRENS


Department o f Geophysics and Planetary Physics,
School of Physics, The University,
Newcastle NE1 7R U, England

Received 21 November 1970


Revised version received 16 January 1971

A temperature-composition diagram has been constructed for the system MgSiO3 FeSiO3 at 1 atm, using a
combination of experimental data obtained at 1100 and 1250 ° and thermodynamic calculations. From the informa-
tion contained in this T - X diagram, and that provided by experimental work on the P - T stability field of ortbo-
ferrosilite, and by studies of the distribution of iron between olivine and orthopyroxene, a series of T - X sections of
the system have been calculated for pressures of 0 (X 2.5) to 15 kbars. The ferrosilite content of a metamorphic
orthopyroxene coexisting with quartz and olivine increases by a constant 2.6% Fs per kbar over this pressure range,
and decreases by approximately 1% for each 100 ° increase in temperature.
The T - X sections are thought to be sufficiently accurate to give the absolute pressure of crystallisation to within
1 kbar if the temperature of formation of an olivine-pyroxene-silicaassemblage is known to -+ 100°C. Application
of this geobarometer to the orthopyroxenes of metamorphic rock suggests pressures of crystallisation ranging up to
8.5 kbars. Intermediate values of 3 to 5 kbars are indicated for contact metamorphosed rocks and deep-seated acid
intrusions. The corresponding depths of crystallisation of orthopyroxene-bearing metamorphic rocks are 15 to
35 km ifg has remained constant. In geologically favourable circumstances the geobarometer is sufficiently accurate
to permit secular variations of 1(~o in the universal gravitational constant (G) to be detected.

1. Introduction garnets are complex and largely undetermined. The


AI2SiO 5 solid solutions, although binary, are stable
There are a number o f examples o f binary solid only over a limited pressure range, and they are not
solutions in which one component (A) is stable over a therefore very useful in geobarometry.
wide range o f pressures, whilst the other (B) is stable By contrast, the enstatite-ferrosilite system is
only at high pressures. Under these conditions, the relatively free o f these problems, since b o t h olivine
composition of a saturated solid solution o f B in A at and orthopyroxene solid solutions may be regarded as
any given temperature is a sensitive function o f the binary, and there is only one relevant reaction if the
pressure of crystallisation. Examples quoted by ortho-clino inversion is ignored:
Strens [ 1 ] in a recent review o f geothermometry and
geobarometry include enstatite-orthoferrosilite, olivine + silica polymorph ~ pyroxene.
a c m i t e - j a d e i t e , a l m a n d i n e - p y r o p e and certain
AI2SiO 5 solid solutions, and others exist, e.g. The large volume change and small entropy change of
j a d e i t e - d i o p s i d e [2]. The behaviour of the a c m i t e - this reaction imply a strong sensitivity to pressure,
jadeite system is complicated in many natural samples combined with insensitivity to temperature, which are
by the presence o f large amounts of diopside in solid desirable properties o f a geobarometer [1, 3].
solution, leading to the crystallisation o f the structur- Nafziger and Muan [4] have shown that pyroxene
ally complex omphacites. P y r o p e - a l m a n d i n e garnets compositions up to Fs57 are stable at 1250°C and
commonly contain significant amounts of spessartine 1 atm, and Lindsley [5] and Akimoto et al. [6] have
and andradite, and the reaction relations o f these shown that orthoferrosilite ( F s l 0 0 ) is stable above
2 B.J. Wood, R.G.J. Strens, The orthopyroxene geobarometer

17 kbar at the same temperature, indicating an obtained from the (130) spacing, using the SnO 2
average change in pyroxene composition of 2.6% Fs (I01) reflection as an internal standard [8], and by
per kbar. Since the Fs content of orthopyroxene analysis with the electron probe. In the absence of a
coexisting with olivine and silica can be determined to reliable method of determining the composition of
-+ 1%, the orthopyroxene geobarometer is potentially small pyroxene grains, the Fs content was assumed to
accurate to -+ 0.4 kbar. correspond to the initial Fe/(Fe + Mg) ratio,
A method of calculating T - X sections at any pres- provided magnesiowustite and olivine were absent. In
sure from appropriate phase equilibrium data and an attempt to grow pyroxene crystals large enough
thermodynamic constants has been developed [7], for accurate probe analysis, some runs were made at
and this was first applied to the calculation of a T - X 1250°C using small amounts of sodium tetraborate as
section at 1 atm from Lindsley's P - T d a t a . The a flux. These runs yielded pyroxene plus glass, and
results, and particularly the slope of the boundary set a lower limit of Fs54 on the maximum Fs content
of the pyroxene field, did not agree with Nafziger and of pyroxene at this temperature, since the reaction
Muan's results [4], and we have therefore redeter- silica + borate flux -~ liquid has negative AG.
mined the 1 atm T - X relations at 1100 and 1250°C, Cristobalite and clinopyroxene rather than tridym-
with results that agree with the calculation. ite and orthopyroxene were the dominant polymorphs
formed in these experiments. Both pyroxene and
olivine crystallise during the early stages of sintering
2. Experimental and it is not strictly necessary to reverse the reactions
as all the phases involved are present throughout.
Preliminary experiments on pyroxene stability at However, one sample (Fs57.5) was first equilibrated at
1 atm were made with stoichiometric mixtures of 1 I O0°C in the pyroxene field, and then held at
composition MSiO 3 where M = (MgxFe l - x ) , 1250°C in the 3-phase region: olivine grew, confirm-
prepared from MgO, SiO2, and Fe203 previously ing the location of the boundary.
dried at 700°C, and electrolytic iron powder dried at
120°C (analysis: 99.47% Fe). The mixture was ground,
pressed into pellets, and held in sealed iron capsules 3. Results
for 1 to 2 days at 1100 or 1250°C. The sintered
pellets were darkened by unreacted wtistite. This The results of the experiments at 1 atm are shown
partial failure to react was attributed to the fact that in fig. 1, where they are also compared with data
many oxide grains in a stoichiometric mixture would obtained by other workers. Agreement with the
not be within diffusion range of a silica grain if(as is pyroxene field determined by Nafziger and Muan [4]
probable) the mean diffusion distance (d) is com- is good at 1250°C, but the divergence increases at
parable with the mean grain size (~). Later experi- lower temperatures. We attribute this in part to the
ments therefore utilised a mixture with a considerable diffusion effects noted earlier, and partly to the wide
excess of silica (1MO:2SiO2) , yielding colorless (5% Fs) spacing of the experimental points in the
pellets after a day at 1100°C. Optical and X-ray earlier work. Smith's data [9] (corrected to 1 atm)
examination confirmed that reaction of oxides with are in reasonable agreement with ours, although we
silica was complete. Pellets were ground, pressed, and suspect that some of his runs using pyroxene as
re-run until no further change in the intensity or reactant may exhibit metastable persistence of
spacing of the olivine (130) reflection was discernible. pyroxene into the 3-phase field, and some low tem-
There was some metastable persistence of olivine in perature runs with olivine + silica apparently failed to
the pyroxene field, the amount decreasing with run react: possibly £ > d at these temperatures.
time, and with increasing proportion of excess silica.
The olivine (130) peak was of normal width, implying
a close approach to homogeneity in most samples. 4. Calculation of T - X sections at high pressure
The reaction products were identified optically,
and by X-ray methods. Olivine compositions were The problem is to calculate AG as a function of
B.J. Wood, R.G.J. Strens, The orthopyroxene geobarometer 3

(P, T, X) for the reaction: suggest a slight preference of iron for M(2), and there
is also some evidence of ordering from Mossbauer
(Fe, Mg)2SiO 4 + silica ~ (Fe, Mg) (Mg, Fe)Si206 . work [ 11 ]. For the present, we assume AG~ = 0,
(1) i.e. that olivine forms a quasi-ideal binary solid
The reaction has been written in this way to empha- solution.
size the quasi-ideality of olivine, and the non- By contrast, the Fe-Mg distribution in ortho-
ideality of orthopyroxene, in which iron is strongly pyroxene is strongly ordered, with iron preferring the
ordered into the M(2) position. The thermodynamic M(2) position. The data obtained by Virgo and
properties of both phases depend sensitively on the Hafner [ 12], who equilibrated orthopyroxene at
way in which the Fe and Mg are distributed between 500 to 1000°C, quenched, and determined the result-
the M(1) and M(2) positions, and we start by relating ing intersite distribution of iron by the Mossbauer
the thermodynamic properties to the site populations, method, yield 14 < G ~ x < 19 kjoule/mole. Similar
and to the exchange energies which determine these values are obtained from the experimentally deter-
populations. mined distributions of iron between orthopyroxene
Consider the intersite exchange equilibrium: and olivine [13, 14] and by calculation from the
structure data. The value adopted here is 18. I kjoule/
Fe 2+Mg 1 ~-Fe I +Mg 2 . mole at 1000°C and 1 atm.
The intracrystalline exchange energies being
The partition coefficient, assuming ideal solution known, it is then necessary to calculate the energies
within each cation lattice is: for exchange between pseudo-ideal olivine and the
orthopyroxene M(I) and M(2) positions. These have
K12 = [Fe 2] [Mgl]/[Fe 1] [Mg2] been obtained by fitting isotherms through the
distribution data obtained by Medaris [ 13 ] and by
where the [Fei] and [Mgi] are the appropriate Grover and Orville [14]. The results agree with
atomic fractions. The standard free energy change for calculations [7] from structure data to within
this exchange reaction is then: 2 kjoule/mole:

AG12 =-RTlnK12 AGo 1 = G°l _ VMl~°PX= _ 14.8 kjoule/mole

and AG02 = G61 -


~opx = + 3.3 kjoule/mole
VM2

AGI2 ~, AtHcf- T~hSel + constant both at 1000°C and 1 atm. The standard free energy
change of reaction (1)may then be written:
where ncf is the crystal field stabilisation enthalpy,
Sel is an electronic entropy contribution arising from AG ° = AsU° - T A S ° + P A V - T ( S °l - S~cPX)
the distribution of the iron 3d electrons over the five
3d orbitals, and the constant contains electrostatic and + AG01x 1 + AG02x 2
strain energy terms. Given accurate structure data,
each of these terms can be calculated, providing a where AU° and AS° are the standard internal energy
useful check on the experimental values of AGI2 [7]. and entropy changes for the reaction of pure ortho-
The first step in the calculation of AG for reaction ferrosilite to fayalite plus silica, the S c are the
(1) is the estimation ofAG~ I and AG~I~x, i.e. the entropies of mixing of(Fe, Mg) in olivine and ortho-
exchange energies which characterise the intracrystal- pyroxene, and the x i are the iron contents of the
line distributions (Fe-Mg order) in olivine and ortho- M(I) and M(2) positions in orthopyroxene.
pyroxene. Both X-ray [10] and phase equilibrium The calculations of(T, X) sections were based on
studies [4] suggest that AG~1 as defined before is the assumption that AG = 0 at Fs62, 1 atm, 870°C
negligible for minerals crystallised near 1200 °, 1 atm. (fig. 1). The ortho-clino inversion was ignored in the
However, our calculations from structure data [7] calculation as the energy terms are insignificant. Other
4 B.J. Wood, R.G.J. Strens, The orthopyroxene geobarometer

"c
D ~3oc
Jz,~
\ "%3---- -m IIoc T

90C
-5
\ \ D S -6 70C
rnpg \
mOI" 50C
x\ "6
\

50 60 70 %FS % FS 60 70 80 90 I00

Fig. I. Temperature-compositiondiagramofpa~ of the Fig. 2. Temperature-composition sections of the En-Fs


En.-Fs systemat I arm. The ortho-clino inversionhas been system at pressures from 1 atm to 15 kbar, showing the move
ignored. Diamond symbolsgive the compositionof olivine ment of the boundary of the pyroxene field to higher Fs
(from d 130, and electronprobe analysis)and pyroxene contents with increasing pressure. The fields of low (~) and
(electron probe analysisof pyroxenefrom fluxed run) of high (#) quartz, and of tridymite (T) are indicated. The
limiting compositions.The rectanglesrepresentthe composi- boundary of the pyroxene field is slightly displaced from
tions of the reactants.The boundariesof the olivine and that shown in fig. 1 by the application of a correction for
pyroxene fields determined by Nafziger and Muan are shown the presence of metastable cristobalite, rather than stable tri-
for comparison (NM), as are Smith's data (S) for those runs dymite, in our run products. Smith's run (S) with Fs95 at an
which showed definite evidence of reaction of pyroxene to uncorrected pressure of 12.5 kbar and 9000(2 is plotted for
olivine + silica (or vice-versa). To render the results directly comparison with the calculated 12.5 kbar isobar. The result
comparable, the Fs content of Smith's reactants has been of our hydrothermal run at 2.00 kbar, 7500(2 is also shown
shifted towards the origin by 2.6% Fs pei"kilobar for those (WS): run time was 2 weeks using Fe:Fe304 buffer.
runs made at 0.5 or 1.0 kbar.

thermodynamic data, notably molar volumes, perature of crystallisation known to + 50 ° gives the
thermal expansion coefficients, and compressibilities pressure o f crystallisation to -+ 0.5 kbar, plus any
are taken from refs. [8], [15] and [16]. The volume systematic errors in our calibration. The following
change of reaction (1) was assumed linear between evidence seems to indicate that systematic errors are
3.60 cm 3 for Mg2Si206 and 4.22 cm 3 for Fe2Si206. small: first, the calculation fits both the observed
Olivine, pyroxene, and a silica polymorph are in 1 atm T - X section (fig. l ) a n d Lindsley's [5] P - T
equilibrium for those (P, T, X) values for which AG stability curve for orthoferrosilite within the experi-
of reaction (1) is zero. These points were located by mental error. Second, electron probe analyses o f large
plotting AG against pyroxene composition at some olivine and pyroxene crystals grown hydrothermally
chosen pressure and temperature. Temperature and at 2.00 kbar, 750°C, are in excellent agreement with
pressure were incremented independently in steps o f calculated values (fig. 2). Third, Smith [9] found
100 ° and 2.5 kbar respectively, each pair o f (P, 7') Fs95 to be stable above an uncorrected pressure o f
values yielding one X for which AG = 0. The resultant 12.5 kbar, at 900°C, and this point plots on the
graph (fig. 2) shows the movement o f the boundary calculated 12.5 kbar isobar. Finally, the pressures o f
of the pyroxene field as the pressure is increased. crystallisation calculated below appear geologically
At constant temperature, the variation in the iron reasonable. We conclude that the error in our cali-
content o f pyroxene coexisting with olivine and bration is largely determined by the errors in the
quartz is nearly linear, averaging 2.6% Fs per kbar at available experimental data, rather than errors in the
1100°K. At constant pressure, the ferrosilite content assumptions made, or in the calculation.
decreases by l to 2% Fs for each 100 ° increase in
temperature, the decrease being larger in the tri-
5. Effect of additional ions on pyroxene stability
dymite field. Over most of the (T, X) range, a deter-
mination of the Fs content to + 1%, with the tem- The replacement o f (Fe, Mg) in M(2) by large ions
B.J. Wood, R. G.J. Strens, The orthopyroxene geobarometer 5

such as Ca and Mn, and of (Mg, Fe) in M(1) by A1 or 3.5 kbar if the temperature of crystallisation was
other small ions affects the observed distributions in 800°C.
at least three ways. The first is a simple blocking ef- The determination of palaeogravity, and thus of
fect, which can be allowed for in the calculations by the secular variation (if any) of the universal gravita-
reducing the pyroxene site multiplicities from 1 to tional constant is of considerable importance to the
( 1 - x ) , where x is the mole fraction of the blocking development of cosmological theories, and theories of
ion. The effect of a substitution on the relative planetary origin. On Dirac's postulate [20] that
stability of the pyroxene and olivine structures is G ¢~ T - 1, a 20% change in G would have occurred
much more difficult to evaluate, particularly when a during the last 2000 my. The orthopyroxene geo-
heterovalent substitution such as 2Al for Mg + Si is barometer is sufficiently sensitive to detect this
involved, but the formation of pyroxene appears to change. Geological conditions favourable to the
be favoured in all cases, and the presence of large ions application of this method will not be common
in M(2) probably stabilises clino-relative to ortho- [1, 21] but the preferred method is to use:
pyroxenes. Finally, the phase capable of accommo-
dating the larger amount of a third ion (Ca, Mn, Al) GM 10OAFs , 2
is stabilised by the larger entropy of mixing. The r E - g = 2.6--~-~ m/sec
blocking effect tends to reduce the Fe:Mg ratio,
whilst the "structural" and "entropy" effects increase where AFs is the difference in iron content (in % Fs),
this ratio: the net result is an extension of the stability and Ah the vertical separation (in kin), at the time of
field of pyroxene to higher Fs contents, especially at crystallisation, of two orthopyroxene-olivine-quartz
high tempe~'atures. The monoclinic iron-rich pyroxenes assemblages. The right conditions for the application
from lunar rocks [17] may exemplify this, pyroxenes of this method might be provided by an intrusion
near Wo25 En 5 Fs70 having crystallised at ~ 1125°C several kilometers thick, with thermally metamor-
and low pressure. phosed iron-rich rocks top and bottom, e.g. Biwabik
I18].

6. Applications of the orthopyroxene geobarometer


7. Discussion and conclusions
Excluding minerals containing > 2.5 mole percent
of Ca, Mn, or octahedral Al in solid solution (rather The importance of developing continuous-scale
than in lamellae), the highest iron contents have been geothermometers and geobarometers to replace the
reported for metamorphic orthopyroxenes from discontinuous scales embodied in the concepts of the
eulysites, a number of which have 77 to 88% Fs. metamorphic facies and the petrogenetic grid has
Assuming the temperature of formation of these been emphasised by Strens [1]. The orthopyroxene
rocks to have lain in the range 200-700°C, then the geobarometer appears to be one of the best of the
pressures of crystallization were 4 to 9 kbar, implying possible continuous scale geobarometers, and ortho-
depths of 15 to 33 krn. Detailed petrographic studies pyroxene-olivine-quartz assemblages are common
of pyroxenes from the Biwabik complex [18] in enough for the technique to be applied to such
which iron-rich rocks were metamorphosed at the problems as the determination of palaeogeothermal
contact with a large basic intrusion, show that early gradient and palaeogravity [ 1], pressures of formation
orthopyroxene (Fs77) coexists with quartz and of rocks, and even to checking experimentally
olivine (Fa9 l). The temperature of crystallisation has determined (P, 7) relations, such as those between the
been determined by oxygen isotope thermometry as AI2 SiO 5 minerals.
700 to 750°C. From fig. 2, the pressure of crystallisa- The computational techniques developed in this
tion was 5.0 kbar, corresponding to a depth of 18 kin. work are now being applied to reactions of very com-
Finally, orthopyroxene (Fs70-75) coexists with plex systems such as almandine-pyrope + silica (four
olivine (Fa90) and quartz in a batholithic adamellite ferromagnesian phases, six inequivalent sites) and
from southern Natal [ 19], indicating a pressure of cummingtonite + silica + water (four phases, eight
6 B.J. Wood, R.G.J. Strens, The orthopyroxene geobarometer

sites). Apart from their use in g e o t h e r m o m e t r y and [8] G.W.Fisher and L.G. Medaris, Cell dimensions and X-ray
geobarometry, the ability to calculate (T, X) sections determinative curve for synthetic Fe-Mg olivines, Am.
tbr such systems promises a very substantial reduction Mineralogist 54 (1969) 741.
[9] D.Smith, Stability of iron-rich orthopyroxene, Carnegie
in the n u m b e r of experimental runs required to reach
Inst. Washington Yearbook 68 (1970) 229.
a desired degree of precision. For example, experi- [ 10] J.D.Birle, G.V.Gibbs and J.V.Smith, Crystal structures
mental runs on c u m m i n g t o n i t e w o u l d have to be of natural olivines, Am. Mineralogist 53 (1968) 807.
spaced at intervals of 2% in F e / ( F e + Mg) and 5°C [ 11 ] W.R. Bush, S.S. Hafner and D. Virgo, Some ordering of
in T to reveal the fine detail in the calculations, but iron and magnesium at the octahedrally coordinated
sites in a magnesium-rich olivine, Nature 227 (1970)
with calculated (T, X) sections available tile experi-
1339.
ments can be planned to give the m a x i m u m of infor- [121 D.Virgo and S.S.Hafner, Fe-Mg order-disorder in heated
m a t i o n from the m i n i m u m n u m b e r of runs. orthopyroxenes, Mineral. Soc. Amer. Spec. Paper 2
(1968) 67.
[ 13 ] L.G.Medaris, Partitioning of Fe 2+ and Mg2+ between
References coexisting synthetic olivine and orthopyroxene, Amer.
J. Sci. 267 (1969) 945.
[ 14} J.E. Grover and P.M. Orville, The partitioning of cations
[ 1 ] R.G.J.Strens, Application of geothermometry and geo- between coexisting single and multi-site phases with
barometry to the determination of the palaeogeothermal application to the assemblages orthopyroxene-clino-
gradient and palaeogravity, in: Palaeogeophysics, ed. pyroxene and orthopyroxene-olivine, Geochim. Cos-
S.K.Runcorn (Academic Press, London, 1970) 377. mochim. Acta 33 (1969) 205.
[2] l.Kushiro, Clinopyroxene solid solutions formed by [ 151 S.P.Clark (ed.), Handbook of Physical Constants, Geol.
reactions between diopside and plagioclase at high Soc. Amer. Mere. 97 (1966).
pressures, Mineral. Soc. Amer. Spec. Paper 2 (1969) [ 16 ] C.W.Burnham, Ferrosilite, Carnegie Inst. Washington
179. Yearbook 64 (1965) 202.
[ 3 ] E.D. Olsen and R.F. MueUer, Stability of orthopyroxenes [ 17] J.A.V.Douglas, M.R.Dence, A.G. Plant and R.J. Traill,
with respect to pressure, temperature and composition, Mineralogy and deformation in some lunar samples,
J. Geol. (1966) 620. Science 167 (1970) 594.
[4] R.H.Nafziger and A.Muan, Equilibrium phase composi- [18] B.Bonnischen, Metamorphic pyroxenes and amphiboles
tions and thermodynamic properties of olivines and in the Biwabik iron formation, Dunka River area,
pyroxenes in the system MgO-FeO-SiO:, Am. Mineralogist, Minnesota, Mineral. Soc. Amer. Spec. Paper 2 (1969)
52 (1967) 1364. 217.
[5 ] D.H.Lindsley, Ferrosilite, Carnegie Inst. Washington [ 19] J.R. Mclver, Orthopyroxene-bearing granitic rocks from
Yearbook 64 (1965) 148. Southern Natal, Trans. Geol. Soc. South Africa 69
[6] Syun-lti Akimoto, Y. Syono, H. Fujosawa and E. Komada, (1966) 99.
Polymorphic transitions of pyroxenes FeSiO a and [20] P.A.M.Dirac, A new basis for cosmology, Proc. Roy.
CoSiO3 at high pressures and temperatures, J. Geophys. Soc. Set. A165 (1938) 199.
Res. 70 (1965) 5269. [21 ] A.D.Stewart, Palaeogravity, in: Palaeogeophysics, ed.
[ 7 ] B.J.Wood and R.G.J.S trens, Inter- and intra-crystalline S.K. Runcorn (Academic Press, London, 1970) 413.
equilibria in ferromagnesian silicates (in preparation).

You might also like