S. Gilder $^{a}$, M. LeGoff $^{a}$, P. Besson $^{a}$, J. Glen $^{b}$, J. Peyronneau $^{a}$ and J.C. Chervin $^{c}$
$^{a}$ Institut de Physique du Globe de Paris, Paris, France. $^{b}$ U.S. Geological Survey, Menlo Park, USA. $^{c}$ Universite Pierre et Marie Curie, Paris, France.
Although fluid motion in the outer core is commonly thought to generate the Earth's magnetic field, its geometry and secular variation may not rely on fluid mechanics alone. Some numerical models suggest that the stability of the geodynamo also depends on a finitely conducting solid inner core (e.g., Hollerbach and Jones, 1994; Glatzmaier and Roberts, 1995). While the phase of iron in the inner core is a subject of active research, most studies conclude that the hexagonal closed packed (hcp) phase is the best candidate. Karato (1993) proposed that the toroidal field intensity can orient the crystallographic c-axis of hcp Fe parallel to the rotation axis, if the metal is paramagnetic with a certain degree of crystalline anisotropy of magnetic susceptibility. Clement and Stixrude (1995) linked the magnetic characteristics of hcp Fe to field behavior by proposing that deviations from a geocentric axial dipole field, such as the far sided effect and preferred trajectories of virtual geomagnetic pole paths during polarity transitions, can be explained if the inner core is composed of an aggregate of preferentially oriented hcp Fe crystals. Such an aggregate can potentially explain why seismic waves traveling parallel to the rotational axis appear 1-4 percent faster than those traversing equatorial paths. Thus, magnetic properties of hcp Fe may influence magnetic field behavior and the structure of the inner core. Previous studies on the magnetic state of hcp Fe have come to diverse conclusions. Employing electron microscopy and shock wave techniques, Wasilewski (1977) concluded hcp Fe is antiferromagnetic. These findings are consistent with extrapolations of Mossbauer effect and neutron diffraction studies that predict antiferromagnetic ordering (e.g., Ohno and Mekata, 1971; Pearson and Williams, 1979). More recent experiments employing Mössbauer effect measurements did not detect either ferromagnetic or antiferromagnetic ordering (e.g., Taylor et al., 1991). Band structure calculations predict hcp Fe should not order magnetically (Fletcher and Addis, 1974). To help solve this problem, we studied the magnetic properties of Fe at high pressures employing two very different approaches. The first allowed direct visual characterization of the magnetization by immersing Fe particles in a fluid pressure medium then observing the interaction (or lack thereof) of the particles with an external magnet (Gilder and Glen, 1998). Stress in the cell was perfectly hydrostatic, e.g., no stress gradient existed. The attraction of Fe to an induced magnetic field at 17.7 GPa and 262 C suggests hcp Fe exists either in a paramagnetic or ferromagnetic state. Clumping of the particles during motion supports the ferromagnetic option. In the second approach we built a system that measures magnetic hysteresis parameters in a diamond anvil cell. Hysteresis loops of 99 percent pure Fe were determined from 0 to 21 GPa. We find that saturation remanence and maximum susceptibility decrease from 0 to 10 GPa, after which no change is observed. These parameters are irreversible upon pressure release. At first glance the interpretation would be that the transformation from bcc to hcp involves a non-magnetic phase transition, e.g., that hcp Fe is not ferromagnetic. The irreversibility of the magnetic parameters prompted us to observe the iron particles before and after compression using transmission electron microscopy. Images before compression show a coherent crystalline structure. Images after compression, however, show that the original grain size has undergone intense reduction, probably during martensitic phase transformation as seen in hcp analogs (Poirier and Langenhorst, 2002). In other words, the observed change in magnetic properties and their irreversibility is due to the passage from single domain to super paramagnetic grain sizes. Thus, non-hydrostatic experimental conditions, which are the case for almost every diamond cell experiment, will arrive at a conclusion that hcp Fe is non-ferromagnetic. To the best of our knowledge, the only experiment achieved at purely hydrostatic conditions is that of Gilder and Glen (1998). They found either a ferromagnetic or paramagnetic state were possible for hcp Fe, although they could not distinguish between the two. For the paramagnetic case, the minimum susceptibility value (0.001 SI) is very similar to hcp transition metal analogs, and supports the viability of the Clement and Stixrude (1995) model. Maximum estimates would increase the influence of the inner core on geomagnetic field behavior. In addition, the results of Hollerbach and Jones (1993) and Glatzmaier and Roberts (1995) predict that a finitely conducting solid inner core tends to stabilize the geodynamo. This arises because the inner core has a diffusive timescale independent of that of the outer core. We speculate the same could be true if the paramagnetic relaxation time of the inner core lags behind external field changes from the outer core. The outcome is that short term fluctuations are damped out, and the dynamo is steadied. As a result, brief breakdown of the dynamo would not lead to a successful reversal of the field. This inhibition to reversing may result directly from the inner core's stabilizing influence on outer core convection, and is feasible if the estimates are valid at core temperatures.