Thermal State and Thickness of the Lithospheric Mantle Beneath the Northern East-European Platform: Evidence from Clinopyroxene Xenocrysts in Kimberlite Pipes from the Arkhangelsk Region (NW Russia) and Its Applications in Diamond Exploration

Abstract

This paper presents the reconstruction of the architecture of the lithospheric mantle, including its thermal state and thickness, as well as the scale and efficiency of its sampling by four kimberlites from the Arkhangelsk diamondiferous province: Arkhangelskaya, Lomonosovskaya, V. Grib, and TSNIGRI-Arkhangelskaya. These kimberlites differ in terms of their composition, diamond content, and location. Data presented include the major-element composition of clinopyroxene xenocrysts (>2000 grains), P–T calculations from compositionally filtered Cr-diopside grains, and the reconstruction of local paleogeotherms. Additionally, we used available data on Ni content in peridotitic garnet xenocrysts to calculate their T values and project them onto local Cr-diopside-derived geotherms to reconstruct the vertical distribution of mantle xenocrysts and assess the efficiency of lithospheric mantle sampling by different kimberlites. We identified the presence of a >200 km-thick lithospheric mantle beneath the region at the time of kimberlite emplacement. We also found that the diamond content of the studied pipes was, to some extent, dependent on the following set of factors: (1) the thermal state of the lithospheric mantle; (2) the width of the real “diamond window” marked by mantle xenocrysts, especially by diamond-associated garnets; and (3) the efficiency of lithospheric mantle sampling by kimberlite. The results of this study can be used to inform diamond exploration programs within the region.

1. Introduction

Kimberlite pipes can contain a wide variety of deep-seated xenoliths and their fragments (xenocrysts), which can provide researchers with an opportunity to reconstruct the architecture and evolution of the lithospheric mantle (LM) beneath ancient cratons. An evaluation of the main features of the LM beneath kimberlite pipes located within the same kimberlite field/province allows for the assessment of the vertical and lateral heterogeneity of the LM, including its temporal evolution (if LM was sampled by kimberlites of different ages); in particular, understanding the reasons behind the differences in the diamond content of these pipes within this region is of great interest [1,2]. Reconstruction of the architecture of the LM is impossible without establishing the thermal state of the LM at the time of kimberlite emplacement, which can be determined by calculating the pressure (P) and temperature (T) of the mantle xenoliths and their disaggregated minerals (xenocrysts). The P–T conditions can be directly retrieved from single clinopyroxene grains [2,3,4] after applying the compositional filter criteria outlined in [1,3,4,5,6].

There are three main ages of kimberlite magmatism in the northern East-European Platform: first, the Paleoproterozoic Kimozero (1.92 Ga) and Kostomuksha (1.2 Ga) kimberlites of central Karelia in Russia [7] and the Kuhmo-Lentiira (~1.2 Ga) kimberlites in Finland [8,9]; second, the Neoproterozoic Kuusamo (757 ± 2 Ma) and Kaavi-Kuopio (626–589 Ma) kimberlites of eastern Finland [8,9]; and, third, the Devonian–upper Carboniferous kimberlites of the Terskii Coast (380–360 Ma; [10]), Timan Ridge [11] and Arkhangelsk Diamondiferous Province (ADP, 410–340 Ma; [11,12,13,14]) in the European part of Russia.

Much of what is known about the thermal state and thickness of the LM beneath the East-European Platform comes from studies of mantle xenoliths and xenocrysts from the Finnish kimberlites [8,9,15,16,17,18] (Karelian craton). Thermobarometry of mantle xenoliths and xenocrysts from these kimberlites indicates low geothermal gradients (36 mW/m² [8,9]) and considerable variations in LM thickness of the Karelian craton from margin to core [8,9,15,16,17,18]. The mantle root of the Karelian craton reaches its greatest thickness (>250 km) in the central portion (Lentiira–Kuhmo–South Kuusamo pipes) and thins (220–230 km) towards the SW and NE edges (Kaavi–Kuopio and North Kuusamo pipes) [8,9,15,16,17,18].

The Arkhangelsk diamondiferous province (ADP) hosts ~100 magmatic bodies of ultramafic and mafic composition, including ~70 kimberlites, of which only 7 are economically diamondiferous [11,12,13,14,19]. Despite the large number of kimberlites in the ADP, mantle-derived xenoliths is only found in the highly diamondiferous V. Grib kimberlite pipe [20,21,22,23,24,25,26,27,28]. The peridotite-derived geotherm for the LM beneath the V. Grib pipe was proposed to involve 36–40 mW/m² heat flows [20,21,24] based on the model derived by Pollack and Chapman (1977) [29]. The clinopyroxene-derived geotherm was previously reconstructed for the LM beneath the Arkhangelskaya [30] and the V. Grib kimberlite pipes [31] and was modelled as 35–40 mW/m² heat flows [29]. Afanasiev et al. (2013) [32] proposed an “average” filtered geotherm for the ADP LM and noted the heating of the base of the LM beneath the majority of the kimberlite pipes of the Zolotitsa field and the V. Grib pipe, as well as the heating of the entire mantle column beneath the Pobeda and Yurasskaya pipes of the Kepino field.

To improve our understanding of the thermal regime and the thickness of the LM beneath the northern East-European Platform, we analyzed the major-element composition of clinopyroxene xenocrysts from four ADP kimberlite pipes (Arkhangelskaya, Lomonosovskaya, V. Grib and TSNIGRI-Arkhangelskaya) that have differing locations, compositions, and diamond contents. We used single-grain thermobarometry for Cr-diopsides proposed by Nimis and Taylor (2000) [2,3] and Sudholz et al. (2021) [4] in conjunction with the recommendations proposed by Nimis and Grutter (2010) [5], the protocol presented by Ziberna et al. (2016) [6], and the lithosphere heat production model described by Hasterok and Chapman (2011) [33]. To revisit and understand the thermal state of the lithospheric mantle beneath these regions, we integrated our new P–T dataset with previously published P–T data for mantle peridotites and Cr-diopside xenocrysts from the ADP kimberlites. We also used available data on Ni content in mantle garnet xenocrysts from the studied pipes [30,34,35,36] to calculate T values [37] and obtain P parameters by projecting the Ni-in-garnet temperatures onto reconstructed local geotherms [33]. This approach allows us to comprehensively compare the architecture of the LM (peridotitic section) in different parts of the ADP and trace its differences beneath high-, moderate-, and low-diamond-bearing kimberlite pipes. These reconstructions are more relevant than ever because the search for diamonds within the ADP requires a significant amount of information about the underlying lithospheric mantle to determine whether it has diamond-bearing potential.

2. Geological Background

The Devonian–Upper Carboniferous ADP (410–340 Ma; [11,12,13,14]) is located in the northern region of the East-European Platform and is confined to the possible south continuation of the Kola craton [38,39]. The ADP is conventionally divided into six magmatic fields (Figure 1) represented by different types of ultramafic (e.g., kimberlite, olivine melilite/picrite, and carbonatite) and mafic (basalt) rocks in the form of pipes, dykes, and sills [19]. The kimberlite pipes are located within the Zolotitsa (including the M. V. Lomonosov diamond deposit), Kepino, and Verhotina (including the V. Grib diamond deposit) fields (Figure 1). These pipes intrude into Vendian sediments and are overlain by middle Carboniferous to Permian terrigenous and carbonate rocks as well as unconsolidated Cenozoic sediments with a total thickness ranging from 20 to >100 m [12].

The Arkhangelskaya and Lomonosovskaya kimberlite pipes are located within the Zolotitsa field about 8.5 km apart and are a part of the M. V. Lomonosov diamond deposit. The peak diamond content in the Arkhangelskaya (highly diamondiferous) and Lomonosovskaya (moderately diamondiferous) kimberlite pipes have been assessed to be 1.35 car/t and 0.67 car/t, respectively [40]. The kimberlites in both pipes exhibit low contents of TiO₂ (<1 wt.%), light rare earth elements (L-REE) and high-field-strength elements (HFSE). Their Sr-Nd isotope compositions are characterized by negative ε_Nd values (from −2.2 to −5.3) and an initial ⁸⁷Sr/⁸⁶Sr(t) isotope ratio ranging from 0.70362 to 0.70662, which is evidence of an ancient enriched mantle source (EM1) [12,41,42,43]. The heavy mineral fraction of both kimberlites possess a relatively low amount of kimberlite indicator minerals (KIMs) [44] and was dominated by Cr-spinel (up to 410 g/t), minor Cr-diopside (up to 90 g/t) and pyrope (up to 55 g/t; [40]). However, Mg-ilmenite was rare [30] or even absent [40,44] compared to kimberlites in the Kepino and Verhotina fields. The age of the kimberlite was determined by the K-Ar dating of a whole-rock sample from the Lomosovskaya pipe (355 ± 10 Ma; [45]); thus far, no geochronological data are available on the age of the Arkhangelskaya pipe. The ages of the kimberlite emplacements in other pipes in the Zolotitsa field are 380 ± 10 Ma for the Pionerskaya pipe (Rb-Sr whole rock-phlogopite-groundmass isochron [13]), 380 ± 2 Ma and 375 ± 2 Ma for the Karpinskogo-1 and Karpinskogo-2 pipes, respectively (Rb-Sr dating of phlogopite and a whole-rock sample [14]).

The V. Grib kimberlite pipe is located within the Verhotina field, which is ~21 km northeast of the Zolotitsa field (Figure 1). The peak diamond content of the pipe is 1.27 car/t (highly diamondiferous; [40]). Kimberlites of the V. Grib pipe have a moderate TiO₂ content (1–2.5 wt.%) and a relatively moderate LREE and HFSE content that falls between the low- and high-TiO₂ kimberlites of the Zolotitsa and Kepino fields, respectively [46]. The V. Grib kimberlites have ε_Nd of −1.0 to + 1.5 and ⁸⁷Sr/⁸⁶Sr(t) ratios ranging from 0.70425 to 0.70648 [46], which provide evidence for the presence of diluted melts generated from an enriched asthenospheric mantle source within the LM. Consequently, the isotopic composition of the V. Grib kimberlites is relatively similar to the Bulk Earth. The V. Grib kimberlites are also very rich in KIMs [40], and are dominated by Mg-ilmenite (up to 49 kg/t), pyrope (up to 14 kg/t), Cr-diopside (up to 3.2 kg/t), and trace Cr-spinel (40 g/t). The emplacement of the V. Grib kimberlite was calculated to be 372 ± 8 Ma through the Rb-Sr dating of kimberlite [47] and 376 ± 3 Ma by the Rb-Sr dating of phlogopite [14].

The TSNIGRI-Arkhangelskaya kimberlite pipe is located within the Kepino field, approximately 40 km south of the V. Grib pipe and to the east of the Zolotitsa field (Figure 1). It is generally accepted that the Kepino field kimberlites are uniform in terms of bulk and isotopic compositions [11,12,41,42,43,46]: they are the most enriched in TiO₂ (>2 wt.%), LREE, and HFSE, have positive ε_Nd ranging from +2.8 to +1.2 and ⁸⁷Sr/⁸⁶Sr(t) ratios ranging from 0.70342 to 0.70518, which provide evidence of a depleted mantle source that was metasomatized prior to melting by fluids and/or melts [42]. However, the kimberlites of the TSNIGRI-Arkhangelskaya pipe differ significantly from other Kepino kimberlites by lower TiO₂ (1.3–1.8 wt.%), LREE, and HFSE contents. Indeed, TSNIGRI-Arkhangelskaya kimberlites have bulk compositions that are more similar to the V. Grib kimberlites [48,49]. The Sr-Nd isotope compositions (ε_Nd from 0 to −0.6 and ⁸⁷Sr/⁸⁶Sr(t) from 0.7068 to 0.7073) of the TSNIGRI-Arkhangelskaya kimberlites are evidence of an enriched mantle source (EM2) [48] and are relatively similar to the isotopic composition of the Nakyn field kimberlites in the Siberian craton [50]. The TSNIGRI-Arkhangelskaya kimberlite pipe is poorly diamondiferous (0.056 car/t; [51]) and contains a moderate amount of KIMs (100 g/t [49,51]) that are dominated by Mg-ilmenite and pyrope, scarce Cr-spinel, and trace Cr-diopside [49,51]. The age of this pipe has not been yet determined. Currently, the only age-related to the Kepino pipes was obtained from the adjacent No 697 kimberlite sill using the Rb-Sr isotopes of phlogopite (397 ± 2 Ma [14]). However, palaeoflora studies [52] suggest that the Kepino pipes may have been emplaced between 410 and 380 Ma.

3. Materials and Methods

Twenty kilograms of kimberlite samples from diatremes in the Arkhangelskaya, Lomonosovskaya, and V. Grib pipes were used for the analysis. In addition, core samples from a borehole on the crater of the V. Grib pipe were obtained. These samples were extracted from rocks found 60 to 174 m beneath the surface and are composed of sandstones with varying modes of magmatic material [19]. The rocks were crushed and processed via magnetic and heavy liquid separation. All visually identifiable clinopyroxene grains in the −1 + 0.5 fraction were handpicked using a binocular microscope. All selected grains were free of alteration, inclusions and compositional zoning. A total of 155 grains were obtained from the Arkhangelskaya pipe samples, 796 from the Lomonosovskaya pipe samples; and 831 and 238 from the diatreme kimberlite and crater sandstones of the V. Grib pipe samples, respectively. An additional 114 grains of clinopyroxene xenocrysts from the TSNIGRI-Arkhangelskaya pipe were kindly provided by the TSNIGRI Federal State Budgetary Institution and represent all visually identified clinopyroxene grains obtained from a heavy mineral concentrate of kimberlite samples from three boreholes between a 77 and 185 m interval beneath the surface. The handpicked clinopyroxene grains were mounted in epoxy, polished, and coated with carbon.

The major-element compositions were measured using a JEOL JXA-8100 electron probe microanalyzer (EPMA) at the Analytical Center for Multi-Element and Isotope Research at the Siberian Branch Russian Academy of Sciences. The grains were analyzed using a 20 kV accelerating voltage and a 50 nA beam current with a beam size of 1 µm [53,54]. Data were acquired for 10 s on-peak as well as for 10 s on either side of the background; ZAF correction was applied. In-house natural mineral IGM SB RAS standards [54] were used for calibration. The relative standard deviations were all within 1.5%. The detection limits of the instrument were <0.05 wt.% for all elements analyzed, including 0.01 wt.% for Cr and Mn, 0.02 wt.% for Ti and Na, and 0.05 wt.% for K.

To calculate the P–T values for single clinopyroxene grains, we applied the single-clinopyroxene thermobarometer described by Nimis and Taylor (2000) [3] in conjunction with the pressure corrections presented by Nimis et al. (2020) [2] for grains equilibrated at >50 kbar (hereafter NT00 [2,3]) as well as the recalibrated version of this correction by Sudholz et al. (2021) (hereafter NT00/SUD21 [4]). Compositional filter criteria outlined by Nimis and Grutter (2010) [5] and Ziberna et al. (2016) [6] were applied before the P–T calculations. The lithospheric heat production model proposed by Hasterok and Chapman (2011) [33] was used. The best-fit geotherm was determined using the Gtherm software (https://gtherm.ru/index.html; “URL” accessed on 20 August 2024). The MATLAB (7.0 version) code for the geotherms was kindly provided by Dr. Derrick Hasterok, the author of the model. This program estimates the heat flow by optimizing the deviation of the measured temperature and pressure values (obtained from xenoliths and xenocrysts) from those calculated based on several initial heat flow values (e.g., 35, 35.2, 35.4, …, 40 mW/m²), approximating the intermediate values of the deviations using third-order splines, and applying “Golden Section”—a one-dimensional optimization method—to the obtained spline. This optimization allows for the calculation of the geotherm based on the optimal heat flow. The spline approximation of misfit represents the assessment of the deviation between the xenoliths and xenocrysts measured (obtained from the thermobarometer) and the calculated P–T approximations. Importantly, the misfit for the optimal geotherm to the input P–T array was calculated using a root mean square distribution of ΔT from the calculated geotherm line. The lower boundary of the lithospheric mantle (lithosphere-asthenosphere boundary; LAB) was estimated by finding the intersection of the calculated conductive geotherm with the mantle adiabat with a surface temperature of 1315 °C. A visualization of the geotherms was obtained using depth–temperature relationships as well as the following pressure–depth formula: depth (km) = 30.4 × P (GPa) + 6.3 [55]. It should be noted that the best-fit calculated geotherms obtained by the Gtherm software were within 0.5–0.6 mW/m² of the geotherms obtained by the PT_Gtrm program proposed by Dr. D. Dolivo-Dobrovolsky (http://www.dimadd.ru/ru/Programs/ptgtrm; “URL” accessed on 20 August 2024), which was based on the approximation of the conductive geotherms proposed by Hasterok and Chapman (2011) [33].

4. Results

The major-element composition of the clinopyroxene xenocrysts is presented in Table S1. Calculated P–T values for clinopyroxene xenocrysts are presented in Table S2. The results of the P–T calculations are also summarized in

Table 1. Calculated pressure–temperature (P–T) values for Cr-diopside xenocrysts from the Arkhangelskaya, Lomonosovskaya, TSNIGRI-Arkhangelskaya, and V. Grib kimberlite pipes as well as the V. Grib garnet peridotites, garnet-clinopyroxene intergrowths, and clinopyroxene megacrysts.

P-T calculation according to NT00
Object of study	matching geotherms (mW/m2)	depth range of sampling (km)	width of sampling (km)	best-fit geotherm (mW/m2)	missfit (±ΔT °)	LAB (km)
Arkhangelskaya (this study) *	34–37	120–187	67	35.3	48	257
Arkhangelskaya [30] *	34–37	117–184	67	35.8	40	247
Arkhangelskaya all data	34–37	117–187	70	35.7	45	251
Lomonosovskaya	34–37	128–214	86	35.3	51	259
TSNIGRI-Arkhangelskaya	37–40	157–196	39	38.1	46	211
V. Grib crater part	35–39	96–197	101	36.4	46	236
V. Grib diatreme part	34–38	93–211	118	35.7	42	250
V. Grib crater + diatrem parts	34–39	93–211	118	36.0	46	243
P-T calculation according to NT00/SUD21
Object of study	matching geotherms (mW/m2)	depth range of sampling (km)	width of sampling (km)	best-fit geotherm (mW/m2)	missfit (±ΔT °)	LAB (km)
Arkhangelskaya (this study) *	33–36	129–196	67	34.3	53	278
Arkhangelskaya [30] *	34–37	135–190	55	34.7	50	271
Arkhangelskaya all data	33–37	129–196	67	34.6	51	274
Lomonosovskaya	33–36	131–221	90	34.6	53	273
TSNIGRI-Arkhangelskaya	36–39	186–203	17	37.2	48	224
V. Grib crater part	33–39	106–211	105	35.2	57	261
V. Grib diatreme part	33–37	106–223	117	34.6	44	275
V. Grib crater + diatreme parts	33–39	106–223	117	34.8	57	268
P-T calculations for the V. Grib pipe garnet peridotites
Data source	matching geotherms (mW/m2)	depth range of sampling (km)	width of sampling (km)	best-fit geotherm (mW/m2)	missfit (±ΔT °)	LAB (km)
data from [21]	36–38	76–161	85	36.7	25	231
data from [21,24]	35–40	76–216	140	36.8	47	230
P-T calculations for the V. Grib pipe garnet peridotites, cpx xenocrysts, cpx-grt intergrowths and cpx megacrysts
Calculation method	matching geotherms (mW/m2)	depth range of sampling (km)	width of sampling (km)	best-fit geotherm (mW/m2)	missfit (±ΔT °)	LAB (km)
P-T calculation according to NT00	33–40	76–216	140	36.2	52	241
P-T calculation according to NT00/SUD21	33–40	76–223	147	35.1	64	262

* without Cr-diopsides originated from spinel peridotites.

.

4.1. Arkhangelskaya Kimberlite Pipe

About 99% of the clinopyroxene xenocrysts (153 grains) fall into the low-Al “on-craton garnet peridotite” field on the Cr₂O₃/Al₂O₃ classification diagram (Figure 2a) [56]. These clinopyroxenes are primarily diopsides (Wo_42.2–48.7En_48.6–52.8Fs_2.6–6.4) with Cr₂O₃ content varying between 0.53 and 3.65 wt.% and an Mg# (Mg/Mg + Fe molar ratio) ranging from 0.89 to 0.95. After applying the compositional filter criteria outlined in [3,5,6], 14 out of 153 “on-craton garnet peridotite” Cr-diopsides were found to be suitable for P–T calculations (9% of the total number of selected grains). The P–T values calculated based on the NT00 thermobarometer revealed a range of T values from 676 to 1064 °C and P values from 2.0 to 6.0 GPa, which is consistent with a conductive geotherm of 34–37 mW/m² [33]. The best-fit geotherm was found to be 35.3 mW/m² (±ΔT 48°). The P values calculated using the recalibrated version of the NT00 Cr-in-clinopyroxene barometer proposed by Sudholz et al. (2021) (NT00/SUD21; [4]) varied from 2.3 to 6.3 GPa; the reconstructed paleogeotherm is slightly colder and more consistent with the 33–36 mW/m² geotherms of [33], while the best-fit geotherm was defined as 34.3 mW/m² (±ΔT 53°). The range of calculated depths sampled by the majority of the grains ranges from ~120 km to ~190 km according to NT00 and from ~129 km to ~196 km according to NT00/SUD21. Two grains appear to be sourced from a “shallow” sampling depth of ~70 km (NT00) or ~80 km (NT00/SUD21) within the graphite stability field. Furthermore, the calculated P–T values of these grains match the “hotter” geotherm with a heat flow of >40 mW/m².

We also applied the compositional filter criteria [3,5,6] to a clinopyroxene xenocryst dataset (238 analyses) from the Arkhangelskaya pipe published by Lehtonen et al. (2009) [30]. All grains fell within the low-Al “on-craton garnet peridotite” field based on the Cr₂O₃/Al₂O₃ classification diagram [56]; these grains are dominated by Cr-diopsides with a Cr₂O₃ content of 0.57–3.84 wt.% and an Mg# of 0.91–0.96 [30]. Out of 238 analyses, only 23 grains (9.3%) passed the compositional filter criteria [3,5,6]. The calculated P–T values for these Cr-diopsides based on NT00 ranged between 626 and 1168 °C and 1.8 and 5.9 GPa, which is consistent with the 34–37 mW/m² geotherms [33] and with the best-fit calculated heat flow of 35.8 mW/m² (±ΔT 40°). The P values calculated according to NT00/SUD21 ranged between 2.2 and 6.1 GPa, which are still consistent with the 34–37 mW/m² as well as a best-fit calculated heat flow of 34.7 mW/m² (±ΔT 50°). The ranges of the calculated sampling depths of these Cr-diopsides presented by [30] are similar to those obtained from our dataset of clinopyroxene xenocrysts: ranging from ~117 km to ~184 km based on NT00 and from ~135 km to ~190 km based on NT00/SUD21 for most of the grains. Five grains appear to be sourced from a “shallow” sampling depth within the graphite stability field of ~61–70 km based on NT00 and ~73–85 km based on NT00/SUD21. Like our dataset, the calculated P–T values for these grains were consistent with the heat flows of the “hotter” geotherm, which were approximately or greater than 40 mW/m² (Figure 3). These Cr-diopsides could have originated from spinel peridotites rather than garnet peridotites and may have erroneously survived compositional filtering [2]; consequently, their P–T estimates were considered to be meaningless and were not used in further discussion. Among the grains with P values between 5.5 and 6.0 GPa, two grains from the dataset published by [30] exhibited relatively high T values compared to the other grains (1130 and 1170 °C); their P–T values were more consistent with the ~37 mW/m² geotherm [33].

By combining our data as well as the dataset from [30], the clinopyroxene-derived geotherm for the LM beneath the Arkhangelskaya kimberlite pipe can be modelled as ranging between the 34 and 37 mW/m² heat flows of [33] with a best-fit calculated geotherm of 35.7 mW/m² (±ΔT 45°) according to NT00 and the 33 and 37 mW/m² heat flows of [33] with a best-fit calculated geotherm of 34.6 mW/m² (±ΔT 51°) according to NT00/SUD21. The maximum calculated P values for Cr-diopside ranged between 5.8 and 6.0 GPa (NT00) and 6.0–6.3 GPa (NT00/SUD21), corresponding to a depth of ~182–188 km and ~188–196 km, respectively. No clinopyroxene grains appeared to have been entrained by kimberlite from the 90–110 km depth range. The LAB is expected to lie at depths of ~251 km and ~274 km according to NT00 and NT00/SUD21, respectively.

4.2. Lomonosovskaya Kimberlite Pipe

99.4% (791 grains) of clinopyroxene xenocrysts from this pipe fall directly into the low-Al “on-craton garnet peridotite” field on the Cr₂O₃/Al₂O₃ classification diagram (Figure 2b) [56]. These clinopyroxenes are diopsides (Wo_41–50.1En_42.5–54.6Fs_3–7.3) with Cr₂O₃ content ranging from 0.89 to 4.67 wt.% and an Mg# of 0.89–0.94. Only 75 of the 791 grains (9.4% of selected grains) passed the compositional filter criteria [3,5,6]. The calculated P–T values for these Cr-diopsides (Figure 4a,b) ranged between 897 and 1262 °C and 4.0 and 6.9 kbar (NT00), consistent with the 34–36 mW/m² geotherms [33]. The best-fit geotherm for these samples was 35.3 mW/m² (±ΔT 51°). Two grains were consistent with “hotter” geotherms of 38–40 mW/m² [33]. The P values calculated according to NT00/SUD21 are in the range of 4.1–7.1 GPa, and consistent with the 33–36 mW/m² geotherms [33]; the best-fit geotherm is 34.6 mW/m² (±ΔT 53°). The maximum calculated P values for Cr-diopside are ~6.7–6.9 GPa according to NT00 and ~6.8–7.1 GPa according to NT00/SUD21, which correspond to depths of ~210–214 km and ~213–221 km, respectively. All clinopyroxene grains (except for a single grain) were entrained by kimberlite from the depths of the diamond stability field. The LAB is expected to lie at depths of ~259 km (NT00) and ~273 km (NT00/SUD21).

4.3. TSNIGRI-Arkhangelskaya Kimberlite Pipe

Only 35% of clinopyroxene grains fell into the low-Al “on-craton garnet peridotite” field on the Cr₂O₃/Al₂O₃ classification diagram (Figure 2c) [56]. The majority of these clinopyroxenes are typical diopsides (Wo_42–48.8En_47.9–51.4Fs_2.7–4.4) with Cr₂O₃ content ranging from 1.3 to 3.7 wt.% and an Mg# of 0.92–0.95. The other clinopyroxenes possessed a higher En component (Wo_36.1–40.9En_52.9–57.8Fs_4.6–6.4), consistent with the “augite” field on the Wo-En-Fs diagram [58]. These grains have a Cr₂O₃ content of 0.61–1.57 wt.%, an Mg# of 0.89–0.92, and Ca/(Ca + Mg) < 0.5. Five clinopyroxene grains passed the compositional filter criteria [3,5,6]. The calculated P–T values for these grains (Figure 4c,d) ranged between 1144 and 1265 °C and 5.0 and 6.3 GPa (NT00), which was consistent with the 37–40 mW/m² geotherms of [33]; the best-fit geotherm was defined as 38.1 mW/m² (±ΔT 46°). The P values calculated based on NT00/SUD21 ranged between 5.9 and 6.5 GPa, which was consistent with the 36–39 mW/m² geotherms of [33] with a best-fit geotherm of 37.2 mW/m² (±ΔT 48°). The calculated P values suggest that these grains were entrained by kimberlite at depths ranging from ~157 to 196 km (NT00) and ~186 to 203 km (NT00/SUD21). The LAB is expected to lie at depths of ~211 (NT00) and ~224 km (NT00/SUD21).

Figure 4. P–T plot of the mantle-derived clinopyroxene xenocrysts from the (a,b) Lomonosovskaya and (c,d) TSNIGRI-Arkhangelskaya kimberlite pipes. P–T calculations according to (a,c) NT00 [2,3] and (b,d) NT00/SUD21 [4]. Conductive model geotherms are based on [33]. The light grey adiabat is represented by a potential temperature of 1300 °C with a gradient of 0.3 °C/km [33]. The diamond (D)–graphite (G) transitional curve follows the findings of [57]. The black line presents the best-fit calculated geotherm; the misfit for each calculation can be found in Table 1. The black dotted line presents the LAB taken from the intersection point between the adiabat and best-fit calculated geotherm. N refers to the number of grains that passed the compositional filter criteria [3,5,6].

Figure 4. P–T plot of the mantle-derived clinopyroxene xenocrysts from the (a,b) Lomonosovskaya and (c,d) TSNIGRI-Arkhangelskaya kimberlite pipes. P–T calculations according to (a,c) NT00 [2,3] and (b,d) NT00/SUD21 [4]. Conductive model geotherms are based on [33]. The light grey adiabat is represented by a potential temperature of 1300 °C with a gradient of 0.3 °C/km [33]. The diamond (D)–graphite (G) transitional curve follows the findings of [57]. The black line presents the best-fit calculated geotherm; the misfit for each calculation can be found in Table 1. The black dotted line presents the LAB taken from the intersection point between the adiabat and best-fit calculated geotherm. N refers to the number of grains that passed the compositional filter criteria [3,5,6].

4.4. V. Grib Kimberlite Pipe

4.4.1. Crater Part

A total of 91.6% (218 grains) of the 238 clinopyroxene xenocrysts fell within the low-Al “on-craton garnet peridotite” field (Figure 2d; [56]). A majority of these samples (200 grains) were diopsides (Wo_41–49En_48.3–54.2Fs_2.2–7) with Cr₂O₃ content ranging from 0.51 to 3.23 wt.% and an Mg# of 0.88–0.96. The other clinopyroxenes possessed a higher En component (Wo_39.7–41.0En_52.2–54.2Fs_5.3–7.3) and fell within the “augite” field on the Wo-En-Fs diagram [58]. These grains had Cr₂O₃ contents of 0.58–1.04 wt.%, an Mg# of 0.88–0.91, and Ca/(Ca + Mg) < 0.5. 51 of the 218 xenocrysts (21.4% of selected grains) passed the compositional filter criteria [3,5,6]. The calculated P–T values for these Cr-diopsides (Figure 5a,b) ranged between 668 and 1170° C and 3.0 and 6.3 GPa (NT00), consistent with the 35–38 mW/m² geotherms of [33]; it should be noted that two grains were more consistent with a 39–40 mW/m² geotherm. The best-fit geotherm was found to be 36.4 mW/m² (±ΔT 46°). The P values calculated according to NT00/SUD21 are within the range of 3.3–6.7 GPa, which was consistent with 33–39 mW/m² geotherms [33] and a best-fit geotherm of 35.2 mW/m² (±ΔT 57°). The range of clinopyroxene sampling depths ranged from ~96 to 197 km (NT00) and ~106–211 km (NT00/SUD21). The LAB is expected to lie at depths of ~236 km and ~261 km according to NT00 and NT00/SUD21, respectively.

4.4.2. Diatreme Part

A total of 715 of the 831 clinopyroxene xenocrysts (86%) fell within the low-Al “on-craton garnet peridotite” field (Figure 2e; [56]). A majority of these clinopyroxenes (696 grains) were diopsides (Wo_41.0–49.2En_46.9–54.9Fs_2.5–8.1) with a Cr₂O₃ content of 0.51–3.76 and an Mg# of 0.86–0.95. The remaining 19 clinopyroxene grains exhibited elevated En contents (Wo_37.5–40.1En_52.8–55.5Fs_5.5–8.9) and lower Mg# (0.86–0.91), consistent with the “augite” field [58]. Only 59 of the 715 clinopyroxene xenocrysts from the “on-craton garnet peridotite” field (7% of selected grains) passed the compositional filter criteria [3,5,6]. The calculated P–T values for these clinopyroxenes (Figure 5c,d) ranged between 710 and 1178 °C and 2.9 and 6.7 GPa (NT00), consistent with the 34–38 mW/m² geotherms of [33] and a best-fit geotherm of 35.7 mW/m² (±ΔT 42°) The P values calculated according to NT00/SUD21 are within the range of 3.3–7.1 GPa, and consistent with 33–37 mW/m² geotherms [33] and a best-fit geotherm of 34.6 mW/m² (±ΔT 44°). The range of clinopyroxene sampling depths ranged from ~93 to 211 km (NT00) and ~106 to 223 km (NT00/SUD21). The LAB is expected to lie at depths of ~250 km and ~275 km according to NT00 and NT00/SUD21, respectively.

Combining the P–T data of clinopyroxenes from the crater and the diatreme (Figure 5e,f) obtained from [2,3] and [4], the main features of the calculations are: the range of sampling depths varies from ~93 to 211 km (NT00) and ~106 to 223 km (NT00/SUD21), the best-fit geotherm is 36.0 mW/m² (±ΔT 46°) (NT00) and 34.8 mW/m² (±ΔT 57°) (NT00/SUD21), and the LAB is expected to lie at the depths of ~243 km (NT00) and ~268 km (NT00/SUD21).

4.4.3. Garnet Peridotites, Garnet-Clinopyroxene Intergrowths and Cr-Rich Clinopyroxene Megacrysts

We used readily available P–T values and major-element composition data on garnet peridotites [21,24,25], garnet-clinopyroxene intergrowths, and Cr-rich clinopyroxene megacrysts [59] to create a more robust reconstruction of the thermal state of the LM beneath the V. Grib pipe.

Previously published data on the P–T values of garnet peridotites [21,24,25] showed that coarse-granular peridotites yielded T values ranging between 639 and 1070 °C and P values ranging between 2.3 and 5.1 GPa, which corresponds to a depth interval ranging from ~76 to 161 km. The P–T values from [21] were consistent with the 36–38 mW/m² geotherms of [33] with a best-fit calculated heat flow of 36.7 mW/m² (±ΔT 25°) (Figure 6a). The P–T data of four peridotites from [24] were consistent with the 35–40 mW/m² geotherm (Figure 6a). The calculated P–T value of a single sheared peridotite sample [25] was 1221 °C and 6.9 GPa, consistent with the 35.5 mW/m² geotherm, suggesting that it sampled the deeper regions of the LM at ~216 km. Combining the data from [21,24,25], the best-fit calculated heat flow was 36.8 mW/m² (±ΔT 47°) with an expected LAB depth of ~230 km (Figure 6a).

We also applied compositional filter criteria [3,5,6] to the dataset published by [59] and identified nine clinopyroxene grains in intergrowths with garnet and five grains of Cr-rich clinopyroxene megacrysts that satisfied the criteria. The calculated P–T values for these clinopyroxene intergrowths with garnet ranged between 796 and 1054 °C and 3.4 and 5.7 GPa (NT00), consistent with the 35–39 mW/m² geotherms of [33]. The P values calculated according to NT00/SUD21 were consistent with 34–38 mW/m² geotherms [33]. The estimated P–T values for Cr-rich clinopyroxene megacrysts ranged between 649 and 1119° C and 3.3 and 5.9 GPa (NT00) and 655 and 1129° C and 3.7–6.4 GPa (NT00/SUD21). These P–T values were consistent with a wide range of fitted geotherms (33 to >40 mW/m²) [2,3,4].

Figure 6b,c show the relative position of the P–T values of all studied clinopyroxene xenocrysts, as well as the garnet peridotites, clinopyroxene-garnet intergrowths, and clinopyroxene megacrysts from the V. Grib pipe on fitted geotherms. The mantle sampling depths were found to be between ~76 and 216 km (Figure 6b), the best-fit geotherm was 36.2 mW/m² (±ΔT 52°) and the LAB is expected to lie at a depth of ~241 km (NT00). In the case where P–T calculations were obtained using NT00/SUD21, the range of mantle sampling depths was nearly the same (from ~76 to 223 km), but the best-fit geotherm was 35.1 mW/m² (±ΔT 64°) with a deeper expected LAB at ~262 km (Figure 6c).

5. Discussion

5.1. Comparison of the P–T Calculation Results for Single, Mantle-Derived, Clinopyroxene Xenocrysts

The NT00 geothermobarometer presented by Nimis and Taylor (2000) [3] based on the Cr-in-clinopyroxene barometer and the enstatite-in-clinopyroxene thermometer has been an effective tool for the reconstruction of P–T conditions and mantle lithospheric geotherms for over 20 years. The fact that P–T conditions can be directly retrieved from single clinopyroxene grains is especially important when working with highly altered mantle xenoliths, diamond inclusions, and grains recovered from samples collected from modern river and stream sediments during diamond exploration programs [3]. The Cr-in-clinopyroxene barometer has been calibrated in experimental studies [60,61,62] that cover a wide range of possible natural lherzolite compositions over a P–T range of 850–1500 °C and up to 6.0 GPa [3]. However, it has been shown that NT00 progressively and systematically underestimates P for samples with P values > 4 GPa [4,6,63,64]. In particular, Sudholz et al. (2021) [4] noted that the majority of the experiments used in the NT00 calibration were conducted at P < 4 GPa and the reliability of the NT00 application was still relatively high between 2.0 and 4.0 GPa. Recent empirical corrections of NT00 have been proposed [2] as an interim measure by recalibrating the original NT00 geobarometer against pressures obtained by the Al-in-orthopyroxene geobarometer of Nickel and Green (1985) [65] modified by Carswell (1991) [66]. According to [4], this correction results in an overall improved agreement with Al-in-orthopyroxene P estimates but still underestimates pressures above 4.5 GPa.

Sudholz et al. (2021) [4] proposed an experimental recalibration of the Cr-in-clinopyroxene geobarometer to improve its precision and reliability above 4.5 GPa based on a set of new experiments conducted at 3.0–7.0 GPa and 1100–1400 °C in conjunction with data obtained from [61,62,67]. The application of this geobarometer together with the NT00 geothermometer on a set of clinopyroxene xenocrysts from the Diavik-Ekati kimberlite pipe (Canada) and Argyle lamproite pipe (Australia) shows that the estimates using the NT00 calibration thermobarometer suggested shallower paleogeotherms that intersected with the isentrope at shallower depths [4]. In the case of Diavik-Ekati, the LAB was expected to be 25–30 km deeper when calculated using NT00/SUD21 compared to the NT00 calibration, which was more consistent with data obtained from mantle xenoliths and geophysical studies. Shaikh et al. (2024) [68] showed that the application of the NT00/SUD21 geobarometer on a set of clinopyroxene xenocrysts from the kimberlites of the Nxau Nxau cluster in Botswana gave P values that were 0.3–0.4 GPa higher relative to the NT00 calibration, yielding “colder” geotherms of 35–37 mW/m² [33] compared to the NT00 geothermobarometer (37–38 mW/m²).

The calculated P–T values of the clinopyroxene xenocrysts analyzed in this study showed that the use of the NT00/SUD21 thermobarometer yielded “colder” geotherms (the best-fit calculated geotherms were lower by 0.7–1.2 mW/m²) that resulted in a deeper expected LAB by approximately 13–25 km compared to the NT00 calibration (Table 1). The misfit of the best-fit geotherms was lower in the NT00 version compared to the NT00/SUD21. The widths of clinopyroxene sampling were comparable between the two P–T calculation versions (±4 km) for the Arkhangelskaya, Lomonosovskaya, and V. Grib kimberlite pipes, but differed significantly for the TSNIGRI–Arkhangelskaya pipe (39 km based on NT00 compared to 17 km based on NT00/SUD21). The shallowest LM boundary according to the NT00/SUD21 was deeper than its NT00 counterpart by 3 km (Lomonosovskaya pipe), 10–13 km (Arkhangelskaya and V. Grib pipes) and 29 km for the TSNIGRI–Arkhangelskaya pipe; and lowest boundary of the LM was deeper by 7–14 km for all pipes. The differences in the P values between the NT00 and NT00/SUD21 versions were at all calculated P ranges, both below and above 4.0 GPa (Figure 7a), and exhibit a negative correlation with Cr#^cpx and α_CaCrTs^cpx and a positive correlation with Al^cpx (Figure 7). A comparison of P values derived from the V. Grib garnet peridotites [21] using the combination of TA98/NG95 [62,65] with those obtained from NT00 and NT00/SUD21 versions using clinopyroxenes in these samples is presented in Figure 8. The calculated P values obtained from NT00 and NG95 were within ±0.3 GPa for 8 out of 12 samples, ±0.4–0.5 GPa for three samples, and +0.7 GPa for one sample (Figure 8a). In contrast, the calculated P values obtained from NT00/SUD21 and NG95 were within ±0.3 GPa for four samples (Figure 8b), and the remaining P values were higher by 0.3–0.6 GPa (six samples) and 0.8–1.1 GPa (two samples). The expected depth of the LAB beneath the V. Grib pipe, calculated from the intersection of the best-fit xenolith-derived geotherm (±25 ΔT°) with the mantle adiabat, was 231 km (Figure 6a) compared to the 241 km and 262 km of NT00 (Figure 6b) and NT00/SUD21 (Figure 6c), respectively (both depths were obtained from a combination of data on garnet peridotites, clinopyroxene xenocrysts, garnet-clinopyroxene intergrowths and Cr-rich clinopyroxene megacrysts). Based on the present-day global S-wave tomography [69], high-velocity anomalies have been recognized beneath the northern East-European Platform in the uppermost 200 km, which is evidence of a thick lithospheric mantle in the region at the present day. No high-velocity seismic anomalies have been observed beneath 260 km in this region [69]. Several seismic datasets and thermal models [70,71] also suggest the existence of an upper mantle down to approximately 200–250 km in this region. If so, the expected LAB depth of >260 km beneath the Arkhangleskaya, Lomonosovskaya, and V. Grib pipes as suggested by the NT00/SUD21 recalibration appears to be overestimated. Thus, considering the more reasonable P–T values obtained by the NT00 version compared to the NT00/SUD21 recalibration based on data from mantle xenoliths and geophysical studies, we will base the following discussion on the NT00 data only.

5.2. The Efficiency of Mantle Sampling with Depth

To assess the efficiency of mantle sampling at different depths as well as to ensure a more robust reconstruction of the architecture of the LM beneath the studied pipes, we used the available data on Ni content in garnet xenocrysts of lherzolite and harzburgite paragenesis [30,34,35,36,72] to calculate the T parameters [37] and further project T values onto the best-fit local cpx-derived geotherm. The results of this analysis are presented in Figure 9.

The thickest LM was evaluated to be beneath the Arkhangelskaya and Lomonosovskaya kimberlite pipes of the Zolotitsa field, where the LAB is estimated to lie at depths of ~251 km and ~259 km, respectively. Based on these LAB values, the expected width of the “diamond window” (expected “diamond window” (E “DW” in Figure 9), i.e., the depth range between the graphite–diamond transition to the LAB under reconstructed local geothermal conditions) is ~131 km and ~139 km, respectively. However, the deepest garnet and Cr-diopside xenocrysts from the Arkhangelskaya pipe are indicative of the depths of ~208 km and ~190 km, respectively, which limit the width of the real “diamond window” (R “DW”, i.e., the depth range between the graphite–diamond transition to the maximum P values found in garnet and clinopyroxene xenocrysts) to ~88 km (Figure 9a). The deepest garnet and Cr-diopside xenocrysts from the Lomonosovskaya pipe are indicative of the same depth of ~220 km and limit the width of the real “diamond window” to ~100 km (Figure 9b). The reconstruction of the vertical distribution of garnet and Cr-diopside xenocrysts showed that the most efficient mantle sampling was at an LM depth range of ~120–165 km beneath the Arkhangelskaya pipe (Figure 9a) and ~170–190 km beneath the Lomonosovskaya pipe (Figure 9b). Beneath the Arkhangelskaya pipe, mantle sampling depths between ~110 and 190 km were marked by both garnet and Cr-diopside xenocrysts, while mantle sampling depths between ~80 and 110 and ~190 and 210 were marked only by garnet xenocrysts. At depths between ~170 and 190 km, Cr-diopside xenocrysts prevail over the garnet ones (Figure 9a). The peaks in the vertical distribution of garnet and Cr-diopside xenocrysts from the Lomonosovskaya pipe mostly do not coincide with each other: ~145–190 km for the garnet xenocrysts and ~170–210 km for the Cr-diopside xenocrysts (Figure 9b). The LM depth ranges of ~120–140 km and ~210–220 km beneath the Lomonosovskaya pipe are poorly sampled, with no xenocrysts marking the shallow part of the LM (<120 km).

The thickness of the LM beneath the V. Grib pipe is 10–18 km less than its thickness beneath the Arkhangelskaya and Lomonosovskaya pipes; the LAB is expected to be at ~241 km and the expected width of the “diamond window” is ~116 km. The deepest garnet and Cr-diopside xenocrysts are sourced from LM depths of ~220 km, consistent with the depth value of the single sheared lherzolite sample and limiting the width of the real “diamond window” to ~95 km (Figure 9c). The reconstruction of the vertical distribution of garnet and Cr-diopside xenocrysts showed that the entire section of the LM between ~90 and 220 km was continuously sampled: the most efficient mantle sampling was found to be at the ~135–180 km LM depth range. The major peak in Cr-diopside in terms of depth distribution was at ~110–120 km, which is within the depth range of the most efficient peridotite sampling. The ~75–90 km depth range of the LM is only marked by peridotite xenoliths, and the deepest parts of the LM (~200–220 km) have been poorly sampled (Figure 9c).

The thickness of the LM beneath the TSNIGRI-Arkhangelskaya pipe was the smallest compared to the other studied pipes: ~40–48 km less than the Arkhangelskaya and Lomonosovskaya pipes and ~30 km less than the V. Grib pipe. The expected width of the “diamond window” is ~76 km (Figure 9d). The deepest garnet and Cr-diopside xenocrysts are indicative of LM depths of ~200 km, limiting the width of the real “diamond window” to ~65 km. The most efficient mantle sampling was at an LM depth range of ~160–200 km (Figure 9d). The LM depth range between ~140 and 160 km was poorly sampled, with no xenocrysts sourced from the shallow parts (<140 km) of the LM.

5.3. The Thermal State and Thickness of the Lithospheric Mantle and Their Relationship with the Diamond Content of Kimberlite Pipes

Numerous studies on mineral inclusions in diamonds [73,74,75,76,77,78,79,80,81,82] have shown that their source rocks are primarily peridotites (both harzburgites and lherzolites) and eclogites, with rare websterites occurring in the deeper (>130–140 km) parts of the thick (up to 230 km), predominantly cool LM [81]. Such strong, thick mantle roots are known to underlie ancient cratons, the oldest parts of the continents, from which almost all diamonds are produced [39].

The paleogeotherms calculated from Cr-diopside geothermobarometry revealed the presence of a >200 km-thick LM in the northern part of the East-European platform within the ADP (a possible southern extension of the Kola craton [38,39]) that is consistent with typical LMs beneath ancient cratons [1,2,3,6,68] including diamond-producing regions [39,83]. The presence of thick LM is certainly a positive sign for the diamond potential of a region, but it is not an absolute indicator of high diamond grades in kimberlites that sample this LM. For example, the thickness of the LM beneath kimberlite pipes in Finland (Karelian craton) was estimated to be >200 km [8,9,15,16,17,18], but all the known pipes in the region are poorly diamondiferous or diamond-free [9]. The thickness of the LM underlying the south-eastern and south-western parts of the Kola craton within the eastern Fennoscandian shield is also expected to be relatively high (>190 km) [84], but no kimberlites have been discovered there. Furthermore, the only known kimberlite pipe in the southern part of the Kola region (Ermakovskaya) is poorly diamondiferous, and the thickness of the LM beneath this pipe is estimated to be no more than 150 km [84].

P–T data from Cr-diopside have revealed variations in the thickness of the LM beneath the studied ADP kimberlite pipes (Figure 9). The LM beneath the poorly diamondiferous TSNIGRI-Arkhangelskaya pipe (0.056 car/t) is approximately ~30–48 km thinner than the LM beneath the highly and moderately diamondiferous ADP kimberlite pipes: this can be considered to be the first indicator for the low diamond content of this pipe. However, there were no correlations observed between the thickness of the LM beneath other studied ADP kimberlite pipes and their diamond content: the thickest LM (LAB at ~259 km) was found beneath the moderately diamondiferous Lomonosovskaya pipe (0.67 car/t), whereas the thickness of LM beneath the highly diamondiferous Arkhangelskaya (1.35 car/t) and V. Grib (1.27 car/t) pipes was approximately 10 km and 18 km less than the LM beneath the Lomonosovskaya pipe, respectively.

Older, thicker, and more stable cratonic regions also have lower geothermal gradients [83] that are equivalent to surface heat flow models (i.e., geotherms that depict pressure–temperature relationships for thermally equilibrated lithosphere [39] up to 41 mW/m² (Figure 9) according to the reference geotherms of Hasterock and Chapman (2011) [33]). Under these conditions, diamonds can be stable in the LM over a relatively wide range of pressures and temperatures. The lower the geotherm, the thicker the LM, and the more likely there is to be a “diamond window”. As the geotherm increases, the potential range of depths for diamond stability within the LM is significantly reduced: this is exactly what is observed in the LM beneath the poorly diamondiferous TSNIGRI-Arkhangelskaya kimberlite pipe. The “hotter” geotherm (38.1 mW/m²) and the lower width of the expected “diamond window” (Figure 9) can be regarded as a second indicator for the lower diamond content of the pipe. In contrast, the geotherms of the LM beneath Arkhangelskaya, Lomonosovskaya, and V. Grib kimberlite pipes are moderately low (35.3–36.2 mW/m²) with a relatively thicker width for the expected “diamond window”. In summary, the thermal state and thickness of the LM beneath highly and moderately diamondiferous ADP kimberlite pipes favor diamond formation. The expected “diamond window” is thicker beneath the pipes of the Zolotitsa field compared to the “diamond window” beneath the V. Grib pipe, and the thickest “diamond window” was observed beneath the moderately diamondiferous Lomonosovskaya pipe. In other words, there are still no clear correlations between the thermal state and thickness of the LM beneath highly and moderately diamondiferous kimberlite pipes and their diamond content.

5.4. The Amount of “Diamond-Associated” Garnets and Their Relationship to the Diamond Grade of Kimberlite Pipes

Mantle peridotites and eclogites are considered to be the key diamond sources as determined by mineral inclusions in diamonds [79] and diamond-bearing xenoliths in kimberlite pipes [83]. Therefore, it is reasonable to expect a positive correlation between the amount of diamond present and the abundance of fragments in the diamondiferous host rocks in the LM within a single kimberlite pipe. One of the factors that controls the amount of diamond the kimberlite contains is the quantity of diamond peridotite and eclogite that it sampled as well as the average grade of said diamond peridotite and eclogite [83]. The number of diamond peridotites and eclogites in the LM sampled by a kimberlite pipe should be reflected in the amount of these xenoliths and/or disaggregated mineral grains (xenocrysts). In addition, analyses should also take into account the specific composition of minerals: the major- [85,86,87] and trace- [34,35,79,88,89,90,91,92,93] elements can be used to identify favorable conditions for the formation and preservation of diamonds in the LM.

An interpretation of the composition of the peridotitic garnet and clinopyroxene xenocrysts in the kimberlite pipes allows conclusions to be drawn about the potential diamond content of mantle peridotites, though this cannot be extended to the eclogites. However, several studies on ADP mantle xenoliths and xenocrysts [20,21,22,23,24,25,30,31,32,34,35] as well as KIMs from modern alluvial samples [38,91] showed that the LM beneath the ADP is predominantly represented by peridotites with subordinate eclogites. Mineral inclusions in the diamonds of the ADP [44,94,95,96,97], as well as the isotopic composition of the carbon in said diamonds [96,98,99], revealed the source rocks of these diamonds were dominated by peridotites. Only the Arkhangelskaya and Karpinskogo-1 kimberlite pipes in the Zolotitsa field are thought to contain a significant proportion of diamonds of eclogitic origin [44,100]. Based on these data, it can be assumed that garnet and Cr-diopside xenocrysts of peridotite origin from the ADP kimberlites should, to some extent, reflect the diamond potential of the LM and, consequently, the pipes themselves.

An important question to answer is: “Are there any correlations or relationships between the amount of diamond-associated garnets and Cr-diopsides in the heavy mineral concentrate of the ADP kimberlite pipes and their diamond content?” Previous studies [30,34,35,72] have shown that the number of pyropes of diamond-associated harzburgite-dunite paragenesis [85] (i.e., G10D pyropes [87]) was low for all studied ADP kimberlites: 1.3–1.6% for the highly diamondiferous Arkhangelskaya pipe [30,72]; 3.6% for the moderately diamondiferous Lomonosovskaya pipe [72]; 4% for the highly diamondiferous V. Grib pipe [35]; and 0.7% for the poorly diamondiferous TSNIGRI-Arkhangelskaya pipe [35], where all values presented refer to the percentage of the total selected garnets associated with mantle paragenesis from the heavy mineral fraction. Based on additional data on trace elements in garnets, the amount of high-chromium pyropes with sinusoidal chondrite-normalized [101] patterns (the most common type of garnets found as inclusions in diamond [79,80], i.e., Lz4 and Hz groups, see [35] for details) associated with lherzolite and harzburgite paragenesis is lowest in the TSNIGRI-Arkhangelskaya pipe (12%), and relatively similar for the Arkhangelskaya (30%; [30]), Lomonosovskaya (33%; [72]) and V. Grib (31% [35]) kimberlite pipes; where all values presented refer to the percentage of the total analyzed pyropes with peridotite and megacryst associations (see [35] for details). It should be noted that the number of garnet xenocrysts with a peridotite and megacryst origin from the Arkhangelskaya pipe analyzed for their trace-element composition is significantly lower (52 grains [30]) compared to the number of grains from the V. Grib (>200 grains [34,35]), TSNIGRI-Arkhangelskaya (>150 grains [35]), and Lomonosovskaya (>200 grains [72]) pipes; this may be statistically unrepresentative and require further investigation. Nevertheless, this study confirms the low diamond content of the TSNIGRI-Arkhangelskay pipe based on the lower amount of “diamond-associated” garnets. However, no clear differences in the amount of diamond-associated garnets both in terms of major- and trace-element composition compared to other diamond-bearing ADP pipes. Data on Cr-diopside xenocrysts reveal that a majority (60% of the V. Grib pipe, 94% of the Arkhangelskaya pipe, and 100% of the Lomonosovskaya pipe) of filtered grains have calculated P–T values that fall within the diamond stability field (Figure 9).

5.5. The Efficiency of Mantle Sampling and Its Relationship to the Diamond Content of Kimberlite Pipes

Reconstructing the vertical distribution of garnet and Cr-diopside xenocrysts within the LM beneath the studied ADP kimberlite pipes (Figure 9) showed that the width of the real “diamond window” marked by the mantle samples entrained in these kimberlites is narrower compared to those of the expected “diamond window” evaluated based on the depth of graphite-diamond transition to the LAB under reconstructed local geothermal conditions.

The narrowest real “diamond window” was observed in the LM beneath the poorly diamondiferous TSNIGRI-Arkhangelskaya pipe (depth ranging from ~135 to 200 km, i.e., ~65 km). However, the ~135–160 km interval is practically unmarked by xenocrysts, indicating that should expect an even smaller real “diamond window” with a thickness of just over 40 km. Considering the T_Ni values for diamond-associated garnets of the Hz and Lz4 groups (1100–1220 °C [35]), potentially diamond-bearing peridotites could be localized only within the 20 km wide interval in the LM between ~160 and 180 km. This depth range coincides exactly with the global mode of diamond depth distribution in the LM of ~175 ± 15 km [102] and ~180 ± 10 km [2]. According to [2], even limited xenocryst records from diamond-favorable depths of ~160–190 km may be evidence of the significant diamond potential of the pipe. In the case of the TSNIGRI-Arkhangelskaya pipe, despite the presence of garnet xenocrysts at diamond-favorable depths, the amount of potentially diamond-bearing peridotites in the LM could be too low and the diamond formation processes may have been limited, as concluded from the reconstruction of metasomatic processes in the LM beneath the pipe before the kimberlite emplacement [35].

The next widest real “diamond window” was identified in the LM beneath the highly diamondiferous Arkhangelskaya kimberlite pipe (depth range of ~120–208 km, i.e., ~88 km). The depth of the most efficient sampling of both garnet and Cr-diopside xenocrysts ranged from ~120 to 165 km (Figure 9), which did not match the global modes of diamond depth distributions (~160–190 km [2,102]). The projection of the T_Ni (800–1000° C) of diamond-associated garnets of harzburgite paragenesis fell within the same depth range of ~120–165 km. However, the T_Ni (880–1190° C) of high-Cr lherzolitic garnets with sinusoidal REE patterns [30], i.e., the Lz4 group according to [35], is evidence of their entrainment from a wider depth interval within the LM from ~140 to 210 km. The number of Lz-4 garnets was 3–4-fold greater than Hz garnets, and we cannot exclude the presence of potential diamond-bearing lherzolites hosted in the deep part of the LM (>165 km). Thus, the depth range of the LM between ~120 km and at least ~200 km (i.e., ~80 km) should be regarded as being capable of hosting potentially diamond-bearing peridotites that were further sampled by the kimberlite of the Arkhangelskaya pipe.

The next widest real “diamond window” was identified in the LM beneath the highly diamondiferous V. Grib kimberlite pipe (a depth range from ~125 to ~220 km, i.e., ~95 km). In contrast to the Arkhangelskaya pipe, the LM beneath the V. Grib pipe was continuously sampled by the kimberlite, especially within the diamond-favorable depth range of ~125 km to at least ~200 km (i.e., ~75 km). Although the LM was poorly sampled beyond a depth of 200 km, this is not necessarily an issue since the deeper parts of the LM are regarded as being unfavorable for diamond formation due to the negative effect of the infiltration of carbon-undersaturated melts or fluids [2]. The most efficient garnet and Cr-diopside sampling fell within the ~135–180 km depth range, while the depth range between ~160 and 180 km was sampled by garnet alone (Figure 9). These depth ranges match the global modes of diamond depth distribution (~160–190 km [2,102]). The projection of the T_Ni of diamond-associated garnets of harzburgite paragenesis (850–1150 °C) and high-Cr lherzolitic Lz-4 garnets (970–1100 °C [34,35]) revealed their entrainment from a depth range between ~130 and 190 km, with the major peak at ~160–180 that clearly matches the global modes of diamond depth distribution [2,102]. It can be concluded that the LM beneath the V. Grib kimberlite pipe has all the features of a “diamond-favorable” region. And effectively sampling these “diamond-favorable” depths by kimberlite clear correlates with high diamond content of the V. Grib pipe.

The widest real “diamond window” was identified for the LM beneath the moderately diamondiferous Lomonosovskaya pipe (depth range between ~120 and 220 km, i.e., ~100 km). However, considering the very poor sampling of the ~120–140 km depth interval of the LM by kimberlite (Figure 9) and the unfavorable conditions for diamond formation at the depths >200 km, the width of the real “diamond window” was limited to ~60 km (i.e., from ~140 to 200 km). A reconstruction of the vertical depth distribution of xenocrysts reveals the approximately uniform entrainment of Cr-diopside xenocrysts across the ~140–210 km interval and the limited entrainment of garnet xenocrysts across the depth range between ~150 and 190 km. All values of T_Ni in the diamond-associated garnets of harzburgite paragenesis and the high-Cr lherzolitic Lz-4 garnets fall between 900 and 1100 °C, which is equivalent to a 40 km wide interval in the LM between ~150 and 190 km. The correlation between the garnet sampling depth interval in the LM to the global modes of diamond depth distribution [2,102] and the presence of a high amount of Cr-diopside within this interval is consistent with the diamondiferous nature of the Lomonosovskaya pipe. However, the limited range of depths sampled by the potentially diamond-bearing garnet peridotites (~40 km) may explain its lower diamond content compared to the Arkhangelskaya (~80 km) and V. Grib (~75 km) pipes.

5.6. Unsampled Range in the Deepest Part of the Lithospheric Mantle

As was shown above, the deepest levels of the LM sampled by kimberlites differ from the expected depth of LAB obtained from the intersection of the local geotherm with the mantle adiabat (Figure 9): additional conductive material should have been present at the time of kimberlite emplacement within the deepest parts of the LM, i.e., at ~11 km (TSNIGRI-Arkhangelskaya pipe), ~21 km (V. Grib pipe), ~39 km (Lomonosovskaya pipe), and ~43 km (Arkhangelskaya pipe). Grutter (2009) [1] concluded that the thermal boundary (i.e., the LAB) must exist far below the deepest depleted peridotite in all cratonic lithospheres; this has been proven in numerous cratonic settings, and kimberlites apparently never sample deep lithospheric materials in a conductively equilibrated state [1]. Ziberna et al. (2013) [90] concluded that the maximum sampling depth reflects the depth at which the rising magma becomes fast enough to initiate the mechanical disaggregation and incorporation of xenoliths. The differences in the width of the unsampled range in the deep LM beneath the studied ADP kimberlite pipes may be caused by both the LM architecture and the source of the kimberlites.

The composition of Sr and Nd isotopes in the TSNIGRI-Arkhangelskaya kimberlites [48] suggest that their source rocks were enriched in incompatible elements relative to mid-ocean ridge basalt (MORB)-type and ocean island basalt (OIB)-type asthenospheric mantle [103,104], were stored in the lithospheric mantle, and were isolated from convective mixing for a long period of time before the generation of kimberlite magmas. The Nd_DM model ages suggest that the enrichment of the source occurred between 0.86 and 0.93 Ga (minimum age) and 1.15 and 1.51 Ga [48]. Furthermore, the LM likely experienced long-term heating, resulting in its gradual thinning and creating conditions necessary for subsequent kimberlite magmatism. The heating of the LM beneath some kimberlite pipes of the Kepino field, as recorded in the composition of the mantle xenocrysts, has also been previously proposed [30,32,88]. However, it is problematic to draw conclusions about the spatial and temporal relationships of thermal anomalies at this scale based on any tectono-thermal events due to the lack of large-scale tectono-thermal reconstructions in the region and the absence of the systematic isotopic geochronological dating of Kepino kimberlites.

The LM beneath the Arkhangelskaya, Lomonosovskaya, and V. Grib kimberlite pipes exhibited similar, moderately low geothermal gradients; in addition, there were no signs of long-term heating at the time of kimberlite emplacement. However, the difference in the width of the interval of the “missing samples” observed in the LM beneath the Arkhangelskaya and Lomonosovskaya kimberlites is almost twice as large as the interval beneath the V. Grib pipe. The Sr and Nd isotope composition of the Zolotitsa field kimberlites indicates an ancient EM1 enriched mantle source [12,41,42,43], contrasting with the V. Grib kimberlites that require a mixing of asthenospheric and lithospheric sources. Consequently, the relatively thick interval of “missing samples” in the LM beneath the Arkhangelskaya and Lomonosovskaya pipes could be explained by the location of the kimberlite melting source above the LAB as well as by the interval of the LM in which the rising magma was unable to initiate the mechanical disaggregation and incorporation of xenoliths. The latter could be due to the lower CO₂ content of the Zolotitsa kimberlites compared to the V. Grib kimberlites, which is similar to the differences proposed for group I and II kimberlites [90,105,106].

6. Conclusions

The best-fit geotherms derived from the single-grain thermobarometry of Cr-diopsides proposed by Nimis and Taylor (2000) [2,3] and the lithospheric heat production model proposed by Hasterok and Chapman (2011) [33] were found to be 35.7 mW/m² (±ΔT 45°) for the Arkhangelskaya pipe, 35.3 mW/m² (±ΔT 51°) for the Lomonosovskaya pipe, 36.2 mW/m² (±ΔT 52°) for the V. Grib pipe, and 38.1 mW/m² (±ΔT 46°) for the TSNIGRI-Arkhangelskaya pipe. The reconstructed Cr-diopside-derived geotherms indicate the presence of a >200 km-thick lithospheric mantle (LM) beneath the northern part of the East-European platform at the time of kimberlite emplacement within the Arkhangelsk diamondiferous province (ADP). The narrowest LM was found beneath the poorly diamondiferous (0.056 car/t) TSNIGRI-Arkhangelskaya pipe: the lithosphere-asthenosphere boundary (LAB) was determined to lie at approximately ~211 km, which was between 30 and 48 km shallower than the LAB beneath highly and moderately diamondiferous kimberlite pipes. However, there was no correlation between the thickness of LM beneath other studied ADP kimberlite pipes and their diamond content: the thickest LM (LAB at ~259 km) was found beneath the moderate-diamondiferous Lomonosovskaya pipe (0.67 car/t), while the thickness of the LM beneath highly diamondiferous Arkhangelskaya (1.35 car/t) and V. Grib (1.27 car/t) pipes was 10 and 18 km narrower compared to the Lomonosovskaya pipe, respectively.

The reconstruction of the vertical distribution of garnet and clinopyroxene xenocrysts within this section of the LM highlights the differences in LM architecture and the efficiency of mantle sampling by kimberlites in regions of diverse diamond content located within different parts of the ADP.

The low diamond content of the TSNIGRI-Arkhangelskaya pipe can be explained by the following set of factors: (1) a high surface heat flow of 38.1 mW/m² (±ΔT 46°), which limits the width of the LM with the LAB at ~211 km and an expected “diamond window” of ~76 km; (2) the real “diamond window” marked by garnet and Cr-diopside xenocrysts is limited to a ~65 km LM interval (depth range from ~135 to 200 km); however, considering that the interval between ~135 and 160 km is practically unmarked by xenocrysts, we propose an even lower thickness for the real “diamond window” of ~40 km; (3) the depth distribution of potentially diamond-bearing peridotites falls within the LM depth interval of ~160–180 km, which further limits the width of a real “diamond window” to ~20 km; (4) despite the presence of diamond-associated garnets at diamond-favorable depths, the amount of potentially diamond-bearing peridotites in the LM could be too low and/or diamond formation processes may have been limited.

The high diamond content of the V. Grib pipe could be explained by the presence of “diamond-favorable” LM as well as the greater efficiency of its sampling by kimberlite: (1) the moderately low geothermal gradient of 36.2 mW/m² (±52 ΔT °) suggests a thicker LM with a LAB at ~241 km and an expected “diamond window” that is ~116 km wide; (2) a thick real “diamond window” (~95 km) marked by garnet and clinopyroxene xenocrysts that are continuously sampled by kimberlites; (3) the most efficient garnet and Cr-diopside sampling was found to be within the diamond-favorable depth range of ~135–180 km; (4) the presence of potentially diamond-bearing peridotites between ~130 and 190 km with a major peak at ~160–180 km that is consistent with the global modes of diamond depth distributions.

The presence of diamonds in the Arkhangelskaya and Lomonosovskaya kimberlite pipes is correlated with several “diamond favorable” features of the LM: (1) moderate low geothermal gradients of 35.7 mW/m² (±ΔT 45°) and 35.3 mW/m² (±ΔT 51°), suggesting a thicker LM with an LAB at ~251 km and ~259 km and an expected “diamond window” width of ~131 km and ~139 km, respectively; (2) thick real “diamond windows” with widths of ~88 km and ~100 km, respectively, as well as the presence of garnet and clinopyroxene xenocrysts at diamond-favorable depths of ~160–190 km. The difference in diamond content of the pipes can be explained by the sampling of potentially diamond-bearing peridotites from a wider interval in the LM by the Arkhangelskaya kimberlite (from ~120 km to at least ~200 km, i.e., ~80 km) compared to the Lomonosovskaya pipe (from ~150 to ~190 km, i.e., ~40 km). In the case of the Arkhangelskaya pipe, eclogites could be an additional source of diamonds [44,100].

Reconstructed models of the architecture of the LM and the efficiency of its sampling by kimberlites in regions of varying diamond content located within different parts of the ADP can be successfully used to inform diamond exploration programs within the northern territories of the East-European Platform to assess the diamond potential of areas of interest and discovered magmatic objects in the region.