Melting modeling of mixed peridotitic and mafic lithologies at shallow depths of the continental metasomatized lithospheric mantle: Implementation for the Early Cretaceous volcanic rocks of Eastern Mongolia

Размещено на http://www.allbest.ru/

Размещено на http://www.allbest.ru/

Melting modeling of mixed peridotitic and mafic lithologies at shallow depths of the continental metasomatized lithospheric mantle: Implementation for the Early Cretaceous volcanic rocks of Eastern Mongolia¹

Maksim V. Kuznetsov1, Valery M. Savatenkov1, 2

Annotation

The Eastern Mongolia volcanic area formed in the Late Mesozoic-Early Cenozoic within Central Asian Orogenic Belt. The main volcanic events of the area occurred in the Early Cretaceous when alkaline basaltic lavas erupted and formed the so-called cover volcanic complex. Geochemical and isotope features of the cover volcanic complex allowed researchers to identify the following mantle rocks as their source: metasomatized peridotites, eclogites, and pyroxenites. Thermodynamic modeling in alphaMELTS program was performed to determine whether the simultaneous melting of these rocks with subsequent processes of crystallization differentiation could lead to the formation of the studied rocks. The modeling results show that the melting of the most enriched with incompatible trace elements peridotites, eclogites, and pyroxenites cannot produce the rocks of the cover volcanic complex. At the same time, the mixing of peridotite- and eclogite-derived melts corresponds most closely to the mechanism of rock formation. However, Ti, K, P, Rb, and Sr enrichment of the studied rocks also requires participation in magma generation processes of mantle metasomatic veins enriched with rutile, apatite, phlogopite, and amphibole.

Keywords: Eastern Mongolia, Early Cretaceous volcanism, thermodynamic modeling, eclogites, pyroxenites, mantle metasomatic veins.

Аннотация

Моделирование процессов смешения расплавов перидотитового и мафического субстратов на малых глубинах континентальной метасоматизированной литосферной мантии. К вопросу о генезисе раннемелового вулканизма Восточной Монголии²

Кузнецов Максим Викторович1, Саватенков Валерий Михайлович1'2

Восточно-Монгольская вулканическая область сформировалась в позднем мезозое-раннем кайнозое в пределах Центрально-Азиатского складчатого пояса. Основные вулканические события в области произошли в раннем мелу, когда щелочно- базальтоидные лавы сформировали так называемый покровный вулканический комплекс. Изотопно-геохимические особенности данного комплекса позволили исследователям установить следующие мантийные породы в качестве его источника: метасоматизированные перидотиты, эклогиты и пироксениты. Чтобы определить, действительно ли одновременное плавление данных пород с последующими процессами кристаллизационной дифференциации могло привести к формированию вулканитов, было проведено термодинамическое моделирование в программе alphaMELTS. Результаты моделирования свидетельствуют о том, что плавление наиболее обогащенных несовместимыми редкими элементами перидотитов, эклогитов и пироксенитов не могло формировать расплавы покровного вулканического комплекса. В то же время, процессы смешения расплавов перидотитов и эклогитов в наибольшей степени соответствуют механизму формирования вулканитов. Однако обогащение вулканических пород Ti, K, P, Rb и Sr по отношению к моделируемым расплавам требует участия в процессах магмогенерации мантийных метасоматических жил, обогащенных рутилом, апатитом, флогопитом и амфиболом.

Ключевые слова: Восточная Монголия, раннемеловой вулканизм, термодинамическое моделирование, эклогиты, пироксениты, мантийные метасоматические жилы.

Introduction

The Eastern Mongolia Volcanic Area (EMVA) is a part of a volcanic province formed during the Late Mesozoic-Early Cenozoic in the east of Asia within the Central Asian Orogenic Belt (CAOB). Despite its name, a small part of the area is located within Russia (Eastern Transbaikalia). The sources of the volcanic rock formation are still debatable. This is especially characteristic of the Early Cretaceous differentiated basaltic rocks (SiO2 > 47 wt.%, MgO < 4.5 wt.%) of the area comprising the so-called cover volcanic complex (Kuznetsov et al., 2022; Kuznetsov et al., 2023; Yarmolyuk et al., 2020) (Fig. 1). Sheldrick et al. (2020a) and Sheldrick et al. (2020c) identified eclogites, pyroxenites, and peridotites of the continental metasomatized lithospheric mantle (CMLM) as the sources of these rocks. Later, Kuznetsov et al. (2022) came to a similar conclusion using data on the rocks' Sr, Nd, and Pb isotope compositions.

Thus, previous studies suggest that the Early Cretaceous basalts of the EMVA were formed because of the simultaneous melting of the CMLM represented by peridotites, eclogites, and pyroxenites. However, these conclusions are based on comparing Mongolian volcanic rocks with well-studied most primitive rocks of other intraplate volcanic provinces. Such an approach has limitations because the variations in the chemical composition of volcanic rocks result at least from different degrees of melting and fractional crystallization. Not considering both factors can lead to incorrect estimates of the parental melts' compositions. Also, it is worth noting that only Peretyazhko et al. (2018) and Sheldrick et al. (2020a) tried to evaluate the role of melting and fractional crystallization using numerical modeling based on equations with fixed parameters (source mineralogy and cumulate composition). However, these parameters change according to thermodynamic equilibrium conditions in the “restite-melt” or “cumulate- melt” systems in natural processes. Thus, a more rigorous approach is required to consider the magmatic system's thermodynamic features. Such an approach should use numerical thermodynamic modeling of phase equilibria in the magmatic system based on experimental data of the melting processes of different lithologies.

The numerical thermodynamic modeling was widely applied to the characterization of magmatic processes in recent decades through the use of alphaMELTS software package, which includes thermodynamic models of Asimow and Ghiorso (1998), Ghiorso et al. (2002), Ghiorso and Sack (1995). It is an effective tool for a more realistic characterization of the influence of various processes (partial melting, fractional crystallization, assimilation, mixing, e.g.) on the evolution of the magmatic melts' composition.

Thus, the thermodynamic modeling approach (alphaMELTS) is used in this paper to determine whether the simultaneous melting of various CMLM lithologies produced parental magmas of the Early Cretaceous rocks of the EMVA.

Geological setting

The EMVA is similar in structure, development features, and composition of its magmatic products to other volcanic areas that occurred within the East Asia province (Fig. 1). The age of the volcanic fields of the EMVA spans from 140 to 48 Ma (Bars et al., 2018; Dash et al., 2015; Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020), which corresponds to the Early Cretaceous-Early Cenozoic. However, this paper only considers the genesis of the Early Cretaceous volcanic rocks (140 - 100 Ma). Their structural features are discussed in detail in Yarmolyuk et al. (2020) and Kuznetsov et al. (2022). Further, we briefly summarize the main conclusions of these works.

The main structure-forming and volcanic events happened in the EMVA at the beginning of the Early Cretaceous. First, the structural framework of this area was formed, whose style is defined by a system of northeast-trending depressions and grabens. Second, mafic lavas formed a thick lava cover. This phase of volcanism (140 - 120 Ma) ended with felsic magmatism, which formed groups of large extrusions, small central volcanoes, and lava domes. In the second half of the Early Cretaceous (120-100 Ma), the lava piles of mafic rocks were formed only. Volcanic rocks are found in the area in various combinations, but the dominant rock associations are always basalt-trachybasalt-trachyandesite and rhyolite- trachyrhyolite. These rocks were produced during the main volcanic phase, encompassing most of the Early Cretaceous, and associated with active graben-forming processes. They dominate the area and make up its lava cover; hence, Kuznetsov et al. (2022) and Yarmolyuk et al. (2020) refer to them as the cover volcanic complex (CVC).

The volcanic area was formed within a territory, including terranes of different origins and ages. However, most Early Cretaceous volcanic fields occurred within the Ereendavaa microcontinent. The rocks with high degrees of metamorphism, including granitic gneisses and migmatites (Badarch et al., 2002), previously justified the Paleoproterozoic age of the Ereendavaa. However, based on the dating of detrital zircons, Miao et al. (2017) have suggested that the formation of the Ereendavaa occurred in the Paleozoic.

Methods

AlphaMELTS is a software package that allows modeling the phase (minerals and melts) relations in magmatic processes. It is based on an approach that assumes the minimization of the Gibbs energy in a system in thermodynamic equilibrium. AlphaMELTS can simulate partial melting, equilibrium crystallization, fractional crystallization, and assimilation using different thermodynamic algorithms. The task of alphaMELTS is to find a model where minerals and melts with specific chemical compositions coexist at minimum energy and given temperature (T) and pressure (P). The software package uses the thermodynamic data presented by Berman (1988).

We use the pMELTS algorithm of alphaMELTS in our calculations (Ghiorso et al., 2002). It is possible to calculate equilibrium phase relationships for magmatic systems in temperature range of 1000 to 2500°C and pressure range of 10 to 30 kbar in pMELTS. This thermodynamic model was calibrated in the system SiO2-TiO2-AhO3-Fe2O3-Cr2O3-FeO-MgO- CaO-Na2O-K2O-P2O5-H2O. Therefore, using other oxides in modeling is unpredictable and not recommended.

It should be noted that the results of numerical calculations cannot strictly correspond to real natural processes. They are prognostic. In addition, pMELTS has significant limitations in the quantitative assessment of accessory phases (rutile, apatite, etc.), which are stable during the melting of mantle or crustal substrate and determine the behavior of trace elements in the resulting melts. Thus, modeling in pMELTS allows us to characterize the dominant trends in the chemical evolution composition of the formed melts, but these models do not describe magmatic processes accurately.

To determine the genesis of CVC rocks, pMELTS was used to simulate partial melting and fractional crystallization processes.

It is unlikely that eclogites and pyroxenites simultaneously were the sources of CVC rocks with a uniform chemical composition. According to the classification proposed by Lambart et al. (2012) and Lambart et al. (2016), these lithologies differ in silica saturation and the nature of the melts they produce. Silica-deficient pyroxenites are the source of nepheline- normative melts. On the other hand, eclogites are analogous to silica-excess pyroxenites and serve as a source of quartz-normative melts. Therefore, the following scenarios of melting modeling in pMELTS were considered:

- simultaneous melting of metasomatized peridotites and eclogites of CMLM;

- simultaneous melting of metasomatized peridotites and pyroxenites of CMLM.

Further, compositions were selected from the obtained melts to simulate fractional crystallization.

Selection of the chemical composition of starting materials

For the numerical simulations of the melting of metasomatized peridotites and pyroxenites/eclogites, the average compositions of the depleted mantle (DM) (e.g., McKenzie and O'Nions, 1991) and mid-ocean ridge basalts (MORBs) (e.g., Gale et al., 2013) can be used, respectively. However, although the main composition of peridotites is relatively homogeneous, the contents of AhO3, FeOtot, CaO, K2O, and P2O5 in xenoliths of peridotites representing the CMLM of Mongolia (Carlson and Ionov, 2019; Kononova et al., 2002; Kourim et al., 2021; Wiechert et al., 1997) sometimes exceed the corresponding values in DM. In addition, the above researchers' data show that Mongolia's CMLM is sometimes enriched in LILEs and LREEs, and depleted in HFSEs compared to DM peridotites. The eclogites/pyroxenites of Mongolia and the CAOB may also differ in their chemical composition from the average composition of MORBs. Sometimes, these rocks are significantly enriched in TiO2, AhO3, FeOtot, CaO, Na2O, K2O, P2O5, LILEs, and LREEs (Ancuta et al., 2017; Barry et al., 2003; Gianola et al., 2019; Saktura et al., 2017; Stosch et al., 1995; Xu et al., 2009). To get results that are more consistent with the real ones, we use the CMLM compositions of Mongolia and the CAOB during the modeling of the formation mechanisms of the considered volcanic rocks. The compositions most enriched in incompatible trace elements were selected. We assume that this approach helps to determine whether the formation of CVC rocks with the participation of the most enriched peridotites, pyroxenites, and eclogites is possible.

It was revealed that most CMLM peridotite xenoliths have negative Ba, Nb, Ta, and Ti anomalies on the primitive mantle-normalized diagrams (Fig. 2). In addition, it can be seen that many of the CMLM peridotites of Mongolia are enriched in Th and U. There are compositions enriched with LREEs among the samples. The last criterion was one of the keys in selecting samples for modeling in pMELTS. As a result, a sample S-16 of spinel lherzolite was selected (Appendix 1) (Carlson and Ionov, 2019).

To date, there is no information in the literature about the findings of xenoliths in the territory of Mongolia represented by pyroxenites or eclogites. These rocks were described as a part of exhumed complexes representing relics of the mantle or Moho (Gianola et al., 2019; Skuzovatov, 2021). However, there is little information in the above sources about the chemical composition of these rocks. Also, these rocks are characterized by a relatively high content of SiO2 (>51 wt.%). Therefore, the rock compositions of the Xuzhou-Suzhou pyroxenite-eclogite complex in Central China were used in the modeling. These rocks represent either relics of subducted oceanic crust (Saktura et al., 2017) or relics of the North China Craton basement (Xu et al., 2009). Therefore, compositions of pyroxenite L4 and eclogite JG2-2 xenoliths were selected (Xu et al., 2009). Garnet, clinopyroxene, and amphibole are the main minerals of these rocks. However, sample JG2-2 also contains plagioclase, rutile, titanite, and quartz. Sample JG2-2 is characterized by a lower content of SiO2, MgO, K2O and a higher content of TiO2, AhO3, FeOtot, and P2O5 (Appendix 1). Both samples are enriched in REEs and depleted in HFSEs (Fig. 3). However, JG2-2 is also highly enriched in Ba, Th, and U.

It should be noted that the CMLM rocks have elevated H2O contents compared to rocks of the depleted mantle. In the case of CMLM peridotites, the H2O content can reach 3737 ppm (~0.37 wt.%) (Green et al., 2014; Turner et al., 2017; Xia et al., 2019). Eclogites can contain up to 3070 ppm H2O (~0.31 wt.%) (Javoy, 1997; Katayama et al., 2006; Radu, 2018; Ragozin et al., 2014). Also, it is known that H2O in the metasomatized peridotites of the mantle wedge is one of the main triggers for their melting (Kelley et al., 2010). Moreover, the melting degree of rocks becomes higher with increasing their H2O content (Hirschmann et al., 2009; Kelley et al., 2010). In this regard, the content of H2O in the CMLM lithologies was also considered during the modeling in pMELTS.

However, data in Appendix 1 shows that H2O content was not determined in selected samples of lherzolite, pyroxenite, and eclogite. Therefore, for the lherzolite sample, the mean value of H2O (0.161 wt.%) in the CMLM peridotites of Mongolia (Wiechert et al., 1997; Lesnov et al., 2009) was used in modeling. The same value was used in the modeling of pyroxenite and eclogite melting.

It is worth noting that this paper does not consider melting modeling with CO2 addition. First, this is since small CO2 contents (<1 wt%) in peridotites and eclogites have been recognized in generating silica-poor, magnesium-rich, strongly alkalic magmas such as kimberlites, melilitites, and nephelinites (Hirose, 1997; Dasgupta et al., 2007, 2013; Malik and Dasgupta, 2014). However, CVC rocks sometimes belong to the subalkaline series and contain >48.5 wt.% SiO2 and <4.2 MgO wt.% (Bars et al., 2018; Dash et al., 2015; Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020). Second, pMELTS is not calibrated for CO2 (Ghiorso et al., 2002).

Determination of P-T parameters for the modeling

A case that includes a constant pressure value and a changing temperature value during the formation of parent melts (isobaric melting) was considered in the paper.

According to Dash et al. (2015), Kuznetsov et al. (2022), and Sheldrick et al. (2020a), the formation of CVC parental magmas in the upper mantle was triggered by heat flow of rising asthenosphere after delamination of the lower parts of the CMLM. Since the isotope features of CVC rocks do not testify to the role of the depleted component in their source (Bars et al., 2018; Dash et al., 2015; Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020), we assume that the areas of magma genesis at the stage of volcanism formation were located at shallow depths in the CMLM far from the lithosphere-asthenosphere boundary. These suggestions are consistent with the conclusions of Yucel et al. (2017) and Kolb et al. (2012) that alkaline magmas generally form due to the high melting degree of the lithospheric mantle at a shallower depth during ongoing lithospheric thinning and associated asthenospheric upwelling. However, we need to appeal to the thickness estimates of the continental crust within Eastern Mongolia to understand what pressure values in the upper parts of the CMLM could have been.

According to Zorin et al. (1999), the thickness of the continental crust in the studied region is 42.5 - 45 km. However, the crust was thicker after the closure of the Mongol-Okhotsk Ocean and the subsequent collision (Meng et al., 2003). Therefore, CVC parental melts could be produced at higher depths. Nevertheless, the most intensive extensional deformations in the region occurred between 130 Ma and 120 Ma (Daoudene et al., 2013). According to the most reliable ⁴⁰Ar/³⁹Ar dating estimates, the peak of EMVA volcanism was at 120.7 (Sheldrick et al., 2020) or 114 Ma (Dash et al., 2015). The results of K-Ar dating indicate that some volcanic fields of CVC were formed even later at 100 Ma (Kuznetsov et al., 2022; Yarmolyuk et al.,

2020) . Therefore, we can conclude that most CVC magmas formed after the main phase of the extension at the depths close to contemporaneous (Zorin et al., 1999).

If we assume the continental crust is homogeneous (granitic) in composition with an average mass of 2691 kg/m³ (Aqua-Calc...), the pressure under one cubic kilometer of the continental crust rocks will be 0.2637 kbar. Considering the estimates of the continental crust thickness in the region (Zorin et al., 1999), the approximate pressure at the boundary of the crust and the mantle is 11.2 - 11.9 kbar. Therefore, the pressure in the upper horizons of the CMLM was more than 12 kbar in the Late Mesozoic. We use P=13 kbar in the thermodynamic modeling of rock melting in pMELTS.

The temperature range of 1200 - 1300°C was chosen for melting modeling. This is because of the following reasons. First, the isotopic and geochemical features of CVC rocks show the role of the CMLM in their formation and exclude the participation of deep mantle sources (Bars et al., 2018; Dash et al., 2015; Kuznetsov et al., 2022; Sheldrick et al., 2020a;

Yarmolyuk et al., 2020). Thus, the role of the mantle plume is unlikely at the stage of CVC formation. Therefore, the temperature of the formation of the parental melt in the CMLM is less than the temperature of the asthenospheric mantle, which, according to (Hamza and Vieira, 2012), is 1280 - 1480°C. Second, at P=13 kbar and T<1200°C, only andesitic melts are formed during the melting of the mafic lithologies (Klemme et al., 2002; Lambart et al., 2013; Pertermann and Hirschmann, 2003; Takahashi et al., 1998). Third, at P=13 kbar and T>1300°C, the resulting melts are not enriched in incompatible trace elements because of the high degree of rock melting. However, one of the main features of CVC rocks is their significant enrichment in trace elements, exceeding the OIBs level (Bars et al., 2018; Dash et al., 2015; Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020).

Selection of distribution coefficients for the melting modeling

To get reliable modeling results for trace elements, it is necessary to ensure that the program uses the correct distribution coefficients (mineral/meltD). The default mineral/meltD values of pMELTS are calculated based on experimental data on the peridotites melting with the formation of basaltic composition melts (McKenzie and O'Nions, 1991; 1995). At the same time, the experiments data on the melting of mafic lithologies at P<15 kbar (Klemme et al., 2002; Lambart et al., 2013; Pertermann and Hirschmann, 2003; Qian and Hermann, 2013) indicate that basalt - basaltic andesite - andesite melts can form. However, to date, mineral/meltD values have been most thoroughly studied only for the “mineral-andesite melt” system (Bedard, 2006; Klemme et al., 2002; Pertermann and Hirschmann, 2003; Qian and Hermann, 2013).

To determine how much the simulation results will differ when using different mineral/meltD, melting modeling of the mean MORB composition (Gale et al., 2013) was carried out at P=13 kbar, T=1200, 1250 and 1300°C with an addition of 0.161 wt.% H2O. In the first case, mineral/meltD values of McKenzie and O'Nions (1991, 1995) were used. In the second case, mineral/meltD values given in Bedard (2006) were used. It should be noted that the latter source provides the most complete data on mineral/meltD values for the “mineral-andesite melt” system during the melting of metabasites. The modeling results are shown in Fig. 4. It is noticeable that the normalized trace element contents mainly do not differ when using different mineral/meltD values. However, significant differences are observed in the HREEs area of spectra. This is because the garnet/meltD values for REEs, Ti, and Y given in Bedard (2006) are significantly higher than the pMELTS default coefficients. For example, according to Bedard (2006) and McKenzie and O'Nions (1991, 1995), garnet/meltDLu equals 24.1 and 5.5, respectively. Because garnet is a stable restite phase at T=1200 and 1250°C, produced melts are strongly depleted in

HREEs, Ti, and Y when modeling with mineral/meltD values of Bedard (2006). At T=1300°C, garnet is unstable, so the modeling results using different mineral/meltD values almost do not differ.

Considering that the modeling results with different mineral/meltD values have little differences, we use mineral/meltD values of McKenzie and O'Nions (1991, 1995) in pMELTS to model the melting of pyroxenite and eclogite. However, mean mineral/meltD values for garnet were used (Table 1). They are calculated using data from McKenzie and O'Nions (1991, 1995) and Bedard (2006).

pMEL TS modeling strategy

The modeling of melting and mixing melts included three stages.

In the first stage, at a constant pressure (13 kbar) and varying temperature (1200 - 1300°C), the melting of pyroxenite/eclogite was simulated using ^garnet/meltD values presented in Table 1. The step of changing temperature was +10°C. Then the program generated the main- and trace element composition of the resulting melt and coexisting residual mineral phases at each step of changing temperature.

In the second stage, the melting of peridotite was simulated following the same principles. However, the ^garnet/^meltD values given in McKenzie and O'Nions (1991, 1995) were used in this case.

In the third stage, melts of different genesis obtained at the same pressure and temperature were mixed. To estimate correctly the chemical composition of the melts formed during mixing, it is necessary to understand the proportions in which the peridotites and eclogites/pyroxenites coexist in the CMLM. However, at present, there are no relevant data in the literature. Based on the fact that the isotopic composition of the studied volcanic rocks (Bars et al., 2018; Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020) is strongly different from that of peridotites of the lithospheric mantle of Mongolia (Carlson and Ionov, 2019; Kononova et al., 2002; Kourim et al., 2021; Wiechert et al., 1997), this paper considers the melting of equal parts of peridotite and eclogite/pyroxenite substrates (1:1). Thus, the chemical composition of the melts formed during mixing was estimated using the formula:

^Cmix = ^Cpyr/ecl * F + Cih * ⁽¹ - F)

Where:

Cmix - concentration of a given component in a hybrid melt; C_vyr/_eci - concentration of a given component in the melt of pyroxenite/eclogite; С_ш - concentration of a given component in the melt of lherzolite; F - weight proportion of pyroxenite/eclogite melt, (1 - F) - weight proportion of lherzolite melt.

Results of the pMELTS modeling

Composition of melts of selected samples

The results of the pMELTS melting modeling at P=13 kbar and T=1200 - 1300°C of eclogite, pyroxenite, and lherzolite are presented in Fig. 5, 6.

The variation diagrams (Fig. 5) show that all modeled melts are depleted in K2O and P2O5 compared to CVC rocks. However, it is worth noting that melts of eclogite at T~1200°C are most similar in main composition to the studied rocks. The only difference is that eclogite- derived melts contain more FeOtot (~18 wt.%), less MgO (~2 wt.%), K2O (~0.3 wt.%), and P2O5 (~0.8 wt.%).

The lherzolite-derived melts correspond to CVC rocks concerning CaO and Na2O contents. However, these melts are strongly depleted in TiO2 and FeOtot and rich in Al2O3. In turn, the pyroxenite-derived melts correspond to CVC rocks in terms of MgO only. The main feature of these melts is their silica-deficient composition.

The primitive mantle-normalized diagrams (Fig. 6) demonstrate that all modeled melts attend the enrichment levels of the studied rocks only at T~1200C°. All spectra have negative anomalies of Ta and Nb, but both positive and negative anomalies of Sr and P. Significant Ti negative anomalies are observed in lherzolite- and pyroxenite-derived melts diagrams. In most cases, the modeled melts are far from the required enrichment levels for the most incompatible trace elements (Rb, K, Ba). Unlike lherzolite-derived melts, pyroxenite- and eclogite-derived melts at T=1200 and 1250C° have the slope of the spectra in the REEs area that coincides with that of CVC rocks. This is because garnet is one of the stable residual phases during the melting of mafic rocks at a given temperature range. During the melting of the spinel lherzolite, the composition of the melt is controlled by a spinel.

Composition of mixing (hybrid) melts

Fig. 7 and 8 show the results of the mixing modeling of CMLM-derived melts. It can be seen that the mixing melts slightly differ in chemical composition from the pure melts of pyroxenite or eclogite. The reason for this is the high melting degree of these rocks compared to peridotites. According to the melting modeling results, at T=1200°C, the melting degree of the eclogite is 0.29, the pyroxenite is 0.12, and the lherzolite is 0.04.

As expected, during the mixing of eclogite- and lherzolite-derived melts at T<1250°C, the resulting melts are more similar in the main composition to the studied rocks relative to hybrids with pyroxenite-derived melts (Fig. 7). The spider diagrams show that the mixing of eclogite- and lherzolite-derived melts does not lead to the required enrichment in trace elements, even at T=1200°C (Fig. 8). Another difference between the compositions of modeled melts and CVC rocks is a significant depletion in fluid-mobile components (Rb, K) of the former. However, hybrids with eclogite-derived melts are characterized by sufficient Ba enrichment. Pyroxenite- and lherzolite-derived hybrids are less similar to CVC rocks. These hybrid melts are more silica-deficient and contain less TiO2 and CaO (Fig. 7). At the same time, they are more enriched in AbO3 and Na2O. The levels of trace elements enrichment of these hybrids are comparable to that of eclogite- and lherzolite-derived hybrids (Fig. 8). However, the former are more enriched in Rb, K, Pb, Sr and less enriched in Ba, Nb, P, and Ti relative to latter. Generally, the primitive mantle-normalized diagrams of pyroxenite- and lherzolite-derived hybrids are similar in pattern to that of CVC rocks.

Thus, the results of mixing modeling show that all hybrid melts lack sufficient enrichment in many main and trace elements. However, the simultaneous melting of eclogites and peridotites of the CMLM is more suitable for CVC rocks forming. The lack of enrichment in some elements can be leveled during crystallization differentiation. To test whether it is possible to enrich hybrid melts during the fractional crystallization to the levels of the studied rocks, a simulation of this process was carried out in pMELTS. Most likely, extremely alkaline and silica-deficient compositions can be formed as a result of crystallization differentiation of pyroxenite- and lherzolite-derived hybrids. However, to confirm this assumption, we also modeled this process in pMELTS.

Modeling of crystallization differentiation

Modeling of crystallization differentiation was carried out as follows. We selected two compositions among the hybrid melts, which can evolve into the compositions of CVC rocks during fractional crystallization. The key criterion in choosing the starting melts was their MgO content. Since the MgO content at the first stages of crystallization differentiation is actively reduced during the fractionation of olivine, its content should be sufficient to form rocks with a magnesian number and trace elements enrichment similar to these in CVC rocks. Therefore, for modeling, it is necessary to choose compositions containing over 4.2 wt.% MgO (the highest value in CVC rocks).

The eclogite-derived hybrids (Fig. 7) with >4 wt.% MgO have much lower contents of the other main components than CVC rocks. For example, when MgO content is ~ 5 wt.%, SiO2 content is ~ 45 wt.%, TiO2 content is ~ 2 wt.%. This may indicate that the crystallization differentiation of eclogite-derived hybrids cannot lead to the formation of melts that correspond to CVC rocks in all main components. It is already possible to cease the simulation of this mechanism in pMELTS at this stage. However, it is necessary to determine whether crystallization differentiation can lead to the required enrichment in trace elements. To model this process, the hybrid composition obtained at T=1200°C was used. It is marked with a red asterisk in Fig. 7 and a bold red line in Fig. 8. Despite this composition having low MgO content, it corresponds most closely to the studied rocks in terms of the content of most main components and the pattern of distribution and enrichment of trace elements.

The pyroxenite-derived hybrids formed at T=1200°C contain more MgO than the CVC rocks. At the same time, these melts are not as silica-deficient as the eclogite-derived hybrids. Therefore, to simulate the process of crystallization differentiation in pyroxenite-derived hybrids, the composition obtained at T=1230°C was used (MgO=5.45 wt.%). This composition is marked with a black asterisk in Fig. 7 and a bold black line in Fig. 8.

During the modeling of crystallization differentiation in pMELTS, a starting temperature corresponded to a temperature at which a particular hybrid melt was initially formed. The temperature step was -10°C. pMELTS generated the main and trace element composition of the fractionated mineral phases and the remaining melt at each step of decreasing temperature. The process of crystallization differentiation was modeled for each hybrid melt at P=5 and 1 kbar (the middle and upper continental crust levels, respectively). The ^mineral/^meltD values given in McKenzie and O'Nions (1991, 1995) were used.

The results of the crystallization differentiation modeling show that the mixing of pyroxenite- and lherzolite-derived melts cannot be responsible for the formation of CVC rocks. Fig. 9 and 10 show that these melts become more silica-deficient and alkaline during their evolution in the continental crust.

As expected, the differentiates of eclogite-derived hybrids have low MgO content relative to CVC rocks. Only the Na2O content in these melts approaches the corresponding values in CVC rocks both at P=5 kbar and P=1 kbar (Fig. 9 and 10). At P=5 kbar, the differentiates have the same CaO content as the studied rocks. In both cases, P2O5 content in melts increases sharply to the levels of CVC basaltic trachyandesites in the first stages of crystallization differentiation. However, neither the melting of P2O5-rich mantle lithologies nor the process of crystallization differentiation can lead to the formation of the most primitive studied rocks regarding P2O5 content. The mechanism of crystallization differentiation of eclogite-derived hybrids at different pressures does not lead to the necessary enrichment in TiO2 and K2O. In addition, the differentiates are excessively rich in FeOtot as well as their parental melts.

In contrast to the results for main components, the trace elements results of crystallization differentiation modeling show that the mixing of lherzolite- and pyroxenite- derived melts can form compositions like those of CVC rocks. Fig. 11 shows that the enrichment levels in trace elements of these differentiates sometimes even exceed the levels of the studied rocks. Moreover, the spider diagrams pattern of these differentiates almost wholly corresponds to that of studied volcanic rocks. The only difference is an insufficient enrichment in Ba and excessive enrichment in Th and U. An insignificant difference is also more pronounced negative Zr and Hf anomalies in the spectra of the differentiates. The degree of the spectra slope in the REEs area also corresponds to that in the studied volcanic rocks.

Modeling results for eclogite-derived hybrids also indicate that crystallization differentiation can lead to trace elements enrichment in them to the level of CVC rocks (Fig. 11). The spider diagrams pattern of the differentiates is very similar to that of studied volcanic rocks. There is almost complete correspondence in the area of REE. However, the required levels of Rb, K, and Sr enrichment are not achieved. As in the case of differentiates of pyroxenite-derived hybrids, there is an excessive enrichment in Th and U. This is due to the fact that initially modeled compositions are enriched in these elements.

Discussion

Role of metasomatic veins in the volcanic rocks formation

An analysis of the modeling results allows us to conclude that the case of mixing of eclogite- and lherzolite-derived melts is most consistent with the studied rocks. At the same time, this process does not produce magmas, which completely correspond in chemical composition to the basalts of the CVC. Thus, the modeling results of melting and crystallization differentiation using the most enriched CMLM compositions in incompatible elements indicate that melts with the required enrichment of Ti, K, P, Rb, and Sr are not formed. Therefore, besides metasomatized peridotites and eclogites, other CMLM rocks were involved in the formation of CVC rocks.

Positive P and Ba anomalies in the primitive mantle-normalized trace element spectra of CVC basalts (Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020) may reflect the involvement of phlogopite- and apatite-bearing mantle metasomatic veins during magma generation processes (Kontak et al., 2001). Bedard et al. (2021) pointed out that the formation of alkaline rocks of the High Arctic igneous province in Canada enriched in TiO2 (34 wt.%), K2O (~3 wt.%), and P2O5 (~1.5 wt.%) resulted from melting of mantle metasomatic lithology of mainly pyroxenite composition with such additional mineral phases as apatite, rutile, and phlogopite. CVC rocks sometimes contain even more TiO2 and P2O5 (Bars et al., 2018; Dash et al., 2015; Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020). Farmer et al. (2020) came to a similar conclusion about the role of rutile- and apatite-bearing CMLM in forming mafic continental rocks of the Sierra Nevada and Rio Grande Rift with Ta/Th ratio between 0.2 and 0.6. This ratio varies from 0.23 to 0.52 in CVC basalts (Kuznetsov et al., 2022; Sheldrick et al., 2020a; Yarmolyuk et al., 2020).

O'Reilly and Griffin (2013) concluded that magmas highly enriched in Sr, P, and K could only be produced by melting of metasomatized peridotites with a significant content of apatite and phlogopite+amphibole, respectively. Furthermore, experiments on the melting of the phlogopite-bearing peridotite (Condamine and Medard, 2014) show the formation of melts rich in K (3.05 - 6.68 wt.%) and Ti (0.29 - 1.45 wt.%). In turn, melts containing up to 5 wt.% TiO2 can be formed during the melting of the substrate consisting of clinopyroxene and amphibole (Pilet et al., 2008). In addition, the studies of veins in peridotite massifs also testify to the role of atypical mantle rocks and minerals in the origin of K-rich magmas. Veins of the lithospheric mantle may be represented by glimmerites, the main minerals of which could be phlogopite and apatite (Becker et al., 1999). These rocks have high K2O (up to 9.13 wt.%) and Rb content (up to 879 ppm), which suggests that their melting could produce melts enriched in these elements.

Thus, the lack in Ti, K, P, and sometimes Rb and Sr of eclogite- and lherzolite-derived hybrid melts relative to studied rocks requires participation in the volcanic rocks' formation of other CMLM lithologies (metasomatic veins) enriched with rutile, apatite, phlogopite, and amphibole. What could be the origin of these metasomatic veins? Considering that CVC rocks have high LILE/HFSE and LREE/HREE ratio values, we can conclude that the formation of veins in the CMLM could be related to previous subduction processes (Pearce et al., 2005). This conclusion is consistent with the fact that the territory of Eastern Mongolia was involved in subduction processes until the end of the Early Mesozoic (Arzhannikova et al., 2022; Yarmolyuk et al., 2019).

It should be noted that the melting modeling of metasomatic lithologies is associated with specific difficulties. For example, when selecting the starting materials representing metasomatized peridotites, pyroxenites, or eclogites, we can refer to the compositions of these rocks that are common within the region. However, this is not possible with other metasomatic formations since the volume of one mineral in metasomatic veins may vary significantly. Therefore, it is difficult to determine which minerals should be chosen for modeling and in what proportions they should be used. Nevertheless, the data of the melting modeling of the enriched in trace element CMLM lithologies lead us to the conclusion that rutile-, apatite-, phlogopite-, and amphibole-bearing metasomatic veins were involved in the magma generation processes.

Conclusion

thermodynamic modeling magmageneration

Thermodynamic modeling using pMELTS showed that the mixing of peridotite- and eclogite-derived melts corresponds most closely to the formation mechanism of Early Cretaceous rocks of the CVC of Eastern Mongolia. However, the lack in Ti, K, P, Rb, and Sr of the hybrid melts requires participating in magma generation processes of mantle metasomatic veins enriched with rutile, apatite, phlogopite, and amphibole. Thus, it could be concluded that peridotites, eclogites, and metasomatic veins of the CMLM were the sources of the studied rocks.

References

thermodynamic modeling magmageneration

Aqua-Calc [online]. Available at https://www.aqua-calc.com/calculate/volume-to- weight/substance/granite-coma-and-blank-solid [Accessed 05.04.2023].

Ancuta, L. (2017) Toward an Improved Understanding of Intraplate Uplift and Volcanism: Geochronology and Geochemistry of Intraplate Volcanic Rocks and Lower-Crustal Xenoliths. Doctoral dissertation, Lehigh University, Bethlehem, PA 18015 USA.

Arzhannikova, A.V., Demonterova, E.I., Jolivet, M., Mikheeva, E.A., Ivanov, A.V., Arzhannikov, S.S., Khubanov, V.B., Kamenetsky, V.S. (2022) Segmental closure of the Mongol-Okhotsk Ocean: Insight from detrital geochronology in the East Transbaikalia Basin. Geosci. Frontiers, 13(1), 1674-9871, https://doi.org/10.1016/i.gsf.2021.101254

Asimow, P., Ghiorso, M. (1998) Algorithmic modifications extending MELTS to calculate subsolidus phase relations. Am. Min., 83(9-10), 1127-1132,https://doi.org/10.2138/am-1998-9-1022

Badarch, G., Cunningham, D., Windley, B. (2002) A new terrane subdivision for Mongolia: implications for the Phanerozoic crustal growth of Central Asia. J. Asian Earth Sci., 21, 87-110, https://doi.org/10.1016/S1367-9120(02)00017-2

Barry, T., Saunders, A., Kempton, P., Windley, B., Pringle, M., Dorjnamjaa, D., Saandar, S. (2003) Petrogenesis of Cenozoic basalts from Mongolia: evidence for the role of asthenospheric versus metasomatized lithospheric mantle sources. J. Petrol., 44(1), 55-91, https://doi.org/10.1093/petrology/44.1.55

Bars, А., Miao L., Fochin, Z., Baatar, M., Anaad, C., Togtokh, K. (2018) Petrogenesis and tectonic implication of the Late Mesozoic volcanic rocks in East Mongolia. Willey Geol. J.53(6), 1-22, https://doi.org/10.1002/gi.3080

Becker, H., Wenzel, T., Volker, F. (1999) Geochemistry of glimmerite veins in peridotites from Lower Austria - implications for the origin of K-rich magmas in collision zones. J. Petrol., 40(2), 315-338, https://doi.org/10.1093/petroi/40.2.315

Bedard, J.A. (2006) A catalytic delamination-driven model for coupled genesis of Archean crust and sub-continental lithospheric mantle. Geochim. Cosmochim. Acta, 70(5), 1188-1214, https://doi.org/10.1016/i.gca.2005.11.008

Bedard, J., Troll, V., Deegan, F., Tegner, C., Saumur, B., Evenchick, C., Grasby, S., Dewing, K. (2021) High Arctic Large Igneous Province Alkaline Rocks in Canada: Evidence for Multiple Mantle Components. J. Petrol, 62(9), https://doi.org/10.1093/petrology/egab042 Carlson, R. and Ionov, D. (2019) Compositional characteristics of the MORB mantle and bulk silicate earth based on spinel peridotites from the Tariat Region, Mongolia. Geochim. Cosmochim. Acta, 257, 206-223, https://doi.org/10.1016/i.gca.2019.05.010

Condamine, P. and Medard, E. (2014) Experimental melting of phlogopite-bearing mantle at 1 GPa: Implications for potassic magmatism. Earth and Planetary Sci. Lett., 397, 8092, https://doi.org/10.1016/i.epsl.2014.04.027

Daoudene, Y., Ruffet, G., Cocherie, A., Ledru, P., Gapais, D. (2013) Timing of exhumation of the Ereendavaa metamorphic core complex (north-eastern Mongolia) - U-Pb and ⁴⁰Ar/³⁹Ar constraints. J. Asian Earth Sci., 62, 98-116,

https://doi.org/10.1016/i.iseaes.2011.04.009

Dasgupta, R., Hirschmann, M.M., Smith, N.D. (2007). Partial melting experiments of peridotite + CO2 at 3 GPa and genesis of alkalic ocean island basalts. J. Petrol., 48(11), 20932124, https://doi:10.1093/petrology/egm053

Dasgupta, R., Mallik, A., Tsuno, K., Withers, A.C., Hirth, G., Hirschmann, M.M. (2013). Carbon-dioxide-rich silicate melt in the Earth's upper mantle. Nature, 493(7431), 211215, https://doi:10.1038/nature11731

Dash, B., Yinb, A., Jiang, N., Tseveendori, B., Han, B. (2015) Petrology, structural setting, timing, and geochemistry of Cretaceous volcanic rocks in eastern Mongolia: Constraints on their tectonic origin. Gondwana Res., 27, 281-299,

https://doi.org/10.1016/i.gr.2013.10.001

Dawson, J. and Smith, J. (1977) The MARID (mica-amphibole-rutile-ilmenite- diopside) suite of xenoliths in kimberlite. Geochim. Cosmochim. Acta, 41(2), 309-323, https://doi.org/10.1016/0016-7037(77)90239-3

Farmer, G.L., Fritz, D.E., Glazner, A.F. (2020) Identifying Metasomatized Continental Lithospheric Mantle Involvement in Cenozoic Magmatism From Ta/Th Values, Southwestern North America. Geochem. Geophys. Geosyst., 21(5), https://doi.org/10.1029/2019GC008499 Gale, A., Dalton, C., Langmuir, C., Su, Y., Schilling, J.-G. (2013) The mean composition of ocean ridge basalts. Geochem. Geophys. Geosyst., 14.

https://doi.org/10.1029/2012GC004334

Ghiorso, M., Sack, R. (1995) Chemical Mass-Transfer in Magmatic Processes IV. A Revised and Internally Consistent Thermodynamic Model for the Interpolation and Extrapolation of Liquid-Solid Equilibria in Magmatic Systems at Elevated-Temperatures and Pressures. Contrib. Mineral. Petrol., 119 (2-3), 197-212, https://doi.org/10.1007/BF00307281 Ghiorso, M.S., Hirschmann, M.M., Reiners, P.W., Kress, V.C. (2002) The pMELTS: A revision of MELTS aimed at improving calculation of phase relations and major element partitioning involved in partial melting of the mantle at pressures up to 3 GPa. Geochem. Geophys. Geosyst., 3(5), https://doi.org/10.1029/2001GC000217

Gianola, O., Schmidt, M., Jagoutz, O., Rickli, J., Bruguier, O., Sambuu, O. (2019) The Crust-Mantle Transition of the Khantaishir Arc Ophiolite (Western Mongolia). J. Petrol., 60(4), 673-700, https://doi.org/10.1093/petrology/egz009

Green, D., Hibberson, W., Rosenthal, A., Kovacs, I., Yaxley, G., Falloon, T., Brink, F. (2014) Experimental study of the influence of water on melting and phase assemblages in the upper mantle. J. Petrol., 55, 2067-2096, https://doi.org/10.1093/petrology/egu050

Gregoire, M., Bell, D., Le Roex, A. (2002) Trace element geochemistry of phlogopite- rich mafic mantle xenoliths: their classification and their relationship to phlogopite-bearing peridotites and kimberlites revisited. Contrib. Mineral. Petrol., 142, 603-625,

https://doi.org/10.1007/s00410-001-0315-8

Hamza, V. and Vieira, F. (2012) Global distribution of the lithosphere-asthenosphere boundary: a new look. Solid Earth, 3, 199-212, https://doi .org/10.5194/se-3-199-2012

Hirose, K. (1997). Partial melt compositions of carbonated peridotite at 3 GPa and role of CO2 in alkali-basalt magma generation. Geophys. Res. Lett., 24(22), 2837-2840, https://doi.org/doi:10.1029/97gl02956

Hirschmann, M., Tenner, T., Aubaud, C., Withers, A. (2009) Dehydration melting of nominally anhydrous mantle: The primacy of partitioning. Physics of the Earth and Planetary Interiors, 176, 54-68, https://doi.org/10.1016/i.pepi.2009.04.0Q1

Javoy, M. (1997) The major volatile elements of the Earth: Their origin, behavior, and fate. Geophysical Research Letters, 24(2), 177-180, https://doi.org/10.1029/96GL03931

Katayama, I., Nakashima, S., Yurimoto, H. (2006) Water content in natural eclogite and implication for water transport into the deep upper mantle. Lithos, 86 (3-4), 245-259, https://doi.org/10.1016/nithos.2005.06.006

Kelley, K., Plank, T., Newman, S., Stolper, E., Grove, T., Parman, S., Hauri, E. (2010) Mantle Melting as a Function of Water Content beneath the Mariana Arc. J. Petrol., 51(8), 1711-1738, https://doi:10.1093/petrology/egq036

Klemme, S, Blundy, J., Wood, B. (2002) Experimental constraints on major and trace element partitioning during partial melting of eclogite. Geochim. Cosmochim. Acta, 66, 31093123, https://doi.org/10.1016/S0016-7037(02)00859-1

Kolb, M., Paulick, H., Kirchenbaur, M., Munker, C. (2012) Petrogenesis of mafic to felsic lavas from the Oligocene Siebengebirge volcanic field (Germany): implications for the origin of intracontinental volcanism in Central Europe. J. Petrol., 53, 2349-2379, https://doi.org/10.1093/petrology/egs053

Kononova, V., Kurat, G., Embey-Isztin, A., Petrov, V.A., Koeberl, C., Brandstatter, F. (2002) Geochemistry of metasomatised spinel peridotite xenoliths from the Dariganga Plateau, South-eastern Mongolia. Mineralogy and Petrology, 75, 1-21,

https://doi.org/10.1007/s007100200012

Kontak, D., Jensen, S., Dostal, J., Archibald, D., Kyser, T. (2001) Cretaceous mafic dyke swarm, Peary Land, Northernmost Greenland: geochronology and petrology. The Canadian Mineralogist, 39, 997-1020, https://doi.org/10.2113/gscanmin.39.4.997

Kourim, F., Wang, K.-L., Beinlich, A., Chieh, C.-J., Dygert, N., Lafay, R., Kovach, V., Michibayashi, K., Yarmolyuk, V., Iizuka, Y. (2021) Metasomatism of the off-cratonic lithospheric mantle beneath Hangay Dome, Mongolia: Constraints from trace-element modeling of lherzolite xenoliths. Lithos, 400-401, https://doi.org/10.1016/Uithos.2021.106407 Kuznetsov, M.V., Savatenkov, V.M., Shpakovich, L.V., Yarmolyuk, V.V., Kozlovsky, A.M. (2022) Evolution of the Magmatic Sources of the Eastern Mongolian Volcanic Area: Evidence from Geochemical and Sr-Nd-Pb Isotope Data. Petrology, 30(5), 457-479, https://doi.org/ 10.1134/S0869591122050034