Differentiation of Chemical Components in Basaltic Melts During Eruptions: An Example from Tolbachik–Hawaiian Fissure Zones and Etna Vent

The study on differentiation of chemical components in quenched volcanics from the Etna, Tolbachik and Hawaiian fissure zones was carried out under simulated temperature/pressure (T/P) conditions reflecting real-time conditions in the early stages of magmatic eruptions. The simulation was performed under high-pressure chamber apparatus, and the rocks were then subjected to high T/P conditions resulting in basaltic melts. Various temperature and pressure conditions were adjusted from the apparatus to allow for different rates of magmatic cooling and crystallization. The scanning electron microscopy (SEM) coupled with energy-dispersive spectroscopy (EDS-microanalysis) was then used to study the quenched glasses with the aim of identifying major element partitioning based on micro-heterogeneity of melts structures. The microscopic study of the experimental melts and their comparisons with natural melts under the electron microscopes indicate the existence of two distinct liquids which fractionated during cooling on the basis of their chemical composition (compositional melts): Fe, Mg liquids concentrated in the poorly polymerized and mobile parts of the melt and K, Na and Al liquids were concentrated in more highly polymerized parts of the melts. Partitioning of Ca and Si appears to be changeable and more dependent on bulk melt composition. Fractionation of non-crystalline melts was evident under post-explosion sharp quenching and high-pressure conditions.


Introduction
The chemical composition of the Earth's crust is inherently different from that of the mantle, and therefore, questions pertaining to the nature and factors influencing crystallization and placement of magmatic melts remain equivocal [1,2].Hack and Thompson [3] pointed out the influence of magmatic viscosity to its migration speed.In addition, the amount of H 2 O dissolving in silicate melts influences meltviscosity.Fractionation of non-crystalline magmatic melts has been a debatable scientific problem over the past three decades shortly after the establishment of micro-heterogenetic clustering structure of magmatic melts.The importance of liquid differentiation in petrogenetic processes has remained a controversial topic [4,5].
In many fields of research and most importantly in material science, behaviors of chemical components in silicate melts have been extensively studied due to mineral processing and production of tech glasses, crystals and ceramics [6,7].Inoue et al. [8] studied migration of the liquids containing cations due to appearance of negatively charged surfaces of framework clusters in aluminosilicate melts.Their findings agreed that the separation of volatiles in melts tends to accelerate the differentiation process during migration of chemical components into less polymerized surroundings [9].
However, there remains obscurity to this day with regards to the importance of silicate liquid differentiation during magmatic evolution.Although the distribution of chemical components in melts determines most of the magmatic features such as viscosity, mechanism of crystal growth, behavior of volatiles, and components mobility, the rate, however, of melt differentiation remains poorly known [10].The basis for describing the structure, physicochemical properties of magmatic melts and the geological processes is to study natural columnar basalts and modeling of silicate melts and glasses [11].The glass state of silicate material is commonly considered as a structural analog to its molten state through saving of cation structural arrangements.Therefore, a big part of the research on this subject has rather been mainly focused on silicate glasses [12].
In this study, quenched volcanic products of the modern eruptions of the Tolbachik fissure zone, scoria from the 2018 Etna vent and quenched lava of the Hawaii fissure zone were studied (Fig. 1a, b). Figure 2 shows a well-pronounced correlation of major element trends during eruption events with a degree of polymerization of aluminum saturation index (ASI), where Fe, Ca, and Mg are concentrated in poorly polymerized melt while K, Na, Si are mainly found in highly polymerized melts.The ASI demonstrates structural-chemical peculiarities of melt components and is calculated as the molecular relationship of Al 2 O 3 (Al in tetrahedral position as Si) to network-forming cations (usually the alkaline and alkaline earth elements): ASI = Al 2 O 3 /(Na 2 O + K 2 O + CaO).For example, a general sequence of the Tolbachik fissure zone constitutes of two extreme types of basalts which differ primarily in the content of Mg and Al: middle-K, high-Mg (Mg/Al:2.1;ASI:0.45)basalts and high-K, high-Al (Mg/ Al:0.45;ASI:0.8)basalts [11].Similar changes in chemical compositions of the volcanic products have been noticed in other volcanic areas such as in old and recent lavas of the Etna and the Kilauea (Fig. 2b, c).The data shown in Fig. 2 may not always be modeled through the crystal fractionation process.And therefore, the aim was to find the features of melts disintegration in natural glasses and to compare with experimental products.

Sampling and Methodology
Basaltic lavas are prone to rapid crystallization in an open system due to degassing and cooling [25].Thus, to accurately determine the melt compositions and preserve the initial relations between the chemical components, a rapid quenching is necessary before nuclei-intergrowths, to avoid phases transformations and different degree of the iron  [14], where NEC-northeast crater, VOR-Voragine, BN 1 and 2-Bocca Nuova, SECsoutheast crater, NSEC-north-southeastern crater oxidation which may alter the Fe-ion coordination from octahedral to tetrahedral.It was therefore necessary in this study to collect quenched volcanic products such as green scoria from the 1941 Tolbachik vent, the sub-water quenched lava from the 1976 Tolbachik southern vent, the snow quenched lava from the 2012-13(Kilauea), scoria from the 2018 Etna vent, and the quenched-porous lava of the Rift East Zone.Sample materials used for the experiments were powders of homogeneous glasses from the 1941 basalts and the natural volcanic glasses of underwater basalts from the Juan de Fuca Ridge.The absence of crystals was proved through the use of X-ray powdered diffraction.Specimen preparation was In this section, we took our own laboratory measurements and compared them with data from previous articles to deduce the relationships [15][16][17][18][19][20][21][22][23][24] checked by using standard XRF-XRD methods and the size of powder particles was 5-7 μm [26].
Laboratory based-simulation studies aimed at recreating some of the alleged conditions of eruption process were carried out to compare the structural-chemical features of the experimental and natural samples.The experiments were carried out under Anvil high-pressure chamber system (a high-pressure apparatus (HPA)) [27].HPA is placed between two metal block-matrices in the DO-138B press high-pressure.The device included calcite container and graphite-based heater with diameter of the reaction space reaching up to 11.5 mm.In our case different modes of melting within various pressure (4-7 MPa) and temperature (1400-1700 °C) conditions created by the electric current through graphite heater were used without any additional mixture.The duration of the experiments was shortened to avoid powder grains from undergoing substantial changes.The material compaction was then induced.After exposure to the sample, the powder was removed from the HPA for air-quenching.Sample preparation for SEM studies included the following stages: 1. Firstly, samples were mounted in an epoxy and then prepared as polished section only with diamond paste.They were then studied and examined under reflected light using Olympus GX-71 petrographic microscopy.Some samples were carbon-coated and then studied under SEM carried out on a JEOL JSM 7001F equipped with a field-emission cathode and the Oxford XMax 80 mm 2 , which is a large-area silicon drift detector (SDD) for the routine microanalysis in energy-dispersive spectroscopy (EDS).2. The most optimal conditions chosen were as follows: acceleration voltage from 15 to 25 kV, a beam current of 10A, and a working distance of 10 mm.The conditions of the electron beam varied depending on the purpose and material properties.The scanning area mode and the acceleration voltage from 15 to 20 kV were used for EDS analysis of glasses compositions.Fully focused beam with the material-dependent beam current of 8-10-10-10 A, acceleration voltage 25 kV were used for back-scattered electrons (BSE) imaging of local variations in average number of the specimen.
The backscattering topographical mode (TOPO) and the secondary electron detection (SEI) were used in order to check the quality of the surface of samples, and in locating micro-morphologies and nano-heterogeneities.
For optimal observation of glass heterogeneities, beams were frequently adjusted in every area to obtain the maximum contrast in BSE mode.Besides the high-contrast backscattered electrons (BSE) imaging of all samples, chemical analyses have been made in selected areas.Compositional information on glasses was acquired within 60-120 s in realtime for the major element analyses; the quant optimization was produced on the basis of Co.
The element interrelationship assessment was executed by means of chemical profiling across the boundaries of different areas of the samples.Chemical profiling by line scan-mode was performed at a focused beam in different directions to insure the accuracy of compositional changes.

Results and Discussion
Our experimental studies of the melt samples were carried out under simulated P/T conditions, reflecting those in realtime magmatic chambers and the crystallized melts were then subjected to SEM-EDS to further study their crystal morphologies and partitioning of chemical components during different phases of crystallization.Many glass-forming multi-components oxide systems tend to spontaneously separate into two liquid phases.The BSE images of scoria, ash and glass-rich samples contain similar inhomogeneities [28].In basaltic melts, liquid differentiation occurs as a result of contrasting roles of two types of structural components: network-formers and network modifiers.According to modern state of science, the basic structural unit of the SiO 4 tetrahedron can form one-, two or three-dimensional networks and clusters.The network is completely polymerized and each Si-O-Si bridging oxygen is shared by two SiO 4 tetrahedra.Network-forming cations (in silicate melts primarily Si 4+ , including Al 3+ , Fe 3+ , Ti 4+ ) are surrounded by oxygen ions in bridging position (Si-O-Si).Network-modifying cations are surrounded by oxygen ions in non-bridging positions (Si-O-M, where M denotes the six-fold coordinated by oxygen network-modifying cations, as a rule of Mg, Fe (II), Ni etc.).It is known that the widths of miscibility gaps in metal oxide-silicate systems and the cations mobility increase with increasing ionic potential, but decrease with increasing the metal cations radius [29].
During melt separation the chemical elements are distributed among the two conjugates according to melt structure, which depends on composition, the number of volatiles, pressure and temperature.The ratio of the ionic potential Z to radius of ions (Z/r) is considered to be a principal driving force of silicate immiscibility [7,[30][31][32].The distributions of micro-heterogenetic structure of basaltic melts or/and crystals, enrichment of element Mg and Fe increase with increasing Z/r.
Analytical results are reported in the form of BSE images, tables with the EDS point analysis and demonstrated in X-ray lines across boundaries between compositionally diverse areas to illustrate chemical elements distribution.The BSE imaging reveals that these areas consist of two compositionally different melt parts.The bright crystalclasts with poor glass structure comprise of numerous flow bands and delicate schlieren of Mg-Fe rich glass.They are regularly distributed in glass-rich quenched samples and sometimes tend to occur as vesicular parts, coatings around pore-spaces and dendritic drainage pattern (tube-like) network of pathways (Fig. 3).In all cases, the amorphous structure is proved by microdiffraction (transmission electron microscopy), which showed diffusion and concentric rings, typical for amorphous material in sample 1941.
The SEM investigations of the basaltic sample 1941 indicated the separation of chemical elements.We found the existence of similar areas in slabs of quenched Hawaiian lava.However, inhomogeneities in them were observed by other scientists, but the descriptions of this process were not discussed [28].
The indicator ratios of elements in dark and bright areas in the sample 1941 and in experimental glasses are similar (Fig. 2a).The behavior all of chemical elements reflects the same compositional differentiation in glass-rich quenched samples of lava.Under the EDS analysis, main tendencies of indicator ratios were compared (Fig. 4).Matrix is always enriched in Al and alkaline elements.Sometimes these areas consist not only of glass, but also of micro/dendritic crystals.Thus, matching of point glass analysis and the area analysis has shown similar trends (Fig. 5a), suggesting that this heterogeneity structure is a consequence of the crystallization of bright and dark areas like experimental glasses and in the 1941th sample, when experimental samples were The EDS analysis in contrast parts  Table 1 Major element composition of various parts of the samples and their indicator ratios analysed using EDS microanalyser (for Fig. 4)

Conclusions
The paper presents and discusses the SEM-EDS compositional results obtained from both natural and experimental basaltic melts (glasses).Based on the experimental tests carried out in this study, two distinct varieties were defined based on their chemical compositions: Fe, Mg-rich liquids concentrated in movable part of the melt and K, Na and Alrich liquids which are mainly concentrated in more polymerized part of the melts.Partitioning of Ca and Si is less pronounced and is dependent on bulk melt composition.The Mg, Fe-containing parts have more mobility.The features of this process with similar chemical allocation were detected in most of quenched products of basaltic eruptions.Separation of micro-heterogenetic structure of basaltic melts appears to begin occurring during the extreme-rising of basaltic magmas.Hence, the separation is not only limited to melts of variable composition (unmixing in a miscibility gap) but influences also the liquidus state of melts.This separation in two-liquids is a much more common process in evolution of basaltic melts.Similar melt inhomogeneities were produced by different modern volcanoes and were detected in the rapid quenched basaltic glasses.The distribution consequence between elements is the same as in HPA experiments.

Fig. 1 a
Fig. 1 a Geological structure of the Tolbachik Volcanic Field showing the distribution of various lava flows and sample locations modified from [13].b Satellite image of the Mt.Etna indicating the extents of lava flow fields from 1991-1993 SE flank eruption, the 2001 upper

Fig. 3
Fig. 3 The bright crystal poor glass samples with dendritic structural network of Mg-Fe (a and b scoria of the 1941 Tolbachik samples; c quenched lava from the 1976 Tolbachik south vent; d quenched lava from the 2012-13 Tolbachik eruption; e and f sample of Etna region; g scoria of Etna region; f and h sample of Hawaii region; i and j experiment of the 1941 Tolbachik samples)

Fig. 4
Fig. 4 The BSE images and the EDS analysis of different volcanic melts in contrast parts and the main indicator ratios (a scoria of 1941 Tolbachik samples; b and c quenched lava from the 1976 Tolbachik south vent; d quenched lava from the 2012-13 Tolbachik eruption; e Etna region and f Hawaii region)

Fig. 5
Fig. 5 Crystallization of glasses (a natural samples, b the EDS analysis of volcanic melts on contrasting areas c natural sample, d experimental sample)

Fig. 6
Fig. 6 Line-scans across typical cluster-areas of inhomogeneities and portions with the dendritic network (a in experimental on basis of 1941 Tolbachik samples, b in quenched lava from the 1976 Tolbachik south vent, c in quenched lava from the 2012-13 Tolbachik eruption, d in scoria from the 2017 Etna eruption) ◂