Paleointensity of the ancient Martian magnetic field


  • Benjamin P. Weiss,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
    Search for more papers by this author
  • Luis E. Fong,

    1. Department of Physics and Astronomy, Vanderbilt University, Nashville, Tennessee, USA
    Search for more papers by this author
  • Hojatollah Vali,

    1. Department of Anatomy and Cell Biology and Facility for Electron Microscopy Research, McGill University, Montreal, Quebec, Canada
    2. Department of Earth and Planetary Sciences, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Eduardo A. Lima,

    1. Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
    Search for more papers by this author
  • Franz J. Baudenbacher

    1. Department of Physics and Astronomy, Vanderbilt University, Nashville, Tennessee, USA
    2. Department of Biomedical Engineering, Vanderbilt University, Nashville, Tennessee, USA
    Search for more papers by this author


[1] Mars today has no core dynamo magnetic field. However, the discovery of remanent magnetization in Martian meteorites and intense crustal magnetization suggests that Mars once had a global field. Here we present high resolution maps of the magnetic field of Martian meteorite ALH 84001. These maps are the most sensitive yet quantitative study of natural remanent magnetization (with resolved anomalies as weak as 1 × 10−14 Am2). ALH 84001 likely contains a 4 billion year old (Ga) thermoremanence partially overprinted by one or more poorly understood secondary components. Our data suggest that the paleointensity of the local paleofield was within an order of magnitude of that of the present-day Earth. If this field were global in extent, it should have played a key role in Martian atmospheric and climatic evolution. However, it is still too weak to easily explain the intensity of Martian crustal paleomagnetic anomalies.

1. Introduction

[2] The intensities of surface magnetic fields on Mars during most of the planet's history are essentially unknown. Magnetic anomalies reside primarily in crustal terranes >3.8 billion years old (Ga) and are weak or absent within younger impact basins. These data have been used to argue that an early field had decayed by ∼3.9 Ga [Acuna et al., 1999], but the low spatial resolution of spacecraft field data combined with the lack of precise ages for the crustal sources make this timing highly uncertain. Understanding the history of Martian fields is critical for understanding the planet's thermal, tectonic, and climatic evolution.

[3] Of the nearly 50 known Martian meteorites, only a single sample—the meteorite ALH 84001—dates back to the early field epoch. ALH 84001 indeed contains a natural remanent magnetization (NRM). The NRM resides mainly in single domain stoichiometric magnetite- and pyrrhotite-bearing carbonate [Antretter et al., 2003; Rochette et al., 2005; Weiss et al., 2002a; Weiss et al., 2004], as well as possibly carbonate-associated hematite [Steele et al., 2007] and free magnetite [Steele et al., 2007], sulfides [Kirschvink et al., 1997] and chromite [Weiss et al., 2002a]. All paleomagnetic studies of ALH 84001 have found that the NRM is heterogeneously oriented at fine scales [Antretter and Fuller, 2002; Antretter et al., 2003; Collinson, 1997; Kirschvink et al., 1997; Weiss et al., 2000; Weiss et al., 2002b]. Bulk (tens of mg and larger) samples do not show significant remanence or susceptibility anisotropy [Collinson, 1997; Gattacceca et al., 2005; Weiss et al., 2000].

[4] 40Ar/39Ar data suggest that the characteristic NRM in ALH 84001, which has been observed as an origin-trending component blocked above coercivities of ∼40 mT [Antretter et al., 2003; Gattacceca and Rochette, 2004], was likely acquired as a thermoremanence on Mars during a shock event that strongly heated the meteorite at ∼4 Ga [Shuster and Weiss, 2005; Weiss et al., 2002a]. Noble gas geochronology and a variety of geochemical and petrographic data collectively indicate that the meteorite has not been heated to temperatures above ∼500°C since this time. The heterogeneous pattern of magnetization and evidence for one or more shock events following carbonate formation suggest that it likely later acquired secondary overprints (see auxiliary material). The characteristic NRM could have been the product of either a dynamo or local crustal paleofield. However, given that any strong crustal paleofields would likely have been the product of an earlier dynamo, this suggests that a dynamo existed on Mars at or before 3.9 Ga.

[5] Although the age of the characteristic NRM in ALH 84001 is reasonably well known, the intensity of the paleofield that magnetized it is not. Previous experiments on the meteorite using the ratio of the NRM to saturation isothermal remanent magnetization (sIRM) method have obtained paleointensity values ranging from several to several tens of microteslas (μT) (for comparison, the Earth's surface field today has an intensity of ∼50 μT), with a striking dependence on the mass of the sample used for the analysis (Figure S1). Although the NRM/sIRM method should only yield an order-of-magnitude estimate of the paleointensity even for single-component NRM (see auxiliary material) [Weiss et al., 2007a], a key reason for this large range of values must also be the nonunidirectional orientation of the fine scale magnetization in ALH 84001 [Weiss et al., 2000; Weiss et al., 2002b], which would mean that measurements of bulk grains provide only a lower limit on the true paleointensity. It is therefore possible that the Martian paleofield at 4 Ga was in fact significantly stronger than that of the Earth today. A strong field would then make it much easier to explain the intensity of the Martian crustal anomalies, which are ten times as intense as those in the Earth's continental crust [Connerney et al., 1999].

[6] At the outset of this study, we hypothesized that this mass dependence of paleointensity estimates is not accidental but is rather causal. If so, then there are at least two possible reasons why small samples would give higher paleointensities: (a) the smaller samples have larger magnetic anisotropy which could result in an overestimate of their paleointensities [Selkin et al., 2000] (and possibly also explain the nonunidirectional NRM [Antretter et al., 2003]), or (b) the small samples are magnetically isotropic and the net NRM of large samples underestimates the true fine-scale thermoremanent NRM and therefore the true paleointensity.

[7] To distinguish between hypotheses (a) and (b), we used the low-transition temperature Superconducting Quantum Interference Device Microscope (SM) [Fong et al., 2005; Weiss et al., 2007a, 2007b] to resolve the fine-scale heterogeneity of the magnetization and obtain more accurate paleointensity estimates from resolved anomalies. The masses of these anomalies are only several hundred nanograms and they contain millions of ferromagnetic crystallites. These maps afford us the opportunity to study the paleomagnetism of samples with effective masses ∼10,000 times less than that of the next largest samples previously analyzed.

2. Methods

[8] All of the magnetic data presented here were acquired using the SQUID Microscope, a low-transition temperature magnetometer that maps the vertical component of the magnetic fields of room temperature samples [Weiss et al., 2007b]. The SM was equipped with a monolithic DC SQUID sensor with an effective diameter of 80 μm and was rastered over the sample at a constant height of 140 μm to map the field at a rectangular grid of thousands of locations spaced by 40 μm. All data were acquired inside a two-layer mu-metal shielded room (DC field <50 nT and AC field peak-to-peak variations <0.05 nT at 0.01 Hz). In this configuration, the SM can resolve individual dipoles with a moment resolution of ∼10−15 Am2 and a spatial resolution of ∼140 μm. Unlike standard SQUID moment magnetometers, the SM does not measure the sample's bulk moment but rather maps its magnetic field in order to resolve magnetization variation within the sample. This requires that the sample magnetization be constrained from least squares fits to the field data.

[9] We used the SM to map ALH 84001,227b,2, a 30 μm thin section taken from directly underneath thin section 227b,1 that we previously studied using a lower resolution (250 μm) SM [Weiss et al., 2002b]. The thin section represents a partial cross section of the parent meteorite (Figure 1), beginning with a semi-continuous rim of fusion crust at left produced during passage through the Earth's atmosphere and proceeding to the meteorite's interior unaffected by atmospheric heating at right.

Figure 1.

Natural remanent magnetization (NRM) field of ALH 84001. (a) Reflected light photograph of 30 μm thin section 227b,2 showing pyroxene (light brown), fusion crust along left side (black) and chromite (black interior grains). (b) Vertical component of the NRM field as measured 140 μm above the sample. Positive (out-of-the-page) fields are red and yellow and negative (into-the-page) fields are blue. The field is dominated by the fusion crust. (c) Same as Figure 1b except with color scale stretched to show weak magnetism of meteorite interior. Approximate boundary of baked zone is shown by dashed line.

[10] Using the SM, we imaged the evolution of the NRM during progressive three-axis AF demagnetization from 5 to 100 mT in steps of 5 or 10 mT. We also measured stepwise IRM acquisition in the out-of-the-thin-section-plane direction (IRMz) from 20 to 545 mT, as well as IRM 545 mT in the two in-plane directions (IRMx and IRMy) (see auxiliary material). Following the magnetic analyses, we imaged the sample using optical and electron microscopy at McGill University (see auxiliary material).

3. Measurements

[11] SM measurements of the NRM revealed three main zones of magnetization associated with the fusion crust, an adjacent baked zone, and the meteorite interior (Figure 1). An intense, relatively uniform field oriented into the plane of the thin section is present above the fusion crust at left and extends ∼0.1 mm into the interior, surrounded by a belt of oppositely oriented (“return flow”) field. Above the meteorite's deep interior at right are much weaker fields with higher spatial frequencies and multiple zero-crossings. Located between the fusion crust and deep interior zones and extending ∼2 mm into the meteorite is a region exhibiting intermediate behavior: like the deep interior, it is unmelted and has weak magnetic anomalies associated with carbonate and chromite, but like the fusion crust, individual anomalies are concentrically zoned with interior downward fields ringed by belts of upward fields.

[12] Fields above the interior are often spatially correlated with carbonate and chromite grains. Our electron microscopy and IRM data (see auxiliary material) indicate that pyrrhotite and magnetite intimately associated with the nonmagnetic macroscopic carbonate and chromite are the actual sources of this magnetization. In particular, high-resolution transmission electron microscopy analysis of an ultrathin section prepared by focused ion beam milling from an individual grain of chromite (Figure S3) detected enclosed sulfide containing lattice fringes with equation image-spacings indicative of monoclinic pyrrhotite (see auxiliary material).

[13] We conducted separate unidirectional least squares fits to both the full NRM scan as well as to the interior-only (no fusion crust) portion (see auxiliary material). From both inversions (data not shown), we obtained residuals that far exceeded our measurement noise, confirming that the meteorite is not unidirectionally magnetized. To characterize the NRM pattern in more detail, we then selected 41 relatively isolated anomalies (Figure S4) and conducted unidirectional least squares fits to solve for each anomaly's magnetization intensity and direction. NRM intensities from individual anomalies range from 4 × 10−14 Am2 (deep interior anomaly 41) to 6 × 10−10 Am2 (fusion crust anomaly 14). As expected, the fit NRM directions are widely scattered (Figure S9a).

[14] The two fusion crust anomalies (3 and 14) collectively carry the majority of the magnetization in the sample. A uniformity test [Watson, 1956] demonstrates that the remaining 39 anomalies are nonrandomly distributed to >99% confidence and are preferentially oriented in the direction of the fusion crust. However, when the 16 anomalies from the fusion crust and intermediate zone (within 3 mm of the fusion crust), which are mainly magnetized downward like the fusion crust, are excluded from the population, randomness cannot be rejected with 99% confidence (Figure S9a). This result serves as a joint conglomerate and baked contact test, demonstrating that the deep interior of the meteorite has a preterrestrial NRM that survived atmospheric passage, sample handling by NASA, and the production of the thin section. By analogy with our previous SM studies of ALH 84001, the fusion crust magnetization is a thermoremanence acquired during atmospheric passage in the Earth's field. Its great intensity relative to the interior is mainly due to the relatively high concentration of cryptocrystalline magnetite that forms in the fusion crusts of stony meteorites [Genge and Grady, 1999]. Previous studies have shown that thermal remagnetization from atmospheric passage extends from 0.1 to several mm into the unmelted interiors of stony meteorites [Butler, 1972; Sugiura and Strangway, 1983; Weiss et al., 2000; Weiss et al., 2002b].

[15] The sample field as imaged during progressive AF demagnetization (Figure 2) was broadly similar to the NRM, with intense, downward fields above the fusion crust and weaker, heterogeneously oriented fields above carbonates and chromite in the interior. As AF demagnetization progressed to higher peak fields, the intensity of the fields decreased and the magnetization weakened. Again, our inversions (Figure 3) show that the meteorite cannot be unidirectionally magnetized even after AF demagnetization to 5–30 mT (Figures S9bS9e). By AF 40 mT (Figure S9fS9l), many anomalies (including essentially all those dominated by magnetite; see Figures 3d and 3e) have been significantly demagnetized and are either too weak to yield high fidelity inversions or else have directions dominated by spurious remanence induced by our AF system. Weak NRM after AF 99 mT remained in anomalies dominated by pyrrhotite, as expected for this high-coercivity mineral. Such high stability has also been observed from studies of bulk ALH 84001 grains [Antretter et al., 2003; Collinson, 1997; Gattacceca and Rochette, 2004; Kirschvink et al., 1997]. By AF 99 mT, the intensity of anomaly 41 had dropped to 1.2 × 10−14 Am2. This is by nearly two orders of magnitude the weakest measured NRM in the history of paleomagnetism.

Figure 2.

Evolution of natural remanent magnetization field of ALH 84001 during progressive three axis alternating field (AF) demagnetization. Shown is the vertical component of the field as measured 140 μm above the sample; positive (out-of-the-page) fields are red and yellow and negative (into-the-page) fields are blue. All images share 1 mm scale bar and intensity scale at lower right. (a) After a peak AF of 5 mT. (b) AF 10 mT. (c) AF 20 mT. (d) AF 30 mT. (e) AF 40 mT. (f) AF 50 mT. (g) AF 60 mT. (h) AF 70 mT. (i) AF 80 mT. (j) AF 90 mT. (k) AF 99 mT.

Figure 3.

Evolution of natural remanent magnetization (NRM) during alternating field (AF) demagnetization of selected anomalies as inferred from least squares inversion of SQUID Microscope maps of ALH 84001 (Figures 1 and 2). Closed (open) symbols represent end points of magnetization projected onto horizontal (vertical) planes. Shown for each anomaly is the NRM and moment after AF demagnetization to 5 mT, 10 mT, 20 mT, 30 mT, 40 mT, 50 mT, 60 mT, 70 mT, 80 mT, 90 mT, and 99 mT. (a) Anomaly 1. (b) Anomaly 15. (c) Anomaly 16. (d) Anomaly 19. (e) Anomaly 20. (f) Anomaly 29. The scatter in moments visible at high AF steps (e.g., anomaly 20 at AF > 20 mT) is due to a combination of spurious remanence induced by our AF system and uncertainties in the magnetization fit due to surrounding magnetization.

[16] Using these data, we calculated 41 individual NRM/sIRM paleointensities for the thin section. The fusion crust and baked zone have Earth-strength paleointensities, giving a median value of 34 μT. This is as expected given that the magnetization in these zones was acquired at 13 ka during landing on Earth [Jull et al., 1995]. The interior anomalies have a median paleointensity of 46 μT, with 95% of the anomalies having a paleointensity within a factor 6 of the median (Figure S10). As observed for a SM paleointensity study of terrestrial basalt [Weiss et al., 2007a], there is a sparsely populated high-field tail to the paleointensity distribution which may be a reflection of the expected variability of grain sizes and volumes amongst the anomalies. Our SQUID microscopy study of anisotropy of IRM (see auxiliary material) found no evidence that the anomalies have strong magnetic anisotropy that might bias these paleointensity estimates or be responsible for the heterogeneous orientation of their NRM directions. Assuming that the heterogeneous magnetization in the meteorite is the product of early shock events (see auxiliary material), these data would suggest that a field of order Earth-strength magnetized the interior of ALH 84001 at 4 Ga.

4. Conclusions

[17] Our new paleointensity estimate matches those derived from the smallest previously studied samples [Gattacceca and Rochette, 2004; Kirschvink et al., 1997]. This indicates that our study has captured (indeed, well exceeded) the smallest characteristic spatial scale of the NRM and that our paleointensity estimates are not lower limits in this sense. The main limitation of our analysis is that we have obtained total moment paleointensities rather than paleointensities for each component of magnetization. On the other hand, paleointensities derived from multicomponent and single component anomalies have median values that are well within our uncertainties (38 and 64 μT, respectively). We also note that a multicomponent NRM/sIRM study on bulk ALH 84001 grains [Gattacceca and Rochette, 2004] measured a characteristic component paleointensity within a factor of 2 of our single component median paleointensity.

[18] We conclude that within the uncertainties of the NRM/sIRM method, our paleointensities are accurate indicators of the ancient Martian paleofield. This ∼50 μT field was most likely the product of either an active core dynamo or crustal magnetization that was the product of a previous dynamo that had since decayed. Even if such a field was responsible for magnetizing the Martian crust, the large intensity of the crustal anomalies would still have required surprisingly large concentrations of ferromagnetic minerals (perhaps 10 times that of the Earth's continents). Regardless of whether the field was of local or global extent, it would likely have shielded the local Martian atmosphere from loss due to impingement of the solar wind. The end of the early Martian dynamo may therefore have played an important role in climate change on Mars.


[19] We thank the NSF Geophysics and Instrumentation and Facilities Programs and the NASA Mars Fundamental Research and Planetary Major Equipment Programs (B.W.) and the Natural Science and Engineering Research Council (NSERC) of Canada (H.V.) for their support of this work. Two anonymous reviewers provided insightful comments that improved the manuscript.