Is iron at the Earth ’ s core conditions hcp-structured ?

Iron is the main component of the Earth’s core and its structure and properties are important for interpretation of geophysical observations and modeling dynamics of the core. We argue that the diffraction lines in the high temperature high pressure X-ray diffraction pattern, presented by Tateno et al., 2010 and interpreted as those of solely hot hcp-Fe, correspond indeed to the insufficiently heated part of the sample. We show that observed diffraction spots are either due to bcc-Fe or carbides.


Introduction
The Earth's core is the most remote region of our planet.The boundary of the core is at about 2900 km in depth.Not only we do not have samples from the core, but we do not even expect to get any.To date, the most direct observations of the core 74 have come from seismological studies using remote-sensing techniques.Due to the complex internal structure of the Earth, seismic investigations require extensive data coverage and appropriate models.Decoding geochemical signature of the core carried by mantle plumes faces similar challenges.Experimental and computational simulations have been hindered by the necessity to approach pressures over 140 GPa and temperatures above 3000 K prevalent in the core.For these reasons, many fundamental issues concerning the Earth's core remain controversial and poorly understood.
Composition of the Earth's core is a geochemical parameter crucial for understanding the evolution and current dynamics of our planet.Since the discovery of the Earth's core about a century ago, the idea of iron being the dominant component of the core gained firm support from cosmochemical and geochemical observations, seismic data, theory of geomagnetism, and high-pressure studies.Cosmochemical data and studies of iron meteorites give evidence that the Earth's core contains significant (5 to 15 %) amounts of nickel.Already in the early 1950s it was recognized that the outer core is less dense than iron or an iron-nickel alloy at corresponding conditions.Current estimates for the density deficit relative to solid iron vary between 6% to 10% for the outer core and 2% to 5% for the inner core (Dubrovinsky et al., 2000;Dubrovinsky et al., 2007;Dewaele et al., 2010).Uncertainties are in part related to the experimental difficulties in measurements of the properties of iron (or iron-nickel alloys) at multimegabar pressure range and high temperatures, but the fundamental problem is the absence of a universal pressure scale that has led to significant discrepancies in the inter-laboratory studies (see, for example, Dubrovinsky et al., 2007).Nevertheless, the main element of the core is iron and its properties at multimegabar pressures and thousands degrees are vital for interpreting mineralogy and geochemistry of the Earth core.
Based on the X-ray diffraction, iron has been reported to adapt several structures at high pressure and temperature:  and , body-centered cubic (bcc); , facecentered cubic (fcc); ε, hexagonal close-packed (hcp); , double-hcp phase (dhcp) or orthorhombic phase.The -, -, and -phases at lower pressures are well established and broadly accepted.The -phase, observed at pressures above 30 GPa and high temperature, is probably metastable or stabilizes due to contaminations.The εphase has been proved to be a dominant phase in a wide pressure-temperature range approaching the Earth's core conditions.Theoretical calculations suggest that iron and iron-based alloys at sufficiently high pressures (above 100 GPa) and temperatures, close to the melting point, transform into the bcc structured phase (Belonoshko et al., 2003).Moreover, both experimental (Lin et al., 2003) and theoretical (Vočadlo et al. 2003) works pointed out towards stabilization of the bccstructured phase due to the elements (particularly, silicon) important for the Earth's inner core chemistry.The hcp-bcc transformation of the Fe 0.9 Ni 0.1 alloy was observed at pressures above 225 GPa and temperatures over 3400 K (Dubrovinsky et al., 2007).Recently Tateno et al., 2010 described breaking through experiments on pure iron at conditions of the Earth's inner core.The authors reported stability of hexagonal close-packed (hcp) Fe at pressures up to 377 GPa and 5700 K and a very weak temperature dependence of the ratio of the lattice parameters c/a on heating.In view of importance of these results for understanding structure and dynamics of the Earth's core we re-analyze the diffraction data reported by Tateno et al., 2010 and demonstrate that Tateno's et al. conclusions regarding stability of hcp Fe at multimegabar pressures and temperatures over 5000 K are not sustained by their own data.

Effect of laser heating in DAC on appearance of diffraction lines in 2D images
One of the main methodological problems with experiments in laser-heated diamond anvil cells (DACs) is in determining pressure at high temperature.In x-ray diffraction experiments with laser-heated diamond anvil cells (DACs) pressure is usually determined using pressure markers (materials with known thermal equations of state (tEOS) added to the sample) (Dubrovinsky et al., 2007;Ma et al., 2004).Tateno et al., 2010 did not have any pressure marker and used known tEOS of hcp-Fe itself (Dubrovinsky et al., 2000).Such an approach could be reasonable, keeping in mind really extreme conditions of the experiments (Tateno et al., 2010), but only in case if a clear evidence of measuring both the diffraction and temperature from the same hot part of the sample is provided.
A number of specific features of the diffraction pattern indicate the state of the heated sample: formation of "arcs" or intensive spotting (due to re-crystallization of the sample) of the diffraction lines, as well as their asymmetry and even splitting due to temperature gradients at inhomogeneous heating (in most cases at temperatures above 2500 K) or at misalignment of the X-ray and laser beams.Figure 1 illustrates these indicative features on example of the diffraction patterns of Ni we obtained on its high pressure high temperature (HPHT) study.As seen in the upper panel of Fig. 1, at HPHT conditions the diffraction pattern of Ni is a superposition of diffraction patterns of cold (continues lines) and hot (spots appearing to the right of continues lines) parts of the sample.This superposition is simply due to the fact that some part of the sample irradiated by X-rays is not affected by laser heating.Due to the thermal expansion of the unit cell of Ni at high temperature, the spots representing the hot part of the sample are shifted to the lower 2-theta values.After quenching the unit cell shrinks, so that at room temperature (lower panel) the spots from previously heated (and re-crystallized) part of the sample lie almost exactly on the lines from the cold (not heated) part of the Ni powder.
Fig. 2 shows two 2D X-ray diffraction images presented by Tateno et al., 2010 we could align to each other according to Kikuchi lines.Despite claimed huge difference in temperature (4820 K) the appearance and the texture of continues lines from hcp-Fe are the same (some obvious futures are marked by arrows).This observation means that the part of the sample (continues lines of hcp-Fe in Tateno et al., 2010) probed by X-ray was not affected by laser-heating.One cannot com-76 pletely exclude that this is a result of a misalignment of an extremely small sample, the X-ray and laser beams in the extremely difficult experiments.However, based on our own experience, we suggest that this is rather an intrinsic problem of laser-heating experiments in DACs at very high pressures -due to very thin thermal insulating layer between the diamond anvil and the sample, only a very thin layer of the sample is actually heated, while the rest remains (relatively) cold.

Effect of laser heating in DAC on molar volume and c/a ratio of hcp-Fe
Pressure and temperature affect lattice parameters of hcp-Fe, but in different wayspressure tends to decrease the lattice parameters c/a ratio (but very little, almost indistinguishable above 100 GPa), while elevated temperature leads to the increase of the c/a ratio (Bubrovinsky et al., 2000;Dewaele et al., 2010;Boehler et al., 2008).Tateno et al., 2010 reported some values of the lattice parameters and the c/a ratio obtained in their experiments.Table S2 in Tateno et al., 2010 contains an erroneous value of c/a = 1.6019 at 332 GPa and 4820 K that should be 1.5956 instead.When the arithmetic mistake is corrected, Table S2 (Tateno et al., 2010) demonstrates an unprecedented (not observed in previous studies) constancy of the c/a ratio over the great temperature interval of 2300 to 4820 K.At the same time the behavior of pressure, as reported by Tateno et al., 2010, is also remarkable.Particularly, Tateno et al., 2010 state that the sample was under pressure of 303 GPa and upon heating to 4820 K the pressure of 332 GPa developed.According to the tEOS of hcp-Fe, reported in our previous paper (Dubrovisnky et al., 2000) and used by Tateno et al., 2010, it means that the molar volume (4.144 cm 3 /mole) of the sample did not change during heating.Applying similar reasoning, for the sample compressed to 321 GPa its heating to 5520 K at constant volume should result in pressure of 358 GPa, and Tateno et al., 2010 indeed give a value of 356 GPa (i.e.within uncertainty it is the same pressure that we calculated for the unchanged volume).Such a "constant volume" and "constant c/a" behavior of hcp-Fe suggests that vast majority of the sample was not heated to the declared temperatures while the diffraction data were collected, and the observed "constant volume" of the sample (at ambient or relatively low temperatures) was incorrectly used in the hcp-Fe tEOS with improper high temperatures.

Phase composition of the laser-heated samples
According to the phase analysis conducted by Tateno et al., 2010 samples heated at pressures up to 377 GPa and 5700 K contain hcp-Fe, high-pressure phase of SiO 2 (used as a thermal insulator), and iron carbide.Fig. 3 shows simulated in GSAS the X-ray diffraction profile of a mixture of hcp-Fe, Fe 3 C, and Fe 7 C 3 in proportion 100:1:1.Lattice parameters of hcp-Fe (a=2.1528Ǻ, c=3.4351Ǻ, at P=332 GPa and T=4820 K) and Fe 3 C (a=3.843 Ǻ, b=4.411Ǻ, c=5.713Ǻ at P=332 GPa and T=4410 K) are taken from Tateno et al., 2010.Lattice parameters of Fe 7 C 3 for the same pressure are estimated based on works by Nakajima et al., 2009. While Tateno et al., 2010 reported the appearance of several spots in the diffraction patterns upon heating at high pressures and explained them either by crystal growth of hcp-Fe, or by formation of Fe 3 C (or by crystallization of pyrite-type SiO 2 ), the Fig. 3 shows that (a) Fe 3 C has in fact many more diffraction lines than the number of lines observed and indexed by Tateno et al., 2010 and, thus, the identification of Fe 3 C in Tateno et al., 2010 is ambiguous;(b) a number of lines of Fe 3 C and Fe 7 C 3 are overlapping that means that Fe 7 C 3 can not be ruled out; (c) Fe 7 C 3 has lines coinciding with ( 100) and ( 101) lines of hcp-Fe and this could be a reason for appearing spots on corresponding hcp-Fe lines; and (d) the (002) line of hcp-Fe, if present, could not be confused with any one of Fe 3 C or Fe 7 C 3 .Note, that positions of the diffraction lines of fcc-Fe with the same molar volume as that of hcp-Fe (4.15 cm 3 /mole) reported by Tateno et al., 2010 for 332 GPa and 4820 K would coincide with the diffraction lines of iron carbides, and in this sense the presence of fcc-Fe cannot be unambiguously excluded.
As we stated above, the (002) line of hcp-Fe, if present, would not coincide with any lines of Fe 3 C or Fe 7 C 3 .According to Tateno et al., 2010, this line, as appeared in their diffraction patters, is distinctly different from other lines of iron -the "(002)" appears only as spots and only upon heating (Fig. 2).The position of this "(002)" spotty line does not agree well with positions of other, ( 100) and ( 101), continues lines (as appeared in the 2D images, Fig. 2) of hcp-Fe.For example, for the only high-P,T diffraction pattern presented by Tateno et al., 2010, the value of the c lattice parameter calculated from the (100) and ( 101) reflections is 3.4351 Ǻ, while the one obtained from the "(002)" peak is 3.4404 Ǻ that is higher than any other c values reported in Tateno et al., 2010(note, that Tateno et al., 2010 did not use at all the "(002)" line for calculating lattice parameters reported in Table S2).At the same time, if one assumes that spots are coming from the hot part of the sample (i.e. at 4820 K measured spectroradiometrically and 303 GPa as determined from continues diffraction lines of hcp-Fe), the expected value of the (110) reflection of 79 bcc-Fe would be 1.71-1.72Ǻ (the range is due to possible 0-2% difference in volumes of hcp-and bcc-phases of iron (Dubrovinsky et al., 2007) at these conditions) that coincides with the reflection assigned by Tateno et al., 2010 (1) as (002) of hcp-Fe.The quantitative full-profile treatment of the spot-rich part of the pattern confirms this analysis (Fig. 4).

Conclusions
In summary, from the data published by Tateno et al., 2010 we can draw the following conclusions: -Major part of the sample (continues lines of hcp-Fe at 2D X-ray images) probed by X-rays has the same texture before and after laser heating and, thus, was not affected by laser-heating at reported temperatures and pressures; -Heated portion of the sample intensively reacted with carbon forming carbide(s), as reported previously (Prakapenka et al., 2003);

Fig. 1 .
Fig.1.The diffraction pattern of Ni during laser-heating at 280(10) GPa and 3800(100) K (upper panel) and that from the same sample quenched to ambient temperature (300 K) after heating at 285(10) GPa (lower panel).Positions of some spots due to heating at upper panel are highlighted by dashed arrows.As a pressure medium Ne was used, but due to its small amount, comparably low Z, and high pressure, the diffraction lines of Ne are almost invisible.
Fig. 2. X-ray diffraction images presented by Tateno et al., 2010 aligned to each other according to Kikuchi lines.Despite claimed huge difference in temperature the appearance and the texture of continues lines from hcp-Fe are the same (some obvious futures are marked by arrows).