Despite significant progress in palaeomagnetic research over the last century, the origin, evolution and long-term behaviour of the geomagnetic field remains poorly understood. One significant open question is when and how the inner core nucleated. Since geomagnetic field behaviour is intrinsically linked to the thermal evolution of the core, scientists have turned to the global palaeointensity record to search for proxies for inner core nucleation. From this record, two signals have been identified as possible indicators of inner core nucleation: (1) a spike in magnetic field strength between 1.5–1.0 Ga, and (2) an initially strong, but gradually decreasing field strength that resulted in a weak dynamo in the Ediacaran. Although both these hypotheses are vastly different, they do have one common challenge hindering rigorous testing: a paucity of palaeointensity data. This is especially true for the Precambrian time period for which well-preserved outcrops are scarce and weathering/alteration is nearly inescapable. Despite making up almost 90per cent of Earth's history, data from this super eon comprise<10 per cent of the global palaeointensity database. This lack of data for most of Earth's history represents a considerable gap in our knowledge and greatly impedes our ability to understand the origin and evolution of our planet and its magnetic field. In an effort to fill in this gap, we performed palaeointensity experiments on Precambrian-aged mafic dykes from India (Malani Igneous Suite and Bastar, Dharwar and Bundelkhand Cratons) with ages ranging from ∼740 Ma to ∼2.36 Ga. To monitor thermal alteration and minimize the effects of non-ideal grain sizes, the Thellier method following the IZZI protocol was used. Successful results were obtained for samples from the Bundelkhand (∼740 Ma) and Bastar (∼1.89 Ga) cratons. The Bastar results fall in a ∼40Myr gap in the database and corroborate field trends predicted by the Monte Carlo axial dipole moment model, which suggests that intensity values were moderately low (2–4×1022 Am2) in the middle Palaeoproterozoic. The Bundelkhand result suggests that the field may have been rapidly decaying in the late Tonian to early Cryogenian.
]]>The Precambrian encompasses nearly 90% of Earth's history and was characterized by the formation of the planet and the origin of life. During this period, pivotal geologic processes such as the formation of continents and oceans, the onset of plate tectonics, and the oxygenation of the atmosphere laid the foundation for the planet's subsequent evolutionary trajectory. Therefore, studies focused on this period are vital as they provide crucial context for understanding the planet's long-term evolution and the emergence of complex life. Studies of the geomagnetic field during the Precambrian are intrinsically linked to our understanding of the thermal evolution of the planet, the onset of the geodynamo, and its role in shaping Earth's internal dynamics and long-term geological processes. One significant open question in this area is related to the timing of inner core nucleation (ICN). Knowing the timing of ICN is key to studying the long-term evolution of the Earth as it marked a significant change in the planet's interior.
The proposed timing of ICN varies widely primarily due to uncertainty regarding the physical parameters within Earth's interior, such as the thermal conductivity of the core (Stacey & Loper 2007;de Koker et al. 2012;Pozzo et al. 2012;Gomi et al. 2013;Olson 2013;Driscoll & Davies 2023). Consequently, numerical models yield disparate results, suggesting inner core formation times that span billions of years.
Older models using thermal conductivity values of 60 Wm -1 K -1 predicted ICN at ~3.5 Ga (Gubbins et al. 2004). Conversely, recent models using revised higher thermal conductivity values (90-226 Wm -1 K -1 ) generally favor much younger ages of < 1 Ga, some as young as ~400-500 Ma (Gomi et al. 2013;Davies 2015;Labrosse 2015;Driscoll 2016;Ohta et al. 2016;Landeau et al. 2017;Zhang et al. 2020). However, this becomes problematic when considering that the geomagnetic field is thought to have persisted since at least ~3.5-3.7 Ga (Tarduno et al. 2014;Nichols et al. 2024) as a core with high thermal conductivity would likely be incapable of sustaining a field long-term (Buffett 2003;Gubbins et al. 2003). These higher estimates of thermal conductivity are also inconsistent with experimental data from heated diamond anvil cells, which indicate values as low as 18-44 Wm -1 K -1 (Konôpková et al. 2016). Attempts to reconcile experimental and computational estimates of thermal conductivity suggest that there are issues with both low and high estimates, and true values may be between 75-100 Wm -1 K -1 (Pozzo et al. 2022;Yin et al. 2022). Despite these instrumental attempts to determine the timing of ICN, it still remains unresolved.
To independently estimate the age of the inner core, scientists are exploring data-driven approaches, including the analysis of paleomagnetic records, to identify signatures of ICN. Changes in the adiabatic processes within the churning, liquid metal outer core modulate geomagnetic field strength. Because nucleation of the inner core would have resulted in a change in the convective regime of the interior, some studies suggest its formation would result in an observable change in the field strength at Earth's surface (e.g., Stevenson et al. 1983;Labrosse & Macouin 2003;Aubert et al. 2009;Christensen et al. 2009;Driscoll 2016). Biggin et al. (2015) identified an increase in paleointensity values during the Mesoproterozoic that they postulated was due to ICN. This increase was later contested by Smirnov et al. (2016) where the authors argued that the observed spike in paleointensity was largely due to biased data from two localities which overestimated the field strength. Upon removal of these data, the analyses of Smirnov et al. (2016) suggested that the low number of high-quality paleointensity estimates available at the time were insufficient to constrain ICN timing. Further paleointensity studies of rocks from this time period have increased the number of high-quality data points, but unfortunately still leave open questions about the behavior of the field and its association with ICN. For example, Zhang et al. (2022) and Sprain et al. (2018) reported high intensities of 56-129 × 10 22 Am 2 , which would have required a geodynamo with a strong power source at ~1.1 Ga whereas other studies found the opposite -ultra-low paleointensities ranging from 0.3-1.8 × 10 22 Am 2 at ~1.1 Ga (Lloyd et al. 2021b;Shcherbakova et al. 2022).
Conversely, one time period that provides consistent paleointensities is the Ediacaran. A study by Bono et al. (2019) identified an extremely low paleointensity of ~0.7 × 10 22 Am 2 at ~565 Ma. This finding was the catalyst for further studies on Ediacaran rocks which yielded paleointensities ranging from 0.31-2.25 × 10 22 Am 2 (Shcherbakova et al. 2020;Thallner et al. 2021aThallner et al. , 2021bThallner et al. , 2022)). Bono et al. (2019) identified a weakening trend from initially high field strength in the Archean to the ultra-low Ediacaran values. It has been suggested that this period of weak field may reflect a collapsing dynamo prior to the formation of the inner core (Bono et al. 2019;Lloyd et al. 2021a;Thallner et al. 2021aThallner et al. , 2021bThallner et al. , 2022)), which may have allowed for atmospheric oxygenation and associated radiation of Ediacaran fauna (Meert et al. 2016;Huang et al. 2024). These observed weak values are broadly consistent with thermal evolution models which predict a decay in dipole moment prior to the nucleation of the inner core (Aubert et al. 2009;Driscoll 2016). These simulations also predict a subsequent increase in field strength upon the nucleation of the inner core. Paleointensity estimates from Zhou et al. (2022Zhou et al. ( , 2024) ) and Thallner et al. (2021bThallner et al. ( , 2022) ) support an increase in field strength following the Ediacaran low-field, with Zhou et al. (2022) reporting early Cambrian (~532 Ma) intensities that are five times higher than the time-averaged Ediacaran field strength. In contrast, Lloyd et al. (2022) found extremely low paleointensity values at 532 Ma, suggesting that the Ediacaran ultra-low field behavior persisted into the Cambrian. We note that the accuracy of the data presented in Lloyd et al. (2022) is contested (Zhou et al. 2024). However, it has been shown by other studies that the field hyperactivity (e.g., extremely rapid reversal rates) observed in the Ediacaran continued well into the Cambrian (Pavlov & Gallet 2001;Gallet et al. 2019), so some discordance in paleointensities in the early Cambrian may also be expected.
Although the aforementioned studies offer varying results and conclusions, they do have one common challenge: a paucity of paleointensity data. Intensity is the most difficult component of the magnetic field to constrain due to time-intensive measurements, experimental and geologic alteration, and notoriously high failure rates. Despite making up almost 90% of Earth's history, data from this super eon comprises only ~10% of the PINT global paleointensity database (Bono et al. 2021). This extreme sparsity in data for the majority of Earth's history represents a huge gap in our knowledge and greatly impedes our ability to understand the origin and evolution of our planet and the fundamental geologic processes that have shaped it, highlighting the urgent need for additional high-quality paleointensity estimates for the Precambrian.
To help fill gaps in our knowledge, we performed paleointensity experiments on mafic dikes from four different Precambrian-aged intrusive events in India. We obtained successful results from two regions, the Bundelkhand (~740 Ma) and Bastar cratons (~1.89 Ga). These new Precambrian data are important additions to the existing paleointensity database and are used to analyze long-term trends in geomagnetic field strength and test hypotheses related to the timing of ICN.
Sampling areas for this study include the Bundelkhand Craton, Malani-Nagar Parkar Igneous Suite, Bastar Craton, and Dharwar Craton of peninsular India (Fig. 1). Samples were all previously collected and studied for directional analysis (Meert et al. 2011(Meert et al. , 2013, in preparation;, in preparation;Pivarunas et al. 2019). In the field, core samples were collected from mafic dikes using a water-cooled, gasoline-powered hand drill and oriented using magnetic and sun compasses. Core samples were cut into standard sized specimens and stored in the magnetically shielded room at the University of Florida paleomagnetic laboratory.
As these samples were originally collected for directional analyses, much of the remaining material had been demagnetized during those studies, limiting the available samples for this study. Samples selected for this study were chosen based on two criteria: (1) the availability of remaining un-demagnetized material and (2) previous studies indicating that sister samples (see subsections below) yielded highquality paleomagnetic directional data for paleogeographic reconstructions and had ideal magnetic mineralogy and rock magnetic behavior for paleointensity study.
The Bundelkhand craton is located to the east of the Aravalli fold belt (AFB in Fig. 1), and is divided into several distinct units (Sharma & Rahman 2000;Pati 2020): (1) Archaean-aged granite-greenstone and gneiss belts (the Enclave Suite); (2) Relatively undeformed granitoid plutons and large quartz reefs (Granite Suite); (3) Mafic dike swarms and other smaller scale intrusions (Intrusive suite). The undeformed Bundelkhand and Berach granites are dated to ~2.5 Ga and these ages reflect the 'stabilization age' of the Bundelkhand craton (Wiedenbeck et al. 1996;Mondal et al. 2002). The Bundelkhand Igneous Complex hosts several suites of mafic dikes. Rao (2004) favored a two-phase emplacement, one at 2.15 Ga and the second at 2.0 Ga, based on 40 Ar/ 39 Ar isotopic analyses. Pradhan et al. (2012) provided U-Pb ages for a NW-SE trending swarm at 1.979 ± 0.008 Ga and for younger, E-W dikes at 1.113 ± 0.007 Ga. Pradhan et al. (2012) and Radhakrishna et al. (2013) noted three other potential dike swarms (based on paleomagnetic directional groups) but were unable to provide age constraints on those dikes. Shankar et al. (2023) reported a mean age for NW-SE trending dikes in the easternmost region of the Bundelkhand craton of 2.004 ± 0.0035 Ga (three dikes). A recent review paper (Mohanty 2023) cited 13 different ages for dikes in Bundelkhand. The array of ages reflects different generations of magmatism as well as re-setting due to tectono-thermal events after emplacement. Due to the wide age range of dikes in Bundelkhand, we performed geochronologic analyses on the Bundelkhand samples in this study.
The Bundelkhand samples used for this study were collected in 2019 from a dike (site I1929) near the Betwa River south of Orchha, Madhya Pradesh. Samples from many other dikes were also collected as part of a paleodirectional study, the results of which will be presented in a future publication (Meert et al. in preparation). The dikes ranged from 5 to 8 m in width. The directional data from our site I1929 pass the McFadden & McElhinny (1990) reversal test (Rc) and have a positive baked contact test. Site I1929 yields a mean direction of D = 306°, I = +58° (k = 225, α95 = 3.2°) (Fig. 2D), similar to the "Steep 3" directions from nearby dikes reported in Radhakrishna et al. (2013). A total of 11 specimens from site I1929 were selected for paleointensity analysis. Individual specimens unblock between 560-580° C (Fig. 2B) and yield nearly reversible Curie-temperature (Tc) spectra (Fig. 2C) with Tc = 570° C and a small Hopkinson peak during heating consistent with low-Ti magnetite as the main remanence carrier. Similar behavior was observed in the Radhakrishna et al. (2013) study. Furthermore, we present a new 740 ± 21 Ma 207 Pb/ 206 Pb weighted mean age from zircons of one of the examined dikes, representing the youngest magmatism in the Bundelkhand craton (See Results section).
The Malani-Nagar Parkar Igneous Suite (MNP) is in present-day Rajasthan (NW India) and southeastern Pakistan and represents the youngest phase of Precambrian magmatism on the Marwar block. The MNP emplacement took place in three phases with overlapping ages: (1) an initial phase of extrusive volcanic activity consisting of mafic and felsic lava flows (780-760 Ma), (2) a secondary phase dominated by granitic plutonism (767-750 Ma; Ashwal et al. 2013), and (3) a final phase that featured emplacement of both felsic and mafic dike swarms (~750 Ma; Wang et al. 2017). The MNP is bracketed by two unconformities that separate Pre-Malani basement rocks of Paleo-to Mesoproterozoic age and post-Malani, Marwar sedimentation (Pandit et al. 1999). Felsic lavas emplaced during the first stage of Malani magmatism are grouped by chemistry into peraluminous (Jalore Granite) and peralkaline (Siwana Granite) types which occupy distinct areas in the MNP to the north and west of Jodhpur, respectively (Meert et al. 2013).
The Malani samples used for this study were collected by Meert et al. (2013) from mafic dikes that intruded the Siwana granite near Redana, which is located west of Jodhpur, India. The dikes are ~5 m in width and a baked contact test at site I9-2 provides conclusive evidence for primary magnetization (fig. 5 of Meert et al. 2013). Thermomagnetic analyses showed the presence of low-Ti titanomagnetite along with some small amounts of pyrrhotite in some specimens (figs 7a, 7e, and 7f of Meert et al. 2013), namely those from site I9-16. Sister specimens from these samples were avoided for paleointensity study.
Isothermal remanence acquisition and backfield coercivity of remanence measurements for MNP samples show saturation by 0.2-0.4 T, indicating the presence of magnetite (fig. 7b of Meert et al. 2013).
Directional data is near univectorial, stable above 550-560°C (fig. 4b of Meert et al. 2013), and yielded robust paleomagnetic poles. A total of six specimens from three sites in the MNP were selected for this study. Samples for geochronology of the MNP were collected from the mafic dike at site I9-2 by Meert et al. (2013) and provided a weighted average 207 Pb/ 206 Pb age of 752 ± 18 Ma.
The Bastar Craton is a ~500 km 2 crustal block in southeastern India bordered by the Central Indian Tectonic Zone (CITZ) to the north and the Godavari rift to the south. Early Archean gneisses make up the basement rocks in Bastar, which are then intruded by Proterozoic granitoids. The Bastar Craton is intruded by numerous mafic dike swarms that crosscut the various granitoids and supracrustal rocks of the region (Srivastava et al. 2021). Many of the dikes in the southern Bastar craton trend NW-SE, paralleling the Godavari rift, and these dikes are thought to have exploited preexisting faults. The northern dikes are oblique to the Mahanadi rift and strike NNW-SSE (French et al. 2008;Shellnutt et al. 2019).
Geochronologic studies identified distinct intrusive intervals for the dikes in the Bastar craton with the following ages: (a) 2.365 Ga NW-SE trending bonititic Bhanupratappur dikes (Liao et al. 2019)
Ga ENE-trending Churra dikes (Srivastava et al. 2021); (c) 1.891-1.883 Ga NW-SE trending Keshkal dike swarm (French et al. 2008); (d) 1.851 NNW-trending Sonakhan dike swarm (Shellnutt et al. 2019);
(e) 1.46-1.44 Ga N-S trending Lakhna swarm (Ratre et al. 2010;Pisarevsky et al. 2013). The Keshkal dike swarm is the focus of the current study.
The Bastar samples used for this study were collected by Meert et al. (2011) from doleritic dikes within the Keshkal area which cross-cut Archean basement rocks. The dikes show only low-grade alteration and generally range in width from 15-20 m thick, with some up to 200 m. Thermomagnetic curves for these dikes are nearly reversible and have Tc of ~568-575°C (figs 3a and 3b of Meert et al. 2011). The authors noted trace amounts of pyrrhotite, pyrite, and/or chalcopyrite in some dikes. Samples showing evidence for Fe sulfides were not used in this study. Paleomagnetic results from these samples were of high-quality and indicated a dual-polarity magnetization that does not match younger magnetic directions from India, providing support for a primary magnetization (fig. 2 of Meert et al. 2011). A total of 26 specimens from two sites in Bastar were selected for this study.
The Dharwar Craton, located in southern India, is the largest of the Indian cratons. When combined with the Southern Granulite Terrane (SGT, directly to the south), this area covers ≥238,000 km 2 (Goodwin 1991). The Dharwar is split into eastern and western portions by the N-S trending Closepet granite (Friend & Nutman 1991;Ramakrishnan & Vaidyanadhan 2010;Meert & Pandit 2015). Both regions are dominated by metamorphic rocks including tonalite-trondhjemite gneisses and greenstone schist belts that range in age from 3.4-2.55 Ga. The eastern portion also contains 3.0-2.55 Ga granites (e.g., Balakrishnan et al. 1990;Kumar et al. 1996;Jayananda et al. 2000). Adjacent to the eastern portion is the crescentshaped Cuddapah basin, which is comprised of igneous and sedimentary rocks of Meso-Neoproterozoic age (Lakshminarayana et al. 2001). Throughout the Proterozoic, magmatism occurred across the craton and included kimberlites, lamproites, and numerous mafic dike swarms that cross-cut the metamorphic basement of the Dharwar.
Of the numerous generations of mafic intrusions in the Dharwar (Söderlund et al. 2019), the samples used in this study come from the Bangalore (Bengaluru) dike swarm in the northern section of the SGT and were collected by Pivarunas et al. (2019). Two sites were chosen for this study: one near the city of Tiruvannamalai and one near Hogennakal. Thermomagnetic curves are highly reversible with Tc consistent with unaltered magnetite (fig. 5a of Pivarunas et al. 2019). Hysteresis properties of all dikes suggested magnetic grain sizes ranging from pseudo single domain to single domain (fig. 6 of Pivarunas et al. 2019). Directional data indicated a multicomponent NRM with the presence of a shallow secondary overprint (fig. 7 of Pivarunas et al. 2019). Baked contact tests were conducted at site I1530/I1531 and provided evidence for primary magnetization (fig. 8 of Pivarunas et al. 2019). A total of five specimens from two sites in Dharwar were selected for this study.
The Bangalore dikes from Dharwar that are used in this study have been reliably dated using U-Pb zircon geochronology to 2.363 ± 0.0066 Ga (Pivarunas et al. 2019). This time period is represented in the database by only one study of Dharwar dikes (Valet et al. 2014), which is currently the only paleointensity study of Precambrian India.
In summary, sampling areas for this study include the Bundelkhand Craton (~740 Ma; This study), Malani Igneous Suite (~752 Ma; Meert et al. 2013), Bastar Craton (~1.89 Ga; Meert et al. 2011), and Dharwar Craton (~2.36 Ga;Pivarunas et al. 2019) of peninsular India (Table 1). Overall, the ages of these samples cover multiple time periods where data are sparse or completely missing in the Precambrian PINT database (Bono et al. 2021).
In all cases, specimens were pre-selected for this study based on the site's demagnetization behavior, rock magnetic properties, and evidence for primary magnetization as indicated by the previous paleomagnetic studies of the samples. In some cases, sample selection was limited based on availability of material.
Overall, a total of 48 specimens from nine sites across the four localities were selected for paleointensity experimentation.
All experiments were performed at the University of Florida Paleomagnetic laboratory. Paleointensity experimentation was completed using the stepwise Thellier-Thellier method (Thellier & Thellier 1959) following the IZZI protocol (Tauxe & Staudigel 2004;Yu & Tauxe 2005). Partial thermoremanent magnetization (pTRM) checks were conducted throughout the experiment to check for heating-induced magnetomineralogical alteration. Heating steps were completed in air using a Magnetic Measurements TD80A paleointensity furnace. Specimens were placed in the same positions within the sample boat during each heating step. During each in-field step, samples were placed in the same orientation with respect to the applied field. Two separate experiment sets were completed using applied fields of 45 μT and 60 μT to check for nonlinear remanence. Any residual field inside the furnace was measured at ≤10 nT for zero-field steps. Remanence measurements were performed using a 2G Enterprises Superconducting Quantum Interference Device (SQUID) rock magnetometer. The aforementioned instruments are all housed in a magnetically shielded room at the University of Florida. During paleointensity experimentation, all samples were stored in the magnetically shielded room.
The resulting IZZI paleointensity data were analyzed using the thermal Thellier analysis tools on paleointensity.org (Béguin et al. 2020). The auto-interpret function was used along with our selection criteria (see below) to select fits and calculate Arai plot statistics and paleointensity estimates. Specimens from the same site were averaged before converting into a Virtual Dipole Moment (VDM). All VDMs were calculated using inclinations obtained from samples from the same sites (Meert et al. 2011(Meert et al. , 2013;;Pivarunas et al. 2019). For Bundelkhand site I1929, the calculated paleolatitude is 39 degrees.
Paleointensity results were assessed using a set of quantitative selection criteria as outlined in Standardized Paleointensity Definitions (SPD; Paterson et al. 2014). To determine whether a paleointensity estimate was accepted, the following criteria had to be met:
(1) a maximum angular deviation (MAD; Kirschvink 1980) of < 10° for grade A results and < 15° for grade B results;
(2) the difference between anchored and free-floating fits of directional data to the origin (α) of < 15°;
(3) number of measurements used for paleointensity determination of n ≥ 4;
(4) fraction of NRM used for the best fit on the Arai plots determined by vector difference sum calculations (FRAC; Shaar & Tauxe 2013) of ≥ 0.35 for grade A results and ≥ 0.25 for grade B results;
(5) scatter parameter (β; Coe et al. 1978) values of < 0.10;
(6) number of pTRM checks of NpTRM ≥ 2;
(7) the maximum absolute difference produced by a pTRM check normalized by the total TRM (δCK; Leonhardt et al. 2004) Similar selection criteria have been used in other paleointensity studies on similar age rocks (e.g., Shcherbakova et al. 2017b;Lloyd et al. 2021aLloyd et al. , 2021b;;Thallner et al. 2021aThallner et al. , 2021b)). A summary of these criteria can be found in Table 2. In addition to selection criteria, we calculated the angular difference between the pTRM acquired in the last step used for the paleointensity estimate and the direction of the applied laboratory field. This is known as the γ parameter and can be used to estimate any bias due to anisotropy of TRM (Paterson et al. 2014).
Zircons were separated from the sampled dike (I2217-HS1) following routine mineral separation techniques of crushing, pulverizing, sieving, density separation (using Wilfley Table and methylene iodide) and magnetic separation (with Frantz isodynamic separator). Approximately 35 zircons were handpicked under a binocular microscope and mounted in a 1" epoxy puck. The puck was then sanded to expose the interior structure of the zircon grains and polished to a 0.3-micron aluminum carbide grit. were ablated using 2% laser power and 5 Hz repetition rate for 30 seconds. FC-1 (Paces & Miller 1993) and 91500 (Wiedenbeck et al. 2007) zircons were used as primary and secondary reference materials, respectively. Data processing and reduction was completed with an in-house spreadsheet "Calamari", and the final age calculations, common-Pb correction and plotting of the diagrams were completed using IsoplotR (Vermeesch 2018).
From the ~752 Ma MNP, all six specimens (three sites) failed. The Arai plots exhibited two slopes that appear to be due to an overprint below 505°C (Fig. 3A inset). The temperature range above 505°C displayed sagging in the Arai plot (high β and |k'|). The few Arai segments that were relatively linear could not be used due to failed pTRM checks (high DRAT). Our results indicate that a combination of non-ideal behavior (e.g., multidomain grains) and thermal alteration during heating caused these samples to be unreliable paleointensity recorders.
All five specimens (four sites) from the ~2.36 Ga Dharwar craton also failed, primarily due to multidomain behavior as evidenced by dramatic zig-zagging in the Arai plots (Fig. 3B). This zig-zagging persisted from Tmin to the 530°C temperature range and then exhibited erratic behavior, precluding the use of any higher temperature segments.
For the Bastar craton, two separate sites (I527 and I531) yielded passing results for a success rate of 31%. Of the 10 specimens from site I527, two passed grade A selection criteria and two passed grade B. Those that passed grade A (specimens 4C and 1C) provided paleointensity estimates of 18.9 and 18.1 μT, respectively. The grade B results (specimens 7B and 10C) yielded paleointensities of 15.8 and 12.1 μT, respectively. The average of these results is 16.2 ± 3.0 μT, corresponding to a VDM of 4.1 ± 0.77 × 10 22 Am 2 for Bastar site I527 (~1.89 Ga). Excluding grade B gives an average of 18.5 μT and VDM of 4.7 × 10 22 Am 2 , a 14% difference. Passing specimens from site I527 exhibited two slope behavior in the Arai plots (Fig. 4A, B). The lower temperature slope, which extends from 0 to approximately 435-505°C, is positive. This is followed by a negative slope through the rest of the Arai plot. Directional data on Zijderveld diagrams (Fig. 4A, B insets) clearly show two separate directions that coincide with these temperature ranges, so we interpret this low-temperature positive slope in the Arai as an overprint and only use the higher temperature slope for paleointensity estimation.
For Bastar site I531, four specimens out of 16 passed our selection criteria. Two specimens passed grade A criteria and two passed grade B. Those that passed grade A (specimens 5C and 6B) yielded paleointensity estimates of 7.4 and 7.8 μT, respectively. The grade B results from specimens 2B and 3B
were significantly higher at 72.9 and 69.8 μT, respectively. Averaging these results provides a paleointensity of 39.5 ± 36.8 μT, or a VDM of 10.2 ± 9.5 × 10 22 Am 2 . However, despite passing selection criteria, we do not believe that the grade B results from site I531 are reliable estimates of the paleofield.
First, the Arai plots have multiple slopes and a "hook" shape (Fig. 4D) which indicates that the specimen did not follow the Law of Reciprocity and may have altered during experimentation (Tauxe et al. 2020).
In addition, the directional data is completely uni-vectorial. This behavior suggests that these specimens were entirely remagnetized at some point, either in-situ or during their time in storage. As such, these samples were excluded from analysis. The average paleointensity value excluding the remagnetized grade B results for site I531 is 7.6 μT, or a VDM of 2.0 × 10 22 Am 2 (~1.89 Ga).
Passing specimens from site I531 also exhibited two slope behavior in Arai plots. The steep lower temperature slope extends from 0 to 300-505°C followed by a shallower negative slope (Fig. 4C). Above ~570°C there is some erratic behavior in the Arai plots. Therefore, the maximum temperature used for fits was 560°C. Directional data indicate the presence of two components which correspond with these two temperature ranges. As such, only the higher temperature slope was used for paleointensity estimation.
Of the 11 specimens from the Bundelkhand site I1929, three passed our selection criteria for a 27% success rate. Specimen 10B passed our grade A selection criteria, yielding a paleointensity estimate of 28.9 μT. The other specimens, 11B and 5Z, passed grade B selection criteria providing paleointensity estimates of 23.9 and 20.0 μT, respectively. The average of these three results is 24.3 ± 4.5 μT, corresponding to a VDM of 4.3 ± 0.79 × 10 22 Am 2 for Bundelkhand site I1929 (~740 Ma).
The Arai plots for the passing Bundelkhand specimens exhibit mostly linear behavior (Fig. 4E, F) with satisfactory pTRM checks. At temperatures higher than ~550-560°C, pTRM checks start to increasingly deviate, suggesting thermal alteration. As such, we avoided use of the higher temperature steps for paleointensity estimates. However, we note that because we did not perform pTRM checks on every step, some of our selected fits could be affected by alteration. Consequently, we interpret the site mean for Bundelkhand as an estimate of minimum field strength.
Overall, a total of nine specimens from three sites yielded paleointensity estimates which satisfied our selection criteria (Table 3). To determine if our estimates were biased by anisotropy, we calculated the gamma statistic (γ) for each specimen. This statistic is the angle between the pTRM acquired at the last step used for the best-fit segment on the Arai plot and the laboratory field direction (Blab). Passing specimens from this study have γ values ranging from 1.7-8.2° (Table 2). Median and average values are 2.9° and 3.9°, respectively. These values are quite low and because the specimens were from quicklycooled dikes, which are less likely to be affected by anisotropic remanence, we do not believe our results to be biased by anisotropy. Furthermore, we do not apply a cooling rate correction as rock magnetic analyses suggest our primary recorders are in the vortex state size range, which recent experimental and theoretical work suggests are unlikely to be affected by cooling rate (Winklhofer et al. 1997;Yu 2011;Biggin et al. 2013;Ferk et al. 2014). Although we performed experiments with two different laboratory fields to check for the effects of non-linear remanence acquisition, we did not have passing specimens from both for any single site. However, the effects of non-linear remanence are typically caused by magnetic grains with strong shape anisotropy that is uncommon in rapidly cooled bodies (Selkin et al. 2007).
To further assess the quality of our paleointensity estimates we used the Quality of Paleointensity (QPI) criteria (Biggin & Paterson 2014). QPI is a set of reliability criteria for site-level paleointensity data and originally included eight criteria (AGE, STAT, TRM, ALT, MD, ACN, TECH, and LITH) but has since been expanded to include MAG and DIR introduced by Biggin et al. (2015) and Kulakov et al. (2019), respectively. These 10 criteria are used in the PINT database (Bono et al. 2021), therefore we assessed our data for all of them. Our results do not pass STAT (a minimum of 5 individual estimates per unit with low dispersion) due to a low amount of material available for this study, TECH (averaging results from more than one paleointensity technique), or TRM (microscopic evidence of primary igneous texture). Low experimental success rates, a common problem for rocks of Precambrian age, also contributed to not meeting STAT. In the case of the TRM criterion, the primary nature of our TRM is supported by rock magnetic analyses, baked contact tests, and paleomagnetic data from previous studies. Nonetheless, dedicated petrographic analyses and/or microscopy were not completed so our specimens do not meet this criterion. Our estimates also do not satisfy LITH (averaging results from more than one lithology which possess different unblocking behavior). This criterion is generally known as the most difficult to achieve.
In this study, the targets were dikes which were typically intruded into heavily weathered granites or other metamorphic rocks. Therefore, another lithology was not available at these sites. All of our estimates satisfy the AGE criteria (high-quality U-Pb and/or Pb-Pb ages), ALT (passing pTRM checks), MD (no portions of the Arai with sagging or zig-zagging were used in fits), DIR (well-constrained paleomagnetic directions), ACN (no evidence for bias due to anisotropy or cooling rate), and MAG (data publicly available). Overall, our estimates have a QPI score of 6 which is higher than 92% of the data currently in the PINT database.
The zircons separated from the Bundelkhand basaltic dike sample I2217-HS1 are 100-200 μm in size, euhedral and display oscillatory, laminar and sector zoning, typical of magmatic zircons. Although a few grains had metamict cores, the magmatic texture is persistent. Furthermore, well developed core-rim texture with distinctive bright boundaries were observed in several of the analyzed zircons. The magmatic character was also inferred by the Th/U ratios that vary from 0.1-0.9 (Table S1) for the majority of the analyzed spots with a negligible portion falling into the magmatic-metamorphic transitional zone (e.g., Rubatto 2002).
The LA-MC-ICPMS analysis revealed >1 Ga ages from core spots, defining a discordia, with an upper intercept of 2.538±0.054 Ga (2σ, n = 77/77). This age agrees well with the age of the Bundelkhand craton in which the sampled dikes intruded into (e.g., Mondal et al. 2002). The remaining spot analyses results from the rims of the zircons (excluding the spots with inclusions and/or Pb-loss) yielded a lower intercept age of 726 ± 24 Ma (2σ, n = 48/77, analytical uncertainty) with Pb anchored to ( 207 Pb/ 206 Pb)c = 1.106 (Fig. 5A). During the analyses, an excess of common-Pb was observed in the time series that is interpreted to be incorporated due to a thermal event before or during the dike emplacement. The common-Pb corrected (with methods specified by Tera & Wasserburg 1972;Stacey & Kramers 1975;Vermeesch 2020) isotopic ratios of the same population from spots on the rims of the zircons yielded a 207 Pb/ 206 Pb weighted mean age of 740 ± 21 Ma (2σ, MSWD = 5.9, n = 48/77) (Fig. 5B). The common-Pb uncorrected lower intercept age, and the common-Pb corrected 207 Pb/ 206 Pb weighted mean age agree well within error (MSWD = 0.77, probability = 0.38). Therefore, we interpret the ca. 740 Ma date to represent an inferred crystallization age for this basaltic dike.
The magnetic field varies on a range of timescales and to assess long-term trends in magnetic field behavior, magnetic data would ideally average out short-term variations, known as secular variation. It is unlikely that our data, e.g., two to three specimens from one to two dikes per locality, average out secular variation. Instead, our data are best interpreted as individual snapshots of field behavior at the time that the dikes cooled. Although we cannot determine average field behavior, our results can still be used to make inferences about field behavior in the Precambrian in addition to adding more high-quality data to the Precambrian database.
Our data represent the first paleointensity estimates from Precambrian-aged rocks from the Bundelkhand and Bastar cratons of India. Although our estimates do not have high N, they do show that the rocks from these areas yield successful paleointensity results and thus could be targeted for future sampling for a variety of Precambrian age ranges. The VDM obtained for the Bundelkhand craton (~740 Ma) is 4.3 ± 0.8 × 10 22 Am 2 while VDMs from the Bastar craton (~1.89 Ga) are 4.1 ± 0.8 × 10 22 Am 2 and 2.0 × 10 22 Am 2 .
We note that the discrepancy between the paleointensity values at the two Bastar sites is unknown, but most likely is due to secular variation. The newly obtained VDMs are much lower than the modern-day field of ~8 × 10 22 Am 2 but are not significantly different than the time-average field strength over much of the Phanerozoic of ~4-5 × 10 22 Am 2 (Tauxe et al. 2013;Biggin et al. 2015;Hawkins et al. 2024).
Additionally, our Bastar results add new data for a ~200 Myr interval in the paleointensity database where previously there was only one other data point at ~1.90 Ga (Shcherbakova et al. 2014).
Comparing our results to the PINT database, filtering to Q PI scores ≥ 3 (Fig. 6), the data closest in age to our 1.89 Ga Bastar results are a ~1.90 Ga estimate of 2.82 ± 0.12 × 10 22 Am 2 from Paleoproterozoic dikes from the Kaapvaal Craton, South Africa (Shcherbakova et al. 2014) and estimates from the 1.860 ± 0.004 Ga dolerite sills of the Murmansk Craton, NE Fennoscandia, which range from 1.16-2.50 × 10 22 Am 2 (Veselovskiy et al. 2019). Our VDM from site I531 of 2.0 × 10 22 Am 2 is consistent with these values. The VDM for site I527 is higher at 4.1 ± 0.8 × 10 22 Am 2 but is within the range of expected variation in field strength due to secular variation. It is also possible that the dikes at site I527 could be slightly older than those at I531 and reflect an overall decreasing field from 2.10 to 1.68 Ga, as predicted by the Monte Carlo Axial Dipole Average Model (MCADAM) dipole moment model (Bono et al. 2022).
Although we did not obtain successful results from the Malani-Nagar Parkar Igneous Suite or the Dharwar Craton, both regions host multiple generations of Precambrian dikes that may yield robust paleointensity estimates. Success also likely varies among individual swarms; for example, Valet et al. (2014) obtained successful paleointensity results from the same Dharwar swarm studied here. Therefore, the negative results reported here may not fully represent the potential of all Precambrian-aged dikes in these areas, highlighting significant potential for future investigations.
Our ~740 Ma Bundelkhand result of 4.3 ± 0.8 × 10 22 Am 2 is bracketed by three estimates from the ~755 Ma Mundine Wells dike swarm in western Australia (Lloyd et al. 2021b) and a cluster of 11 estimates from the ~726-720 Ma dikes of the Franklin large igneous province (Lloyd et al. 2021a). At ~755 Ma the VDMs range from 6.08-6.41 × 10 22 Am 2 with an average of 6.3 × 10 22 Am 2 . Our result ~15 Myr later represents a 32% decrease in the average field strength. At ~726-720 Ma, VDMs range from 0.36-1.9 × 10 22 Am 2 with an average of 1.1 × 10 22 Am 2 . These VDMs, occurring another 14-20 Myr later, represent a further decrease in average field strength of 74%. Overall, these data between 755-720 Ma show the field losing approximately 83% of its strength in ~30 Myr. We note that 1) paleointensity estimates for this time period are sparse and likely insufficient to average out secular variation, and 2) our estimate from Bundelkhand may represent a minimum intensity (see Results section). As a consequence, the true magnitude of the decay in field strength remains uncertain, and the percentages stated above reflect changes between discrete snapshots of the field. Nevertheless, the apparent rapid decay in intensity between these records (Fig. 7) suggests that the period of ultra-low field strength identified in the Ediacaran (Bono et al. 2019;Shcherbakova et al. 2020;Thallner et al. 2021aThallner et al. , 2021bThallner et al. , 2022;;Huang et al. 2024) may have begun in the late Tonian to early Cryogenian. Broadly, this supports numerical geodynamo simulations and thermal evolution models that propose a weak geodynamo prior to ICN (e.g., Aubert et al. 2009;Landeau et al. 2017). More specifically, this earlier onset of field collapse is in agreement with the model of Driscoll (2016) which predicts that the geodynamo slipped into a weak-field regime for ~350 Myr prior to ICN. Based on currently available data, Huang et al. (2024) stated that the Ediacaran ultra-low field lasted at least 26 Myr, but the onset of this period of weak field strength has not yet been identified.
If the magnetic field's collapse began earlier than currently established, such as in the late Tonian, it may have remained in a weakened state for ~175 Myr until the reported increase in field intensity during the late Ediacaran to early Cambrian (Zhou et al. 2022(Zhou et al. , 2024;;Huang et al. 2024). This longer duration, on the order of 10 2 Myr rather than 10 1 Myr, aligns more closely with numerical predictions. However, the length of the ultra-low field period remains unknown and cannot be constrained to the late Tonian/early Cryogenian yet as the paleointensity estimates at ~755 Ma, ~740 Ma, and ~720 Ma are unlikely to be averaging secular variation. Furthermore, there are no robust paleointensity data for the 120 Myr period between 719-599 Ma. Currently, the only data in this time range is one estimate at ~682 Ma from a meteorite impact structure -but, the authors note a bias towards high values in their Arai plots and large uncertainty in the age (Salminen et al. 2006). Therefore, this data point was excluded from Fig. 7.
Similarly, as shown in Figures 6 and 7, high-quality data are also missing for the 950-760 Ma interval. To further test this hypothesis of a late Tonian to early Cryogenian field collapse, additional data from the 950-600 Ma time range are certainly needed. Future studies should target rocks in this 950-600 Ma age range to (1) constrain when secular variation in intensity diminished to levels characteristic of the mid-tolate Ediacaran, and (2) identify the true onset and duration of the weak-field regime. This information will be essential to understand geodynamo evolution during this particularly critical period in our planet's history.
To examine what our Bastar results may indicate about the Precambrian magnetic field, we will consider both the MCADAM.1b model of Bono et al. (2022) and the long-term trend of Bono et al. (2019) (Fig. 6). MCADAM is based on site level results from the PINT database (Bono et al. 2021) and in the authors' preferred model (MCADAM.1b), moderately restrictive data selection requiring that paleointensity site records meet at least three of the QPI criteria was used. Our new data is in agreement with the predictions of MCADAM (Bono et al. 2022) and fall within the 95% confidence interval of the model. However, we note the MCADAM model is informed by existing data and, in time intervals which lack data, it is constrained by its nearest neighbors. Therefore, it may not reflect the true scale of variability in the field.
Despite this limitation, it is a useful tool for analyzing trends in the existing data.
A more restrictive data selection method was used in Bono et al. (2019) including the removal of data solely incorporating low unblocking temperatures in paleointensity fits, the requirement that results from rapidly cooled units are based on more than one value (e.g. >1 dike, flow, or sill) and have more than one determination for each cooling unit, and the exclusion of data where alteration is apparent. Focusing on the interval of time most closely associated with our new Bastar results (2.1-1.85 Ga), among the studies with QPI ≥ 3 included in MCADAM, the ~2.07 Ga results from Sumita et al. (2001) were excluded from Bono et al. (2019) due to the use of low-temperature fits, and the ~2.07 Ga results from Halls et al. (2004) and Macouin et al. (2003), and the ~1.90 Ga results from Shcherbakova et al. (2014) were also excluded due to lack of specimen or dike level replication. Additionally, the ~1.86 Ga results of Veselovskiy et al. (2019) came out after Bono et al. (2019) was published, but were not included in updated versions presented in Zhou et al. (2022) and Huang et al. (2024). The estimates from these excluded studies are all noticeably lower than Huang et al.'s (2024) new 2.04 Ga estimate from the Bushveld (1.16-4.86 × 10 22 Am 22 vs. 8.04 × 10 22 Am 22 , respectively) in addition to field values from the Midcontinent Rift at ~1.1 Ga (~6 × 10 22 Am 22 ). Our new field strength estimates from Bastar are consistent with the values from these studies excluded in Bono et al. (2019), adding support to their validity. Additionally, our estimates are in agreement with more recent studies of Proterozoic aged rocks which report low VDMs (Shcherbakova et al. 2023(Shcherbakova et al. , 2024;;Lloyd et al. 2024).
The rationale for the highly restrictive data selection of Bono et al. (2019) was to create a dataset with only paleointensity estimates which represent time-averaged field values in order to relate models of field evolution to observations. As such, it is intended to capture the primary signal of secular cooling by a primarily thermally driven dynamo. While secular cooling played a major role in the early geodynamo, the consistent observation of low VDMs in the Proterozoic suggest other mechanism(s) were also significantly affecting the efficiency of the Precambrian geodynamo. In addition to the "Proterozoic dipole low" (Biggin et al. 2015), other dipole lows have been identified in the Paleozoic (e.g., Shcherbakova et al. 2017a;Hawkins et al. 2019Hawkins et al. , 2021) ) and Mesozoic (e.g., Prévot et al. 1990;Biggin & Thomas 2003;McElhinny & Larson 2003;Tauxe et al. 2013;Shcherbakova et al. 2015;Anwar et al. 2016). Some have postulated that this is due to cyclical behavior in the field after ICN, possibly due to mantle convection (e.g., Tarduno & Cottrell 2005); but the existence of a similar period in the Proterozoic raises questions about what kind of core conditions are needed to result in cyclical behavior in the absence of an inner core. Unfortunately, the Precambrian paleointensity database is still too sparse to come to any definitive conclusion.
In this work, we presented three new high-quality (QPI = 6) paleointensity estimates for Precambrian-aged samples. As a whole, these new data are 46-75% lower than the modern-day field, are consistent with the long-term average for the Phanerozoic, and are in good agreement with paleointensity estimates of similar age.
Our new results from the Bastar are consistent with other low Proterozoic paleointensity estimates (Sumita et al. 2001;Macouin et al. 2003;Halls et al. 2004;Veselovskiy et al. 2019;Shcherbakova et al. 2023Shcherbakova et al. , 2024;;Lloyd et al. 2024) adding support for the Proterozoic dipole low (Biggin et al. 2015).
Additionally, our data for ~740 Ma suggest that the field collapse that led to the ultra-low Ediacaran field may have started earlier than the Ediacaran, possibly in the late Tonian to early Cryogenian. In this scenario, the weak geodynamo regime could have persisted for at least ~175 Myr before regaining strength in the early Cambrian. This more protracted period of weakness in the field is in good agreement with numerical predictions and thermal models suggesting a weakening/low-field between 1.0 Ga and 650 Ma prior to the nucleation of a young inner core. However, the onset of the weak field regime is currently unknown due to sparse data between 950-600 Ma. Future investigations should target rocks with ages in this time range to test this hypothesis in order to improve our understanding of the evolution of the geodynamo, the nucleation of the inner core, and the evolution of Ediacaran and Cambrian fauna.
Overall, the sparsity of paleointensity data during the Precambrian hinders our ability to make broad conclusions about the long-term trends in the field. Despite this issue, the addition of this data is a step towards a more representative paleointensity record for the Precambrian. More studies on Precambrianaged rocks are needed to continue to build upon existing hypotheses of long-term trends in the geodynamo. India is a great candidate for continued future work as, 1) it has a wide range of Precambrian-aged outcrops and intrusions that are capable of recording paleointensities (Meert et al. 2021), and 2) it has been understudied -only this and one other published study on paleointensity of Precambrian India currently exist. Table 3. A summary of the paleointensity results for Bastar and Bundelkhand (abbrev. Bund.) cratons. Omitted results are indicated by italicized text. Tmin and Tmax are the temperature ranges of the best fits used for paleointensity determination. Blab is the lab field used during experimentation (μT). Banc is the acquired intensity from individual specimens (μT). VDM is the Virtual Dipole Moment (x10 22 Am 2 ) of the average intensities. SD indicates standard deviation, when applicable. The following parameters are the statistics of the best-fit used for paleointensity determination: N is the number of points on the Arai used, FRAC is the fraction of NRM used, β is the scatter parameter, |k'| is the curvature parameter, MAD
is the maximum angular deviation, α is the angle between anchored and unanchored fits of the directional data. The following parameters are statistics related to partial thermoremanent magnetization (pTRM) checks: NpTRM is the number of pTRM checks, δCK is the maximum absolute difference of a pTRM check normalized by the total TRM, DRAT is the maximum absolute difference of a pTRM check normalized by the length of the best-fit line, CDRAT is the sum of the absolute pTRM differences, and γ is the angle between the pTRM at the last step used for the best-fit and the applied field direction (an indicator of anisotropic TRM). For full definitions, see Paterson et al. (2014). Bundelkhand, Malani-Nagar Parker Igneous Suite (MNP), Bastar Craton, and Dharwar Craton. Sampling locations are indicated by red stars. Also included are: Aravalli Fold Belt (AFB), Marwar Basin (MB), Vindhyan Basin (VB), Central Indian Tectonic Zone (CITZ), Chattisgarh Basin (ChB), Central Indian Suture Zone (CIS), Prahnita-Godavari Basin (PG), Cuddapah Basin (CuB), Mahandi Rift (MR), Eastern Ghats Mobile Belt (EGMB), Western Dharwar Domain (WDD), Eastern Dharwar Domain (EDD), Closepet Granite (CG), and the Southern Granulite Terrane (SGT). Zijderveld diagram showing primary and secondary directions. B) Demagnetization spectra indicating unblocking between 560-580°C and majority of signal carried by magnetite. C) Thermomagnetic curves showing reversibility and Curie temperature consistent with magnetite. D) I1929 site mean direction on orthographic projection. Nagar Parkar Igneous Suite and B) Dharwar Craton. Blue (pink) circles on the Arai plots indicate zerofield/in-field (in-field/zero-field) steps, or "ZI" ("IZ"). Gray triangles show the pTRM checks. On the Zijderveld diagrams, the closed and open circles represent horizontal and vertical projections, respectively. Paleointensity data were analyzed using Paleointensity.org (Béguin et al. 2020). specimens from (A, B) Bastar site I527, (C, D) Bastar site I531, and (E, F) Bundelkhand site I1929. Blue (pink) circles on the Arai plots indicate zero-field/in-field (in-field/zero-field) steps, or "ZI" ("IZ"). Gray triangles show the pTRM checks. On the Zijderveld diagrams, the closed and open circles represent horizontal and vertical projections, respectively. Paleointensity data were analyzed using Paleointensity.org (Béguin et al. 2020). diagram, displaying the discordant distribution of the analysis results. The upper intercept age is defined using the entirety of the analysis, whereas the lower intercept age is defined using the analysis of the spots on the rims of the zircons. (B) The common-Pb corrected 207 Pb/ 206 Pb age distribution of all the zircons analyzed and the weighted mean age of the spots from the zircon rims (same spots used in the lower intercept age calculation). Representative cathodoluminescence zircon images displaying the welldeveloped core-rim textures. The scale bar is 100 μm, and the spots are 15 μm for the spots on the zircon cores, and 10 μm for the spots on the zircon rims. The ages reported on the zircons are common-Pb uncorrected 206 Pb/ 238 U ages for the respective spots. MSWD -mean square weighted deviation; nnumber of analyses. 750-720 Ma followed by the Ediacaran ultra-low field values. The red line is the mean from the MCADAM.1b model (Bono et al. 2022) with shaded regions representing the 95% confidence interval. Error bars show age uncertainties -when not shown, the uncertainty is smaller than the data point. As shown by the arrow, the new Bundelkhand estimate may represent a minimum intensity. References for other plotted data are: Mundine Wells (Lloyd et al. 2021b), Franklin LIP (Lloyd et al. 2021a), Grenville (Thallner et al. 2021a), Volyn Traps (Shcherbakova et al. 2020; Thallner et al. 2022), Skinner Cove (Thallner et al. 2021b), Single Crystal (Time-averaged data from Bono et al. 2019; Zhou et al. 2022, 2024; Huang et al. 2024). Please note that data published after 2022 are not yet incorporated into the MCADAM model.