The Structure of Iron in Earth’s Inner Core

See allHide authors and affiliations

Science  15 Oct 2010:
Vol. 330, Issue 6002, pp. 359-361
DOI: 10.1126/science.1194662


Earth’s solid inner core is mainly composed of iron (Fe). Because the relevant ultrahigh pressure and temperature conditions are difficult to produce experimentally, the preferred crystal structure of Fe at the inner core remains uncertain. Static compression experiments showed that the hexagonal close-packed (hcp) structure of Fe is stable up to 377 gigapascals and 5700 kelvin, corresponding to inner core conditions. The observed weak temperature dependence of the c/a axial ratio suggests that hcp Fe is elastically anisotropic at core temperatures. Preferred orientation of the hcp phase may explain previously observed inner core seismic anisotropy.

Determining the crystal structure of iron (Fe) under ultrahigh pressure and temperature (P-T) conditions is a key piece of information required to decipher the complex seismic structures observed in Earth’s inner core (13). Fe adopts body-centered cubic (bcc) structure at ambient conditions and transforms into the hexagonal close-packed (hcp) phase above 15 GPa. Although hcp Fe can persist under core pressures at 300 K (4, 5), a phase transition at elevated temperature is a possibility. Both theory and experiments have proposed different forms of Fe at simultaneously high P-T conditions, which include bcc (6, 7), face-centered-cubic (fcc) (8), and hcp structures (5, 9). The structure of Fe has never been examined experimentally at the inner core P-T conditions (>330 GPa and ≥5000 K), because the techniques previously used to produce such extreme conditions—dynamical shock-wave experiments—impede the ability to make simultaneous structure measurements on the order of a microsecond.

Based on a combination of static compression experiments in a laser-heated diamond-anvil cell (DAC) and synchrotron x-ray diffraction (XRD) measurements (Fig. 1), we determined the structure of Fe up to 377 GPa and 5700 K (10). A temperature gradient was relatively large when the sample was heated to more than 5000 K at >300 GPa (Fig. 2A); nevertheless, the variations were less than ±10% in the 6-μm region across the hot spot, which corresponds to the x-ray beam size at full width at half maximum, considering the fluctuations in temperature with time (Fig. 2B). We calculated the sample temperature by averaging the variation in the 6-μm area probed by x-rays. Pressure was determined from the unit-cell volume of hcp Fe, using its P-V-T (where V is volume) equation of state (11). The ±10% temperature variation leads to about ±2% uncertainty in pressure (377 ± 8.5 GPa at 5700 K). The pressure gradient in the sample was <5 GPa at ~300 GPa in a 10-μm area after heating.

Fig. 1

Representative XRD patterns of hcp Fe at (A) 332 GPa and 4820 K and at (B) 356 GPa and 5520 K. The peak positions of the bcc and fcc phases were calculated for volumes larger by 0 to 1% than that for the observed hcp phase. hcp, hcp Fe; py, pyrite-type SiO2 (pressure medium); C, Fe3C cementite; Re, rhenium (gasket).

Fig. 2

(A) Typical temperature profiles across the laser-heated spot obtained at pressures above 300 GPa. (B) Fluctuations in temperature with time in 60 s.

To construct the phase diagram of Fe at inner core conditions, we conducted six separate sets of experiments (Fig. 3). The first experiment at 303 GPa and room temperature resulted in an XRD pattern that included peaks from hcp Fe and Re (the gasket material) (fig. S1A) (10). Subsequently, we heated this sample to 4820 K at 332 GPa. The one-dimensional (1D) XRD pattern did not change except for the appearance of an hcp 002 line (Fig. 1A). On the other hand, the 2D image became spotty (fig. S1B), indicating crystal growth and hence the stability of hcp Fe at these P-T conditions. After these measurements, the sample was temperature-quenched and then further compressed to 321 GPa at room temperature. We again heated the sample to 5520 K at 356 GPa. The XRD pattern was still dominated by the hcp phase (Fig. 1B), but minor peaks assigned to pyrite-type SiO2 (12) (the pressure medium) and Fe3C cementite appeared. The measured lattice parameters and volumes of Fe3C are in agreement with earlier experimental results to 187 GPa (13) (table S1). Similar observations were made in five other experiments, which were conducted in a wide P-T range from 135 GPa and 2690 K to 377 GPa and 5700 K (Fig. 3). In all measurements, we obtained no evidence of a phase transition to bcc or fcc Fe phases.

Fig. 3

Phase diagram of Fe and the inferred temperature profile inside Earth (19, 23, 24). Open symbols indicate the present results (different symbols indicate different runs), and solid diamonds indicate data from previous experimental work (5). The low-pressure solid-solid phase transition boundaries and melting curve are from Boehler (19). Liq., liquid. (Inset) Sample photograph at 335 GPa in the DAC.

The presence of Fe3C in the XRD pattern indicates contamination by carbon from the diamond anvils, which has been reported in several earlier DAC studies (1416). Nevertheless, the maximum solubility of carbon in solid Fe is already low (<0.6%) at 44 GPa and further decreases with increasing pressure (17). Moreover, although previous ab initio calculations (18) demonstrated that the addition of a small amount of carbon to Fe strongly stabilizes the bcc phase relative to the hcp phase, we did not observe XRD peaks from bcc Fe. These suggest that the effect of carbon contamination on the phase relation of Fe was negligible.

These experiments were performed at temperatures near the melting curve of Fe (19, 20) (Fig. 3). We observed a temperature jump from 3220 K to ~4000 K at 135 GPa by a small increase in laser output power (fig. S2). Large fluctuations in temperature and the elevation of background intensity in the XRD patterns were also noticed after the temperature jump. These are usually interpreted as a sign of melting (21, 22). The melting temperature at 135 GPa should thus be little higher than 3220 K, which is somewhat higher than the previous experimental determinations by Boehler (19, 22) but lower than those by Ma et al. (20), considering the effect of thermal pressure contributions. At >350 GPa, on the other hand, 2D XRD images showed the extensive grain growth above 5700 K, suggesting that this temperature is close to the melting point. In addition, we found no anomalies up to 4120 K at 210 GPa. These observations place constraints on the melting curve of Fe, which previously existed only up to 200 GPa.

These results indicate that hcp Fe is a stable form of Fe up to 377 GPa and 5700 K, which is compatible with previous ab initio calculations by Vočadlo et al. (9). The estimation of temperature at the inner/outer core boundary ranges from 4850 to 5700 K, depending on the melting temperature of Fe and the effect of light alloying elements (19, 23, 24). The temperature gradient should be very small within the inner core (24). Our experiments thus represent a range of inner core P-T conditions. One limitation of these experiments is that chemical impurities such as silicon and sulfur, which have the ability to change the stable crystal structure (9, 18, 25), were not accounted for.

Strong seismic anisotropy exists in the inner core, with longitudinal waves traveling ~3% faster along the polar axis than in the equatorial plane (1). This was originally attributed to the preferred orientation of hcp Fe, which exhibits a strong single-crystal elastic anisotropy, at least at low temperature (26). However, Steinle-Neumann et al. (27) demonstrated that the c/a axial ratio of hcp Fe increases substantially with increasing temperature (Fig. 4), which has a significant influence on its elastic anisotropy. More recent calculations (6, 28) reported that the c/a ratio approaches the value of 1.6299 for the ideal hcp structure at high temperature, and consequently elastic anisotropy of hcp Fe no longer exists at inner core conditions. On the other hand, experimental evidence previously suggested weak temperature dependence of the c/a ratio at 140 GPa (22), as do our data at ~330 GPa (Fig. 4). The c/a ratio of 1.602 at 332 GPa and 4820 K, which is substantially lower than the ideal value, suggests that hcp Fe should be elastically anisotropic even at the high temperature conditions of the inner core. The observed seismic anisotropy may therefore result from the preferred orientation of the hcp phase with the c axis parallel to Earth’s rotation axis (26).

Fig. 4

Temperature dependence of the c/a axial ratio of hcp Fe collected at 135 GPa (red circles) and ~330 GPa (red diamonds) (table S2). Previous experimental results at 84, 106, and 140 GPa are from Boehler et al. (22) (blue open symbols), and those at 161 GPa are from Ma et al. (20) (purple open squares). The results of theoretical calculations are also shown by the dot-dashed curve (27), dashed curve (6), and dotted curve (28).

Supporting Online Material

Materials and Methods

Figs. S1 to S3

Tables S1 and S2


References and Notes

  1. Materials and methods are available on Science Online.
  2. We thank N. Sata, T. Komabayashi, and Y. Tanaka for technical support. The synchrotron XRD measurements were conducted at beamline BL10XU of the SPring-8 synchrotron radiation facility (proposal nos. 2009B0087 and 2010A0087). S.T. was supported by the JSPS Research Fellowships for Young Scientists.
View Abstract

Stay Connected to Science

Navigate This Article