Rev. High Pressure Sci. Technol., Vol. 7 (1998) 9•`11

A Temperature-Pressure Calibration Grid for Multiauvil Experriments Based on

Phase Relations in the System CaO-MgO-SiO2

T. GasparikCenter for High Pressure Research and Department of Geosciences, State University of New York at Stony Brook,

Stony Brook, New York 11794, USA

The subsolidus phase boundaries and melting curves in the system CaO-MgO-SiO2 produce a grid of sufficient

density and coverage at temperatures up to 2600•Ž and pressures up to 25 GPa to be suitable as a calibration

grid for multianvil experiments. The relative positions of these phase boundaries have been determined with high

precision in high-pressures experiments using the same multianvil apparatus, sample assembly, experimental

procedures and calibrations to maximize the internal consistency of the results. The grid can be easily adjusted

in the future to reflect a consensus based on new advances in our understanding of the absolute temperatures and

pressures for selected calibration points.

[calibration, phase relations, high-pressure experiments, multianvil press, system CaO-MgO-SiO2]

1. Introduction

In experiments conducted with solid-media high-pres-

sure devices, the relationships between the reading of the

thermocouple emf and the sample temperature and between

the reading of the pressure gauge and the sample pressure are

unique for every apparatus and sample assembly, and have

to be calibrated. A good calibration should assure that

the experimental temperatures and pressures are independent

of the experimental conditions, including the apparatus,

sample assembly, laboratory or the scientist. This can be

achieved if the same calibration grid is used in all high-

pressure laboratories, and the density of the calibration points

and the coverage of the entire temperature-pressure space

explorable in the experiments is sufficient for carrying out

a detailed calibration. A grid based on phase relations in the

system CaO-MgO-SiO2 appears to be particularly suitable for

calibration. The grid was determined in a large number of

experiments over an extended period of time using the

same multianvil apparatus, sample assembly, experimental

procedures and calibrations to maximize the internal consis

tency of the data. This emphasis allowed to locate, with high

precision, the relative positions of many melting curves and

subsolidus phase boundaries, thus producing a grid with suffi

cient density and coverage at temperatures up to 2600•Ž and

pressures up to 25 GPa.

2. Multianvil Apparatus and Sample Assembly

The experiments were carried out using the split-sphere an-

vil apparatus at the Stony Brook High Pressure Laboratory

(USSA-2000). The apparatus consists of a 2000-ton uniaxial

press with a two-stage anvil system capable of achieving pres

sures up to 25 GPa at temperatures in excess of 2600•Ž . The

first stage is a steel sphere split in six parts, which enclose a

cubic cavity holding the second stage. The second stage con

sists of 8 tungsten carbide cubes separated by pyrophyllite gas

kets, teflon back-up gaskets and balsa wood spacers. The cubes

are truncated at the corners and enclose an octahedral cavity

which holds the sample assembly. The sample assembly was

made of an MgO octahedron with the edge length of 10 mm .

A lanthanum chromite sleeve was used as the resistance heater.

Temperature was measured by a W3%Re vs. W25% Re ther

mocouple introduced axially. Additional details were re-

ported in [1]. This assembly was used in two configurations:

the 10/5 assembly for the experiments at 7-16.5 GPa [1], and

the 10/4 assembly for 16-23 GPa [2]. The 10/5 assembly

uses Kennametal Grade K313 tungsten carbide cubes with 5

mm truncation edge length and pyrophyllite gaskets 2.4 •~ 3.5

mm in cross section, while the 10/4 assembly uses Toshiba

Tungaloy Grade F tungsten carbide cubes with 4 mm trunca

tion edge length and gaskets 3.4 •~ 3.4 mm.

3. Temperature calibration

The relationship between the thermocouple electromotive

force (emf) and the measured temperature is based on the 1

bar calibration provided by the manufacturer (Engelhard).

This calibration is typically performed to 1800•Ž and in

cludes extrapolation to 2300•Ž. Further extrapolation can be

carried out by the user for experiments at higher temperatures.

The pressure effect on the thermocouple emf is not known

and, therefore, not included. The temperature (emf) is con-

trolled during an experiment by a Eurotherm controller, which

has a limit set by the manufacturer at 39.99 mV, approxi-

mately corresponding to 2380•Ž. Experiments at higher tem-

peratures are rarely needed or performed, because the

temperatures in such experiments are not controlled and are

based on extrapolations far beyond the calibrated range. An

electronic ice point is used at the cold junction. The contact

at the hot junction is provided by a rhenium sample capsule.

The, fluctuations around the set point of the temperature con-

troller are usually trivial with respect to other uncertainties.

The sample enclosed in a metal capsule is placed in the

assembly slightly off center, so that the hot spot is in the hot

end of the sample. Calibration of the temperature distribution

in the sample was performed at the nominal temperatures of

1400, 1600 and 1700•Ž using the compositions of two

coexisting pyroxenes on the join Mg2Si2O6,-CaMgSi2O6 [1]. It

was found that the temperature controlled by the thermocouple

10

Fig. 1. Schematic cross-section of an experimental charge in

a rhenium capsule. Solid isotherms show the tempera-ture distribution in a standard-size sample, dashed rectangle

outlines the distribution in a reduced-size sample. After [9].

(nominal T) was approximately the same as the temperature

in the center of the sample. The temperature in the hot spot

was about 50•Ž higher than the nominal temperature, and it

was 150•Ž lower at the cold end. The resulting temperature

distribution shown in Fig. 1 can be used to estimate the tem-

perature at various levels in the sample. Due to the low

temperature gradient, the temperature estimates around the

hot spot are more reliable than the estimates from the cold

end. Hence, for example, the best constraints on melting

temperatures are obtained from experiments showing minimal

melting. The reproducibility of the experimental temperatures

around the hot spot was estimated at •}30•Ž [1]. In some ex-

periments, particularly with Na or H2O, chemical gradients can

develop in response to the temperature gradient [3,4]. If this

is deemed unsuitable with respect to the intended goals, the

simplest solution is to decrease the size of the sample, as i

ndicated in Fig. 1, thus reducing the total temperature

range in the sample from 200•Ž to 50•Ž or less .

4. Pressure Calibration

Phase transitions in metals (Bi, Ba, Pb) or semiconductors

(ZnTe, ZnS, GaAs, GaP) are commonly used for the pres-

sure calibration at room temperature . However, such calibra-

tion is of a limited use in the high-temperature experiments,

because the assemblies are different and the temperature ef-

fect on the sample pressures at high temperatures can be a

complex function of temperature with little apparent relation

to the room temperature calibration. The pressure calibration

of the 10/5 assembly at high temperatures (Fig. 2) is based

on 3 univariant transitions [1]. Reversals of the coesite to

stishovite transformation were carried out at 1200•Ž (9 .3

GPa) and 1400•Ž (9.5 GPa). Although, the original calibration

by [5] turned out to be incorrect and was replaced later by

[6], the change in the investigated temperature range was

Fig. 2. Pressure calibration of the 10 mm assembly at high

temperatures in the 10/5 and 10/4 configurations. After [1, 2].

negligible. The transition from forsterite to beta phase

(Mg2SiO4) was used at 1400•Ž (14.3 GPa) and 1600•Ž (15.2

GPa), based on [7]. The transition from enstatite (MgSiO) to

beta phase and stishovite was observed at 1400•Ž, and asigned

to 16.5 GPa according to [8]. The pressure calibration of the

10/4 assembly (Fig. 2) was anchored at 16.6 GPa by the

observation of the transition from enstatite to beta phase

and stishovite at 1500•Ž [2]. The formation of MgSiO,

perovskite from majorite was observed at 2000•Ž and later

asigned to 21.5 GPa [9]. The transition from beta phase to

perovskite and periclase was observed at 2000•Ž and the cor-

responding reading of the pressure gauge was asigned to 22.4

GPa [2]. Later experiments [10] indicated that the formation

of perovskite from ilmenite at 1800•Ž occurred at about the

same gauge pressure as previously found at 2000°C [2], de-

spite the well documented negative slope of the reaction [11].

Hence, the efficiency of the pressure generation decreased with

temperature between 1800 and 2000•Ž. It was assumed that

this temperature dependence of the pressure calibration was

limited only to the maximum pressures and the pressure cali-

bration was independent of temperature at lower pressures.

The reproducibility of the experimental pressures was esti-

mated at •}0.3 GPa [1].

5. Calibration grid based on the system CaO-MgO-SiO2

Fig. 3 shows the proposed calibration grid based on the phase

relations in the system CaO-MgO-SiO2. They were determined

during a systematic investigation of the phase relations relevant

to the Earth's mantle, which is the primary focus of research

for many multianvil experiments. The most important among

the simple chemical systems for the Earth's mantle is the sys-

11

Fig. 3. Temperature-pressure phase diagram showing the pro-posed calibration grid. Symbols: Bt, Mg2SiO4 beta phase; CaPv, C

aSiO3 perovskite; CEn, clinoenstatite; Cpx, clinopyroxene; Cs, coesite; Di, diopside; Fo, forsterite; Ga, garnet; Ii, MgSiO3 ilmenite; L, liquid; MgPv, MgSiO3 perovskite; Mj3, majorite; OEn, orthoenstatite; Opx, orthopyroxene; Pc, periclase; Pv, perovskite; Sp, Mg2SiO4 spinel; St, stishovite.

tem MgO-SiO2, which constitutes over 80% of the mantle composition. Building on the early pioneering works [7, 8, 11], the phase diagram for the system MgO-SiO2 was finalized in a series of experimental studies [1, 2, 9,12-16]'. The recom-mended temperature-pressure conditions for selected univariant

points are summarized in Table 1. While the phase relations in the system MgO-SiO2 have been used for calibration in most multianvil studies, the addition of CaO increases the com-

plexity of the calibration grid to allow calibration to a much greater detail. The phase diagram for the CaSiO3 system was reported by [17], and the experimental study of the join Mg2Si2O6 CaMgSi2O6 [9] completed the phase diagram for the system CaO-MgO-SiO2. The resulting calibration grid (Fig. 3) can be constructed using the parameters and information summarized in [9]. For the calibration at lower temperatures, the grid can be improved by including the melting curves of Mg(OH)2 [10], Na2Mg2Si2O, and Na2MgSiO4 [18]. Additional phase relations fully consistent with the calibration grid in Fig. 3 were reported by [19-21]. The proposed calibration

grid can be easily adjusted in the future to reflect a consen-sus based on new advances in our understanding and knowl-edge of the absolute temperatures and pressures for selected

calibration points. Such absolute calibration points could result from the correct identification of the phase transitions responsible for the discontinuities in the seismic velocity pro-files of the Earth's mantle.

Acknowledgements The high-pressure experiments reported in this paper were

performed in the Stony Brook High Pressure Laboratory which is jointly supported by the National Science Foundation Sci-ence and Technology Center for High Pressure Research (EAR 89-20239) and the State University of New York at Stony Brook.

References[1] T. Gasparik, Contrib. Mineral. Petrol., 102, 389 (1989).[2] T. Gasparik, J. Geophys. Res., 95, 15751 (1990).[3] C. E. Lesher, D. Walker, J Geophys. Res., 93, 10295 (1988).[4] T. Gasparik, M.1. Drake, Earth Planet. Sci. Lett.,134, 307 (1995).[5] T. Yagi, S. Akimoto, Tectonophysics, 35, 259 (1976).[6] J. Zhang, B. Li, W. Utsumi, R. C. Liebermann, Phys. Chem. Minerals, 23,1(1996).[7] T. Katsura, E. Ito, J. Geophys. Res., 94, 15663 (1989).[8] H. Sawamoto, In: M. H. Manghnani, Y. Syono (eds.), High-Pressure Research in Mineral Physics, Terrapub., Tokyo, 209 (1987).[9] T. Gasparik, Contrib. Mineral. Petrol., 124, 139 (1996).[10] T. Gasparik, J. Geophys. Res., 98,4287 (1993).[11] E. Ito, E. Takahashi, J. Geophys. Res., 94, 10637 (1989).[12] R. E. G. Pacalo, T. Gasparik, J. Geophys. Res., 95, 15858 (1990).[13] D. C. Presnall, T. Gasparik, J. Geophys. Res., 95, 15771 (1990).[14] T. Gasparik, J. Geophys. Res., 97, 15181 (1992).[15] D. C. Presnall, M. J. Walter, J. Geophys. Res., 98, 19777 (1993).[16] J. Zhang, R. C. Liebermann, T Gasparik, C. T. Herzberg, J. Geophys. Res., 98, 19785 (1993).[17] T. Gasparik, K. Wolf, C. M. Smith, Am. Mineral., 79, 1219 (1994).[18] T. Gasparik, Yu. A. Litvin, Eur. J. Mineral., 9, 311(1997).[19] T. Gasparik, Contrib. Mineral. Petrol., 111, 283 (1992).[20] T. Gasparik, High Temp. High Pres., 25, 245 (1993).[21] T. Gasparik, Phys. Chem. Minerals., 23, 476 (1996).