JOURNAL OF PETROLOGY
VOLUME 55
NUMBER 12
PAGES 2403^2418
2014
doi:10.1093/petrology/egu061
Plagioclase Fractionation inTroctolitic Magma
S. A. MORSE*
DEPARTMENT OF GEOSCIENCES, UNIVERSITY OF MASSACHUSETTS, 611 NORTH PLEASANT STREET, AMHERST, MA
01003-9297, USA
RECEIVED MARCH 29, 2014; ACCEPTED OCTOBER 14, 2014
ADVANCE ACCESS PUBLICATION NOVEMBER 13, 2014
Troctolitic intrusive rocks poor in augite are common in certain
Proterozoic anorthosite complexes and related rocks. The Lower Zone
of the Kiglapait intrusion, Labrador, consists of 1570 km3 troctolite
today and possibly 2900 km3 before erosion. In this augite-free
Lower Zone, plagioclase fractionation is as low as 16% An km^1 of
cumulate thickness and averages 3% An km^1. When augite crystallizes after 84% of the magma has crystallized, the fractionation becomes as much as 17% An km^1 of cumulate. Why such a difference?
It is clear from first principles of phase equilibria that fractionation
accelerates with saturation in augite, but not so clear that the difference
should be so great.The answer is to be found in the silica-poor nature
of troctolitic magma that is critically undersaturated in silica. This
low-silica effect reduces the activity of the NaSi albite component relative to the CaAl anorthite component in the plagioclase, thereby favoring the An component of the liquid and crystals and weakening the
fractionation process. As the normative augite component in the
magma rises from the base of the Lower Zone to the base of the
Upper Zone, the activity of silica also rises slightly and its consequent
effect on plagioclase composition tends to diminish. Liquid fractionation paths derived from observed crystal paths, when plotted in the
system Diopside^Anorthite^Albite, rise across the liquidus fractionation lines toward diopside and reach augite saturation near the 1atm
cotectic. They produce plagioclase compositions 10 mol % higher in
An than pure liquidus fractionation lines predict. The key criterion
for the troctolitic fractionation of plagioclase composition is the absence
of Ca-poor pyroxene from the rocks. Noritic magmas, by contrast,
have higher activities of silica and more effective fractionation
of plagioclase. A parallel fractionation of olivine is also retarded in
the Lower Zone by the accumulation of ferric iron in the liquid until
augite and titanomagnetite crystallize in the Upper Zone.
*Corresponding author. Fax: 413-545-1200. E-mail:
[email protected]
KEY WORD: plagioclase fractionation; troctolite melts; silica activity;
layered intrusions; Kiglapait intrusion
I N T RO D U C T I O N
Troctolite is not an uncommon rock, depending on where
you look. Troctolite of age 43 Ga is one of the oldest rocks
on the Moon (Nyquist et al., 2012), and the freshest troctolite sample in captivity. Troctolite of age 13 Ga is the
major component of the Kiglapait intrusion in Labrador
(Morse, 1969). Other occurrences of troctolite abound in
the Nain Plutonic Suite (Ryan, 1990; Morse, 2006), in the
Sept Iles intrusion (Namur et al., 2010), in the Duluth
Complex of Minnesota (Miller & Ripley, 1996) and at the
base of the Skaergaard intrusion, East Greenland (Wager
& Brown, 1967), among many others. Despite the moderate
abundance of this simple rock composed mainly of plagioclase and olivine, the notion of troctolitic liquids has been
occult in the imagination of many petrologists, for whom
oceanic basalt is a reference point. However, there is
much evidence in the field for the existence of troctolite
melts, especially in the snowflake troctolite of the
Hettasch intrusion (Berg, 1980) in which a very hot melatroctolite magma invaded a troctolite magma and was
quenched to a remarkable variety of plagioclase textures
derived from supercooling. Granitic magmas engulfing
troctolite magma to form giant troctolite pillows in the
Newark Island intrusion were vividly described by
Wiebe (1988).
ß The Author 2014. Published by Oxford University Press. All
rights reserved. For Permissions, please e-mail: journals.permissions@
oup.com
JOURNAL OF PETROLOGY
VOLUME 55
20
NUMBER 12
DECEMBER 2014
10
Sny
der
Falls B
roo Gro
up
kG
rou
p
30
IBZ
40
PCS
Sally Lake
SL
84
MOB
20
10
15
30 40 PCS
20
DB
15
CP
plunge
Kiglapait Sampling
Traverses
Patsy
Bay
Port Manvers
SL Sally Lake
DB David-Billy
CP Caplin-Patsy
40
30 PCS
20
10
Slambang Bay
Fig. 1. Sketch map of the Kiglapait intrusion with sampling traverses. The traverse maps have been shown by Morse (1979b). The contours are
those of volume per cent solidified (PCS) and are based on the strike of layering and the volumes of 13 cross-sections described in the original
GSA Memoir (Morse, 1969). Plunge values of layering are shown in the western part of the synclinal axis; these show shallowing plunges that
limit the probability that the steepness of layering has increased during subsidence: the layering seen must have been within 15 of the present
dip. MOB, Main Ore Band at 935 PCS.
The matter of troctolitic magmas comes particularly
into focus in the Kiglapait intrusion (KI), where an estimated 84% of the intrusion volume is troctolite of the
Lower Zone (Fig. 1). This large body (initially43500 km3;
Morse, 1969) is notable for its very small variation in
plagioclase composition in the Lower Zone (Fig. 2), raising
the question of multiple injections of fresh magma or
some other effect of slow compositional evolution.
2404
MORSE
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
PCS
UBZ
f
99.9
0.01
99
e
94
Antiperthite+
d Ap+
M.O.B.
c
Po+
b Mt+ (88.6)
a
+
90
Fo
Aug
84
An
0.1
Mesoperthite+
UZ
LAYERED SERIES
0.001
Fo in O
FL
An in
Plag
livine
99.99
99.985
LZ
IBZ
70
60
0
50
Plag+, Ol+ at 0
40
30
20
Fo, An (Mol %)
10
0
Fig. 2. Stratigraphic section of the Kiglapait intrusion, showing olivine and plagioclase composition trends superimposed; the crossover near
955 PCS should be noted. The vertical scale is logarithmic to show fine-scale changes in lithology over the last 300 m of stratigraphy (above
9987 PCS). The subzones of the Upper Zone are indicated by letters a^f based on arrivals on the liquidus of augite (Augþ), titanomagnetite
(Mtþ), sulfide globules (Poþ), apatite (Apþ), antiperthite, and mesoperthite. MOB, Main Ore Band at 935 PCS. FL, fraction of liquid. The
dashed excursion near 90 PCS represents the elevated Fo values interpreted as ‘oxygen spikes’ owing to the increasing components of Fe^Ti
oxide minerals present in the system (Morse 1979b, 1980).
2405
JOURNAL OF PETROLOGY
VOLUME 55
S TAT E M E N T O F T H E P RO B L E M
A single, prolonged episode of magma chamber filling is
assumed. Over the interval 0^40 PCS (volume per cent
solidified) the model liquidus plagioclase composition in
the basal Lower Zone of the Kiglapait intrusion varies
only from An 67 to 64, a matter of 3 mol % over a stratigraphic thickness of 1900 m and 40% by volume of the
entire magmatic history (bold line in Fig. 3). Over
the entire Lower Zone, a cumulate thickness of 5 km,
the variation is 29% An per km of cumulate deposited.
In contrast, the (pre-augite) Skaergaard Lower Zone fractionates plagioclase at a rate of 136 mol % An km^1
(McBirney, 1996). The rational mind queries, what is
going on here? The evolution of the Kiglapait liquid is
evidently retarded. Is the cause exotic or intrinsic?
Among exotic causes might be the injection of new
magma, or even better, a leading-edge fractionated
magma that could cause the scatter to low An values.
Such periodic reversals are well known and distinct on a
small scale in many other troctolitic intrusions, particularly at Rum (Emeleus et al., 1996). The only small-scale
study at Kiglapait so far found four paired reversals and
returns in a 65 cm section, but the variations are not large
70
Kiglapait Intrusion Plagioclase
I N T R I NSIC EFF ECTS
The effect of pressure
The inferred central magma depth of the intrusion is
8400 m (Morse, 1979b) at PCS ¼ 0. This depth decreases
outward toward the limbs of the basin-shaped structure
(Morse, 1969, 1979b) so that the ambient pressure at the cumulate interface also decreases. The following discussion
refers only to the central maximum pressure range from
51 to 28 kbar, giving only the maximum effects of pressure
on plagioclase composition. Crystals of plagioclase at
high pressure are closer in composition to their parent liquids than at lower pressures. This relationship is quantified in the direct variation of the exchange coefficient KD
with pressure, as
KD ¼ 0:0516P þ 0:2622
Mol% An
60
50
40
0
0.2
DECEMBER 2014
(51% An; Morse, 1969, p. 127). Because the lens-like layering in the intrusion generally extends for only a few hundreds of meters (Young, 1983), it is inferred that the
section described is a local phenomenon, and intrinsic.
In the absence of more small- to medium-scale traverses,
the case for significant new refreshments of magma above
10 or 20 PCS cannot be made. Given this stand-off, it is
worthwhile to consider all possible intrinsic effects that
could cause retardation of the mineral compositional evolution. If these can be shown to suffice, then the exotic
scenario becomes moot, and unnecessary.
Roof crystals?
Depletion mode
NUMBER 12
Binary
0.4
0.6
-log FL
0.8
1
Fig. 3. Kiglapait plagioclase compositions for 0^90 PCS. Each data
point represents a bulk analysis for the sample. Causes of scatter are
discussed in the text. Various paths can be modelled through or away
from the data, according to varying multiphase Rayleigh fractionation processes (Morse, 2008). The main path (bold continuous line)
is a model for the liquidus at variable depth and pressures, 51^
33 kbar, and hence a variable KD. The uppermost, dotted curve
would satisfy limiting cumulus compositions starting from An69. The
dashed line near the lower bound of the data represents another
Rayleigh fractionation path without the pressure effects; it is of no
concern in this study. Paths for the binary and depletion mode represent special fractionation effects not of interest here except to note
that they may satisfy many of the low values at stratigraphic levels
above 20 PCS (volume per cent solidified). Figure modified from fig.
13 of Morse (2008). The horizontal axis is the negative log of the fraction of liquid remaining; the PCS scale denotes the per cent of liquid
solidified.
ð1Þ
where P is in kbar (Morse, 2013, fig. 7). The systematics of
linear partitioning bearing on equation (1) are reviewed
in Supplementary Material Appendix 1 (supplementary
data are available for downloading at http://www.petrology.oxfordjournals.org). As KD increases, the binary
loop narrows to a limit of zero width at KD ¼10. Lower
values of KD describe fatter loops. The effects of compression in a deep magma chamber can be highly variable
and significant (Table 1). To first order a magma crystallizing plagioclase of composition An 67 at the roof, if transferred instantly and adiabatically to the floor, could be in
equilibrium with plagioclase An 61, a drop of 65 mol %
(Fig. 4). Because the crystal^liquid tielines are shorter at
depth, the magma fractionates more slowly at depth than
at higher levels and lower pressures. This is the intrinsic
pressure effect on fractionation; pressure damps the tendency for the liquid to evolve.
There are other features of interest in this context.
Crystals may grow over time within the magma and
encounter liquid motions leading to reverse, normal, and
oscillatory zoning (Maale, 1976) to cause pre-cumulus
zoning. This effect has been estimated to cause up to 4%
An variation in the An range in the Kiglapait intrusion
(Morse, 2012). If a low-pressure plagioclase crystal suddenly becomes taken down to the floor and survives, it
2406
MORSE
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
Table 1: Pressure effect on plagioclase and liquid
compositions*
Line PCS
P (kbar) KD
–log F(L) X An Xl X An Liq Difference
1
Roof 28
0409
n.a.
07
0484
0216
2
Roof 28
0409
n.a.
067
0451
0219
3
Roof 28
0409
n.a.
0502
0298
0204
4
86
36
0445
0854
0502
0310
0192
5
70
40
0468
0523
0579
0392
0187
6
35
46
04996 0187
0642
0474
0169
7
0
51
05254 0
0671
0519
0152
Effect of trapped liquid
The resident melt trapped in a cumulate presumably has
partition coefficients equal to 10 for all components,
hence it has no capacity to cause the parent magma to
evolve. The residual porosity has been calculated for
single rocks in the KI from the range of An content in
plagioclase (Morse, 2012). It ranges erratically from 35%
to zero in the range from zero to 40 PCS and can therefore
cause the local orthocumulates and mesocumulates to reduce their bulk plagioclase composition by 6 to 2 mol
% An in that range. This effect tends to damp the fractionation process by sequestration. Trapped liquid solidification
has been discussed by Morse (2013).
*Morse (2008, fig. 10).
n.a., not applicable.
ROOF: 9.6 km from surface
2.8
liquids
= 0.22
Floor
3.2 levels
@ PCS:
3.6
P, kbar
4.0
crystals
Magma
Depth,
m:
86
3000
70
4500
35
6770
Modal effect of olivine
4.4
4.8
= 0.154
0
5.2
0.2
0.3
0.4
0.5
X An
0.6
8400
0.7
will be richer in An than the local high-pressure crystals,
and this might be the cause of local high-An crystals as
suggested in Fig. 3, none of which exceeds the main trend
by more than 4% An.
The pressure difference decreases with stratigraphic
height as the cumulate surface rises to lower pressure.
It also decreases from the center of the chamber to the
margins, which is where we see and sample all the rocks
at the present erosion level. However, any pre-cumulus or
complexly zoned crystals may contribute to local variety
anywhere. Perhaps this is the effect of wandering crystals.
0.8
Fig. 4. Pressure effect on plagioclase^liquid tielines. The diagram
shows crystal^liquid tielines at the roof and at four selected pressures
corresponding to several PCS levels in the Lower Zone. The upward
trend in decreasing XAn of the crystal compositions ( symbol) is
taken from the plagioclase model, Fig. 1. The liquid compositions (ellipses) are calculated from the crystal compositions using the linear
partitioning equation and the pressure variation of KD taken as
KD ¼ 00525 P þ 02622 (Morse, 2013); P is in kbar. The point of the
diagram is to illustrate quantitatively the narrowing of the plagioclase
loop with increasing pressure. Liquids descending directly from the
roof would be in equilibrium with crystals 022 ^ 0154 ¼ 0065 XAn
units closer than at the roof. Pressure^depth relations are developed
from the modified PREM table of Stacey & Davis (2009). The emplacement pressure at the top of the UBZ is taken as 25 kbar based
on the phase relations of the contact aureole (Berg & Docka, 1983).
The systematics of multiphase Rayleigh fractionation reveals that a fractionation curve can become steepened
when a given phase component is scarce, having nothing
to do with chemistry (Morse, 2008). This effect causes
the ‘depletion mode’ curve shown in Fig. 3. That effect is
brought into play to some extent when the local cumulate
is richer in olivine than the Ol:Pl cotectic ratio of about
25:75 in oxygen units (Morse et al., 2004). This effect may
account for some rocks with low An in plagioclase over
the range 10^40 PCS, or even in the olivine-rich basal
Lower Zone at 510 PCS. Here the olivine compositions
and abundance tend to be high even where the plagioclase
compositions are low in An, a condition fostering depletion
in An owing to modal scarcity of plagioclase. Where both
Fo and An are low together, recharge with fractionated
magma may be inferred (there are five examples in the
range 0^20 PCS).
Magmatic characteristics
Experimental studies of the inferred parent magma composition, designed to mimic the evolution of the Lower
Zone to its eventual saturation with augite (Morse et al.,
2004), show that the path from very little augite component to saturation can indeed be very long, with about
83% of the primary melt crystallizing as troctolite. That
result is to some extent embedded in the experimental
design. Nevertheless, the estimated bulk composition of
the intrusion is very similar to that of the chilled margin
2407
JOURNAL OF PETROLOGY
VOLUME 55
of the Hettasch intrusion and to the Inner Border Zone
of the Kiglapait intrusion (Morse et al. 2004), so the low
initial concentration of augite (54% in the oxygen norm;
table 8 of the cited paper) is evident.
Experiments on the same inferred bulk composition at
high pressures (McIntosh, 2009) suggest that the troctolitic
Kiglapait magma originated from a spinel harzburgite
source near 15 kbar pressure and temperatures exceeding
1400 C (Morse, 2013). The decompression path for melts
to the site of emplacement at 3 kbar would require cooling by 200 C and might involve a leading-edge fractionation process in which the earliest pulses were somewhat
evolved. Later additions might also be evolved, contributing to some of the low-An samples found at 0^20 PCS.
A R E L E VA N T F O R M A L I S M
Apart from the issues of pressure and competition from
olivine discussed above, we seek a more fundamental
cause of weak evolution of plagioclase composition in
troctolites. Issues involving the augite component will not
work, as discussed below. Instead, an intrinsic property
of the system, the low activity of silica, will do the trick.
If the silica activity is low, the NaSi component, Ab, of
plagioclase, is rendered less effective and instead the role
of An is enhanced. These terms will be given more formal
development in the ensuing treatment.
Silica activity in magmas
From observation, the concept of the activity of silica
in magmas is not frequently encountered in the current
petrological literature. Its history may therefore usefully
be mentioned in a contribution such as this, in which the
burden of the argument revolves around silica activity.
The ‘critical plane of silica undersaturation’ was a description by Yoder & Tilley (1962) of the Albite^Orthoclase
(Ab^Or) join in Bowen’s (1937) ‘Petrogeny’s Residua
System’, the ternary system Nepheline^Leucite^Silica
(Schairer, 1950). Liquids below this line were deemed
‘critically undersaturated’ because the crystallization of
feldspar would drive the residual liquid to eventual saturation with a feldspathoid. Liquids above this Ab^Or line,
in contrast, would evolve toward saturation with quartz
or another form of pure silica. From these principles it is
clear that the feldspar join is a thermal maximum in the
residua system.
This nomenclature is conceptually useful in dividing
intrinsically undersaturated liquids from those that are
capable of becoming saturated with a silica phase. But it
is only a generalized concept with some attached ambiguity. In what dimension is Ab^Or a ‘plane’? (It is actually
a section through a Gaussian negative surface, a saddleshaped temperature^composition surface.) Exactly how
silica-undersaturated is this divide? And how shall we
numerically express its relationship to actual silica
NUMBER 12
DECEMBER 2014
saturation? The answer lies in the quantitative calculation
of silica activity, most notably brought into the light of a
magmatic discussion by Carmichael et al. (1970) and then
made the foundation of all magmatic petrology by
Carmichael et al. (1974) in their classic and revolutionary
textbook on Igneous Petrology.
Despite its enduring utility in discussing magmas and
their evolution, the numerical value of the silica activity
associated with different magma types is rather sensitive
to the thermodynamic parameters used in the calculations.
One result of this sensitivity was that in the actual parameters used by Carmichael et al. (1974), a stable equilibrium
was found to exist between forsterite (Fo) and quartz (Q),
an assemblage contrary to experience because it is known
to be unstable with respect to the orthopyroxene enstatite
(En). Accordingly, Morse (1979a) applied free energies
from the newer literature for En and Fo and used a new
value for the entropy of orthopyroxene based on the mineral^liquid assemblages found in Makaopuhi Lava Lake
in Hawaii, in order to write a new equation for the activity
of silica.
From this practical approach it was then possible also to
show that there exists an activity ratio for oxygen/silica,
thereby linking oxygen fugacity (itself an activity) and
silica activity in such a way that the one could be calculated from the other. In this result, tholeiitic magmas
plotted just above the fayalite^magnetite^quartz (FMQ)
buffer. Troctolitic magmas, however, were found to lie on
the wu«stite^magnetite (WM) buffer, substantially below
the FMQ buffer (Morse, 1980, fig. 10). They then fractionate to slightly less-reduced compositions until titanomagnetite forms, after which the oxygen fugacity runs at a
constant distance between FMQ (40%) and WM
(60%).
In modern terms, the activities of oxygen and silica in
magmas are most usefully calculated from multiple mafic
mineral compositions in the QUILF equilibria of
Andersen et al. (1993), assuming the presence of all relevant
mafic minerals. But it should be emphasized that calculations of phase equilibria can be usefully made even in
compositional^thermal spaces that are not physically saturated with all relevant phases: these are metastable equilibria. An example will be shown in the development that
follows.
Silica saturation in magmas
A chemical system is said to be saturated in silica if a
crystalline phase of silica is present at equilibrium. At relatively low temperature and low pressure, 1atm to
143 kbar, that phase is generally tridymite; at higher
temperatures it is cristobalite, and otherwise at plutonic
magmatic conditions high (b) quartz. Systems deficient
in silica may contain silica-poor phases such as nepheline
or leucite; in petrological parlance such systems are
considered ‘undersaturated’ in silica. Systems containing
2408
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
neither normative quartz nor a feldspathoidal signature
have been called ‘critically undersaturated’. Troctolitic
magmas fractionating to syenite are an example.
Tholeiitic magmas that fractionate to rhyolite or granite
may become saturated in silica.
Complex multicomponent systems can be defined by a
few pertinent reactions leading to a determination of intensive parameters such as the chemical potential, or more
specifically the activity, of silica and oxygen. The chemical
potential mi is defined as the partial molar free energy per
mole of a component i. The activity ai of a component i is
related to the mole fraction Xi of that component by
ai ¼ Xigi where gi is the activity coefficient, to be determined empirically. Ideal solutions are defined as those
for which gi ¼10. The activity is related to the chemical
potential by
mi ¼ mi þ RT ln ai
ð2Þ
where mi is the chemical potential at some standard state
(such as the pure component i at the temperature and pressure of interest), R is the ideal gas constant, and T is the
temperature in degrees Kelvin.
The presence of quartz or its polymorphs in equilibrium
with a silicate melt defines the activity of silica
aL SiO2 ¼ 10 in the system. A carbonatite magma may
have aL SiO2 ¼ 0, and a granite magma crystallizing quartz
will have aL SiO2 ¼10. All phases at equilibrium in such a
system will also have aSiO2 ¼10. The presence of magnesian olivine suggests aSiO2 510 but the presence of fayalite
permits aSiO2 ¼10. Therefore, the activity of silica in a
mafic melt will be expected to increase toward 10 as the
ferrous iron ratio increases. In the Skaergaard intrusion,
that tendency occurs when pigeonite ceases to be a cumulus phase at a coexisting olivine composition of Fo40.
Calculations then yield a value of aSiO2 ¼ 09 (Morse,
1994), or nine-tenths of the way to silica saturation.
Kiglapait petrography
The Kiglapait magma was critically undersaturated to the
extent that it never crystallized either quartz or a feldspathoid at the liquidus. Its final product was a high-temperature syenite with feldspars at the ternary minimum.
Most of the rocks contain normative hypersthene (Fig. 5)
but not actual orthopyroxene: this constituent is occult
in the augite component of the liquid. Many of the rocks
do contain normative nepheline (NE) and after the crystallization of major Fe^Ti oxides near 94 PCS a few
calculated liquids contain normative NE (Fig. 5). The normative NE is accommodated in feldspar and augite
components. Unlike tholeiitic magmas, the Kiglapait
liquid did not encounter an olivine hiatus, conventionally
ascribed to the oxidation accompanying the arrival of cumulus magnetite and leading to the breakdown of ironrich olivine to magnetite plus silica (Morse et al., 1980).
Instead, in the Kiglapait intrusion, olivine remained
Silica Saturation in the Kiglapait Intrusion
Summation Liquids
Rocks
10
Oxygen Norm, % HY
MORSE
LZ UZ
5
HY norm
0
NE norm
40
-5
0
80 90
1
PCS
- log FL
99
99.9
2
3
99.99
4
Fig. 5. Normative silica saturation in the Kiglapait intrusion, based
on whole-rock data from Morse (1981). Values 0 are percentages of
hypersthene (HY) in the oxygen norm of the rocks (open circles) or
summation liquids (filled circles) calculated by summation over all
the rock compositions. Values50 are percentages of nepheline (NE)
in the oxygen norm converted to negative values of HY. The diagram
refers to the petrographic classification of rocks and melts according
to their distance from the so-called critical plane of silica (under-)saturation where HY ¼ NE ¼ 0. This criterion is related to, but very
different from, the scale of silica activity in the thermodynamic
sense of mole fraction or chemical potential. Horizontal axis as in
Fig. 1. Summation liquids are models derived by summing over
the chemical compositions of all the analysed whole-rocks from top
down in the intrusion. Their compositions are reported in table 6 of
Morse (1981).
present at the liquidus but became locally more magnesian
near the Main Ore Band, leading to ‘oxygen spikes’ in
some rocks (Morse, 1979b). These spikes result from the
oxidation of the magma that decreases the activity of
ferrous iron, hence favoring the Mg end-member. The
absence of Ca-poor pyroxene in the Kiglapait cumulates
is diagnostic of the difference between troctolitic and
tholeiitic magma types, and is a major signal of low silica
activity.
That said, traces of Ca-poor pyroxene do occur locally
in troctolites as reaction rims on olivine, revealing the
minor presence of trapped liquid. They also occur in
small (millimeter- to centimeter-scale) pockets where
enough liquid has been trapped to form oxysymplectites
(e.g. Morse, 1969, plate 34; Morse, 1980). These strictly
local intergrowths (rarely more than one, and usually
zero, per thin section) consist of magnetite and hypersthene, partially invading crystals of olivine that have
reacted with trapped liquid. Although fundamentally
caused by the oxidation of olivine with release of silica to
form hypersthene, the bulk analyses of the phases also
show the introduction of Fe, Ti, Al, and Si as well as
oxygen, so the symplectites result from a late, local magmatic reaction above the solidus. The process eventually
runs out of oxygen and stops consuming olivine. The persistence of unaltered olivine exterior to the oxysymplectites
defines a low activity of silica and oxygen in the rock
generally.
2409
JOURNAL OF PETROLOGY
VOLUME 55
These intergrowths, local vestiges of trapped liquid, are
highly concentrated in the evolved components of magnetite, silica, and oxygen that are ordinarily encountered in
crystals only near 90 PCS. They do not represent the ambient magma of the Lower Zone. Nevertheless, they furnish
an opportunity to reveal the chemical reactions that are
pertinent to the issues of plagioclase fractionation encountered here. A silication reaction that leads to an understanding of silica activity is the EFS reaction:
MgSiO3 ¼ MgSi05 O2 þ 05SiO2 :
Enstatite
Forsterite
Fayalite
logK ¼ 2logaFe3 O4 þ 3logaSiO2 6logaFeSi05 O2 logfO2
ð7Þ
Silica
6FeSi05 O2 þ O2 ¼ 2Fe3 O4 þ 3SiO2 :
ð4Þ
Oxygen Magnetite Quartz
Reaction constants used for these equations were developed by Morse (1979a) and summarized by Morse
(1980), for a pressure of 4 kbar. Quartz was used as the
reference state for these calculations. Reactions and their
constants from Morse (1980) may be found in the
Supplementary Material to this paper.
The oxysymplectite reaction is represented by equation
(3) run backwards to make orthopyroxene (hypersthene)
and the redox reaction dissolves the Fa component of the
olivine to make magnetite plus quartz in equation (4) run
forward.
When the condensed phases coexist in equilibrium at P
and T they define the state of the system in terms of silica
and oxygen activities, and if liquid is present it also shares
those activities. Therefore the crystal equilibria also define
the crystal þ liquid state of equilibrium. The equations
can be used to evaluate the intensive parameters of the
magma at equilibrium with only olivine and plagioclase,
as happens in the entire Kiglapait Lower Zone from 0 to
84 PCS.
Application to liquidus equilibria
For the EFS reaction (3) we can write the equilibrium
constant K for a given temperature and pressure as
logK ¼ logaMgSi05 O2 þ 05logaSiO2 logaMgSiO3 :
ð5Þ
Rearranging and multiplying to eliminate the fraction,
we have
logaSiO2 ¼ 2logaMgSiO3 2logaMgSi05 O2 þ 2logK:
ð6Þ
Log K is equal to ^Gr/RT where Gr is the Gibbs free
energy change for the reaction. Empirical constants for
log K in the form A/T þ B þ C(P ^ 1)/T are given in
table 6 of Morse (1980), reproduced here in the
Supplementary Material. Because this reaction involves
the presence of orthopyroxene it furnishes only a
DECEMBER 2014
maximum estimate of the silica activity for the Kiglapait
troctolites, and that result must be adjusted downwards
from the FMQ results to estimate the actual activity of
silica.
Similar relations are also developed for the oxygen
activity called the fugacity, fO2 , in the Supplementary
Material. From this process the oxygen fugacity is obtained
using mineral compositions and temperatures as input.
For the FMQ reaction [equation (4)] we write
ð3Þ
A redox reaction that also locks the silica activity is the
‘FMQ’ reaction:
NUMBER 12
and when the appropriate constants are applied (see
Supplementary Material) we have a solution for the last
term, the log of the oxygen fugacity. When this was determined for the Kiglapait oxide minerals it was also possible,
using the mineral compositions, to isolate the activity of
silica from the above equation (Morse, 1980). This result
in turn required correction to a value 01 lower in the log
scale of silica activity relative to the EFS result.
Upper Zone rocks with Ti-magnetite furnish a direct
estimate of silica activity through the QUILF equilibria
(Andersen et al., 1993):
SiO2 þ Fe2 TiO4 ¼ Fe2 SiO4 þ FeTiO3 :
Quartz Ulvoº spinel Fayalite
ð8Þ
Ilmenite
Application to the Kiglapait Lower Zone
To calculate silica activity in the natural rocks we require
a description of the olivine compositional trend (Fig. 6).
The data are scattered for many reasons, including probable addition of fresh, fractionated magma in the region
0^20 PCS and the abundant crystallization of plagioclase
in felsic layers (‘plag effect’) that drives the olivine composition to lower values simply because it is scarce
(Morse, 2008). A ‘trapped liquid path’ is shown as a dotted
line for the Lower Zone, and this is calculated from
observed residual porosity values (Morse, 2012) that cause
the Fo values to decrease by reaction with trapped liquid.
The upper path (continuous line) in this figure is drawn
to represent a plausible trend of liquidus compositions.
It is this path that will yield the silica activity in the main
body of the magma (see Supplementary Material for the
equation).
A labor-saving device was constructed (Morse, 1979a) by
contouring the compositions of liquidus olivine against
the Kelvin temperature and silica activity referred to
quartz at 5 kbar. A portion of this T^X^activity space is
shown here as Fig. 7. The contours are linear on T(K) and
derived from the EFS reaction. Using an algorithm
relating nominal orthopyroxene compositions to olivine
compositions, XEn 085XFo þ 015, the observed olivine
compositions are translated directly to the silica activities
2410
MORSE
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
80
1000
1100
1300 oC
1200
Kiglapait Olivine 0-90 PCS
75
Liquidus model
0.7
OPX Sat
65
Plag
effect
55
Trapped
liquid path
0-20
Basal LZ
+ magma?
50
UZ trend
Aug+
45
40
0.0
0.1
0.2
0.3
0.4
a SiO
0.2
0.5
0.6
0.7
0.8
FMQ Corr.
SiO2
60
-log a
Mol % Fo
70
0.9
1.0
OL Sat
87.5
PCS 60
of the inferred pyroxene compositions that would coexist
with the observed olivines. Temperatures were taken from
the 5 kbar experimental study of Morse et al. (2004) covering the entire Lower Zone to saturation with augite at
84 PCS. The results are shown in Fig. 7 as open circles
from 0 to 90 PCS.
From the study of Fe^Ti oxide minerals (Morse, 1980) it
was found that at 935 PCS the silica activity obtained
from the FMQ equilibria was lower than that for OPX saturation by 01 units in the log scale. The plotted array of
points was therefore lowered by that amount to give the
ambient silica activity for the LZ troctolites. The result is
shown by the lower bold line in Fig. 7, with labels showing
points of interest in the PCS scale. The evolution of silica
activity is flat to 60 PCS and then climbs by a value of
004 log units to augite saturation at 84 PCS. It then
declines modestly to 875 PCS within the Upper Zone, at
which point the modal augite proportion is about 30 vol.
%, greater than its cotectic proportion (Morse, 1979c).
This metastable excursion of the magma into the augite
field reaches its maximum augite production of 425% at
90 PCS.
Although the graphical method of evaluating this record
of silica activity is based on old and probably outmoded
reaction constants, the general nature of the result is adequate to show an interesting change in the trend, peaking
at 84 PCS. This inflection can only result from the
0.6
84 Aug+
90
- log FL
Fig. 6. Kiglapait olivine compositions in the range 0^90 PCS. The
continuous curve for the liquidus model is given by Fo
% ¼ ^2403X2 ^ 1361X þ 7423 (where X ¼ ^log FL) and is nominally
considered extended by connection to an internal reservoir (Morse,
2008). Alternatively, it may simply be extended by the low activity of
Fe2þSi in the melt. The dotted path represents a correction for
trapped liquid as determined from the An range in plagioclase
(Morse, 2012), not used here. The dashed curve labeled ‘plag effect’
shows the Rayleigh fractionation path for cotectic olivine plus
plagioclase (Morse, 2008; Fig. 4), which may relate to some of the low
Fo values observed. In the olivine-rich region 0^20 PCS some low
Fo values may result from addition of more fractionated magma.
The ‘UZ trend’ indicates more rapid fractionation after the magma
became saturated with augite, which is relatively Mg-rich.
2
KI
0
0.3
0.8
0.5
103/T, K
0.7
0.6
Kiglapait Silica Activity (5 kbar, 0-90 PCS)
Fig. 7. Activity of silica as a function of temperature and olivine composition, taken from the EFS reaction and fig. 1b of Morse (1979a).
The axes are log^linear in silica activity and Kelvin temperature.
The interior space is contoured for olivine composition in nominal
equilibrium with Ca-poor pyroxene (‘OPX Sat’) at 5 kbar, quartz
basis. Data are plotted from experimental temperatures (Morse et al.,
2004) applied to the liquidus curve in Fig. 3, to give the upper array
of circles. This array is then translated downward to the Opx-absent
condition of the Kiglapait Lower Zone using a correction from the
FMQ equilibrium (Morse, 1980), to find the nominal silica activity
for this region of the intrusion. The lower array is labeled with PCS
values that show a flat curve for the early LZ to 60 PCS, then a rise
to 84 PCS, where augite arrives at the base of the Upper Zone, followed by a decreasing trend to 875 PCS (augite mode increasing)
and a final up-tick to 90 PCS where the augite mode reaches its maximum value of 42%. The range of inferred silica activity is impressively small, and the wealth of interpretable detail impressively large.
steepening (and tightening) of the olivine compositional
array toward a lower Fo value that in itself responds to
the new coexistence with augite (Fig. 6).
Stratigraphic results
To put this result into context, the entire history of silica
activity in the intrusion is shown in Fig. 8, modified from
fig. 16 of Morse & Ross (2004). The data for the Lower
Zone and early Upper Zone are taken from Fig. 7 where
they are scaled downward from the EFS reaction. Here
the trend is reversed along with the temperature scale.
It is connected at 90 PCS to the array determined for
the Upper Zone. The LZ^UZ boundary is marked with a
vertical dashed line labeled ‘AUGþ’.
In the UZ array the original estimates from the FMQ
reaction (Morse, 1980) are shown as open rectangles.
New estimates from QUILF, kindly provided by D. H.
Lindsley (personal communication, 2004), are based on
a significantly larger assemblage of equilibria among
fictive quartz, ulv€ospinel, ilmenite, liquid, and fayalite
2411
JOURNAL OF PETROLOGY
Kiglapait silica activity
0.00
VOLUME 55
1.0
most likely
0.9
MLL
0.05
-log aSiO2
2004 QUILF
0.10
Skd
1980 OXIDES
max
0.8
aSiO
2
0.15
0.7
NORTH
0.20
AUG+
0.6
LZ UZ
0.25
less likely
MOB
0.30
-log FL 0
PCS 0
1
90
0.5
2
99
3
99.9
4
99.99
Fig. 8. Activity of silica in the Kiglapait intrusion. The data for the
Lower Zone and early Upper Zone are taken from Fig. 7 where they
are calculated from the EFS reaction. Here the trend is reversed
along with the temperature scale. Data shown in filled rectangles for
the Upper Zone refer to coexisting pairs of augite and olivine at high
temperature, from the QUILF equilibrium of Andersen et al. (1993)
calculated by D. H. Lindsley (personal communication, 2004). The
original calculations by Morse (1980) for the Upper Zone beyond the
Main Ore Band (MOB) gave the open rectangles and lower, dotted
line, but the subsequent calculations from QUILF (black rectangles)
supersede the earlier results. The sharp increase in silica activity
beyond the Main Ore Band arises from the sudden deposition of Fe^
Ti oxide minerals, mostly titanomagnetite, and the resulting silication
of the main magma body. Figure modified from fig. 16 of Morse &
Ross (2004). The polygon labeled ‘MLL’ shows the silica activity for
the tholeiitic Makaopuhi lava lake (Wright & Weiblen, 1967) at an
olivine composition of Fo49, corresponding to 91 PCS in the
Kiglapait intrusion. The MLL liquid is reported to have the phases
Ol, Opx, and Mt at 1293 C. The dotted line at 0115 on the y-axis
labeled ‘Skd’ is a recalculation for the base of the Skaergaard intrusion using T ¼1168 C and olivine Fo72 (Thy et al., 2009, 2013) and the
EFS fig. 1a of Morse (1979a) for conditions at 1atm, tridymite base,
and pigeonite at the liquidus. This is a maximum value because
pigeonite is not an early liquidus phase; accordingly, a small arrow
suggests a range toward a lower value.
NUMBER 12
DECEMBER 2014
minerals, mostly titanomagnetite, and the resulting silication of the main magma body.
Figure 8 also shows a pentagon symbol at 005 on the log
scale (09 on the activity scale) labeled MLL for
Makaopuhi Lava Lake, a natural experiment analyzed by
Morse (1979a) from several data sources. The x-axis location of this point is taken from the olivine composition
(Fo49) applied to the Kiglapait PCS scale. This sample is
quartz-tholeiitic, and is of high silica activity for its olivine
composition; it lies on the FMQ buffer. The sample of
melt with olivine, orthopyroxene, and magnetite was
taken at 1293 C, very hot!
Also shown in Fig. 8 is a horizontal dashed line (‘Skd’)
near 015 in the log scale, representing the calculated initial
silica activity for the Skaergaard intrusion (Morse et al.,
1980). The magma represented here is an olivine tholeiite
with an olivine hiatus from Fo53 to Fo40 with pigeonite
as the low-Ca pyroxene. In this intrusion the mean plagioclase composition varies as 16% An km^1, not inhibited
by low silica activity. The Kiglapait Lower Zone was,
in sharp contrast, at a much lower silica activity, 027 on
the log scale.
A P P L I C AT I O N T O P L A G I O C L A S E
F R AC T I O N AT I O N
If a system of melt and crystals is sufficiently rich in silica
the activity of the NaAlSi3O8 albite component is maximum and, all other things being equal, crystallization
will proceed with maximal enrichment of Ab and depletion of An. If, however, the silica activity is low, the activity of albite itself may become low as in the following
reaction:
NaAlSi3 O8 2SiO2 ¼ NaAlSiO4
(Andersen et al. 1993), and are shown as filled rectangles.
Results from Morse (1980) for the early Upper Zone show
that the north sector of the intrusion, characterized by the
highest modal Fe^Ti oxide contents and culminating in
the Main Ore Band (MOB), has a generally higher range
of silica activity (dashed line with ‘max’ and ‘North’
labels) than the southern sampling traverses. This region
includes one unique sample (KI 4032) from the northern
sector of the intrusion that contains orthopyroxene without
olivine, defining a maximum activity of silica at that
location.
Near the Main Ore Band the mineralogy is dominated
by the peak production of augite and depletion of olivine
in the rocks (Morse, 1979c). The Main Ore Band itself
shows a wide range of silica activity, shown by the vertical
line at 935 PCS. The estimated early UZ trend bisects
this line and is strengthened by the QUILF result. The
sharp increase in silica activity beyond the Main Ore
Band arises from the sudden deposition of Fe^Ti oxide
Albite
Silica
ð9Þ
Nepheline
which we may write with rearrangement
2log aSiO2 ¼ logaNaAlSi3 O8 logaNaAlSiO4 þ logK
ð10Þ
and thereby the activity of silica available to plagioclase is
diminished by loss to nepheline. With this loss, the enrichment in Ab is retarded and the relative activity of anorthite
is enhanced.
At what point does the silica activity become critically
low for this to happen? To a first approximation it is synchronous with the absence of orthopyroxene or pigeonite
among the cumulus crystals. It is ensured in the absence
of intercumulus crystals of orthopyroxene except where
local as in reaction rims or intergrowths.
Other petrographic measures can help to identify the
low-silica effect, using comparisons with the tholeiitic
Skaergaard intrusion as an example. Let us consider the
mean value of 3% An km^1 of cumulate for the Kiglapait
Lower Zone, versus several measures for the Skaergaard
2412
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
intrusion as follows (McBirney, 1996): 14% An km^1 for the
first 1100 m (the Lower Zone) and for the interval LZb
alone, 83% An km^1, even where augite was crystallizing
and magnetite arriving. It may be recalled that the
Skaergaard intrusion has an olivine hiatus where oxygen
and silica activities are high enough to destabilize Mg-olivine until the liquid evolves enough to stabilize Fe-olivine
(e.g. Morse, 1994). The Kiglapait intrusion lacks this
hiatus, but instead generates Mg-rich olivine by oxidation
when magnetite crystallizes. Therefore the intrinsic silica
activity is lower in the Kiglapait system. The LZ and
early UZ path shown in Fig. 8 defines the silica activity
trend over this interval, and it is clearly much lower than
that at Skaergaard, even when augite crystallizes.
However, in the northern Upper Zone sector of the intrusion where Fe^Ti oxide minerals are more conspicuous,
calculated silica activities at 91PCS do rise locally to
the same level as at Skaergaard (Fig. 8). This effect appears
to be a precursor of the rapid rise in silica activity after
the deposition of the Main Ore Band. In any case, it does
not affect our understanding of conditions in the Lower
Zone (0^84 PCS) where the slow evolution of plagioclase
composition occurs.
68
Mol % An
MORSE
Kiglapait Plagioclase
0 - 90 PCS
64
( )
60
Rock oxygen
norms
( )
56
Crystal separates
(Wet Chemical)
52
48
35 PCS 60
44
0
0.20
0.40
70
0.60
80
85
0.80
1
- log FL
Fig. 9. Kiglapait bulk plagioclase crystal compositions from wet
chemistry (Morse, in preparation) compared with oxygen norms
of the rocks (Morse, 1981); the latter overestimate the An content of
the plagioclase component. The systematic difference runs from
6 mol % An at the beginning of crystallization to 2% at 90 PCS. The
rocks contain significant amounts of non-feldspar CaAl component
belonging to the augite component of the liquid, whereas the norm
convention uses all the Al to make An, thereby inflating the An
content of the rocks.
An augite effect
The augite component of the magma has an opposite effect
to that of low silica activity. The CaAl component of the
pyroxene in effect steals from that available to plagioclase
and by lowering the An activity it promotes the evolution
toward Ab. The experimental normative augite content in
the LZ melt rises from 5% to 25% at saturation with
augite crystals (Morse et al., 2004; see Supplementary
Material). Therefore the retardation of plagioclase compositional evolution by low silica activity is increasingly
opposed by the decreasing activity of CaAl in the plagioclase component of the melt. Yet the low silica effect wins
out, until augite actually crystallizes, when the tables are
turned and the CaAl component is extracted from the
melt by both solid phases at once. When augite does first
crystallize, it does so in small amounts, and the mode
rises in the rocks continually from a few per cent at
84 PCS to a maximum overshoot of 42% in the volume
mode at 90 PCS (Morse, 1979c). It then has a very
strong effect on the rapid evolution of plagioclase.
A N A LY S I S O F T H E
C RY S TA L L I Z AT I O N PAT H S
Normative issues
The normative plagioclase composition of Kiglapait rocks
for the region 0^90 PCS is compared with that of separated
feldspars in Fig. 9. The rocks have persistently higher normative An than the actual plagioclase crystals that occur
within them. The excess An in the oxygen norm amounts
to about 6% at 0 PCS, 4% at 30 PCS, and diminishes to
2% at 90 PCS. One reason for this excess is that the norm
convention takes all the Al (after orthoclase and albite
have been formed) to form An to make anorthite. Instead,
a considerable amount of the resident Al resides in the pyroxene component, along with a small amount of the Ab
component NaSi. However, the amount of augite component in the early LZ rocks is minor, so the augite contribution to excess Al becomes important only at higher values
of PCS. That the difference between norms and separated
plagioclase crystals decreases with fractionation progress
shows that the Al in augite is not a major contributor to
the difference. The norm routine inflates the An content of
the plagioclase by at least the amounts shown above.
Accordingly, for purposes of modeling, the differences
shown in Fig. 9 are applied as corrections to the normative
plagioclase compositions, as shown in Table 2.
Projection to the system Diopside^
Anorthite^Albite
Conversion of data
This ternary system (Fig. 10) is well suited to an examination of the paths of natural systems along with theoretical
liquid fractionation paths. The latter are rigorously derived
as tangents to the plagioclaseçliquid legs of three-phase
triangles at the cotectic with the constraint that all paths
in the plagioclase field originate at the An corner
(Presnall, 1969; Morse, 1994). Here these paths are shown
in grayscale for a fixed value of KD ¼ 0468 (Table 1),
2413
JOURNAL OF PETROLOGY
VOLUME 55
Table 2: Kiglapait liquid paths in Di^An^Ab space
KD
–log F(L)
PCS
Oxygen norms
KD loop
Summation liquids
Plag
Liquid
An
An adj.*
f(AUG)
An Xl
An
0525
0
0
581
531
0072
671
519
0506
0155
30
567
525
0087
648
482
0497
0222
40
552
514
0098
637
466
0489
0301
50
539
506
0115
623
447
0479
0398
60
524
497
0142
604
422
0474
0456
65
515
489
0163
593
409
0468
0523
70
506
481
0190
579
392
0462
0602
75
490
468
0225
562
372
0456
0699
80
475
455
0271
540
349
0453
0745
82
466
446
0291
529
337
0445
0854
86
426
406
0330
502
310
*Adjusted for augite effect.
consistent with an assumed average Lower Zone pressure
of 4 kbar [equation (1)].
Two liquid paths for the Kiglapait Lower Zone are
shown in the figure. The continuous curve is calculated
from the normative augite content of the summation
liquid and the normative plagioclase composition adjusted
for the differences shown in Fig. 7. The input data are
given in Table 2. The curve runs dramatically toward Di
across the liquid fractionation paths, simply showing that
the observed data for the intrusion run rapidly toward saturation with augite but slowly in terms of plagioclase
evolution.
To avoid the compromised model An content of the summation liquids it is desirable to plot liquid compositions
that are derived solely from the actual plagioclase compositions as represented by the bold line in Fig. 3. Still using
the normative augite contents, the dashed curve in Fig. 10
labeled ‘KD loop’ was generated from the last three columns
in Table 2, and these in turn from the linear partitioning
equation in Appendix 1 (Supplementary Material). This
path also climbs dramatically toward augite saturation
and nearly mimics the summation path, but is located 1^
10% lower in An content.
Analysis of the results
There are several notable features to Fig. 10. First, the pure
fractionation path (dotted on the grayscale curve) leads to
the cotectic with a plagioclase composition of An 40, as
shown by the three-phase triangle. This end point compares with the plagioclase composition of An 50 at
NUMBER 12
DECEMBER 2014
equilibrium with the last liquid in the ‘KD loop’ path
(Table 2). Hence the natural crystallization path is far less
advanced than in a pure fractionation process, and at a
much higher augite to feldspar ratio (33% augite instead
of 20%). The mean augite to feldspar ratio in the
Kiglapait liquids never drops as low as 20% (Morse, 1981).
Second, the ‘KD loop’ path also crosses the liquidus fractionation lines at a considerable angle, with a final fictive
crystal composition of An 42 that would be defined by the
final tangent to the path. Instead, the actual crystal composition is An 50 (Table 2), so the path ignores the rules.
It is also unlike an equilibrium crystallization path, for
which the total solid composition would have to be An 56
given by a tieline from the last liquid through the starting
composition.
Third, both the calculated paths show a small initial
interval of pure fractionation along a liquidus fractionation
line. This result is consistent with the low initial augite contents shown in Table 2; it also shows that the modeling is
realistic.
Fourth, both paths terminate at saturation with augite
near the 1atm Di^An^Ab cotectic. This near-coincidence
suggests that the pressure effect on the location of the field
boundary is small, as the loop calculation was done for
the pressure range 51^34 kbar. A similar indifference to
pressure is also found with the experimental plagioclase^
olivine cotectic, which does not move significantly with
pressures up to 13 kbar (McIntosh, 2009). This result puts
the Aug^Plag cotectic ratio at 33% Aug for the Kiglapait
composition at augite saturation.
Fifth, the summation path is offset 1^10 mol % higher
in An content than the KD loop path, despite having
been moderately adjusted for the difference between the
norms of rocks and the actual plagioclase compositions,
as seen in Fig. 9. This result can only mean that the
summation liquid is intrinsically more An-rich than
the equilibrium liquid defined by the KD loop, and
that not all of its richness in An resides in the augite
component.
Sixth, as always the case in this system, there is a drastic
change in the liquid direction and the rate of plagioclase
evolution at the cotectic, when the liquid runs directly
away from the Di^An sideline. This is reflected in the
Kiglapait plagioclase trend (Fig. 11).
Finally, the two paths differ by one degree of freedom:
they share the same augite content of the liquid, but the
summation path is subject to closure assumptions and possible conceptual error, whereas the KD loop rigorously
finds the plagioclase liquid composition at equilibrium
with the actual crystallizing plagioclase, according to the
principles of the linear partitioning equation found by experiment (Morse et al., 2004). Still, the fact that the two
paths show the same behavior in crossing the fractionation
lines and becoming steeper with augite content shows
2414
MORSE
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
Di
KI liquids
Oxygen units
Projected from Ol
>
>
Summation
>
>
Augite
>
KD loop:
>
>
>
>
LFL’s
4 kbar
33 %
KD = 0.53-0.44
(P = 5.1 - 3.6
kbar)
>
86
PCS
> Pure FC
>
An 40 v. 50
>
>
>
Ab
>
BC
20
40
Mol %
60
80
An
Fig. 10. Liquidus fractionation lines (LFLs; gray in the plagioclase field) constructed (Presnall, 1969; Morse, 1994) for a plagioclase loop with
KD ¼ 0455 (Morse, 2013; Fig. 7) for a mean pressure of 4 kbar in the Kiglapait magma, and projected into the ternary system Diopside^
Anorthite^Albite (Bowen, 1915; Morse, 1994). Upon these lines are contructed two model fractionation paths for the Kiglapait Lower Zone,
using the cumulus plagioclase compositions from Fig. 1 and the augite contents of the norms of the calculated summation liquids (Morse,
1981), using the oxygen norm conventions of Morse et al. (2004) shown in Table 2. A tieline from the bulk composition (BC) to the initial crystal
composition at An 67 is shown. The maximum augite content of the model summation liquid (Table 2) occurs at 86 PCS as shown, and is
taken as the condition of augite saturation. The summation liquid path (continuous curve) uses the An content of the normative rock composition minus a reduction for the augite effect varying with PCS (Table 2). The path crosses the liquidus fractionation lines steeply. The future
liquid path will shadow the cotectic toward the Ab^Di sideline, leaving strongly fractionated plagioclase compositions as a result. The KD
loop liquid path (dashed) is calculated from the last column in Table 2 as calculated from the actual plagioclase compositions of the cumulus
model (Table 2 and Fig. 1) using variable KD values as shown for variable pressure. This curve has a shape similar to the summation path but
is offset to lower An contents, showing that the summation curve is based on rock compositions that have artificially elevated An contents
even more than ascribed to the augite CaAl component being assigned to normative An.
an intrinsic correlation of An with augite in the natural
fractionation process.
R EV I EW A N D CONC LUSIONS
The nature of troctolites
The clue to this unaccustomed effect lies in two main compositional truths about troctolitic magmas. First, they are
not basalts, which have been defined as rocks composed of
augite plus plagioclase. And second, they are subject to
liquid constraints on crystal fractionation that do not
occur in basalts. Troctolitic magma is a high-temperature
magma type, which, over a range of 40 C, contains
only two main crystallizing phases, olivine and plagioclase, before becoming saturated with augite (Morse et al.,
2004). According to Table 2, this range corresponds to a
variation in the liquid composition of 20 mol % An and
in the crystal composition of 17 mol %.
At the same time the fraction of augite relative to augite þ feldspar in the melt increases dramatically to 30%
in the experiments of Morse et al. (2004) and 31%
(referred to 84 PCS) in the summation liquids of Table 2.
The experiments were intended to find the equilibrium
condition, whereas the summation liquids are affected by
the ‘slow arrival’ and overproduction of augite (Morse,
1979c). That the two paths are similar in shape is
encouraging.
The entire stratigraphic array of plagioclase compositions in the intrusion is shown with a trend line (called
‘actual’) in Fig. 11. The expected path from pure fractional
crystallization (Fig. 10) is shown dotted, and the difference
at 84 PCS is 10 mol % An, as before. The shallower trend
continues until 91PCS, slightly beyond where the augite
content of the metastable liquid path becomes maximum
at 89 PCS (Morse, 1979c). Now the effect of augite crystallization is to steepen dramatically the curve of plagioclase
2415
JOURNAL OF PETROLOGY
VOLUME 55
70
mol % An
actual
50
40
DECEMBER 2014
iron to the feldspar. To the west of the olivine line, the anorthosites are mostly noritic, mostly older, and paler because of fewer inclusions owing to higher oxygen fugacity.
The color of the rocks betrays their origin. The dark troctolitic facies comes more directly from the mantle source.
The pale noritic facies has seen more crustal influence
(Morse, 2006).
Kiglapait Feldspars
60
NUMBER 12
pure FC
30
Differentiation trends in troctolite
Olivine fractionation
20
10
composition, as on the cotectic in the system Di^An^Ab.
This steepening ceases at 96 PCS, precisely where the
excess modal augite curve rejoins the normal curve in fig.
12 of Morse (1979c). The steep curve fractionates plagioclase at a rate of 17 mol % An km^1. The ensuing trend
over the last 1446 m fractionates at a rate of 138 mol %
An km^1.
The principle of fractionation influenced by the activity
of some component in the liquid extends to olivine.
In this case the retarding component within the troctolitic
Lower Zone is ferric iron, which is conserved in the liquid
until augite and then titanomagnetite crystallize at 84
and 886 PCS, respectively. Ferric iron then becomes overproduced at the ‘Main Ore Band’ at 935 PCS (e.g. Morse,
1979b, 1979c). In this stratigraphic region are found at least
eight samples with anomalously high Mg ratios, called
‘oxygen spikes’ by Morse (1979b). These examples provide
evidence that the increase in oxygen fugacity and of ferric
iron occurs at the expense of ferrous iron, so that the activity of the fayalite component is depressed and that of the
forsterite component enhanced as in the FMQ equation
(4). Accordingly, the evolution of olivine in a troctolitic
magma is retarded with respect to that in an augite-bearing magma, as illustrated in the sharp steepening of the
olivine composition curve in the Kiglapait Upper Zone
[fig. 10 of Morse (1979b)].
The nature of norites
Plagioclase fractionation
The alternative companion to troctolite is the silicated rock
called norite, composed of hypersthene and plagioclase.
A norite magma has the silica activity of the EFS reaction.
That result is similar to the upper array in Fig. 7 for the
following reason. Even though there may be no olivine
present, the olivine grid is linked to the related orthopyroxene composition by the algorithm mentioned above.
Olivine is then the proxy for Opx. Its use would only have
to be adjusted with respect to the observed pyroxene
composition. The present solution for troctolite would be
appropriate if the observed pyroxene were appropriately
more magnesian than the olivine plotted. Details can be
ignored if we only wish to make the point: Opx saturation
is essentially equivalent to tholeiitic silica activity, similar
to the Skaergaard estimate.
Norite plays a large role in the Nain anorthosite.
Troctolitic rocks occur east of the ‘Olivine line’ of Morse
(2006). They include major troctolitic intrusions but also
dark-facies anorthosite. These olivine-bearing anorthosites
are dark because they have dark gray plagioclase and that
is because the feldspars have many exsolved rods and
blades of Fe^Ti oxide minerals. These in turn are favored
by the low silica and oxygen activities that supply ferrous
The total combined effects on the retardation of plagioclase compositions in the Kiglapait Lower Zone are those
from pressure, trapped liquid, presence of olivine, and the
low silica activity in the melt. The results shown in Fig. 10
suggest a low-silica effect of 10 mol % An relative to a
pure fractionation path. The critical signature of low silica
activity in troctolites is the absence of Ca-poor pyroxene
in the cumulate rocks.
The question has been raised as to whether noritic
magmas can be shown to fractionate plagioclase faster
than troctolitic ones. Unfortunately, existing noritic examples with stratigraphic sequences comparable with the
Kiglapait troctolite are not known to this writer.
40
0
0
80 90
1
PCS
99
99.9
2
3
-log FL
99.99
4
PL Cum model
Fig. 11. Complete Kiglapait feldspar model showing steepening
trend at 9 PCS after the overproduction of augite, followed by a shallower trend after 94 PCS. The dotted curve to An 40 at 86 PCS
shows the result for pure fractional crystallization without the silica
effect, as shown in Fig. 7.
AC K N O W L E D G E M E N T S
I am indebted to Rais Latypov and many visitors to the
Kiglapait intrusion for puzzling loudly over the slow evolution of the plagioclase composition. Their queries have
spurred on this renewed study of plagioclase fractionation.
Don Lindsley awakened me from a false start to an appreciation of low silica activity to retard plagioclase evolution.
I am grateful to him and to Ron Frost and James Scoates
2416
MORSE
PLAGIOCLASE FRACTIONATION IN TROCTOLITIC MAGMA
for comments greatly improving the rigor and clarity of
this presentation.
FUNDING
This paper is based upon research supported by the US
National Science Foundation under Award No. EAR
0948095.
S U P P L E M E N TA RY DATA
Supplementary data for this paper are available at Journal
of Petrology online.
R EF ER ENC ES
Andersen, D. J., Lindsley, D. H. & Davidson, P. M. (1993). QUILF: A
Pascal program to assess equilibria among Fe^Mg^Mn^Ti oxides,
pyroxenes, olivines, and quartz. Computers and Geosciences 9,
1333^1350.
Berg, J. H. (1980). Snowflake troctolite in the Hettasch intrusion:
evidence for magma mixing and supercooling in a plutonic environment. Contributions to Mineralogy and Petrology 72, 339^351.
Berg, J. H. & Docka, J. A. (1983). Geothermometry in the Kiglapait
aureole, Labrador. AmericanJournal of Science 283, 414^434.
Bowen, N. L. (1915). The crystallization of haplobasaltic, haplodioritic,
and related magmas. AmericanJournal of Science 40, 161^185.
Bowen, N. L. (1937). Recent high-temperature research on silicates
and its significance in igneous geology. American Journal of Science,
5th Series 33, 1^21.
Carmichael, I. S. E., Nicholls, J. & Smith, A. L. (1970). Silica activity
in igneous rocks. American Mineralogist 55, 246^262.
Carmichael, I. S. E., Turner, F. J. & Verhoogen, J. (1974). Igneous
Petrology. McGraw^Hill, 739 p.
Emeleus, C. H., Cheadle, M. J., Hunter, R. H., Upton, B. G. J. &
Wadsworth, W. L. (1996). The Rum Layered Suite. In:
Cawthorn, R. G. (ed.) Layered Igneous Rocks. Elsevier,
pp. 403^440.
Maale, S. (1976). The zoned plagioclase of the Skaergaard Intrusion,
East Greenland. Journal of Petrology 17, 398^419.
McBirney, A. R. (1996). The Skaergaard intrusion. In: Cawthorn, R.
G. (ed.) Layered Igneous Rocks. Elsevier, pp. 147^180.
McIntosh, D. C. B. (2009). An experimental study exploring the magmatic source region of the Kiglapait layered intrusion, Labrador,
Canada. MS thesis, University of Massachusetts, 93 pp.
Miller, J. D., Jr & Ripley, E. M. (1996). Layered intrusions of the
Duluth Complex, Minnesota, USA. In: Cawthorn, R. G. (ed.)
Layered Igneous Rocks. Elsevier, pp. 257^302.
Morse, S. A. (1969). The Kiglapait Layered Intrusion, Labrador. Geological
Society of America, Memoirs 112, 204 p.
Morse, S. A. (1979a). Reaction constants for En^Fo^Sil equilibria: an
adjustment and some applications. American Journal of Science 279,
1060^1069.
Morse, S. A. (1979b). Kiglapait geochemistry I: Systematics, sampling,
and density. Journal of Petrology 20, 555^590.
Morse, S. A. (1979c). Kiglapait geochemistry II: Petrography. Journal
of Petrology 20, 591^624.
Morse, S. A. (1980). Kiglapait mineralogy II: Fe^Ti oxide minerals
and the activities of oxygen and silica. Journal of Petrology 21,
685^719.
Morse, S. A. (1981). Kiglapait geochemistry IV: The major elements.
Geochimica et Cosmochimica Acta 45, 461^479.
Morse, S. A. (1994). Basalts and Phase Diagrams (corrected and reprinted). Krieger, 493 pp.
Morse, S. A. (2006). Labrador massif anorthosites: Chasing the liquids
and their sources. Lithos 89, 202^221.
Morse, S. A. (2008). The internal magma reservoir of large intrusions
revealed by multiphase Rayleigh fractionation. Journal of Petrology
49, 2081^2098.
Morse, S. A. (2012). Plagioclase An range and residual porosity in igneous cumulates of the Kiglapait Intrusion. Journal of Petrology 53,
891^918.
Morse, S. A. (2013). Solidification of trapped liquid in rocks and crystals. American Mineralogist 98, 888^896.
Morse, S. A. & Ross, M. (2004). Kiglapait mineralogy IV: The augite
series. American Mineralogist 89, 1380^1395.
Morse, S. A., Brady, J. B. & Sporleder, B. A. (2004).
Experimental petrology of the Kiglapait intrusion: Cotectic trace
for the Lower Zone at 5 kbar in graphite. Journal of Petrology 45,
2225^2259.
Morse, S. A., Lindsley, D. H. & Williams, R. J. (1980). Concerning
intensive parameters in the Skaergaard intrusion. American Journal
of Science 280-A, 159^170.
Namur, O., Charlier, B., Toplis, M. J., Higgins, M. D., Lie¤geois, J.-P.
& Vander Auwera, J. (2010). Crystallization sequence and magma
chamber processes in the ferrobasaltic Sept Iles layered intrusion,
Canada. Journal of Petrology 51, 1203^1236.
Nyquist, L. E., Shih, C.-Y. & Reese, Y. D. (2012). Redetermination of
the Sm^Nd age and initial eNd of Lunar troctolite 76535: implications for Lunar crustal development. Abstracts Lunar and Planetary
Science Conference, 2416.
Presnall, D. C. (1969). The geometrical analysis of partial fusion.
AmericanJournal of Science 267, 1178^1194.
Ryan, A. B. (1990). Preliminary geological map of the Nain Plutonic
Suite and surrounding rocks (Nain^Nutak, NTS 14 S.W.), scale
1:500 000. St. Johns, NL: Geological Survey Branch, Department
of Mines and Energy, Map 90-44.
Schairer, J. F. (1950). The alkali feldspar join in the system NaAlSiO4^
KAlSiO4^SiO2. Journal of Geology 58, 512^517.
Stacey, F. D. & Davis, P. M. (2009). Physics of the Earth, 4th edn.
Cambridge University Press, 532 p.
Thy, P., Lesher, C. E. & Tegner, C. (2009). The Skaergaard liquid line
of descent revisited. Contributions to Mineralogy and Petrology 157,
735^747.
Thy, P., Lesher, C. E. & Tegner, C. (2013). Further work on experimental plagioclase equilibria and the Skaergaard liquidus temperature.
American Mineralogist 98, 1360^1367.
Wager, L. R. & Brown, G. M. (1967). Layered Igneous Rocks. Freeman,
588 p.
Wiebe, R. A. (1988). Structural and magmatic evolution of a magma
chamber: The Newark Island layered intrusion, Nain, Labrador.
Journal of Petrology 29, 383^411.
Wright, T. L. & Weiblen, P. W. (1967). Mineral composition and paragenesis in tholeiitic basalt, 1965 Makaopuhi lava lake, Hawaii.
Geological Society of America, Abstracts with Programs 242^243.
Yoder, H. S. & Tilley, C. E. (1962). Origin of basalt magmas: An
experimental study of natural and synthetic rock systems. Journal
of Petrology 3, 342^532.
Young, I. M. (1983). Observations on layering styles in the Kiglapait
intrusion. Nain Anorthosite Project, Labrador: Field Report 1981.
University of Massachusetts Department of Geology & Geography
Contribution 40, 105^112.
2417