Lecture 8: The Mantle & The Core

The vast majority of the Earth is inaccessible to us — the deepest

we have ever dug is ~7 km, while the radius of the Earth is ~6400
km. Seismic studies tell us that the Earth gets hotter with depth and
pressures increase with depth. Based on these behaviours, the
composition of crustal rocks, the compositions of xenoliths, and the
compositions of the interiors of other planetary bodies recovered
from meteorites, we can start thinking about the mineralogy of the
deep Earth and why it behaves the way it does.

Photos from The Core 03/09/2008 12:27

Lecture 8: The Mantle & The Core

Now Playing Movie/TV News My Movies DVD New Releases IMDbTV Message Boards Showtimes & The deep Earth consists
Tickets IMDbPro IMDb Resume
Login | Register of spherical shells of
Home | Top Movies | Photos | Independent Film | GameBase | Browse | Help
search
different materials. Below
All more | tips the crust is the mantle,
below that the liquid
outer core, and below
Photos from The Core
2019/20 IA Earth Sciences: What’s the Earth made of? 52
that the solid inner core.
See All 39 Photos Both layers of the core
Lecture 14: Journey to the centre of the Earth
« Prev Next »
are predominantly Fe,
View as Slideshow
What’s in the core? with some Ni (likely
The Earth’s core is composed of Fe, with a few
percent Ni and other lighter elements. It is split
~10%) and some light
into two layers: a liquid outer core (roughly the elements (likely ~10% in
total, likely predominantly
size of Mars) and a solid inner core (roughly the
size of the Moon). The latent heat released by the
solidification of the inner core drives the
geomagnetic field and plate tectonics, without S, some O, Si, C).
which it is unlikely that life on Earth would exist.
The core is continuously evolving as the Earth
cools. However, the temperature, chemical
composition, and viscosity of the inner and outer
cores are subject to much uncertainty. Most
worryingly, we still do not know for sure what
the basic crystal structure of Fe is at core
conditions.

Here we discuss some of the possible candidate

structures for Fe in the inner core. These
structures represent different ways of efficiently
packing identical spheres. Many other high Lecture 8: The Mantle & The Core
pressure minerals in the mantle are based on
these principles of close packing, and therefore
these concepts are of fundamental importance.

A close packed layer

Title: The Core
The Core
The most efficient
Photo 27 of 39: Photo date: 25 March 2003
method of packing
Despite being the most difficult
identical spheres in
two dimensions is
A A A A A part of our planet to study, we
to form a close
B
C
B
C
B
C
B
C
know the core is ~90% Fe and
Ni. Because these elements
packed layer (Fig.
A A A A A
61). Each atom is
C C C C
in contact with 6
neighbouring
A
B
A
B
A
B
A
B
A
behave very similarly, to first
atoms arranged in a
hexagon. C C C C approximation we can treat the
Close packed A
B
A
B
http://www.imdb.com/media/rm1862637568/tt0298814
A
B
A
B
A core as being made of just Fe.
Page 1 of 2

layers can be C C C C
stacked on top of
each other to form A
B
A
B
A
B
A
B
A
Fig. 61
A monoatomic material like this
a variety of close
packed structures,
just involves identically sized
the most common
of which are the hexagonal close packed (hcp) and cubic close packed (ccp) structures.
spheres, so adopts a close-
packed structure.
 B B
Hexagonal close packing A A A

To form the hexagonal close packed A A C C

B
structure we stack close packed layers A A
in the sequence ...ABABAB... Atoms
in both the A and B layers are Lecture 8: The Mantle & The Core
surrounded by 12 neighbours: 3 atoms
in the layer below, 3 atoms in the layer above, and 6 atoms in Fig. 62
The Core
its own layer (Fig. 62). Recall from Table 2 in Lecture 9, that
12-fold coordination is typical of structures where we areDespite beingofthe mostsize
difficult
IA Earth Sciences: What’s the Earth made of? 53
packing atoms identical (i.e. where
so the atoms lie
the radius ratio is 1). part of our planet to study, we
hcp ccp
eated by the layer
B and C in Fig.
know the core is ~90% Fe and
B
Cubic close packing B
(ccp) A A Ni. Becausefcc these elements
A
B
C
B
behave very similarly, to first
cking
To form the cubic B close B packedB structure A A A approximation we can treat the
EarthweSciences:
al close stack theWhat’s
packed layerstheA
the A made of?
inEarth C
B
C
53core as being made of just Fe.
ose packed layers
ms sequence
lie Atoms ...ABCABCABC... Atoms in A, B
B B A A
BABAB...
ayers
the andare C layers havehcp
layer identical 12-fold ccp A monoatomic material like this
ghbours:
n Fig. 3 atoms
coordination. Each atom is surroundedFig. 62 by just involves identically sized
atoms in the layer above, and 6 atoms in
12 other atoms: 3 in the layer below, 3 in A A spheres, so adopts a close-
2). Recall from Table 2 in Lecture 9, that
the layer
is typical abovewhere
of structures
A
(with thepacking
we are triangle inverted),
atoms
C
of identical size (i.e. where packed structure.
B B
and 6 in its own layer. The coordination A A A

gackedenvironment is Atherefore
(ccp)
A
similar (butfcc not C
B
C

d layers
identical) to that of the hcp structures, andA A
ose packed structure
nAtoms
thewe would expect both these structures to
CABC... Atoms in A, B
have
3 atoms
entical
similar
12-fold
high densities.
Fig. 62 Fig. 63
tom layer
the is surrounded
above,by and 6 atoms in
The
the layer
from conventional
below,
Table Lectureview
3 in
2 in
the triangle inverted),
Lecture
9, thatof the ccp structure8:is The Mantle & The Core
r.of astructures
The cubic unit
where
coordination cellwewith atoms at
are packing the corners
atoms andsize
of identical in (i.e.
the where
centre of each face (Fig. 63). This structure is
forealso
similarknown
(but not as face centred cubic (fcc).
he hcp structures, and
h these structures to fcc bcc
Body centred
nsities. cubic (bcc)
d structure Fig. 63
w ofThe body
the ccp centred
structure is cubic structure simply consists of a
cubic
atoms
Atoms at A,unit
inthe B cell
corners andwith
in theatoms
centre ofateach
theface
corners and
(Fig. 63). Thisone at is
structure
entred cubic (fcc).
-foldthe centre of the unit cell (Fig. 64). The coordination
rounded bcc
number
(bcc) by is reduced from 12 to 8, and therefore it
below,would3 in simply
bic structure be expected
consists ofto
a have a lower density than
gle
tomsinverted),
either the hcp
at the corners andor
onefcc
at structures.
ordination
cell (Fig. 64). The coordination
ar (but not
om 12 to 8, and therefore it
High
ouctures,
have a lower
andPressure
density thanPhase Diagram for Fe
structures.
ructures to
The question of which is the most stable structure of
e Diagram for Fe
Fe in the core is not a simple one to63
answer. Fe is an
Fig. Fig. 64
h is the most stable structure of
example
ccp structure
a simple isof a material
one to answer. Fe is an
that can adopt different structures according to the temperature and pressure
Fig. 64
the conditions
corners
l that and
can adopt (much
different like carbon
in thestructures
centre of can
each face
according exist
(Fig.
to the as graphite
63). This
temperature at low
andstructure
pressure is pressures and diamond at high pressures).
bic A phase
(fcc).
e carbon diagram
can exist provides
as graphite a map
at low pressures andof the most
diamond stable
at high form of a substance for a given temperature and
pressures).
vides a map of the
pressure most65).
(Fig. stableThe
form labelled
of a substance for a given
regions temperature
of the phaseanddiagram show the range of P-T space over
he labelled regions of the phase diagram show thebccrange of P-T space over

ure simply consists of a

he corners and one at Lecture 8: The Mantle & The Core
64). The coordination
8, and therefore it The Core 2019/20 IA Earth Sciences: What’s the Earth made of? 54
ower density than
.
So, given its simplistic nature
which a particular structure remains
stable. These regions are bounded by
74 and straightforward composition,
solid lines, where one structure D. Alfe` et al.
Liquid
m for Fe the core almost certainly adopts
transforms to another.
and therefore by integrating dFl/dl one obtains particular its melting curve. So, we obtain the Gibbs free

most stable structure of

either an hcp, or fcc, or bcc
Z At
1
low
! pressures
" there is excellent energy from the Helmholtz free energy, and we apply
@Ul between different bccbe???
structure. However, which one of
DF ¼ F " F ¼ agreement
ne to answer. Fe is an 1 0 0experimental
dl
@l
:
Fig.lstudies.
ð29Þ
64 Fe adopts a bcc
equation (16) to obtain the melting curve of Fe. This can
done in a whole range of pressure spanning the Earth’s
these is in our core is very
structure at room temperature
adopt different structures according to the temperature and pressure and core, and the results are shown in figure 5, where an error
can existThis
as also
graphite difficult to determine, because
represents the reversible
at low pressures
pressure.
work done
as the potential energy function transforms
At high
and diamond
is switchedtofrom
on temperatures,
the system Fe band of about plus or minus 600 K has to be added to the
at highthen
an fccUstructure, pressures). bcc
0 to U1. calculated melting curve. For a more in depth discussion
hcp ?? of
ap of the
In most stable
most cases Fe changes between these
form of
a suitable a substance
choice briefly
for theback for a that
to bcc
function given
before temperature
melting to
mixes and points in these calculations
the technical
hcp, dhcp, distorted
we refer the reader
d regions
U0ofandtheUphase
1 is structures as a function of
diagram
simply f(l) ¼
form the
show
l, and
a liquid.
therange At low
of
thermodynamic
temperatures
P-T
high pressures, Fe transforms from a
space and
over
to the original fcc
papers [42,51 – 53]. We also show on the
temperature and pressure — it
formula (29) takes the simple form:
bcc structure to an hcp structure.
figure a number of experimental points. In the low pressure
region these are diamond
bcc anvil cell hcp
experiments, while in
Z 1
undergoes phase transitions. A
DF ¼ F1 " F0 ¼ However,
dlhU1 " U when
0 il : we come to study Fe
ð30Þ
theat high pressure region they correspond to shock
ambient P and T, Fe adopts a
0 simultaneously high temperature and experiments. The scattering between these points reminds
high pressure uncertainties begin tous how difficult it is to perform Fig. these experiments. The
bcc structure.
This way to calculate free energy creepdifferences
in. Different between two
lines correspond theoretical
to different predictions
experimentalare
65
and in good agreement with the
systems is called thermodynamic integration
theoretical attempts [50]. The
to determine theshock data experiments
boundaries. While there[16,17]
is and with some late diamond
usefulness of the thermodynamic good agreement
integration at low pressures, atanvil
formula high cell
pressures there is [13].
experiments tremendous variation the positions of the
expressed in equation (29) becomes phase boundaries
clear when (such experiments
one are highly non trivial!). The high P-T form of Fe has been
identifies U1 with the DFT potential suggested
(free)toenergy
be either hcp, a distorted form of the hcp structure, or a dhcp (double hcp structure)
function,
withpotential
a 4-layer for stacking 4. Constraints on the composition
Until veryof the Earth’s
most core
and with U0 some classical model whichsequence
the (...ABACABACABAC...). recently, geologists
free energy is easily calculated, wereto beintaken
agreement
as a that hcp is the stable
reference We form for Feturn
can now in thethe
core,discussion
but now anto increasing
the topicamount
of theof
evidence suggests
system. Then equation (29) can be used to calculate the that bcc is the stable phase at extreme pressure and temperature.
composition of the Earth’s core. It will be shown that by
DFT free energy of the system by evaluating the integrand combining first principles calculations of free energies, or
Packing efficiency
hU1 7 U0il using first principles molecular dynamics simula- more precisely chemical potentials, with seismic data, it is
tions at a sufficiently large numberOne way
oftovalues
determine
of ltheand
likely structure ofto
possible Feput
in the core is to on
constraints compare the packing of
the composition efficiency of
the core.
 on top of the holes created by the layer
surrounded by 12 neighbours: 3 atoms
underneath (labelled B and C in Fig.
in the layer
below, 3 atoms in the layer above, Fig. 62
61). A Aand 6 atoms in
its own layer (Fig. A62). Recall from Table 2C in Lecture 9, that
B B
Hexagonal close packing 12-fold coordination is typical of structures
A A whereA we are packing atoms of identical size (i.e. where

the radius ratio is 1).

To form the hexagonal close packed A A C C
B
structure we stack close packed layers A A
in the sequence ...ABABAB... Atoms fcc
in both the A and B layers are Cubic close packing (ccp)
surrounded by 12 neighbours: 3 atoms

Lecture 8: The Mantle & The Core

To form the cubic close packed
in the layer below, 3 atoms in the layer above, and 6 atoms in Fig. structure
62
we stack
its own layer (Fig. 62). Recall from Table 2the layers9,inthatthe
in Lecture
sequence
12-fold coordination is typical of structures ...ABCABCABC... Atoms
where we are packing atoms in A, size
of identical B (i.e. where
the radius ratio is 1). and C layers have identical 12-fold
Packing Efficiency
Cubic close packing (ccp) coordination. Each atom is surrounded
12 other atoms: 3 in the layer below, 3 in
fcc by
To form the cubic close packed structure
As pressure increases,
we stack the layers in the
sequence ...ABCABCABC... Atoms
crystals
the layer above (with the triangle inverted),
and 6in inA, Bits own layer. The coordination
like to adoptand more
coordination. Each atom iscompact
C layers have identical 12-foldenvironment is therefore similar (but not
surrounded by
identical) to that of the hcp structures, and
structures, i.e., they
the layer above become
12 other atoms: 3 in the layer below, 3 in
weinverted),
(with the triangle would expect both these structures to

more dense asenvironmentmore of the (but notunit

have similar high densities.
and 6 in its own layer. The coordination
is therefore similar Fig. 63
identical) to that of the hcp structures, and
cell becomes occupied
have similar high densities.
bya atoms
we would expect both these structuresconventional
The to view of the ccp structure is
cubic unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is
rather than Thefree conventionalspace (the
also known
view of the ccp structure is
Fig. 63
as face centred cubic (fcc). vs
packing efficiency increases). As cubic (bcc)
a cubic unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is
also known as face centred cubicBody
(fcc). centred
bcc

such, we can Body think about

centred cubic (bcc) which
The body centred cubic structure
2019/20
bccsimply consists of a
IA Earth
cubic unit cell with atoms at the corners and one at
Sciences: What’s the Earth made of? 53

of the close-packed
cubic unit cell with structures
The body centred cubic structure simply consists of a
atoms at theEach
the
on
layer
centre
corners
top
number
and of is
is the
one
ofisthe
placed
at unitso
holes from
reduced
the(Fig.
cell
created12by
atoms 64).lieThe coordination
to the
8, andlayer
therefore it hcp ccp
more dense tonumber think about how
the centre of the unit cell (Fig. 64). The coordination
is reduced from 12 to 8,underneath
and therefore
would (labelled
it
be expected toBhave
and aClower
in Fig.density than
would be expected to have a lower density than
the core mighteitherbehave. This is an
61).
either the hcp or fcc structures. A A
the hcp or fcc structures. A C

exercise you High will do for the

B B
Pressure Phase DiagramHexagonal
High
for FePressure close packing
Phase Diagram for Fe A A A
The question of which is the most stable structure
ofofwhich is A A C C

tutorial that weFeexample

will have.
To
The form the hexagonal
question close packed
the most stable structure of
in the core is not a simple one to answer. Fe is an Fig. 64 B
structure
of a material that can adopt the we
Fe indifferent core stack
is notclose
structures packed
a simple
according thelayers
toone to answer.andFepressure
temperature is an Fig. 64 A A
2019/20 IA Earth Sciences: What’s the Earth made of? 53 in the sequence
conditions (much like carbon can example
exist as graphite at...ABABAB...
of a material that can
low pressures andAtoms
adopt
diamonddifferent structures according to the temperature and pressure
at high pressures).
A phase
Each diagram
layer provides
is placed so the aatoms
mapin of the most
both
lie
conditions the stable
A andform
(much B of acarbon
layers
like substance
are canforexist
a given
astemperature
graphite and
at low pressures and diamond at high pressures).
pressure
top of(Fig. 65). The labelled regions
layer of the phase hcp ccp
on the holes created by the
surrounded
A phase diagram by 12diagram show the range of P-T space
neighbours:
provides a map 3 atoms over
of the most stable form of a substance for a given temperature and
underneath (labelled B and C in Fig.
in the layer below, 3 atoms in theregions
layer above, and 6 diagram Fig. 62
atoms in show the
61). pressure (Fig. 65). The labelled Aof the Aphase range of P-T space over
its own layer (Fig. A62). Recall from Table 2C in Lecture 9, that
B B
Hexagonal close packing 12-fold coordination is typical of structures
A A whereA we are packing atoms of identical size (i.e. where

the radius ratio Ais 1). A

To form the hexagonal close packed C C
B
structure we stack close packed layers A A
in the sequence ...ABABAB... Atoms fcc
in both the A and B layers are Cubic close packing (ccp)
surrounded by 12 neighbours: 3 atoms

Lecture 8: The Mantle & The Core

To form the cubic close packed
in the layer below, 3 atoms in the layer above, and 6 atoms in Fig. structure
62
we stack
its own layer (Fig. 62). Recall from Table 2the layers9,inthatthe
in Lecture
sequence
12-fold coordination is typical of structures ...ABCABCABC... Atoms
where we are packing atoms in A, size
of identical B (i.e. where
the radius ratio is 1). and C layers have identical 12-fold
Packing Efficiency
Cubic close packing (ccp) coordination. Each atom is surrounded
12 other atoms: 3 in the layer below, 3 in
fcc by
To form the cubic close packed structure
However, temperature
we stack the layers in the has
sequence ...ABCABCABC... Atoms and 6in in the
the layer above (with the triangle inverted),
A, Bits own layer. The coordination
opposite effect coordination.to Each atom pressure,
and C layers have identical 12-fold environment is therefore similar (but not
is surrounded by
identical) to that of the hcp structures, and
causing thingsthe12 other toatoms:
layer above expand
3 in the layer below, 3 in
(with the triangle and
weinverted),
would expect both these structures to

become less environment

dense. is therefore So, how
and 6 in its own layer. The coordinationhave similar high densities.
similar (but not Fig. 63
identical) to that of the hcp structures, and
might we expect this to The
we would expect both these structures
have similar high densities. aimpact
conventional
to view of the ccp structure is
cubic unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is
the crystal structure
The conventionalof view ofthe alsocore?
known as face centred cubic
the ccp structure is
Fig. 63
(fcc). vs
Also, differentaalsocubicknown
elements adopt
unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is
as face centred cubic Body
(fcc). centred cubic (bcc)
bcc

different crystal
Body centred structures,
cubic (bcc) 2019/20 so
The body centred cubic structure bccsimply consists of a
IA Earth Sciences: What’s the Earth made of?
cubic unit cell with atoms at the corners and one at
53

what would The the

cubic ~10%
unit cell with the
on
light
body centred cubic structure simply consists of a
atoms at theEach layer
centre
corners
top
number
and of is the
one
ofisthe
placed
at unitso
holes from
reduced
the(Fig.
cell
created12by
atoms 64).lieThe coordination
to the
8, andlayer
therefore it hcp ccp
element content numberof the core do?
the centre of the unit cell (Fig. 64). The coordination
is reduced from 12 to 8,underneath
and therefore
would (labelled
it
be expected toBhave
and aClower
in Fig.density than
would be expected to have a lower density than
Fe is very susceptible to alloying
61).
either the hcp or fcc structures. A A
either the hcp or fcc structures. A C

(e.g., iron vs High

steel vs stainless
B B
Pressure Phase DiagramHexagonal
High
for FePressure close packing
Phase Diagram for Fe A A A
The question of which is the most stable structure
ofofwhich is A A C C

steel).
To
The form the hexagonal
question close packed
the most stable structure of
Fe in the core is not a simple one to answer. Fe is an Fig. 64 B
example of a material that can adopt structure
the we
Fe indifferent core stack
is notclose
structures packed
a simple
according thelayers
toone to answer.andFepressure
temperature is an Fig. 64 A A
2019/20 IA Earth Sciences: What’s the Earth made of? 53 in the sequence
conditions (much like carbon can example
exist as graphite at...ABABAB...
of a material that can
low pressures andAtoms
adopt
diamonddifferent structures according to the temperature and pressure
at high pressures).
A phase
Each diagram
layer provides
is placed so the aatoms
mapin of the most
both
lie
conditions the stable
A andform
(much B of acarbon
layers
like substance
are canforexist
a given
astemperature
graphite and
at low pressures and diamond at high pressures).
pressure
top of(Fig. 65). The labelled regions
layer of the phase hcp ccp
on the holes created by the
surrounded
A phase diagram by 12diagram show the range of P-T space
neighbours:
provides a map 3 atoms over
of the most stable form of a substance for a given temperature and
underneath (labelled B and C in Fig.
in the layer below, 3 atoms in the layer above, and 6 diagram Fig. 62
atoms in show the
61). pressure (Fig. 65). The labelled regions Aof the phase
A range of P-T space over
its own layer (Fig. A62). Recall from Table 2C in Lecture 9, that
B B
Hexagonal close packing 12-fold coordination is typical of structures
A A whereA we are packing atoms of identical size (i.e. where

the radius ratio Ais 1). A

To form the hexagonal close packed C C
B
structure we stack close packed layers A A
in the sequence ...ABABAB... Atoms fcc
in both the A and B layers are Cubic close packing (ccp)
surrounded by 12 neighbours: 3 atoms

Lecture 8: The Mantle & The Core

To form the cubic close packed
in the layer below, 3 atoms in the layer above, and 6 atoms in Fig. structure
62
we stack
its own layer (Fig. 62). Recall from Table 2the layers9,inthatthe
in Lecture
sequence
12-fold coordination is typical of structures ...ABCABCABC... Atoms
where we are packing atoms in A, size
of identical B (i.e. where
the radius ratio is 1). and C layers have identical 12-fold
Packing Efficiency
Cubic close packing (ccp) coordination. Each atom is surrounded
fcc
12 other atoms: 3 in the layer below, 3 in
by
To form the cubic close packed structure
To calculate packing efficiency:
we stack the layers in the
sequence ...ABCABCABC... Atoms
the layer above (with the triangle inverted),
and 6in in its own layer. The coordination
A, B
and C layers have identical 12-fold
environment is therefore similar (but not
coordination. Each atom is surrounded by
identical) to that of the hcp structures, and
12 other atoms: 3 in the layer below, 3 in
the layer above (with the triangleweinverted),
would expect both these structures to
have similar high densities.
and 6 in its own layer. The coordination

V olumeAtoms,T otal
<latexit sha1_base64="RbLVBT/anCKBuLJZ0deZwQVktcc=">AAACGXicbZDLSgMxFIYzXmu9jbp0M1gEF1JmpKgbodqNywq9QTsMmTTThmaSITkjlKGv4cZXceNCEZe68m1ML4i2/hD4+c45Sc4fJpxpcN0va2l5ZXVtPbeR39za3tm19/YbWqaK0DqRXKpWiDXlTNA6MOC0lSiK45DTZjiojOvNe6o0k6IGw4T6Me4JFjGCwaDAdqtXnUhhkjUkT2MaZNcgY31ak4D5aPRD64JBhXKDArvgFt2JnEXjzUwBzVQN7I9OVxJziwDCsdZtz03Az7ACRjgd5TuppgkmA9yjbWMFjqn2s8lmI+fYkK4TSWWOAGdCf09kONZ6GIemM8bQ1/O1Mfyv1k4huvQzJpIUqCDTh6KUOyCdcUxOlylKgA+NwUQx81eH9LEJCkyYeROCN7/yommcFb3zYumuVCjfzOLIoUN0hE6Qhy5QGd2iKqojgh7QE3pBr9aj9Wy9We/T1iVrNnOA/sj6/AalMqH7</latexit>

environment is therefore similar (but not Fig. 63

a
identical) to that of the hcp structures, and
P = we would expect both these structuresconventional
have similar high densities.
The to view of the ccp structure is
a cubic unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is
V olumeU nitCell
also known as face centred cubic (fcc).
The conventional view of the ccp structure is
Fig. 63
vs
a cubic unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is bcc
also known as face centred cubic Body
(fcc). centred cubic (bcc)
bccsimply consists of a
The body centred cubic structure
-Volume of oneTheatom in terms ofcellthe length of and
one
Body centred cubic (bcc)
cubic unit with atoms at the corners one at
body centred cubic structure simply consists of a

side of the unitcubic

thecell, aunit(treat atoms
unit cell with atoms at thethe
asfromspheres)
centre
corners and of
onethe
at unit cell (Fig. 64). The coordination
centre of the number
cell (Fig. 64). is reduced
The coordination 12 to 8, and therefore it
number is reduced from 12 to 8,would
and therefore it
be expected to have a lower density than
-Volume of total number ofeither atoms the hcp orin
would be expected to have a lower density than
either the hcp or fcc structures.
unit cell
fcc structures.

-Volume of unitHighcell
Pressure as Phasefunction
DiagramHigh of aPhase Diagram for Fe
for FePressure
The question of which is the most stable structure of
The question of which is the most stable structure of
Fe in the core is not a simple one to answer. Fe is an Fig. 64
Fe indifferent
example of a material that can adopt the core is notaccording
structures a simpletoone to answer.andFepressure
the temperature is an Fig. 64
conditions (much like carbon canexample of a material
exist as graphite that can
at low pressures andadopt
diamonddifferent structures according to the temperature and pressure
at high pressures).
A phase diagram provides a map of the most stable
conditions (much form of acarbon
like substance forexist
can a given
astemperature
graphite and
at low pressures and diamond at high pressures).
pressure (Fig. 65). The labelled regions of the phase diagram show the range of P-T space over
A phase diagram provides a map of the most stable form of a substance for a given temperature and
pressure (Fig. 65). The labelled regions of the phase diagram show the range of P-T space over
 on top of the holes created by the layer
surrounded by 12 neighbours: 3 atoms
underneath (labelled B and C in Fig.
in the layer
below, 3 atoms in the layer above, Fig. 62
61). A Aand 6 atoms in
its own layer (Fig. A62). Recall from Table 2C in Lecture 9, that
B B
Hexagonal close packing 12-fold coordination is typical of structures
A A whereA we are packing atoms of identical size (i.e. where

the radius ratio is 1).

To form the hexagonal close packed A A C C
B
structure we stack close packed layers A A
in the sequence ...ABABAB... Atoms fcc
in both the A and B layers are Cubic close packing (ccp)
surrounded by 12 neighbours: 3 atoms

Lecture 8: The Mantle & The Core

To form the cubic close packed
in the layer below, 3 atoms in the layer above, and 6 atoms in Fig. structure
62
we stack
its own layer (Fig. 62). Recall from Table 2the layers9,inthatthe
in Lecture
sequence
12-fold coordination is typical of structures ...ABCABCABC... Atoms
where we are packing atoms in A, size
of identical B (i.e. where
the radius ratio is 1). and C layers have identical 12-fold

Packing Efficiency
Cubic close packing (ccp) coordination. Each atom is surrounded
fcc by
12 other atoms: 3 in the layer below, 3 in
To form the cubic close packed structure
the layer above (with the triangle inverted),
Two main issues with identifying
we stack the layers in the
sequence ...ABCABCABC... and 6in in
Atoms its own layer. The coordination
A, B

the crystal structure of the core:

and C layers have identical 12-fold
environment is therefore similar (but not
coordination. Each atom is surrounded by
identical)
12 other atoms: 3 in the layer below, 3 in
to that of the hcp structures, and

-We don’t know and 6 inthe light

the layer above (with the triangle
its own layer. element
weinverted),
would expect both these structures to
have similar high densities.
The coordination
Fig. 63
composition identical)
environment is therefore similar (but not
to that of the hcp structures, and
The conventional
we would expect both these structures to view of the ccp structure is
-Reaching thehavetemperatures
similar high densities.
and
a cubic unit cell with atoms at the corners and in the centre of each face (Fig. 63). This structure is
also known as face centred cubic (fcc). Fig. 63
vs
pressures simultaneously during
The conventional view of the ccp structure is
a cubic unit cell with atoms at the corners
Body
and in the centre of each face (Fig. 63). This structure is
(fcc). centred cubic (bcc)
bcc

laboratory experiments that mimic

also known as face centred cubic
The body centred cubic structure bccsimply consists of a
Body centred cubic (bcc)
the core is incredibly
The body centredchallenging.
cubic unit cell with atoms at the corners and one at
cubic structure simply consists of a
cubic unit cell with atoms at thethe centre
corners and of
onethe
at unit cell (Fig. 64). The coordination
number
the centre of the unit cell (Fig. 64). is reduced from 12 to 8, and therefore it
The coordination
number is reduced from 12 to 8,would
and therefore it
be expected to have a lower density than
So we don’t really know what the
would be expected to have a lower density than
either the hcp or fcc structures.
either the hcp or fcc structures.

crystal structure of the inner

High Pressure Phase DiagramHigh
core
for FePressure Phase Diagram for Fe

is…
The question of which is the most stable structure of
The question of which is the most stable structure of
Fe in the core is not a simple one to answer. Fe is an Fig. 64
Fe indifferent
example of a material that can adopt the core is notaccording
structures a simpletoone to answer.andFepressure
the temperature is an Fig. 64
conditions (much like carbon canexample of a material
exist as graphite that can
at low pressures andadopt
diamonddifferent structures according to the temperature and pressure
at high pressures).
A phase diagram provides a map of the most stable
conditions (much form of acarbon
like substance forexist
can a given
astemperature
graphite and
at low pressures and diamond at high pressures).
pressure (Fig. 65). The labelled regions of the phase diagram show the range of P-T space over
A phase diagram provides a map of the most stable form of a substance for a given temperature and
pressure (Fig. 65). The labelled regions of the phase diagram show the range of P-T space over

Lecture 8: The Mantle & The Core

The Mantle
As temperature increases, materials tend to adopt fewer phases and
more symmetric crystal structures. Seismic studies and laboratory
experiments argue that we see this readily as we enter Earth’s
mantle. Unlike the core, we can reach pressures and temperatures of
the mantle fairly easily in the lab, so we can start with a mixture of
FeO, MgO, SiO2, CaO, and Al22019/20
O3 in the correct proportions, and heat 55
IA Earth Sciences: What’s the Earth made of?
it and squeeze it, give it timedown,toand itequilibrate, andusing
is impossible to judge the stability then quench
simple criteria the There
like packing efficiency.

system to see which phases are show present

that the balance canin thein favour
be tipped mantle.
is a delicate energy balance between the different possible structures. Sophisticated computations
of any of the candidate structures (hcp, fcc, or bcc)
depending on the nature and concentration of light elements! The only thing we know for sure is
that pure Fe would have the hcp structure under core conditions.

How hot is the core?

The boundary between the inner and outer cores is an equilibrium phase boundary between the
liquid and solid forms for Fe. Hence the temperature of the boundary can be simply determined
from the phase diagram, given that the pressure at the inner core boundary is 330 GPa (Fig. 66).
This yields an estimate of around 6300 K. The presence of light elements (most likely Si, S,and O)
reduces the melting temperature to around 5400-5700 K. This is the same temperature as the
surface of the Sun (5778 K)!

What’s in the mantle?

Unlike the inner core, the temperatures and pressures within the Earth’s mantle are well within the
range accessible to high-PT experiments. To determine the mineralogy of the mantle, therefore, we
simply take a mixture of MgO, FeO, SiO2, CaO, and Al2O3, in proportions representative of the

Lecture 8: The Mantle & The Core

model composition of the bulk silicate Earth (see Fig. 1) and expose it to the temperatures and
pressures appropriate to the geotherm. Experimentally this can be achieved using, for example, a
laser-heated diamond anvil cell (DAC) or large volume press (LVP) (Fig. 67). The resulting
combination of minerals formed at any given pressure can be analysed in-situ using X-ray
diffraction.

transition zone is now often taken as the shell between the

major seismic discontinuities at 410 km and 660 km depth,

These experiments can be

with the upper mantle ending at 660 km (although slightly
high seismic velocity gradients persist to ~900 km).
conducted using either a
(2) A lower mantle with the defining characteristic of smoothly
large volume press (LVP) to a
and slowly varying velocities, which appear to be consistent
with the compression of a uniform material. The lower
diamond anvil cell (DAC).
mantle is usually taken as starting at the 660 km disconti-
nuity. In early models, the lower mantle was referred to as
Heating can be achieved by
layer D, at the bottom of which is a thin layer of anomalous

shining a focussed laser on

seismic structure, labeled the D“ layer in FIGURE 1B.

Minerals and Other Phases Present at Depth

the sample, or by using a
One of the few mantle rocks found at the surface is kimberlite,
which traditional resistance
contains abundant olivine heater.
and, in some cases, xeno-
liths of peridotite and eclogite. The presence of diamonds
Perform
and other high-pressure measurements
indicator minerals has led to the at
suite of T and P.
general view that kimberlites tap upper mantle sources to
about 200 km depth (a very few of them may tap deeper
sources). These rocks suggest an olivine-rich uppermost
mantle, at least under cratons, with lesser amounts of eclogite.
Thus, early studies of mantle mineralogy focused on the
Fig. 67
mineral olivine, (Mg,Fe)2SiO4. Among the more important FIGURE 2 The temperatures (T) and pressures (P) in the Earth. Pres-
studies were those by Katsura and Ito (1989) and Akaogi et sures at the core–mantle boundary and inner–outer core
boundary are shown. At greater pressure (or depth), the uncertainties in
al. (1989), who showed that magnesium-rich olivine trans- temperature increase (shown by the orange shaded region). Different
forms into a high-pressure phase called wadsleyite (β-phase), fields show the P–T regions that can be explored using a large-volume
at a pressure that corresponds well to the seismic disconti- press (LVP) or a diamond anvil cell (DAC). Temperatures that can be
nuity at 410 km depth. We therefore infer that this olivine achieved using resistance or laser heating with the diamond cell are indi-
cated. Below room temperature is the cryogenic region. FIGURE COURTESY
to wadsleyite phase transformation is responsible, at least in OF G. SHEN
part, for the jump in velocity at 410 km depth. At a pressure
corresponding to ~520 km depth, β-phase transforms into
ringwoodite, or γ-(Mg,Fe)2SiO4, with a cubic structure related
to the mineral spinel, MgAl2O4. However, olivine-composition because it is well stirred by whole-mantle convection, or is
minerals cannot be the only minerals present in mantle it divided into separate partially isolated layers of different
s from this exercise is a remarkably simple picture of what the Earth is made of (Fig.
more remarkable, however, is that this simple picture provides a quantitative
or the origin of the various seismic anomalies that are observed at key depths within
g the 410 km and 660 km anomalies that mark the upper and lower bounds of the
Lecture
e). Each of the major seismic anomalies 8: ThetoMantle
is observed & The
coincide with Core
a fundamental
structure and nature of the minerals present. On this basis the
These measurements paint a comparatively simple Earth can be divided
picture of the
ntle, transition zone, andoflower
inside mantle:
our planet.
e:
A This issue focuses on experiments at high pressure and tem-
perature and what they tell us about Earth’s deep interior,
when combined with seismology data. In such experiments,
the physical and chemical properties of materials are meas-
ured at extreme pressure–temperature conditions that are as
close as possible to those inside the Earth (FIG. 2). As noted
throughout this issue, there is a natural and intimate con-
nection between the disciplines of mineral physics and seis-
mology. Part of this connection comes through the elastic
properties of minerals, which determine the velocities of
ne:
seismic waves throughout the Earth. One of the very earliest
γ) efforts to exploit this relationship was by Williamson and
Adams (1923), who used the available seismic data and elas-
4
ticity measurements on rocks and minerals to calculate a
density
56 distribution for the Earth. They also proposed a
) 2019/20 IA Earth Sciences: What’s the Earth made of?
compositional model for the interior that bears a remarkable
B
similarity to present-day models.
4 What emerges from this exercise is a remarkably simple picture of what the Earth is made of (Fig.
68). What is more remarkable, however, is that this simple picture provides a quantitative
Sources of information on the Earth’s interior include geo-
et (Mj-Gt)
explanation for the origin of the various seismic anomalies that are observed at key depths within
chemical and petrological studies of meteorites, basalts, and
the
Si)[6] (Simantle Lecture
(e.g the 410 km and 660
3)[4]O12 8: The
km anomalies Mantle
that mark & The
the upper Corebounds of themantle rocks, as well as results from the fields of geodesy,
and lower
transition zone). Each of the major seismic anomalies is observed to coincide with a fundamental
geodynamics, and cosmochemistry, to name a few. In the
change in the structure
e (peridotite): Beneath andthe
nature of thethe
crust, minerals
rocky present.
part On
of this
ourbasis the Earth
planet can be divided
is broken into limited
three space of this issue, it would be impossible to cover
into upper mantle, transition zone, and lower mantle:
shells: all of these topics. Even our attempt to focus the issue on
results provided by high-pressure mineral-physics data and
Lower mantle:-Upper mantle
4 olivine seismology is necessarily just an introduction. Space does
This issue focuses on experiments at high pressure and tem-
-Transition zone A not allow a comprehensive discussion of all techniques and
perature and what they tell us about Earth’s deep interior,
perovskite interpretations arising from the increasingly rich datasets
(Mg, Fe)SiO3 -Lower mantle
e (OPx) Lower when combined with seismology data. In such experiments,
these interconnected disciplines provide. Indeed, an important
the physical and chemical properties of materials are meas-
O6 CaSiO3 mantle
part of theured
fieldatof mineral physics is the theoretical calcula-
extreme pressure–temperature conditions that are as
Upper TZ tion of mineral
close asproperties,
possible tooften under
those insidepressure–tem
the Earth (FIGperature
. 2). As noted
e (CPx)
ferropericlase mantle conditions that cannot be attained experimentally.
throughout this issue, there is a natural and intimate con-
2O(Mg,
6 Fe)O Theoreticalnection
advances and the
between ever-increasing
the disciplines speedphysics
of mineral of com- and seis-
puters make computational
mology. Part of this mineral
connectionphysics
comesindispensable
through the elastic
(Gt)
Transition zone: for studying planetary
properties of interiors.
minerals, We
whichregret that this
determine theissue of
velocities of
seismic
Elements lacks an waves
articlethroughout the Earth.
on this topic, and we One of the
hope onevery
willearliest
6](Si3)[4]O12
ringwoodite (γ) appear in efforts
a latertoissue.
exploit this relationship
However, was bytheoretical
several recent Williamson and
(Mg, Fe)2SiO4 Fig. 68 Adams
results related to(1923),
the D”who used
layer of the
the available
mantle areseismic data and
referred to elas-
ticity
in the article bymeasurements
Hirose and Lay on(2008
rocks this
and issue).
minerals to calculate a
density distribution for the Earth. They also proposed a
wadsleyite (β) C compositional model for the interior that bears a remarkable
B GENERAL STRUCTURE OF models.
THE EARTH
(Mg,
% of theFe)2SiO4
Earth’s lower mantle is MgSiO3 perovskite and a further 6% is CaSiO3 similarity to present-day
The major divisions of the Earth—crust, mantle, and core—
aking this mineral the Earth's most abundant silicate. The structure of perovskite can Sources of information on the Earth’s interior include geo-
majorite garnet (Mj-Gt) are defined seismologically (FIG. 1). At shallow depths from
ng the principles of cubic close packing: chemical and petrological studies of meteorites, basalts, and
(Mg3) (Mg Si) (Si3) O12
[8] [6] [4] about 6 km beneath
mantle rocks,ocean
as wellbasins and from
as results 30–50 km
the beneath
fields of geodesy,
continentsgeodynamics,
(with a maximum of ~70 km beneath
and cosmochemistry, Tibet),
to name the In the
a few.
Upper mantle (peridotite): Lecture 8: The Mantle & The Core speeds of limited
P-waves andofS-waves
space jump
this issue, dramatically
it would at the
be impossible to cover
Mohorovičić seismic discontinuity (the Moho), which separates
all of these topics. Even our attempt to focus the issue on
olivine (Ol) Upper mantle the crust and mantle.
results Beneath
provided the mantle mineral-physics
by high-pressure is the liquid outer data and
(Mg, Fe)2SiO4 -Olivine
olivine (~40%) core, a low-velocity
seismologyzone where wave
is necessarily just speeds drop suddenly
an introduction. Space does
not allow
(going to zero a comprehensive
for S-waves) discussion of
at the core–mantle all techniques
boundary. At and
-Clinopyroxene (~15%)
orthopyroxene (OPx) interpretations arising from the increasingly
the boundary with the solid inner core, seismic velocities rich datasets
(Mg, Fe)2Si2O6-Orthopyroxene (~15%) rise again.these
Thisinterconnected disciplines
is all that’s needed provide.
to infer Indeed,
Earth’s anlay-
basic important
part ofFar
ered structure. themore
field of mineral physics
observations is the
go into theoretical
the 1-dimen-calcula-
-Garnet (~30%) tion of mineral properties, often under pressure–tem perature
sional whole-Earth models, like the preliminary reference
clinopyroxene (CPx) conditions that cannot be attained experimentally.
Earth model, PREM (Dziewonski and Anderson 1981)
Ca(Mg, Fe)Si2O6 Theoretical advances and the ever-increasing speed of com-
Opx and cpx transform to garnet as pressure increases. shown in FIGURE 1, and a slightly more complex Earth struc-
puters make computational mineral physics indispensable
ture is suggested. We can describe the stratigraphy of the
pyrope garnet (Gt) for studying planetary interiors. We regret that this issue of
mantle as Elements
follows:lacks an article on this topic, and we hope one will
(Mg3)[8](Al2)[6](Si3)[4]O12 appear in a down
later issue. However, several recent theoretical
(1) An upper mantle, to ~660 km depth. This includes
Fig. 68 results related to the D” layer of the mantle are referred to
a “transition zone,” where velocities increase extremely rap-
in the article by Hirose and Lay (2008 this issue).
idly with depth (that is, velocity gradients are high). The
Perovskites C
GENERAL STRUCTURE OF THE EARTH
About 80 vol% of the Earth’s lower mantle is MgSiO3 perovskite and a further 6% is CaSiO3
The major divisions of the Earth—crust, mantle, and core—
perovskite, making this mineral the Earth's most abundant silicate. The structure of perovskite can
FIGURE 1 (A) Major phases present in the mantle at all depths, and tion to (A) are
and (B), with seismically fast (bluish) and slow (reddish)
defined seismologically (FIG. 1). At shallow depths from
their
be derived using the principles of cubic close approximate proportions. (B) One-dimensional seis-
packing: regions of the mantle compared with the PREM model (Woodhouse and
mic preliminary reference Earth model (PREM) of Dziewonski and about 6 km beneath ocean basins and 30–50 km beneath
Dziewonski 1989). Opx: low-Ca pyroxene (enstatite); Cpx: high-Ca
Anderson (1981). The Moho, at 6–40 km depth, is too shallow and thin continents
pyroxene (diopside); Gt:(with a maximum
garnet; Ol: olivine; βof
: β~70 km beneath
-(Mg,Fe) Tibet), the
2SiO4 or wad-
to be seen in this figure. The original PREM model has a discontinuity at speeds 2of
sleyite; γ: γ-(Mg,Fe) SiOP-waves and S-waves
4 or ringwoodite; Mj-Gt:jump dramatically
majorite-garnet solid at the
670 km depth, but more-recent studies put it at 660 km; the latter value solution; CMB: core–mantle
Mohorovičić boundary.
seismic The vertical
discontinuity line
(the in the which
Moho), Opx field
separates
is used in this figure and throughout the text. Thin lines correlate indicates thethetransition pressure
crust and to Beneath
mantle. a monoclinic
the form
mantleof is
Opx.
theFliquid
IGURES
outer
selected discontinuities in A and B. (C) Interior of the Earth and its rela- COURTESY OF A. DZIEWONSKI AND S. SINOGEIKIN
core, a low-velocity zone where wave speeds drop suddenly
(going to zero for S-waves) at the core–mantle boundary. At
 Lecture 8: The Mantle & The Core

Transition Zone
At a depth of 410 km (corresponding to a certain temperature-
pressure combination), olivine undergoes a phase transition into a
phase called wadsleyite (also called 𝛽-olivine), and garnet transforms
into majorite-garnet. At depth of ~525 km, wadsleyite transforms
into ringwoodite (also called 𝛾-olivine). At a depth 660 km,
ringwoodite and majorite-garnet transform into bridgmanite (only
recently got a mineral name, so often referred to as silicate
perovskite) and ferropericlase (Fe,Mg)O. A Ca-bearing perovskite
also forms (called larnite).

Lecture 8: The Mantle & The Core

2019/20 IA Earth Sciences: What’s the Earth made of? 59
Transition Zone
Olivine-Wadsleyite-Bridgmanite: all three
Wadselyite has a structure containing pairs of corner sharing SiOhave the same chemical
4 tetrahedra, making [Si2O7]6-
formula ((Fe,Mg)
groups, whose 2SiO
charge must 4) butby 4Mg
be balanced different crystal
2- and one O structures,
2- atom not bonded so are
to the Si network.
polymorphs.
The oxygen atoms can become readily hydrated, so that wadsleyite can accept up to 3 wt% water in
its structure, making it an important phase for storing water in the deep mantle.
Olivine (Mg2SiO4) Wadsleyite (β-Mg2SiO4) Ringwoodite (γ-Mg2SiO4)

POSITION AND MINERALOGY OF THE pied by Al, accepting Mg (and Fe) and Si. The Al-free garnet
R MANTLE AND TRANSITION ZONE end member (Mg,Fe)4Si4O12 is called majorite. This substi-
tution is favoured with increasing pressure and tempera-
mparing seismic properties with those estimated for
ture. Garnet accepts Mg and Fe into the octahedral site but
lar mineral assemblages, the physical and chemical
not Ca, so the majorite substitution initially decreases the
ties of the mantle can be constrained. A reasonable
proportion of orthopyroxene; however by mid–transition
g point in such an analysis is to choose an average
zone conditions, all pyroxene components are hosted by
mantle chemical composition and then examine the
garnet. At pressures higher than 18 GPa, CaSiO3 perovskite
tch between the seismic properties observed and cal-
starts to exsolve from garnet. At depths greater than 660
d for a mineral assemblage reflecting this composition
km, garnet also transforms into (Mg,Fe)(Al,Si)O3 perovskite,
nction of depth. A number of methods have been
over a wider pressure interval than the ringwoodite trans-
o estimate upper-mantle composition. The simplest
formation, causing the Al content of perovskite to increase
alyses of xenoliths that appear to have been unaf-
over the first 50 km of the lower mantle.
by melting, while others assume that certain elements
mantle have chondritic ratios (i.e. ratios found in The uniformity in the major-element composition of
ites considered to be the unfractionated building MORB over time is apparent evidence that the upper man- Fig. 70
of the Earth). Pyrolite, on the other hand, a synthetic tle has a relatively homogeneous composition, at least at
omposition conceived by Ringwood as the source the scale at which partial melting takes place. However,
The upper mantle (peridotite)
2019/20 IA Earth Sciences: What’s the Earth made of?
r mid-oceanic ridge basalts (MORB), was constructed trace elements and radiogenic isotopes indicate significant 59
xing a basalt composition with what was considered chemical heterogeneity in the basalt source region. These
In the upper mantle, the pressure is reduced enough such that Si is exclusively in its most favoured
he corresponding mantle-melt residue (see Ringwood characteristics must reflect, to some extent, the more
Differences in major-element concentrations between mobile nature of incompatible trace elements, while varia-
tetrahedral coordination
Wadselyite in containing
has a structure all phases. pairs
The pyroxenes that were
of corner sharing
jority of recently proposed upper-mantle compositions tions in major elements should be much smaller. On theSiOdissolved into the majorite garnet
4 tetrahedra, making [Si2O7]6-
phase
groups,atwhose
high pressure have be
now come outby of solution
and to
quite minor (Walter 2004). There are slightly more- other hand, certain geochemical signatures in
charge must balanced 4Mg oneyield an upper
O atom not mantle
bondedmineralogy
2- mantle-
to the Si network. 2-
s differences between the mineral proportions estimated derived magmas are widely believed to be caused by the
dominated
oxygenbyatoms
olivine,
canorthopyroxene,
become readilyclinopyroxene, andwadsleyite
pyrope garnet.
its structure, makingLecture 8:phase
The Mantle water & Thedeep Core
The hydrated, so that
h compositions at, for example, 1.5 GPa (gigapascals). presence of subducted oceanic crust in the mantle. In some can accept up to 3 wt% water in
ompositions result in olivine contents in the range studies, it has been proposed that partial melts from sub-
it an important for storing in the
%. Models that assume chondritic ratios for some ele- ducted crust are a component of most mantle-derived mag- mantle.
The main Al-bearing phase in the upper mantle is garnet at high pressures. This gives way to spinel
have in general lower olivine contents than models mas and that the relative homogeneity of MORB in fact
(MgAl O ) and then plagioclase (CaAl Si O ) at shallower depths. Ringwoodite (γ-Mg2SiO4)
primarily on mantle samples, while pyroliteOlivine
2 falls
4 (Mg2mixing
reflects SiO4)either in magma chambers Wadsleyite
2 2 8or through (β-Mg melt2SiO4)
y in the middle (56%). Although
rious compositions would become
more distinct if we were comparing
pes of basalts they would produce
tial melting in the MORB source
POSITION AND MINERALOGY OF THE pied by Al, accepting Mg (and Fe) and Si. The Al-free garnet Plagioclase
Rdifferences
MANTLEinAND terms of density
TRANSITION ZONE end member (Mg,Fe)4Si4O12 is called majorite. This substi-
astic properties are likely to be at
mparing
mit of whatseismiccan properties
be discriminatedwith those estimated for
tution is favoured with increasing pressure and tempera- CaAl2Si2O8
ture. Garnet accepts Mg and Fe into the octahedral site but
lar mineral observations.
geophysical assemblages, the physical and chemical
not Ca, so the majorite substitution initially decreases the
ties of the mantle can be constrained. A reasonable
proportion of orthopyroxene; however by mid–transition
g1 point
showsinthe
tions ofchemical
mantle
such variation
a pyrolite
an analysisin mineral
composition
composition
is to choose an average
and as then examine the zone conditions, all pyroxene components are hosted by
Spinel
tionbetween
tch of depth thealong
seismicanproperties
oceanic observed and cal- garnet. At pressures higher than 18 GPa, CaSiO3 perovskite
starts to exsolve from garnet. At depths greater than 660
MgAl2O4
drm.for aThis is most
mineral easily described
assemblage reflecting this composition
ding theofdiagram in number
two andof first km, garnet also transforms into (Mg,Fe)(Al,Si)O3 perovskite,
nction depth. A methods have been
over a wider pressure interval than the ringwoodite trans-
oing the olivine
estimate transitions,composition.
upper-mantle each of The simplest
is related formation, causing the Al content of perovskite to increase
alyses of to a seismicthat
xenoliths discontinuity,
appear to have been unaf-
en melting,
transformations in the remain- over the first 50 km of the lower mantle.
by while others assume that certain elements

go
-mantle
and Al-rich
itesphase
consideredtransformations
Wadsleyite consists of pairs of corner-sharing
minerals,ratios
have chondritic which
over
to be the unfractionated
tetrahedra
(i.e. ratios found in The uniformity in the major-element composition of
Fig. 70 Olivine that make
building MORB over time is apparent evidence that the upper man-
broader depthPyrolite,
of the Earth).
zone begins at
intervals.
~14 GPa
up a [Si O ] unit. This charge is balanced by Ca-rich
onThe the tran-
(410
other hand, a synthetic tle
km
omposition conceived by Ringwood as the source the scale at which partial melting takes place. However,
4 Mg2+ ions and one
clinopyroxene
2 7 6-has a relatively homogeneous composition, at least at
,rwhere (Mg,Fe)
mid-oceanic ridge 2SiO
O that is not bonded to the Si network. The oxygens
olivineThe upper mantle (peridotite)
trans-
4basalts (MORB), Ca-poor orthopyroxene
2-was constructed trace elements and radiogenic isotopes indicate significant in this
into the denser structure wads-
xing a basalt composition with what was considered chemical heterogeneity in the basalt source region. These
sometimes
he referredmantle-melt
corresponding
structure
In the uppercan
to as β-phase or
residue
become
mantle, the pressurereadily
is reducedhydrated,
enough such thatmeaning Al-bearing
Si is exclusively
(see Ringwood characteristics must reflect, to some extent, the more phase
thatinwadsleyite
its most favouredcan
ed spinel. At in~17.5 GPa (520 concentrations
km),
Differences major-element
tetrahedral coordination in all phases. The pyroxenes that were dissolved(garnet, spinel, or
between mobile nature of incompatible trace elements, while varia-
into the majorite garnet
yite
joritytransforms
mes
quite
of recently into
termed
minor γ-phase
proposed
(Walter
hold
or silicatephaseup
ringwoodite,
to pressure
3% water,have nowmaking
come out ofitsolution
a key component mantleof
upper-mantle compositions tions in major elements should be much smaller. On the
2004). spinel.
at high to yield plagioclase
an upper
There are slightly more- other hand, certain geochemical signatures in mantle-
Earth’s whole
depending on
mineralogy
roximately
s differences24
he compositions
breaks down to
GPa (660
between
at,an
water budget.
km), ring-
the mineral
dominated
forassemblage
example, 1.5
proportions estimated derived magmas are widely believed to be caused by the
by olivine, orthopyroxene, clinopyroxene, and pyrope garnet. pressure)
of GPa (gigapascals). presence of subducted oceanic crust in the mantle. In some
kite-structured
ompositions result (Mg,Fe)SiO
in olivine 3 and
contents in the range studies, it has been proposed that partial melts from sub-
%.)OModels
magnesiowüstite
that assume(actually chondritic fer-ratios for some ele- ducted crust are a component of most mantle-derived mag-
lase asinMgO
have general is greater
lower olivine The main Al-bearing phase in the upper mantle is garnet at high pressures. This gives way to spinel
than FeO), contents than models mas and that the relative homogeneity of MORB in fact
marks the on
primarily beginning
mantle ofsamples,the lower while pyrolite falls reflects mixing either in magma chambers or through melt
(MgAl O ) and then plagioclase (CaAl Si O ) at shallower depths.
2 4 2 2 8
y. With
in therespect
middleto(56%). Si- and Al-rich
Although
, by 3compositions
rious GPa pyrolite is composed
would become of Fig. 71
FIGURE 1 Mineral volume fractions for the top 1000 km of a pyro-
lite mantle. Small orange and pink regions in the top-
moreperidotite,
distinct butif wewith increasing
were comparing right-hand corner denote the stabilities of feldspar and spinel, respec-
re ofboth
pes basaltsclinopyroxene
they would produce and tively (Ringwood 1991; Stixrude and Lithgow-Bertelloni 2005). Five
yroxene
tial
nto
meltingcomponents
in the MORB
garnet as ainresult
differences terms
are incorpo-
of of
source
thedensity
garnet
inset images show minerals recovered from high-pressure and high-
temperature experiments where the field of view is typically ~0.2 mm. Plagioclase
The olivine, garnet and pyroxene photographs are of natural samples
dral site,
astic which are
properties is normally
mit of what can be discriminated
likely to occu-be at from peridotite xenoliths. CaAl2Si2O8
geophysical observations.
 Lecture 8: The Mantle & The Core

Majorite-garnet

Garnet (Mg3Al2Si3O12) is the main carrier of Al in the upper mantle.

With increasing pressure, it become preferable to substitute 2 x Al3+
for Mg2+ and Si4+, allowing the pyroxene component to ‘dissolve’ into
the parent to create majorite-garnet (Mg3(MgSi)Si3O12).
In the garnet structure, Al occupies octahedra, so in the new
majorite-garnet structure, some Si is in a 6-fold co-ordination.
Although very uncommon on Earth’s surface, this much more dense
configuration is preferred at high pressures.
This ‘dissolution’ occurs gradually across the TZ as P increase,
leading to a smooth change in Si co-ordination with depth.

Lecture 8: The Mantle & The Core

Perovskite
Bridgmanite makes up ~80% 2019/20 of the lower IA mantle, and CaSiO3
Earth Sciences: What’s the Earth made of? 57
makes up a further ~6%. Both of these have what is called a
How to build the perovskite structure: MgSiO3
perovskite structure, making it Step
the1: most abundant mineral structure
Imagine a cubic closed packed arrangement of O anions:
on Earth. Manmade perovskites are also key from the O
point of view of 1
2

high temperature superconductors and solar panels.

Mg

Si 1 1
2 2

2019/20 IA Earth Sciences: What’s the Earth made of? 57

1
How to build the perovskite structure: MgSiO3 2

Step 1: Imagine a cubic closed packed arrangement of O anions:

Step 2: Replace the atoms at the corners of the unit cell with Mg cations:
1
1
2
O 2
O Mg Mg

Mg Mg
1 1

Lecture 8: The Mantle & The Core

Si 1 1
Si
2 2
2 2

IA Earth Sciences: What’s the Earth made of? 57 1 Mg Mg

1
2 2
uild the perovskite structure: MgSiO3
Step 3: Put a Si cation slap bang in the middle of the unit cell:
magine a cubic closed packed arrangement
Step 2: of O anions:
Replace the atoms at the corners of the unit cell with Mg cations:
1 1 1
2 2 2
O O Mg Mg O Mg Mg

Mg Mg Mg
1
1 1 1 1 1 2 1
Si Si Si
Si
2 2 2 2 2 2

Mg Mg Mg 1
Mg
1 1
2 2 2

Step 3: Put a Si cation slap bang in the middle of the unit cell:
eplace the atoms at the corners of the unit cell with Mg cations:
1
2
Now examine1 the coordination environment of the Mg and Si cations. Taken together, the Mg and
2
O Mg Mg O Mgthe O atoms form a cubicMg closed packed arrangement. As shown in Fig. 62 the coordination number
of Mg within the ccp structure will be 12. This is much higher than the usual value of 6 observed in
Mg Mg low pressure phases (see Fig. 8). Most surprising, however is the fact that Si is in 6-fold octahedral
coordination.1Although Si is far smaller than the ‘no-rattle’ limit for octahedral coordination (see
Si 1 1 1 2 1
Si
2 2 2
Fig. 8), theSi increased packing
2
efficiency associated with the higher coordination number outweighs
the desire to be in tetrahedral coordination at extreme pressure.

If we draw a 2x2 array of the perovskite unit cell we see that the Si octahedra share corners to form
Mg 1
Mg Mga 3 dimensional
1
network,
Mg with the Mg cation sitting in the 12-fold coordinated dodecahedral cage in
2 2
ut a Si cation slap bang in the middle of the unit cell:
1
Now examine the coordination environment of the Mg and Si cations. Taken together, the Mg and
2
 Lecture 8: The Mantle & The Core
Perovskite
This leads to the unusual situation where Si is in an octahedra and
the Mg has 12-fold co-ordination, both of which are much high co-
ordination states than these elements are found in in crustal rocks.
Again, this is due toIA the
2019/20 their dense
Earth Sciences: What’s thestructures
Earth made of? that are favoured
58 at
highthepressure.
middle. In reality
the octahedra are
slightly tilted with
respect to one another
to form a distorted
structure with
orthorhombic
symmetry (Fig. 69).
2019/20
Ferropericlase IA Earth Sciences: What’s the Earth made of? 58
the middle. In reality
Ferropericlase is the
the octahedra
other are in
major phase
slightly tilted
the lower withIt is
mantle.
formedto
respect from
onetwo interpenetrating ccp lattices of Mg and O,
another
towith
formMga in [6]-fold octahedral coordination.
distorted
structure with
The reaction MgO + MgSiO3 = Mg2SiO4 defines the
orthorhombic
boundary between the lower mantle and the transition zone
symmetry (Fig. 69).
at 660 km depth, causing a major seismic discontinuity.
Ferropericlase Lecture
Pyrope-majorite garnet 8: The Mantle & The Core
Ferropericlase is the
Garnet is a major component of the Earth's upper mantle and
other majorzone,
transition phase
andinis the main Al-bearing phase.
the lower mantle. It is
Perovskite
formed
Pyrope:from
(Mg3two interpenetrating
)[8](Al 2)[6](Si3)[4]O12 ccp lattices of Mg and O,
with Mg in [6]-fold octahedral coordination.
ThisGarnet
leads to normally
minerals the unusual
have structuressituation
with isolatedwhere
SiO4 Si
is The
inthe
an octahedra
tetrahedra.
reaction Octahedral
tetrahedral
MgO + MgSiO
sites. Mg
sites and
(here the2SiOMg
containing
3 = Mg
Al3+has the 12-fold
) connect
4 defines
2+ ions reside in large [8]-fold
co-ordination,
atcoordinated
660 km depth,
botha major
dodecahedral
causing sites.
of which are much high
boundary between the lower mantle and the transition zone
seismic discontinuity.
co-ordination states than these elements Fig. 69
With increasing pressure, the substitution 2Al3+ = Mg2+ +
arePyrope-majorite
Sifound inpyroxene
4+ causes the ingarnet
crustal
(MgSiO3)rocks.
componentAgain, this
to ‘dissolve’ is garnet
into the
dueGarnetto isthe
structure their
atomajor
produce dense
majorite
component of thestructures
garnet: thatandare
Earth's upper mantle
favoured at 3high
transition zone,
Majorite: (Mg )[8](Mg pressure.
and is the main Al-bearing phase.
Si)[6](Si3)[4]O12
In Pyrope:
reality,(Mgthe octahedra are slightly rotated,
3)[8](Al2)[6](Si3)[4]O12
Note that Si in majorite is in both tetrahedral and octahedral sites.
which acts to increase the density of the
crystal
Garnet even
minerals
The gradual further.
normally
dissolution have structures
of pyroxenes withoccurs
into garnet
tetrahedra. Octahedral sites (here containing Al3+) connect
isolated SiO4 the transition zone (Fig. 68),
throughout
leading to the transition from pyrope to majorite and a gradual change in the coordination
the tetrahedralofsites.
environment Mgmixed
Si from 2+ ions reside in large [8]-fold
tetrahedral/octahedral at the base of the transition zone to purely
coordinated
tetrahedral atdodecahedral
the top of the sites.
transition zone.
Fig. 69
Olivine-wadsleyite-ringwoodite
With polymorphs
increasing pressure, the substitution 2Alof3+(Mg,
= MgFe) SiO4
2+ 2+

SiThe
4+ causes the pyroxene (MgSiO3) component to ‘dissolve’ into the garnet
transition zone is defined by the transformation between three polymorphs (i.e. alternative
structure
structurestowith
produce majorite
the same garnet:
chemical formula) of Mg2SiO4. Olivine transforms to wadsleyite at 410 km
depth, which then transforms to ringwoodite at 525 km, which then breaks down to a mixture of
Majorite: (Mg3)[8](Mg
MgSiO3 perovskite andSi)
MgO[6](Si )[4]Okm,
at 3660 12 defining the boundary with the lower mantle. All three
polymorphs have Si in tetrahedral coordination.
Lecture
Note that Si in majorite is in both8: The Mantle
tetrahedral & The
and octahedral sites. Core
The gradual dissolution of pyroxenes into garnet occurs throughout the transition
0 zone (Fig. 68),
leading to the transition from pyrope to majorite and a gradual change in the coordination
environment of Si from mixed tetrahedral/octahedral at the base of the transition zone to purely
tetrahedral at the top of the transition zone. 500

Mantle mineralogy can start to explain a lot

of Olivine-wadsleyite-ringwoodite
the features we seepolymorphs seismically of (Mg, Fe)2SiO4
in the 1000
deep Earth.
The transition zoneFor instance,
is defined there
by the transformation are
between a polymorphs (i.e. alternative
three
structures with the same chemical formula) of Mg2SiO4. Olivine transforms to wadsleyite at 410 km
number ofthendiscrete
transforms tochanges
ringwoodite at in seismic
Depth (km)

depth, which 525 km, which then breaks down to a mixture of

wave
MgSiO velocity
3 perovskiteinandthe
MgOupper
at 660 km,~700
definingkm of thewith the
the boundary
1500
lower mantle. All three
Earth. This is shown using the Preliminary
polymorphs have Si in tetrahedral coordination.

Reference Earth Model (PREM). 2000

2500

2 4 6 8 10 12 14
Seismic velocity (km/s)
 Lecture 8: The Mantle & The Core
0

By marking on the depths of some of these, 410 km

we can start to understand what their origin 500 525 km

660 km

could be. There are pronounced jumps in

seismic velocity at 410 and 660 km, which 1000

correspond almost exactly to the phase

Depth (km)
transitions in mantle mineralogy. When the 1500

mineral changes structure, its density and

bulk and shear moduli (resistance to 2000
compression/shear) change, impacting their
seismic wave speed. So we can start to
interpret seismic behaviour through the Earth 2500

using mineralogy.
2 4 6 8 10 12 14
Seismic velocity (km/s)

2019/20 IA Earth Sciences: What’s the Earth made of? 56

What emerges from this exercise is a remarkably simple picture of what the Earth is made of (Fig.
68). What is more remarkable, however, is that this simple picture provides a quantitative
explanation for the origin of the various seismic anomalies that are observed at key depths within
the mantle (e.g the 410 km and 660 km anomalies that mark the upper and lower bounds of the

Lecture 8: The Mantle & The Core

transition zone). Each of the major seismic anomalies is observed to coincide with a fundamental
change in the structure and nature of the minerals present. On this basis the Earth can be divided
into upper mantle, transition zone, and lower mantle:

Lower mantle:
A This issue focuses on experiments at high pressure and tem-
perovskite perature and what they tell us about Earth’s deep interior,
when combined with seismology data. In such experiments,
(Mg, Fe)SiO3
the physical and chemical properties of materials are meas-
CaSiO3 ured at extreme pressure–temperature conditions that are as
close as possible to those inside the Earth (FIG. 2). As noted
ferropericlase throughout this issue, there is a natural and intimate con-
(Mg, Fe)O nection between the disciplines of mineral physics and seis-
mology. Part of this connection comes through the elastic
properties of minerals, which determine the velocities of
Transition zone:
seismic waves throughout the Earth. One of the very earliest
ringwoodite (γ) efforts to exploit this relationship was by Williamson and
(Mg, Fe)2SiO4 Adams (1923), who used the available seismic data and elas-
ticity measurements on rocks and minerals to calculate a
density distribution for the Earth. They also proposed a
wadsleyite (β) compositional model for the interior that bears a remarkable
B
(Mg, Fe)2SiO4 similarity to present-day models.

Sources of information on the Earth’s interior include geo-

majorite garnet (Mj-Gt) chemical and petrological studies of meteorites, basalts, and
(Mg3)[8](Mg Si)[6](Si3)[4]O12 mantle rocks, as well as results from the fields of geodesy,
geodynamics, and cosmochemistry, to name a few. In the
Upper mantle (peridotite): limited space of this issue, it would be impossible to cover
all of these topics. Even our attempt to focus the issue on
olivine (Ol) results provided by high-pressure mineral-physics data and
(Mg, Fe)2SiO4 olivine seismology is necessarily just an introduction. Space does
not allow a comprehensive discussion of all techniques and
orthopyroxene (OPx) interpretations arising from the increasingly rich datasets
these interconnected disciplines provide. Indeed, an important
(Mg, Fe)2Si2O6 part of the field of mineral physics is the theoretical calcula-
tion of mineral properties, often under pressure–temperature
clinopyroxene (CPx) conditions that cannot be attained experimentally.
Ca(Mg, Fe)Si2O6 Theoretical advances and the ever-increasing speed of com-
puters make computational mineral physics indispensable
pyrope garnet (Gt) for studying planetary interiors. We regret that this issue of
Elements lacks an article on this topic, and we hope one will
(Mg3)[8](Al2)[6](Si3)[4]O12 appear in a later issue. However, several recent theoretical
Fig. 68 results related to the D” layer of the mantle are referred to
in the article by Hirose and Lay (2008 this issue).
Perovskites C
GENERAL STRUCTURE OF THE EARTH
About 80 vol% of the Earth’s lower mantle is MgSiO3 perovskite and a further 6% is CaSiO3
The major divisions of the Earth—crust, mantle, and core—
perovskite, making this mineral the Earth's most abundant silicate. The structure of perovskite can
are defined seismologically (FIG. 1). At shallow depths from
be derived using the principles of cubic close packing: about 6 km beneath ocean basins and 30–50 km beneath
continents (with a maximum of ~70 km beneath Tibet), the
speeds of P-waves and S-waves jump dramatically at the
Mohorovičić seismic discontinuity (the Moho), which separates
the crust and mantle. Beneath the mantle is the liquid outer

Lecture 8: The Mantle & The Core core, a low-velocity zone where wave speeds drop suddenly
(going to zero for S-waves) at the core–mantle boundary. At
the boundary with the solid inner core, seismic velocities
rise again. This is all that’s needed to infer Earth’s basic lay-
ered structure. Far more observations go into the 1-dimen-
sional whole-Earth models, like the preliminary reference
Earth model, PREM (Dziewonski and Anderson 1981)
shown in FIGURE 1, and a slightly more complex Earth struc-
ture is suggested. We can describe the stratigraphy of the
mantle as follows:
(1) An upper mantle, down to ~660 km depth. This includes
a “transition zone,” where velocities increase extremely rap-
idly with depth (that is, velocity gradients are high). The

FIGURE 1 (A) Major phases present in the mantle at all depths, and tion to (A) and (B), with seismically fast (bluish) and slow (reddish)
their approximate proportions. (B) One-dimensional seis- regions of the mantle compared with the PREM model (Woodhouse and
mic preliminary reference Earth model (PREM) of Dziewonski and Dziewonski 1989). Opx: low-Ca pyroxene (enstatite); Cpx: high-Ca
Anderson (1981). The Moho, at 6–40 km depth, is too shallow and thin pyroxene (diopside); Gt: garnet; Ol: olivine; β: β-(Mg,Fe)2SiO4 or wad-
to be seen in this figure. The original PREM model has a discontinuity at sleyite; γ: γ-(Mg,Fe)2SiO4 or ringwoodite; Mj-Gt: majorite-garnet solid
670 km depth, but more-recent studies put it at 660 km; the latter value solution; CMB: core–mantle boundary. The vertical line in the Opx field
is used in this figure and throughout the text. Thin lines correlate indicates the transition pressure to a monoclinic form of Opx. FIGURES
selected discontinuities in A and B. (C) Interior of the Earth and its rela- COURTESY OF A. DZIEWONSKI AND S. SINOGEIKIN

ELEMENTS 158 J UNE 2008

Lecture 8: The Mantle & The Core

Lecture 8: The Mantle & The Core

Lecture 8: The Mantle & The Core