Multistage Mantle Metasomatism during the Generation of Kimberlite Melts: Evidence from Mantle Xenoliths and Megacrysts of the Grib Kimberlite, Arkhangelsk, Russia

Abstract

Major and trace element compositions of garnet, clinopyroxene, orthopyroxene, phlogopite, and ilmenite from garnet peridotite, ilmenite-bearing peridotite (dunites), and clinopyroxene–phlogopite xenoliths, as well as megacrysts of these minerals from the Grib kimberlite, Arkhangelsk diamond province, Russia, have been analyzed. These data are used to propose a model for mantle metasomatism of lithospheric mantle by kimberlite melts, including their generation and evolution, geochemical enrichment of depleted lithosphere mantle, and formation of megacrystic assemblage. The lithospheric mantle beneath the Arkhangelsk diamond province, from its base (depth ~180–210 km) to a depth of ~100–120 km (corresponding to a pressure of 3.5 GPa) experienced extensive metasomatism along the main kimberlite melt channel. Petrography of the peridotite xenoliths indicates a progressive refertilization of depleted harzburgite into garnet lherzolite, phlogopite–garnet wehrlite, and clinopyroxene–phlogopite rocks. Metasomatic refertilization occurred shortly before the capture of these xenoliths by the kimberlite melt. The model melt compositions calculated from garnet–clinopyroxene equilibria in different types of xenoliths and megacrysts show that alkaline–carbonate–ultramafic kimberlite melt acted as a metasomatic agent in the sheared peridotite at the base of the lithospheric mantle. High-Ti garnet and high-Cr clinopyroxene megacrysts in the middle part of the lithospheric mantle, as well as the main volume of garnet lherzolite xenoliths were formed in geochemical equilibrium with the kimberlite melts, which demonstrate an increase of silicate components and fractionation of Fe–Ti phases. The modification could be related to the interaction of ascending carbonate-rich protokimberlite melts with surrounding lithospheric mantle. The similarities in the compositions of garnet, clinopyroxene, phlogopite, and ilmenite megacrysts with minerals of peridotite xenoliths in the Grib kimberlite suggest that these megacrysts are disintegrated fragments of coarsest grained metasomatized garnet lherzolite, ilmenite-bearing peridotite, and clinopyroxene-phlogopite mantle rocks or formed under the same conditions as xenoliths or directly crystallized from metasomatic melts.

INTRODUCTION

Kimberlites are scarce alkaline–ultrabasic rocks, the main natural sources of diamond. In addition to economic significance of diamond deposits, kimberlite rocks provide insight into the composition and structure of lithospheric mantle, since bring up numerous mantle xenoliths and xenocrysts. In spite of the over 50-year study of kimberlites and related rocks worldwide, many petrological, geochemical, mineralogical, and volcanological aspects of these rocks remain controversial:

(1) the composition of kimberlite melt (initially mainly carbonatite or ultrabasic silicate) and its evolution en route through the lithospheric mantle, including dissolution of orthopyroxene and assimilation of mantle peridotites (e.g., Price et al., 2000; Kopylova et al., 2007; Russell et al., 2012; Kamenetsky et al., 2014; Kamenetsky and Yaxley, 2015; Kamenetsky, 2016; Stamm and Schmidt, 2017; Soltys et al., 2018; Giuliani et al., 2020);

(2) role of asthenospheric and lithospheric sources in the generation of kimberlite magmas (Mitchell, 1995; Burgess and Harte, 2004; Tappe et al., 2011; Kostrovitsky et al., 2013; Solovieva et al., 2019), role of mantle plumes (Howarth et al., 2014; Sun et al., 2014; Pokhilenko et al., 2015; Tappe et al., 2016), as well as a contribution of vein metasomatic assemblages MARID (mica–amphibole–rutile–ilmenite–diopside) or PIC (phlogopite–ilmenite–clinopyroxenite) or their analogues (Foley, 1992; Mitchell, 1995; Gregoire et al., 2002; Fitzpayne et al., 2018a, 2018b; Lebedeva et al., 2020a), in kimberlite genesis;

3) relationship of lithospheric mantle metasomatism recorded in mantle xenoliths with kimberlite melt, including stages and timing of metasomatism, as well as the composition of metasomatic agent (e.g., Griffin et al., 1999; Burgess and Harte, 2004; Nimis et al., 2009; Ionov et al., 2010, 2018; Howarth et al., 2014; Giuliani et al., 2014, 2016; Pokhilenko et al., 2015; Sazonova et al., 2015; Shchukina et al., 2015; Kargin et al., 2017b; Nosova et al., 2017; Bussweiler et al., 2018; Solovieva et al., 2019);

(4) origin of megacrysts (individual grains of garnet, clinopyroxene, phlogopite, ilmenite, olivine, and other minerals, over 10 mm in size, and their fragments (Mitchell, 1995)), causes of their compositional variations, as well as genetic relationship with mantle metasomatism, sheared peridotite xenoliths, and kimberlite melts (e.g., Moore and Lock, 2001; Burgess and Harte, 2004; Kostrovitsky et al., 2004; Moore and Belousova, 2005; Kopylova et al., 2009; Kostrovitsky et al., 2013; Kargin et al., 2017a; Solovieva et al., 2019; Lebedeva et al., 2020b).

Mantle xenoliths brought up by kimberlites reflect a complex multiple history of metasomatic reworking of lithospheric mantle. Mantle reworking first of all causes a geochemical enrichment of depleted garnet harzburgites and formation of garnet lherzolites, i.e., refertilization of a depleted mantle protolith, as well as the introduction of new minerals atypical of mantle rocks such as apatite, phlogopite, ilmenite, and others (review in O’Reilly and Griffin, 2013). Metasomatic agent may be a melt/fluid: (1) compositionally close to asthenosphere melts of basic or ultrabasic composition (e.g., Burgess and Harte, 2004; Solovieva et al., 2008; Shchukina et al., 2015); (2) of carbonatite and alkaline-carbonatite compositions (Yaxley et al., 1998; Griffin et al., 1999; Pokhilenko et al., 2015; Ionov et al., 2018), as well as (3) of kimberlitic composition (Boyd et al., 1997; Giuliani et al., 2013; Bussweiler et al., 2018; Fitzpayne et al., 2018b).

Mantle xenoliths bear evidences for several stages of mantle metasomatism, which were driven by diverse metasomatic agents at different stages of tectonothermal transformations of lithospheric mantle (e.g., Pokhilenko et al., 2015; Shchukina et al., 2015; Ionov et al., 2018). The mantle metasomatism was likely accompanied by a gradual change of metasomatic agent or fluid during its interaction with lithospheric mantle, as well as fractionation crystallization (Burgess and Harte, 2004; Ionov et al., 2006; O’Reilly and Griffin, 2013; Kamenetsky and Yaxley, 2015; Giuliani et al., 2016, 2020). The generation and subsequent evolution of kimberlite melt during ascent through lithospheric mantle lead to its transformation along the main transport channel (Bussweiler et al., 2018), because alkaline–ultrabasic melts, including carbonatites and kimberlites, are geochemically disequilibrium with lithospheric mantle (e.g., Kopylova et al., 2007; Yaxley et al., 2017). The timing of transformation of lithospheric mantle and its interaction with kimberlite melt can be determined from isotope studies of kimberlites, whole-rock samples of mantle xenoliths, and individual minerals in them (e.g., Gregoire et al., 2002; Ionov et al., 2015, 2020; Fitzpayne et al., 2018a, 2020, and references therein). For instance, the recent studies of Rb–Sr and Sm–Nd isotope systems in minerals from mantle xenoliths of the Grib pipe, Arkhangelsk province, in spite of intense mantle metasomatism, indicate a heterogeneity of these isotope systems in xenoliths due to the different extent of rock re-equilibration with kimberlite melt, as well as provide insight into pre-kimberlite transformations of lithospheric mantle beneath this province (Lebedeva et al., 2020a).

This work generalizes data on mantle xenoliths from the Grib kimberlite, Arkhangelsk diamond province (ADP), Russia and proposes a model of the ADP lithospheric mantle metasomatism, which includes the formation and evolution of kimberlite melts, enrichment of depleted lithospheric mantle, and generation of minerals of megacryst assemblage.

GEOLOGICAL BACKGROUND

The ADP is located in the northeastern part of the East European Platform (Fig. 1) and comprises over 80 bodies (explosion pipes and sills) of kimberlites and related alkaline–ultrabasic rocks (Kononova et al., 2007; Tretyachenko, 2008), which are usually subdivided into magnesian–aluminous (alkaline–ultrabasic rocks of the Zolotitsa and Verkhotina fields, picrites and olivine melilitites of the Chidviya, Izhma, Nenoksa, and Suksoma clusters) and Fe–Ti (Chernoozero kimberlites (Girb pipe), alkaline–ultrabasic rocks and kimberlites of the Kepino cluster, alkaline picrites of the Megorsk cluster, and carbonatites–kimberlites of the Mela cluster) (Sablukov et al., 2000; Mahotkin et al., 2000; Kononova et al., 2007; Sablukov and Sablukova, 2008). The rocks of the Mg–Al series (<1.1 wt % TiO₂) are restricted mainly to the western and marginal parts of ADP, whereas Fe–Ti series rocks (>1.2 wt % TiO₂) are localized in its eastern and central parts (Fig. 1). Kimberlites of the Grib pipe and Zolotitsa cluster are economically diamondiferous and form, respectively, the Grib and Lomonosov deposits.

Fig. 1.

Schematic map showing the distribution of Devonian magmatic rocks in the northeastern East European Platform with indication of the alkaline-ultrabasic magmatic occurrences according to (Tretyachenko, 2008; Arzamastsev and Wu, 2014). Ages of the provinces are shown according to (Arzamastsev and Wu, 2014; Larionova et al., 2016). (1–4): Precambrian crust (Bogdanova et al., 2016): (1) Archean crust reworked within the Lapland–Kola collisional orogeny; (2) Paleoproterozoic volcanic belts and sedimentary basins (2.50–1.95 Ma); (3) Paleoproterozoic crust (1.83–1.82 Ma); (4) aulacogens and intracratonic basins (1.50–0.70 Ma).

It remains unknown whether the ADP magmatic rocks were formed in one or several stages. The age of high-grade kimberlites is 375 ± 2 Ma, whereas the ages of the rocks of the Kepino cluster and carbonatites of the Mela cluster are estimated as 397 ± 1.2 and 393 ± 8 Ma, respectively (Larionova et al., 2016). The geological position, petrography, and geochemical characteristics of the ADP magmatism, including the Grib kimberlites, was characterized in detail in (Sablukov et al., 2000; Verichev et al., 1999, 2003; Mahotkin et al., 2000; Kononova et al., 2007; Tretyachenko, 2008; Sablukov et al., 2009; Larionova et al., 2016; Kargin et al., 2020).

The Grib pipe is located in the central part of ADP (Fig. 1). Geological position, structure, and petrographic-geochemical characteristics are reported in (Verichev et al., 1999, 2003; Golubeva et al., 2006; Kononova et al., 2007; Larionova et al., 2016).

Kimberlites cut across weakly lithified Neoproterozoic (Ediacaran) sedimentary rocks and are overlain by a sequence of the Middle Carboniferous weakly cemented fine-grained quartz sandstones and Quaternary loose sediments 65–70 m thick. The diatreme has NE-trending rhomobohedral rounded shape 570 × 480 m in size and is subdivided into crater and vent facies. The crater facies 110 m thick is filled with sandstones, tuffstones, and breccias of sedimentary rocks, kimberlite tuffs, and tuffites. The vent facies is mainly occupied by kimberlites of two phases and traced by holes to a depth of 950 m. The rocks of the first phase are represented by kimberlite tuff breccias and xenotuff breccias, compose the near-contact relict zones, and occupy approximately 25–30 vol % of the pipe. The rocks of the second phase compose the most part of the kimberlite diatreme and are pyroclastic kimberlites with wide variations of pyroclasts and litoclasts of olivine at low content of fragments of host rocks. The kimberlites of this phase have the elevated contents of mantle and crustal xenoliths and megacrysts. The contacts of the pipe are distinct sharp, with brecciation of host rocks. The thickness of the outer contact rock reaches 10 m (Verichev et al., 1999, 2003).

METHODS

The detailed studies of microtextural features of the studied rocks, as well as minerals were carried out on a JEOL JSM 6480LV scanning electron microscope (SEM) equipped with an INCA Energy 350 energy dispersive analyzer at an accelerating voltage of 15 kV, beam current 15 ± 0.1 nA, and beam diameter of 4 μm at the Laboratory of Local Analytical Methods, Petrology and Volcanology Department, Geological Faculty, Moscow State University. The back-scattered electron images were obtained. The analytical errors of all analyzed elements were no more than ±10 rel % for elements with concentrations from 1 to 5 wt %, up to ±5 rel % for elements with concentrations from 5 to 10 wt %, and up to ±2 rel % for contents more than 10 wt %. The detection limit varies from 0.1 to 0.3 wt % depending on element.

The major-element contents in the studied minerals were determined on an JEOL JXA-8200 electron microprobe at the Laboratory for Analysis of Mineral Materials of the IGEM RAS. The operating conditions were 20 kV accelerating voltage, 20 nA beam current, and 1–2 μm probe diameter. The counting time for major and trace elements was 10 and 20–40 s, respectively. The measurements were corrected with the GEOL ZAF correction routine. The instrument was calibrated using standard compounds close in composition to studied phases.

The trace-element contents were measured in the minerals previously analyzed by electron microprobe using secondary-ion mass spectrometry (SIMS) at the Institute of Microelectronics of the Russian Academy of Sciences (Yaroslavl) following technique described in (Nosova et al., 2002).

Detailed description of analytical methods is given in (Kargin et al., 2016, 2017a, 2017b).

STUDY OBJECTS

The Grib kimberlite contains numerous mantle xenoliths of peridotites, eclogites, and clinopyroxene–phlogopite rocks, as well as megacrysts of garnet, olivine, clinopyroxene, and their intergrowths (Kostrovitsky et al., 2004; Sablukova et al., 2004, 2009; Golubkova et al., 2013; Sazonova et al., 2015; Shchukina et al., 2015; Kargin et al., 2016, 2017a, 2017b, 2019, 2020; Nosova et al., 2017; Lebedeva et al., 2020a, 2020b).

The main volume of mantle xenoliths and megacrysts (Fig. 2) was studied in the core of borehole 1/1000, which recovered mainly pyroclastic kimberlites of the second phase to a depth of 920 m. In general, the pyroclastic kimberlites of the studied section are characterized by the elevated content of olivine lithoclasts of different size and shape (up to 45–50 vol %), which are represented by rounded macrocrysts and megacrysts, as well as by olivine fragments of different shape and size (see Supplementary, ESM_1.pdf ). Most part of olivine, as mega- and macrocrysts of other minerals and xenoliths represent the core of pyroclasts, which can be subdivided into two groups: (1) large rounded or anhedral pyroclasts up to 5–6 cm in size representing crystallized kimberlite melt and (2) thin kimberlite rims around cores (Golubev et al., 2006; Sazonova et al., 2015). The pyroclastic material has microporphyritic or aphyric texture and consists of olivine phenocrysts, less common phlogopite flakes (up to 50 μm), and grains of ore minerals embedded in a serpentine or carbonate–serpentine groundmass. The kimberlite matrix is fine-grained and consists of serpentine, more rare carbonate, and the great amount of ore high-Ti minerals: picroilmenite, rutile, as well as perovskite frequently replaced by titanite. The aggregate of secondary minerals in the groundmass consists of carbonate, serpentine, and high-Mg chlorite (Sazonova et al., 2015).

Fig. 2.

Photos of the studied peridotite xenoliths and clinopyroxene-phlogopite xenoliths (a) and garnet, clinopyroxene, ilmenite, and phlogopite megacrysts (b) in kimberlite groundmass, and intergrowths of ilmenite and garnet (c). (Kmb) kimberlite groundmass, (Xen) xenolith, (Grt) garnet, (Cpx) clinopyroxene, (Ilm) ilmenite, (Phl) phlogopite, (Srp) serpentine.

In general, the xenoliths of mantle and crustal rocks, megacrysts, and their fragments were found throughout the entire section of pyroclastic kimberlites, but the depth interval from 550–600 to 750–800 m is characterized by the elevated contents of relatively well-preserved mantle xenoliths and megacrysts (see Supplementary, ESM_1.pdf)¹, while interval of 400–450 m has the high content of olivine lithoclasts (up to 50 vol %).

Mantle Xenoliths

Detailed petrographic characteristics of the studied xenoliths is reported in (Sazonova et al., 2015; Kargin et al., 2016, 2017a, 2017b, 2020; Nosova et al., 2017). Most part of the studied xenoliths have a rounded shape and size from 2–3 to 10–15 cm. Some xenoliths represent the cores of pyroclasts. Garnet peridotite xenoliths show wide variations in proportions of rock-forming minerals, varying in composition from garnet harzburgites to garnet, more rarely, ilmenite–garnet lherzolites. Xenoliths are made up of olivine, orthopyroxene, clinopyroxene, garnet, phlogopite, and ilmenite (in order of decreasing content) and have inequigranular-medium, from medium to coarse-grained texture with the predominance of varieties with grain size from 2 to 7 mm (Figs. 3a–3d). Olivine is usually represented by rounded highly-fractured grains, orthopyroxene forms scarce equant euhedral grains (Figs. 3a–3c). In most cases, olivine and orthopyroxene are replaced by secondary serpentine-group minerals, more rarely by chlorite and carbonates, which complicate the determination of these minerals and their proportions in altered xenoliths. Clinopyroxene usually occurs as elongated grains located in interstices between serpentinized olivine and orthopyroxene. It frequently replaces orthopyroxene and more rarely olivine in fractures and along rims (Figs. 3a–3d), which indicates that clinopyroxene was formed shortly before the xenolith entrainment into kimberlite melt. With increasing the clinopyroxene content in garnet lherzolites, it forms xenomorphic grains or large equant grains, which could occur in intergrowths with later phlogopite (Figs. 3c–3d). Garnet in mantle peridotite xenoliths usually occurs as rounded large grains in equilibrium intergrowths with olivine and orthopyroxene. Phlogopite was found as individual grains, and as clusters of fine grains around garnet crystals, replacing the latter (Fig. 3e).

Fig. 3.

BSE photomicrographs of mantle peridotite xenoliths. (а) equant grains of olivine (Ol) and orthopyroxene (Opx), as well as elongated clinopyroxene grains between olivine and orthopyroxene, sample Gr1-754; (b) extended grains of clinopyroxene (Cpx) developed between olivine grains, as well as along fractures in them, sample Gr-3; (c, d) gradual increase of clinopyroxene and garnet (Grt) in peridotite xenoliths, where clinopyroxene forms both large grains and grains extended along boundaries of olivine crystals, samples Gr1-754 and Gr1-745; (e) replacement of garnet by phlogopite (Phl) and spinel (Spl); (Serp) pseudomorphs of serpentine after olivine, sample Gr1-634; (f) zoned garnet, sample Gr1-466.

Sheared garnet lherzolite xenolith (Fig. 2) shows intense ductile deformations, which led to the formation of porphyroclastic texture, where large porphyroclasts of olivine, orthopyroxene, and garnet are embedded in a microgranular mosaic mass consisting of olivine and orthopyroxene neoblasts (Figs. 4a, 4b). Clinopyroxene develops after neoblasts and porphyroclasts of orthopyroxene (Fig. 4a), more rarely olivine (Kargin et al., 2017b), which indicates its formation after the main deformation stage and shortly before the xenolith entrapment by kimberlite melts, as indicated by the preservation of its disequilibrium textures. Garnet porphyroclasts show sharply expressed zoning in BSE image (Fig. 4b). Equant ilmenite up to 0.6 mm in size occurs as scarce grains.

Fig. 4.

Back-scattered electron (BSE) images of sheared lherzolite (a, b), ilmenite peridotites (c, d), and clinopyroxene–phlogopite (e, f) xenoliths. (a) replacement of orthopyroxene porphyroclasts (Opx), more rarely olivine (Ol), by clinopyroxene (Cpx), sample Gr106-644; (b) zoned garnet porphyroclasts (Grt) with sharp boundaries between zones, sample Gr106-644; (c, d) fragment of ilmenite veinlet in peridotite xenolith, where ilmenite (Ilm) is represented by polygonal crystals, with small apophyses branching from the main veinlets and cementing serpentinized olivine grains, samples Gr1-536 and Gr1-639; (e) granoblastic texture of clinopyroxene–phlogopite xenolith, as well as development of late veinlets filled with calcite (Cb), phlogopite (Phl), spinel and serpentine (Cb + Srp + Spl), sample Gr1-715; (f) rounded inclusions of Mg-ilmenite in clinopyroxene, sample Gr1-715.

Ilmenite-rich peridotite xenoliths usually consist of ilmenite widely varying up to 60–70 vol %, serpentinized olivine and orthopyroxene (?) (Kargin et al., 2020). The studied xenoliths usually lack garnet, in spite of the wide abundance of ilmenite-garnet intergrowths among xenoliths in the Grib kimberlite. The complete replacement of silicate minerals by serpentine prevents their reliable assignment to peridotites or dunites. Ilmenite has an anhedral morphology and frequently forms small veinlets between silicate minerals (Fig. 4c). In the latter case, ilmenite is represented by polygonal grains 100–500 μm in size (Figs. 4c–4d). Ilmenite veinlets frequently contain rounded inclusions of serpentinized olivine (Fig. 4d). Sometimes, small apophyses of the ilmenite veinlets brecciate olivine grains. In general, the studied ilmenite xenoliths are texturally similar to the polymictic breccia xenoliths, which are widespread among the Kimberley kimberlites, South Africa (Giuliani et al., 2013, 2014), as well as in the Udachnaya kimberlites, Yakutia (Pokhilenko, 2009).

Clinopyroxene–phlogopite xenoliths (Kargin et al., 2017a) have inequigranular medium to coarse grained granoblastic texture (Fig. 2). They consist mainly of clinopyroxene and phlogopite, while subordinate olivine, Cr-spinel, and ilmenite are obsered as inclusions in clinopyroxene (Figs. 4e, 4f). Some xenoliths have higher content of olivine and large equant garnet, and, correspondingly, can be classified as phlogopite wehrlites or phlogopite–garnet wehrlites.

Most part of the studied mantle xenoliths is cut by numerous systems of thin fractures (up to 0.2–0.3 mm), which are filled with carbonates, serpentine, phlogopite, and spinel (Fig. 4e). The xenolith minerals around veinlets are transformed. In particular, clinopyroxene acquires spongy texture, while garnet and phlogopite, and more rarely orthopyroxene, are overgrown by thin marginal zones, which are clearly distinguishable in the BSE images, thus indicating their compositional changes. At the contact with veinlets, garnet is frequently replaced by aggregate of phlogopite, carbonates, and Cr-spinel. In the least altered samples, the neoblastic clinopyroxene and orthopyroxene are also observed at the contact with garnet, which can be caused by the interaction of garnet megacryst with a kimberlite melt (Bussweiler et al., 2016). The veinlets are likely related to kimberlite melt bringing xenoliths, while the marginal zones in minerals have reaction origin.

Megacrysts

Garnet, clinopyroxene, phlogopite, and ilmenite megacrysts usually have rounded, extended shape and sizes up to 2–3 mm (Fig. 2). The Grib kimberlite contains numerous disintegrated fragments of megacrysts less than 0.5 mm in size, as well as their intergrowths, among which we studied ilmenite–garnet and orthopyroxene–ilmenite–garnet intergrowths. The garnet megacrysts frequently are intensely fragmented and contain mono- and polymineral inclusions of different composition (Lebedeva et al., 2020b). The large ilmenite megacrysts could contain small (<0.5 mm) rounded inclusions of serpentine represented by fragments of mantle peridotites (Kargin et al., 2020). Like the studied mantle xenoliths, the megacrysts are cut by numerous secondary veinlets filled with carbonates, serpentine, and spinel, whose origin by analogy with those in mantle xenoliths is related to the influence of the host kimberlite melt (Kargin et al., 2017b).

MINERAL COMPOSITION

The detailed description of rock-forming minerals of mantle xenoliths and megacrysts is presented in (Kargin et al., 2016, 2017a, 2017b, 2019, 2020). This work generalizes data on composition of the studied minerals. Compositions of minerals used in this work are given in Supplementary, ESM_2.xlsx.

Major Elements

Most part of the studied minerals are characterized by the wide variations of Mg number Mg# (Mg/(Mg + Fe²⁺)). Therewith, harzburgitic minerals have the higher Mg# and low TiO₂, whereas minerals from sheared lherzolite have low Mg# and Cr₂O₃ at high TiO₂. Minerals from garnet lherzolite xenoliths and megacrystic assemblage usually have intermediate Mg# and TiO₂.

Orthopyroxene. Composition of orthopyroxene is well illustrated by a trend in Fig. 5: from garnet harzburgites to sheared lherzolites, its Mg# varies from 0.94 to 0.90, while TiO₂ content varies from 0.04 ± 0.03 to 0.17 ± 0.03 wt % (n = 211). In addition, orthopyroxene shows the wide variations of Cr₂O₃ and Al₂O₃ (Figs. 5b, 5c). Orthopyroxene in association with garnet and ilmenite has compositions close to orthopyroxene neoblasts from sheared lherzolite xenoliths (Figs. 5a–5c). The marginal zones of orthopyroxene, which were re-equilibrated with host kimberlite melt, have the lowest Mg#, as well as high TiO₂ and Cr₂O₃ contents at low Al₂O₃ (Figs. 5a–5c). The least magnesian orthopyroxenes approach the composition of matrix orthopyroxene from polymictic breccia xenoliths (Fig. 5), which represent kimberlite melt crystallized at mantle depths (Pokhilenko, 2009; Giuliani et al., 2014).

Fig. 5.

Compositions of orthopyroxene from mantle xenoliths of the Grib pipe. Arrows show variation trend of orthopyroxene during mantle metasomatism (see text for explanation). Fields show the compositions of orthopyroxene from: polymictic breccia xenoliths (field PB) from the Kimberley province, South Africa (Giuliani et al., 2014), sheared peridotite (field SP) and garnet peridotite (field P) xenoliths from the Udachnaya pipe (Ionov et al., 2010).

Clinopyroxene. Like orthopyroxene, clinopyroxene shows Mg# decrease from 0.93 to 0.89 and TiO₂ increase from 0.20 ± 0.03 to 0.35 ± 0.03 wt % (n = 65) from garnet harzburgite to sheared lherzolite (Fig. 6). Thereby, clinopyroxenes from sheared peridotite are close in composition to clinopyroxene from the PIC-type mantle xenoliths, which could represent lithospheric mantle reworked by kimberlite melts (Fitzpayne et al., 2018b), and are overlapped with the highest Mg clinopyroxene from sheared peridotite xenolith in the Udachnaya Pipe, Yakutia (Ionov et al., 2010).

Fig. 6.

Composition of clinopyroxene from the studied mantle xenoliths and megacrysts of the Grib pipe. Arrows show variation trends of clinopyroxene during mantle metasomatism (see text for explanation); dashed arrow in Fig. (a) shows variations of clinopyroxene from late spongy zones (Kargin et al., 2017a); pink arrow (upper) in Figure (b) shows a change of clinopyroxene composition developed along fractures in orthopyroxene and olivine in garnet harzburgite xenoliths. Fields show clinopyroxenes from: sheared peridotite xenoliths (field SP) from the Udachnaya pipe, Yakutia (Ionov et al., 2010); phlogopite–ilmenite–clinopyroxene xenoliths (field PIC), and phlogopite–amphibole–rutile–ilmenite–clinopyroxene xenoliths (field MARID) from South Africa kimberlites (Fitzpayne et al., 2018b); garnet peridotite xenoliths from the Grib kimberlite, which experienced silicate (field P1) and carbonatite (field P2) metasomatism according to (Shchukina et al., 2015).

Large clinopyroxene megacrysts differ from typical megacrysts in the higher Ca# and Mg# and lower TiO₂ (Kargin et al., 2017a) and are correlated with high-Cr megacrysts (Pivin et al., 2009; Bussweiler et al., 2016). In terms of composition, they are completely overlapped with clinopyroxene from clinopyroxene–phlogopite xenoliths from the Grib pipe and have moderate Mg# (0.91–0.93) and lowered TiO₂ (<0.25 wt %), and are close to clinopyroxenes from garnet peridotite xenoliths from some kimberlite provinces (Nimis et al., 2009; Pivin et al., 2009). Clinopyroxenes crystallizing along orthopyroxene and olivine boundaries, as well as along fractures in them show a wide scatter of Cr₂O₃ content at constant Mg#, whereas clinopyroxene forming large crystals petrographically equilibrated with other peridotite minerals have less variable Cr₂O₃ (Fig. 6c). Clinopyroxene from intergrowths with garnet megacrysts and inclusions in them has intermediate composition between clinopyroxene from sheared peridotite xenolith and large megacrysts from the Grib pipe and is overlapped in composition with clinopyroxene from sheared peridotite xenoliths from kimberlite worldwide (Fig. 6).

Clinopyroxene with porous (spongy) microtextures differs from unaltered domains in a sharp increase of TiO₂ content at decreasing Al₂O₃, Na₂O, Cr₂O₃ (Fig. 6), which is typical of clinopyroxene megacrysts with similar structures (Bussweiler et al., 2016). Late rims of zoned crystals, which were re-equilibrated with host kimberlite melt have the low Mg# and high TiO₂, which make them similar to clinopyroxene from sheared lherzolite, except for higher Cr₂O₃ (Fig. 6b). In garnet lherzolite xenolith, clinopyroxene shows an increase of Na₂O and Cr₂O₃ along grain boundaries and fractures (Fig. 6b, 6d). Such variations likely reflect the interaction of xenolith with late Na-rich kimberlite melt.

Garnet. Unlike pyroxene, garnet shows no a gradual increase of TiO₂ with decreasing Mg# (Fig. 7). It is characterized by wide compositional variations at pyrope content varying from 63 to 78 mol %. In the CaO–Cr₂O₃ diagram (Fig. 7a), most part of the studied garnets fall in the field of lherzolitic garnets from the Yakutian kimberlites (Sobolev, 1977), in spite of the fact that some studied garnets are from garnet harzburgite xenoliths. Some garnets from harzburgite and wehrlite xenoliths have the elevated CaO contents, which displaces their compositions in the field of wehrlite xenoliths. In terms of TiO₂ content, all studied garnets could be subdivided into two groups: (1) high-Ti garnets (0.86 ± 0.13 wt % TiO₂, n = 252) from sheared lherzolite xenoliths and garnet megacrysts and their intergrowths with ilmenite with low CaO and Cr₂O₃, which is typical of garnets from low-Cr megacryst assemblage (Griffin et al., 1999; Grütter et al., 2004); (2) low-Ti garnets (0.12 ± 0.07 wt % TiO₂, n = 451) with wide variations of CaO and Cr₂O_3; these are garnets from harzburgite and lherzolite xenoliths (Pivin et al., 2009; Ziberna et al., 2013).

Fig. 7.

Composition of garnet from the studied mantle xenoliths and megacrysts from the Grib pipe. Fields in the classification diagram (a) show compositions of garnet widespread in mantle xenoliths: (W) wehrlite, (L) lherzolite, and (H) diamondiferous harzburgite assemblage from the Yakutian kimberlites (Sobolev, 1977). Figures (c, d) show the composition fields of garnet from mantle xenoliths that experienced diverse mantle metasomatism according to (Griffin et al., 1999), where arrows show core-to-rim variations of garnet. Gray field (М) shows the compositions of garnet megacrysts (gray point) from (Kargin et al., 2016; Shchukina et al., 2017).

Note that all studied garnets from garnet harzburgite xenoliths do not fall in the garnet field from similar xenoliths of the Yakutian province (Fig. 7a), but instead are shifted to the field of lherzolitic garnet. Similar compositional evolution is typical of garnet from metasomatically reworked mantle xenoliths (Howarth et al., 2014). The latter indicates that the composition of separate disintegrated grains does not provide reliable information on their mantle source.

Phlogopite. Data on phlogopite from different mantle xenoliths and kimberlite groundmass from the Grib pipe are shown in (Kargin et al., 2019). Two groups of this mineral can be distinguished (Fig. 8a–8c): (1) high-Mg (Mg# = 0.93) phlogopite with low TiO₂ (0.56 ± 0.06 wt %, n = 137) and Cr₂O₃ (0.56 ± 0.11 wt %, n = 137) and (2) low-Mg (Mg# = 0.91) phlogopite with high TiO₂ (2.67 ± 0.36 wt %, n = 37) and Cr₂O₃ (1.49 ± 0.23 wt %, n = 37). Phlogopite of the first group was found as large individual grains in garnet peridotite and clinopyroxene–phlogopite xenoliths, and as large megacrysts. Phlogopite of the second group usually forms marginal zones of high-Mg phlogopites and more frequently occurs in association with high-Ti garnets. Phlogopite data are overlapped with compositions of phlogopite from polymictic breccia xenoliths and high-Ti phlogopite from kimberlites worldwide (Fig. 8). Phlogopite from kimberlitic matrix and kimberlite shells of pyroclasts is close in composition to phlogopites from low-Mg group and is characterized by decreasing Cr₂O₃ with increase of Al₂O₃ (Fig. 8c), FeO (Fig. 8b), BaO (Fig. 8d), which is typical of phlogopite crystallizing from kimberlite melt together with other minerals of kimberlite matrix (Mitchell, 1995).

Fig. 8.

Composition of phlogopite from the studied mantle xenoliths and megacrysts from the Grib pipe. Fields show the compositions of phlogopite from: phlogopite-ilmenite-clinopyroxene (field PIC) and phlogopite–amphibole–rutile–ilmenite–clinopyroxene (MARID field) xenoliths from South Africa kimberlites (Giuliani et al., 2016; Fitzpayne et al., 2018b); polymictic breccia xenoliths (field PB) from the Kimberley province, South Africa (Giuliani et al., 2014); compositions of low-Cr phlogopites from the kimberlite groundmass (KG field) and composition of high-Ti phlogopites ( Ti-K field) from South Africa kimberlites (Giuliani et al., 2016). Dashed arrow shows the composition evolution of phlogopite from kimberlite matrix in the Grib pipe, which reflects the fractionation of kimberlite melt during crystallization matrix minerals (Kargin et al., 2019).

Ilmenite. Megacrystic ilmenite and ilmenite from garnet-free peridotite xenoliths have similar composition and are characterized by high MgO (14.35 ± 0.49 wt %, n = 422) (Fig. 9a) at Cr₂O₃ variations from 1.34 to 3.16 wt % (on average, 2.22 ± 0.41 wt %, n = 422), Al₂O₃—0.53 ± 0.13 wt % and MnO—0.24 ± 0.05 wt % (n = 422). Individual ilmenite grains, including sheared grains, from garnet lherzolite xenoliths are identical to ilmenite from ilmenite peridotites. In general, a range of compositional variations of ilmenite from the Grib kimberlite is overlapped with that of ilmenite from the PIC-type mantle xenoliths and polymictic breccia xenoliths from the Kimberley kimberlites, South Africa (Giuliani et al., 2013; Fitzpayne et al., 2018b).

Fig. 9.

Composition of ilmenite from the studied mantle xenoliths and megacrysts from the Grib pipe. Fields show compositions of phlogopite from: phlogopite–ilmenite–clinopyroxene (field PIC) and phlogopite–amphibole–rutile–ilmenite–clinopyroxene (field MARID) xenoliths from the South Africa kimberlites (Fitzpayne et al., 2018b); polymictic breccia xenoliths (field PB) from kimberlites of the Kimberley province, South Africa (Giuliani et al., 2013), kimberlites of the Kepino field (field KEP, pipes Stepnaya and TsNIGRI-Arkhangelskaya), according to (Kargin et al., 2020).

Trace Elements

In terms of rare-earth element (REE) and trace-element composition, garnet and clinopyroxene are subdivided into several types (Fig. 10). Additional C1 chondrite-normalized and primitive mantle-normalized trace element patterns in the studied garnet and clinopyroxene are given in Supplementary, ESM_3.pdf.

Fig. 10.

C1 chondrite- and primitive mantle-normalized (PM) (McDonough, Sun, 1995) REE and trace-element patterns for the average compositions of garnet, clinopyroxene from different mantle xenoliths, and megacrysts. LREE are light REE.

Garnet. The studied garnets can be subdivided into following types:

(1) High-Ti garnet megacrysts and intergrowths of garnet with other minerals of megacrystic assemblage with typical megacrystic REE distribution patterns characterized by LREE depletion and gradual MREE and HREE enrichment up to 6–10 times chondrites (Fig. 10a). These garnets are also characterized by positive anomalies of high-field strength elements (Nb, Ta, Zr, Hf, Ti) relative to REE (Fig. 10b). In the Zr–Ti and Y–Zr diagrams (Figs. 7c, 7d), they fall into the field of garnet from mantle peridotite xenoliths, which experienced high-temperature mantle metasomatism (Griffin et al., 1999). Similar trace-element patterns were also found in some garnets from lherzolite xenoliths.

(2) Low-Ti garnets from harzburgite and lherzolite xenoliths with moderate to highly sinusoidal REE patterns, which are typical of garnet from lherzolite and harzburgite xenoliths, respectively, from kimberlites worldwide (Fig. 10a). These garnets are characterized by the absence of positive HFSE anomalies (Zr, Hf, and Ti) relative to REE (Fig. 10b). In the Zr–Ti diagrams, they (Fig. 7c) fall in the composition field of garnet from depleted mantle peridotite xenoliths. In the Y–Zr diagram (Fig. 7d), they occupy an intermediate position between garnets from depleted and high-temperature mantle peridotites (Griffin et al., 1999). Some studied garnets have zoned structure with rim-to-core change from high to moderately sinusoidal REE patterns (Kargin et al., 2016).

(3) Some low-Ti garnets from lherzolite xenoliths with REE pattern (Fig. 10a) typical of garnet megacrysts, with the absence of positive Ti anomaly at positive Zr and Hf anomalies relative to REE (Fig. 10b). In addition, these garnets overlap field of megacrystic garnets in the CaO–Cr₂O₃ diagram (Fig. 7a).

(4) High-Ti garnet from sheared lherzolite xenolith (outer zones of large garnet porphyroblasts with harzburgitic REE pattern), enriched in LREE (Fig. 10a), Ti, Zr, and Y (Fig. 10b). Thereby, MREE and HREE patterns are comparable with those of garnet megacrysts (Fig. 10a). In the Zr–Ti and Y–Zr diagrams (Figs. 7c, 7d), as other high-Ti garnets, they fall in the field of garnet from mantle peridotite xenoliths subjected to high-temperature mantle metasomatism (Griffin et al., 1999).

Clinopyroxene. Like garnet, clinopyroxene from the studied xenoliths is also subdivided into several groups in terms of trace-element content (Figs. 10c, 10d):

(1) clinopyroxene from garnet peridotite xenoliths with low-Ti garnet has a fractionated REE pattern and (La/Sm)_n from 0.4 to 7.4 (Fig. 10c). Multielement patterns display sharp negative Zr–Hf anomalies and moderate negative Ti and Nb–Ta anomalies (Fig. 10d).

(2) clinopyroxene in equilibrium with high-Ti garnet from garnet peridotite xenoliths has a moderately fractionated REE pattern (Fig. 10c) with elevated contents of Nd and Sm, negative Zr and Hf anomalies relative to REE, and the absence of Ti anomaly (Fig. 10d). The degree of REE fractionation is comparable to that of clinopyroxene from sheared peridotite xenolith (Fig. 10c).

(3) clinopyroxene from clinopyroxene–phlogopite xenoliths, as well as high-Cr megacrysts show LREE and MREE-enriched patterns with maximum in the Pr–Nd region and flat LREE patterns (Fig. 10c). It is also characterized by the negative HFSE (Nb, Ta, Zr, Hf, and Ti) anomalies relative to REE (Fig. 10d).

Late spongy clinopyroxenes have highly fractionated REE patterns (Fig. 10c) and the negative Zr–Hf and Ti anomalies (Fig. 10d).

Ilmenite. Concentrations of Nb, Ta, Zr, Ni, Zn, and V in the studied ilmenites widely vary and correlate with MgO content. In particular, Nb, Ta, Zr, Zn, and V contents increase, while Ni, decreases with decreasing MgO (Kargin et al., 2020). Similar behavior of trace-elements in the studied ilmenites in general is comparable with that of Mg-ilmenites from kimberlites worldwide (Moore et al., 1992; Castillo-Oliver et al., 2017). Note that the studied ilmenites from the Grib pipe in general have low Zr contents (Fig. 9b) compared to ilmenites from kimberlites of the Kepino cluster of the ADP and ilmenites from the PIC-type xenoliths and polymictic breccia from the Kimberley pipe, South Africa (Giuliani et al., 2013; Fitzpayne et al., 2018b).

Orthopyroxene. By analogy with garnet, orthopyroxene shows an increase of Zr and Y with increasing TiO₂ (Supplementary, ESM_2.xlsx) from garnet harzburgite xenolith to sheared lherzolite xenolith. An increase in TiO₂ content is also accompanied by an increase of Zr/Nb ratio; and orthopyroxene neoblasts from sheared lherzolite xenolith and orthopyroxene from intergrowths with garnet in the TiO₂–Zr/Nb diagram occupy an intermediate position between orthopyroxenes from ganet peridotite and polymictic breccia (Fig. 5d).

Thus, based on the mineral composition of mantle peridotites and megacrysts, two rock types can be distinguished: (1) rocks with high-Ti, mainly low-Mg garnet and clinopyroxene and (2) rocks with low-Ti mainly high-Mg garnet and clinopyroxene. The xenoliths of the first type usually contain ilmenite, while the second type rock more frequently contains low-Ti and low-Cr phlogopite. In most cases, Mg# positively correlates with Cr₂O₃ content.

P-T ESTIMATES

Pressure (Р) and temperature (Т) were calculated for peridotite and clinopyroxene–phlogopite xenoliths, as well as for clinopyroxene megacrysts and garnet–ilmenite–clinopyroxene intergrowths (Supplementary, ESM_2.xlsx, Table S6) using geothermometers (Brey and Köhler, 1990; Taylor, 1998) with correction according to (Nimis and Grütter, 2010); and geobarometers (MacGregor, 1974; Nickel and Green, 1985). P-T equilibrium conditions for high-Cr clinopyroxene megacrysts and clinopyroxene from clinopyroxene–phlogopite xenoliths were calculated using thermobarometer (Nimis and Taylor, 2000). Detailed results are given in (Kargin et al., 2016, 2017a, 2017b).

The lherzolitic assemblage of sheared lhezolite xenolith yielded the highest Р = 7 GPa and Т = 1220°C. Garnet peridotite xenoliths with low-Ti garnets gave the wider Р–Т range: 2.2–5.0 GPa and 730–1070°C. High-Cr clinopyroxene megacrysts and clinopyroxene from clinopyroxene–phlogopite xenoliths were formed within 3.6–4.7 GPa and 760–920°C, which in general is overlapped with conditions of formation of garnet peridotites with low-Ti garnet. Orthopyroxene–ilmenite–garnet xenolith yielded the following Р–Т estimates: 4.4 ± 0.3 GPa and ~1020°C. Conditions of formation of high-Ti garnet megacrysts were estimated at 5.5 GPa and 1150°C ± 30 (Lebedeva et al., 2020b).

In general, obtained data (Fig. 11) correspond to previously published Р-Т parameters for xenoliths and megacrysts from the Grib kimberlite (Kostrovitsky et al., 2004; Golubkova et al., 2013; Shchukina et al., 2015) and demonstrate that the majority of the studied xenoliths was brought up from the middle part of lithospheric mantle (depths of 90–150 km), where temperatures were between 35 and 40 mW/m² geotherms (Pollack and Chapman, 1977).

Fig. 11.

P-T parameters of equilibrium of mineral phases from the studied mantle xenoliths. Line 2 (APG) is the mantle geotherm of ADP (Afanasiev et al., 2013), line 3 is G/D (graphite–diamond) transition boundary. Field 1 according to (Afanasiev et al., 2013).

DISCUSSION

The study of mantle xenoliths and megacrysts from the Grib kimberlite suggests that the ADP lithospheric mantle experienced several stages of metasomatic transformations. The widest spread type of metasomatism is expressed in the geochemical enrichment (fertilization) of depleted garnet harzburgites with their transformation into garnet lherzolites (Fig. 12). Similar type of mantle metasomatism is typical of mantle xenoliths from kimberlites worldwide (Griffin et al., 1999; Burgess, Harte, 2004; O’Reilly, Griffin, 2013). The degree of involvement of the depleted mantle in metasomatic process (i.e., fluid/rock ratio) depends on the distance from metasomatic melt conduit and can determine the wide variations in proportions of garnet and clinopyroxene in mantle xenoliths (Bussweiler et al., 2018). Subsequent transformation of mantle peridotites leads to the formation of phlogopite peridotites and clinopyroxene–phlogopite rocks (O’Reilly and Griffin, 2013).

Fig. 12.

Photos of the studied peridotite xenoliths demonstrating an increase of clinopyroxene content, which reflects the enrichment of depleted harzburgites (a) and their transformation into garnet lherzolites (b, c). (Cpx) clinopyroxene, (Grt) garnet, (Ol) olivine, (Kmb) kimberlite groundmass.

Based on analysis of available data on mantle xenoliths and xenocrysts from kimberlites worldwide, two approaches to their interpretation can be distinguished. The first approach is based on the assumption that the majority of mantle xenoliths brought up by kimberlites and the megacrystic minerals were formed prior to the formation of kimberlite melts or with contribution of melts of other composition. This approach can be applied to reconstruct the structure of lithospheric mantle at the moment of kimberlite formation. According to this approach, kimberlite melts themselves do not cause significant transformations of mantle xenoliths, but serve only as carriers of disintegrated fragments of lithospheric mantle. Multiple metasomatic transformation of lithospheric mantle caused by different agents, not kimberlites, was proposed for xenoliths from the Grib kimberlite (Shchukina et al., 2015). The complex studies of xenoliths from the ADP kimberlites were used to construct the lithospheric mantle section beneath ADP (Sablukov et al., 2000, 2004) and to predict the large thickness of lithospheric mantle (210–180 km) beneath the economically diamondiferous and poorly diamondiferous kimberlites of the province.

The second approach suggests that the formation and ascent of kimberlite melts are a multiple process, during which some portions of kimberlite melt do not reach the Earth’s surface and crystallize in mantle (Giuliani et al., 2016). Correspondingly, the next portions of kimberlite melts cause prograde metasomatism of lithospheric mantle, thus forming a permeable zone, conduit, which is in more geochemical equilibrium with kimberlite melt. In this case, kimberlite melts ascend to the surface by percolation through lithospheric mantle (e.g., Harte, 1983), fractionally crystallizing megacrystic mineral assemblage and causing metasomatism of surrounding lithospheric rocks. The latter will be metasomatically transformed from depleted garnet harzburgites into enriched garnet lherzolites (e.g., Bussweiler et al., 2018). According to this interpretation, most part of mantle xenoliths brought by kimberlites are fragments of walls of mantle conduits, along which kimberlite melts reached surface (Bussweiler et al., 2018), and, respectively, are not representative of the lithospheric mantle. Similar conclusions were made from study of garnet peridotites and clinopyroxene–phlogopite xenoliths from the Grib kimberlite (Kargin et al., 2016, 2017a, 2017b).

The study of metasomatic transformation of mantle rocks encountered also the problem of origin of megacrystic mineral assemblage. Megacrysts could be genetically related to kimberlitic or protokimberlitic melts (Moore and Lock, 2001; Moore and Belousova, 2005; Kopylova et al., 2007) or were formed through fractional crystallization of asthenospheric silicate melts, which have existed prior to the formation of kimberlite melts and represented a preparatory stage for kimberlite generation (Burgess and Harte, 2004; Solov’eva et al., 2008).

Origin of minerals of megacrystic assemblage from the Grib kimberlite remains controversial. On the one hand, the chemical composition of megacrystic garnet indicates its crystallization from a silicate melt compositionally close to the ADP alkaline picrites (Mahotkin et al., 2000; Shchukina et al., 2017). On the other hand, the geochemical and isotope-geochemical characteristics of ilmenite megacrysts (Golubkova et al., 2013), clinopyroxene, and phlogopite (Kargin et al., 2019) from the Grib kimberlite suggest that these minerals were in equilibrium with kimberlite melts. Kostrovitsky et al. (2004) proposed that megacrysts from the Grib kimberlite were derived by the interaction of asthenospheric protokimberlite melts with lithospheric mantle.

Below, we consider the nature and peculiarities of mantle metasomatism of the ADP lithospheric mantle and its relationship with kimberlite or protokimberlite melts and megacrystic assemblage.

Mantle Metasomatism

Several stages of mantle metasomatism were deduced from mantle xenoliths in kimberlites (O’Reilly and Griffin, 2013): (1) modal metasomatism expressed in the introduction of new minerals atypical of peridotite paragenesis, for instance, phlogopite, amphibole, carbonate, ilmenite, apatite, and others (Harte, 1983); (2) cryptic metasomatism leading to the compositional changes of existing mineral phases, for instance, the emergence of chemical zoning or lherzolitic composition of garnet in harzburgitic xenoliths. Thereby, pre-kimberlite metasomatism of xenoliths frequently results in the decoupling in the trace- and major element variations (e.g., Howarth et al., 2014). (3) cryptic or stealth metasomatism (O’Reilly and Griffin, 2013), which suggests an introduction of new mineral phases typical of mantle peridotites, i.e., garnet and clinopyroxene.

The study of the mantle xenoliths from the Grib kimberlite allowed us to identify all above mentioned types of mantle metasomatism. The modal mantle metasomatism is expressed in the formation of phlogopite, clinopyroxene–phlogopite aggregates, and ilmenite (Figs. 3, 4), producing phlogopite peridotite, clinopyroxene–phlogopite, and ilmenite peridotites (dunite) xenoliths.

The studied mantle xenoliths show the following textural and structural relationships: (1) formation of clinopyroxene, more rarely garnet, along boundaries between olivine and orthopyroxene, as well as along margins and fractures in these minerals in garnet harzburgite xenoliths (Figs. 3a–3c); (2) replacement of olivine and orthopyroxene by clinopyroxene along fractures and grain boundaries; replacement of neoblasts of these minerals by clinopyroxene in sheared lherzolite xenoliths (Figs. 3a–3d and Fig. 4a); (3) olivine inclusions in clinopyroxene neoblasts (Fig. 3d); (4) sharp chemical zoning in BSE images (Fig. 3f), which could indicate a disequilibrium state of major minerals of mantle peridotites, mainly clinopyroxene. All these facts suggest that the lherzolitization of depleted peridotites occurred shortly before the xenolith entrapment by kimberlite melts. The Sr–Nd isotope composition of garnet, clinopyroxene, and orthopyroxene from garnet lherzolites, garnet harzburgites, and sheared garnet lherzolites from the Grib kimberlite is consistent with mixing between isotopically contrasting components: ancient metasomatized mantle and kimberlite melts (Lebedeva et al., 2020a).

The wide chemical variations of minerals in mantle peridotites could be used for the identification of cryptic metasomatism. From garnet harzburgite xenoliths to sheared lherzolite and intergrowths with high-Ti garnet and ilmenite, orthopyroxene shows a gradual increase of TiO₂ content with a decrease of Mg# (Figs. 5 and 13a). Correlation r between TiO₂ and Mg# is ‒0.79 (n = 211). The same trend is observed for clinopyroxene (Figs. 13b, 13c).

Fig. 13.

Statistical distribution of TiO₂ (a, b) and Mg# (c) in orthopyroxene and clinopyroxene from the studied xenoliths, in order of increasing metasomatic agent/lithospheric mantle ratio.

During geochemical enrichment (refertilization) of mantle peridotites, the major and trace element contents in garnet are also controlled by cryptic metasomatism (Kargin et al., 2016): on the one hand, the garnet demonstrates a decrease of CaO and Cr₂O₃, which is consistent with observed equilibrium with clinopyroxene, i.e., reveals a lherzolitic trend (Fig. 7a); on the other hand, REE patterns change from sharply sinusoidal harzburgitic profile (Griffin et al., 1999) through weakly sinusoidal lherzolitic (Lazarov et al., 2009) to typical megacrystic profiles (Fig. 10).

A continuous trend of mineral variations in the Mg#–TiO₂ diagram could be interpreted as indicator of fluid/rock ratio, i.e., variations of metasomatic agent to lithospheric mantle ratio, because the interaction with lithospheric mantle leads to the increase of Mg# in a melt (review in Giuliani et al., 2020). In this case, an increase of Mg# indicates an increase of this ratio or increasing distance from a metasomatic source. For instance, TiO₂-rich low-Mg pyroxenes from sheared lherzolite xenolith were in equilibrium with metasomatic agent, whereas high-Mg pyroxenes depleted in TiO₂ suggest the low fluid/rock ratio, i.e., the greater contribution of depleted lithosphere. Such metasomatic zoning can be explained by the interaction of alkaline–ultrabasic melts (kimberlite or protokimberlite melts, which represent the early portions of kimberlite melts and could differ in composition from those of kimberlite body) with surrounding lithosphere mantle, which led to the formation of megacrystic minerals (high- and low-Cr assemblages), as well as the introduction of new clinopyroxene and garnet in mantle peridotites, as was proposed for kimberlites of Canada (Bussweiler et al., 2018). During this interaction, kimberlite melt did not reach the Earth’s surface and was crystallized in the lithospheric mantle with formation of ilmenite-rich polymictic breccia (Giuliani et al., 2014; Pokhilenko, 2009; Bussweiler et al., 2018) and/or could serve as mantle metasomatic source responsible for the formation of ilmenite and PIC-type assemblage (Gregoire et al., 2002, 2003; Fitzpayne et al., 2018b).

Thus, the petrographic and mineralogical features of the studied xenoliths show that the mantle metasomatism of the ADP lithospheric mantle caused the transformation of depleted peridotites from garnet harzburgites to garnet lherozlites, phlogopite–garnet wehrlites, and clinopyroxene–phlogopite rocks (Fig. 12). This type of metasomatism is widespread among xenoliths brought up by kimberlites (review in O’Reilly and Griffin, 2013).

Relationship of Mantle Metasomatism with Megacrysts

Clinopyroxene and phlogopite megacrysts from the Grib kimberlite are close in composition to high-Cr megacrysts from kimberlite provinces worldwide (Kargin et al., 2017a, 2019). The clinopyroxene megacrysts are also similar to clinopyroxenes from some garnet lherzolite xenoliths (Shchukina et al., 2015) and studied clinopyroxene–phlogopite xenoliths (Figs. 6 and 10b). The phlogopite megacrysts are identical to “primary” phlogopite from peridotite xenoliths in kimberlites worldwide (Carswell, 1973), as well as to phlogopite from the studied clinopyroxene–phlogopite xenoliths (Fig. 8). Ilmenite megacrysts (Fig. 9) in composition are also similar to ilmenites from garnet-free peridotite xenoliths from the Grib kimberlite (Kargin et al., 2020). Garnets similar to high-Ti megacrystic garnets more rarely occur in peridotite xenoliths than their low-Ti varieties (Kargin et al., 2016). The high-Ti garnet megacrysts are most close to the garnet neoblasts from sheared lherzolite xenoliths (Fig. 7), while clinopyroxenes from these xenoliths are close to the low-Cr clinopyroxene megacrysts. In general, the similarity of minerals from sheared peridotites with low-Cr megacrysts was mentioned for xenoliths from other kimberlites (e.g., Moore and Lock, 2001; Kostrovitsky et al., 2013).

The observed similarity of megacrysts and minerals from peridotite and other xenoliths suggests that the megacrysts are disintegrated fragments of coarse-grained varieties of these rocks, and that the formation of megacrysts and transformation of mantle peridotites occurred under similar conditions during a single petrological process. The latter is consistent with suggestion by (Bussweiler et al., 2018) that crystallization of megacrysts occurs during percolation of kimberlite melts through depleted lithospheric mantle in a main channel, while further away from the channel, this process causes the enrichment of depleted peridotites and introduction of garnet and clinopyroxene.

Obtained results show that megacrysts, as minerals from mantle xenoliths, could be subdivided not only on high- and low-Cr, as accepted for megacrysts from kimberlite provinces worldwide (Moore and Belousova, 2005), but also on high-Ti low-Mg and low-Ti high-Mg groups (Figs. 5–8). Difference in the Ti and Fe contents could be related to the formation of ilmenite, ilmenite veinlets, or ilmenite-bearing assemblages in the lithospheric mantle during separation of Fe–Ti liquids (or melts enriched in these elements) from initial metasomatic agent or with its peculiar evolution (Giuliani et al., 2014; Soloviev et al., 2019).

Evolution of Metasomatic Agent

One of the widely applied approaches to the identification of composition of metasomatic agent and its evolution is the calculation of trace-element composition of equilibrium melts from the trace-element composition of minerals formed during this process using mineral/melt partition coefficient (Kd). The results of such calculations strongly depend on the choice of partition coefficients, which in turn is determined by the inferred composition of parental melts. However, there are difficulties with a choice of partition coefficients for kimberlite-forming systems, because the compositions of kimberlite melts involved in mantle metasomatism and melts responsible for the transformation of mantle rocks and formation of megacrysts are controversial (see Introduction). Moreover, melts are evolved from carbonatite to alkaline–ultrabasic ones (e.g., Giuliani et al., 2020). The melt composition can be determined from geochemical equilibrium of the coexisting mineral phases, for instance, garnet and clinopyroxene in the studied mantle xenoliths and megacrysts calculated as the clinopyroxene/garnet concentration ratio for element i $\left( {{{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}} \right).$ The most representative are the garnet/clinopyroxene REE ratios (Ziberna et al., 2013) calculated for each xenolith group (Fig. 14a). Available experimental determinations of partition coefficients for silicate and carbonate melts can be used for deciphering megacrysts and xenolith minerals that were in equilibrium with definite melts. The comparison of observed ${{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}$ ratios with experimental one $\left( {{{Kd_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{Kd_{i}^{{Grt}}} {Kd_{i}^{{Cpx}}}}} \right. \kern-0em} {Kd_{i}^{{Cpx}}}}} \right)$ makes it possible to estimate the mineral equilibria conditions in nature.

Fig. 14.

(a) Variations of D^Grt/Cpx = ${{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}$ for REE in the peridotite xenoliths from kimberlites of the Grib pipe compared to experimental data on carbonatite (Dasgupta et al., 2009) and alkaline-basic melts at 900°C (Burgess and Harte, 2004) and 1430°C (Johnson, 1998); (b, c) C1 chondrite-normalized (McDonough and Sun, 1995) model melt composition, which could occur in equilibrium with: (b) high-Ti garnet and clinopyroxene from sheared lherzolite xenoliths calculated using mineral/melt partition coefficient according to (Dasgupta et al., 2009), (c) garnet megacrysts calculated using mineral/melt partition coefficients (Burgess and Harte, 2004); (d) primitive mantle-normalized (PM) (McDonough and Sun, 1995) trace-element patterns in model melts, which could be in equilibrium with high-Cr clinopyroxene megacrysts calculated using mineral/melt partition coefficients (Dasgupta et al., 2009). Composition of juvenile pyroclasts from the Grib kimberlite after (Golubeva et al., 2006). Estimated compositions of primary portions of kimberlite melts from the Jericho pipe, Canada, and South Africa group 1 kimberlites according to (Price et al., 2000) and (Becker and Roex Le, 2006), respectively.

The ${{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}$ ratio for low-Ti garnet and clinopyroxene from garnet peridotite xenoliths is most close to ${{Kd_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{Kd_{i}^{{Grt}}} {Kd_{i}^{{Cpx}}}}} \right. \kern-0em} {Kd_{i}^{{Cpx}}}},$ obtained for garnet and clinopyroxene in equilibrium with silicate alkaline–ultramafic melts at Р ~ 3 GPa and Т ~ 1000°C (Johnson, 1998; Burgess and Harte, 2004). In turn, the LREE-rich high-Ti garnet and coexisting clinopyroxene from sheared lherzolite xenolith have ${{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}$ values (Fig. 14a) typical of equilibrium with carbonatite melts (Dasgupta et al., 2009).

The values of ${{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}$ for megacrysts of high-Ti garnets and high-Cr clinopyroxenes are inconsistent with equilibrium curves for silicate and carbonate melts. They differ in the lower ${{C_{i}^{{Grt}}} \mathord{\left/ {\vphantom {{C_{i}^{{Grt}}} {C_{i}^{{Cpx}}}}} \right. \kern-0em} {C_{i}^{{Cpx}}}}$ values for REE and are inconsistent for Ti, Nb, and other trace elements (see Supplementary, ESM_3.pdf), which suggests that the minerals are in disequilibrium and were likely formed at different evolution stages of the kimberlite-forming system.

The model compositions of melts in equilibrium with high-Ti garnet and clinopyroxene from sheared lherzolite xenolith were calculated using mineral/carbonatite partition coefficients according to (Dasgupta et al., 2009). They show strong REE fractionation (Fig. 14b) and are close in composition to the model 1.5% melts (corresponding to carbonatites) from carbonated depleted mantle according to (Grassi and Schmidt, 2011). This confirms the assumption that a melt in equilibrium with high-Ti garnet and clinopyroxene from sheared lherzolite xenolith likely represented a carbonate-rich component of alkaline–ultrabasic melt (Kargin et al., 2017b).

The high-Ti garnet megacrysts have lower LREE contents than high-Ti garnets from sheared lherzolite xenolith formed at greater depths, which could be caused by an increasing content of silicate component in equilibrium melt or metasomatic agent (Burgess and Harte, 2004) during melt ascent through lithospheric mantle (Giuliani et al., 2020 and references therein). The compositions of melts equilibrium with garnet megacrysts were calculated using Kd for silicate melts of asthenosphere nature (Burgess and Harte, 2004). They are most close to the kimberlite pyroclasts from the Grib pipe (Fig. 14c), which represent crystallized kimberlite melt (Golubeva et al., 2006). The latter suggests that garnet megacrysts could be in equilibrium with kimberlite melt.

The model composition of melts in equilibrium with high-Cr clinopyroxene megacrysts (Kargin et al., 2017a) calculated using partition coefficients (Burgess, Harte, 2004) is overlapped with compositions of primary kimberlite melts estimated for the Jericho Pipe, Canada (Price et al., 2000) and to lesser extent, with those of South Africa group I kimberlites (Becker and Roex Le, 2006).

Thus, our calculations revealed that metasomatic agent equilibrium with high-Ti garnet and clinopyroxene from sheared lherzolite xenolith at the base of lithospheric mantle (pressure near 7 GPa) had a silicate–carbonate composition, whereas the megacrysts of high-Ti garnet and high-Cr clinopyroxene at the middle horizons of lithospheric mantle (pressure from 5 to 3.5 GPa) were in equilibrium with the alkaline–ultrabasic melt. Based on assumption that metasomatic transformations occurred shortly before the xenolith entrapment by kimberlite melt and could occur during formation of magmatic channel by kimberlite melts, such metasomatic agent could be kimberlite or protokimberlite melt, which changed its composition en route through lithospheric mantle. Obtained conclusions are consistent with isotope-geochemical equilibrium of kimberlites from the Grip pipe with kimberlite melt (Golubkova et al., 2013), while the evolution of kimberlite melt is consistent with previously proposed model that megacrysts from the Grib kimberlite were formed through the interaction of asthenospheric protokimberlite melts with lithospheric mantle (Kostrovitsky et al., 2004).

Mantle Metasomatism during Formation of Kimberlite Melts

Revealed genetic link between kimberlite melts and mantle metasomatism and formation of megacrysts suggests that these processes represent a single episode in the evolution of the kimberlite-forming system (Kostrovitsky et al., 2013; Kargin et al., 2017a; Bussweiler et al., 2018; Solovieva et al., 2019). Obtained data allowed us to reconstruct the following sequence of metasomatic transformation of lithospheric mantle during formation of kimberlite conduit (Fig. 15).

Fig. 15.

Schematic sequence of metasomatic transformations of lithospheric mantle during formation and evolution of kimberlite melts. See text for explanation.

(1) Interaction of the early portions of LREE and Fe–T-rich kimberlite melts of carbonate-silicate composition with surrounding depleted harzburgites at the base of lithospheric mantle beneath the ADP (at a depth of 180–210 km). These melts can be exemplified by alkaline–ultramafic lamprophyres of the ailikite–carbondatite series. This stage was responsible for the metasomatic transformation of sheared peridotite with introduction of low-Mg and high-Ti orthopyroxene and olivine neoblasts, as well as high-Ti garnets and low-Mg clinopyroxene (Kargin et al., 2017b) at ~1220°C and 7 GPa (Fig. 15, stage I).

(2) Further interaction of kimberlite with surrounding mantle rocks during ascent through lithospheric mantle from its base to depths of 150–120 km. Formation of the Fe- and Ti-rich megacrysts of garnet and clinopyroxene, ilmenite, and intergrowths of these minerals. The “new” minerals are close in composition to that of low-Cr megacrysts, which are widespread in kimberlites worldwide (Fig. 15, stage II). The fraction of carbonate component in a melt decreases owing to the interaction with lithospheric mantle (e.g., Russell et al., 2012; Giuliani et al., 2020). Р-Т parameters of this stage were T = 1000–1100°C and Р = 4–5 GPa (Sablukov et al., 2009; Lebedeva et al., 2020b). This stage likely produced the greater amount of ilmenite-bearing peridotites (Kargin et al., 2020) owing to the separation of Fe–Ti melt (Solovieva et al., 2019) or direct crystallization of kimberlite melts in mantle with formation of polymictic breccias (Pokhilenko, 2009; Giuliani et al., 2014).

(3) Formation of Fe–Ti minerals led to the depletion of kimberlite melts in these elements. Further ascent of these kimberlite melts through lithospheric mantle caused the transformation of depleted peridotites into garnet lherzolites with formation of lherzolitic low-Ti garnet and clinopyroxene similar to the high-Cr megacrysts from kimberlites worldwide (Kargin et al., 2017a). The metasomatic transformations of this stage occurred at 730–1070°C and 3–5 GPa (Fig. 15, stage III). With increasing distance from the main kimberlite conduit at this stage, the contribution of lithospheric mantle increased, which is consistent with an increase of Mg# and high variations of Cr₂O₃ content in clinopyroxene (Fig. 6).

(4) Transformation of garnet lherzolites into phlogopite peridotites, clinopyroxene–phlogopite rocks, and low-Cr phlogopite megacrysts (Kargin et al., 2019) by residual melts that remained after formation of garnet and clinopyroxene and are enriched in K and H₂O ± CO₂ (Fig. 15, stage IV). Formation of phlogopite is accompanied by the intense replacement of garnet (Fig. 3f), which facilitates the transformation of garnet lherzolites into phlogopite wehrlites and clinopyroxene–phlogopite rocks (Kargin et al., 2017a, 2019). The Rb-Sr isotope system of phlogopite indicates that the mineral was in isotope-geochemical equilibrium with a melt similar to the Girb kimberlite (Kargin et al., 2019).

(5) Ascent of new portions of kimberlite melts along magmatic conduit metasomatized by the early melt portions facilitates the entrapment of mantle xenoliths and megacrysts with formation of Ti-rich outer rims in phlogopite, pyroxene, and polyphase inclusions in garnet megacrysts (Lebedeva et al., 2020b).

CONCLUSIONS

The study of composition of garnet, clinopyroxene, orthopyroxene, phlogopite, and ilmenite from garnet peridotite, ilmenite peridotite (dunites), and clinopyroxene–phlogopite xenoliths and megacrysts of these minerals in combination with their petrographic features provided insight into relationship of mantle metasomatism with kimberlite-type alkaline ultrabasic melts and their evolution during formation of magmatic conduit.

(1) Petrographic features of peridotite xenoliths show that mantle metasomatism of depleted garnet harzburgites leads to their geochemical enrichment and transformation into garnet lherzolites, phlogopite-garnet wehrlites, and clinopyroxene–phlogopite rocks. This process occurred shortly before the entrapment of these xenoliths by kimberlite melts.

(2) Calculation of model compositions of melts that could be in equilibrium with garnet and clinopyroxene from different types of mantle xenoliths and megacrysts demonstrates that the sheared lherzolites at the base of lithospheric mantle were metasomatized by kimberlite melts of alkaline–carbonate–silicate composition or their analogues, for instance, ailikitic melts. The formation of high-Ti garnet and high-Cr clinopyroxene megacrysts in the middle part of lithospheric mantle and the global lherzolitization of depleted harzburgites occurred in equilibrium with kimberlite melts, in which the fraction of silicate component increased while the content of Fe–Ti component decreased. The modification of equilibrium melt could be related to the interaction with mantle peridotites on passing/ascent of the melt through them.

(3) Significant variations of TiO₂ content and Mg# in pyroxenes, with formation of continuous trends in the binary diagrams are consistent with variations of metasomatic agent/lithospheric mantle (fluid/rock) ratio, which reflects the distance from the main channel for the ascent of metasomatizing melts.

(4) Similarity of garnet, clinopyroxene, phlogopite, and ilmenite megacrysts with minerals from peridotite xenoliths suggests that the megacrysts either represent disintegrated fragments of the coarsed grained varieties of these xenoliths or were formed at the same conditions that caused mantle metasomatism of these rocks or crystallized directly from a metasomatic agent.

Thus, lithospheric mantle beneath the ADP from its base (depth around 180–210 km) to depths corresponding to pressure of 3.5 GPa (near 100–120 km) experienced intense mantle metasomatism under the influence of alkaline–ultrabasic melts, which could be kimberlites or their precursors responsible for the formation of magma channels, along which the main portions of kimberlite melts reached the Earth’s surface. During formation of magma channel, the melt interacted with lithospheric mantle and evolved from protokimberlite melts enriched in REE, Ti, Fe, and carbonate component to silicate-rich Fe–Ti-depleted ultrabasic kimberlite melts. The proportions of these end compositions are determined by the degree of reworking of mantle channel, which served as pathways for primary kimberlite melts.

It was established that mantle xenoliths found in the Grib kimberlite were formed during several stages of mantle metasomatism, which were related to the several pulses of kimberlite melts. Thereby, the later portions of kimberlite melts could interact with lithospheric mantle reworked by the earlier portions. Such complex history of the interaction of kimberlite/protokimberlite melts led to the formation of magmatic conduits, whose walls reached a complete equilibrium with kimberlite melts in the lower part of lithosphere mantle.

Undoubtedly, the revealed intense transformation of lithospheric mantle by kimberlite melts was not a single episode of mantle metasomatism of lithospheric mantle, and some xenoliths from the Grib pipe, which were not described in this study, suggest the older pre-kimberlite mantle metasomatism (Shchukina et al., 2015). The study of Rb-Sr, Sm-Nd, and oxygen isotope systems of minerals in the mantle xenoliths from the Grib pipe made it possible to determine the ages of pre-kimberlite stages of mantle metasomatism in the lithosphere beneath the Arkhangelsk Diamond Province. It was established that the ages of these stages coincide with Neoproterozoic tectonothermal events in Fennoscandia, which were related to the Rodinia breakup (610–550 Ma) and an episode of alkaline ultrabasic magmatism in Karelia and East Finland at ~1.2–1.0 Ga (Lebedeva et al., 2020a).