Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 We designed an Isothermal Remanent Magnetization (IRM) acquisition system based on permanent magnets and sized to accommodate an amagnetic hydrostatic pressure cell. This pressure cell fits in a superconducting rock magnetometer, allowing for the measurement of remanent magnetization of pressurized samples. With this system, we determined the coercivity of remanence (Bcr) at different hydrostatic pressures up to 1.4 GPa for rock and dispersed mineral samples with various magnetic mineralogy and domain state. IRM and Bcr are nearly identical before compression and after decompression, indicating no permanent changes in the magnetic properties during pressure cycling. Hydrostatic pressure up to 1.4 GPa does not significantly increases IRM under pressure except for multidomain pyrrhotite and magnetite which show an increase of about 40%. Relative increase of Bcr under pressure is mild, except for a near single domain titanomagnetite where Bcr doubles.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Investigating the magnetic properties of minerals under high pressures has two main goals. First, during high velocity collision, the instantaneous high pressure produced by the impact can change the magnetic properties of magnetic minerals within the impacted body. Many studies have shown the effects of impact on the target materials magnetization [e.g., Gattacceca et al., 2007]. In addition to magnetic mineral deformation, the temporary change of magnetic properties will induce the acquisition of a pressure-induced remanent magnetization in the ambient field and a demagnetization of previous natural remanent magnetization. Consequently, this will influence the characteristics of the magnetic anomaly produced on the planetary surface by the impact. Impact pressures can reach over 100 GPa, but at such high pressure, thermal effects likely erase the pressure effects. On the other hand, impact pressures of less than 5 GPa produce a negligible thermal effect and little deformation [Melosh, 1989]. Second, modeling of large scale magnetic anomalies measured by satellite around Earth and Mars implies deeply buried (e.g., up to 50 km [see Arkani-Hamed and Strangway, 1987; Frost and Shive, 1986; Voorhies et al., 2002]) magnetic bodies. Their magnetic properties and natural magnetization probably differ from those measured at atmospheric pressure. The effect of temperature was already considered [e.g., Dunlop and Arkani-Hamed, 2005] but little was done for taking hydrostatic pressures into consideration. The range of relevant pressure on Earth is up to 1.5 GPa, i.e., the lithostatic pressure at 50 km depth. A number of studies have been devoted to the effect of uniaxial pressure in a lower pressure range [e.g., Pozzi, 1975; Revol et al., 1977; Martin and Noel, 1988], but the magnetic effect of deviatoric pressure is something rather independent of the hydrostatic pressure effect [Nagata, 1966]. Moreover, as soon as deviatoric stress is used, nonelastic irreversible strain may be generated, leading to other mechanisms of magnetization changes [e.g., Borradaile, 1997].
 Up to now, studies led on the effects of high pressures on magnetic properties have used diamond [Gilder et al., 2002] or moissanite anvil cell [Gilder et al., 2004; Gilder and Le Goff2008; Gilder et al., 2011], up to 6 GPa. Despite characteristics close to hydrostatic conditions, the use of anvil cells with a solid confining media implies deviatoric stresses which necessarily produce magnetic mineral deformation. The existence of nonhydrostatic deviatoric stresses is evidenced in these experiments by the dependence of the results on the maximum compression axis [Gilder and Le Goff, 2008] and by direct pressure measurements [Gilder et al., 2011]. This deviatoric component has been quantified for diamond anvil cells: it may reach 10% of the pressure at 1 GPa [Wu and Basset, 1993]. Although deviatoric stress and strain are also possible in deep crustal conditions, the much lower values of maximum deviatoric stress and strain rate, as well as the healing effect of temperature on strain-induced defects indicate that purely hydrostatic experiments may be a better analog for natural in situ conditions.
 The present study presents the production and measurement of isothermal remanent magnetization (IRM) for different samples in a pressure cell with liquid confining media at pressures up to 1.4 GPa, as well as stepwise backfield demagnetization to determine the coercivity of the remanence. Our goal is to characterize the effect of pressure on rock magnetic properties avoiding irreversible magnetic mineral deformation.
2 Material and Experimental Protocol
 From the set of samples previously investigated by Bezaeva et al. , we selected those containing pseudo-single domain (PSD) titanomagnetite (bba) composed of two equal phases with different substitution rate (0.8 and 0.6), PSD titanomagnetite (be-3a) with a substitution rate of 0.5, near single domain (SD) titanomagnetite (kil-2) with a substitution rate of 0.6, multidomain (MD) magnetite (est), and MD pyrrhotite (pyr-a). For MD hematite (BG-8603), we selected a sample of banded iron formation (Ouabego, M., Y. Quesnel, P. Rochette, F. Demory, E. M. Fozing, T. Njanko, J.-C. Hippolyte, and P. Affaton, Rock magnetic investigation of possible sources of the Bangui magnetic anomaly, submitted to Physics of the Earth and Planetary Interiors, 2013). Hematite was confirmed by a large Morin transition and typical hysteresis. Except pyr-a, which is a sized fraction from Dekkers  embedded in epoxy, all other selected samples are natural rocks dominated by a well characterized magnetic mineralogy [Bezaeva et al., 2010, and references therein]. All samples (except BG-8603) were already subject to hydrostatic pressure (up to 1.2 GPa) demagnetization of IRM acquired at atmospheric pressure [Bezaeva et al., 2010]. The possible effects of this preliminary study are discussed in section 4. All samples are soft enough to be nearly saturated in 260 mT. The initial data of Bezaeva et al. , the description of samples, and the results of the present study are compiled in table S1 presented in supporting information.
 To apply hydrostatic pressures, samples were inserted in a nonmagnetic pressure cell containing a liquid pressure transmitting medium (polyethilsiloxane liquid PES-1) providing effective hydrostatic conditions up to 1.8 GPa at room temperature [Kirichenko et al., 2005; Sadykov et al., 2008]. Samples have variable shape (some are roughly spherical and some are cylindrical; see masses in the table in supporting information) and occupy much less than 50% of chamber volume. The fact that the samples are fully surrounded by a liquid phase ensures that they are submitted at their borders to pure hydrostatic pressure. As the samples are polyphasic assemblage of solid phases with different and eventually anisotropic compressibilities, some deviatoric stresses may exist at microscopic scale. Therefore, the pure hydrostatic conditions refer to the pressure applied on the bulk sample, not necessarily to the microscopic scale. The pressure was applied following the protocol of Bezaeva et al.  using a SPECAC press and a piston of 7 mm diameter. The pressure was cross-calibrated with another cell that was calibrated using a manganin sensor [Sadykov et al., 2008]. For 1 ton of applied force, the effective pressure in the cell is 0.23 GPa. The maximum loading in the present study is 6 tons, with a corresponding pressure of 1.4 GPa. Considering the unknown repeatability of the calibration versus loading condition and the precision of the press gauge, we estimate that uncertainty on pressure is about 5%. To impart IRM to the samples under pressure in the cell, we used an aluminum frame holding face to face two mobile rare earth magnets of 40 mm diameter (see Figure S1). The pressure cell, guided by the plastic rails, is placed at the center of the frame with the magnets initially placed at 24 cm from the center. In this configuration, the residual field in the center does not exceed 1 mT (Figure 1a). The distance between magnets is then reduced to 4 cm to produce a magnetic field of 260 mT at the center (Figure 1a). At closest position, the volume in which the magnetic field is homogeneous (i.e., between 260 and 270 mT) is 5 mm large and 8 mm high (Figure 1b). The magnets are then moved away back to their initial position and the pressure cell removed from the system. Field gradients in the sample zone decreases rapidly as a function of magnet distance. The hydrostatic chamber is 6.9 mm large and 20 mm high, and the samples, much smaller than the chamber (the biggest sample is 8 mm high for 5 mm large and its long axis is placed in vertical position), are centered. Therefore, all samples are affected by a homogeneous field within the cell.
 To demagnetize IRM260mT, the pressure cell was again placed in the center of the aluminum frame but in the opposite direction to the IRM260mT acquisition, and the distance between magnets was decreased step by step. IRM260mT and its stepwise demagnetization were measured by placing the hydrostatic pressure cell in a superconducting rock magnetometer with a moment sensitivity of ~10−11 Am2 (SRM760R, 2G Enterprises). A Micromag vibrating sample magnetometer (VSM) with 1 T maximum field was used for hysteresis measurements after decompression (see Table S1). All experiments were carried out at Centre Européen de Recherche et d'Enseignement de Géosciences de l'Environnement.
 Acquisition and stepwise demagnetization of the IRM260mT were measured until the acquisition of the IRM260mT in the opposite direction (Figure 2, left). For each sample measurement, the signal of the empty cell was subtracted. IRM curves for the empty cell are quite constant with pressure. IRM of the cell reaches a maximum of 1.88 10−6 Am2 at 260 mT (Figure 2, left) that corresponds to 23.8 % of the signal of the weakest sample (MD magnetite est, Table S1). A first quality estimate was given by the absolute value of the IRM260mT which has to be the same in both directions of acquisition. The quality of the demagnetization procedure was also estimated using stereographic projection of the resulting magnetic vectors showing the antipodality of the IRM260mT acquired in the opposite field (Figure 2, right). Bcr could therefore be calculated from the demagnetization curve with good precision, as quantified by standard deviations calculated from repeated procedures.
 All samples except BG8603 were subjected to two compression-decompression experiments: the first one up to 1.2 GPa by Bezaeva et al.  and the second one up to 1.4 GPa in the present study. The constant magnetic characteristics before the first compression and after the last decompression confirm that repeated hydrostatic pressure experiments up to 1.4 GPa do not affect irreversibly the magnetic properties, at odds with Gilder and Le Goff  and Gilder et al.  experiments for similar maximum pressures (Figure 3). Most of the changes observed on IRM intensity can be explained by the anisotropy of the samples which is quantified by a standard deviation resulting from IRM measurements and Bcr calculations performed at final stage in two or three perpendicular directions for all samples (see Table S1). The largest relative change reaches 11%. It concerns the IRM of the pyrrhotite sample (pyr-a) and may be attributed to its anisotropy reaching 20%. Few changes (none of them exceeding 10% of the IRM) can also be due to a slight loss of material related to friability of some samples (this is probably the case for hematite sample BG-8603). Concerning Bcr, only Be-3a presents a significant change with an increase of 25% mostly acquired during the compression-decompression experiment performed by Bezaeva et al. . As the sample is soft, the absolute change is only of 4.5 mT.
 The evolution of IRM260mT with pressure shows complex behavior (Figure 4a). Mild variation (<10%) is shown by hematite (decrease) and PSD titanomagnetite (increase). SD titanomagnetite (Kil-2) shows first a decrease of the IRM260mT of about 25% at 0.45 GPa, and then an increase at steps 0.9 and 1.4 GPa up to values equivalent to the magnetization before pressure experiment. This nonmonotonic behavior is unique among our data set. Interestingly, the same sample was the only one to show a distinct sigmoidal shape of IRM demagnetization curve, with an inflection point near 0.4 GPa [Bezaeva et al., 2010]. This may sign the competition of two pressure sensitive processes. MD magnetite (est) and pyrrhotite samples display a large progressive increase of IRM260mT up to 40–50% of the original value at 1.4 GPa.
 All samples show a regular increase of coercivity under pressure (Figure 4b), with the least effect shown by hematite followed by MD magnetite and PSD titanomagnetite samples, all with relative increases at 1.4 GPa below 20%. Larger effects are observed for pyrrhotite (40%) and SD titanomagnetite (110%).
 The use of a hydrostatic pressure cell prevents irreversible deformation of the sample so that initial rock magnetic characteristics are preserved after decompression (Figure 3). This result is important because deformation may cause strong changes in rock magnetic characteristics [e.g., Gattacceca et al., 2007]. This nonhydrostaticity is discussed in a recent paper [Gilder et al., 2011] and likely explains the strong and nonreversible increase of the IRM and coercivity measured after decompression, as demonstrated in Figure 3.
 Under pressure, the magnetic signature of deformation as well as of deviatoric stress cannot be separated from the effect of the hydrostatic pressure itself, thus rendering the interpretation of previous experiments with solid confinement ambiguous. For MD magnetite and pyrrhotite, we qualitatively reproduced the increase of IRM and Bcr under pressure reported by Gilder and Le Goff  and Gilder et al. , but to a lesser extent. The discrepancy is, for example, obvious for the IRM of some titanomagnetite and the Bcr of pyrrhotite (Figures 4c and 4d). The increase in coercivity is attributed to changes in the crystalline anisotropy and therefore in the magnetostrictive anisotropy [Kamimura et al., 1992; Gilder et al., 2011]. The present study shows that part of this change is reversible and purely due to the effect of hydrostatic pressure. Another difference between the two sets of experiments for (titano-)magnetite is that Gilder and Le Goff  used highly concentrated assemblages of magnetic crystals, while our experiments on rocks ensure a much lower potential effect of magnetic interactions, which are known to strongly affect coercivity and remanence for magnetite [Sugiura, 1979].
 Our PSD titanomagnetites (bb-a and be-3a) display constant IRM260mT whatever the hydrostatic pressure applied. This result differs strongly from pressure experiments conducted by Gilder and Le Goff , reporting an increase of the IRM by a factor up to three for MD titanomagnetite in equivalent pressure applications. This difference may be again partially attributed to deviatoric stress and strain. In addition, the present study deals with pseudo single domains whereas Gilder and Le Goff  deal with multidomains so that changes of domain state during under pressure could play a role in the magnetization enhancement. Magnetic domain reorganization may explain the increase of Bcr of about 10% observed in the present study.
 The increase by a factor two of Bcr for SD titanomagnetite (kil-2) is coherent with the results of Gilder et al. . On the other hand, the present study shows first a decrease at low pressure and an IRM260mT at 1.4 GPa equivalent to the IRM260mT at 0 GPa whereas Gilder et al.  report a strong increase of more than twice the original magnetization at 1.5 GPa. This strong increase may again be related to magnetic mineral deformation as the magnetization after decompression is 60% to 80% higher than magnetization before the experiments of Gilder et al. .
 The slight decrease of IRM260mT (Figure 4a) for the MD hematite sample (BG-8603), at odds with the behavior of MD magnetite and pyrrhotite, may be related to the Morin transition. Indeed, it has been observed that the temperature of the Morin transition increases with pressure of about 30°C per GPa [Parise et al., 2006]. The measurements performed at 1.4 GPa and at room temperature are close to the transition from antiferromagnetism to paramagnetism, that would explain the decrease of the IRM260mT, pressure application being somehow equivalent to cooling near the Morin transition [Özdemir et al., 2008]. However, the relationship between pressure and Morin transition temperature needs to be better constrained (ongoing study with our pressure cell).
 The pressure experiments conducted using a liquid confining media pressure cell up to 1.4 GPa show that magnetic properties come back to their original values after decompression, contrary to previous studies conducted with solid confining media. Since this procedure avoids permanent deformation and deviatoric stress, the evolution of magnetic properties is therefore only induced by the effect of hydrostatic pressure and is thus a better analog of the in situ conditions at large depths. We confirm the general increase of remanent coercivity under pressure previously observed on magnetite, titanomagnetite, and pyrrhotite, but to a much lower extent than observed in nonhydrostatic pressure: this increase remains below 20% up to 1.4 GPa (except for MD pyrrhotite and SD titanomagnetite, with 30 and 110%, respectively). Saturation remanence increases by more than 5% only for MD magnetite and pyrrhotite (up to 50%), but again much less than in previous experiments. Still, these changes must be taken into account for estimating the deep in situ magnetizations. However, one needs also to predict the combined effect of pressure and temperature. These increases support again the need for taking into account geologically stable remanence in deep crustal magnetization modeling [Shive, 1989; McEnroe et al., 2004]. Hematite, studied for the first time at such pressures, as well as SD titanomagnetite, may show a decrease of saturation remanence under pressure. Our results are also the first obtained in that pressure range on rocks with diluted magnetic grains rather than on assemblages of strongly interacting crystals.
 We thank M. Fuller and S. Gilder for their reviews, which helped to improve the submitted manuscript.
 The Editor thanks an anonymous reviewer and Stuart Gilder for their assistance in evaluating this paper.