magmatic processes

1
magmatic processes

• a principal goal of igneous petrology is to
  understand the origin of the compositional
  diversity of igneous
• in this set of lectures, we will explore the
  processes most commonly thought to be most
  important

2
recall the distribution of igneous rock
compositions
3
what processes will we consider?

• crystal-liquid fractionationpartial
  meltingcrystallization differentiationassimilati
  on
• liquid-liquid processesmagma mixing liquid
  immiscibilitythe Soret effect
• liquid-gas fractionation

4
differentiation by partial melting

• effects of degree of melting
• effects of pressure
• how can we distinguish batch fusion from
  fractional fusion?
• effects of source heterogeneity

5
fractionation by isobaric partial melting
6
a compositional range is also generated by
melting for a binary loop
so, batch partial melting generates continuous
ranges in liquid and solid compositions
7
effects of variable pressure
8
isentropic melting
T
T
Liq FoEn
P20 kbar
Liq En
Liq Fo
P10 kbar
liq
FoLiq En
Fo liq
Qtz liq
P1 bar
En liq
Fo En
En Qtz
10
20
Fo
En
SiO2
P
incongruent melting
congruent melting
weight
9
how can we distinguish batch from fractional
fusion?1 - phase equilibrium insights
fractional fusion in this case generates a
continuous range in solid compositions, but melts
are discontinuous compositionally and thermally
10
does this apply to isentropic melting?
T
Liq En
Liq FoEn
Liq Fo
FoLiq En
P
thus, although it is a little more difficult to
analyze, isentropic fractional fusion also
generates a continuous range in solid
compositions, but melts are discontinuous
compositionally and thermally
11
a compositional range is also generated by
melting for a binary loop
so, batch and fractional partial melting both
generate continuous ranges in liquid and solid
compositions, but the fractional case produces
extended ranges in both and melts over a larger
temperature range
12
the discontinuity in melt composition that
results from fractional fusion leads to yet
another possible explanation of the Daly gap (due
to Yoder)
P1 bar
T(C)
liq
Fo liq
Qtz liq
En liq
Fo En
En Qtz
Fo
En
SiO2
weight
instead of a peridotitic composition, consider
fractional fusion of a quartz-normative
composition
? whereas batch melting generates a continuum in
melt compositions, fractional fusion generates a
bimodal distribution, which could account for the
Daly gap
13
Yoder actually developed this concept using a
ternary diagram
although it is clever, I doubt that this is the
explanation of the Daly gap, but it may provide
the clue
14
consider a magma chamber in a basaltic volcanic
edifice
basaltic magma
heat flow from magma to volcanic edifice
basaltic edifice
melting of the basaltic edifice will produce
silica-rich (i.e., rhyolitic) partial melts, and
the bimodality is simply the contrast between the
basaltic melt (from mantle melting) in the magma
chamber and the composition of low degree partial
melts of the same basalt
15
trace elements

• up to now, we have focused on the behavior of
  major elements (i.e., components that make up
  stoichiometric components of the major phases),
  but significant insights into processes of
  petrogenesis can come from consideration of
  minor and trace elements that typically occur
  only as impurities in the major phases
• here we will develop the principles and equations
  for describing the behavior of trace elements and
  show in particular how they play a critical role
  in distinguishing batch from fractional fusion

16
thermodynamic background

• so far we have focused on one-component systems,
  but to understand trace elements, we need to
  develop a little background for multicomponent
  systems
• we will use a binary system as an example

17

• the key equation for a binary system is

18

• going back to the first equation

19

• so, lets look at how varies with composition
  in a binary system at constant P and T

curve must be concave up (same arguments we gave
for S-H diagrams) since must be at a minimum
subject to all fluctuations ?
20

• there is a convention for how to describe the
  compositional dependence of chemical potential

m2
m1
2
X2
1
mole fraction of component 1
standard state
activity of component 1
21

• how do we portray equilibrium between two phases

A
B
m2
m1
A
B
AB
2
X2
1
22

• suppose we have olivine and liquid in
  equilibrium, and we are interested in the
  partitioning of Ni between the two phases

23

• although trace element partitioning is usually
  described by a constant D, the reality is that
  neither the liquid nor the solid are ideal
  solutions with respect to an NiO component this
  is particularly easy to envision for the olivine
  -- can we add/subtract NiO from an olivine? no,
  but Ni2SiO4 is a legitimate Ni-olivine component,
  so this can probably be reasonably approximated
  as ideal
• we know that the following relation must hold

what is the chemical potential of NiSi0.5O2 in
the liquid?
24
so, the lever rule tells us that
25

• how does D(Ni) actually vary with SiO2 content?

this simple case illustrates how partition
coefficients can vary with melt composition in a
predictable fashion (although people still tend
to assume constant Ds)
26

• the simple graphical construction demonstrates a
  general principle we can generalize our
  equilibrium condition that any stoichiometric
  relationship among phase components also must
  hold among chemical potentials this is often
  more useful than the statement that chemical
  potentials of all components must be the same in
  all phases, since different phases do not always
  have the same components

how might we describe Fe-Mg partitioning between
olivine and melt?
27

• the same exchange equilibrium approach can be
  used to describe the partitioning of Ni between
  olivine and melt bypassing the SiO2 content of
  the melt

I suspect that the KD(Ni/Mg) for olivine is
nearly constant, although I am not sure it really
is
28
but we can rearrange this equation as follows to
predict a dependence of D(Ni) on MgO content
what do actual data look like?
29

• trace elements are usually characterized by
  whether their partition coefficient is lt1
  (incompatible) vs. gt1 (compatible)
• their behavior will depend on the phases that are
  present
• this will vary with the degree of melting, since
  phases disappear sequentially from the residue on
  progressive melting
• minor phases can dramatically alter the behavior
  (e.g, Zr will behave very differently if zircon
  is present than if it is not)
• behavior will differ for different rock types,
  with significantly different minerals (e.g.,
  melting of the mantle will differ significantly
  from melting of the crust)

30
typical values of Ds
31
in addition to effects of melt composition, there
are crystal chemical controls on Ds
availability of sites, mixing on different sites,
whether the cation is a good fit to the size of
the site, charge balance issues, crystal-field
stabilization effects, etc.
olivine
Onuma diagram(s)
32
often see diagrams where the horizontal axis is
D, compatibility increases from left to right
spider diagrams
33

• equations governing the behavior of elements with
  constant D are relatively easy to derive

lets consider first batch fusion, assuming D is
constant
34
what about equilibrium crystallization?
note that this is identical to batch
(equilibrium) melting, except that the
composition of the system is the initial
composition of the melt instead of the solid the
same is true when you use a phase diagram
equilibrium melting is just the reverse of
equilibrium crystallization
35
what do these functions look like?
36
how about fractional fusion?
ns mass element in solid nL mass of element
in liquid ms mass of solid mL mass of
liquid m total mass (constant) y mass
fraction of solid in the system ms/ m x mass
fraction of melt in the system as a
whole Csconcentration in solid ns /ms D Cs/CL
37
how do batch and fractional fusion compare?

• equilibrium and fractional fusion are identical
  at F0
• for incompatible elements, the difference between
  batch and fractional liquids is extreme even at
  low F
• for compatible elements, batch and fractional
  fusion only become distinguishable at high F

38
simple model of ocean crust formation
even if melts are produced by fractional fusion
during decompression and escape upwards, they
would generally mix together in a magma chamber
i.e., magmas are blends of all the melts
delivered to the magma chamber
39
can we model the integral of melts generated by
fractional fusion?
refer back to the derivation for fractional
fusion ns mass element in solid nL mass of
element in liquid ms mass of solid mL mass
of liquid m total mass (constant) y mass
fraction of solid in the system ms/ m x mass
fraction of melt in the system as a
whole Csconcentration in solid ns /ms D Cs/CL
40
how do batch, fractional, and integrated
fractional fusion compare?

• equilibrium and integrated fractional fusion are
  essentially identical, especially for
  incompatible elements

41
so how can we tell batch from fractional fusion?

• cannot tell by looking at melts, since if they
  are blends, fractional and batch fusion cannot be
  distinguished
• however, if we could find the residues, the
  differences would be striking

42
abyssal peridotites (mostly dredged from
fracture zones)
43
abyssal peridotite photomicrographs
44
evidence from abyssal peridotites - Ti vs. Zr

• melts could be either batch melts or products of
  integrated fractional fusion
• looking at the residues, however, they can only
  plausibly be related by fractional fusion

45
rare earth element (REE) partition coefficients
in cpx
light REE (LREE)
heavy REE (HREE)
46
typical rare earth partition coefficients
47
schematic evolution of REE patterns for mantle
melting
48
evidence from abyssal peridotites -- rare earth
elements

• melts can be explained by integrated fractional
  fusion
• the residues can only plausibly be related by
  fractional fusion

49
evidence from abyssal peridotites -- rare earth
elements

• melts that could be in equilibrium with the
  residual cpx are very different from actual
  mi-ocean ridge basalts
• melts of this sort are found in melt inclusions
  in olivines in mid-ocean ridge basalts, so such
  melts are formed and trapped at depth, but
  usually not expressed in magmas because of the
  blending process

50
isotope systems can also provide insights into
magmatic processes
51
evolution of Nd isotope ratios with time
low Sm/Nd
high Sm/Nd
La
Nd
Sm
Lu
La
Nd
Sm
Lu
52
system can be used to determine age
53
isotopic evolution of the earth
high Sm/Nd
La
Nd
Sm
Lu
unfractionated Sm/Nd
La
Nd
Sm
Lu
low Sm/Nd
La
Nd
Sm
Lu
54
multistage evolution of the mantle and crust
55
application to MORB/abyssal peridotite system
abyssal peridotite cpx
how are the MORBs and abyssal peridotites
related? are the magmas integrated fractional
fusion products and the peridotites the residues?
or maybe the MORBS are melts of pyroxenite?
melts in equilibrium with cpx
field for normal MORB
MORB from region of abyssal peridotites
56
how can we test these hypotheses for the
relationship between magmas and peridotites?

• if the MORBs are integrated fractional melts and
  the peridotites residues from a single
  homogeneous peridotitic source, then they should
  all have the same isotopic characteristics (i.e.,
  they all come from a single source over a narrow
  time interval, so they must have the same
  isotopic ratio)

• if the source was a mixture of pyroxenite (low
  Sm/Nd ratio) and residual peridotite (high Sm/Nd)
  that is old, such that the pyroxenites have a low
  Nd isotope ratio and the peridotites have a high
  Nd isotope ratio, then MORBs and abyssal
  peridotites will have different isotopic ratios
  (assuming they did not reequilibrate during
  melting and melt segregration)

if the source has contained basaltic
(pyroxenitic) rocks and peridotitic residues for
a long time, the melts of the pyroxenite will
have 143Nd/144Nd at E and the periodites will be
at D
melts and residues of a homogeneous source will
have 143Nd/144Nd at C
57
how can we test these hypotheses for the
relationship between magmas and peridotites?
peridotites
basalts
? the abyssal peridotites are systematically more
radiogenic in Nd isotopes than are the basalts
from the same part of the ridge suggesting that
the basalts come from pyroxenites that have been
isolated from the depleted peridotites
(previously depleted by fractional fusion) for
enough time to build up the observed isotopic
differences
58
how could we put this together as a melting model?
a key element of this is that all the pyroxenite
is melted out at depth, so none are ever observed
in abyssal peridotites -- very convenient!
pyroxenite ends melting
depleted peridotite
peridotite begins melting
pyroxenite veins
pyroxenite begins melting
59
how much of the compositional variability of
magmas could reflect variability in the
compositions of their sources?
lets look at a complementary radiogenic isotopic
system
this isotopic system is complementary to the
Sm-Nd system, except Rb is highly incompatible
relative to Sr, so partial melts typically have
high Rb/Sr relative to the source, so basalts
evolve to elevated 87Sr/86Sr (whereas they evolve
to low 143Nd/144Nd)
60
how much of the compositional variability of
magmas could reflect variability in the
compositions of their sources?
? there is considerable heterogeneity in the
mantle sources of basalts, extending from the
depleted sources of MORB (DMM) to several
distinguishable types of enriched sources it
is unclear to what extent these isotopic
signatures or fingerprints are associated with
major element variability, but we at least can be
certain that sources of ocean island basalts and
MORBs are not compositionally homogeneous
61
hypotheses to explain source heterogeneity?

• recycling of sediments and oceanic crust
  (pyroxenite again!) by subduction
• mantle metasomatism (e.g., by hydrous fluids or
  mobile small degree melts such as carbonatites)
• ancient depletion and/or enrichment events by
  extraction or addition of basalt

62
variations on the partial melting theme

• zone refining
• filter pressing
• melt-peridotite reaction -- an important
  deviation from the adiabatic assumption

63
we will consider for this example Hawaii, the
classic hot spot volcano
64
recall the plume hypothesis for ocean islands
65
the origin of plumes
66
what is the internal structure of a plume?
67
although the plume head is complex, the main stem
is predicted to be concentrically zoned

• Concentric zoning (temperature and/or
  composition) of the plume maps into temporal
  evolution of magmas and vertical zonation in the
  volcano
• Possibly high variable intermingling of sources
  both vertically and horizontally on many length
  scales

68
plume geometry based on simple physical models

• Considerable variation in temperature, starting
  and ending depth of melting, and average degree
  of melting as a function of position relative to
  the center of the plume, which varies from
  volcano to volcano and as a function of time for
  individual volcanoes
• We thus expect considerable compositional
  variability related to process even without
  variations in source

3D Model by Ribe and Christensen
69
melting as the plume nears the surface/lithosphere

70
what would temperature and melt production look
like at the top of the plume?

• Considerable variation in temperature, starting
  and ending depth of melting, and average degree
  of melting as a function of position relative to
  the center of the plume, which varies from
  volcano to volcano and as a function of time for
  individual volcanoes
• We thus expect considerable compositional
  variability related to process even without
  variations in source

71
lets map this into a simple two-dimensional
melting model
72
what happens when melt ascends through the
horizontally moving zone above the zone of melt
generation?

• Allow ascending melts from depth to interact
  chemically and thermally with mantle as they
  ascend.
• Superheat from ascending magma relative to
  residual peridotite generates additional melt,
  resorbing pyroxenes and crystallizing olivine.

73
a simple two-dimensional melting model, including
melt-residue chemical and thermal interaction
74
can we use some simple phase diagrams to analyze
this process?
L10
L20
T
S
Liq FoEn
P20 kbar
L20
L10
?
P10 kbar
10
20
Fo
En
P

• note the approximation of using an S-X diagram,
  when an H-X diagram should be used for the final
  step
• the process produces extra melt
• interaction between the ascending melt and
  residual peridotite leads to increasing SiO2
  content of the melt (in this case, pinned to the
  FoEnliquid univariant curve) by resorbing En
  and precipitating Fo this is why porous flow
  leads to dunitic zones
• this cannot be analyzed by a T-X diagram

75
is there evidence of this process in Hawaiian
lavas?
lets look at data from drilling in 1993 and 1999
into the flank of Mauna Kea volcano

• Continuous coring into the Mauna Kea volcano at
  sites in Hilo, Hawaii
• gt95 recovery, to a total depth of 3.1 km below
  sea level
• Penetration through 1 km of subaerial lavas, 2
  km of submarine deposits, both hyaloclastites and
  pillows

76
there are significant variations in SiO2 contents
of Mauna Kea lavas with depth

• SiO2 contents of lavas are highly variable.
• SiO2 contents in the submarine section are
  bimodal.
• Transitions between high- and low-SiO2 magmas are
  generally abrupt.

77
other compositional characteristics correlate
with SiO2 content
78
mass balance calculation to generate high-SiO2
magmas from low-SiO2 magmas

• the average high-SiO2 magma can be generated from
  the low-SiO2 magma by resorption of opx and cpx
  and crystallization of olivine

but is this thermodynamically feasible or
reasonable?
79
use MELTS to model this process in a
thermodynamically valid and consistent manner

• Reaction coefficients forward modeled by MELTS
  can generate the full suite of major elements in
  high-SiO2 melts by interaction of low-SiO2 magmas
  generated at 30 kbar with lherzolite at 20-25
  kbar, leaving dunite or harzburgite residue

80
crystallization differentiation

• as you extract heat, melts will crystallize, but
  since the crystals and liquid usually differ in
  composition, the liquid composition changes, and
  if the crystals and liquid can be separated, the
  composition of the magma will change
• note that magma ascending adiabatically will
  usually have superheat, but most magmas do not
  erupt in superheated condition

81
mechanisms for separating crystals and liquid

• gravitational segregation
• mantling of early-formed crystals
• crystal growth on magma chamber boundaries
• flow segregation

82
primary vs. parental magmas

• primary magmas are magmas that have not changed
  composition since being formed by partial
  melting until recently, this was a very clear
  concept, but if magmas coming from the mantle are
  blends of melts of heterogeneous sources and/or
  fractional melts from over a pressure range
  and/or melt-peridotite interaction during magma
  ascent, the whole concept become somewhat muddied
• parental magmas are magmas from which an
  observed magma was derived by crystal
  fractionation

83
how could we recognize magmas that are
unfractionated since leaving the mantle?
basically, we need to compare observed magmas
with what we expect based on our knowledge of
partial melts of peridotite -- note, however,
that if magmas are melts of pyroxenites/eclogites,
these criteria are invalid

• multiple saturation at low P vs. high P
• trends on variation diagrams
• MgO contents
• Fe/Mg, NiO -- compatible elements are most useful
  here
• variation diagrams
• accumulation of crystalsphenocrysts vs.
  xenocrysts

most magmas are fractionated, relative to liquids
produced by partial melting
84
multiple saturation at high pressure
ol-in
opx-in
T
T
Liq FoEn
P20 kbar
P10 kbar
liq
FoLiq En
Fo liq
Qtz liq
P1 bar
En liq
Fo En
En Qtz
10
20
Fo
En
SiO2
P
weight
primary magmas must have olopx on their high-P
liquidus the P and T of multiple saturation give
the conditions of melting and the phase
compositions give the compositions of the
residual phases caveats what if magmas are
blends? what if there is a reaction relation?
85
are basaltic melts multiply saturated with ol
opx at high pressure?
MORB
a few basalts have olopx on their liquidi, but
most do not ? most cannot be primary assuming a
lherzolitic source note that a more stringent
test is to have olopxcpxgt or sp
simultaneously on the liquidus (can always get 2
phases on the liquidus 4 would be highly
coincidental)
86
what are the MgO contents of peridotite partial
melts?
results of melting experiments on peridotite
16 wt.
8 wt.
primary magmas produced at pressures gt10 kbar
(roughly the the base of the continental crust)
will have MgO contents of at least 10 wt.
picritic
87
MgO contents of MORBs
? most MORBs (and other basalts) are too low in
MgO to be in equilibrium with mantle peridotite
at the pressures thought to apply to mantle
melting
88
can basalts be in in Fe/Mg equilibrium with
mantle olivine?

• from studies of mantle xenoliths we know that
  mantle olivines are about Fo90
• we showed earlier that (Fe/Mg)ol/(Fe/Mg)liq0.30-0
  .33
• ? so liquids in equilibrium with mantle olivine
  will have Fe/(FeMg)0.25-0.27

? most basalts have Fe/Mg ratios too high to be
in equilibrium with mantle olivine
89
can trace elements provide any insights?
how about fractional crystallization?
nL amount of element in liquid (e.g., in
grams) x mass fraction of melt (e.g., in
grams) ? CL concentration in melt nL /x D
Cs/CL
90
what do these functions look like?
the key point is that compatible elements are
most sensitive to crystallization (whether
equilibrium or fractional), whereas incompatible
elements only increase significantly after
extensive crystallization
91
lets look at NiO, which we know is a compatible
element with respect to olivine (the major
crystallizing phase)
the NiO contents of MORBs are too low to be
primary melts in equilibrium with mantle
olivines, suggesting again that most erupted
magmas evolved from primary magmas by extensive
crystallization of olivine note that this gives
an indication of how difficult it would be to
distinguish equilibrium from fractional
crystallization
92
using trace elements to distinguish melting from
crystallization
93
what can we learn from the observed compositional
variations of magmas?
94
lets look at glasses from the high- and low-SiO2
magmas from Mauna Kea we looked at earlier
95
summary of the information content of variation
diagrams

• can clearly see the effects of crystallization
  differentiation and the liquid line of descent
• can identify crystallizing phases in the glass
  trends (can do same with whole rocks, but it is
  more difficult)
• especially useful to couple the chemical
  variations in the glasses with petrography
• can identify the parental or most primitive
  liquid compositions
• can explore what fractionation conditions (P, T,
  volatiles, etc.) are needed to match the observed
  liquid line of descent

96
could we estimate the composition of primary
magmas from these trends?

• identify the most primitive end of the trend
• if you can confidently conclude that olivine was
  the only phase crystallizing from such liquids
  generally impossible to rule out other phases
  were involved, but it is usually plausible, then
  the primary liquid can be reconstructed by
  incremental addition of equilibrium olivine
  (KD0.33) until the olivine is Fo89-91 (i.e.,
  corresponding to mantle olivine)

MgO 16 in reconstructed primitive Hawaiian
magmas
most primitive liquid (in equilibrium with Fo85
olivine)
ol plag
reconstructed primary liquid (in equilibrium with
Fo89 olivine)
Al2O3
ol
MgO
Fo85
87
89
97
variation diagrams for Hawaiian whole rocks show
a distinctive and different trend from the glasses

• note the linear trends from 7 MgO to 23 MgO
• based on what we have presented so far, we might
  conclude that the 23 MgO end of the trend
  represents the most primitive liquid
• but this end of the trend is more MgO-rich than
  the reconstructed primary liquid based on the
  glasses
• what is going on?

98
can we distinguish the following possibilities?
evolved liquid
evolved liquid
crystallization of olivine
Al2O3
evolved liquid
accumulation of olivine
primitive liquid
crystallization of olivine
cumulate
accumulation of olivine
primitive liquid
cumulate
MgO
99
Hawaiian lavas are typically rich in olivine, so
it is plausible that they have variably
accumulated olivines, but it is also possible
that the olivines crystallized from a liquid
corresponding to the whole rock, but failed to
separate from the magma
100
linear vs. curved arrays could help in principle
evolved liquid
crystallization of olivine
evolved liquid
Al2O3
primitive liquid
accumulation of olivine
cumulate
MgO
in principle, the curvature of the liquid line of
descent vs. the linearity of a trend due to
addition of olivine could be used to distinguish
these possibilities, but in practice, as we
showed previously for incompatible trace elements
(Al2O3 is an incompatible element during olivine
crystallization), the curvature is too small to
be useful
101
this approach can work for compatible elements
like NiO
the NiO vs. MgO trends are linear, demonstrating
the importance of olivine accumulation
102
FeO is also a compatible element, so trend in
FeO-MgO space are also useful in this respect
fractional crystallization trend

• the trend in FeO-MgO (as with NiO-MgO) argues for
  mixing rather than fractionation

103
trends in Fe/Mg ratio variations are also useful
for distinguishing liquids from cumulates
lets use the Fo-Fa system as a model to see how
this works

• suppose we have a whole rock with composition A
• if the rock represents a liquid, then the most
  Mg-rich olivine in the rock cannot be more
  Fo-rich than B
• if the rock was formed by accumulation of olivine
  D into liquid D, the the most magnesian olivines
  in the rock (D) would be more Fe-rich than the
  olivine (B) on the liquidus of the whole rock
• note that if the cores of olivines reequilibrate
  on cooling, this could also account for the most
  magnesian olivine being more Fe-rich than the
  calculated liquidus olivine

A
C
B
D
104
what does this approach tell us when applied to
Hawaiian lavas?
??
cumulates

• according to this test, the most magnesian Mauna
  Kea lavas (gt12 MgO) are all consistent with
  olivine accumulation
• the few lavas with olivines more magnesian than
  would crystallize from a whole rock liquid could
  represent entrainment of earlier crystallized
  olivines (xenocrysts rather than phenocrysts)
  or partial retention of earlier formed olivines

105
where is the estimated primary magma for
Hawaiaan tholeiites relative to the trend of
whole rock compositions?
cumulates fractionated liquids
cumulates
primary magma
106
multiple saturation at low pressuresimple phase
equilibrium insights

• primitive liquids inexorably evolve at low
  pressure from crystallizing one phase, to
  crystallizing two phases, and all wind up in this
  simple system at the eutectic, crystallizing
  olplagcpx

107
experimental studies of MORBs show that most are
multiply saturated at 1 atm

• these results for MORBs are typical there is not
  a large crystallization interval of olivine only
  as would be expected for mantle-derived melts,
  and olplagcpx all begin to crystallize within a
  few 10s of degrees

108
pseudo-liquidus diagrams confirm that most
MORBs are near 1 atm phase saturation boundaries
(cotectics) and thus have been fractionated at
low P

• note that this approach to modeling
  crystallization sequences of complex natural
  liquids is only approximate, but it leads
  nevertheless to the correct result in this case

109
a similar example from a single volcanic province
-- the Cascades
110
how can we distinguish equilibrium from
fractional crystallization? phase equilibrium
insights
P1 bar
T(C)
liq
Fo liq
Qtz liq
En liq
Fo En
En Qtz
Fo
En
SiO2
weight
fractional crystallization generates a larger
range of residual liquid compositions and a
larger temperature range than equilibrium
crystallization note that all liquids wind up at
the eutectic
111
a compositional range is also generated by
crystallization for a binary loop
fractional crystallization generates a larger
range of residual liquid and solid compositions
and a larger temperature range than equilibrium
crystallization note that all liquids wind up at
the low temperature end member
112
the same insights follow from ternary systems
equilibrium crystallization generates a narrower
range of liquids over a narrower temperature
range for fractional crystallization, all
liquids wind up at the eutectic, corresponding to
rhyolitic model liquids
113
can trace elements help to distinguish
equilibrium and fractional crystallization?
the only chance would be with compatible
elements, as incompatible elements only build up
differences after large amounts of
crystallization realistically, it is very
difficult to do, and it is unclear how realistic
equilibrium crystallization can be
114
in principle, NiO is an example of a compatible
element where one might think it possible to
distinguish equilibrium and fractional
crystallization
but I suspect in practice it would be nearly
impossible
115
magma bodies (magma chambers, dikes, sills,
flows, etc)

• these are the environments in which
  crystallization differentiation occurs

116
magma chamber processesthe simplest view --
cooling through the roof

• magma chambers are likely to convect
• they will crystallize from the bottom up

117
magma chamber processesloss of heat through the
roof can melt the country rock

• if the magma is not superheated, the heat needed
  to melt the country rock is released by the
  crystallization of the magma -- i.e.,
  crystallization and melting are coupled
• the partially or totally molten country rock is
  likely to be SiO2-rich, so it will be
  intrinsically liqht, and therefore will stay at
  the top of the chamber
• an interface will develop between the more basic
  magma below and the felsic magma above that will
  be hard to break down (density contrast,
  viscosity contrast), but diffusive mixing will
  mute the contrast

118
graphical thermodynamic analysis of thermal
interaction between melt and country rock
P
magma at its liquidus (TgtT0)
country rock at its solidus (T0)
the independent variables in the zone in which
the magma and country rock interact are H and P
(S is maximize at equilbrium) heat flows from
the magma to the country rock, with the country
rock melting and the magma crystallizing such
that the enthalpy is unchanged (mcrDHf,crmmagmaDH
f,magma)
119
magma chamber processes
but heat is lost through all the chamber walls,
and crystallization at the magma chamber
boundaries has predictable consequences for
convection and the development of stratification
of magma chambers
120
the consequences of crystallization at magma
chamber wallsdevelopment of a zoned magma
chamber
121
zoned magma chambers can result from injection of
fresh magma
overturn could lead to eruption of complex, mixed
magma (see later discussion of magma mixing)
122
injection of fresh magma can also lead to
eruption of the overlying magma
A
B
?
B
AB
mixed zone
A

• suppose the overlying magma, A, is initially
  vapor undersaturated, heating by injection of the
  hotter magma B can drive A into the liquidvapor
  field, leading to exsolution of vapor, eruption
  of A, entraining B as well, which degasses on
  depressurization
• this leads to a characteristically chemically
  zoned pyroclastic deposit, with B ? BA ? A

A
B
123
more complex convection patterns can develop in a
zoned magma chamber because thermal diffusivity
is higher than chemical diffusivity and the
complex effects of temperature and chemistry on
density (double diffusive convection)
124
accumulation of crystals at the boundaries of
magma chambers igneous sediments ? cumulates
125
lets look a little closer at crystal growth and
processes at the boundary of a magma chamber
harrisite

• known as crescumulates with a distinctive
  texture, these are thought to represent crystal
  growth into supersaturated liquid at at the
  cooling boundaries of magma bodies

126
more crescumulates
127
comb layering -- a related feature
128
how do these features form?

• growth on nuclei among crystals on the wall/floor
  into supersaturated melt that has descended from
  the top of the chamber

129
crystal settling vs. growth in place at the
boundary of a magma chamber
crystal settling
what happened here?
note the evidence of replenishment based on
repeated cycles
130
evolution of cumulates in a thermal gradient --
the origin of compositional gradients in cumulate
layers

• a mechanism for generating compositional
  gradients in cumulates
• a mechanism for removing melt from cumulates
• note that this can be complicated by variations
  in diffusivities of individual components

131
experimental confirmation of this process
132
chemical separation in a thermal gradient the
Soret effect

• if you put a homogeneous, multicomponent liquid
  in a temperature gradient, it will develop a
  compositional gradient
• although considered a possible mechanism for
  differentiation in the early 20th century, it was
  dismissed by Bowen as irrelevant, but it has been
  recently revived as a possible factor
• experiments have shown that temperature gradients
  of several hundred degrees over several mm can
  produce significant fractionation

133
is the Soret effect ever important?

• could the temperature gradients across convecting
  layers pump chemical components across the
  boundaries because of the Soret effect? it has
  been suggested that elevated opx contents of some
  gabbros in ophiolites are related to this
  phenomenon

134
what if the Soret effect operates at magma
chamber boundaries?could it influence the
evolution of chemical gradients in cumulates?
135
assimilationthermal and chemical interaction
between magma and country rock or xenoliths --
the eutectic case

• the proper variables for evaluating this process
  are H, S, and X, where H and X are fixed by the
  temperatures of the liquid and B and S in
  maximized at equilibrium
• it is clear from the H-X diagram that McBirneys
  conclusion based on the T-X diagram need not be
  valid, and for the schematic H-X diagram shown,
  it is not correct unless the temperature of B is
  less than the eutectic

T1
liq
H
T2
Bliq
Aliq
ABliq
AB
X
136
assimilationthermal and chemical interaction
between magma and country rock or xenoliths --
the case of a thermal divide
T1
liq
H
ABliq
Bliq
ABliq
T2
ABBliq
Aliq
AABliq

• as in the previous example, the actual trends
  depend in detail on the H-X diagram and cannot be
  determined from the T-X diagram

ABB
AAB
X
137
assimilationthermal and chemical interaction
between magma and country rock or xenoliths --
the case of a continuous solid solution
H
T1
T2

• the key point is that a variety of behaviors are
  possible, but they are easily evaluated with the
  H-X diagrams, but only with difficulty if at all
  using a T-X diagram

T3
X
138
magma mixing

• we have seen already that magma mixing is a major
  factor in the compositional evolution of basaltic
  magmas (i.e., most/all are blends of liquids
  produced over a range of pressure) either in the
  mantle or the crust
• the usual emphasis, however, is on mixing of
  basic and felsic magmas, which largely occurs in
  crustal environments

• we will review here
• physical evidence (phenocryst assemblages,
  zoning, blobs/inclusions)
• textural evidence (zoning patterns, bimodal
  phenocryst compositions, resorption, etc.)
• linear trends on variation diagrams vs. curved
  fractionation trends usefulness of compatible
  elements
• melt inclusions
• relation to magma chamber processes, eruptions,
  time scales

139
magma mixinghand-specimen-scale evidence
140
magma mixingthin-specimen-scale evidence
141

• we developed previously the following analysis of
  how textures could relate to mixing

• mixing of magmas to produce the magma will
  produce a superheated magma, dissolving
  phenocrysts and other nuclei
• mixing of magmas to produce the magma, will
  produce a subliquidus magma, preserving
  phenocrysts and other nuclei

142
this was the explanation offered for the
contrasting textures of the mid-ocean ridge
basalts in the two right-hand panels
143
magma mixing -- chemical trends

• this observation was key in identifying the
  importance of magma mixing in island arc
  environments

144
magma mixing -- phase equilibrium complications

• although the mixed magmas define a linear trend,
  their fractionation products will usually deviate
  from the this trend

145
magma mixing -- magma chamber dynamics

• there are many other possible complexities and
  relationships depending on the relative densities
  of the preexisiting and intruding magmas, their
  fractionation trends, and the role of degassing
• for example, if the magma near the top of the
  chamber degasses, it will become significantly
  denser (and often crystallize because of the
  increase in the liquidus on loss of water), which
  will result in its sinking into the chamber,
  inducing mixing

146
magma mixing - physical processes in the crust
mingled hornblende gabbro and biotite-hornblende
tonalite-granodiorite from the eastern ring
complex, Sierra Nevada
147
magma mixing related to magma emplacement
magma stoping
zone refining is an industrial process it is
sometimes invoked in mantle melting, but it is
unlikely to be important there, although it could
work in crustal environments as described here
148
liquid immiscibility -- an important chemical
phenomenon
GXXgt0
liquid
GXX0 GXXX0
critical point
spinodal
L2
L1
L1 L2
GXX0 spinodes
GXXlt0
L2
L1
L1 L2
GXXgt0
X
149
liquid immiscibility -- is it an important
petrological phenomenon?

• it unquestionably occurs in late-stage
  interstitial liquids in magmas, but does it occur
  on a larger scale, contributing to the diversity
  of igneous rocks?
• are ocelli like those on the right immiscible
  liquids, or mixed magmas that mixed then cooled
  too rapidly for homogenization (remember the
  evidence of mixing we have seen, where it is
  definitely not immiscibility)
• remember the eruptive carbonatites we looked at
  earlier -- plausible that carbonate liquids and
  alkaline silicate liquids are immiscible, and
  this is how carbonatites form in the crust, but
  it is unclear whether this is important for major
  rock types

150
liquid-vapor fractionation at high pressure

• at high pressures and temperature, water-rich
  vapor (if you can even call it vapor) can
  dissolve significant quantities of silicate
  components (especially alkalies, silica, alumina)
• this is especially true for alkalic magmas
• could vapor transport (e.g., as bubbles) move
  these components from one magma to another?
• this explains why quartz and feldspathic veins
  are so common
• it has also been proposed in alkalic magmas as a
  mechanism for enrichment of alkalis and other
  components
• mobility of such fluids plays a role in producing
  heterogeneity of the mantle sources of basalts
  (mantle metasomatism)
• mobility of such fluids play a critical role in
  transporting chemical components from the
  subducting slab to the overlying mantle,
  contributing to the distinctive chemical
  characteristics of island arc magmas

T
liquid/vapor
liquid
QtzL
vapor
LV
QtzV
SiO2
H2O
151
liquid-vapor fractionation at low pressure

• CO2 is strongly fractionated into the vapor at
  low P, so when a magma depressurizes, the vapor
  has a high CO2/H2O ratio and the CO2/H2O ratio of
  the residual melt goes down dramatically
  decreasing CO2 with roughly constant H2O in
  magmas is a clear indication of progressive
  degassing at all but the lowest pressures
• other components (e.g., He, Ne, S, Cl, etc.) can
  also be strongly fractionated by degassing