Vanpeteghem 2008

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Phys Chem Minerals (2008) 35:493–504

DOI 10.1007/s00269-008-0244-4

ORIGINAL PAPER

The effect of oxygen vacancies and aluminium substitution


on the high-pressure properties of brownmillerite-structured
Ca2Fe2-xAlxO5
Carine B. Vanpeteghem Æ Ross J. Angel Æ Jing Zhao Æ
Nancy L. Ross Æ Günther J. Redhammer Æ
Friedrich Seifert

Received: 18 December 2007 / Accepted: 27 April 2008 / Published online: 24 May 2008
Ó Springer-Verlag 2008

Abstract The structural evolution with pressure and the The unit-cell compression is anisotropic, with the c-axis
equations of state of three members of the brownmillerite being stiffer than a or b, and the anisotropy increases with
solid solution, Ca2(Fe2-xAlx)O5, have been determined by increasing Al content of the structure. The structural
single-crystal X-ray diffraction up to a maximum pressure response to pressure of all samples is similar. The
of 9.73 GPa. The compositions of the samples were (Al,Fe)O4 tetrahedra and the (Al,Fe)O6 octahedra undergo
x = 0.00 and x = 0.37 (with Pnma symmetry) and approximately isotropic compression. There is an increase
x = 0.55 (with I2mb symmetry). No phase transitions were in the twists of the chains of corner-sharing (Al,Fe)O4
observed in the experiments. The equation of state tetrahedra, and an increase in the tilts of the (Al,Fe)O6
parameters determined from the pressure-volume data are octahedra, because these framework polyhedra are stiffer
K0T = 128.0 (7) GPa, K0 0 = 5.8 (3) for the sample with than the Ca–O bonds to the extra-framework Ca site. The
x = 0.00, K0T = 131 (2) GPa, K0 0 = 5.5 (4) for x = 0.37, alignment of the two shortest Ca–O bonds sub-parallel to
and K0T = 137.5 (6) GPa, K0 0 = 4 for x = 0.55. The bulk [001] accounts for the relative stiffness of the c-axis and
modulus therefore increases with Al content, being 11% thus the elastic anisotropy.
higher in the x = 0.55 sample than in the Al-free sample.
Keywords Single-crystal  X-ray diffraction 
Oxygen vacancies  Perovskite
Electronic supplementary material The online version of this
article (doi:10.1007/s00269-008-0244-4) contains supplementary
material, which is available to authorized users. Introduction

C. B. Vanpeteghem  R. J. Angel (&)  J. Zhao  N. L. Ross


Over the past 10 years, there has been a growing interest in
Crystallography Laboratory, Department of Geosciences,
Virginia Polytechnic Institute and State University, the effect of Al substitution on the physical properties of
Blacksburg, VA 24061, USA (Fe,Mg)SiO3 perovskite (e.g. McCammon 1997; Richmond
e-mail: rangel@vt.edu and Brodholt 1998; Wood 2000; Brodholt 2000; Lauter-
bach et al. 2000; Andrault et al. 2001; Andrault et al. 2007;
G. J. Redhammer
Department of Materials Engineering and Physics, Nishio-Hamane et al. 2007; Kojitani et al. 2007; Nishiyama
Division of Mineralogy, University of Salzburg, et al. 2007), because it has been suggested that a small
Hellbrunnerstrasse 34, 5020 Salzburg, Austria amount of Al can affect significantly the properties and
characteristics of mantle perovskites. For example, Zhang
F. Seifert
Bayerisches Geoinstitut, Universität Bayreuth, and Weidner (1999) have claimed that 5 mol% of Al2O3
95440 Bayreuth, Germany incorporated in silicate-perovskite makes it 10% more
compressible than pure MgSiO3.Wood and Rubie (1996)
C. B. Vanpeteghem
and Frost and Langenhorst (2002) indicated that the pres-
School of Chemistry and Biochemistry,
Georgia Institute of Technology, ence of Al creates a significant increase in the partitioning
Atlanta, GA 30332, USA of Fe into Fe–MgSiO3 perovskite co-existing with

123
494 Phys Chem Minerals (2008) 35:493–504

magnesiowustite. This can be understood in the light of a understanding of the physical properties of mantle Fe–Al–
single-crystal X-ray diffraction structure determination of MgSiO3 perovskites.
Fe–Al–MgSiO3 that revealed that the Fe is incorporated
into the perovskite structure by the coupled substitution
MgA2+ + SiB4+ , FeA3+ + AlB3+. Excess Fe is then Experimental methods
accommodated by the substitution MgA2+ , FeA2+. Nei-
ther mechanism requires the formation of oxygen vacan- Details of the sample syntheses of the Ca2(Fe2-xAlx)O5
cies (Vanpeteghem et al. 2006). However, little is known single crystals with x [ 0 are given in Redhammer et al.
about how the incorporation of Al, Fe and of oxygen (2004). The amount of Al3+ substituted for Fe3+ was
vacancies change the structure of Fe–Al–MgSiO3 at the determined by electron microprobe analyses and single-
pressure and temperature conditions of the lower mantle crystal X-ray diffraction (Redhammer et al. 2004). A new
and how the changes influence the chemistry of the deep sample of end-member Ca2Fe2O5 was synthesized by solid-
Earth’s mantle. state reaction from homogenized mixtures of CaCO3,
One of the experimental difficulties in addressing these Fe2O3 and Al2O3 as described in Ross et al. (2002).
issues is that MgSiO3 and Fe–Al–MgSiO3 perovskites Selected high-quality single crystals of approximately
quenched to ambient pressure and room temperature may 120 lm 9 80 9 35 lm of three different compositions
not retain the full defect structures and chemistry from (x = 0.00, 0.37 and 0.55) were used in the high-pressure
the conditions of synthesis. Crystal chemical insights in to studies. The samples were loaded, separately, in an ETH-
the substitution mechanisms and physical properties at the type diamond-anvil cell (Miletich et al. 2000) mounted
atomic scale can also be acquired by studying analogue with 600 lm diameter culet diamonds. A T301 steel gasket
systems stable at lower pressure and temperature condi- was indented to a thickness of about 90 lm and a
tions. For example, an ideal system to explore substitution 250 lmm diameter hole was drilled at the center. A piece
mechanisms and the effect of oxygen vacancies into of quartz and a ruby sphere were both used as the internal
perovskites is CaTiO3–Ca2Fe2O5. In this system Ti4+ is pressure calibrants and were loaded together with the
substituted by Fe3+ and creates oxygen vacancies (Vo) sample and a mixture of 4:1 methanol:ethanol as the
through the charge-balanced reaction 2 Ti4+ , 2 Fe3+ + pressure transmitting medium. The unit-cell parameters
Vo. The state of order of these vacancies depends on the were determined by least-squares fits, at each pressure, to
specific conditions of pressure, temperature and their the corrected setting angles of about 15 reflections, mea-
concentration. Details of these mechanisms are reported by sured by the eight-position centering technique on a Huber
Becerro et al. (1999, 2000) and McCammon et al. (2000). four-circle diffractometer. Pressure was determined from
The resulting Ca2Fe2O5 has the brownmillerite struc- the measured unit-cell volume of the quartz using the
ture-type, ‘‘brownmillerite’’ being the mineral name for equation of state of Angel et al. (1997).
the 50%-Al-substituted composition Ca2AlFeO5. The Ambient and high-pressure intensity data were collected
structure can be considered to be a perovskite-like for all three samples with an Xcalibur-1 diffractometer
structure of a theoretical ‘‘CaFeO3’’ composition from equipped with a point detector and a MoKa radiation
which oxygen atoms have been removed to create a source operated at 50 kV and 40 mA. Crystal offsets from
framework structure with ordered oxygen vacancies. the center of the goniometer were determined by measuring
Consequently, the structure is made of corner-linked FeO4 approximately 20 strong low-angle reflections and calcu-
tetrahedra and FeO6 octahedra. Exchange of Al for Fe on lating the crystal offsets from the reflection positions with
both the octahedral and tetrahedral sites leads to the solid the WinIntegrStp program (Angel 2003). The DAC was
solution Ca2(Fe2-xAlx)O5 that includes brownmillerite then adjusted accordingly to minimize the offsets before
itself. With increasing Al content the space group changes data collection. The intensities of all reflections were
from Pnma to I2mb symmetry when x is between 0.45 measured in fixed-phi mode (Finger and King 1978) with
and 0.70, but the general structural topology is retained constant-precision step scans in x. The WinIntegrStp
(Redhammer et al. 2004). The same transition occurs in software was used for peak fitting and integration.
pure Ca2Fe2O5 upon heating to 724 (4)°C. A complete Absorption corrections were performed by numerical
temperature-composition phase diagram can be found in integration in ABSORB 6.0 (Angel 2004). After averaging
Fig. 4 of Redhammer et al. (2004). In this manuscript we the symmetrically equivalent reflections, the remaining
report the results of a structural study of the solid solution independent reflection with (F [ 4r(F)) were used to
at high pressures in order to investigate in detail the effect refine structures with RFINE99, an updated version of
of both Al and oxygen vacancies on the compressibility RFINE4 (Finger and Prince 1974). Unit-cell parameters
of brownmillerite structures. The aim is to provide measured on the Huber diffractometer were also used in the
information that can be used towards developing a better structure refinements.

123
Phys Chem Minerals (2008) 35:493–504 495

Results and discussion 5.44

0 % Al
5.42 30 %Al
Lattice parameters and compressibilities 55 %Al
a
5.4

The cell parameters measured at room conditions in this


5.38
study are in excellent agreement with the previously pub-

Å
lished data (Redhammer et al. 2004). The parameters a and 5.36
b decrease with increasing Al content, while c remains
5.34
nearly unchanged, leading to a decrease in unit-cell volume
with increasing x. Note that this also results in a decrease 5.32
in density with increasing Al-content (Redhammer et al.
5.3
2004).
The unit-cell parameters and volumes of all three sam-
ples decrease smoothly with pressure (Fig. 1; Table 1)
0 % Al
without any discontinuities or changes in slope. There is 30 %Al
14.7 b 55 %Al
thus no evidence of any phase transition in any of the
samples over the pressure range explored, a conclusion
confirmed by the structure refinements to the intensity data 14.6
that are described below. The normalized pressure F (Birch

Å
1978) plotted against the Eulerian strain measure f (e.g.
Angel 2000) provides a direct indication whether higher 14.5

K0 0 and K00 0 are significant in the equation of state. In both


the Al-free (x = 0) and the x = 0.37 samples, the F–f plot
14.4
indicates that K0 0 [ 4, and therefore a third-order equation
has been used. However, the P–V data for x = 0.55 sample
result in values of F that do not change with f, which means
5.58
K0 0 = 4. The resulting fit of the P–V data with the appro- 0 % Al
30 %Al
priate equation of state for each sample shows that the bulk c 55 %Al
modulus of pure Ca2Fe2O5 end-member is 128.0 (7) GPa 5.56

with K0 0 = 5.8 (3) which is in good agreement with Ross


et al. (2002). For the x = 0.37 and x = 0.55 samples the
Å

5.54
bulk moduli are K0T = 131 (2) GPa with K0 0 = 5.5 (4) and
K0T = 137.5 (6) GPa with K0 0 = 4, respectively. The bulk
5.52
modulus therefore increases with Al content, being 11%
higher in the x = 0.55 sample than the value obtained for
5.5
the Al-free sample (Fig. 2). This trend of stiffening of the
structure with Al substitution, even though the density
decreases with increasing Al content (Redhammer et al.
450
2004), is similar to, but stronger than, that found for true
Ca–perovskites in which the bulk modulus increases uni- 445
0 % Al
30 %Al
variantly with decreasing molar volume (Ross and Angel V 55 %Al

1999, Ross et al. 2002). 440

As the brownmillerite structure is a perovskite structure


Å3

435
with vacancies, we can use the current results to infer the
effect of oxygen vacancies on the compressibility of per- 430
ovskites, and especially MgSiO3. There has been a long
425

420

Fig. 1 Variation of lattice parameters a, b and c, and unit-cell c


volume, V, of the brownmillerite samples as a function of pressure. 0 2 4 6 8 10

The lines are the equations of state (Table 2) fit to the data Pressure (GPa)

123
496 Phys Chem Minerals (2008) 35:493–504

Table 1 Unit cell parameters of Ca2(Fe2-xAlx)O5 at different pressures

P (GPa) 0.0001 1.539 (7) 2.249 (1) 3.251 (7) 3.863 (8) 4.942 (8) 5.564 (8) 6.326 (8) 7.190 (9) 8.167 (9) 9.73 (1)
x = 0.0 Pnma

a (Å) 5.4271 (2) 5.4054 (3) 5.3956 (3) 5.3818 (3) 5.3744 (5) 5.3595 (2) 5.3522 (2) 5.3422 (3) 5.3324 (2) 5.3246 (2) 5.3035 (2)
b (Å) 14.7619 (5) 14.7025 (6) 14.6756 (6) 14.6392 (5) 14.6175 (2) 14.5831 (6) 14.5621 (9) 14.5385 (9) 14.5114 (7) 14.4819 (2) 14.4389 (3)
c (Å) 5.5978 (2) 5.5787 (2) 5.5703 (3) 5.5594 (2) 5.5526 (3) 5.5412 (2) 5.5354 (2) 5.5277 (3) 5.5195 (1) 5.5110 (1) 5.4979 (1)
V (Å3) 448.46 (3) 443.36 (4) 441.08 (4) 437.99 (3) 436.21 (5) 433.09 (2) 431.42 (4) 429.32 (4) 427.10 (3) 424.63 (3) 421.01 (3)

P (GPa) 0.0001 0.995 (6) 2.015 (6) 3.140 (7) 3.729 (6) 4.919 (9) 6.267 (5) 8.89 (1)
x = 0.37 Pnma

a (Å) 5.4042 (6) 5.3904 (5) 5.3769 (6) 5.3624 (6) 5.3549 (6) 5.3403 (6) 5.3246 (6) 5.2953 (5)
b (Å) 14.6835 (2) 14.6429 (2) 14.6032 (2) 14.5617 (2) 14.5410 (2) 14.4992 (2) 14.4558 (2) 14.3759 (3)
c (Å) 5.5947 (3) 5.5826 (3) 5.5712 (3) 5.5591 (3) 5.5524 (3) 5.5409 (3) 5.5273 (2) 5.5037 (2)
3
V (Å ) 443.96 (5) 440.65 (4) 437.45 (5) 434.08 (5) 432.34 (5) 429.04 (5) 425.44 (5) 418.97 (4)

P (GPa) 0.0001 0.488 (5) 1.49 (5) 2.375 (6) 3.21 (1) 4.511 (8) 4.88 (1) 6.27 (8) 6.97 (1) 7.73 (1)
x = 0.55 I2mb

a (Å) 5.383 (3) 5.374 (1) 5.361 (1) 5.3500 (8) 5.3394 (9) 5.324 (1) 5.3189 (9) 5.303 (1) 5.294 (1) 5.2866 (9)
b (Å) 14.610 (2) 14.4251 (2) 14.5523 (3) 14.5173 (2) 14.4853 (2) 14.4380 (3) 14.4251 (2) 14.3778 (3) 14.3574 (3) 14.3321 (2)
c (Å) 5.5920 (2) 5.5870 (5) 5.5958 (5) 5.5672 (4) 5.5584 (4) 5.5450 (6) 5.5420 (4) 5.5290 (6) 5.5226 (4) 5.5163 (4)
V (Å3) 439.7 (2) 438.14 (7) 434.99 (7) 432.39 (6) 429.91 (7) 426.24 (9) 425.22 (7) 421.57 (8) 419.80 (8) 417.96 (7)

et al. 2006) showed that the dominant mechanism for


170
introduction of Al and Fe in to MgSiO3 perovskite involves
a coupled cation substitution and therefore does not result
160 c-axis
directly in the creation of vacancies. Nonetheless, as
pointed out by Ross et al. (2002), the bulk modulus of pure
Ca2Fe2O5 falls below the trend for stoichiometric Ca–
150 perovskites, so the presence of any oxygen vacancies in
Kd(GPa)

perovskites, whatever the mechanism of formation, will


140
soften the structure. Our new data show that this conclusion
remains true even for the Al-substituted samples, although
Volume the degree of softening will be less for higher Al contents
130 of the perovskite.
a-axis The elastic linear modulus of each unit-cell axis for all
three samples was obtained by fitting a second or third
120
(when the F–f plot showed K0 0 [ 4) order Birch–Murna-
b-axis ghan equation of state to the cubes of each cell parameter
(Angel 2000). All three samples show strong anisotropy of
0 0.1 0.2 0.3 0.4 0.5 0.6
unit-cell compression (Table 2). In the Al-free sample, the
x in Ca2(Fe2-xAlx)O5 a- and b-axes are the softest and it is the c-axis that is the
Fig. 2 The variation of linear and bulk moduli with composition
least compressible (12% stiffer than the b-axis). These
(Table 2) in Ca2(Fe,Al)2O5 from the fit of third order Birch– results agree with those of Ross et al. (2002) when
Murnaghan EoS to the unit cell parameters and volume variation allowance is made for them using a cell setting in which
with pressure. The lines are linear fits to the moduli the a- and c-axes are swapped with respect to the setting
used in this paper. As noted previously (Ross et al. 2002)
the introduction of oxygen vacancies not only softens the
debate (e.g. Brodholt 2000; Andrault et al. 2001) on how perovskite structure, but the ordering of those vacancies
the presence of Al and Fe, and of oxygen vacancies, affects (Becerro et al. 1999, 2000) significantly increases the
the elasticity of MgSiO3. Our previous work (Vanpeteghem elastic anisotropy over that of true perovskites. Our new

123
Phys Chem Minerals (2008) 35:493–504 497

Table 2 The bulk modulus and linear compressibilities of brown-


millerite with various amounts of Al
Ca2(Fe2-xAlx)O5 x=0 x = 0.37 x = 0.55

Number of data points 12 8 10


Pmax(GPa) 9.82 8.89 7.73
Volume
V0 (Å3) 448.48 (3) 443.95 (4) 439.71 (4)
K0 (GPa) 128.0 (7) 131 (1) 135 (1)
K00 5.8 (3) 5.5 (3) 4.6 (4)
a-axis
a0 5.4271 (3) 5.4039 (4) 5.3829 (3)
Ka0 122.1 (5) 128.6 (8) 122 (3)
K0 4 (fixed) 4 (fixed) 6 (1)
b-axis
b0 14.762 (8) 14.6831 (2) 14.6104 (5)
Kb0 120.5 (9) 117.6 (6) 118.8 (4)
K0 5.8 (2) 5.5 (1) 4 (fixed)
c-axis
c0 5.59 (6) 5.5951 (4) 5.5921 (3)
Kc0 134 (6) 148 (3) 166 (4)
K0 11 (1) 8.1 (8) 6 (1)

data show that the degree of elastic anisotropy is further


increased by the incorporation of Al in to the structure
because the c-axis becomes stiffer while there is little
significant change in the linear moduli of the a- and b-axes
(Fig. 2). Therefore, the pattern of elastic anisotropy is not Fig. 3 Polyhedral representation of the two polymorphs of the
just the result of ordering of the oxygen vacancies, but brownmillerite structure-type, viewed down [010], showing the two
must also be dependent upon the details of the structural (010) layers of tetrahedral chains on either side of a octahedral (010)
response to applied stresses. We will explore this theme sheet. Ca positions are represented by spheres. Note that the relative
rotation of the tetrahedral chains in the two layers is different in the
further in the following section. two structures. The (001) a-glide planes are marked by the thick
broken lines. The O2–Ca–O3 strut that stiffens the c-axis is marked by
Crystal structure a solid thick line in the upper diagram, and the O3–O3–O3 linkage that
defines the tetrahedral chain rotation is indicated in the lower diagram
The crystal structures of Ca2(Fe2-x Alx)O5 are comprised
of alternating (010) layers of corner-sharing MO6 inversion center at the M site, so the tetrahedra are twisted
(M = Fe, Al) octahedra and TO4 (T = Fe, Al) tetrahedra, in the same direction in adjacent layers (Fig. 3). By con-
with the tetrahedra forming isolated single chains that trast, these chains are related by a twofold axis passing
extend along [100] (Fig. 3) in the cell setting used here through the M site in I2mb and thus have the opposite of
(with a \ c). The Ca atoms occupy the interstitial sites rotation when viewed down [010] as seen in Fig. 3.
within the framework, at positions analogous to those The details of the structural evolution with composition
found in true perovskites. The vacancies on the oxygen were previously reported by (Redhammer et al. 2004), and
positions result in the Ca site being 8-coordinated. The our structural refinements on intensity data collected at
samples with x = 0.00 and x = 0.37 have Pnma symmetry, ambient pressure in the diamond-anvil cells are in good
and the x = 0.55 sample has I2mb symmetry (Redhammer agreement with those results (Tables 3, 4, 5 and 4a, 4b, 4c,
et al. 2004). The two space groups result in the same 5a, 5b, 5c, 6a, 6b, 6c in Electronic supplementary material).
structural topology, with the only significant difference Aluminium preferentially substitutes in to the tetrahedral
being that the relative orientations of the tetrahedral chains site, with Al occupancies typically three times those in the
are reversed between Pnma and I2mb. In Pnma consecutive octahedral site (Redhammer et al. 2004). There is no evi-
layers of chains at y = 1/4 and y = 3/4 are related by an dence from either Mossbauer spectroscopy (Redhammer

123
498 Phys Chem Minerals (2008) 35:493–504

Table 3 Refinement parameters, refined fractional occupancies, refined atomic coordinates and equivalent isotopic temperature factor (Beq) of
Ca2Fe2O5
0.01 GPa 2.25 GPa 3.86 GPa 5.56 GPa 6.33 GPa 8.17 GPa 9.73 GPa

Nb. reflections 941 1,019 972 1,023 1,110 1,070 902


Rint 0.054 0.052 0.052 0.058 0.052 0.055 0.069
Refinement
Nb. of reflections 378 434 410 330 354 326 280
Ru 0.042 0.044 0.044 0.048 0.043 0.042 0.043
Rw 0.043 0.047 0.048 0.054 0.044 0.045 0.043
Ca
x 0.4810 (2) 0.4803 (2) 0.4808 (2) 0.4802 (2) 0.4802 (2) 0.4802 (2) 0.4804 (2)
y 0.1081 (2) 0.1082 (1) 0.1080 (1) 0.1080 (2) 0.1083 (2) 0.1088 (2) 0.1089 (2)
z 0.0226 (2) 0.0224 (2) 0.0223 (2) 0.0218 (2) 0.0216 (2) 0.0217 (2) 0.0219 (2)
Beq 0.73 (2) 0.65 (2) 0.66 (2) 0.51 (2) 0.65 (2) 0.60 (2) 0.65 (2)
M (Fe = 1.00, Al = 0.00)
x 0 0 0 0 0 0 0
y 0 0 0 0 0 0 0
z 0 0 0 0 0 0 0
Beq 0.36 (2) 0.35 (2) 0.36 (2) 0.19 (2) 0.35 (2) 0.28 (2) 0.37 (2)
T (Fe = 1.00, Al = 0.00)
x 0.9451 (2) 0.9457 (2) 0.9459 (2) 0.9446 (2) 0.9454 (2) 0.9455 (2) 0.9449 (2)
y 0.25 0.25 0.25 0.25 0.25 0.25 0.25
z 0.9335 (2) 0.9347 (2) 0.9349 (2) 0.9344 (2) 0.9351 (2) 0.9353 (2) 0.9352 (2)
Beq 0.44 (2) 0.41 (2) 0.44 (2) 0.29 (2) 0.46 (2) 0.42 (2) 0.45 (2)
O1
x 0.2633 (5) 0.2629 (6) 0.2621 (5) 0.2627 (7) 0.2624 (6) 0.2617 (6) 0.2622 (6)
y 0.9821 (6) 0.9830 (4) 0.9850 (4) 0.9827 (8) 0.9829 (5) 0.9836 (6) 0.9825 (7)
z 0.2368 (6) 0.2371 (6) 0.2372 (6) 0.2371 (6) 0.2377 (6) 0.2380 (6) 0.2376 (6)
Beq 0.47 (6) 0.42 (5) 0.49 (5) 0.31 (7) 0.43 (6) 0.36 (5) 0.59 (6)
O2
x 0.0227 (6) 0.0255 (7) 0.0250 (7) 0.0270 (8) 0.0273 (7) 0.0266 (7) 0.0276 (8)
y 0.1406 (5) 0.1394 (5) 0.1397 (4) 0.1404 (7) 0.1409 (5) 0.1414 (5) 0.1417 (8)
z 0.0731 (8) 0.0741 (8) 0.0734 (7) 0.073 (8) 0.0773 (8) 0.0761 (7) 0.0756 (8)
Beq 0.88 (7) 0.89 (6) 0.89 (7) 0.62 (8) 0.92 (7) 0.72 (7) 0.90 (7)
O3
x 0.6001 (9) 0.5976 (9) 0.5955 (9) 0.596 (1) 0.5923 (9) 0.5916 (9) 0.5945 (9)
y 0.25 0.25 0.25 0.25 0.25 0.25 0.25
z 0.8737 (9) 0.8737 (9) 0.8746 (9) 0.873 (1) 0.8759 (9) 0.8756 (9) 0.8756 (9)
Beq 0.53 (8) 0.42 (8) 0.66 (9) 0.4 (1) 0.62 (9) 0.61 (9) 0.52 (9)

et al. 2004) or X-ray diffraction for any additional short or appropriate coordination environment for the Ca (Red-
long range ordering of Fe and Al within either site. The hammer et al. 2004). Note that this is the same mechanism,
substitution of Al for Fe results in an overall decrease in the but with the opposite effect, as that found on substituting
volumes of both the TO4 and MO6 polyhedra by reductions Al + Fe in to MgSiO3 perovskite (Vanpeteghem et al.
in the mean hT–Oi and hM–Oi bond lengths. If there were 2006). In that case, the Al is substituting for a smaller (Si)
no change in the tilts of the tetrahedra or octahedra, then octahedral cation, and results in an increase in the tilts of
this decrease in the size of the polyhedra would lead to the octahedra. The transition from Pnma to I2mb in
decreased Ca–O bond lengths. Therefore, both the tilts of Ca2(Fe2-x Alx)O5 results in changes in some structural
the octahedra decrease and the tetrahedral chains extend trends with composition. But, for the x = 0.55 sample
with increasing Al content in order to maintain an studied here, the only significant changes from the Pnma

123
Phys Chem Minerals (2008) 35:493–504 499

Table 4 Refinement parameters, refined fractional occupancies, refined atomic coordinates and equivalent isotopic temperature factor (Beq) of
Ca2Fe1.63Al0.37O5
0.01 GPa 2.01 GPa 3.72 GPa 6.27 GPa 8.89 GPa

Nb. reflections 843 899 777 758 808


Rint 0.052 0.053 0.063 0.054 0.061
Refinement
Nb. reflections 319 338 301 240 244
Ru 0.050 0.047 0.055 0.040 0.042
Rw 0.058 0.054 0.062 0.043 0.045
Ca
x 0.4840 (5) 0.4838 (5) 0.4920 (6) 0.4832 (6) 0.4824 (7)
y 0.10830 (9) 0.10830 (8) 0.10822 (1) 0.10832 (9) 0.10837 (9)
z 0.0248 (2) 0.0240 (2) 0.239 (2) 0.0238 (2) 0.0233 (2)
Beq 0.65 (2) 0.63 (2) 0.96 (3) 0.64 (2) 0.54 (3)
M (Fe = 0.93, Al = 0.07)
x 0 0 0 0 0
y 0 0 0 0 0
z 0 0 0 0 0
Beq 0.45 (3) 0.45 (2) 0.72 (3) 0.47 (2) 0.37 (3)
T (Fe = 0.70, Al = 0.30)
x 0.9472 (5) 0.9468 (5) 0.9491 (1) 0.9470 (8) 0.9459 (8)
y 0.25 0.25 0.25 0.25 0.25
z 0.9318 (2) 0.9325 (3) 0.9332 (3) 0.9334 (2) 0.9338 (3)
Beq 0.34 (3) 0.32 (3) 0.69 (4) 0.36 (3) 0.30 (3)
O1
x 0.260 (1) 0.259 (1) 0.264 (2) 0.264 (2) 0.259 (2)
y 0.9854 (3) 0.9844 (2) 0.9840 (3) 0.9837 (3) 0.9826 (3)
z 0.2402 (8) 0.2404 (7) 0.2404 (9) 0.2384 (8) 0.2394 (8)
Beq 0.70 (7) 0.57 (7) 0.91 (9) 0.63 (7) 0.51 (7)
O2
x 0.023 (1) 0.024 (1) 0.026 (1) 0.0274 (2) 0.026 (2)
y 0.1417 (8) 0.1419 (3) 0.1426 (3) 0.1417 (3) 0.1420 (3)
z 0.0711 (8) 0.0735 (9) 0.075 (1) 0.0729 (7) 0.0763 (8)
Beq 0.79 (8) 0.85 (7) 1.05 (9) 0.64 (7) 0.73 (8)
O3
x 0.602 (3) 0.602 (2) 0.598 (3) 0.606 (3) 0.608 (3)
y 0.25 0.25 0.25 0.25 0.25
z 0.8705 (9) 0.8691 (9) 0.877 (1) 0.871 (1) 0.870 (1)
Beq 0.41 (9) 0.41 (9) 1.0 (1) 0.55 (1) 0.5 (1)

structures are an increased distortion of the MO6 octahe- effect of Al-substitution within the structure at room
dron and a decrease in the tilts of the octahedra, which pressure, except that the compression is isotropic whereas
result in changes to the Ca–O bond lengths. In the fol- Al substitution increases the distortion (Redhammer et al.
lowing we compare the structural evolution with pressure 2004). Under pressure, there is also significant compression
to these trends with composition. of the T–O bonds but they become less compressible as the
Within the framework under pressure there is significant Al content increases (Fig. 4). Similar to the MO6 octa-
compression of the M–O bonds and the O–O edges of the hedra, the tetrahedra do not deform internally with pressure
octahedra, with no large changes in the O–M–O angles as the O–T–O angles remain nearly constant and the tetra-
(Tables 4a, 4b, 4c in ESM). The resulting reduction in the hedral volume reduction is due to bond shortening alone.
octahedral volume with pressure therefore mimics the Thus the response of the framework to increased pressure,

123
500 Phys Chem Minerals (2008) 35:493–504

Table 5 Refinement parameters, refined fractional occupancies, refined atomic coordinates and equivalent isotopic temperature factor (Beq) of
Ca2Fe1.45Al0.55O5
0.01 GPa 1.49 GPa 3.21 GPa 4.51 GPa 6.27 GPa 7.73 GPa

Nb. reflections 494 512 574 533 411 453


Rint 0.053 0.051 0.054 0.055 0.061 0.055
Refinement
Nb. reflections 374 376 315 281 228 269
Ru 0.047 0.050 0.042 0.042 0.045 0.048
Rw 0.049 0.054 0.048 0.044 0.051 0.053
Ca
x 0.491 (1) 0.491 (1) 0.489 (1) 0.488 (1) 0.492 (1) 0.495 (1)
y 0.10833 0.10830 0.10824 0.10849 0.1085 0.10827
z 0.0266 0.0261 0.0264 0.0260 0.0256 0.0258
Beq 0.71 0.69 0.62 0.54 0.72 0.71
M (Fe = 0.87, Al = 0.13)
x 0 0 0 0 0 0
y 0 0 0 0 0 0
z 0 0 0 0 0 0
Beq 0.54 (3) 0.57 (3) 0.52 (2) 0.46 (2) 0.60 (3) 0.56 (3)
T (Fe = 0.58, Al = 0.42)
x 0.955 (1) 0.953 (1) 0.952 (2) 0.952 (1) 0.952 (1) 0.958 (1)
y 0.25 0.25 0.25 0.25 0.25 0.25
z 0.9295 (3) 0.9300 (3) 0.9302 (3) 0.93078 (2) 0.9309 (3) 0.9312 (3)
Beq 0.57 (3) 0.56 (3) 0.49 (3) 0.49 (3) 0.58 (4) 0.58 (4)
O1
x 0.246 (3) 0.244 (4) 0.243 (4) 0.237 (7) 0.247 (4) 0.258 (4)
y 0.9861 (2) 0.9862 (3) 0.9853 (3) 0.9842 (3) 0.9848 (3) 0.9844 (3)
z 0.256 (1) 0.256 (1) 0.259 (2) 0.260 (2) 0.258 (2) 0.253 (2)
Beq 0.58 (8) 0.72 (9) 0.66 (8) 0.54 (8) 0.7 (1) 0.67 (8)
O2
x 0.024 (2) 0.026 (3) 0.028 (3) 0.025 (3) 0.025 (3) 0.037 (3)
y 0.1424 (3) 0.1429 (3) 0.1420 (3) 0.1429 (3) 0.1430 (4) 0.1425 (4)
z 0.0699 (9) 0.0699 (9) 0.0708 (9) 0.0718 (8) 0.072 (1) 0.072 (1)
Beq 0.67 (9) 0.71 (8) 0.80 (9) 0.78 (8) 0.9 (1) 0.8 (1)
O3
x 0.612 (2) 0.613 (3) 0.6178 (3) 0.611 (3) 0.609 (3) 0.613 (4)
y 0.25 0.25 0.25 0.25 0.25 0.25
z 0.863 (1) 0.869 (1) 0.871 (1) 0.868 (1) 0.869 (1) 0.871 (1)
Beq 0.7 (1) 0.7 (1) 0.6 (1) 0.2 (1) 0.8 (2) 0.6

and increased Al substitution, cannot be described in terms to chains. Consecutive O3 atoms, and consecutive T sites,
of rigid polyhedra, but must be analysed in terms of along a chain are related to one another by the a-glide
polyhedra each undergoing different amounts of approxi- present in both space groups (Fig. 3). The O2 atoms of the
mately isotropic compression. tetrahedra lie above and below the mirror plane, and thus
The framework polyhedra also undergo changes in tilts the O2–O2 edge of the tetrahedron is constrained to be
with both Al substitution and increasing pressure, but the exactly parallel to [010]. As a consequence, the tetrahedron
tilting patterns are severely restricted by the symmetry can only rotate around the O2–O2 edge (i.e. around [010]),
present in both Pnma and I2mb space groups. The critical with the rotation angle from the
h straight configuration
i (O3–
constraint is that the T sites lie on the (010) mirrors at O3–O3 = 180°) given by 180  ðO3 O23 O3 Þ : Provided
y = 1/4, 3/4, as do the O3 atoms that link the tetrahedra in there is no internal distortion of the tetrahedra, changes in

123
Phys Chem Minerals (2008) 35:493–504 501

1.9
240

Polyhedral Bulk Modulus (GPa)


1.88
42 % Al
200
<T-O> bonds (Å)

0 % Al TO4
1.86
0 % Al MO
160 6 7 % Al
0 % Al
1.84 30 % Al
37 % Al 13 % Al

120
55 % Al CaO8
1.82

80
1.8

0 2 4 6 8 10
0 0.1 0.2 0.3 0.4 0.5 0.6
Pressure (GPa)
x in Ca2(Fe2-xAlx)O5
Fig. 4 The variation of average \T–O[ bond lengths with pressure.
The dotted lines are linear fits to the data Fig. 5 Polyhedral bulk moduli for MO6, TO4 and CaO8 polyhedra
determined by fitting the pressure-induced variation of the polyhedral
the length of the a-axis are thus directly related to changes volumes with a Birch-Murnaghan second-order EoS. The dashed lines
are linear fits to the moduli. The percentage of Al indicates the Al
in the size of the tetrahedra and their rotation angle. Further
occupancy of MO6 and TO4 at each composition
consideration of Fig. 3 shows that the distance between the
O2 oxygens in neighbouring tetrahedra is also determined decreases in both the M–O1–M and M–O2–T bond angles
by the tetrahedral size and the rotation angle. Since the O2 with increasing pressure. At the same time, all three sam-
oxygens link the tetrahedral and octahedral sheets by also ples show changes in T–O3–T angles and O3–O3–O3 angles
forming the apical oxygens of the octahedra, it follows that (Tables 5a, 5b, 5c in ESM) that, within the uncertainties,
the rotation of the MO6 octahedra around [010] is also appear to indicate that the chains become more twisted
determined by the M–O1 bond length and the tetrahedral (less extended) with pressure which is opposite to the effect
size and rotation angle (Fig. 3; Redhammer et al. 2004). of increasing Al content at room pressure.
Lastly, because the two O1 atoms forming the octahedral In summary, both the MO6 and TO4 polyhedra within
edges parallel to [100] are also related by the a-glide, the the framework behave in the same way. Their volumes
octahedra cannot rotate about [001], but only around [100] both decrease with both higher Al content and pressure,
in addition to the [010] rotation described above (Red- but the tilting and rotation of the framework polyhedra
hammer et al. 2004). The [100] rotation can be calculated increases with increasing pressure, whereas it decreases
either from the  coordinates of the O2 apical oxygen as with increasing Al content. As a consequence, the changes
zðO2 Þ:c
arctan yðO 2 Þ:b
or from
 thetilt of the equatorial plane of O1 in the Ca coordination with pressure differ from those
1 Þ:b
oxygens as arctan yðO zðO1 Þ:c : The fact that the deviations of
associated with Al substitution for Fe. The Ca-site is
the O–M1–O bond angles from 90° are small means that irregularly coordinated by eight O atoms, including four O1
these two estimates agree to within 0.5°. In a similar atoms that form the links between the octahedra and three
fashion the [010] rotation can be  determined
 from the O2 atoms that link the octahedra with the tetrahedra. One of
zðO1 Þ:c
coordinates of O1 as 45  arctan xðO 1 Þ:a
: Both tilts con- these Ca–O2 bonds, and the bond from the Ca to the
tribute to changes in the M–O1–M angle away from the bridging O3 atom of the tetrahedral chains, are less than
value of 180° that would be found in the un-tilted structure. 2.35 Å in length, significantly shorter than the other Ca–O
Aluminium substitution leads to decreases in both the bonds. As the Al content increases, these two shortest
[100] and [010] tilts of the octahedra (see Fig. 12 in Red- bonds become slightly shorter, while the Ca–O1 and Ca–O2
hammer et al. 2004), and corresponding increases in the bonds of intermediate length become longer and the lon-
M–O1–M and M–O2–T bond angles. By contrast, pressure gest bonds become shorter (Fig. 17 in Redhammer et al.
results in no change, within the considerable experimental 2004). Increasing pressure, however, reverses the tilting
uncertainties, in the [010] tilt in any of the three samples, pattern and, in combination with the effect of decreasing
but an increase in the [100] tilt by between 1° and 1.7° in polyhedral size, results in a decrease in all individual Ca–O
the x = 0.37 and x = 0.55 samples. This results in bond lengths (Tables 6a, 6b, 6c in ESM).

123
502 Phys Chem Minerals (2008) 35:493–504

Clearly, the effect of Al substitution and pressure on the the structural basis for development of elastic anisotropy
structure are not the same, and the structural response with the ordering of oxygen vacancies within the brown-
therefore cannot be described in terms of site sizes or bond millerite structure type. The anisotropy increases with Al
lengths alone. The rotation and tilting patterns observed in content, because Al substitution results in further shorten-
our study can, however, be explained in terms of relative ing of these two short Ca–O bonds, while four of the Ca–O
compressibilities of the MO6 octahedra, TO4 tetrahedra and bonds of intermediate length become longer (Fig. 17 of
the CaO8 polyhedra. By fitting an appropriate equation of Redhammer et al. 2004).
state to the volume of the polyhedra, we can obtain an The Pnma structure of Ca2Fe2O5 has previously been
estimate of their compressibilities and bulk moduli. Given shown to invert to I2mb phase at high temperature (Red-
the uncertainties in the data and the limited pressure range, hammer et al. 2004). But in our experiments at high
the polyhedral volume was fit with pressure using the same pressures, there were no changes in the pattern of sys-
K0 0 determined for the unit-cell volume variation with tematic absences (I vs. P lattices) in the diffraction patterns
pressure. The bulk moduli of the MO6 and CaO8 remain within the experimental pressure range. This clearly indi-
roughly constant over the compositional range studied, cates that the space-group symmetries remain unchanged
with CaO8 being on average 18% more compressible than and no transition occurred in any of the three samples.
MO6. There is a clear increase in the bulk modulus of TO4 Consideration of the structure, and results from studies of
as the Al content increases (Fig. 5), as a result of the T–O structurally similar transitions in pyroxenes, suggests that
bonds getting shorter with increasing Al content. Conse- this should not be taken as evidence to infer either the
quently in the most Al-rich sample, at x = 0.55, the TO4 position or slope of the equilibrium phase boundary in P–T
tetrahedra are stiffer than the MO6 octahedra. space. Figure 3 shows that the transition from Pnma to
The experimental data therefore show that, as in Ca– I2mb symmetry only requires the reversal of the rotation of
perovskites (Andrault and Poirier 1991; Zhao et al. 2004a, alternate (010) layers of tetrahedral chains. Such a reversal
b), the extra-framework Ca site is softer than the polyhedra would only require the breaking of a single Ca–O3 bond
comprising the framework. This is confirmed by the per tetrahedron, the rotation of the entire chain and the
observation that bond-valence matching is maintained in reforming of a new Ca–O3 bond on the other side of the
Ca2Fe2O5 at high pressure within the uncertainty of the chain. Such a mechanism is very similar to the M2–O3
measurements, as holds for normal perovskites in which bond breaking, chain rotation, and M2–O3 bond formation
the bond-valence increase on the framework cation induced that accompanies the P21/c to C2/c structural phase tran-
by compression matches that of the extra-framework cation sition in clinopyroxenes (e.g. Angel et al. 1992; Arlt et al.
(Zhao et al. 2004b). Given that bond-valence strain is 1998; Arlt and Angel 2000a). Such transitions in clinopy-
equally partitioned across all three sites (Ca, M and T) in roxenes are first-order in character and exhibit varying
end-member Ca2Fe2O5 one can obtain (cf. Zhao et al. degrees of hysteresis as a consequence of having to break
2004b) a reasonable estimate of the relative compressibil- the M2–O3 bonds. For example, the hysteresis in spodu-
ities of bCa:bM:bT * 1:0.55:0.62. Thus, as in perovskites, mene under compression at room temperature is essentially
the tilts of the frameworks of the brownmillerite structures zero (Arlt and Angel 2000b), that in FeSiO3 is less than
increase with increasing pressure because the Ca site is 0.2 GPa (Hugh-Jones et al. 1994), whereas the same tran-
softer than the framework. sition in MgSiO3 exhibits approximately 2 GPa hysteresis
Interaction between the framework and the Ca site also (Angel and Hugh-Jones 1994). Clearly, the degree of
accounts for the significant anisotropy of compression in hysteresis is related to the precise atomic mechanism of the
the brownmillerite structure, with the c-axis being signifi- transition (e.g. Arlt and Angel 2000b). Thus our failure to
cantly stiffer than the a- or b-axes. There is a similar observe either the Pnma to I2mb transition in the Fe-rich
pattern of cell parameter change with increasing Al content samples, or the reverse transition in the x = 0.55 sample,
within the Pnma phase; the a- and b-cell parameters presumably merely means that its mechanism results in it
decrease by *1% whereas c decreases by only 0.2%. This being kinetically hindered at our experimental conditions.
anisotropy cannot be due to the topology of the framework Therefore, it should not be considered evidence that the
structure alone because the three allowed rotations of the Pnma-I2mb equilibrium phase boundary has not been
framework polyhedra, the [100] and [010] rotations of the crossed in any of our experiments.
octahedra and the [010] rotation of the tetrahedral chains,
can produce equal compression of all three axes. Figure 3,
however, shows that the two shortest Ca–O bonds, to O2 Summary
and O3, lie sub-parallel to [001]. This specific orientation
of the shortest, and therefore stiffest, Ca–O bonds to form a The key conclusion we can draw from this study of the
strut (cf. Megaw 1970) between the tetrahedral chains is brownmillerite structure at various compositions and

123
Phys Chem Minerals (2008) 35:493–504 503

pressures is that it behaves like a perovskite, with the Arlt T, Angel RJ (2000a) Displacive phase transitions in C-centred
‘‘A-site’’ (i.e. Ca) being softer than the ‘‘B-site’’ (i.e. clinopyroxenes: spodumene, LiScSi2O6 and ZnSiO3. Phys Chem
Minerals 27(10):719–731
the framework). Consequently, the framework exhibits Arlt T, Angel RJ (2000b) Pressure buffering in a diamond anvil cell.
increasing tilts with increasing pressure to compress the Mineral Mag 64(2):241–245
Ca-site. The effect of an increase of Al and an increase of Becerro AI, McCammon CA, Langenhorst F, Seifert F, Angel RJ (1999)
pressure are in some ways similar; both result in a decrease Oxygen-vacancy ordering in CaTiO3–CaFeO2.5 perovskites: from
isolated defects to infinite sheets. Phase Trans 69:133–146
in the volumes of the framework polyhedra. But, increasing Becerro AI, Langenhorst F, Angel RJ, Marion S, McCammon CA,
Al content leads to a decrease in the tilts of the framework Seifert F (2000) The transition from short-range to long-range
as a consequence of having to maintain an appropriate ordering in oxygen vacancies in CaFexTi1-xO3-x/2 perovskites.
coordination for the Ca site. All of the samples measured in J Phys Chem 2:3993–3941
Birch F (1978) Finite strain isotherm and velocities for single-crystal
this study are significantly softer than the equivalent Ca and polycrystalline NaCl at high pressure and 300 K. J Geophys
perovskites as a consequence of the introduction of O Res 83:1257–1268
vacancies in to the perovskite framework. The elastic Brodholt JP (2000) Pressure-induced changes in the compression
anisotropy of the brownmillerite structure arises from the mechanism of aluminous perovskite in the Earth’s lower mantle.
Nature 407:620–622
ordering of these vacancies, so that infinite tetrahedral Finger LW, Prince E (1974) US Nat Bur Stand. NBS Tech Note 854
chains are formed, the twisting of which lead to short and Finger LW, King HE (1978) Revised method of operation of single-
strong Ca–O bonds which form a strut sub-parallel to [001] crystal diamond cell and refinement structure of NaCl. Am
and thus stiffen this direction compared to the other axes. Mineral 63:337–342
Frost DJ, Langenhorst F (2002) The effect of Al2O3 on Fe–Mg
Our measurements have led to several insights into the partitioning between magnesiowutsite and magnesium silicate
effect of Al and oxygen vacancies on the elastic properties perovskite. Earth Planet Sci Lett 199:227–241
of perovskite-type structure, including MgSiO3. Further Hugh-Jones DA, Woodland AB, Angel RJ (1994) The structure of
studies on analogue systems to higher pressures and tem- high-pressure C2/c ferrosilite and crystal—chemistry of high-
pressure C2/c pyroxenes. Am Mineral 79(11–12):1032–1041
peratures, similar to mantle conditions, would certainly Kojitani H, Katsura T, Akaogi M (2007) Aluminum substitution
expand our current knowledge and understanding of the mechanisms in perovskite-type MgSiO3: an investigation by
properties of mantle perovskites. Rietveld analysis. Phys Chem Miner 34:257–267
Lauterbach S, McCammon CA, van Aken P, Langenhorst F, Seifert F
Acknowledgments This work was supported by NSF grant EAR- (2000) Mossbauer and ELNES spectroscopy of (Mg,Fe)
0408460 to NLR and RJA. (Si,Al)O3 perovskite: a highly oxidized component of the lower
mantle. Contrib Mineral Petrol 138:17–26
McCammon C (1997) Perovskite as a possible sink for ferric iron in
the lower mantle. Nature 387:696–699
References McCammon CA, Becerro AI, Langenhorst F, Angel RJ, Marion S,
Seifert F (2000) Short-range ordering of oxygen vacancies in
Andrault D, Poirier JP (1991) Evolution of the distortion of CaFexTi1-xO3-x/2 perovskites (0 \ x \ 0.40). J Phys C: Cond
perovskites under pressure—an EXAFS study of BaZrO3, Matter 12:2969–2984
SrZrO3 and CaGeO3. Phys Chem Miner 18:91–105 Megaw H (1970) Structural relationship between coesite and feldspar.
Andrault D, Bolfan-Casanova N, Guignot N (2001) Equation of state Acta Crystallogr B26:261–265
of lower mantle (Al,Fe)–MgSiO3 perovskite. Earth Planet Sci Miletich R, Allan DR, Kuhs W (2000) High-pressure single-crystal
Lett 193:501–508 techniques. Rev Mineral Geochem 41:445–519
Andrault D, Bolfan-Casanova N, Bouhifd MA, Guignot N, Kawamoto Nishio-Hamane D, Fujino K, Seto Y, Nagai T (2007) Effect of the
T (2007) The role of Al-defects on the equation of state of Al– incorporation of FeAlO3 into MgSiO3 perovskite on the post-
(Mg,Fe)SiO3 perovskite. Earth Planet Sci Lett 263:167–179 perovskite transition. Geophys Res Lett 34. doi:L12307/
Angel RJ (2000) Equations of state. Rev Mineral Geochem 41:35–60 1-L12307/4
Angel RJ (2003) Automated profile analysis for single-crystal Nishiyama N, Yagi T, Ono S, Gotou H, Harada T, Kikegawa T (2007)
diffraction data. J Appl Cryst 36:295–300 Effect of incorporation of iron and aluminum on the thermo-
Angel RJ (2004) Absorption corrections for diamond-anvil cell elastic properties of magnesium silicate perovskite. Phys Chem
implemented in the software Absorb 6.0. J Appl Cryst 37:486– Miner 34:131–143
492 Redhammer GJ, Tippelt G, Roth G, Amthauer G (2004) Structural
Angel RJ, Hugh-Jones DA (1994) Equations of state and thermo- variations in the brownmillerite series Ca2(Fe2-xAlx)O5: single-
dynamic properties of enstatite pyroxenes. J Geophys Res crystal X-ray diffraction at 250 C and high-temperature X-ray
99:19777–19783 powder diffraction (250C \ T \ 1,000C). Am Mineral 89:405–
Angel RJ, Chopelas A, Ross NL (1992) Stability of high-density 420
clinoenstatite at upper-mantle pressures. Nature 358:322–324 Richmond NC, Brodholt JP (1998) Calculated role of aluminum in the
Angel RJ, Allan DR, Miletich R (1997) The use of quartz as internal incorporation of ferric iron into magnesium silicate perovskite.
pressure standard in high-pressure crystallography. J Appl Cryst Am Mineral 83:947–951
30:461–466 Ross NL, Angel RJ (1999) Compression of CaTiO3 and CaGeO3
Arlt T, Angel RJ, Miletich R, Armbruster T, Peters T (1998) High- perovskites. Am Mineral 84:277–281
pressure P21/c-C2/c phase transitions in clinopyroxenes: influ- Ross NL, Angel RJ, Seifert F (2002) Compressibility of brownmille-
ence of cation size and electronic structure. Am Mineral 83(11– rite (Ca2Fe2O5): effect of vacancies on the elastic properties of
12):1176–1181 perovskites. Phys Earth Planet Int 129:141–151

123
504 Phys Chem Minerals (2008) 35:493–504

Vanpeteghem CB, Angel RJ, Ross NL, Jacobsen SD, Dobson DP, Zhang J, Weidner DJ (1999) Thermal equation of state of aluminum-
Litasov KD, Ohtani E (2006) Al, Fe substitution in the MgSiO3 enriched silicate perovskite. Science 284:782–784
perovskite structure: a single-crystal X-ray diffraction study. Zhao J, Ross NL, Angel RJ (2004a) Tilting and distortion of
Phys Earth Planet Int 155:96–103 perovskite structure CaSnO3 from single-crystal X-ray diffrac-
Wood BJ, Rubie DC (1996) The effect of alumina on phase tion study at high pressure up to 7 GPa. Phys Chem Miner
transformations at the 660 km discontinuity from Fe–Mg 19:299–305
partitioning experiments. Science 273:1522–1524 Zhao J, Ross NL, Angel RJ (2004b) New view of the high-pressure
Wood BJ (2000) Phase transformations and partitioning relations in behaviour of GdFeO3-type perovskites. Acta Crystallogr B
peridotite under lower mantle conditions. Earth Planet Sci Lett 60:263–271
174:341–354

123

You might also like