Paleomagnetism provides a unique means to probe Earth’s deep interior over the billions of years across which rocks preserve primary magnetizations. Estimates of the strength of the Earth’s magnetic field (its paleointensity) are essential to understanding long-term changes in the geodynamo. Therefore, accurate and broad paleointensity surveys are needed across the entirety of the rock record.
Initially, just a single step was used to replace the specimen’s magnetic field with a field of known strength (Koenigsberger 1936). Thereafter, the method was expanded to include multiple steps and a best-fit line for a more robust estimate (Thellier and Thellier 1959). Next, checks for thermochemical alteration (pTRM checks) were added because alteration may not always cause a sharp change in the slope of the Arai plot data [e.g., Coe (1967)]. Thereafter, checks for non-single domain behavior (pTRM tail checks) were added because their non-ideal behavior can cause sagging (concave-up) curvilinear Arai plots (Riisager and Riisager 2001). The addition of each of these steps increased the required amount of time to complete the experiment, but aimed to improve the fidelity of the resulting data.
The problem with large magnetic grains stems from their non-unique lowest energy domain configurations. Large grains contain multiple magnetic domains, whose domain walls deteriorate and then reform each time the grain is remagnetized [e.g., Butler (1992)]. These domain walls do not necessarily reform into the same configuration, so precisely replicating the previous domain state is nearly impossible. Absolute paleointensity methods require the excitement of magnetic grains to remove and replace their natural remanent magnetization (NRM) with a known field (Thellier and Thellier 1959), giving them a thermoremanent magnetization (TRM). Since multi-domain grains cannot necessarily be returned to the same internal domain wall configuration, the data extracted from paleointensity experiments containing multi-domain grains are, at best, not necessarily fully reproducible and, at worst, severely biased and unusable.
Hodgson et al. (2018) showed that repeatedly heating synthetic specimens containing interacting single domain or multi-domain (MD) grains can cause increased deviation from ideal single domain (SD) behavior. The deviation, in turn, causes an Arai plot to become curvilinear over the course of a paleointensity experiment, which is an issue that has been known for over 40 years (Levi 1977). Hodgson et al. (2018) used the Coe (1967) variant of the Thellier protocol, which consistently gives concave-up Arai plots (Nagata et al. 1963) when non-SD effects are strong (e.g., Shcherbakova et al. (2000)). The result can be a paleointensity overestimate if the low-temperature portion of an Arai plot is used (Coe et al. 2004; Dunlop and Ozdemir 2001; Xu and Dunlop 2004) and an underestimate if the high-temperature portion of an Arai plot is used (Biggin and Thomas 2003; Dunlop et al. 2005; Shcherbakov and Shcherbakova 2001).
One of the major findings in Hodgson et al. (2018) was that the Arai plot straightened out in the temperature region approximately 20 °C below a given specimen’s Curie temperature. Their Fig. 12 is reproduced in Fig. 1.
In addition, Hodgson et al. (2018) showed non-SD effects can be reduced by minimizing the number of heatings and restricting the experiment to only temperatures near the Curie temperature (TC) of the specimens. The specimens in Hodgson et al. (2018) were artificially created from naturally occurring magnetite and oxyexsolved titanomagnetite. As a result, these specimens had well-defined Curie temperatures and rock magnetic properties, which made doing high-temperature paleointensity experiments relatively straightforward. This paper develops and tests the method proposed by Hodgson et al. (2018) using natural basalt specimens from the well-studied Scientific Observation Hole (SOH1) bore hole from the island of Hawaiʻi.
In the purest form of Hodgson et al. (2018)’s method, only temperatures in the range \(T \in \left[ {T_{C} - 20} \right., \left. {T_{C} } \right]\) are used, and alteration checks are omitted to minimize the number of heatings. However, the lack of systematic pTRM checks means that alteration cannot be detected, which would limit the reliability—as expressed by, for example, the QPI score (Biggin and Paterson 2014) of these experiments. This technique will be tested further herein by including systematic alteration tests (pTRM checks) in the experimental protocol.
The purpose of this current study is to investigate the aforementioned Hodgson et al. (2018) high-temperature paleointensity method by taking the Coe (1967) paleointensity method to its extreme. This study therefore undertook experiments with a range of initial temperatures, from a relatively standard 200 °C to a maximum of 560 °C.