American Mineralogist, Volume 89, pages 1553–1556, 2004
Piston-cylinder calibration at 400 to 500 MPa: A comparison of using water solubility in
albite melt and NaCl melting
DON R. BAKER*
Earth and Planetary Sciences, McGill University, 3450 rue University, Montréal, QC Canada H3A 2A7
ABSTRACT
A crushable alumina-pyrex-NaCl solid-media assembly for the piston-cylinder apparatus was
calibrated by NaCl melting at 500 MPa, 920 °C, and used to assess a new calibration technique based
upon the solubility of water in albite melt at 400 to 500 MPa, 800 and 1200 °C. Use of the “difference
from 100” technique to measure water solubility with the electron microprobe produces a calibration
accuracy of ± 25 MPa. Both calibration techniques agree and demonstrate that the real pressure is
50 MPa greater than the nominal pressure. The water solubility in albite calibration is advantageous
because it can be used to demonstrate that pressure, temperature, and time have no effect on the pressure calibration. Additionally, a single capsule containing albite + water can act as an in situ pressure
monitor in solid-media assemblies containing other experiments. These calibrations demonstrate that
the piston-cylinder apparatus can be used reliably at middle-to-upper crustal conditions for long-duration experiments.
INTRODUCTION
The invention of the piston-cylinder apparatus by Boyd and
England (1960) provided us with an easy-to-use, reliable tool
to investigate processes occurring in the deep crust and upper
mantle. Routine use of the piston cylinder at pressures between
800 MPa and 5.0 GPa in laboratories around the world has resulted in thousands of scientific contributions that have shaped
our understanding of igneous petrogenesis and high-temperature
geochemistry in the lower crust and upper mantle. However,
these machines are not routinely used at lower pressures. At
pressures below 800 MPa, the internally heated pressure vessel
is commonly the preferred apparatus for high-temperature geochemical studies (Holloway and Wood 1988), but these vessels
are expensive and difficult to operate and maintain. In comparison to most internally heated pressure vessels, the piston-cylinder
apparatus has the advantage of more-rapid heating and cooling;
this rapidity is important in kinetic studies (e.g., H2O and CO2
diffusion) and equilibrium studies involving low-viscosity melts
that may not quench to a glass unless cooled very rapidly. Here
I show that the piston-cylinder apparatus can be calibrated and
used reliably at pressures as low as 400 MPa and temperatures
from 800 to 1200 °C, making them useful machines for the study
of processes occurring in the middle-to-upper crust.
The principle behind the piston-cylinder is that a moderate
oil pressure, at most a few hundred megapascals, exerts force on
a large diameter piston that then pushes a small diameter piston.
The small piston compresses a solid-media assembly inside a
cylindrical pressure vessel to high pressure. The solid-media
assembly is composed of nested cylindrical bushings, and the
sample is situated at the center of the assembly (Fig. 1). One of
the bushings is made of graphite through which a current runs
that can heat the sample to temperatures in excess of 2000 °C;
* E-mail: [email protected]
0003-004X/04/0010–1553$05.00
because of the small thermal mass of the assembly, it heats and
quenches rapidly. The nominal pressure on the sample is determined by the ratio of the piston areas and the pressure applied
to the large piston.
However, it is well known that the sample pressure is typically
different from the nominal pressure. The discrepancy between
the nominal and actual pressure is attributed to friction effects
between the various portions of the assembly, between the assembly and the wall of the pressure vessel, and between the piston
and the wall of the pressure vessel; the extent of these effects
depends upon the pressurization and heating history (e.g., Boyd
and England 1960; Johannes et al. 1971; Mirwald et al. 1975). In
addition, differential loading due to bushings of different strength
in the assembly also can cause a discrepancy between the nominal
and actual pressure on the sample and create pressure differentials
across the assembly (Mirwald et al. 1975).
The value of the pressure correction necessary to convert from
the nominal to the real pressure is determined by comparison
of the nominal pressure of a selected phase transition measured
in the piston-cylinder with the actual pressure of the transition.
The actual pressure of the transition is measured in another
type of apparatus, ideally an internally heated pressure vessel
(cf., Holland 1980). Commonly used transitions are the albite
breakdown reaction (e.g., Johannes et al. 1971; Holland 1980),
the melting points of metals (e.g., Mirwald et al. 1975), and of
alkali halides (e.g., Bohlen 1984). Each type of phase transition
provides a unique calibration point in pressure and temperature
space. Importantly, none of these phase transitions allow us to
calibrate the piston-cylinder apparatus isobarically at multiple
temperatures, thus possible temperature effects on the pressure
calibration remain largely unknown.
In order to minimize, or even eliminate, measurable differences between the nominal and sample pressures, the solid-media
assemblies used inside the piston-cylinder apparatus have been
1553
1554
BAKER: PISTON–CYLINDER CALIBRATION USING WATER SOLUBILITY IN ALBITE
Thermocouple
Alumina ceramic
Tool steel
Pyrex glass
NaCl
Graphite
Crushable alumina
Samples in noble metal
capsules
Pb foil
Pyrex glass
19.1 mm
FIGURE 1. 19.1 mm crushable alumina-pyrex-NaCl assembly used
in these experiments.
modified by the replacement of bushings originally made of
boron nitride, pyrophyllite, and talc by ones composed of NaCl,
or other salts or carbonates (Johannes et al. 1971; Mirwald et
al. 1975, Boettcher et al. 1981; Bohlen 1984; Rosenbaum et al.
1994). Because the low-pressure limit of the piston-cylinder apparatus is determined by the strength of the materials comprising
the solid-media assembly, the use of NaCl bushings allows the
piston-cylinder apparatus to reach pressures as low as 500 MPa
(Bohlen 1984). However, the maximum sample temperature
is limited by the NaCl solidus at the pressure of interest (e.g.,
918 °C at 500 MPa, Bohlen 1984). Because many geochemical
processes occur at higher temperatures, other materials more
refractory than NaCl must used in the assemblies. But these
materials are stronger than NaCl and their use might increase
the minimum pressure attainable in a piston-cylinder apparatus
and create a significant difference between the nominal and the
sample pressures due to differential loading. Additionally, the
strength of materials is temperature sensitive and a pressure
calibration at one temperature may not be applicable to experiments at another.
This paper presents the results of low-pressure calibration of
the crushable alumina-pyrex-NaCl assemblies normally used in
our laboratory. I show that the water solubility in albite melt determined by Behrens et al. (2001) can be used for piston-cylinder
apparatus pressure calibration with an easily attainable precision
of ±25 MPa. This method of calibration is consistent with that
determined from the melting of NaCl, but more flexible in that
calibrations can be performed at constant pressure and varying
temperature. The calibrations demonstrate that the assemblies
have an average friction correction of +50 MPa over the range
of temperatures from 800 to 1200 °C and 400 to 500 MPa.
EXPERIMENTAL AND ANALYTICAL TECHNIQUES
All experiments were performed in 19.1 mm crushable alumina-pyrex-NaCl
assemblies (Fig. 1) in a piston-cylinder apparatus. For each run, the assembly
was cold pressurized to a nominal pressure of 320 MPa and then simultaneously
heated and continuously pressurized to run conditions such that at 500 °C the
nominal pressure was 400 MPa, and at 800 °C the nominal pressure was 450
MPa. For 500 MPa experiments, this nominal pressure was maintained until the
experiment reached the desired final temperature; for experiments that needed to
reach 550 MPa, the sample was additionally pressurized to a nominal pressure
of 500 MPa at 1000 °C and this pressure maintained until run temperature was
reached. Pressure was controlled to within 32 MPa during the experiments, which
in some cases required either episodic pumping or venting of the piston cylinder.
Temperatures were measured with type C, W-Re thermocouples and controlled
to within 3 °C of the setpoint. Experiments were quenched isobarically to below
600 °C in less than 20 s.
I initially calibrated the assemblies using the melting temperature of NaCl
at 500 MPa, 918 °C (Bohlen 1984). Reagent-grade NaCl plus a Pt sphere of approximately 0.5 mm diameter were placed into a 3 mm diameter Au75Pd25 capsule
with the sphere at the top. The open capsule was dried at 110 °C for a minimum
of 1 h before being welded closed. Melting during experiments was detected by
falling of the sphere. Based upon 20 minute duration experiments at 920 °C and
nominal pressures of 475 MPa, where no melting occurred, and 425 MPa, where
melting occurred, I determined the need to add 50 MPa to the nominal pressure
to find the real pressure; all pressures for the albite water solubility experiments
include this correction.
The starting material for albite water-saturation experiments was an albite
glass synthesized from carbonates and oxides by repeated melting and grinding;
the analyzed composition of the starting material is homogenous and almost pure
albite: Na0.97Al1.00Si3.03O8. Powder of this glass plus 11 to 16 wt% distilled H2O were
placed inside 3 mm diameter Pt or Au75Pd25 capsules and welded closed without
volatile loss. Sealed capsules with albite glass and water were heated at 110 °C for
a minimum of one hour to check for leaks and to homogenize the water along the
grain boundaries of the powder. The capsules were then placed into holes drilled in
the crushable alumina of the solid media assembly and backfilled with pyrophyllite
to minimize water loss from the capsules by hydrogen diffusion.
Albite water solubility experiments at 500 MPa (AB-31 and -32) were pressurized and heated to 500 MPa, 1200 °C and held at those conditions for 1 hour
before quenching. All other experiments were pressurized and heated to 550 MPa,
1200 ˚C, where the melts were water undersaturated, and held at those conditions
for 1 hour to allow all the water present in the capsule to dissolve completely into
the melt. Following this step, the pressure was reduced at a rate of approximately
1 MPa/s to the pressure of interest and held at the pressure and temperature of
interest for the durations listed in Table 1. Experiments whose final temperature
was 800 °C (AB-19, -21, -22) were hydrated at 1200 °C, 550 MPa for one hour
before isobarically lowering the temperature to 800 °C, followed by decreasing the
pressure to 450 MPa. All experiments were highly vesiculated and most contained
a free water phase when opened.
The water concentration was determined by the “difference from 100” technique. This technique assumes that the difference between the analytical sum
determined by electron microprobe analysis and 100% is the concentration of
dissolved water in the melt. Although not as accurate as infrared techniques and
Karl Fischer analysis (see Behrens et al. 2001), this technique has been shown to be
accurate to within approximately 0.5 wt% H2O (Devine et al. 1995; Thomas 2000);
this value is similar to the analytical uncertainty based upon multiple analyses of
the experiments from this study (Table 1). Electron microprobe analysis at McGill
University used albite as the standard for Na, and orthoclase for Al and Si. Counting times for Na were 10 s on the peak and 5 s on each background, whereas Al
and Si were each counted for 20 s on the peak and 10 s on each background. A 15
kV accelerating voltage with a 5.5 nA current on the sample and a beam diameter
of 20 micrometers was used for the analysis. These conditions were chosen such
that Na volatilization during analysis was not observed; such volatilization would
produce artificially high water concentration values. The lack of Na volatilization
was determined by comparing the stoichiometry of the hydrated glass with the
anhydrous starting glass. Our analyses were repeated in at least four locations in
the sample and demonstrate the homogeneous distribution of water.
Water-undersaturated experiments quenched directly from 550 MPa, 1200 °C
with 11.1 wt% added water produced micrometer-sized quench bubbles. These
experiments were imaged on the electron microprobe using back-scattered electrons
and the bubble areas counted. These experiments produced approximately 1.5 vol%
quench bubbles, which when combined with a maximum estimate of the capsule
volumes after the experiment indicates that the water loss during quenching was
less than our analytical uncertainty. Measured water concentrations were not corrected for water loss during quenching.
RESULTS AND DISCUSSION
The results listed in Table 1 and displayed in Figure 2 present
the measured water solubilities in albite melts at 392, 450, and
500 MPa, 1200 and 800 °C. Comparison between these data and
BAKER: PISTON–CYLINDER CALIBRATION USING WATER SOLUBILITY IN ALBITE
TABLE 1.
Albite water solubility determinations in calibration experiments
15
14
Experiment
the albite water solubility measurements of Behrens et al. (2001),
obtained from experiments in cold-seal and internally heated
pressure vessels, demonstrates excellent agreement. Almost all
average water solubilities measured in this study are within 1σ
of the 4th-order polynomial fit to the Behrens et al. (2001) measurements (Fig. 2). This relationship was inverted to calculate
the pressure from the water solubility:
P = –67.061 + 48.777 H2O – 2.2576 H2O2 + 0.42747 H2O3
– 0.019803 H2O4,
where P is in MPa and H2O is the wt% water measured in the
glass.
Comparison of the calculated pressures from the water
solubility with the calibrated pressure demonstrates that both
the NaCl calibration and the water solubility pressure calibration techniques typically agree to within 25 MPa (Table 1). The
average difference between the nominal pressures of the experiments, 50 MPa below those listed in Table 1, and the pressures
measured by water solubility indicate that the true pressure in the
experiments was equal to the nominal pressure +48 MPa. This
difference is well within the uncertainty of the NaCl calibration
technique and demonstrates the agreement of both calibration
techniques. Replicate albite water-solubility experiments at the
same temperature and pressure conditions, but in different assemblies, indicate a maximum difference of 31 MPa between
experiments (Table 1). This variation is only slightly above the
analytical uncertainty.
I performed duplicate experiments in the same assemblies to
investigate the possibility of significant pressure differentials in
the assembly and to assess the reproducibility of the experiments.
These experiments yielded pressures with overlapping uncertainties. The measured pressure differences between the duplicate
experiments are attributed to analytical uncertainties rather than
a true pressure differential within the assembly.
The albite water-solubility technique allows us to calibrate
Water solubility in albite melt
13
12
H2O (wt%)
T
Pressure* Duration No.
H2O†
Pcalc‡
(MPa)
(°C)
(MPa)
(h) analyses (wt%)
1200
500
1
9
11.29 (0.39) 489 (19)
AB-31§
1200
500
1
10 11.95 (0.57) 519 (26)
AB-32§
AB-18
1200
450
32
6
10.35( 0.30) 443 (16)
1200
392
2
8
9.95 (0.31) 422 (16)
AB-25§
1200
392
2
8
9.74 (0.47) 411 (25)
AB-26§
AB-27
1200
392
2
8
9.38 (0.47) 391 (25)
AB-29
1200
392
8
8
9.40 (0.38) 392 (21)
AB-19
800
450
2
5
10.17 (0.41) 433 (22)
800
450
32
4
9.84 (0.42) 416 (22)
AB-21§
800
450
32
6
10.28 (0.46) 439 (24)
AB-22§
* Nominal pressure + 50 MPa, based upon calibration with NaCl melting at 500
MPa, 920 °C.
† Water solubility in albite melt determined by the difference between the electron microprobe analytical total and 100 wt%. The average water concentration
is followed in parentheses by the 1-sigma standard deviation of the electron
microprobe analytical totals.
‡ Pressure is calculated from the measured water solubility and a fourth-order
polynomial equation fit to the solubilities reported in Behrens et al. (2001).
Numbers in parentheses are the uncertainties in the pressure based upon the
1-sigma standard deviation of the water concentration.
§ These sequentially numbered runs were performed simultaneously in the
same 19.1 mm assembly.
1555
11
10
9
8
o
7
1200 C
o
800 C
Behrens et al. (2001)
6
5
250
300
350
400
450
500
550
Pressure (MPa)
FIGURE 2. Water solubility in albite melt as a function of the
experimental pressure based upon calibration using the melting of NaCl
at 500 MPa. Error bars represent uncertainty in the measured water
concentrations based upon the 1σ standard deviation of the analyses
from this study and from Behrens et al. (2001).
the piston cylinder at constant pressure and temperatures that
span the range of most igneous processes. An advantage of using
water solubility in albite is its virtual insensitivity to temperature
at the investigated pressures (Behrens et al. 2001). The pressures determined from water concentrations in experiments at
450 MPa, 1200 and 800 °C, are within 28 MPa of each other,
slightly larger than the measurement uncertainties. This variation is less than that seen in replicate experiments performed
in different assemblies at 392 MPa, 1200 °C. Thus, there is no
significant temperature effect on the pressure correction for
these assemblies.
An advantage of using albite water solubility to calibrate
pressures in piston-cylinder apparatus is that the rapid kinetics of
water diffusion and vesiculation in silicate melts ensures that the
pressure measured by albite water solubility represents the final
pressure in the capsule. Because of this property, we can easily
investigate the possibility of temporal changes in the pressure
calibration, which is impossible to do using the falling sphere
technique to determine NaCl melting (although more complicated techniques such as electrical conductivity and differential
thermal analysis can be used with NaCl melting to investigate
temporal changes in pressure calibration). Changes in the pressure calibration might be expected as, with time, the crushable
alumina inside the assembly sinters into a hard ceramic and
reacts with the pyrex pedestal. Experiments of different durations
were performed at 1200 and 800 °C to investigate whether the
relationship between the nominal and the real pressure changed
with time. However, no such effect was observed and the pressure correction observed in short experiments is applicable to
longer ones.
DISADVANTAGES AND ADVANTAGES OF THE WATER
SOLUBILITY PRESSURE CALIBRATION
The albite water solubility calibration technique for the
piston-cylinder apparatus is accurate to approximately ±25
MPa when the water concentration in the quenched glasses is
determined by electron microprobe analysis. This precision is
limited by the ± 0.5 wt% uncertainty in the electron microprobe
“difference from 100” technique. This accuracy in pressure is
about an order of magnitude less than that attainable in cold-seal
1556
BAKER: PISTON–CYLINDER CALIBRATION USING WATER SOLUBILITY IN ALBITE
pressure vessels or internally heated pressure vessels where the
sample pressure is measured directly. However, this level of
accuracy is sufficient for most petrologic and kinetic studies
where experiments are commonly performed at approximately
100 MPa intervals; additionally, the piston-cylinder apparatus
has the advantage of a higher temperature limit and more-rapid
heating and cooling compared with cold-seal and internally
heated pressure vessels. Other, more-accurate and time consuming methods of water analysis (e.g., infrared spectrometry, Karl
Fischer titration) may be used to increase the precision of the
calibration by a factor of two to three (see Behrens et al. 2001).
Such techniques were not used in this study because of my desire
to develop a routine, rapid, easy, and inexpensive, technique for
piston-cylinder calibration.
These results demonstrate that the piston-cylinder apparatus
with a crushable alumina-pyrex-NaCl solid media assembly can
be calibrated easily by measuring water solubility in albite melt
and routinely used for long-duration experiments at pressures as
low as 400 MPa. Attempts to perform experiments below 392
MPa failed because of poor electrical contact between the piston
and the assembly resulting in furnace failures. The maximum
temperature limit of these assemblies is at least 1300 °C at 500
MPa based upon diffusion experiments we have performed in
them. Neither temperature nor time affects the pressure calibration of the solid-media assembly used in this study. The albite
water solubility technique allows calibration of a piston-cylinder
apparatus in a single experiment; this is a slight advantage compared to the minimum of a pair of experiments needed for the
NaCl melting calibration with a falling sphere. However, unlike
the NaCl melting calibration, the water solubility technique can
determine the effects of temperature and duration on the pressure calibration of a solid-media assembly. Additionally, the
ability to calibrate with just a single capsule containing albite +
water suggests the possibility of including such a capsule in the
solid-media assembly for use as an in situ pressure monitor for
critical experiments.
ACKNOWLEDGMENTS
Thoughtful reviews from Dana Johnston and David Walker were very much
appreciated and resulted in an improved manuscript. Lang Shi is thanked for his
help in choosing the best analytical parameters for electron microprobe analysis
and keeping the machine in excellent operating condition. Comments on this manuscript by Carmela Freda were very helpful. Some of the experiments performed in
this study were done with the aid of Phyllis Lang, Jean-Francois Bergevin, Leslie
Gold, Yanan Liu, and Geneviève Robert. Funding for this study was provided by
an NSERC Discovery grant to D.R.B.
REFERENCES CITED
Behrens, H., Meyer, M., Holtz, F., Benne, D., and Nowak, M. (2001) The effect
of alkali ionic radius, temperature, and pressure on the solubility of water in
MAlSi3O8 melts (M = Li, Na, K, Rb). Chemical Geology, 174, 275−289.
Boettcher, A.L., Windom, K.E., Bohlen, S.R., and Luth, R.W. (1981) Low-friction,
anhydrous, low- to high-temperature furnace assembly for piston-cylinder
apparatus. Reviews of Scientific Instrumentation, 52, 1903−1904.
Bohlen, S.R. (1984) Equilibria for precise pressure calibration and a frictionless
furnace assembly for the piston-cylinder apparatus. Neues Jahrbuch für Mineralogie Monatshefte, 9, 404−412.
Boyd, J.R. and England, J.L. (1960) Apparatus for phase-equilibrium measurements at pressures up to 50 kilobars and temperatures up to 1750 °C. Journal
of Geophysical Research, 65, 741−748.
Devine, J.D., Gardner, J.E., Brack, H.P., Layne, G.D., and Rutherford, M.J. (1995)
Comparison of microanalytical methods for estimating H2O contents of silicic
volcanic glasses. American Mineralogist, 80, 319−328.
Holland, T.J.B. (1980) The reaction albite = jadeite + quartz determined experimentally in the range 600–1200 °C. American Mineralogist, 65, 129−134.
Holloway, J.R. and Wood, B.J. (1988) Simulating the Earth, 196 p. Unwin Hyman,
Winchester, Massachusetts.
Johannes, W., Bell, P.M., Boettcher, A.L., Chipman, D.W., Hays, J.F., Mao, H.K.,
Newton, R.C., and Seifert, F. (1971) An interlaboratory comparison of pistoncylinder pressure calibration using the albite-breakdown reaction. Contributions to Mineralogy and Petrology, 32, 24−38.
Mirwald, P.W., Getting, I.C., and Kennedy, G.C. (1975) Low-friction cell for
piston-cylinder high-pressure apparatus. Journal of Geophysical Research,
80, 1519−1525.
Rosenbaum, J.M., Kyser, T.K., and Walker, D. (1994) High temperature oxygen
isotope fractionation in the enstatite-olivine-BaCO3 system. Geochimica et
Cosmochimica Acta, 58, 2653−2660.
Thomas, R. (2000) Estimation of water contents of granite melt inclusions by
confocal laser Raman microprobe spectroscopy. American Mineralogist, 85,
868−872.
MANUSCRIPT RECEIVED FEBRUARY 12, 2004
MANUSCRIPT ACCEPTED APRIL 22, 2004
MANUSCRIPT HANDLED BY JOHN AYERS