1. Introduction
Potassium (K) behaves as a large ion lithophile element during crust-mantle differentiation, leading to its enrichment in the Earth’s crust (average K = 12930 ppm) and significant depletion in the mantle (average K = 191 ppm) (Rudnick and Gao, Reference Rudnick and Gao2003). This characteristic makes K a valuable geochemical tracer of crust-mantle interactions (e.g. Hu et al., Reference Hu, Teng, Plank and Chauvel2020; Pan et al., Reference Pan, Xiao, Su, Robinson, Li, Wang and Liu2024). Recent development in high-precision K isotope analytical methods has enabled the determination of K isotope compositions in K-rich reservoirs of the Earth (Tuller-Ross et al., Reference Tuller-Ross, Marty, Chen, Kelley, Lee and Wang2019a; Huang et al., Reference Huang, Teng, Rudnick, Chen, Hu, Liu and Wu2020; Hu et al., Reference Hu, Teng, Plank and Chauvel2020, Reference Hu, Teng, Helz and Chauvel2021a; Moynier et al., Reference Moynier, Hu, Wang, Zhao, Gérard, Deng, Moureau, Li, Simon and Teng2021; Wang et al., Reference Wang, Li, Li, Tian, Koefoed and Zheng2021; Li et al., Reference Li, Cui, Pan, Wang, Gao, Liu, Yuan, Su, Zhao, Teng and Han2022; Ding et al., Reference Ding, Liu, Su, Li, Bai, Pan, Hu and Pang2023; Gamaleldien et al., Reference Gamaleldien, Wang, Johnson, Ma, Anbar, Zhang, Olierook and Kirkland2024). However, the K isotope composition of the deep mantle remains poorly constrained due to its low K content (190–260 ppm; Meybeck, Reference Meybeck2003) and the mobility of K. Available K isotope data for mantle peridotite rocks are scarce and exhibit significant variation (Ionov and Wang, Reference Ionov and Wang2021; Wang and Ionov, Reference Wang and Ionov2023). This limited understanding hampers our ability to comprehend K recycling processes within the Earth and restricts the application of K isotopes in geological studies.
The studies of K isotopes in various basalt types and mantle xenoliths have provided valuable insights into the isotope composition of the Earth’s mantle beneath both oceanic and continental regions (e.g. Tuller-Ross et al., Reference Tuller-Ross, Marty, Chen, Kelley, Lee and Wang2019a, Reference Tuller-Ross, Savage, Chen and Wangb; Hu et al., Reference Hu, Teng and Chauvel2021b; Wang and Ionov, Reference Wang and Ionov2023; Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025). Specifically, mid-ocean ridge basalts (MORBs) exhibit a remarkable degree of isotopic homogeneity, with an average δ41K value of −0.44 ± 0.17 ‰ (Tuller-Ross et al., Reference Tuller-Ross, Marty, Chen, Kelley, Lee and Wang2019a). Studies of Kilauea Iki lavas from Hawaii, despite varied bulk chemical compositions, reveal limited δ41K variation (−0.42 to −0.37 ‰), indicating no fractionation of K isotopes during basaltic magma differentiation (Hu et al., Reference Hu, Teng, Helz and Chauvel2021a). These findings suggest a uniform isotopic signature in the mantle beneath oceans, allowing for estimates of K isotope compositions for the pristine mantle (δ41K = −0.42 ± 0.08 ‰) and the bulk silicate Earth (BSE, δ41K = −0.42 ± 0.07 ‰) (Hu et al., Reference Hu, Teng, Helz and Chauvel2021a). In contrast, alkali basaltic lavas derived from the continental mantle display a broader range of δ41K values, from −1.330 to 0.126 ‰, attributed to source heterogeneity and/or fractionation of K-rich minerals during magma differentiation (e.g. Sun et al., Reference Sun, Teng, Hu, Chen and Pang2020; Du et al., Reference Du, Luo, Wang, Palmer, Ersoy and Li2024; Su et al., Reference Su, Pan, Bai, Li, Cui and Pang2024, Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025). Direct measurements of mantle xenoliths from the lithospheric mantle demonstrate highly variable and generally low δ41K values (Ionov and Wang, Reference Ionov and Wang2021; Wang and Ionov, Reference Wang and Ionov2023), reflecting the complex history of metasomatic processes within the sub-continental lithospheric mantle. These findings underscore the significance of subducted materials and metasomatism in shaping the isotopic composition of the continental mantle and the magmas derived from it.
It is essential to note that materials recycled through subduction can reach the asthenosphere or lower mantle (e.g. Su et al., Reference Su, Hu, Teng, Xiao, Zhou, Sun, Zhou and Chang2017), significantly deeper than the sub-continental lithospheric mantle. Currently available K isotope data are primarily obtained from rocks and melts derived from shallow continental mantle (Ionov and Wang, Reference Ionov and Wang2021; Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025). In contrast, kimberlite magmas are believed to originate from the upper convecting mantle at depth <250 to 300 km rather than the lithospheric mantle (Giuliani et al., Reference Giuliani, Schmidt, Torsvik and Fedortchouk2023). Their sources may contain volatile-rich recycled materials (Mitchell, Reference Mitchell1995; Gernon et al., Reference Gernon, Jones, Brune, Hincks, Palmer, Schumacher, Primiceri, Field, Griffin, O’Reilly and Keir2023), and melting is likely triggered by the oxidation of reduced carbon during the upwelling of mantle peridotites (Giuliani et al., Reference Giuliani, Schmidt, Torsvik and Fedortchouk2023). Fractional crystallization commonly occurs in kimberlites due to the presence of olivine, garnet, ilmenite, perovskite and phlogopite phases (Mitchell, Reference Mitchell2013; Sparks, Reference Sparks2013). In addition, kimberlites are well-known as sources of diamonds. Conducting K isotope studies on kimberlites provides an opportunity to explore the composition of the deep mantle and investigate the effects of subducted slabs. This study presents K isotope compositions of diamondiferous kimberlites from Mengyin in the North China Craton, along with petrographic and elemental investigations, to elucidate K isotope fractionation during post-emplacement alteration and magma differentiation and to constrain the K isotope composition of deep mantle beneath the North China Craton.
2. Geological background of Mengyin diamondiferous kimberlites and samples
The Mengyin Palaeozoic kimberlite field is located in the eastern North China Craton (Fig. 1a) and consists of three belts: Changma, Xiyu and Poli (Lu et al., Reference Lu, Wang, Chen and Zheng1998; Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010; Tian et al., Reference Tian, Wang, Tian, Shan, Wang, Chi, Ma, Chu, Li and Lv2023). These kimberlites intruded into the Archaean granitic gneisses of the Taishan Group, which originated from the melting of juvenile materials, as evidenced by low initial 87Sr/86Sr ratios and high positive εNd(t) values (e.g. Jahn et al., Reference Jahn, Auvray, Shen, Liu, Zhang, Dong, Ye, Zhang, Cornichet and Mace1988). The Mengyin kimberlites are characterized by irregular dikes and vents and are significant hosts for major diamond deposits in China (Lu et al., Reference Lu, Wang, Chen and Zheng1998). The emplacement age of the Mengyin kimberlites has been established through various methods. A phlogopite megacryst from the Mengyin kimberlite yielded a mean plateau age of 466.3 ± 0.3 Ma and a39Ar-40Ar isochron age of 464.9 ± 2.7 Ma (Zhang and Yang, Reference Zhang and Yang2007). This age aligns with U-Pb perovskite ages of 456 ± 8 Ma (Dobbs et al., Reference Dobbs, Duncan, Hu, Shee, Colgan, Brown, Smith, Allsopp, Meyer and Leonardos1994), 470 ± 4 Ma (Yang et al., Reference Yang, Wu, Wilde, Liu, Zhang, Xie and Yang2009) and 480.6 ± 2.9 Ma (Li et al., Reference Li, Wu, Li, Qiu, Liu, Yang and Tang2011). The samples analysed in this study were collected from the Shengli 1# pipe (35°40′08”N, 117°46′40”E) within the Changma belt (Fig. 1b), which represents the largest diamond mine in the Mengyin kimberlite field. The kimberlites in this pipe exhibit a predominantly porphyritic texture (Fig. 1c), featuring megacrystic olivine and phlogopite (Fig. 1d, e). The matrix is composed of relatively fine-grained olivine, phlogopite, perovskite, apatite, carbonate and Fe-Ti-oxides.
Figure 1. (a) Location of Mengyin diamond-bearing kimberlite in North China Craton. (b) Open pit of diamond deposit in Mengyin kimberlite pipe. (c) Occurrence of breccia and matrix in typical Mengyin kimberlite. (d, e) Occurrence of calcite (Cal), apatite (Ap), phlogopite (Phl) and olivine (Ol) in less altered kimberlites. (f) Completely altered kimberlite.
The Mengyin kimberlites show variable degrees of alteration (Fig. 1e), with differing amounts of chlorite and serpentine. This alteration has been attributed to self-alteration by cognate fluids derived from the magmas rather than post-emplacement alteration by crustal fluids (Liu et al., Reference Liu, Zhang, Sun and Ye2004; Zhang and Yang, Reference Zhang and Yang2007; Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010). Several lines of evidence support this interpretation: (1) preservation of fresh primary minerals (Fig. 1d, e), (2) elemental compositions comparable to unaltered kimberlite fields globally, rather than to carbonated or highly carbonated kimberlite fields (Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010), (3) constant Nd and Hf isotope ratios (εNd(t) = −1.5 to 2.1, εHf(t) = −2.37 to −0.30; Lu et al., Reference Lu, Wang, Chen and Zheng1998; Zhang and Yang, Reference Zhang and Yang2007; Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010) and (4) limited variation in C isotopes (δ13CPDB = −7.6 to −4.8 ‰; Liu et al., Reference Liu, Zhang, Sun and Ye2004). Based on these observations, most of the Mengyin kimberlite samples are considered to preserve their primary geochemical compositions and are referred to as pristine kimberlite samples. However, a few samples lack fresh phases in the matrix, showing only olivine pseudomorphs (Fig. 1f), which are indicative of post-emplacement alteration; these are thus classified as altered kimberlites.
3. Analytical methods
Whole-rock major elements and some trace elements were determined using a Shimadzu X-ray fluorescence spectrometer (XRF-1500) on fused glass beads at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS), Beijing, China. Analytical uncertainties were 1–3 % for elements present in concentrations >1 wt.% and about 10 wt.% for elements present in concentrations <1.0 wt.%.
Potassium isotopic analyses were conducted at Metallogenic Elements and Isotopes Lab at IGGCAS, following the protocol described in Li et al. (Reference Li, Cui, Pan, Wang, Gao, Liu, Yuan, Su, Zhao, Teng and Han2022, Reference Li, Zhao, Su, Gao, Wang and Liu2023). Approximately, 5–20 mg of sample powders were weighed and digested using a mixture of concentrated HNO3 and HF. The digested sample solutions were evaporated to dryness and then treated sequentially with aqua regia and 6 mol L−1 HNO3. After evaporating the solutions, the final residues were fully dissolved in 0.5 mol L−1 HNO3 twice prior to column separation. The sample solution was loaded onto pre-conditioned 2 mL Bio-Rad AG50W-X8 (200–400 mesh) resin and then eluted with 15 mL of 0.5 mol L−1 HNO3 to remove the matrix elements. The same purification process was repeated twice for silicate lavas and four times for carbonatites to ensure complete matrix removal. The final K solution was redissolved with 2 % HNO3, ready for measurement. The total procedure blank for K isotope analyses is < 30 ng K, which is negligible compared with tens of μg of K in the solution from sample chemical purification.
Potassium isotopic measurements were performed on the Nu Sapphire CC-MC-ICP-MS (Nu Instruments, Wrexham, UK) using the low-energy path. The hexapole collision cell utilizes He and H2 gas to greatly reduce various Ar-based polyatomic species to very low levels; hence, K isotopic ratios can be measured in the low-resolution mode. An auto-sampler SC-2DX (Elemental Scientific, USA) was connected to an Apex Omega desolvation nebulizer with uptake of 100 μL min−1 (Elemental Scientific, U.S.A.) system for sample introduction. One Faraday cup is connected to a pre-amplifier fitted with a 1010 Ω resistor for the collection of 39K+ ion beam, while the other two Faraday cups using 1011 Ω resistors collect 41K+ and mass 40 beams, respectively. Potassium isotopic data are reported in δ notation relative to SRM 3141a, using the sample-standard bracketing technique for instrumental mass fractionation correction. The 200 ng mL−1 K solution was used during our measurements, yielding an ion intensity of ∼200 V on 39K (20 V relative to a 1011 Ω amp). The K concentration of each sample and standard was matched to within 5%. Each analysis consisted of 1 block of 50 cycles with 4 s integrations. Five repeated analyses were conducted on each sample solution. Geostandards yielded δ41K values of −0.398 ± 0.015 ‰, −0.382 ± 0.032 ‰ and −0.463 ± 0.029 ‰ for BCR2, −0.531 ± 0.016 ‰ for GSR1, −0.427 ± 0.031 ‰ for G-2, −0.309 ± 0.037 ‰ for GSR17 and −0.278 ± 0.030 ‰ for DNC-1, consistent with literature data (Chen et al., Reference Chen, Tian, Tuller-Ross, Korotev-Randy and Wang2019; Xu et al., Reference Xu, Hu, Chen, Huang, Sletten, Zhu and Teng2019; Li et al., Reference Li, Han, Zhang and Miao2020, Reference Li, Cui, Pan, Wang, Gao, Liu, Yuan, Su, Zhao, Teng and Han2022; Moynier et al., Reference Moynier, Hu, Wang, Zhao, Gérard, Deng, Moureau, Li, Simon and Teng2021). Replicated analyses of samples are consistent within analytical uncertainty.
4. Results
The pristine kimberlite samples from Mengyin exhibit variable compositions, with SiO2 content ranging from 24.19 to 35.94 wt.%, Al2O3 from 1.20 to 6.79 wt.%, MgO from 21.99 to 37.84 wt.%, CaO from 1.75 to 15.88 wt.% and K2O from 0.02 to 1.17 wt.%. Their TiO2, MnO and TFe2O3 (total Fe) contents range from 0.81 to 1.32 wt.%, 0.09 to 0.17 wt.% and 6.51 to 9.44 wt.%, respectively. The trace elements in these samples also show considerable variability, with Co contents of 43.2–98.0 ppm, Ni of 759–1747 ppm and Rb of 13.8–118.5 ppm. The elemental ratios vary significantly: SiO2/Al2O3 ranges from 4.72 to 29.7, TFe2O3/MgO from 0.20 to 0.36, MgO/CaO from 1.39 to 21.7 and Ni/Co from 14.5 to 21.6, all of which are comparable to the majority of global kimberlites (Fig. 2a, b). The K isotope compositions of the pristine kimberlite samples show a δ41K variation between −0.494 ± 0.057 and −0.270 ± 0.048 ‰. The altered samples demonstrate broader ranges of element contents (e.g. TFe2O3: 4.70–9.63 wt.%; MgO: 16.45–39.64 wt.%; Ni: 292–2445 ppm), elemental ratios (e.g. SiO2/Al2O3: 3.11–59.2; MgO/CaO: 0.99–57.7; Ni/Co: 6.31–30.1) and δ41K values ranging from −1.257 ± 0.020 to −0.114 ± 0.029 ‰ (Table 1), indicating significant differences compared to the pristine kimberlite samples.
Figure 2. Bivariate plots of (a) Si/Mg vs. SiO2/Al2O3, (b) Ni/Co vs. K2O, (c) δ41K vs. MgO/CaO and (d) δ41K vs. CIA (chemical index of alteration) values for Mengyin kimberlites. Data of global kimberlites from the GEOROC Database (https://georoc.mpch-mainz.gwdg.de//georoc/new-start.asp) are plotted in Fig. 2a, b for comparison.
Table 1. Element contents and K isotope compositions of kimberlites in Mengyin, Shandong province, China
Note: r, replicated analysis; N, number and analysis; nd, not detected; CIA, mole ratio of bulk Al/(Al+Ca+Na+K)×100.
5. Discussion
5.a. Effects of post-emplacement alteration on chemical compositions of kimberlites
Alterations in kimberlites commonly include serpentinization, carbonatization, silicification, chloritization and pyritization, which can significantly modify rock compositions (Stripp et al., Reference Stripp, Field, Schumacher, Sparks and Cressey2006; Vasilenko et al., Reference Vasilenko, Kuznetsova, Minin and Tolstov2012; Mitchell, Reference Mitchell2013; Afanasyev et al., Reference Afanasyev, Melnik, Porritt, Schumacher and Sparks2014). For instance, the serpentinization reaction [3Mg2SiO4 (olivine) + SiO2 (fluid) + 4H2O = 2Mg3Si2O5(OH)4 (serpentine)] introduces silica and water into the rocks, resulting in an increased Si/Mg ratio (Sparks et al., Reference Sparks, Brooker, Field, Kavanagh, Schumacher, Walter and White2009; Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010). Silicification and carbonatization typically add Si and Ca to the rocks, respectively, leading to elevated Si/Mg ratios and reduced MgO/CaO ratios (Tompkins et al., Reference Tompkins, Meyer, Han, Hu, Armstrong, Tayer, Gurney, Gurney, Pascoe and Richardson1999; Vasilenko et al., Reference Vasilenko, Kuznetsova, Minin and Tolstov2012). Pyritization and serpentinization can decouple Ni and Co contents in the rocks, causing contrasting variations in the Ni/Co ratio (Cui et al., Reference Cui, Su, Wang and Yuan2024). In many cases, multiple types of alteration can be observed within a single kimberlite occurrence, as evidenced by the compositions of altered kimberlite samples from Mengyin. The Si/Mg, SiO2/Al2O3, Ni/Co and TFe2O3/MgO ratios in these altered kimberlites exceed the restricted ranges found in pristine kimberlite samples (Table 1). These element ratios exhibit covariations in the altered kimberlites (Fig. 2a, b), while in pristine kimberlite samples, they remain relatively constant despite varying K2O contents.
Potassium is a fluid-mobile element, and its isotopes can be sensitive to weathering processes and hydrothermal fluid activities (e.g. Santiago Ramos et al., Reference Santiago Ramos, Coogan, Murphy and Higgins2020). During the chemical weathering of igneous rocks, 41K is preferentially partitioned into fluids relative to 39K, enriching the fluids in 41K and the solid residue in 39K (Chen et al., Reference Chen, Liu and Wang2020; Teng et al., Reference Teng, Hu, Ma, Wei and Rudnick2020; Liu et al., Reference Liu, Wang, Sun, Xiao, Xue and Tuller-Ross2020). Similarly, hydrothermal alteration can lead to K isotope fractionation of up to 1.4 ‰ (Parendo et al., Reference Parendo, Jacobsen and Wang2017; Li et al., Reference Li, Wang, Huang, Wang and Wu2019; Li et al., Reference Li, Coogan, Wang, Takahashi, Shakouri, Hu and Liu2024). In the case of the altered kimberlites from Mengyin, the overall low δ41K values (except for sample 22CM04 with δ41K = −0.114‰) relative to those of the pristine kimberlite samples are consistent with the behaviour of K isotopes during weathering and alteration processes, as anticipated from theoretical models and principal investigations. The lack of correlation between δ41K and MgO/CaO or chemical index of alteration (CIA) (Fig. 2c, d) further confirms that these altered kimberlites underwent multiple alteration processes. Conversely, the δ41K values of the Mengyin pristine kimberlite samples do not show covariations with these geochemical proxies of alteration (Fig. 2c, d), indicating that their K isotope compositions retain their primary features. In addition, negative correlations between δ41K and K2O and Rb in the altered samples (Fig. 3a, b) suggest the removal of fluid-mobile elements K and Rb during alteration.
Figure 3. Bivariate plots of δ41K vs. (a) K2O and (b) Rb for Mengyin kimberlites. Panels c and d are magnified version of panels a and b, respectively. Bulk silicate Earth value (−0.42 ± 0.07 ‰) is from Hu et al. (Reference Hu, Teng, Helz and Chauvel2021a). (e) Modelling of K isotopic variations during magmatic differentiation of Mengyin kimberlite magma. Solid orange lines represent calculated K isotopic compositions of residual melts during phlogopite fractional crystallization by assuming a Rayleigh fractionation process. Dashed blue lines represent calculated mixing lines between the residual melt and phlogopite phenocrysts. The δ41K values of primary melt and phlogopite are assumed as − 0.458 ‰ (sample 22CM18 with the lowest δ41K value and K2O content of 0.17 wt.%; Table 1) and − 0.576 ‰ (average K isotopic compositions of phlogopite separates from basaltic lavas with K2O content of 8.72 wt.%; Su et al., Reference Su, Pan, Bai, Li, Cui and Pang2024), respectively, with phlogopite-melt fractionation factors (Δδ41Kphlogopite-melt = δ41Kphlogopite − δ41Kmelt) of − 0.175 ‰. The orange stars represent the increased K2O contents in melts (0.34 wt.%, 0.51 wt.%, 0.68 wt.%, 0.85 wt.%, 1.70 wt.%, 2.55 wt.%, 3.40 wt.% and 4.25 wt.%) caused by K-poorly mineral accumulations prior to phlogopite.
5.b. K isotope fractionation during magma differentiation of kimberlites
During the magma differentiation of kimberlites, minerals such as olivine, pyroxene and spinel crystallize at an earlier stage, followed by the crystallization of carbonate minerals and hydrous phases like phlogopite (Mitchell, Reference Mitchell2013; Giuliani et al., Reference Giuliani, Schmidt, Torsvik and Fedortchouk2023). In the Mengyin pristine kimberlite samples, K is primarily hosted by phlogopite, which has a high K2O content of approximately 9 wt.%. The contrasting K2O contents between phlogopite and bulk rocks (0.02–1.17 wt.%, Table 1) suggest that the crystallization of solely phlogopite from the magma would reduce the K2O content of the evolved melt. However, the amount of phlogopite in the Mengyin kimberlites is lower than that of other kimberlites worldwide (Lu et al., Reference Lu, Wang, Chen and Zheng1998; Tompkins et al., Reference Tompkins, Meyer, Han, Hu, Armstrong, Tayer, Gurney, Gurney, Pascoe and Richardson1999; Kjarsgaard et al., Reference Kjarsgaard, Pearson, Tappe, Nowell and Dowall2009; Pearson et al., Reference Pearson, Woodhead and Janney2019). The crystallization of K-poor phases (e.g. olivine, garnet and spinel) increases the K2O content of the residual melt. Consequently, K2O tends to increase with the differentiation of kimberlite magmas (Zhang and Yang, Reference Zhang and Yang2007; Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010). Rb, which behaves similarly to K, also shows covariation with K2O contents. In addition, there is no increase in K or Rb with a decrease in Mg/Fe or MgO (not shown), which can be attributed to the abundance of olivine and the effects of phlogopite addition during magma ascent (Brett et al., Reference Brett, Russell and Moss2009; Arndt et al., Reference Arndt, Guitreau, Boullier, Le Roex, Tommasi, Cordier and Sobolev2010).
Phlogopite, due to its higher coordination numbers of K (7–11, Cibin et al., Reference Cibin, Mottana, Marcelli and Brigatti2005; Li et al., Reference Li, Wang, Huang, Wang and Wu2019) compared to silicate melts (5–7, Greaves, Reference Greaves1985), is expected to exhibit a lower δ41K value than the magma from which it crystallizes (Li et al., Reference Li, Wang, Huang, Wang and Wu2019). This expectation aligns with reported K isotope compositions in volcanic rocks (Δ41Kphlogopite-silicate rock = −0.502 to −0.109 ‰, Su et al., Reference Su, Pan, Bai, Li, Cui and Pang2024). The δ41K values of the Mengyin pristine kimberlite samples correlate with K2O contents (correlation coefficient R 2 = 0.66, Fig. 3a, b) and Rb contents (R 2 = 0.70, Fig. 3c, d), suggesting that the fractional crystallization of phlogopite is a primary controller of K isotope fractionation in these kimberlites.
Quantitative modelling is conducted based on Rayleigh fractionation and mixing calculation for the kimberlites (Fig. 3e). The equation for isotopic fractionation in a Rayleigh fractionation process is:
$$[{\delta ^{{\rm{41}}}}{{\rm{K}}_{melt}} = \left( {{\delta ^{{\rm{41}}}}{{\rm{K}}_0} + 1000} \right){f^{\left( {\alpha - 1} \right)}} - 1000]$$
where δ 41 K 0 , the initial K isotope composition of melts, is assumed as −0.458 ‰ (sample 22CM18 with the lowest δ41K value and K2O content of 0.17 wt.%; Table 1), and the initial K2O contents of melts is set as 0.17 wt.%. The fraction of K2O remaining in the melt and other phases is given by f = (F × C melt )/C 0 , where C melt and C 0 represent the K2O concentration in the remaining melts and the initial melts, and F is the fraction of melts remaining. The fractionation factor α is calculated by:
$$[{\Delta ^{{\rm{41}}}}{{\rm{K}}_{phl - melt}} = {\delta ^{{\rm{41}}}}{{\rm{K}}_{phl}} - {\delta ^{{\rm{41}}}}{{\rm{K}}_{melt}} \approx {10^{\rm{3}}}\ln {\alpha _{phl - melt}}]$$
where the phlogopite-melt fractionation factor (Δ 41 K phl-melt = δ 41 K phl -δ 41 K melt ) is −0.175 ‰ in our model, calculated using δ41K values of phlogopite (−0.576 ‰, average K isotopic compositions of phlogopite separates from basaltic lavas with K2O content of 8.72 wt.%; Su et al., Reference Su, Pan, Bai, Li, Cui and Pang2024) and melt (−0.458 ‰).
In the mixing model, the equation for mixing two end-members is:
$$[{C_A} = {\left( {{C_A}} \right)_{melt,0}} \times {F_{melt,0}} + {\left( {{C_A}} \right)_{phl}} \times {F_{phl}}]$$
where C A is the fraction of K2O from end member 1, end member 2, and the mixing phase, and F is the fraction in each member. The K isotope composition of the kimberlites can be calculated by:
$$[{\delta ^{{\rm{41}}}}{{\rm{K}}_{lava}} = {\delta ^{{\rm{41}}}}{{\rm{K}}_{residual{\rm{ }}melts,0}} \times {(C)_{residual{\rm{ }}melts,0}} \times {F_{residual{\rm{ }}melts,0}} + {\delta ^{{\rm{41}}}}{{\rm{K}}_{Phl,0}} \times {(C)_{Phl,0}} \times {F_{Phl,0}}]$$
where the δ 41 K residual melt, 0 and δ 41 K Phl could identify with δ 41 K 0 and δ 41 K lava in the model for Rayleigh fractionation.
The modelling results demonstrate that the obtained K isotopic compositions of the kimberlites are a mixture between phlogopite and evolved melts with phlogopite fractionation. The elevated K2O contents of the samples compared to the modelling results (Fig. 3e) are due to K elevation from the crystallization of K-free minerals, which is consistent with the mineral assemblage of the kimberlites.
5.c. K isotope features of kimberlite source and possible formation mechanism
Since the K isotope variation observed in the pristine kimberlite samples from Mengyin is attributed to magma differentiation (Fig. 3), the lowest δ41K value (−0.458 ‰) of sample 22CM18 can represent the K isotope composition of a relatively primary kimberlite melt, falling within the range of BSE (Hu et al., Reference Hu, Teng, Helz and Chauvel2021a) (Fig. 3c, d). In contrast, mantle xenoliths from the continental lithospheric mantle (non-subduction zone) exhibit overall low δ41K values, ranging from −2.15 to −0.37 ‰ and low K2O contents (< 0.1 wt.%) (Ionov and Wang, Reference Ionov and Wang2021). Among these mantle xenoliths, some highly-metasomatized peridotites align well along the varying trends of the Mengyin kimberlites (Fig. 4a, b). These samples, characterized with higher K2O and Rb contents and the presence of phlogopite, may represent the mantle source of kimberlite melts. The δ41K value of the kimberlite source can be estimated from the average compositions of these samples (δ41K = −0.414 ‰) (Fig. 4), which is indistinguishable from the pristine mantle (δ41K = −0.42 ± 0.17 ‰) and BSE (δ41K = −0.42 ± 0.07 ‰) (Hu et al., Reference Hu, Teng, Helz and Chauvel2021a).
Figure 4. Correlation diagrams of (a) δ41K vs. (a) K2O and (b) Rb for Mengyin pristine kimberlite samples (this study) and mantle peridotites (Ionov and Wang, Reference Ionov and Wang2021). Five highly-metasomatized samples (with high K and Rb contents) are plotted along the varying trends of the Mengyin kimberlites, and their average values (K2O = 0.088 wt.%; Rb = 5.56 ppm; δ41K = − 0.414 ‰) are considered as source composition of the kimberlite melt.
The BSE-like K isotope composition of kimberlite source does not align with the previously proposed mixed source of a carbonated asthenosphere, lithospheric keel and a subduction-dehydrated oceanic slab located at the diamond-stable mantle level or deeper, as inferred from diamond-bearing mineralogy and Sr-Nd-Hf-C-O isotope compositions (Lu et al., Reference Lu, Wang, Chen and Zheng1998; Liu et al., Reference Liu, Zhang, Sun and Ye2004; Zhang and Yang, Reference Zhang and Yang2007; Zhang et al., Reference Zhang, Zhou, Sun and Zhou2010). This discrepancy may be related to varying metasomatic media at different depths in the mantle. In subduction zones, metasomatism in the mantle wedge mainly occurs shallower than 75 km in the spinel-facies environment. Hydrous melts/fluids released from subducting slabs in these conditions tend to be enriched in heavy K isotopes (e.g. Liu et al., Reference Liu, Wang, Sun, Xiao, Xue and Tuller-Ross2020, Reference Liu, Xue, Geldmacher, Hoernle, Wiechert, An, Gu, Sun, Tian, Li and Wang2024), resulting in higher δ41K values in both subduction zone peridotites (Wang and Ionov, Reference Wang and Ionov2023) and arc lavas (e.g. Parendo et al., Reference Parendo, Jacobsen, Kimura and Taylor2022; Pan et al., Reference Pan, Xiao, Su, Robinson, Li, Wang and Liu2024) (Fig. 5). As dehydration progresses, the residual slabs represented by eclogites become isotopically lighter (δ41K = −1.64 to −0.24 ‰; Liu et al., Reference Liu, Wang, Sun, Xiao, Xue and Tuller-Ross2020), which could contribute to lighter isotope compositions in the metasomatized mantle. This inference is consistent with the low δ41K values observed in non-subduction zone mantle peridotites, which range from −2.15 to −0.27 ‰ (Ionov and Wang, Reference Ionov and Wang2021). Notably, these mantle peridotites are derived from the lithospheric mantle at depths from 60 to <150 km (Ionov and Wang, Reference Ionov and Wang2021). The K isotope features at such depths are likely inherited by intra-continental alkali basaltic lavas (−1.330 to 0.126 ‰, Sun et al., Reference Sun, Teng, Hu, Chen and Pang2020; Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025; Fig. 5), where heavy K isotope compositions arise from the fractionation of isotopically light K-rich minerals during magma differentiation (Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025).
Figure 5. K isotope compositions of kimberlites in this study and comparisons with data of various rocks in literature (Tuller-Ross et al., Reference Tuller-Ross, Marty, Chen, Kelley, Lee and Wang2019a, b; Hu et al., Reference Hu, Teng, Plank and Chauvel2020, Reference Hu, Teng, Helz and Chauvel2021a, b; Huang et al., Reference Huang, Teng, Rudnick, Chen, Hu, Liu and Wu2020; Liu et al., Reference Liu, Wang, Sun, Xiao, Xue and Tuller-Ross2020, Reference Liu, Xue, Wang, Sun and Wang2021; Santiago Ramos et al., Reference Santiago Ramos, Coogan, Murphy and Higgins2020; Sun et al., Reference Sun, Teng, Hu, Chen and Pang2020; Ionov and Wang, Reference Ionov and Wang2021; Wang and Ionov, Reference Wang and Ionov2023; Parendo et al., Reference Parendo, Jacobsen, Kimura and Taylor2022; Pan et al., Reference Pan, Xiao, Su, Robinson, Li, Wang and Liu2024; Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025).
In comparison to MORBs, intra-continental basaltic lavas and arc lavas, both oceanic island basalts and kimberlites originate from the deep asthenospheric mantle (> 150 km), with the involvement of recycled materials, and exhibit relatively narrow δ41K ranges (Fig. 5). The uniformity of K isotopes in these two rock types, originating from similar mantle depths but beneath different settings (ocean vs. craton), suggests a homogeneous K isotope composition within the deep mantle. This may result from efficient mixing and homogenization processes, such as mantle convection. On one hand, K may readily achieve homogeneity due to its high mobility relative to other elements. On the other hand, K in subducted slabs might become exhausted after a long subduction journey, leading to dilution of light K isotope features, which occurs more rapidly than in the metasomatized shallow mantle.
5.d. Implication for the destruction of North China Craton
Geophysical and geochemical studies have extensively documented that the lithospheric thickness in the eastern North China Craton has decreased from ∼200 km during the Palaeozoic to the present 60–80 km (Menzies et al., Reference Menzies, Fan and Zhang1993; Zhu et al., Reference Zhu, Chen, Wu and Liu2011). This lithospheric destruction has been attributed to the circular subduction of the Palaeo-Tethyan, Palaeo-Asian and Pacific oceans since the Mesozoic (Zhu, Reference Zhu2023), with a peak destruction age identified at ∼125 Ma (Zhu et al., Reference Zhu, Chen, Wu and Liu2011). These events are well-constrained by compositional variations observed in igneous rocks within the North China Craton (Zhang et al., Reference Zhang, Zhu, Santosh, Ying, Su and Hu2013). Non-traditional stable isotopes, such as those of Li, Fe, Mg and Ca, have emerged as powerful tools for understanding lithospheric processes in this region (e.g. Tang et al., Reference Tang, Zhang, Deloule, Su, Ying, Xiao and Hu2012; Su et al., Reference Su, Hu, Teng, Xiao, Zhou, Sun, Zhou and Chang2017). These well-documented processes provide an excellent opportunity to evaluate the application of K isotopes in tracing secular geological events.
Recent studies have reported K isotope data for Mesozoic to Cenozoic basaltic lavas in the North China Craton (Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025). These data, along with findings from this study, are illustrated in Fig. 6. There is a notable increase in δ41K values from Palaeozoic kimberlites to Mesozoic basalts (up to 0.126 ‰). The elevated δ41K values in the Mesozoic basalts overlap with the ranges observed in igneous rocks formed during initial subduction and typical arcs (Parendo et al., Reference Parendo, Jacobsen, Kimura and Taylor2022; Pan et al., Reference Pan, Xiao, Su, Robinson, Li, Wang and Liu2024; Rodney et al., Reference Rodney, Tacail, Lewis, Andersen and Elliott2024; Fig. 5), indicating a significant addition of isotopically heavy melts/fluids from the subduction zone to the magma sources. Since 125 Ma, the δ41K values have decreased with age, down to −0.926 ‰ in Cenozoic basalts (Fig. 6). This decline is likely due to the incorporation of melts from dehydrated (eclogitized) slabs, as these Cenozoic basalts were generated in a big mantle wedge setting (Su et al., Reference Su, Hu, Teng, Xiao, Zhou, Sun, Zhou and Chang2017; Zhu, Reference Zhu2023). This temporal variation in δ41K values of mantle-derived magmas correlates with the timing of lithospheric destruction in the North China Craton. Consequently, K isotope systematics can be utilized to elucidate the influence of subduction, mantle dynamics and the complexities of the region’s geological history.
Figure 6. Variation of K isotope compositions of kimberlites (this study) and basaltic lavas (Su et al., Reference Su, Bai, Tang, Xiao, Li, Zhao and Moynier2025) in the North China Craton through geological time.
6. Conclusions
-
(1) Based on the findings of this study, the K isotope compositions of diamondiferous kimberlites from Mengyin in the North China Craton reveal significant insights into the processes of alteration and magma differentiation. The observed variations in δ41K values indicate that post-emplacement alterations can lead to substantial K isotope fractionation, while pristine kimberlite samples exhibit restricted K isotope compositions primarily driven by the crystallization of K-rich minerals, particularly phlogopite.
-
(2) Combining our new measurements with literature data, we estimate a δ41K of −0.458 ‰ for the primary kimberlite melt and of −0.414 ‰ for the kimberlite source. It appears that the deep mantle (>150 km) exhibits a more homogenous K isotope composition than the shallow mantle, likely due to the efficiency of convection flow at different depths and K behaviour during subduction processes.
-
(3) The temporal variation of K isotope compositions in mantle-derived magmas from the Palaeozoic to the Cenozoic in the North China Craton is linked to the timing and mechanism of lithospheric destruction. These characteristics enhance the application of K isotopes in crustal recycling and geochemical evolution of the mantle.
Data availability statement
The data are available via Figshare https://doi:10.6084/m9.figshare.27308157.
Acknowledgements
We gratefully acknowledge the constructive reviews provided by editor Prof. Sarah Sherlock, Hamed Gamaleldien and one anonymous reviewer, which have significantly improved the quality of this paper. This work was supported by National Natural Science Foundation of China (42350001).
Competing interests
We declare that no conflict of interest exists.
Open access
