Nature of Cretaceous dolerite dikes with two distinct trends in the Damodar Valley, eastern India: Constraints on their linkage to mantle plumes and large igneous provinces from 40 Ar/ 39 Ar geochronology and geochemistry

Two distinct sets of Cretaceous dolerite dikes intrude the Chhotanagpur gneissic complex of eastern India, mostly within the Damodar Valley Gondwanan sedimentary basins. One dike set trends NNE to ENE, whereas the other set, which includes the prominent Salma dike, trends NW to NNW. One dike from each set in the Raniganj Basin was dated using the 40 Ar/ 39 Ar method in order to resolve a controversy concerning the emplacement age of the Salma dike. The NE-trending dike yielded a plateau age of 70.5 ± 0.9 Ma, whereas the NNW-trending Salma dike is much older, with a plateau age of 116.0 ± 1.4 Ma. These results demonstrate that the Salma dike was emplaced at ca. 116 Ma and not at ca. 65 Ma, as suggested in an earlier study. Geochemical characteristics of the two dikes are also distinct and indicate that they belong to previously identified high-Ti and low-Ti dolerite groups, respectively. The observed geochemical characteristics of both dike sets are comparable with the geochemistry of basalts of the Kerguelen Plateau, Bunbury Island, and Rajmahal Group I and suggest a connection to mantle plumes. The new age data presented herein indicate that these two magmatic episodes in the eastern Indian Shield were related to the ca. 120–100 Ma Kerguelen mantle plume and its associated Greater Kerguelen large igneous province and the ca. 70–65 Ma Réunion plume and its associated Deccan large igneous province, respectively.


INTRODUCTION
Due to their vertical depth and lateral extent, mafic dike swarms often survive when the volcanic component of short-lived, mantlegenerated magmatic events or large igneous provinces is lost to erosion. Therefore, the study of dikes is of great importance in determining the original extent of large igneous provinces and in linking these events to mantle plumes (Ernst andBuchan, 2001a, 2001b;Bleeker and Ernst, 2006;Ernst et al., 2010;Srivastava et al., 2010Srivastava et al., , 2019aErnst, 2014). Precise age determinations are essential for identifying components of a large igneous province, determining its source plume, and comparing ages of magmatic events on different cratonic blocks (i.e., magmatic barcoding) to test continental reconstructions (e.g., Bleeker, 2003Bleeker, , 2004Bleeker and Ernst, 2006;Srivastava et al., , 2019aErnst, 2014).
More importantly in the context of the present study, Srivastava et al. (2014) pointed out that the CGC is the only terrane in the Indian Shield that appears to have experienced magmatic activity associated with both the Kerguelen and Réunion plumes. Although most of the Cretaceous magmatism in the region is related to the Kerguelen plume (Kent et al., 2002;Srivastava et al., , 2014Basu, 2011, 2013;Chalapathi Rao et al., 2014;Srivastava, 2020; and references therein), Kent et al. (2002) reported an age of 65.4 ± 0.3 Ma for the prominent NNW-trending Salma dolerite dike, suggesting that it may be related to the Réunion plume. However, on the basis of paleomagnetic data, Patil and Arora (2008) questioned the reported ca. 65 Ma age of the Salma dike and suggested that it was emplaced at ca. 117 Ma, consistent with other magmatic activity of the region. Geochemistry of the Salma dike is also not consistent with a ca. 65 Ma emplacement age . Srivastava et al. (2014) suggested that further geochronological study of the Salma dike is needed to resolve the uncertainty in its emplacement age. In this study, we report the results of a geochronological and geochemical study of the Salma dike, as well as a similar study of a NE-trending dike from the same region. In addition, we examined the possible link between the two dike sets and mantle plumes and associated large igneous provinces.

GEOLOGICAL BACKGROUND
The eastern and northeastern parts of the Indian Shield consist of the Singhbhum craton and the Shilling Plateau-Mikir Hills (Fig. 1). The northern portion of the Singhbhum craton is represented by the CGC, which is separated from the Singhbhum granite complex by the Singhbhum mobile belt ( Fig. 1; cf. Naqvi and Rogers, 1987;Sharma, 2009;. This part of the Indian Shield also includes two alkaline provinces, the Damodar Valley and the Shillong Plateau ( Fig.  1; Srivastava, 2020).
The geology of the CGC has been described and discussed in detail in several publications (Naqvi and Rogers, 1987;Mahadevan, 2002;Sharma, 2009;Ramakrishnan and Vaidyanadhan, 2010;Rekha et al., 2011;and references therein). Therefore, only a brief summary is provided here. The CGC is considered to be a cratonized mobile belt of Archean age (cf. Naqvi and Rogers, 1987;Kumar and Ahmad, 2007;Sharma, 2009;Srivastava et al., , 2012. The cratonic nature of the CGC is supported by a number of ENE-to E-trending intracontinental rift/shear zones responsible for distinct magmatism in the region . In general, the CGC is dominated by foliated granitoids (dominantly granite to granodiorite, occasionally tonalite) that are intrusive into amphibolites and granulite-facies gneisses and schists and occur as mesoscopic to regional-scale enclaves. The craton includes gabbros, massif anorthosites, peraluminous granite plutons, rapakivi granites, leptynites, komatiitic and tholeiitic sills/dikes, and flood basalts that range in age between late Paleoproterozoic and Early Tertiary (cf. Chatterjee et al., 2008;Ghose and Chatterjee, 2008;Chatterjee and Ghose, 2011). Cretaceous magmatic events are also recorded (cf. Baksi, 1995;Kent et al., 1997Kent et al., , 2002Chalapathi Rao et al., 2014;Srivastava et al., 2014). Sedimentary rocks of the Gondwana Supergroup were deposited during the Carboniferous (ca. 295-250 Ma; Mahadevan, 2002) and occur mainly in troughlike depressions arranged in a more or less en echelon fashion along distinct shear/fault zones. These include the NW-trending Mahanadi basin, and the E-trending Damodar Valley basins and subsidiary basins to the north and northeast of Damodar Valley (Mahadevan, 2002;Chakraborty et al., 2003).
Mafic dikes of different ages (Mesoproterozoic to Cretaceous) are well exposed in the CGC and the Damodar Valley sedimentary basins of eastern India (cf. Kent et al., 2002;see Fig. 1B). Several ENE-to E-and NNW-to NW-trending Mesoproterozoic mafic dikes intrude the basement gneisses of the CGC (Kumar and Ahmad, 2007;. Cretaceous mafic dikes intrude the Damodar Valley sedimentary basins, as well as Precambrian basement rocks, and trend either NNW to NW or NNE to ENE (Kent et al., 2002;Paul, 2005;Srivastava et al., 2014). The two Cretaceous dike sets have distinct geochemical characteristics . Both sets of dikes are fresh and rarely show any hydrothermal alteration.
In this study, we focused on the dikes of the Raniganj Basin (Fig. 2). Kent et al. (2002) obtained 40 Ar/ 39 Ar ages for three Cretaceous dikes. Two of these dikes were from the Raniganj Basin and have NNW trends. One of these was interpreted to be the Salma dike, and the other was interpreted to be a separate dike parallel to the Salma dike. The third dike, also trending NNW, was from the Koderma area (see Fig. 1B). The two dikes studied by Kent et al. (2002) from the Raniganj Basin yielded quite different ages, with the one interpreted to be the Salma dike dated at 65.4 ± 0.3 Ma and the other dated at 112.5 ± 0.5 Ma. The third dike, from the Koderma area, yielded an age of 115.3 ± 0.4 Ma. There are several other NNW-to NW-trending mafic dikes exposed in the Raniganj Basin that could be related to the Salma dike . A couple of NEtrending dikes have also been reported from the Raniganj Basin (see Fig. 2).

Sampled Dikes
Since this work aimed to settle the disagreement concerning the emplacement age of the Salma dike, a very careful sampling of dikes in the Raniganj Basin was carried out. The NNWtrending Salma dike is >50 km long, and its width varies between 40 and 100 m (Fig. 3A). It is one of the prominent geological features 3 SRIVASTAVA ET AL. | 40 Ar/ 39 Ar geochronology of Cretaceous mafic dikes, eastern India in the basin. Figure 3 shows locations where geochronological and geochemical samples were collected for the present study, as well as locations for the geochronological samples of the earlier study by Kent et al. (2002). The four samples in the present study were used for geochemistry, and two of these were also selected for 40 Ar/ 39 Ar dating. One dated sample (RG09/ SM14) came from the Chinchuria area and definitely belongs to the Salma dike (Fig. 3C). The other (RJ13/3) is from a NE-tending dolerite dike ~600 m north of the Domohani Railway Colony, and very close to the location of the dated sample that Kent et al. (2002) interpreted to be from the Salma dike. Our NE-trending dike likely intersects the Salma dike ( Fig. 3B), although the actual intersection is not exposed in the field. The mafic dike samples are coarse grained, show ophitic/subophitic textures, and consist mainly of augite/titan augite, plagioclase, and ilmenite, with a few grains of rutile and apatite.

Ar/ 39 Ar Dating
In the present study, as noted above, two dolerite samples were collected from the mafic dikes of the Raniganj Basin for 40 Ar/ 39 Ar geochronological analysis. Sample RG09/SM14 came from the NNW-trending Salma dike, and sample RJ13/3 came from a NE-trending dike (see locations in Fig. 3). In order to constrain the emplacement ages and avoiding excess argon, the whole-rock fragments from the 60-80 mesh (200-280 μm in diameter) were checked carefully under a binocular microscope to remove impurities, enclaves, and phenocrysts (see Appendix for details). The final groundmass separates were dated at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS), Beijing.
Aliquots from both the samples, each containing 10 grains, were wrapped in aluminum foil to form wafers and stacked in quartz vials with the international standard YBCs (29.286 ± 0.045 Ma; Wang et al., 2014). Neutron irradiation was carried out in position H8 of the 49-2 Nuclear Reactor (49-2 NR), Beijing (China), with a flux of ~6.5 × 10 12 n (cm 2 s) -1 for 24 h. The CO 2 laser fusion step-heating technique was used for 40 Ar/ 39 Ar analyses.
Isotopic measurements were made on the Noblesse mass spectrometer at IGGCAS. Ca and K correction factors were: [ 36 Ar/ 37 Ar] Ca = 0.000261 ± 0.0000142, [ 39 Ar/ 37 Ar] Ca = 0.000724 ± 0.0000281, [ 40 Ar/ 39 Ar] K = 0.00088 ± 0.000023. All argon isotopes were measured using electron multipliers in ion-counting mode. Ages were calculated using the decay constant (5.531 × 10 -10 yr -1 ) reported by Renne et al. (2011), and all errors are quoted at the 2σ level. Different decay constants (Steiger and Jäger, 1977;Min et al., 2000) were used to recalculate the data for sample RG09/SM14 in order to make a direct comparison with the previous results of Kent et al. (2002), and to check for any bias in the ages associated with the different decay constants (Table A1). The recalculation shows that the bias in ages arising from different decay constants is minor (<0.25 Ma; in other words, <0.36% for both the samples) and undetectable within errors (see Table A1; Appendix).
Plateau ages were determined from three or more contiguous steps comprising >50% of the 39 Ar released, revealing concordant ages at the 95% confidence level. The age errors reported here are internal errors, including analytical errors and errors on blanks, Ca and K correction factors, mass-discrimination, and J value; the error on the total decay constant was not propagated into the age error. Uncertainties on all data reported herein are at the 95% confidence level (2σ). The data were processed using ArArCALC (Koppers, 2002). The detailed protocol of the 40 Ar/ 39 Ar analysis technique is described in the Appendix.

Whole-Rock Geochemistry
Although geochemical data from the Cretaceous mafic dikes of the Damodar Valley (including the Salma dike) were presented in Srivastava et al. (2014), four new fresh mafic dike samples were analyzed for whole-rock   geochemistry herein, including the two samples that were later used for 40 Ar/Ar 39 analyses (for sample locations, see Fig. 3). Geochemical data for the four freshly analyzed samples and a sample analyzed earlier by Srivastava et al. (2014) are presented in Table 1. All analyses were carried out at the Activation Laboratories, Ltd., Ancaster, Ontario, Canada. An inductively coupled plasma-optical emission spectrometer (ICP-OES; model: Thermo-JarretAsh ENVIRO II) was used to analyze major elements, whereas an inductively coupled plasma-mass spectrometer (ICP-MS; model: Perkin Elmer Sciex ELAN 6000) was used to determine trace and rare earth element (REE) concentrations. The precision was ~5% and 5%-10% for the major oxides and trace elements, respectively, when reported for concentrations at 100× detection limit. Several geostandards were run with the studied samples to check accuracy and precision. The analytical procedure was described by Gale et al. (1997), and details are available on the Activation Laboratories Ltd. website (http://www.actlabs.com).

Geochemical Characteristics
The geochemical characteristics of the Cretaceous mafic dikes exposed in Damodar Valley were comprehensively studied by . They determined that the dikes trending NNE to ENE have geochemical signatures that are distinct from the dikes trending NW to NNW. The NE-trending dikes have higher concentrations of Ti, Fe, and high field strength elements (HFSEs) and are referred to as high-Ti dolerite dikes. The NW-trending dikes have comparatively lower concentrations of Ti, Fe, and HFSEs and are referred to as low-Ti dolerite dikes. It is significant that both have comparable MgO contents, despite different HFSEs values, suggesting their derivation from manifestly different mantle melts.
Geochemical data obtained for samples in the present study (Table 1) are comparable with the high-Ti dolerite and low-Ti dolerite groupings identified in the earlier geochemical study . Samples from the NEtrending mafic dike studied herein showed high concentrations of Ti, Fe, and HFSEs compared to the NNW-trending Salma dike samples. The NE dike samples fall in the basalt field on a total alkali-silica (TAS) classification diagram, whereas the NNW Salma dike samples fall in the basaltic andesite field ( Fig. 4A; Irvine and Baragar, 1971;Le Maitre et al., 2002).
Geochemical distinctions between the NWand NE-trending sets of dikes in the present study are also very evident in variation diagrams and comparable with the previously studied Cretaceous mafic dikes of this region. Two such variation plots are shown in Figures 4B and 4C. The multi-element (ME) and REE patterns of the high-Ti dolerite and low-Ti dolerite samples are also distinct (Fig. 5). The REE pattern of the NE-trending dike is comparatively more inclined (La N /Lu N ratios ~3.4) than the REE patterns of the NNW-trending Salma dike (La N / Lu N ratios ~2.4). This suggests that the high-Ti dolerite samples crystallized from a melt generated through a lower percentage melting of a mantle source than the low-Ti dolerite samples (cf. Cullers and Graf, 1984;Hirschmann et al., 1998).
The Zr/Y and Nb/Y ratios were examined to identify any involvement of a plume component in their genesis ( Fig. 6A; Fitton et al., 1997;Baksi, 2000); the samples of both groups showed a possible plume connection and less likelihood of any crustal involvement. Correspondingly, the Th/Ta versus La/Ta ratio plot ( Fig. 6B; cf. Kent et al., 1997; also indicates less possibility of any crustal contamination, as samples of both dike sets plot well away from the average continental crust and close to the primordial mantle value. Their geochemistry is also comparable with the geochemistry of basalts of the Kerguelen Plateau, Bunbury Island, and Rajmahal Group I ( Fig. 6B; cf. Storey et al., 1992;Salters et al., 1992;Mahoney et al., 1995;Frey et al., 1996). Therefore, it appears that crustal contamination has not played a significant role in the petrogenesis of the samples studied herein.
The observed geochemical characteristics of the dike samples studied herein are consistent with those of the earlier study (cf. ; however, they appear to be inconsistent with the emplacement age reported by Kent  (Kent et al., 2002), has geochemistry that is indistinguishable from that of another NNW-trending (low-Ti dolerite group) dike that was dated at 112.5 ± 0.5 (Kent et al., 2002), but quite different from the geochemistry of NE-trending (high-Ti dolerite group) dikes. The paleomagnetic pole for the Salma dike is similar to that of the ca. 117 Ma Rajmahal Traps (Patil and Arora, 2008), also suggesting that there could be a problem with the reported age of the Salma dike. To resolve this issue, we carried out 40 Ar/ 39 Ar dating for two samples, one from the Salma dike and the other from a NE-trending dike, as described in the next section.

Ar-39 Ar Ages
The 40 Ar/ 39 Ar data for the two geochronological samples are listed in Table 2, and age spectra and inverse isochron plots are shown in Figure 7. The sample from the NNW-trending Salma dike (RG09/SM14), collected from the Chinchuria area (Fig. 3C), yielded consistent plateau (116.0 ± 1.4 Ma) and inverse isochron (117.8 ± 2.9 Ma) ages within errors, suggesting there is no excess argon residing in the sample. This is also indicated by the trapped argon composition of 277.4 ± 25.8, which is within error of atmospheric Ar (295.5 ± 0.5). Therefore, we used the more precise age of 116.0 ± 1.4 Ma as the emplacement age of this sample. This age is quite different from the 40 Ar/ 39 Ar age of 65.4 ± 0.3 Ma for the Salma dike reported by Kent et al. (2002).
The sample from the NE-trending dike (RJ13/3) was collected ~600 m north of the Domohani Railway Colony (Fig. 3B); as noted earlier, this dike likely crosscuts the NNW-trending Salma dike, although the actual intersection is not observed. Similar to our analysis of the Salma dike data, we prefer the plateau age of 70.5 ± 0.9 Ma, rather than the inverse isochron age of 69.5 ± 3.5 Ma, as representing the time of emplacement of this dike. It is noteworthy that the age of the NE-trending dike is close to the age that was reported for the NNW-trending Salma dike by Kent et al. (2002).
Our NE-trending dike sample (RJ13/3) was collected close to the sampling site (i.e., 2 km northeast of Asansol; as mentioned in Kent et al., 1997;see Fig. 3B) that Kent et al. (2002) interpreted to be on the Salma dike and where they obtained an age of ca. 65 Ma. The similar 40 Ar/ 39 Ar ages of the NE-trending dike (sample RJ13/3) in this study and the dike of Kent et al. (2002) strongly suggest that the sample in the Kent et al. (2002) study was actually collected from the NE-trending dike (or perhaps another nearby NE-trending dike, given that there is a small difference of ~5 m.y. between the ages for the two dikes), and was not collected from the Salma dike. This interpretation is further strengthened by considering the location of another dated sample (RJ 1-19-1; ~1 km west of Kalidaspur; Fig. 3D) collected by Kent et al. (1997) from a NNW dolerite dike, which yielded a 40 Ar/ 39 Ar age of 112.5 ± 0.5 Ma (Kent et al., 2002). The location of this sample indicates that it was likely collected from the SE extension of the Salma dike (see Figs. 3A and 3D), not from another dike parallel to the Salma dike at Kalidaspur, as suggested by Kent et al. (1997).
All these observations indicate that the sample that Kent et al. (2002) dated and interpreted to be from the NNW-trending Salma dike, based on the description of Kent et al. (1997), was actually from a NE-trending dike, rather than from the Salma dike. In addition, it is clear that the correct emplacement age of the Salma dike is 116.0 ± 1.4 Ma, not 65.4 ± 0.3 Ma.

Linkage to Mantle Plume Tectonics and Large Igneous Provinces
From the new 40 Ar/ 39 Ar geochronology and geochemical data presented in this work and the discussion above, the following two broad inferences can be drawn: (1) The NNW-trending Salma dike was emplaced at 116.0 ± 1.4 Ma and not at 65.4 ± 0.3 Ma (as suggested earlier; Kent et al., 2002).

RESEARCH
(2) The eastern and northeastern Indian Shield hosts magmatic activity related to both the Kerguelen and Réunion plumes (cf. Srivastava et al., 2014), as evident from the Early Cretaceous (ca. 118-101 Ma) and Late Cretaceous (ca. 70-65 Ma) ages obtained by Kent et al. (2002) and in the present study.
The younger Late Cretaceous magmatic event (70-65 Ma; Kent et al., 2002;present work) of eastern and northeastern India was likely connected to the Réunion plume and the associated Deccan large igneous province of western India (Kent et al., 2002;Paul, 2005;Srivastava et al., 2014). While it is well established that the bulk (>90%) of Deccan volcanism occurred at 65 (±1) Ma, close to the Cretaceous-Tertiary boundary (Duncan and Pyle, 1988;Baksi, 1994;Allègre et al., 1999;Hofmann et al., 2000;Chenet et al., 2007;Schoene et al., 2015Schoene et al., , 2019Sprain et al., 2019), more limited Deccan volcanism occurred before and after the main event between 69 and 62 Ma (Widdowson et al., 2000;Pande, 2002;Hooper et al., 2010;. Volcanic rocks that are thought to be associated with the Réunion plume have also been identified far from the main Deccan large igneous province in western India. For example, Mahoney et al. (2002) obtained a 72 Ma age for Réunion hotspot-related lavas in the South Tethyan suture zone of Pakistan that have very similar isotopic and trace-element geochemistry to the Deccan large igneous province (Mahoney, 1988). In addition, 64.7 ± 0.5 Ma Rajahmundry lavas along India's eastern coast, >400 km from the main Deccan lavas, are thought to have been part of the Deccan event (Knight et al., 2003). It appears likely that the huge (2000-2500-kmdiameter) Réunion mantle plume head and/or heat associated with it was responsible for all these volcanic eruptions, as well as for continental breakup of greater India with the Seychelles (cf. White and McKenzie, 1989;Mahoney et al., 2002).
Overall, the new 40 Ar/ 39 Ar ages obtained from the two geochemically distinct sets of doleritic dikes in the CGC of eastern India indicate that they were intruded in extensional tectonic settings associated with the Kerguelen and Réunion mantle plumes. However, in both the cases, the plume was likely not directly involved, but rather provided additional heat necessary to melt the mantle source. More robust geochronological, radiogenic isotope geochemical, and paleomagnetic data are required to determine the overall extent of these two events and better constrain the relative contributions of the plumes to their genesis.

CONCLUSIONS
The NNE to ENE and NW to NNW Cretaceous mafic dikes exposed in the Chhotanagpur gneissic complex of eastern India show very distinct geochemical characteristics and are identified as high-Ti and low-Ti dolerites, respectively. In the Raniganj Basin, dikes of these two sets were dated by the 40 Ar/ 39 Ar method to resolve a controversy concerning the emplacement age of the long NNW-trending Salma dike.
The NE-trending dike has a 40 Ar/ 39 Ar plateau age of 70.5 ± 0.9 Ma, whereas the NNW-trending Salma dike has a plateau age of 116.0 ± 1.4 Ma, indicating that the Salma dike was emplaced at ca. 116 Ma, not at ca. 65 Ma as suggested previously.
The genesis of the two dike sets is thought to be from melts generated during extensional tectonic regimes developed by sizeable mantle plume heads, with the plume heads providing the additional heat necessary to melt the mantle source.
The ca. 116 Ma Salma dike is interpreted as part of the Greater Kerguelen large igneous province associated with the Kerguelen mantle plume, whereas the younger ca. 70 Ma mafic dike likely belongs to the Deccan large igneous province linked to the Réunion plume.

Sample Preparation and Irradiation
The groundmass of a volcanic rock is usually quenched from melt at or near the surface during eruption. Therefore, it is a proxy for the time of eruption, and its initial argon composition is similar to that of the atmosphere (avoiding contamination from deep excess argon). In contrast, phenocrysts, such as plagioclase, may crystallize in a magma chamber in the lower crust or mantle sometime prior to its emplacement, and, hence, may trap deep argon that is different from the atmosphere (excess argon). Therefore, to constrain the emplacement age and avoid excess argon, the groundmass portions of the dolerite samples were analyzed in this study.
The samples were crushed in a steel jaw crusher followed by grinding in a steel mill. Grinding was accomplished in 1-2 second steps and alternated with sieving at 200 μm until 90% of the rock powder was reduced to <200 μm. The remaining 200-280 μm fraction included the hardest components of the altered samples and is assumed to be the freshest parts of the samples. The groundmass aliquots were then handpicked under a binocular microscope from this 200-280 μm fraction. The aliquots of every groundmass sample were inspected carefully to remove phenocrysts or xenocrysts, and all other impurities.
The separates of groundmass were further cleaned by acid-leaching with 2 N HNO 3 for 30 min in an ultrasonic bath and heated to 50 °C in order to remove the calcite and other nitric-soluble phases. The leached separates were washed in an ultrasonic bath in deionized water three times, each for 30 min, and then in acetone three times (for 30 min), and dried in preparation for irradiation.
All sample aliquots were wrapped in aluminum foil to form wafers and stacked in quartz vials with the international standard YBCs. The vials were 40 mm in length, with an inner diameter of 6.5 mm. The position for each sample was recorded as the distance from the bottom of the vial. Then, the vials were sealed in vacuo and put in a quartz canister. The canister was wrapped with cadmium foil (0.5 mm in thickness) for shielding slow neutrons in order to prevent interface reactions during the irradiation.
Neutron irradiation was carried out in position H8 of 49-2 Nuclear Reactor (49-2 NR), Beijing (China), with a flux of ~6.5 × 10 12 n (cm 2 s) -1 for 24 h, yielding J values of ~0.0017. The H8 position lies in the core of the reactor and receives flux from all directions. Specimens were rotated during the fast neutron irradiation to get homogeneously distributed neutrons. Six to eight replicate analyses of the monitors from each vial were conducted to constrain the vertical neutron fluence gradient to within ±0.7%. This additional uncertainty was propagated into the plateau and inverse isochron ages. External uncertainties arising from the decay constants and primary K-Ar standards were not propagated. Errors are reported at the 2σ confidence level.

Gas Extraction and Mass Spectrometry Analyses
Argon isotope analyses were carried out on a mass-spectrometry system at the 40  the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS), Beijing, China. The system, a product of NU Instrumental Company based in the UK, consists of a CO 2 laser (50 W, 10.6μm wavelength, 999 Hz), a gas purification preparation line, and a Noblesse mass spectrometer. The irradiated samples were moved into 2-mm-diameter wells in a copper sample holder, which was then placed in a ZnSe windowed chamber. After preheating under pumping at 140 °C for 72 h by using heating tapes, the samples were heated in stepwise fashion to extract gas by using a continuous 2-mm-spot CO 2 laser beam. Twelve heating steps, each for 3 min, were performed for each sample with laser outputs from 1.6 to 2.7 W at steps of 0.1 W until the samples were completely molten (the grains turned into a black ball that did not glitter any longer under lasing). Each gas fraction was purified simultaneously during the 3 minutes using two SAES NP10 Zr-Al getters built in the prep-line, one of which ran at 400 °C to remove active gases (e.g., N 2 , CO, CO 2 ), and the other one of which was kept at room temperature to absorb H 2 . After purification, the valve between the prep-line and the mass spectrometer was switched on, and the gas was introduced into the mass spectrometer for argon isotope analyses. The entire volume of gas in the laser chamber, prepline, and mass spectrometer was ionized and analyzed. The purity of gas and a low level of residual active gases in the mass spectrometer were maintained by another two Zr-Al getters at room temperature. Argon isotopes were measured in the order of 40 Ar, 39 Ar, 38 Ar, 37 Ar, and 36 Ar, and involved 13 cycles of replicate measurements. For each peak and baseline measurement, the ion beam of 40 Ar, 39 Ar, 38 Ar, 37 Ar, and 36 Ar was integrated for 5, 10, 5, 5, and 20 s, respectively, with baselines for 5 s. Baseline measurements were taken at a half mass unit away from the peaks, except between 39 Ar and 40 Ar, where the baseline was set at 39.3. The final measurement for each isotope was extrapolated to zero time using the leastsquare regression from the 13 cycle measurements. Usually the "memory effect" of the instrument was negligible; the variation of 40 Ar/ 39 Ar was less than 1% in one run, and the analytic error was less than 0.1%.
The Noblesse mass spectrometer at IGGCAS is configured as a multicollector with a high-mass Faraday cup with a 10 11 Ω resistor and three discrete dynode secondary electron multipliers performing in ion-counting mode. Due to its well-focused ion beam and wide collector-slit spacing, the Noblesse mass spectrometer yields steep-margined and flat-topped broad peaks, with a peak flatness of ≤1 part in 10 3 over ±300 ppm on the Faraday cup and ≤2 parts in 10 3 over ±150 ppm on multipliers (without retardation filter). Electrostatic filters are positioned at the entrance of the multipliers to suppress stray ions. Interferences produced by HCl and hydrocarbon species overlap the high-mass side of all argon peaks. Argon isotopic measurements were obtained on the unaffected low-mass sides of the peaks to obtain 36 Ar measurements that were free of interference species (Saxton, 2015). Static backgrounds of the IGGCAS Noblesse are typically ~4.5 × 10 -17 , 1.3 × 10 -18 , 3.5 × 10 -19 , 5.7 × 10 -18 , and 4.2 × 10 -19 mol for 40 Ar through 36 Ar. Procedural blanks (i.e., simulated analyses without sample gas) are ~3.8 × 10 -16 , 2.0 × 10 -18 , 5.1 × 10 -19 , 5.5 × 10 -18 , and 1.9 × 10 -18 mol for 40 Ar through 36 Ar, respectively. Blanks were monitored every two measurements for correction. The efficiency (gain) of multipliers against the Faraday cup, which depends on amount of gas and is governed by "dead time" of the multiplier, is monitored every month. Mass discrimination was monitored using an online air pipette from which multiple measurements were made before and after each experiment. The mean over this period was 1.00100 ± 0.00021 per amu, and the uncertainty of this value was propagated into all age calculations.
Ages were calculated using the decay constant of Renne et al. (2011), and all errors are quoted at the 2σ level. In order to make our results comparable directly with the previous study (Kent et al., 2002) on the same dike sets, the age results for RG09/SM14 were recalculated (Table A1) using the different decay constants reported by Steiger and Jäger (1977) and Min et al. (2000). The results show that the bias in ages caused by the different decay constants is minor (<0.25 Ma; Table A1) and is undetectable within errors (Table A1).