the importance of temperature on ree and other - USRA

Lunar and Planetary Science XLVIII (2017)
1535.pdf
THE IMPORTANCE OF TEMPERATURE ON REE AND OTHER TRACE ELEMENT PARTITIONING
IN PLAGIOCLASE WITH APPLICATIONS TO LUNAR MAGMA OCEAN SOLIDIFICATION. C. Sun1,2
and Y. Liang1, 1Department of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI
02912 ([email protected]), 2Earth Science Department, Rice University, Houston, TX 77025 ([email protected]).
Database: Data used in our model calibration are from
29 plagioclase-melt trace element partitioning experiments from [5-11]. The 29 experiments were conducted at 1127 – 1410 °C and 1 atm to 1.5 GPa under nominally anhydrous or water-saturated conditions. Plagioclase crystals and melts from these experiments cover
a large range of composition (An = 41 – 98 in plagioclase; SiO2 = 41.82 – 62.50 wt% and Mg# = 49 – 100
in melt). To ensure a self-consistent calibration dataset,
we excluded partitioning experiments that are suspect
of kinetics and/or analytical artifacts.
Models: Given the compiled plagioclase-melt trace
element partitioning data, we developed parameterized
lattice strain models for the partitioning of 1+, 2+ and
3+ cations and noble gas following a protocol that includes rigorous statistical treatment of partitioning data
through parameter swiping and simultaneous nonlinear
least squares inversion of all the selected partitioning
data. Fig. 1 compares our model predicted D’s with
experimentally measured values. In the lattice strain
model for REE partitioning, the strain-free partition
coefficient (D0) is given by the expression
2
19.45+1.17P 2
ln D 3+ = 16.05−
× 104 −5.17 ( X ) ,
0
Ca
RT
where P is in GPa and XCa is the Ca content in plagioclase per eight-oxygen. The ideal radius (r0) and effective Young’s modulus (E) for REE are constant
(1.179Å and 196 GPa, respectively). The new models
indicate that REE and other trace element partitioning
in plagioclase depends strongly on temperature and the
abundances of Ca and Na in plagioclase. Particularly,
Na content in plagioclase primarily controls divalent
element partitioning, while temperature and Ca content
in plagioclase are the dominant factors for REE partitioning in plagioclase. For a given plagioclase composition, DREE decrease with the decrease of temperature.
This is an unique and important feature of REE partitioning in plagioclase and it has important implications
for REE fractionation during LMO crystallization.
1
10
χ2p = 6.0
0
10
Predicted D
Introduction: Plagioclase is one of the major rockforming minerals constituting the crusts of Earth, the
Moon and other planetary bodies. The lunar highland
crust is dominated by ferroan anorthosites (FANs) with
>90% anorthitic plagioclase (An > 95; An =
100×Ca/(Ca+Na+K) in mol; e.g., [1-2]). According to
the canonic model of lunar formation and evolution,
FANs are products of crystallization and floatation of
plagioclase in a global-scale lunar magma ocean
(LMO; [2-3] and references therein). Thus, compositions of plagioclase record the differentiation processes
of terrestrial and lunar magmas.
A key parameter for unraveling the magmatic processes involving plagioclase is the plagioclase-melt
trace element partition coefficient (D), which generally
is a function of temperature (T), pressure (P), and
compositions of the mineral and melt (X). Since Drake
and Weill’s pioneering work [4], many experimental
studies have been conducted to determine trace element partitioning between plagioclase and silicate melt
(e.g., [5-11]). Several empirical or semi-empirical
models have also been calibrated for predicting plagioclase-melt DREE (e.g., [5-9, 12]). However, these partitioning models are incomplete, especially for lunar
relevant melt compositions and temperatures.
In this study, we developed parameterized lattice
strain models for the partitioning of 1+, 2+, 3+, cations
and noble gas between plagioclase and silicate melt
over a wide range of T, P, and X. These new partitioning models are broadly applicable to plagioclase-melt
fractionation during terrestrial and lunar magmatisms.
Application of our models to LMO fractional crystallization indicates that the temperature effect on DREE is
fundamental to understanding chemical fractionation
during LMO solidification and must be considered in
future modeling studies.
−1
10
−2
10
1
2:
−3
10
−3
10
1
1:
1+ cations (Li, Na, K)
2+ cations (Ca, Mg, Sr, Ba, Ra)
3+ cations (REE and Y)
2
1:
−2
10
−1
10
0
10
1
10
Measured D
Figure 1. Comparison between experimentally determined
D’s and model predicted values.
Solidification of LMO: Fractional crystallization of a
global lunar magma ocean was calculated using the
MAGFOX program for a lunar upper mantle composition (LPUM; [13]). Anorthite begins to crystalize at
78% solidification of the LMO at 1250 °C and 4.6
Lunar and Planetary Science XLVIII (2017)
1535.pdf
kbar. As crystallization proceeds from 78% to 94%,
the anorthite content in plagioclase slightly decrease by
~2 mol% and the temperature drops from 1250 °C to
995 °C. Using the new partitioning models, we calculated plagioclase-melt D’s for three representative
stages of LMO solidification (F): plagioclase saturation (F = 78%, at 1250 °C and 4.6 kbar), spinel saturation (F = 89%, at 1144 °C and 2.0 kbar), and the maximum extent solidification that MAGFOX is capable of
handling (F = 94%, at 995 °C and 1.0 kbar). The results are shown in Fig. 2.
Plagioclase-melt partition coefficient
10 1
Phinney &
Morrison
(1990)
10 0
Jones (1995)
10 -1
10 -2
F=
10
-3
F=7
8%
89%
F=9
4%
10 -4
LMO solidification
1250
°C
1144
°C
995°
C
(cf. Fig. 2), the estimated REE abundances in the parental magmas appear to increase from 100 to 2000
times chondrite values with increasing Eu negative
anomalies. Interestingly, the REE compositions of
FAN parental magmas calculated using D’s at 89%
LMO solidification are comparable to those of the
KREEP basalts (K, REE, and P rich; [17]).
Because temperature is the dominate facter controlling anorthite-melt DREE, the anomalous REE enrichments and apparent Eu negative anomalies in FAN
parental magmas may be attributed to (1) subsolidus
re-equilibration; (2) more complicated crystallization
process, or (3) inappropriate T used for FAN crystallization. Assuming a higher T, REE abundances in the
estimated parental magmas become more comparable
to previous estimations (grey regions; calculated using
D’s from [18]) and display flat patterns without significant Eu anomalies (see dashed outline in Fig. 3).
Thus, determination of FAN crystallization temperatures is a prerequisite for estimating their parental
magmas and further assessing formation of the lunar
crust. Effects of (1) and (2) will also be assessed.
Ar* Li Rb Sr Ra La Pr Sm Gd Dy Ho Tm Lu
Na K Cs Ba Mg Ce Nd Eu Tb Y Er Yb
Parental melts of FANs: Using the anorthite-melt
partitioning data derived from LMO solidification
(78%, 89% and 94%), we reassessed the parental
magmas of FANs with reported anorthite REE data
from [15-16]. Fig. 3 displays one example of the estimated parental magma compositions. For each set of
DREE, the estimated REE abundances vary by a factor
of five, which may be attributed to different extents of
LMO solidification (e.g., [15-16]). Because of the systematic decreases of DREE during LMO solidification
F=94% (995°C)
103
F=89% (1144°C)
KREEP
102
F=78% (1250°C)
1350°C
The partition coefficients of divalent cations (Sr,
Ba, Ra, and Eu2+) display small changes across the 250
°C temperature interval, because of the strong dependence on Ca content in plagioclase and the nearly invariant plagioclase composition; however, the small increase of albite content in plagioclase gives rise to
about a factor of two decrease in DMg. As crystallization proceeds from 78% to 94%, the predicted noble
gas D increases by a factor of three, monovalent element D’s increase by a factor of three to four, whereas
DREE decrease by 1.5 orders of magnitude. Given the
nearly constant anorthite content in plagioclase, these
variations mainly result from the 250 °C temperature
reduction. Together with our recent study on REE partitioning in low-Ca pyroxene [14], the large variations
of plagioclase-melt DREE underscore the importance of
temperature and composition in quantifying trace element fractionation during LMO solidification.
104
CI normalized FAN parental melt
Figure 2. Calculated plagioclase-melt D’s for LMO fractional crystallization. Ar* denotes the calculated noble gas D’s.
101
P&M90
La Ce Nd Sm Eu Gd Dy Y Ho Yb
Figure 3. Chondrite normalized REE abundances in FAN
parental magmas calculated using D’s at different temperatures. KREEP compositions are from [18].
References: [1] McGee (1993) JGR 98, 9089-9105. [2]
Papike et al. (1998) RiMG 36, 5-1. [3] Warren (1985) Annu.
Rev. EPS 13, 201-240. [4] Drake & Weill (1975) GCA 39,
689-712. [5] Bindeman et al. (1998) GCA 62, 1175-1193. [6]
Bindeman & Davis (2000) GCA 64, 2863-2878. [7] Tepley et
al. (2006) JVGR 157, 147-165. [8] Aigner-Torres et al.
(2007) CMP 153, 647-667. [9] Fabbrizio et al. (2009) EPSL
280, 137-148. [10] Blundy et al. (1998) EPSL 160, 493-504.
[11] Graff et al. (2013) LPSC 44th, #1641. [12] Wood &
Blundy (2003) Treatise on Geochem. 2, 395-424. [13]
Longhi (2006) GCA 70, 5919-5934. [14] Sun & Liang (2013)
GCA 119, 340-358. [15] Papike et al. (1997) GCA 61, 23432350. [16] Floss et al. (1998) GCA 62, 1255-1283. [17] Warren (1989) Apollo 14, KREEP, and Evolved Lunar Rocks,
pp.149-153. [18] Phinney & Morrison (1990) GCA 54, 16391654.