Highlights in Experimental Techniques

During the previous quadrennium an impressive list of developments in diamond anvil cell (DAC) and large volume press techniques representing the culmination of great effort to obtain experimental data at deep mantle and core conditions was compiled [ Wolf et al. 1991]. In this quadrennium, progress in research techniques has been no less impressive.

Badding et al. [1992] measured iron hydride up to 80 GPa using the diamond anvil cell (DAC). Yagi et al. [1994] reported on the formation and structure of iron-hydride with a cubic anvil apparatus combined with x-rays from synchrotron radiation for up to 6 GPa. Williams and Jeanloz [1990] reported on measurements of iron-sulfur up to 120 GPa using the DAC.

Boehler et al. [1990] (the Mainz Group) reported on DAC measurements on the of iron up to 120 GPa and found a triple point (t.p.) connecting the (hcp), (fcc), and liquid () phases at about 100 GPa and 2800 K. This triple point appeared in Boehler's experiments because he found extensive curvature in the -liquid boundary.

Williams et al. [1987] did not see this curvature along the -liquid boundary in their data on , and, as a consequence, their was relatively high (4000 K) compared with Boehler et al.'s. They [ Williams et al., 1991] reported the -- triple point to be at 300 GPa and 7500 K.

Using a DAC, Boehler [1993] extended his measurements on the of iron from 100 GPa to 190 GPa and from 2800 to 3800 K. In 1994 he confirmed the existence of the t.p.\ at 100 GPa by experimental data showing all three branches. Boehler's experiments at high pressure were done by plating his iron specimen on a ruby disk immersed in a pressure medium made of ruby powder. Temperature was measured using the radiation from the iron through the ruby disk, to fit a Planck's radiation function. Melting was detected visually as the onset of convective motion.

The Uppsala group ([ Saxena et al., 1993]), using a DAC, obtained experimental results on of iron up to 60 GPa, confirming the curve of Boehler et al. [1990], who, as previously mentioned, had reported of iron up to 100 GPa (see also Figure 5, Anderson [1993]).

The Uppsala group experiment was much like that of the Mainz group, except that Boehler et al. [1990] used argon as the pressure medium below 100 GPa, and Saxena et al. used MgO powder. Saxena et al. used an abrupt change in the slope of versus laser power (for heating) to find . Boehler et al. employed three methods to find , including that used by Saxena's group, all yielding the same curve .

Mao et al. [1990] found the volume of iron to above 300 GPa at using x-ray measurements by synchrotron radiation to accurately pin down the lattice constants. Their results confirmed and extended the measurements of Jephcoat et al. [1986] made up to 70 GPa. Mao et al. also found virtually no difference between the V- results on iron-nickel Fe Ni and those on pure iron up to 300 GPa.

Because the values for iron found by Boehler using a DAC were low compared with those found by the CalTech group [ Bass et al., 1987, 1990; Ahrens et al., 1990] using shock compression, an analogous shock wave study was made by the Livermore shock group ([ Yoo et al., 1993a, 1994a]). Their results were in good agreement with the Bass et al. [1987, 1990] curve except for one datum. For that datum, the Bass et al.\ result for was 9000 K at 300 GPa, 1500 K higher than the Yoo et al. results (see highest point in Figure 1). Both the CalTech group and the Livermore group used shock wave radiance to find . The radiance shock wave measurements gave a more than higher than an equivalent point on the shock wave curve determined by Brown and McQueen (B&M) [1986], who used standard thermodynamics to find from the Hugoniots.

In order to test the Boehler data on , Yoo et al. [1993b], undertook to measure by the DAC and found that they could reproduce the curve of Boehler up to the limit of their experiment (40 GPa) (see comment in abstract, Yoo et al. [1993b]). Though their DAC measurements of were in good agreement with Boehler's DAC results, they disagreed with their own shock wave measurements. Further, their shock wave measurements of using radiance to find disagreed with the shock wave measurement of by Brown and McQueen [1986] (who used the classical method of finding along the Hugoniot).

The importance of the high datum in the Bass et al. [1987] shock wave analyses (9000 K at 300 GPa) is that by using it with their other data, Bass et al. [1987] estimated that is , and this influenced Williams et al. [1987] to report . For comparison, the Yoo et al. [1994a] estimate using the Livermore shock wave data is .

Ahrens [personal communication, 1994] suggests that the Bass et al. high datum at 300 GPa should be disregarded, in view of new, unpublished results. At the 1994 Fall meeting of the AGU, Gallagher and Ahrens [1994] presented new data that substantially decrease the CalTech shock wave values of of iron from values reported by Bass et al. [1987] and Ahrens et al. [1990]. Their new measurements allowed the thermal diffusivity of AlO to be determined at high AlO is a window through which the radiance of the shocked iron must pass before it is detected).

Five papers, those of Boness and Brown [1990], McQueen and Isaak [1990], Nellis and Yoo [1990], Tan and Ahrens [1990], and Duba [1994] discuss problems in the radiative shock wave data reduction to find (all indicating the importance of thermal diffusivity of AlO in the data reduction for ).

These new determinations of thermal diffusivity substantially lowered the calculated values of iron (see crosses and arrows in Figure 1) from those proposed by Bass et al. [1990]. Gallagher and Ahrens concluded, ``Thus the melting point of iron inferred from previous shock temperature measurement can be decreased by as much as .'' The new 1994 CalTech radiance measurements of connect with the upper DAC measurement of Boehler [1993] (see Figure 1), thus verifying Boehler's measurements. ( Duba [1992] was the first to publicly support Boehler's measurements over those of Williams et al. [1987].)

We see that the Gallagher and Ahrens data on have a larger slope on the high pressure side of 200 GPa than that of the Boehler data on coming into 200 GPa. The value of along a melting curve is a monotonically decreasing function. The presence of a triple point causes to increase suddenly [ Weathers and Bassett, 1990]. Thus the discontinuity in at 200 GPa demands a t.p. in the vicinity of 200 GPa.

A t.p. requires three branches. The Boehler branch is a liquidus, and the Gallagher and Ahrens branch is a liquidus, so the third branch must be a solid-solid transition. We therefore look for a solid-solid transition in the vicinity of 200 GPa to make the t.p. complete, and this is satisfied by the solid-solid transition of Brown and McQueen [1986], who reported 4400 K with wide error bars and 200 GPa.

Such a triple point was proposed by Anderson [1993, 1994] (Figure 1, Figure 2). He located it at 190 GPa. If the solid-liquid boundary for below 200 GPa separates the liquid from the hcp phase, as customarily assumed, then the solid-solid transition of Brown and McQueen separates the hcp phase from another phase, which cannot be hcp, nor can it be bcc (as we shall see below). It is therefore fcc or some unknown phase of iron. This new phase is undoubtedly the same as identified as by Boehler [1986] and later by Anderson [1994].

Since the early CalTech ([ Bass et al., 1987]) shock wave work was used to substantiate the Williams et al. [1987] estimate of (extrapolated value, ), the new work of Gallagher and Ahrens must decrease the Williams et al. [1987] estimate of the value, . Duba's [1992] analysis anticipated the new results of Gallagher and Ahrens.

There is a large discrepancy between the reported DAC measurements of by Williams et al. [1987] up to 60 GPa and those of Boehler [1993] up to 100 GPa (shown in Figure 1). This discrepancy was evaluated by Chen and Ahrens [1994], who computed between the (hcp) solid phase and the liquid phase of iron by means of the Gibbs energy, G, at high and . The of liquid iron was calculated from the measurements reported by W.W. Anderson and Ahrens [1994]. The effect of on is strong, and in turn is strongly affected by the ambient bulk modulus, . Thus the of the solid phase is sensitive to the value of the equation of state parameters.

Since along the melting curve must vanish, Chen and Ahrens were able to use the melting curves from the two DAC experiments ([ Williams et al., 1987 and Boehler, 1993]) to find respective values of the EoS parameters of iron. From this analysis, they found that the from Williams et al. required an unreasonably low value of (30 GPa) and a high value of (7.5) for the solid hcp phase. They found from the of Boehler [1993] that and ; is reasonably close to the value and measured by Mao et al. [1990] for iron. Chen and Ahrens concluded, ``our calculations favor Boehler's melting curve as more thermodynamically consistent with the various equations of state of the iron phase.''

While there is still a discrepancy between the various experimental results on , there has been substantial convergence in the last two years, resulting especially from the Gallagher and Ahrens work.