Geos 306, Fall 2009, Lecture 14
Mineralogy of the Upper Mantle
In our last lecture we discussed the mineralogy of the lower mantle. There
appears to be three phases of importance there, perovskite, magnesio-wüstite,
and minor stishovite, though we have little direct evidence for this. Instead
we assume the P and T conditions and, with the help of laboratory experiments,
we deduce that these are the major phases.
The mineralogy of the upper mantle is better constrained. We can observe
many of the phases directly in rocks, like those from San Carlos, Arizona.
The major phases of the upper mantle include olivine (~60%) and pyroxene
(~25%), with a minor aluminous component (~10%) (plagioclase, spinel or garnet,
depending upon depth).
In general, the deeper in the earth, the greater the density of the minerals.
At the relatively shallow depths of the upper mantle, we have distorted
closest-packed arrays of O atoms, with Si in the tetrahedral sites and
Mg/Fe in the octahedral sites. As pressure increases, the atoms must pack
even more tightly, so the collection of atoms undergo phase transitions
to structure with intrinsically denser packing. One of the ways to do this
is to put Si into octahedral coordination. The transition zone is
made up of minerals with a mixture of 4 and 6 coordinated Si. Below the
transition zone, in the lower mantle, all Si is 6 coordinated. Above the
transition zone, in the upper mantle, all Si is 4 coordinated. In order
for Si to be able to fit in a six-coordinated site, the radius of the O
atoms must decrease with increasing pressure, at a greater rate than the
Olivine: (Mg,Fe)2SiO4, with an assumed Fe component
equal to 8-12%. This is the observed concentration from many localities,
such as at San Carlos, Arizona. The structure is based upon a hcp array
of O atoms, with Mg/Fe in octahedral sites and Si in tetrahedral.
It is the most common phase in the upper mantle. The structure is capable
of taking up some Ca into an octahedral site. The amount of Ca that it
can take is dependent upon pressure, so it may be possible to use the Ca
content of olivine as a geobarometer. However, the amount of Ca is also
very sensitive to temperature, and we do not know temperature very well.
Therefore, knowledge of the Ca content is not really very useful at this
Mg2SiO4 is called
Fe2SiO4 is called
fayalite, Fa; and
Ca2SiO4 is called
Larnite is not common. We summarize the composition of mantle olivines by writing,
for example, Fo0.89Fa0.10La0.01
Pure forsterite is colorless, but with minor amounts of Fe it is green.
With major amounts of Fe it is black. Thus color can be an indicator of
To remember the formula, start with SiO2, then double it to
Si2O4, then substitute 2Mg or 2Fe for 1Si for charge
balance. The result is Mg2SiO4. It is also possible
to make an olivine with cations of +1 and +3 substituting for the +4 Si.
For instance, LiScSiO4.
Olivines are orthorhombic, and thus are intrinsically anisotropic. However,
olivine displays anisotropic behavior to a greater extent than many other minerals.
It is twice as compressible along the b-cell edge as along the a or c directions.
Olivine is an important component of oceanic ridges and the seismic data
indicate that the crystals display preferred orientation in that environment.
If you have an aggregate of grains of a mineral that displays anisotropic compression, then the most
compressible direction of the grains will tend to align themselves in the direction of the applied stress.
Thus olivine will tend to align itself with its b-axis oriented in the vertical direction.
The animation below shows the crystal structure of olivine cycling back and forth as a function of pressure.
The red spheres are O atoms that become more closest packed with increasing pressure.
The stacking direction of the close packed
layers is vertical, and the b-axis is horizontal.
At pressures around the 410-km discontinuity, Fo-rich olivine transforms
to a ccp structure called
wadsleyite. Most researchers believe that this
transformation is the source of the discontinuity. Iron rich olivines do
not undergo this transformation. At higher pressures, both the Fa-rich
olivine and wadsleyite transform to a spinel structure, (Mg,Fe)2SiO4,
These phases are frequently called the a
-b -g phases of Mg2SiO4.
According to Kudoh and Takeuchi (1985), the pressure at which the olivine transformation occurs (13.7 GPa) coincides
with the pressure at which the structure becomes as closest-packed as it
can get. Hence, in order to become more dense it must transform to a new
Pyroxene: (Mg,Fe)2Si2O6. The important end members
ferrosilite, FeSiO3, and
enstatite, MgSiO3. (Note
that these are the same formula as for perovskite). Pyroxene is a distorted ccp array of O atoms with
Si in tetrahedral coordination,
and Fe/Mg in octahedral. The SiO4 groups form chains that are
linked together through octahedral chains.
One of the cation sites is extremely distorted and the structure is often
illustrated as if this site contained a sphere, as shown above. The coordination
of the cations in this site can range from 4 to 8, depending upon chemistry
and P and T conditions.
Link to spodumene at pressure
There are a wide variety of elements that substitute into the pyroxene
structure. For instance, jadeite, NaAlSi2O6, is another
example. There are also several different symmetries. Ferrosilite and enstatite
are orthorhombic at room conditions, however, at high pressure, (below
220 km) they transform to monoclinic. This transformation is likely to be the cause of the 220 km discontinuity.
Pyroxenes are categorized in terms
of their symmetries, as orthopyroxenes and clinopyroxenes.
The mixing of Fe with Mg, and of clinopyroxenes with orthopyroxenes is
an extremely sensitive indicator of pressure and temperature conditions.
Thus, we can determine the pressure and temperature conditions of the formation
of a mantle rock by examining the mixing states of pyroxene.
Other important pyroxenes are:
To remember the formula, start with SiO2, then triple it to
Si3O6, then substitute 2Mg or 2Fe for a Si for charge
balance. The result is then Mg2Si2O6.
You can also substitute a +1 and a +3 cation for the +4 Si, for instance,
- Link to Downs (2003) pyroxene topology paper
This is a framework structure constructed of corner-linked octahedra (the
B-site) and tetrahedra, with a larger cation in an 8-coordinated site (the
A-site, or dodecahedral site). Important garnets fall along two well-known
trends that are given the nicknames, pyralspite (pyrope,
(contain Al) and ugrandite (uvarovite, grossular,
(contain Ca). They display all sorts of colors, though blue is quite rare. Link
Pyrope is the most important of the mantle garnets, and is indicative of
high pressure environments such as in diamond-bearing kimberlites.
The symmetry is cubic at low pressures, but at higher pressures, in the
transition zone, Si can occupy up to 1/2 of the octahedral sites and so
ordering of the octahedral cations induces tetragonal symmetry. For instance,
Mg3MgSi(SiO4)3 = MgSiO3,
(note the same chemistry as enstatite and perovskite).
With pressure we find that MgSiO3 pyroxene transforms to majorite-garnet
in the transition zone and then to perovskite in the lower mantle.
The garnet structure cannot be catagorized in terms of closest-packing.
Garnet is also an important high tech material. It forms the crystals that
are the pixels of a TV or computer screen. By putting minor amounts of
a rare-earth element into the dodecahedral site, the crystals will glow
different colors when excited by electrons. For instance, one composition
produces red, another produces green and another can produce blue. So,
the pixels of a TV screen are made of 3 separate crystals, each producing
one of the colors. Exciting the crystals to varying degrees produces the
colors seen on the screen.
The following animation shows the structure of garnet changing as a function of volume.
The volume change is a condensation of data from various garnets as a function of pressure, temperature and composition.
Notice that the garnet structure can be classified as a mixed framework of octahedra and tetrahedra and most of the
volume change can be correlated with the bending of the octahedral-Oxygen-tetrahedral angle.
We will see more of the framework structures and this kind of behavior when we examine silica and feldspars in the crust.
Spinel: MgAl2O4. This is an important phase in the upper part of the upper mantle,
and again in the transition zone. It is also an important structure
type with a large number of chemical compositions. There are at least 44 different minerals known with the spinel structure.
The structure is based upon a ccp array of O atoms, with cations in the octahedral and tetrahedral sites.
Many of the spinel minerals have variable cation occupancy in the octahedral and tetrahedral sites. For instance, MgAl2O4, can vary
from the normal state with only Mg in the tetrahedral site, and only Al in the octahedral site, to the inverse state, with only Al in the tetrahedral
site, and all the Mg and the rest of the Al in the octahedral site.
If we define X as the divalent cation, and Y as the trivalent cation, then we say that the normal state is X[Y2]O4, and the inverse state
is Y[XY]O4, where the brackets [ ] indicate the octahedral site. Most natural spinels are disordered and lie between these two end-members.
The disordered configuration can be written X1-iYi[XiY2-i]O4, where the disorder is defined by the
inversion parameter, i. Thus, i = 0 for end-member normal spinel, and i = 1 for end-member inverse spinel.
Spinels that cool slowly will tend to be ordered as normal, with i = 0, or inverse, with i = 1,
while spinels that cool fast are disordered and cations are randomly found at the tetrahedral and octahedral sites, and i = 2/3.
In the Uchida et al (2005) study of the San Carlos spinels
(Link to pdf), they found that 0.10 < i < 0.21.
This led to an estimate of the cooling rates of the San Carlos xenoliths.
In contrast, magnetite is an inverse spinel. The Fe2+ strongly prefers the octahedral site. So the Fe3+ then
fully occupies the tetrahedral site, and half-occupies the octahedral site.
Not only is the inversion parameter an important geological property, but the various cation substitutions that take place are also of interest.
The San Carlos spinels show two significant coupled substitutions, Cr with Al, and Fe2+ with Mg.
The Composition of the Upper Mantle
The dominant minerals found in the deeper upper mantle are Mg/Fe containing olivines,
garnets and pyroxenes with minor Ca components. There is little debate
about this. However, the relative proportions of each phase are not well
understood and are a source of much controversy. With increasing depth,
olivine (a-phase) transforms to wadsleyite (b-phase),
then to the spinel structure (g-phase) and then
to perovskite + magnesio-wüstite. These transformations occur at P
and T conditions that match the 410 km, 520 km and 660 km discontinuities
and are viewed as the cause of the seismic jumps.
Therefore, the magnitude of these seismic jumps are an indication of the
olivine content in the mantle. We can "measure" the olivine content by
matching the magnitude of experimental velocity and density jumps with
those measured by seismology.
For instance, the measured sound velocity increase for a
to b phases at 13.7 GPa at ambient T is about
12%, while the observed seismic change is 4.4%. Thus the abundance of olivine
is estimated to be 4.4/12 = 37%.
A deficiency in this model is (1) the lack of high-T data, and (2) only using pure forsterite instead of the real composition.
The Composition of the Earth
- Reduce the mineralogy of the Earth to olivine, pyroxene and metal phases using the compostion of the Earth as summarized in our first lecture (O = 49%, Fe = 17%, Mg = Si = 14% by atom).
Wenk & Bulakh, Chapter 35