American Mineralogist, Volume 101, pages 277–288, 2016

CHEMISTRY AND MINERALOGY OF EARTH’S MANTLE

Some thermodynamic properties of larnite (β-Ca2SiO4) constrained by high T/P

experiment and/or theoretical simulation

Zhihua Xiong1,2, Xi Liu1,2,*, Sean R. Shieh3, Sicheng Wang1,2, Linlin Chang4, Junjie Tang1,2,
Xinguo Hong5, Zhigang Zhang6, and Hejing Wang1,2

The Key Laboratory of Orogenic Belts and Crustal Evolution, Ministry of Education of China, Beijing 100871, China
1

2
School of Earth and Space Sciences, Peking University, Beijing 100871, China
3
Department of Earth Sciences, University of Western Ontario, London, Ontario N6A 5B7, Canada
4
College of Earth Science, University of Chinese Academy of Sciences, Beijing 100049, China
5
Mineral Physics Institute, State University of New York, Stony Brook, New York 11974, U.S.A.
6
Key Laboratory of Earth and Planetary Physics, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing 100029, China

Abstract
Pure larnite (β-Ca2SiO4; Lrn) was synthesized at 6 GPa and 1473 K for 6 h by using a cubic press,
its thermal expansivity was investigated up to 923 K by using an X‑ray powder diffraction technique
(ambient P), and its compressibility was investigated up to ~16 GPa by using a diamond-anvil cell
coupled with synchrotron X‑ray radiation (ambient T). Its volumetric thermal expansion coefficient
(αV) and isothermal bulk modulus (KT) were constrained as αV = 4.24(4)×10–5 K–1 and KT = 103(2) GPa
[the first pressure derivative KTʹ obtained as 5.4(4)], respectively. Its compressibility was further studied
with the CASTEP code using density functional theory and planewave pseudopotential technique. We
obtained the KT values as 123(3) GPa (LDA; high boundary) and 92(2) GPa (GGA; low boundary),
with the values of the KTʹ as 4.4(9) and 4.9(5), respectively. The phonon dispersions and vibrational
density of states (VDoS) of Lrn were simulated using density functional perturbation theory, and the
VDoS was combined with a quasi-harmonic approximation to compute the isobaric heat capacity (CP)
and standard vibrational entropy (S0298), yielding CP = 212.1(1) – 9.69(5)×102T–0.5 – 4.1(3)×106T–2 +
5.20(7)×108T–3 J/(mol∙K) for the T range of ~298–1000 K and S0298 = 129.8(13) J/(mol∙K). The micro-
scopic and macroscopic thermal Grüneisen parameters of Lrn at 298 K were calculated to be 0.75(6)
and 1.80(4), respectively.
Keywords: β-Ca2SiO4, compressibility, entropy, heat capacity, larnite, thermal expansivity, thermal
Grüneisen parameter, thermodynamic property

Introduction in diamonds possibly originating from the lower mantle (Joswig

Natural larnite (Lrn; β-Ca2SiO4; space group P21/n with Z = et al. 1999; Nasdala et al. 2003; Brenker et al. 2005). This new
4) was first reported in the Larne district of Great Britain (Tilley appearance argues that some portions of the deep interior of
1929) and soon documented on the island of Muck in Scotland the Earth are significantly Ca-richer than the normal pyrolitic
(Tilley 1947) and the Tokatoka region in New Zealand (Mason mantle (Ringwood 1975). According to the results from some
1957). It usually locates in the contact zone between an igneous high-P experimental studies (e.g., Green and Ringwood 1967;
rock like dolerite or andesite, and a calcitic rock such as chalk or Takahashi 1986; Irifune 1994), the Ca-rich silicate in the normal
limestone (Tilley 1929; Mason 1957). Lrn commonly coexists mantle normally changes from clinopyroxene, majoritic garnet to
with spurrite, melilite, merwinite, and spinel and readily trans- calcium perovskite (CaPv; compositionally close to CaSiO3) as
forms to the olivine-structured Ca2SiO4 phase (the γ-Ca2SiO4 P increases from the surface to the core-mantle boundary of the
phase) if it is shocked or heated at certain temperature (Tilley Earth. To stabilize Lrn, Wal, and Ttn, significantly higher levels
1929). According to Bowen (1940) and Tilley (1951), Lrn gener- of CaO should be introduced into the mantle, which has been
ally represents the high-T stage of the progressive metamorphism presumably accomplished via deep subduction of the CaO-rich
(or decarbonation) of the siliceous limestones. continental crust material (Irifune et al. 1994; Liu et al. 2012a).
This conventional field occurrence of Lrn was recently It is well known that Lrn is not a stable phase at ambient P
supplemented with another completely different appearance, (Fig. 1a). From low T to high T, the composition of Ca2SiO4
tiny mineral inclusions, usually coexisting with the walstromite- can form many polymorphs, namely γ-Ca2SiO4, αʹL-Ca2SiO4,
structured CaSiO3 (Wal) and titanite-structured CaSi2O5 (Ttn), αʹH-Ca2SiO4, and α-Ca2SiO4 (e.g., Barnes et al. 1980; Remy et
al. 1995, 1997a, 1997b; Yamnova et al. 2011). Lrn never forms
during heating, but appears as a metastable phase in the stabil-
* E-mail: [email protected]
Special collection papers can be found on GSW at http://ammin.geoscienceworld. ity field of the γ-Ca2SiO4 phase during cooling. Consequently,
org/site/misc/specialissuelist.xhtml. it was usually synthesized by heating the Ca2SiO4 composition

0003-004X/16/0002–277$05.00/DOI: http://dx.doi.org/10.2138/am-2016-5425 277

278 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE

with the former taking place at ~8 GPa and the latter at ~12
GPa (Kanzaki et al. 1991; Wang and Weidner 1994; Akaogi
et al. 2004; Sueda et al. 2006). To understand these reactions,
especially the one related to the CaPv, which is the dominant
Ca-bearing phase in the lower mantle of the Earth (Mao et
al. 1977; Irifune et al. 1989) but unfortunately unquenchable
to room P for further characterizations such as single-crystal
crystallographic study and calorimetric investigation (Liu and
Ringwood 1975; Mao et al. 1977; Irifune 1994), it is neces-
sary to investigate the thermodynamic properties of Lrn as
extensively and accurately as possibly. These thermodynamic
properties of Lrn then can be combined with the thermodynamic
properties of Ttn to rigorously constrain the thermodynamic
properties of CaPv via an accurate experimental determination
of the P-T conditions of Equation 2. The thermal expansivity
of Lrn has been investigated by different techniques, but the
volumetric thermal expansion coefficient at 298 K (α298) varies
from ~1.80 × 10–5 to 4.25 × 10–5 K–1 (e.g., Remy et al. 1997b;
Swamy and Dubrovinsky 1997). The isothermal bulk modulus
(KT) of Lrn has also been investigated (Remy et al. 1997b;
Swamy and Dubrovinsky 1997), but its value at 298 K varies
from ~119 to 166(15) GPa (the first pressure derivative of KT,
Figure 1. Phase diagram of composition Ca2SiO4: (a) Phase diagram KTʹ, set as 4). Heat capacity measurements have been performed
at 1 atm (T in K); (b) Phase diagram at high-P and high-T conditions
from 52.66 to 296.48 K (Todd 1951) and from 406.0 to 964.6 K
(HP-HT). The phase sequence in a is from Remy et al. (1997a). The
(Coughlin and O’Brien 1957). With an empirical sum equation
solid univariant curves of γ = β, γ = αʹL, and β = αʹL in b are from Remy
et al. (1995). The solid circle in b represents the P-T condition of our of Debye and Einstein functions to extrapolate the heat capacity
Lrn-synthesizing experiments, which locates in the stability field of the data down to 0 K, an S0298 value as 127.6(8) J/(mol∙K) has been
Lrn phase, but near the high P-T extension of the β = αʹL univariant curve. derived, which has never been verified by any means so far. It
In addition, Liu (1978) observed a K2NiF4-structured Ca2SiO4 phase at thus appears that a substantial amount of work has to be con-
~24 GPa and 1273 K, and Wang and Weidner (1994) detected the phase ducted to fully understand the thermodynamic features of Lrn.
transition between the β phase and the αʹL phase at ~9 GPa and 1350 K. In this study, we synthesized pure Lrn in its P-T stability
field by using a massive cubic press, determined its thermal
in the stability field of the phase αʹL-Ca2SiO4, αʹH-Ca2SiO4, or expansivity by using an in situ high-T powder X‑ray diffrac-
α-Ca2SiO4, and then quickly quenching into water (Remy et al. tion method (ambient P), and its compressibility by using a
1997a, 1997b; Fukuda et al. 1997), a process that might introduce synchrotron X‑ray radiation combined with a diamond-anvil
different amounts of defects in the products (e.g., twins are com- cell (DAC; ambient T). Its compressibility was further studied
mon in Lrn; Groves 1983; Kim et al. 1992). To preserve the Lrn with the CASTEP code using density functional theory and
structure, some “stabilizer” such as P2O5 (Saalfeld 1975), B2O3 planewave pseudopotential technique. Using density functional
(Remy et al. 1997a), vacancy of either Ca (Rodrigues 2003) or perturbation theory, its phonon dispersions and vibrational
Si (Yannaquis and Guinier 1959), or other species (Lai et al. density of states (VDoS) were simulated, and the VDoS was
1992) was usually introduced into the system. combined with a quasi-harmonic approximation to compute the
According to the phase diagram of the composition Ca2SiO4 isobaric heat capacity (CP) and vibrational entropy (S). Finally,
at high-P and high-T conditions (Fig. 1b), Lrn has a vast stabil- the microscopic and macroscopic thermal Grüneisen parameters
ity field. At the low-P side, Lrn transforms to the γ-Ca2SiO4 of Lrn at the ambient P-T condition were calculated.
phase at ~2 GPa and room T (Hanic et al. 1987; Remy et al.
1995, 1997b; Reynard et al. 1997), and the invariant point of Experimental and simulating methods
β-Ca2SiO4 + γ-Ca2SiO4 + αʹL-Ca2SiO4 locates at ~0.34 GPa and The polycrystalline Lrn sample used in our thermal expansion and compres-
711 K (Hanic et al. 1987). At the high-P side, the relationship sion experiments was synthesized with the CS-IV 6 × 14 MN cubic press at the
High-Pressure Laboratory of Peking University (Liu et al. 2012b). In the high-P
between Lrn and the αʹL-Ca2SiO4 phase (Wang and Weidner 1994) synthesizing experiments, we used the assembly BJC-1 (Liu et al. 2012c), with its
or K2NiF4-structured Ca2SiO4 phase (Liu 1978) is still poorly pressure calibrated at ambient T by using the high-P phase transitions of Bi (I-II
constrained (Fig. 1b). transition at 2.55 GPa, and II-III transition at 2.69 GPa) and Ba (I-II transition at
Lrn also participates in some other geologically important 5.5 GPa). The temperature in the high-P synthesizing experiments was measured
and controlled by employing a Pt94Rh6-Pt70Rh30 thermocouple (type B), with any
reactions such as
potential pressure effect on the e.m.f. ignored. We prepared the starting material
3Wal = Ttn + Lrn (1) for the high-P synthesizing experiments with the following steps: we first mixed
under acetone the powders of SiO2 and CaCO3 in a molar ratio of 1:2, which
and were pretreated at 1 atm and 573 K for 72 h; we second pressed this mixture into
a pellet and degassed it at 1 atm and 1273 K for about 48 h; we third crushed
Ttn + Lrn = 3CaPv (2) finely the pellet under acetone into a powder, which was later stored at 383 K
 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE 279

in a drying oven. The starting material for the high-P synthesizing experiments Results and discussion
was sealed into an hBN capsule. The synthesizing conditions were 6 GPa and
1473 K with a heating time of 6 h. In total we conducted five high-P synthesizing High-P synthesizing of pure Lrn
experiments under nominally identical physical conditions, and obtained four
coherent samples and one dust-like sample. It appeared that some unknown factor The BSE images taken from the polished surfaces of the
played a role in the high-P synthesizing process. coherent samples suggest two coexisting phases: the predomi-
Some portions of the coherent samples from the high-P synthesizing experi- nant phase has a volume percentage higher than ~98%, appears
ments were processed and examined by a scanning electron microscope (Quanta
relatively dark, and attains a grain size of ~10–50 μm; the trace
650 FEG), a confocal micro-Raman system (Renishaw system RM-1000), and
an electron microprobe (JEOL JXA-8100), and confirmed to be pure Lrn. Other phase appears relatively bright and has a grain size of ~1–2 μm.
portions of these samples were slowly ground down to a fine powder, which was 10 electron microprobe analyses were conducted on 10 arbitrarily
used in our thermal expansion and compression experiments. selected grains of the predominant phase, and their average is
The thermal expansion experiments at ambient P were carried out with an Ca2.02(2)Si0.99(1)O4, suggesting a homogeneous composition closely
X’Pert Pro MPD diffractometer system equipped with an Anton Paar HTK-1200N
oven. A Eurotherm temperature controller (Eurotherm 2604) collected with a type S
matching the stoichiometry of Lrn. Ten Raman spectra were
thermocouple, which was checked against the melting point of NaCl, was used collected on 10 arbitrarily selected grains of the predominant
to control the temperature. Other details of the X’Pert Pro MPD diffractometer phase, and they are similar, and in excellent agreement with
system included a Cu target, operation voltage of 40 kV and current of 40 mA. The the Raman data of Lrn reported by Piriou and McMillan (1983)
whole experimental system was carefully checked up to 1273 K by using quartz
and Remy et al. (1997a). It follows that the predominant phase
as an internal standard (Hu et al. 2011). Following the heating and data-collection
procedures reported in Liu et al. (2010), we carried out the high-T experiments up in the coherent samples is Lrn indeed. On the other hand, the
to 1023 K. At every studied T, X‑ray data were collected from 10 to 70° 2θ, with a trace phase is mostly likely a reaction residue (CaO) left from
scanning step length of 0.017° 2θ and a scanning time of 10 s for each scanning step. the formation of Lrn.
As demonstrated by He et al. (2011) and Wang et al. (2012), unit-cell parameters As suggested by our preliminary Raman spectroscopic data,
with high accuracy were readily extracted by a full powder X‑ray pattern refine-
ment with the MDI’s program Jade 6.0 (Material Data, Inc.).
the dust-like sample is distinctly dominated by the γ-Ca2SiO4
The high-P angle dispersive X‑ray diffraction experiments (ambient T) were phase, which could be produced during temperature quench-
conducted with a symmetrical DAC at the beamline X17C, National Synchrotron ing or pressure release. The Ca2SiO4 system is infamous in its
Light Source, Brookhaven National Laboratory. The experimental techniques back-transformation (either partial or complete) of its high-T
were generally identical to those used in our previous studies (e.g., Liu et al.
polymorphs at room P (Fig. 1a), and the operating factors
2011; He et al. 2012; Xiong et al. 2015): a T301 stainless steel plate was used as
the gasket, a 4:1 methanol-ethanol mixture as the pressure medium, and a couple include the maximum synthesizing T, the cooling kinetics, the
of ruby chips used as the pressure marker (the ruby fluorescence method; Mao et grain size, and the release of strains and charge repulsions in
al. 1978). The incident synchrotron radiation beam had a wavelength of 0.409929 the β-Ca2SiO4 phase (Remy et al. 1997a). Our observation made
Å, and its size was ~25 × 20 μm2. Each X‑ray diffraction pattern at certain P was in this study similarly suggests that the back-transformation
collected for 60 min using an online CCD detector, and later processed to generate process is not always escapable for the samples synthesized at
the conventional one-dimension profile using the Fit2D program (Hammersley
1996). Subsequently, the positions of the diffraction peaks were determined with
high P, in agreement with Remy et al. (1995) and Reynard et al.
the Jandel Scientific PeakFit V4.11 program, and the unit-cell parameters were (1997). Anyhow, the first attempt in synthesizing a large amount
obtained by using the UnitcellWin program. of pure Lrn in its P-T stability field (Fig. 1b) has been at least
The first-principles simulations carried out to investigate the compression partially successful, in spite of the complication caused by the
behavior of Lrn were completed with the CASTEP code using density functional
back-transformation process.
theory (Hohenberg and Kohn 1964; Kohn and Sham 1965) and planewave pseudo-
potential technique (Payne et al. 1992). We treated the exchange-correlation It is worthy to point out that the univariant curve of β = αʹL
interaction by both the local density approximation (LDA) (Ceperley and Alder determined by Klement and Cohen (1974) and Hanic et al. (1987)
1980; Perdew and Zunger 1981) and generalized gradient approximation (GGA) may be reasonably extrapolated to ~6 GPa (Fig. 1b). Assuming
with the Perdew-Burker-Ernzerhof functional (Perdew et al. 1996), and we used there was no any back-transformation from the αʹL phase to the
a convergence criterion of 10–6 eV/atom on the total energy in the self-consistent
field calculations. We employed a planewave basis set with a cutoff of 900 eV to
Lrn phase in our synthetic samples, the P-T condition (6 GPa
expand the electronic wave functions, and a norm-conserving pseudopotential to and 1473 K) of our synthesizing experiments does support this
model the ion-electron interaction (Lin et al. 1993; Lee 1995). We sampled the extrapolation. In contrast, the P-T condition (9 GPa and 1350
irreducible Brillouin zone with a 4 × 3 × 2 Monkhorst-Pack grid (Monkhorst and K) for the phase transition from the Lrn phase to the αʹL phase
Pack 1976). The effects of using larger cutoff and k point mesh on the calculated
constrained by the in situ synchrotron X‑ray experiments of
properties were found to be insignificant. The computation cell contained four
Ca2SiO4 molecules (28 atoms), with the initial structure model from Tsurumi et Wang and Weidner (1994) does not. More high-P experimental
al. (1994). The equilibrium lattice parameters and internal coordinates at differ- investigation on this issue seems necessary.
ent pressures were optimized by minimizing the Hellmann-Feynman force on
the atoms and simultaneously matching the stress on the unit cell to the target High-T phase transition and thermal expansion coefficient
stress. These theoretical techniques were used in our previous studies targeting (room P)
the structures and thermodynamics of some silicate minerals (e.g., Deng et al.
2010, 2011; Chang et al. 2013). High-T X‑ray diffraction experiments were conducted up to
Based on the optimized structure with the LDA method, the phonon dis- 1023 K at ambient pressure (Fig. 2). We found that all major
persions and VDoS of Lrn were calculated by diagonalizing the dynamical
matrix whose elements were obtained using density functional perturbation
X‑ray diffraction peaks observed at temperatures below 973 K
theory (Baroni et al. 2001; Refson et al. 2006). The q-vector grid spacing for could be attributed to the Lrn phase (Jost et al. 1977; Yamnova
interpolation was 0.05 Å–1, which represented the average distance between the et al. 2011), and the new peaks appearing from 973 K on could
Monkhorst-Pack q-points used in the dynamical matrix calculations. The phonon be attributed to the αʹL phase (ideally Ca2SiO4; Saalfeld 1975).
dispersions were obtained at the high symmetry points (Z, G, Y, A, B, D, E, C).
As shown in Figure 1a, the stable form of Ca2SiO4 at relatively
The coordinates of these points on the surface of the Brillouin zone were Z = (0
0 ½), G = (0 0 0), Y = (0 ½ 0), A = (–½ ½ 0), B = (–½ 0 0), D = (–½ 0 ½), E = low T is the γ phase; its formation from the metastable Lrn
(–½ ½ ½), and C = (0 ½ ½). phase was not observed in our experiments though. This phase
280 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE

Figure 2. XRD patterns collected at (a) 300 K, (b) 923 K, and (c) 973 K (ambient P). The material underwent the phase transition from the
β phase to the αʹL phase at a temperature between 923 and 973 K. A few weak peaks, as indicated by the asterisks, do not belong to the Lrn phase,
but are attributable to the γ phase, which might have been generated during the sample-grinding process. The relative intensities of these weak
peaks did not change as the experimental T increased, suggesting that the volume proportion of the γ phase did not increase and the potential phase
transition from the metastable Lrn to the γ phase did not actually take place at low temperatures (Fig. 1a).

transition is thus generally sluggish, in good agreement with the than that defined by this study.
observation made by Remy et al. (1997a). The mechanism in If the phase transition T from the metastable Lrn to the αʹL
this phase transition has been regarded as complicated and not phase on heating is assumed to be ~973 K, the extrapolated unit-
fully understood, with the affecting factors presumably including cell parameters of the Lrn phase at this T are then a = 5.5462 Å,
the synthesizing condition, impurity, and grain size of the used b = 6.8058 Å, c = 9.4156 Å, V = 354.87 Å3, and β = 93.345°.
Lrn material, and the details of the heating process (Remy et al. Compared to the unit-cell parameters of the αʹL phase at 973 K
1997a; references therein). (Table 1), which can be translated as a = 5.532 Å, b = 6.864
The phase transition T from the metastable Lrn to the αʹL phase Å, c = 9.443 Å, V = 358.53 Å3, and β = 90° (Saalfeld 1975),
on heating was located between 923 and 973 K by our experi- the relative changes in the unit cells accompanying this phase
ments (Fig. 2). In contrast, it was constrained between 960(5) transition are suddenly a reduction in the a-axis (~0.26%), some
to 988(5) K by in situ high-T Raman spectroscopy (Remy et al. expansions of the b-axis (~0.86%), c-axis (~0.29%), and volume
1997a) and between 984(10) to 1005(10) K by in situ high-T (~1.03%), and a huge jump of the β angle. It concludes that this
powder X‑ray diffraction (Remy et al. 1997b). Taking into ac- phase transition is indeed of first-order character, in agreement
count the differences in the used materials and heating processes with Barnes et al. (1980), Remy et al. (1997a, 1997b), and Liu
in these studies, the agreement is good. et al. (2002).
The variation of the unit-cell parameters of the Lrn phase To derive the volumetric thermal expansion coefficient, the
with T as determined in this study is shown in Figure 3, and it following equations have been fitted with our T-V data
is almost linear for the investigated T interval. This result is in ⎡T ⎤
general agreement with Fukuda et al. (1997), but contradicts VT = V0 exp ⎢⎢ ∫ α T dT ⎥⎥ (3)
Remy et al. (1997b): the data from Remy et al. (1997b) clearly ⎢⎣ 0 ⎥⎦
demonstrated non-linear correlations between the a-axis and T and
(Fig. 3a), between the b-axis and T (Fig. 3b), between the c-axis α T = a0 + a1T + a2T −2 (4)
and T (Fig. 3c), and between the β and T (Fig. 3e). As T increases
from ~300 to 923 K, our data suggested that the a-axis increases where VT, V0, and αT are the high-T volume, room-T volume,
by 0.61(1)%, the b-axis by 0.73(1)%, the c-axis by 1.12(1)%, and and volumetric thermal expansion coefficient at temperature T,
the volume by 2.62(1)% (Table 1); the ratio of the relative thermal respectively. a0, a1, and a2 are the constants obtained in fitting the
expansions along the a-, b-, and c-axes is 1:1.10:1.84, suggesting experimental T-V data. Due to the generally linear relationship ob-
that the c-axis is much more expandable than the other two axes. served between the volume and T (Fig. 3), a1 and a2 should be close
In contrast, the data from Fukuda et al. (1997) suggested that the to 0, so that their values have been fixed as zero in our data-fitting
a-axis increases by 0.61%, the b-axis by 0.65%, the c-axis by process. The obtained parameter is then a0 = 4.24(1)×10–5 K–1.
1.02%, and the volume by 2.45% for a similar T interval (from Similarly, replacing the volume data in the Equations 3 and 4
298 to 918 K). It follows that the volumetric thermal expansion with the axial dimensions, the axial thermal expansion coefficients
constrained by Fukuda et al. (1997) should be generally smaller can be obtained. Our experimental data yield a0 = 1.00(1)×10–5
 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE 281

Table 1. Unit-cell parameters of Lrn and α’L vs. T (ambient P)

T (K) a (Å) b (Å) c (Å) β (°) V (Å3)
Lrn
300 5.5096(4)a 6.7518(4) 9.3059(8) 94.575(7) 345.07(19)
323 5.5101(2) 6.7529(2) 9.3091(4) 94.533(4) 345.31(11)
373 5.5126(2) 6.7567(2) 9.3178(4) 94.447(3) 346.02(9)
423 5.5157(2) 6.7587(3) 9.3223(5) 94.369(4) 346.52(13)
473 5.5183(2) 6.7649(2) 9.3324(3) 94.289(3) 347.41(9)
523 5.5214(2) 6.7681(3) 9.3377(5) 94.191(5) 348.01(13)
573 5.5246(3) 6.7726(3) 9.3476(7) 94.117(6) 348.85(17)
623 5.5278(3) 6.7781(3) 9.3575(7) 94.025(6) 349.74(15)
673 5.5303(3) 6.7801(4) 9.3643(8) 93.934(7) 350.31(17)
723 5.5332(2) 6.7851(3) 9.3731(6) 93.818(5) 351.11(14)
773 5.5351(2) 6.7887(3) 9.3826(6) 93.731(4) 351.81(11)
823 5.5379(2) 6.7946(2) 9.3912(5) 93.626(4) 352.67(9)
873 5.5407(4) 6.7992(5) 9.4031(9) 93.502(6) 353.57(15)
923 5.5431(2) 6.8009(4) 9.4098(8) 93.379(6) 354.11(12)
α’L
973 5.532(3) 20.591(16) 9.443(6) – 1075.6(12)
1023 5.764(4) 21.019(11) 9.855(4) – 1193.9(11)
a
Number in the parentheses represents one standard deviation in the right-
most digit.

The volumetric thermal expansion coefficient of the Lrn

phase determined in this study is compared to the results from
other studies in Table 2. With the only exception of Fukuda et al.
(1997), all other experimental studies have obtained remarkably
similar values: for example, they vary slightly from 3.97×10–5
K–1 to 4.25(14)×10–5 K–1 when T is ~298 K. The reason why Fu-
kuda et al. (1997) does not agree to other experimental studies is
presently unknown. On the other hand, the values obtained with
a quasi-harmonic lattice dynamic calculation done by Swamy
and Dubrovinsky (1997) and with a thermodynamic optimization
method done by Holland and Powell (1998) are significantly
smaller. The value constrained with the thermodynamic optimiza-
tion method done by Swamy and Dubrovinsky (1997), however,
is in extraordinarily good agreement with the values from most
experimental studies.

Isothermal bulk modulus

High-P compression experiments at ambient T were con-
ducted up to ~16 GPa for the Lrn (Fig. 4). No phase transition
was observed, which is in good agreement with Reynard et al.
(1997) and Remy et al. (1997b). The X‑ray diffraction patterns
obtained at P higher than ~10 GPa did not show any apparent
peak-broadening, which was merely a fortuitous situation; ac-
cording to Klotz et al. (2009), our pressure medium could only
maintain a hydrostatic pressure environment up to about 10 GPa.
β
The unit-cell parameters of the Lrn at different pressures are
summarized in Table 3 and shown in Figure 5. All unit-cell pa-
rameters vary non-linearly with P for the investigated P range. As
P increases from 1 atm to 16.25 GPa, the a-axis decreases from
5.510(1) to 5.297(4) Å [by 3.87(1)%], the b-axis decreases from
Figure 3. Variation of the unit-cell parameters of Lrn with T 6.753(2) to 6.529(1) Å [by 3.32(1)%], the c-axis decreases from
(ambient P): (a) the a-axis; (b) the b-axis; (c) the c-axis; (d) the volume; 9.316(1) to 8.931(2) Å [by 4.14(1)%], the volume decreases from
(e) the angle β. Note that lengths of the error bars are generally equal to 345.6(1) to 307.7(4) Å [by 10.9(2)%], and the β-angle increases
or smaller than the symbols. from 94.48(1) to 94.82(2) Å [by 0.36(1)%].
To determine the isothermal bulk modulus, the third-order
K–1 for the a-axis, a0 = 1.21(1)×10–5 K–1 for the b-axis and a0 = Birch-Murnaghan equation of state (BM-EoS; Birch 1947) has
1.77(2)×10–5 K–1 for the c-axis, which are in excellent agree- been fitted with the P-V data of the Lrn by a least-squares method:
ment with those from Remy et al. [1997b; 1.00(10)×10–5 K–1, 5 ⎡ ⎤
3
1.25(10)×10–5 K–1, and 1.79(15)×10–5 K–1, respectively]. P = 3K T f E (1+ 2 f E )2 ⎢1+ (K T' − 4) f E ⎥ (5)
⎢⎣ 2 ⎥⎦
282 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE

Table 2. Volumetric thermal expansion coefficients of Lrn (ambient P)a

T range (K) a0 (10–5) a1 (10–9) Data source Note
300–923 4.24(4) – This study powder XRD
298–984 4.25(14) – Remy et al. (1997b) powder XRD
298–918 2.87(15) 35(4) Fukuda et al. (1997) powder XRD
293, 993b 3.97 – Barnes et al. (1980) powder XRD
298 1.80 – Swamy and Dubrovinsky (1997) QHLDCc
298–923 2.38(3) 17.8(9) Holland and Powell (1998) TOEd
– 4.11(0) 15.4(0) Swamy and Dubrovinsky (1997) TOEd
a
αT = a0 + a1(T – T0) + a2(T – T0)–2, where αT and a0 are in K–1, a1 is in K–2, and a2
(always zero in the case of Lrn) is in K. T0 is set as 298 K.
b
The reported unit-cell parameters at 293 and 993 K were used in the calculation.
c
Quasi-harmonic lattice dynamic calculation done with the GEMIN algorithm.
d
Thermodynamically optimized estimate.

Table 3. Unit-cell parameters of Lrn vs. P (ambient T)

P (GPa) a (Å) b (Å) c (Å) β (°) V (Å3)
0.0001(1) 5.510(1) a
6.753(2) 9.316(1) 94.48(1) 345.6(1)
3.80(10)b 5.446(6) 6.685(6) 9.197(2) 94.57(2) 333.7(1)
5.09 (10) 5.428(5) 6.666(6) 9.163(2) 94.59(4) 330.5(3)
6.60(10) 5.406(3) 6.647(6) 9.125(2) 94.61(2) 326.8(1)
7.61(10) 5.394(7) 6.637(7) 9.101(6) 94.63(3) 324.7(2)
8.68(10) 5.380(6) 6.626(5) 9.078(2) 94.66(2) 322.5(1)
9.45(10) 5.371(5) 6.615(1) 9.061(2) 94.67(2) 320.9(4)
10.39(10) 5.362(6) 6.605(4) 9.041(5) 94.69(5) 319.1(1)
11.35(10) 5.351(4) 6.591(7) 9.021(2) 94.72(2) 317.1(2)
12.30(10) 5.341(5) 6.577(6) 9.003(2) 94.73(2) 315.2(1)
13.49(10) 5.329(4) 6.556(1) 8.980(8) 94.75(3) 312.7(3)
14.33(10) 5.319(4) 6.549(1) 8.965(2) 94.78(2) 311.2(1)
15.35(10) 5.308(2) 6.538(4) 8.947(1) 94.80(1) 309.4(1)
16.25(10) 5.297(4) 6.529(1) 8.931(2) 94.82(2) 307.7(4)
a
The number in the parentheses represents one standard deviation in the
rightmost digit.
b
Uncertainty in the P measurement at high P assumed to be 0.1 GPa.

Figure 5. Variation of the unit-cell parameters of Lrn with P

(ambient T for the experimental data whereas zero T for the simulated
Figure 4. XRD patterns of Lrn collected at 3.80, 7.61, 12.30, and data): (a) the a-axis; (b) the b-axis; (c) the c-axis; (d) the volume; (e) the
16.25 GPa (ambient T). Note that peak broadening at P > 10 GPa was angle β. Note that lengths of the error bars of the experimental data are
not obvious. generally equal to or smaller than the symbols. The curves, both solid
and broken, are drawn according to the derived third-order BM-EoS.
where P is the pressure, KT the isothermal bulk modulus, KTʹ the
first pressure derivative of KT, and fE the Eulerian definition of F ≡ P/[3fE(1 + 2fE)5/2]. Using F, the third-order BM-EoS can be
finite strain, which is [(V0/V)2/3 – 1]/2, respectively. In the Eule- rewritten as:
rian definition of finite strain, V0 is the volume at zero pressure,
F = KT + 3/2KT(KTʹ – 4)fE (6)
whereas V is the volume at high pressure. When KTʹ is set as 4,
the isothermal bulk modulus (KT) of Lrn is 110.8(8) GPa and so that the slope of the line defined by the experimental data
the zero-pressure volume is 345.5(1) Å3. If KTʹ is not fixed, the should be equal to 3/2KT(KTʹ – 4), and the intercept value is the
results of our best data-fitting are KT = 103(2) GPa, KTʹ = 5.4(4), isothermal bulk modulus. Accordingly, a slope of zero means KTʹ
and V0 = 345.6(1) Å3. = 4, a negative slope KTʹ < 4, and a positive slope KTʹ > 4. Figure 3
The quality of the derived third-order BM-EoS for the Lrn clearly suggests that the KTʹ of the Lrn is larger than 4, in agreement
can be evaluated by using the fE-F plot (Fig. 6); F is defined as with our P-V data-fitting results detailed in the previous paragraph.
 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE 283

Table 4. Comparison between experimental and energy-optimized

crystallographic data of Lrn
Parameters Experimentala Calculated by first-principles simulation
LDA R.D. (%)b GGA R.D. (%)b
a (Å) 5.5041 5.4080 –1.75 5.5678 1.16
b (Å) 6.7622 6.6310 –1.94 6.8103 0.71
c (Å) 9.3281 9.1270 –2.16 9.3873 0.63
β (°) 94.172 95.433 1.34 94.466 0.31
V (Å )
3
346.27 325.87 –5.89 354.87 2.48
Ca polyhedron
Ca1-O1 (Å) 2.719 2.682 –1.02 2.752 1.2
Ca1-O2 (Å) 2.671 2.672 0.04 2.685 0.55
Ca1-O3 (Å) 2.459 2.456 -0.09 2.451 –0.32
Ca1-O4 (Å) 2.508 2.482 –1.01 2.513 0.22
Ca2-O1 (Å) 2.255 2.238 –0.77 2.280 1.12
Ca2-O2 (Å) 2.432 2.418 –0.57 2.464 1.31
Ca2-O3 (Å) 2.871 2.856 –0.53 2.907 1.26
Ca2-O4 (Å) 2.735 2.737 0.04 2.774 1.41
Si tetrahedron
Si-O1 (Å) 1.598 1.562 –2.23 1.617 1.21
Si-O2 (Å) 1.533 1.512 –1.33 1.553 1.34
Si-O3 (Å) 1.470 1.457 –0.89 1.486 1.10
Si-O4 (Å) 1.672 1.660 –0.69 1.692 1.22
O1-Si-O2 (°) 109.051 108.092 –0.88 109.891 0.77
O1-Si-O3 (°) 114.907 114.187 –0.62 115.101 0.17
O1-Si-O4 (°) 104.195 103.581 –0.59 105.373 1.13
O2-Si-O3 (°) 112.622 112.701 0.07 113.343 0.64
O2-Si-O4 (°) 99.790 99.840 0.05 99.900 0.11
O3-Si-O4 (°) 114.932 114.161 –0.67 114.403 –0.46
a
Tsurumi et al. (1994).
b
Relative difference.

to the experimentally determined curves, the curves obtained

with the GGA method have slightly steeper slopes, whereas
those obtained with the LDA method have more gentle slopes,
which is normal. In general, the GGA method should yield a
lower KT and the LDA method should yield a higher KT, with
Figure 6. Eulerian strain-normalized pressure (fE-F) plots based on the experimentally constrained KT falling in between (Deng et
the BM-EoS (experimental data only): (a) is for the a-axis; (b) is for the al. 2011). The KT values obtained by fitting Equation 5 with the
b-axis; (c) is for the c-axis; (d) is for the volume. Estimated standard P-V data essentially meet this expectation: KT = 92(2) GPa [KTʹ
deviations were calculated following the method in Heinz and Jeanloz = 4.9(5), V0 = 355.0(1) Å3; GGA] and KT = 123(3) GPa [KTʹ =
(1984). The solid lines are the weighted linear fits through the data. 4.4(9), V0 = 325.94(9) Å3; LDA]. When KTʹ is set as 4, KT = 96(1)
GPa [V0 = 354.8(1) Å3] and KT = 124(1) GPa [V0 = 325.9(6) Å3]
A linearized third-order Birch-Murnaghan equation of state are derived from the data predicted with the GGA method and
(Angel 2000) was used to obtain the parameters of the equations LDA method, respectively. It concludes that our experimentally
of state for the crystallographic axes, yielding: a0 = 5.510(1) Å, constrained KT value should be a good approximation to the real
KT-a = 99(5) GPa and KTʹ-a = 5.8(10) for the a-axis, b0 = 6.753(3) KT value of the Lrn phase.
Å, KT-b = 137(7) GPa and KTʹ-b = 3.0(1) for the b-axis, and c0 = The KT values of the Lrn phase from different studies are
9.316(1) Å, KT-c = 89(2) GPa and KTʹ-c = 5.9(4) for the c-axis. The compared in Table 5, with the most of them ranging from ~96 to
quality of the derived third-order Birch-Murnaghan equation of 124 GPa (with the assumption KTʹ = 4). Clearly, the result from
state for the axes of the Lrn has been evaluated by using the fE-F Remy et al. (1997b) is the one at odd, which can be adequately
plot as well (Fig. 6). explained by their used pressure medium of silicon oil. Accord-
The geometry parameters of the Lrn phase at zero pressure ing to Klotz et al. (2009), silicon oil can maintain a hydrostatic
obtained from the first-principles simulation are summarized in pressure environment at P < 3 GPa only, suggesting the high-P
Table 4. As usual, the unit-cell parameters a, b, and c calculated data in Remy et al. (1997b; 4.6, 10, and 14.7 GPa) were severely
by the GGA method are larger (0.63 to 1.16%), whereas those affected by deviatoric stress.
calculated by the LDA method are smaller (–1.75 to –2.16%),
compared to the experimental determinations by Tsurumi et Heat capacity and entropy
al. (1994). Additionally, the unit-cell volume calculated by the We calculated the phonon dispersions and VDoS of the Lrn
GGA method is larger by 2.48%, whereas that calculated by the phase by the first-principles method using density functional
LDA method is smaller by –5.89%. Summarily, both methods perturbation theory; the optimized crystal structure with the LDA
reproduce well the unit-cell parameters of the Lrn phase. method was used in the calculation since the geometries of the Ca
The unit-cell parameters of the Lrn phase at high pressures polyhedra and Si tetrahedra have been slightly better reproduced
predicted by our first-principles simulation are shown along with the LDA method (Table 4). The dynamical matrices were
with the experimentally obtained data in Figure 5. Compared computed at 22 wave (q) vectors in the Brillouin zone of the
284 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE

Table 5. Isothermal bulk modulus (KT ) of Lrn (ambient P)

KT (GPa) KT’ Method Data source
114(1) 4a Powder XRD This study
103(2) 5.4(4) Powder XRD This study
166(15) 4a Powder XRD Remy et al. (1997b)
124(1) 4a First-principles simulation (LDA) This study
123(3) 4.4(9) First-principles simulation (LDA) This study
96(1) 4a First-principles simulation (GGA) This study
92(2) 4.9(5) First-principles simulation (GGA) This study
119 4a QHLDACb Swamy and
Dubrovinsky (1997)
119 4a Thermodynamic optimization Swamy and
Dubrovinsky (1997)
120 4a Thermodynamic optimization Holland and Powell (1998)
a
KT’ is set as 4.
b
Quasi-harmonic lattice dynamic calculation done with the GEMIN algorithm.
Figure 7. Phonon dispersions and VDoS of Lrn (LDA).
primitive cell, and interpolated to obtain the bulk phonon disper-
sions. Figure 7 shows the predicted dispersion curves along several Table 6. Heat capacity [J/(mol·K)] of Lrn at selected T
symmetry directions and the VDoS. The results of the phonon T (K) Calculated by first-principlesa Experimentalb Experimentalc
CV CP CP CP
spectra were used to compute the internal energy (E) and isochoric
10 0.06 0.17(1)
heat capacity (CV) as functions of temperature. The temperature 20 1.05 1.26(1)
dependence of the E was obtained by the following equation, 30 4.08 4.38(1)
41 9.16 9.56(1)
hω 51 15.72 16.23(2) 14.36 –
E(T ) = Etot + EZP + ∫ F(ω)dω (7)
⎛ hω ⎞⎟ 101 52.91 53.93(4) 53.05 –
exp⎜⎜ ⎟⎟ −1 152 81.59 83.12(6) 81.75 –
⎝⎜ kT ⎟⎠ 202 101.69 103.73(6 102.60 –
253 116.37 118.92(9 117.73 –
with Etot representing the total electronic energy at 0 K, EZP the 303 127.46 130.52(11) 129.29 –
zero point vibrational energy, h the Planck’s constant, k the 354 135.99 139.56(13) –
404 142.61 147.69(15) 145.50
Boltzmann constant, and F(ω) the vibrational density of states. 455 147.80 151.35(16) 151.91
We evaluated the EZP in Equation 7 using the following equation, 505 151.91 157.01(19) 157.20
556 155.20 160.81(21) 161.59
606 157.86 163.98(22) 165.25
EZP = 12 ∫ F(ω)hω dω (8) 657 160.03 166.67(24) 168.34
707 161.83 168.97(26) 170.98
758 163.32 170.98(28) 173.24
Furthermore, we approximated the lattice contribution to the CV 808 164.58 172.75(30) 175.21
with the following equation: 859 165.65 174.32(32) 177.94
2 909 166.56 175.74(34) 178.46
⎛ hω ⎞⎟ ⎛ ⎞
⎜⎜ ⎟ exp⎜⎜ hω ⎟⎟ 950 167.19 176.79(3) 179.56
⎜⎝ kT ⎟⎟⎠ ⎜⎝ kT ⎟⎟⎠ 1000 167.89 177.99(37) 180.79
CV (T ) = k ∫ 2
F(ω)dω (9) a
This work.
⎡ hω 2 ⎤ b
Todd (1951).
⎢( ) −1⎥ c
Coughlin and O’Brien (1957).
⎢⎣ kT ⎥⎦
The CV result calculated in this way is shown in Table 6.
We calculated the isobaric heat capacity (CP) by adding an modulus, and they decrease from ~6% at ~10 K to ~0.06% at
anharmonic effect to the CV obtained from the above calculations, ~202 K, and then increase to ~0.21% at 1000 K. The calculated
using the following equation: CP data were then expressed using the following polynomial
of temperature,
C P = CV + α T 2 K TVT T (10)
C P = k 0 + k1T −0.5 + k 2T −2 + k 3T −3 + k 4T + k5T 2 + k 6T 3 (12)
where αT, KT, and VT were the thermal expansivity, isothermal
where CP was in J/(mol∙K) and T in K. In the fitting procedure,
bulk modulus, and volume at 1 atm and T (K), respectively. For
the data were divided into three different T intervals, 10–50,
the Lrn phase, αT = 4.24(4)×10–5 K-1 and KT = 103(2) GPa [KTʹ =
50–293, and 293–1000 K. The CP equation coefficients for the
5.4(4)]. The temperature dependence of KT has not been experi-
three T intervals are summarized in Table 7.
mentally determined, thus assumed as zero in our calculation.
Our calculated CP values are compared in Figure 8 with the
VT at T K was calculated with the following equation:
heat capacity measurements by Todd (1951; 52.66–296.48 K) and
⎛ T ⎞ (11)
VT = V298 exp⎜⎜⎜ ∫ α T dT ⎟⎟⎟ Coughlin and O’Brien (1957; 406.0–964.6 K). For the T range of
⎝ 298 ⎠ 52.66–296.48 K, our CP data show excellent agreement with those
where V298 = 51.88 cm3/mol (Remy et al. 1997b). The calculated from Todd (1951), with the relative difference quickly decreasing
CP values are listed in Table 6, along with the experimental re- from ~13% at ~51 K to ~1% at ~303 K (Table 6). For the T range of
sults from Todd (1951) and from Coughlin and O’Brien (1957). 406.0–964.6 K, our CP data are also in good agreement with those
Errors of our CP values were propagated from the uncertainty from Coughlin and O’Brien (1957), with the relative difference
in the thermal expansion coefficients and isothermal bulk varying from ~ –2.0% (859 K) to ~1.5% (404 K).
 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE 285

The obtained CP values have been applied to the calculation Table 7. Coefficients of the CP polynomials of Lrn (ambient P)a
of the vibrational entropy at T K using the following equation T = 10–50 K T = 50–293 K T = 293–1000 K
k0 = 8.0(2)E-1b k0 = –3.56(7)E1 k0 = 2.121(1)E2
k4 = –1.6(2)E-1 k4 = 1.13(1)E0 k1 = –9.69(5)E2
T CP k5 = 9.2(3)E-3 k5 = –2.77(9)E-3 k2 = –4.1(3)E6
ST0 = ∫ dT (13)
0 T k6 = 2.7(1)E-6 k3 = 5.20(7)E8
a
CP = k0 + k1T–0.5 + k2T–2 + k3T–3 + k4T + k5T2 + k6T3 [J/(mol∙K)].
The vibrational entropy at 298 K (S0298) calculated from the CP val- b
E-n represents ×10–n.
ues in the T range of 0 to 298 K obtained in this study is 129.8(13)
J/(mol∙K), in good agreement with the only available value derived
from the existing low-T heat capacity measurements, 127.6(8)
J/(mol∙K) (Todd 1951).

Grüneisen parameter
The Grüneisen parameter is often used to describe the pres-
sure dependences of some thermodynamic, elastic, and transport
properties that are crucial in the investigation of the deep Earth.
It is dimensionless, and varies slowly with P and T, and therefore
appears very attractive to geophysicists.
The mode Grüneisen parameters (γi) of the Lrn phase (298 K)
have been calculated according to the following equation
∂ln ν i KT ⎛⎜ ∂ν i ⎞⎟
γi = − = ⎜ ⎟ (14)
∂lnV ν0i ⎜⎝ ∂ P ⎟⎟⎠T
where ν0i and νi are the frequencies of the vibration mode i of the
Lrn phase at room P and high P, respectively (at ambient T). With Figure 8. Isobaric heat capacity of Lrn. Our CV results from the first-
the high-P Raman data for the Lrn phase from Reynard et al. (1997) principles method were combined with our experimentally determined
and our bulk modulus determined in this study, we calculate the volumetric thermal expansion coefficient and isothermal bulk modulus
value of the γi of the 20 Raman vibration modes, and find that it to derive the CP values.
varies from ~0 (peaks 152 and 520 cm–1) to ~1.90 (peak 255 cm–1),
with an arithmetic average of ~0.69(52). The details are summa-
rized in Figure 9. As expected (Gillet et al. 1991; Fujimori et al.
2002), the γi values for the lattice modes are generally larger than
those for the internal modes of the SiO4 tetrahedra.
The mode Grüneisen parameters can be cast into a microscopic
thermal Grüneisen parameter (γth,1) with the following average
scheme,
∑ C .γ V ,i i
(15)
γ th,1 = i

∑C i
V ,i

where CV,i is the contribution of the vibration mode i to the iso-

choric heat capacity. CV,i is estimated on the basis of a harmonic
Einstein oscillator,
2 2
hvi hv hv
CV ,i = k exp i exp i 1 (16)
kT kT kT
Figure 9. Isothermal mode Grüneisen parameters (γi) of Lrn
where h is the Planck’s constant, and k the Boltzmann constant. (ambient T). High-P Raman data used in the calculation are from Reynard
The γth,1 of the Lrn phase attains 0.77(8), about 12% larger than et al. (1997). Although a full assignment of the Raman-active vibrations
the arithmetic average of the mode Grüneisen parameters. of Lrn has not been available, it is generally appropriate to assign the
Alternatively, a macroscopic thermal Grüneisen parameter peaks with νi > ~500 cm–1 as the internal modes of the SiO4 tetrahedra and
the peaks at lower frequencies as the lattice modes (Remy et al. 1997a).
(γth,2) can be calculated from the following equation
αKTVT
γ th,2 = (17) the microscopic isothermal Grüneisen parameter γth,1 and the
CV macroscopic isothermal Grüneisen parameter γth,2 for the Lrn
With all the values of the variables on the right-hand side of phase, which should by no means exist under ideal circumstance.
Equation 17 constrained in this study, we obtained γth,2 = 1.80(4) In the cases of other minerals, the observed difference is usually
for the Lrn phase at the P-T condition of 298 K and 1 atm. no more than ~25% (Hofmeister and Mao 2002; Tang et al. 2014).
Clearly there is a sharp difference (relatively ~133%) between Several factors may account for this prominent difference. First,
286 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE

the pressure medium (a 16:4:1 methanol-ethanol-water mixture) amounts of impurities in the Lrn phase (such as 0.32% iron
used by Reynard et al. (1997) was not able to maintain a fully and aluminum oxides and 0.14% magnesia; Todd 1951) have
hydrostatic pressure environment at P > ~10 GPa (Klotz et al. insignificant influence on the heat capacity and vibrational en-
2009), which should have led to some overestimate in the P tropy. Our first-principles simulation, combined with the newly
measurements (He et al. 2004). Indeed, the νi-P curves of the experimentally determined volumetric thermal expansion coef-
Lrn phase shown by Reynard et al. (1997; their Fig. 7) bend ficient and bulk modulus, provides an independent examination
toward the P-axis. This situation was exacerbated by the fact on the results from the heat capacity measurements carried out
that they only used the data in the P range of 6–16 GPa to derive with the metastably formed material (Todd 1951; Coughlin and
the P dependence of the Raman frequency, which was thus pos- O’Brien 1957). The agreement between the results from these
sibly substantially underestimated. Consequently, all the mode two completely different lines is good (Fig. 8); for example, the
Grüneisen parameters could have been strongly underestimated relative difference between the values of the standard entropy
according to Equation 14. Second, the factor-group analysis is ~1.6% only.
(Piriou and McMillan 1983; Remy et al. 1997a) predicted 84 Lrn has been discovered as tiny inclusions in diamonds
normal modes for the Lrn phase, with 42 Raman-active modes, originating from the deep interior of the Earth (Joswig et al.
39 infrared-active modes, and 3 acoustic modes. Since the 1999; Nasdala et al. 2003; Brenker et al. 2005), which provides
number of the Raman peaks observed in Reynard et al. (1997) a means to explore the P-T conditions of the diamond formation
was limited to 20 only, it is thus possible that the microscopic (or the P-T condition for trapping the Lrn). Accurate estimates
isothermal Grüneisen parameter γth,1 obtained here might have of the volumetric thermal expansion coefficient, bulk modulus
been strongly biased. Third, the polyhedral bulk moduli of the and Grüneisen parameter are critical in constraining the remnant
Ca polyhedra and the Si tetrahedron in the Lrn phase should pressures of the Lrn inclusions (Ye et al. 2001; Gillet et al. 2002;
be significantly different (Hazen and Finger 1979), a different Joswig et al. 2003; Nasdala et al. 2003). With the formation
conversion and average scheme employing the polyhedral bulk T of the diamond host independently estimated from some
moduli, as the one proposed by Hofmeister and Mao (2002), geothermometer or assumed from some typical geotherm, the
might be more appropriate than derived from Equations 14 and formation P of the diamond host can be accurately constrained
15. Fourth, the assumption of a harmonic Einstein oscillator (Barron 2005). According to Kagi et al. (2009), the coexisting
(Eq. 16) might be a poor approximation to the actual chemical of Lrn with other calcium-rich silicates such as Wal and Ttn
bonds in the Lrn phase, and the contribution of an anharmonic can be extremely useful, and capable to constrain the formation
component may be significant (Fujimori et al. 2002). Whatever P without any extra estimate of the formation T, provided the
the actual reasons are, more spectroscopic investigation on the volumetric thermal expansion coefficient, bulk modulus and
Lrn phase is deemed necessary to solve the discrepancy between Grüneisen parameter of the coexisting Wal or Ttn known well.
the microscopic isothermal Grüneisen parameter γth,1 and the
macroscopic isothermal Grüneisen parameter γth,2. 16
Cold Hot Normal
Implications 14 slab slab mantle
Pure Lrn of large quantity has been successfully synthe-
12
sized for the first time in its P-T stability field by using high-P CaPv
experimental technique. Its thermal expansivity has been ac- 10 L + Ttn
rn
P (GPa)

curately constrained by some high-T X‑ray powder diffraction L rn + T tn

experiments (ambient P), and the obtained volumetric thermal 8 Wal
expansion coefficient is not very different from that determined 6
with the Lrn metastably formed at ambient P (Table 2). In com-
diamond
parison, its experimentally constrained bulk modulus, verified 4
graphite
by supplementary first-principles simulation, is much lower than
that obtained from the metastably formed material (Remy et al. 2
1997b), by ~38% (Table 5). However, the DAC experiments in 0
Remy et al. (1997b) used silicon oil as the pressure medium, 273 673 1073 1473 1873 2273
which could have resulted in significant overestimation in the P
values (Klotz et al. 2009) and consequently a nominally larger T (K)
bulk modulus (He et al. 2004). The discrepancy in the bulk
moduli is thus possibly irrelevant to any possible difference in the Figure 10. P-T trajectory for trapping the Lrn inclusion by the
materials used. It follows that the difference in the two materials diamonds from the Kankan district of Guinea (thick curve). The P-T
profiles of the cold slab and hot slab are sketched from Thompson (1992),
synthesized in the stability field of Lrn or metastably generated
and the P-T profile for the normal mantle follows the 1600 K adiabat
at ambient P is too small to have any significant impact on the
for the simplified pyrolite composition in Jackson (1998). Since the Lrn
volumetric thermal expansion coefficients (presumably, the bulk inclusion always coexists with the Ttn phase (Joswig et al. 1999), the
moduli as well). Indeed, both materials were shown as heavily phase boundaries between Wal and Lrn + Ttn (Eq. 1), and between Lrn
twined (Groves 1983; Kim et al. 1992; Wang and Weidner 1994; + Ttn and CaPv (Eq. 2) are plotted to show the stability field of the phase
Remy et al. 1995). assemblage Lrn + Ttn (Gasparik et al. 1994; Sueda et al. 2006). The
Furthermore, the defects (such as twins) and the small well-known phase boundary of graphite and diamond is also sketched.
 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE 287

One exercise has been carried out to constrain the P-T system involving diamond. Canadian Mineralogist, 43, 203–224.
Birch, F. (1947) Finite elastic strain of cubic crystals. Physical Review, 71, 809–924.
condition for trapping the Lrn inclusions by the diamonds Bowen, N.L. (1940) Progressive metamorphism of siliceous limestone and dolomite.
from the Kankan district of Guinea (Fig. 10). Joswig et al. The Journal of Geology, 48, 225–274.
(1999) reported the unit-cell parameters of five Lrn inclusions Brenker, F.E., Vincze, L., Vekemans, B., Nasdala, L., Stachel, T., Vollmer, C,. Kersten,
M., Somogyi, A., Adams, F., Joswig, W., and Harris, J.W. (2005) Detection of a
coexisting with the Ttn phase in four diamonds, which, com- Ca-rich lithology in the Earth’s deep (>300 km) convecting mantle. Earth and
bined with our third-order BM-EoS for Lrn, lead to a remnant Planetary Science Letters, 236, 579–587.
pressure of 1.3(3) GPa. Using Equation 1 in Kagi et al. (2009), Ceperley, D.M., and Alder, B.J. (1980) Ground state of the electron gas by a stochastic
method. Physical Review Letters, 45, 566–569.
and with the thermal expansion and compressibility data for Chang, L., Liu, X., Kojitani, H., and Wang, S. (2013) Vibrational mode analysis and
diamond used by them and those for Lrn from this study, the heat capacity calculation of K2SiSi3O9-wadeite. Physics and Chemistry of Miner-
als, 40, 563–574.
P-T trajectory for the diamonds to trap the Lrn inclusions has Coughlin, J.P., and O’Brien, C.J. (1957) High temperature heat contents of calcium
been calculated as orthosilicate. Journal of Physical Chemistry, 61, 767–769.
Deng, L., Liu, X., Liu, H., and Dong, J. (2010) High-pressure phase relations in the com-
position of albite NaAlSi3O8 constrained by an ab initio and quasi-harmonic Debye
P = 0.0057T + 0.17 (18) model, and their implications. Earth and Planetary Science Letters, 298, 427–433.
Deng, L., Liu, X., Liu, H., and Zhang, Y. (2011) A first-principles study of the phase
where P is in GPa, and T in K (the thick curve in Fig. 10). Clearly, transition from Holl-I to Holl-II in the composition KAlSi3O8. American Mineralo-
gist, 96, 974–982.
this P-T trajectory locates in the stability field of diamond. Its inter- Fujimori, H., Komatsu, H., Ioku, K., Goto, S., and Yoshimura, M. (2002) Anharmonic
section with the P-T curve for Equation 1 indicates a minimum P lattice mode of Ca2SiO4: ultraviolet laser Raman spectroscopy at high temperatures.
Physical Review B, 66, 064306.
value of ~9.9 GPa as the inclusion-trapping pressure. Furthermore, Fukuda, K., Maki, I., and Ito, S. (1997) Anisotropic thermal expansion of β-Ca2SiO4
the T that the Lrn and Ttn inclusions and the diamond host once monoclinic crystal. Journal of the American Ceramic Society, 80, 1595–1598.
experienced is of typical mantle character, suggesting that some Gasparik, T., Wolf, K., and Smith, C.M. (1994) Experimental determination of phase
relations in the CaSiO3 system from 8 to 15 GPa. American Mineralogist, 79,
portions of the deep mantle have distinct CaO-rich composition 1219–1222.
but similar T, compared to the normal mantle. Gillet, P., Richet, P., Guyot, F., and Fiquet, G. (1991) High-temperature thermodynamic
The discovery of Lrn and other calcium-rich silicates in some properties of forsterite. Journal of Geophysical Research, 96, 11805–11816.
Gillet, P., Sautter, V., Harris, J., Reynard, B., Harte, B., and Kunz, M. (2002) Raman
diamonds argues that some portions of the deep interior of the spectroscopic study of garnet inclusions in diamonds from the mantle transition
Earth have compositions substantially different to the normal zone. American Mineralogist, 87, 312–317.
Green, D.H., and Ringwood, A.E. (1967) The stability fields of aluminous pyroxene
pyrolitic mantle (Ringwood 1975; Joswig et al. 1999; Brenker et peridotite and garnet peridotite and their relevance in upper mantle structures. Earth
al. 2005; Zedgenizov et al. 2014). The enriching process of CaO and Planetary Science Letters, 3, 151–160.
to concentration levels high enough to stabilize these calcium- Groves, G.W. (1983) Phase transformations in dicalcium silicate. Journal of Materials
Science, 18, 1615–1624.
rich silicates might have been achieved via deep subduction of Hammersley, J. (1996) Fit2D report. Europe Synchrotron Radiation Facility, Grenoble,
the CaO-rich continental crust material into the mantle (Irifune France.
et al. 1994; Liu et al. 2012a). How this material interacts with Hanic, F., Kamarad, J., Stracelsky, J., and Kapralik (1987) The P-T diagram of Ca2SiO4.
British Ceramic Transactions and Journal, 86, 194–198.
the regular mantle, how the calcium-rich silicates such as Lrn Hazen, R.M., and Finger, L.W. (1979) Bulk modulus-volume relationship for cation-
survive the interaction, what new phases form from the reactions anion polyhedra. Journal of Geophysical Research, 84, 6723–6728.
He, D., Shieh, S.R., and Duffy, T.S. (2004) Strength and equation of state of boron
at the P-T conditions of the mantle are largely unknown (Joachim suboxide from radial X‑ray diffraction in a diamond cell under nonhydrostatic
et al. 2011, 2012). It is believed that the thermodynamic proper- compression. Physical Review B, 70, 184121.
ties of the Lrn phase constrained in this study will shed lights on He, Q., Liu, X., Hu, X., Li, S., and Wang, H. (2011) Solid solution between lead fluor-
apatite and lead fluorvanadate apatite: mixing behavior, Raman feature and thermal
understanding these possible reactions. expansivity. Physics and Chemistry of Minerals, 38, 741–752.
He, Q., Liu, X., Hu, X., Deng, L., Chen, Z., Li, B., and Fei, Y. (2012) Solid solutions
Acknowledgments between lead fluorapatite and lead fluorvanadate apatite: compressibility determined
We thank the constructive comments from two anonymous reviewers and by using a diamond-anvil cell coupled with synchrotron X‑ray diffraction. Physics
the editorial handling from K. Crispin. We are grateful for the suggestions from and Chemistry of Minerals, 39, 219–226.
K. Putrika on an early version of the manuscript. The in situ X‑ray diffraction Heinz, D.L., and Jeanloz, R. (1984) The equation of state of the gold calibration standard.
experiments at high pressures were carried out at the National Synchrotron Light Journal of Applied Physics, 55, 885–893.
Source (NSLS), which is supported by the U.S. Department of Energy, Division Hofmeister, A.M., and Mao, H.K. (2002) Redefinition of the mode Grüneisen parameter
of Materials Sciences and Division of Chemical Sciences under Contract No. for polyatomic substances and thermodynamic implications. Proceedings of the
DE-AC02-76CH00016. The operation of X17C is supported by COMPRES, National Academy of the Sciences, 99, 559–564.
the Consortium for Materials Properties Research in Earth Sciences. This work Hohenberg, P., and Kohn, W. (1964) Inhomogeneous electron gas. Physical Review,
is financially supported by the Natural Science Foundation of China (Grant No. 136, 864–871.
41273072 and 41440015). Holland, T.J.B., and Powell, R. (1998) An internally consistent thermodynamic data set
for phases of petrological interest. Journal of Metamorphic Geology, 16, 309–343.
Hu, X., Liu, X., He, Q., Wang, H., Qin, S., Ren, L., Wu, C., and Chang, L. (2011)
References cited Thermal expansion of andalusite and sillimanite at ambient pressure: a powder
Akaogi, M., Yano, Y., Tejima, M., Iijima, M., and Kojitani, H. (2004) High-pressure X‑ray diffraction study up to 1000 °C. Mineralogical Magazine, 75, 363–374.
transitions of diopside and wollastonite: phase equilibria and thermochemistry of Irifune, T. (1994) Absence of an aluminous phase in the upper part of the Earth’s lower
CaMgSi2O6, CaSiO3, and CaSi2O5-CaTiSiO5. Earth and Planetary Science Letters, mantle. Nature, 370, 131–133.
143-144, 145–156. Irifune, T., Susaki, J., Yagi, T., and Sawamoto, H. (1989) Phase transformations in
Angel, R.J. (2000) Equation of state. In R.M. Hazen and R.T. Downs, Eds., High-Tem- diopside CaMgSi2O6 at pressures up to 25 GPa. Geophysical Research Letters,
perature and High-Pressure Crystal Chemistry, vol. 41, p. 35–60. Reviews in Min- 16, 187–190.
eralogy and Geochemistry, Mineralogical Society of America, Chantilly, Virginia. Irifune, T., Ringwood, A.E., and Hibberson, W.O. (1994) Subduction of continental crust
Barnes, P., Fentiman, C.H., and Jeffery, J.W. (1980) Structurally related dicalcium silicate and terrigenous and pelagic sediments: an experimental study. Earth and Planetary
phases. Acta Crystallographica A, 36, 353–356. Science Letters, 117, 101–110.
Baroni, S., de Gironcoli, S., Dal Corso, A., and Giannozzi, P. (2001) Phonons and Jackson, I. (1998) Elasticity, composition and temperature of the Earth’s lower mantle:
related crystal properties from density-functional perturbation theory. Reviews of a reappraisal. Geophysical Journal International, 134, 291–311.
Modern Physics, 73, 515–562. Joachim, B., Gardés, E., Abart, R., and Heinrich, W. (2011) Experimental growth of
Barron, L.M. (2005) A linear model and topography for the host-inclusion mineral åkermanite reaction rims between wollastonite and monticellite: evidence for vol-
288 XIONG ET AL.: THERMODYNAMIC PROPERTIES OF LARNITE

ume diffusion control. Contributions to Mineralogy and Petrology, 161, 389–399. approximations for many-electron systems. Physical Review B, 23, 5048–5079.
Joachim, B., Gardés, E., Velickov, B., Abart, R., and Heinrich, W. (2012) Experimental Perdew, J.P., Burke, K., and Ernzerhof, M. (1996) Generalized gradient approximation
growth of diopside + merwinite reaction rims: the effect of water on microstructure made simple. Physical Review Letters, 77, 3865–3868.
development. American Mineralogist, 97, 220–230. Piriou, B., and McMillan, P. (1983) The high-frequency vibrational spectra of vitreous
Jost, K.H., Ziemer, B., and Seydel, R. (1977) Redetermination of the structure of and crystalline orthosilicates. American Mineralogist, 68, 426–443.
β-dicalcium silicate. Acta Crystallographica, B33, 1696–1700. Refson, K., Tulip, P.R., and Clark, S.J. (2006) Variational density-functional perturba-
Joswig, W., Stachel, T.H., Harris, J.W., Baur, W.H., and Brey, G.P. (1999) New Ca- tion theory for dielectrics and lattice dynamics. Physical Review B, 73, 155114.
silicate inclusions in diamonds-tracer from the lower mantle. Earth and Planetary Remy, C., Guyot, F., and Madon, M. (1995) High pressure polymorphism of dicalcium
Science Letters, 173, 1–6. Ca2SiO4. A transmission electron microprobe study. Physics and Chemistry of
Joswig, W., Paulus, E.F., Winkler, B., and Milman, V. (2003) The crystal structure of Minerals, 22, 419–427.
CaSiO3-walstromite, a special isomorph of wollastonite-II. Zeitschrift für Kristal- Remy, C., Reynard, B., and Madon, M. (1997a) Raman spectroscopic investigations of
lographie, 218, 811–818. dicalcium silicate: polymorphs and high-temperature phase transformations. Journal
Kagi, H., Odake, S., Fukura, S., and Zedgenizov, D.A. (2009) Raman spectroscopic of the American Ceramic Society, 80, 413–423.
estimation of depth of diamond origin: technical developments and the application. Remy, C., Andrault, D., and Madon, M. (1997b) High-temperature, high-pressure X‑ray
Russian Geology and Geophysics, 50, 1183–1187. investigation of dicalcium silicate. Journal of the American Ceramic Society, 80,
Kanzaki, M., Stebbins, J.F., and Xue, X. (1991) Characterization of quenched high 851–860.
pressure phases in CaSiO3 system by XRD and 29Si NMR. Geophysical Research Reynard, B., Remy, C., and Takir, F. (1997) High-pressure Raman spectroscopic study
Letters, 18, 463–466. of Mn2GeO4, Ca2GeO4, Ca2SiO4, and CaMgGeO4 olivines. Physics and Chemistry
Kim, Y.J., Nettleship, I., and Kriven, W.M. (1992) Phase transformations in dicalcium of Minerals, 24, 77–84.
silicate: II, TEM studies of crystallography, microstructure and mechanisms. Journal Ringwood, A.E. (1975) Composition and Petrology of the Earth’s Mantle. McGraw-
of the American Ceramic Society, 75, 2407–2419. Hill, New York.
Klement, W. Jr., and Cohen, L.H. (1974) Determination of the β → αʹ transition in Rodrigues, F.A. (2003) Synthesis of chemically and structurally modified dicalcium
Ca2SiO4 to 7 kbar. Cement and Concrete Research, 4, 939–943. silicate. Cement and Concrete Research, 33, 823–827.
Klotz, S., Chervin, J.C., Munsch, P., and Le Marchand, G. (2009) Hydrostatic limits of Saalfeld, H. (1975) X‑ray investigation of single crystals of β-Ca2SiO4 (larnite) at high
11 pressure transmitting media. Journal of Physics D: Applied Physics, 42, 075413. temperatures. American Mineralogist, 60, 824–827.
Kohn, W., and Sham, L.J. (1965) Self-consistent equations including exchange and Sueda, Y., Irifune, T., Yamada, A., Inoue, T., Liu, X., and Funakoshi, K. (2006) The phase
correlation effects. Physical Review, 140, 1133–1138. boundary between CaSiO3 perovskite and Ca2SiO4 + CaSi2O5 determined by in situ
Lai, G.C., Nojiri, T., and Nakano, K.I. (1992) Studies of the stability of β-Ca2SiO4 doped X‑ray observations. Geophysical Research Letters, 33, L10307.
by minor ions. Cement and Concrete Research, 22, 743–754. Swamy, V., and Dubrovinsky, L.S. (1997) Thermodynamic data for the phases in the
Lee, M.H. (1995) Advanced pseudopotentials for large scale electronic structure calcula- CaSiO3 system. Geochimica et Cosmochimica Acta, 61, 1181–1191.
tions. PhD thesis, University of Cambridge, U.K. Takahashi, E. (1986) Melting of a dry peridotite KLB-1 up to 14 GPa: implications on the
Lin, J.S., Qteish, A., Payne, M.C., and Heine, V. (1993) Optimized and transferable origin of peridotitic upper mantle. Journal of Geophysical Research, 91, 9367–9382.
nonlocal separable ab initio pseudopotentials. Physical Review B, 47, 4174–4180. Tang, J., Liu, X., Xiong, Z., He, Q., Shieh, R.S. and Wang, H. (2014) High temperature
Liu, L. (1978) High pressure Ca2SiO4, the silicate K2NiF4-isotype with crystalchemical X‑ray diffraction, DSC-TGA, polarized FTIR and high pressure Raman spectros-
and geophysical implications. Physics and Chemistry of Minerals, 3, 291–299. copy studies on euclase. Bulletin of Mineralogy, Petrology and Geochemistry,
Liu, L., and Ringwood, A.E. (1975) Synthesis of a perovskite-type polymorph of CaSiO3. 33, 289–298.
Earth and Planetary Science Letters, 28, 209–211. Thompson, A.B. (1992) Water in the Earth’s upper mantle. Nature, 358, 295–302.
Liu, J., Duan, C.G., Mei, W.N., Smith, R.W., and Hardy, J.R. (2002) Polymorphous trans- Tilley, C.E. (1929) On larnite (calcium orthosilicate, a new mineral) and its associated
formations in alkaline-earth silicates. Journal of Chemical Physics, 116, 3864–3869. minerals from the limestone contact-zone of Scawt Hill, Co. Antrim. Mineralogical
Liu, X., He, Q., Wang, H., Fleet, M.E., and Hu, X. (2010) Thermal expansion of kyanite Magazine, 22, 77–86.
at ambient pressure: an X‑ray powder diffraction study up to 1000°C. Geoscience ——— (1947) The gabbro-limestone contact of Camas Mor, Muck, Inverness-shire.
Frontiers, 1, 91–97. Bulletin de la Commission Geologique de Finlande, 140, 97–106.
Liu, X., Shieh, S.R., Fleet, M.E., Zhang, L., and He, Q. (2011) Equation of state of ——— (1951) A note on the progressive metamorphism of siliceous limestones and
carbonated hydroxylapatite at ambient temperature up to 10 GPa: significance of dolomites. Geological Magazine, 88, 175–178.
carbonate. American Mineralogist, 96, 74–80. Todd, S.S. (1951) Low-temperature heat capacities and entropies at 298.16 °K of
Liu, X., Ohfuji, H., Nishiyama, N., He, Q., Sanehira, T., and Irifune, T. (2012a) High-P crystalline calcium orthosilicate, zinc orthosilicate and tricalcium silicate. Journal
behavior of anorthite composition and some phase relations of the CaO-Al2O3-SiO2 of the American Chemical Society, 73, 3277–3278.
system to the lower mantle of the Earth, and their geophysical implications. Journal Tsurumi, T., Hirano, Y., Kato, H., Kamiya, T., and Daimon, M. (1994) Crystal structure
of Geophysical Research, 117, B09205. and hydration of belite. Ceramic Transactions, 40, 19–25.
Liu, X., Chen, J., Tang, J., He, Q., Li, S., Peng, F., He, D., Zhang, L., and Fei, Y. (2012b) Wang, Y., and Weidner, D.J. (1994) Thermoelasticity of CaSiO3 perovskite and implica-
A large volume cubic press with a pressure-generating capability up to about 10 tions for the lower mantle. Geophysical Research Letters, 21, 895–898.
GPa. High Pressure Research, 40, 239–254. Wang, S., Liu, X., Fei, Y., He, Q., and Wang, H. (2012) In situ high-temperature powder
Liu, X., Wang, S., He, Q., Chen, J., Wang, H., Li, S., Peng, F., Zhang, L., and Fei, Y. X‑ray diffraction study on the spinel solid solutions (Mg1–xMnx)Cr2O4. Physics and
(2012c) Thermal elastic behavior of CaSiO3-walstromite: a powder X‑ray diffraction Chemistry of Minerals, 39, 189–198.
study up to 900 °C. American Mineralogist, 97, 262–267. Xiong, Z., Liu, X., Shieh, S.R., Wang, F., Wu, X., Hong, X., and Shi, Y. (2015) Equation
Mao, H.K., Yagi, T., and Bell, P.M. (1977) Mineralogy of the Earth’s deep mantle: of state of a synthetic ulvöspinel, (Fe1.94Ti0.03)Ti1.00O4.00, at ambient temperature.
quenching experiments on mineral compositions at high pressure and temperature. Physics and Chemistry of Minerals, 42, 171–177.
Carnegie Institution of Washington Yearbook, 76, 502–504. Yamnova, N.A., Zubkova, N.V., Eremin, N.N., Zadov, A.E., and Gazeev, V.M. (2011)
Mao, H.K., Bell, P.M., Shaner, J.W., and Steinberg, D.J. (1978) Specific volume mea- Crystal structure of larnite β-Ca2SiO4 and specific features of polymorphic transitions
surements of Cu, Mo, Pt, and Au and calibration of ruby R1 fluorescence pressure in dicalcium orthosilicate. Crystallography Reports, 56, 210–220.
gauge for 0.006 to 1 Mbar. Journal of Applied Physics, 49, 3276–3283. Yannaquis, N., and Guinier, A. (1959) La transition polymorphique β-γ de l’orthosilicate
Mason, B. (1957) Larnite, scawtite, and hydrogrossular from Tokatoka, New Zealand. de calcium. Bulletin de la Societe Francaise de Mineralogie et de Cristallographie,
American Mineralogist, 42, 379–392. 82, 126–136.
Monkhorst, H.J., and Pack, J.D. (1976) Special points for Brillouin-zone integrations. Ye, K., Liou, J.B., Cong, B., and Maruyama, S. (2001) Overpressures induced by coesite-
Physical Review B, 13, 5188–5192. quartz transition in zircon. American Mineralogist, 86, 1151–1155.
Nasdala, L., Brenker, F.E., Glinnemann, J., Hofmeister, W., Gasparik, T., Harris, J.W., Zedgenizov, D.A., Shatskiy, A., Ragozin, A.L., Kagi, H., and Shatsky, V.S. (2014)
Stachel, T., and Reese, I. (2003) Spectroscopic 2D-tomography: residual pressure Merwinite in diamond from Sao Luiz, Brazil: a new mineral of the Ca-rich mantle
and strain around mineral inclusions in diamonds. European Journal of Mineral- environment. American Mineralogist, 99, 547–550.
ogy, 15, 931–935.
Payne, M.C., Teter, M.P., Allan, D.C., Arias, T.A., and Joannopoulos, J.D. (1992) Iterative
minimization techniques for ab initio total-energy calculations: molecular dynamics Manuscript received May 17, 2015
and conjugate gradients. Review of Modern Physics, 64, 1045–1097. Manuscript accepted September 1, 2015
Perdew, J.P., and Zunger, A. (1981) Self-interaction correction to density-functional Manuscript handled by Katherine Crispin