Paleoproterozoic Variolitic Lavas from the Onega Basin, Fennoscandian Shield: Mineralogy, Geochemistry and Origin

Abstract

The Yalguba Ridge volcanic rocks form part of the Middle Paleoproterozoic (ca. 1.97 Ga) volcano-sedimentary sequence within the Karelian Craton in the Fennoscandian Shield. Yalguba variolitic textures are known worldwide and have been previously considered to originate from liquid immiscibility. The present study reveals two new variolite types recognized in the Yalguba sequence: (1) Variolites with unzoned varioles have distinct chemical and mineralogical compositions of varioles and matrix that support an origin by liquid immiscibility. They were recognized in quenched zones of pillows, so it might be assumed that melt separation caused by liquid immiscibility occurred before magma emplacement. The difference from the previously described variolites lies in the variole microtexture and might be caused by the various cooling conditions. (2) Spherulitic variolites have varioles composed of andesine–oligoclase spherulites embedded in the cryptocrystalline matrix with oligoclase–anorthoclase composition, thus the variole and matrix have similar chemical and mineralogical composition. The mineralogical and textural features of these variolites suggest that the spherulites have a primary magmatic origin due to the rapid cooling of superheated magma. The variety of variolitic textures in the Yalguba section might be caused by the different H₂O saturation of parental magma and cooling conditions.

1. Introduction

Variolitic textures referred to as varioles, globules, orbs, orbicules, and ocelli are common in Archean and Paleoproterozoic volcanic rocks, but rare in modern volcanic sequences. The origin of variole-bearing rocks (hereafter, variolites) has been attributed to (1) mixing of contrasting acid and basic melts [1,2]; (2) liquid immiscibility [2,3,4,5,6,7]; (3) rapid cooling of superheated magmas [8,9] and (4) postmagmatic hydrothermal or metamorphic alteration of basaltic glass [10]. Understanding variolite formation is important to retrieve magma chamber processes and volcanic processes, especially in the formation of early Earth crust. Variolites are intensively studied worldwide, but the interpretation of their textural, chemical and mineralogical features still remains a challenge because the overlapping of several processes can lead to variolite formation or a commonly high degree of metamorphic alteration.

Most studies reported on the occurrence of variolites of similar origin within a single volcanic cycle, e.g., variolites of undercooling origin from metabasaltic lavas in the uppermost part of the Palaeoarchaean Hooggenoeg Complex of the Barberton Greenstone Belt [9] or the variolites caused by liquid immiscibility from the Archean basalts of the Pilbara Supergroup, Western Australia [7]. A magma mingling origin has been proposed for the variolites from the ca. 3075 Ma Ivisaartoq greenstone belt, SW Greenland [1,11]. To the best of our knowledge, only the variolitic textures from volcanic rocks of the Abitibi Greenstone belt, Canada, are interpreted to have various origins including liquid silicate immiscibility [12], undercooling of superheated magmas [13] and probably mingling of basalt and rhyolite [14]. Understanding the generation of variety in variolite types within a single volcanic cycle would provide new insights into the mechanism of variolite formation.

In the Karelian Craton (Central Karelia, Russia), variole-bearing rocks are recognized in the various Paleoproterozoic volcanic sequences [15,16]. Among them, the Yalguba Ridge variole-bearing lavas in the Middle Paleoproterozoic (~1.97 Ga) Suisary Formation from the central part of the Karelian Craton are the best studied and worldwide known. The volcanic rocks from the Yalguba are well preserved due to the low-grade metamorphic alteration and are clearly exposed, which makes them an appropriate object to study.

The Yalguba variolites have a long history of geological investigation starting at the end of the XIX century by A.A. Inostrantsev and F.Yu. Levinson-Lessing [17]. The latter was the first to attribute the Yalguba variolite formation to the liquid immiscibility. His hypothesis has been further supported by petrological and geochemical studies [18,19,20,21,22] and model experiments [23]. However, Gudin et al., 2012 [21] revealed some inconsistencies between the trace element distribution observed in the Yalguba variolites with the mechanism of liquid immiscibility revealed in model experiments. Regardless of the intensive studies of the Yalguba variolites, some issues including the morphology of globular secretions and the crystallization mechanism still remain unclear.

This paper reports new data on the several variolite types recognized in the massive and pillow lava flows of the Yalguba Ridge and discusses the formation processes of these newly discovered variolites in comparison with the previously reported. Based on petrological, geochemical and mineralogical studies of various variolite types, the models of the Yalguba variole-bearing rocks formation are proposed providing new insights into the origin and formation of variolites.

2. Materials and Methods

2.1. Geological Settings

The Onega Basin is located in the southeastern part of the Archean Karelian Craton, Fennoscandian Shield within an area of more than 35,000 km² (Figure 1A) [24,25]. Meso-Neoarchean granite-gneiss and greenstone complexes prevail within the Karelian Craton [26]. These complexes are overlapped with angular unconformity by the Paleoproterozoic volcano-sedimentary sequence (2500–1700 Ma) in the Onega Basin [27].

Mafic volcanic rocks of the Suisari Formation (SF) comprise a significant part of the Onega Basin Paleoproterozoic succession. The SF in the Onega Basin is exposed mainly in the western part of the Onega Basin (Figure 1B) within an area of a. 2000 km² [27]. SF overlaps with a low-angle discordance or subconcordantly the volcano-sedimentary succession of the Zaonega Formation (ZF) (2060–1980 Ma) [24,25,27,28,29]. The mafic rocks of SF are dated 1976–1956 Ma [30,31].

SF succession has a thickness ranging from 0.3 to 1.0 km [24,27,32] and consists of massive, amygdaloidal, brecciated, and pillow lava flows interbedded with tuffs, hyaloclastites, tuffites, and tuff-conglomerates. The SF is accumulated in shallow-water environments and is related to continental flood basalts [30]. The intrusive component of SF includes gabbro, dolerite, and peridotite dykes and sills.

Paleoproterozoic volcano-sedimentary sequence (ZF and SF) within the Onega Basin was deformed into a system of roughly parallel, northwest–southeast trending folds, that overall form a synclinorium [24]. Metamorphism and hydrothermal activity in the Onega Basin are associated with Svecofennian orogeny 1890–1790 Ma [24,28]. According to the Rb-Sr isotopic data for the Onega Basin metasomatites, three peaks of hydrothermal activity have been established (1780–1700, 1600, and 1500 Ma) [33]. The SF volcanic rocks within the Onega Basin were metamorphosed under the prehnite-pumpellyite facies conditions (T: 290–320 °C, P: 1–3 kbar) [34].

The maximum thickness of the SF of the Onega Basin that is confined to the Ukshozero synform structure is more than 600 m. The SF successions of the Konchozero–Yalguba Ridge area have a thickness of about 340 m (Figure 1C) and mainly comprise lavas (up to 75%) [27]. The SF is subdivided into 5 subsections [24,27] (Figure 1C):

SF1 (thick 15–130 m) The lower unit of the SF1 at the border with ZF is composed of basaltic and picrobasaltic tuffs, tuffites, and tuff-conglomerates. The middle unit of SF1 comprises basaltic massive and pillow lavas (with lava flow thickness 3–15 m), interbedded with agglomerate, silty-lithic and psammitic tuffs. The upper unit is represented by a series of variolitic lavas with a thickness of 5–25 m.

SF2 (thick 0–90 m) is formed by tuffs, tuffites with cross and parallel bedding, interbedded with picrobasaltic massive lava flows.

SF3 (thick 5–140 m) is represented by lava flows (1.5 to 18 m thick) of massive and brecciated Pl-Cpx phyric basalts, interbedded with tuffs and tuffites.

SF4 (thick 20–200 m) lavas of massive augite basalts interbedded with crystal-clastic tuffs.

SF5 (thick 0–40 m) is formed by a series of basaltic pillow lava flows with agglomeratic tuffs.

2.2. Methods

Mineral, textural and chemical characterization of the Yalguba variolites was performed using optical microscopy, scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS), X-ray diffraction (XRD), X-ray fluorescence analysis (XRF) and inductively coupled plasma mass spectrometry (ICP-MS). All studies were carried out at the Institute of Geology, Karelian Research Centre, RAS (IG KRC RAS, Petrozavodsk, Russia).

In order to study the chemical composition of the variole and matrix separately, samples of the variolites with zoned varioles from massive lava and variolites with unzoned varioles from pillow lavas were cut into 5 mm thick slabs. The slab sections of these variolites were broken into millimetre-sized fragments, which were separated into light-coloured variole material and dark-coloured matrix material by hand-picking under a stereomicroscope. The accuracy of separation was controlled by X-ray analysis, which showed that matrix separates did not have any feldspar suggesting that separation was sufficient (Table S1, Supplementary Materials). In the case of the slab of spherulitic variolites with gradual variole–matrix interfaces, 5 mm diameter cylinders were drilled from the variole and matrix areas using a hand drill. Further, all samples were powdered in an agate mortar. Hand separation was not applied to the variolites with unzoned varioles from massive lava because of the small size of the varioles (less than 2 mm).

Scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS) was applied to determine the mineral composition of studied rocks. The SEM-EDS study was carried out on the polished sections with scanning electron microscope VEGA II LSH (Tesсan, Brno, Czech Republic) and energy dispersive microanalyzer INCA Energy 350 (Oxford Instruments, Oxford, UK). The analyses were performed under the following analytical conditions: W cathode, 20 kV accelerating voltage, 20 nA beam current, 2 µm beam diameter, and a counting time of 90 s. The following standards were used: calcite, albite, MgO, Al₂O₃, SiO₂, FeS₂, wollastonite, Fe, Zn, and InAs. SEM-EDS quantitative data and determination of the analysis accuracy were acquired and processed using the Microanalysis Suite Issue 12, INCA Suite version 4.07 (v.4.07, Oxford Instruments Analytical Limited, High Wycombe, UK). The accuracy of analyses was better than 1% for the major elements and 5% for the minor elements. Thin sections were coated by a carbon layer.

Powder XRD analysis was applied to determine the quantitative mineral composition of rocks and to check the accuracy of variole–matrix separation. Powder XRD analysis was carried out using a Thermo Scientific ARL X’TRA (Thermo Fisher Scientific, Ecublens, Switzerland) diffractometer (CuKα-radiation (λ = 0.1790210 nm), voltage 40 kV, current 30 mA). Samples were scanned from 2° to 90° 2θ, with a step size of 0.02° 2θ and a scan rate of 0.6° 2θ/min. The quantitative mineral composition of 5 samples was determined by the modelling of experimental diffraction curves using Siroquant software (v.3.0, Sietronics, Mitchell, Canberra, Australia). The detection limit for XRD phase identification was 3%.

The major element composition of the variolite samples was determined by X-ray fluorescence (XRF) using an ARL ADVANT’X-2331 (Thermo Fisher Scientific, Ecublens, Switzerland) wavelength-dispersive spectrometer with a rhodium tube, working voltage of 60 kV, working current of 50 mA, and resolution of 0.01°. Preliminarily, 2 g of each powdered sample was heated in ceramic crucibles at 1000 °C in a muffle furnace for 30 min. The loss of ignition was determined by a change in the mass of the sample after heating. For XRF measurements, 1 g of heated sample was mixed with Li-tetraborate flux and heated in an Au-Pt crucible to 1100 °C to form a fused bead.

Inductively coupled plasma mass spectrometry (ICP-MS) was used to determine the content of trace and rare earth elements. The ICP-MS analysis was carried out using the Thermo Scientific X-SERIES-2 quadrupole mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) following the standard procedure [35]. The powder samples obtained from the variolite and variole and matrix separates were decomposed using acid digestion in an open system. The analysis was controlled by measurement of USGS standard BHVO-2 (Table S2, Supplementary Materials). The concentrations of the trace element and rare earth elements were normalized to the concentrations in the primitive mantle and CI1 chondrite, respectively [36]. Several parameters were calculated: Nb anomaly as Nb/Nb* = NbN/(Th_N × La_N)^1/2, Sr anomaly as Sr/Sr* = Sr_N/(Pr_N × Nd_N)^1/2 and Eu anomaly as Eu/Eu* = Eu_N/ (Sm_N × Gd_N)^1/2.

3. Results

3.1. Field Study

The studied outcrops in the Yalguba Ridge represent the lower unit (SF1) of the Suisari Formation (Figure 1C). It is a continuous volcanic sequence, corresponding, probably, to a single eruptive cycle [27]. It consists of interbedding basaltic massive and pillow lavas (Figure 2A). In the lower parts of the section, the massive lavas have a thickness ranging from 1 to 5 m, whereas, in the middle and upper parts of the section, the thickness of lava flows of both massive and pillow lavas increases up to 15 m. The massive lavas are homogenous in composition and contain rare amygdales at the top of the lava flows. The pillow lavas comprise elongated pillows that vary in size from 0.3–0.4 to 4–5 m. The pillows have thin chilled zones (up to 1 cm) and occasionally have a well-developed sagging tail.

Variole-bearing rocks are well represented in the Yalguba Ridge section and are recognized in both massive and pillow lavas (Figure 2A). Based on the different morphology of varioles, the variole-bearing volcanic rocks from the Yalguba section can be subdivided into three main types: (1) variolites with well-pronounced zoned varioles (VZ-type), (2) variolites with unzoned varioles (VU-type) and (3) spherulitic variolites (S-type).

The VZ-type variolites are recognized in the massive lavas, including the thin (~2 m) lava flow in the lower part of the section (horizon 3) and lava flows with thickness varying from 4 to 25 m in the upper part of the section (horizons 9–11) (Figure 2A). VZ-type variolites from the lava flows 9–11 have been described in previous studies [19,22].

Here we present a detailed description of VZ-type variolites from lava flow 3. In the VZ-type variolites, 2–10 mm diameter spherical leucocratic varioles are embedded in the dark matrix (Figure 2B). The varioles exhibit sharp boundaries with the matrix and are concentric-zoned, with a greyish-green core and a lighter-coloured rim. The VZ-type variolites contain single varioles as well as juxtaposed and coalesced varioles (Figure 2C), which occasionally form lenticular clusters. In the VZ-type variolites, the varioles are mainly concentrated at the top of the lava flows with a variole/ matrix ratio of about 1:1.

The VU-type variolites occur in both massive and pillow lavas from the middle part of the section (horizons 6–8) with a total thickness of 30 m (Figure 2A). The light-coloured spherical varioles have sharp boundaries with the dark matrix (Figure 2D–G). They do not display any interior zoning. In massive lavas, the varioles have a small diameter of 1–2 mm. They are heterogeneously distributed with the areas contained sparse varioles and areas with large amounts of varioles commonly coalesced (Figure 2D). In the pillow lavas, variolitic texture is recognized in both the margins and core of pillows (Figure 2F). The varioles have larger diameters varying from 2 to 7 mm (Figure 2G). The abundance and size of varioles increase towards the centres of pillows, where the varioles are predominantly coalesced or juxtaposed.

The S-type variolites are recognized in the single thin (~2 m) massive lava flow in the lower part of the Yalguba section (horizon 4). Light-coloured varioles are embedded in the darker-coloured matrix; the contrast between the varioles and matrix is not as sharp as in the case of the VZ- and VU-type variolites (Figure 2H). They are unzoned with a slightly elongated shape and a diameter varying from 3 to 6 mm (Figure 2I). No clear (sharp) boundaries between varioles and matrix are observed in the S-type variolites. Varioles form about 15% of the rock and are uniformly distributed in the lava flow.

3.2. Petrography and Mineral Chemistry

All mafic volcanics in the Yalguba Ridge area preserved primary magmatic minerals and textures. Prehnite-pumpelleite to lower greenschist facies alteration resulted in the development of secondary albite, chlorite, prehnite, epidote, calcite and titanite.

Optical microscopy observations reveal that all variolitic rocks from the Yalguba Ridge are fine-grained clinopyroxene-phyric rocks. Clinopyroxene (augite) phenocrysts vary in size and composition and demonstrate various types of zoning [37]. The groundmass in all studied samples comprises augite microcrysts, plagioclase needles and altered volcanic glass. The amount of glass varies in various flows increasing in most fine-grained varieties, e.g., chilled zones of pillows. Sometimes groundmass contains fan-shaped grains of augite and clinopyroxene-plagioclase intergrows. In variolitic lavas, the varioles and matrix are well distinguished. Gas vesicles are recognized in both varioles and matrix in all variolite types, and their concentration is less than 5%. They have rounded, irregular or elongated shapes and sizes of 150–1000 μm. The vesicles are filled with quartz and chlorite and commonly have albite rims. Because studied rocks are fine-grained or cryptocrystalline, optical diagnosis is complicated. So, X-ray diffraction and SEM methods were applied to study the composition of varioles, matrix and minerals in detail.

X-ray diffraction analysis shows that clinopyroxene and feldspars constitute the essential mode in all types of Yalguba variolites (Table S1, Supplementary Materials). In the VZ- and VU-type variolites, the matrix does not contain feldspars but is enriched in clinopyroxene, chlorite and quartz compared to varioles. In contrast, in the S-type variolites, no significant differences in mineral composition between varioles and matrix are revealed.

SEM studies indicate, that in all samples, clinopyroxenes are mainly presented as zoned and unzoned augite phenocrysts (>100 μm) and microcrysts (<100 μm) with Mg# = Mg/(Mg + Fe²⁺) = 0.64–0.89. The amounts of clinopyroxene phenocrysts and microcrysts both in varioles and matrix are equal. The composition and textures of clinopyroxenes in varioles and matrix show no distinct differences (Figure S1, Table S3, Supplementary Materials). Clinopyroxene phenocrysts are abundant in the VZ-type and S-type variolites but rarely recognized in the VU-type variolites. Clinopyroxene phenocrysts occur as euhedral to subhedral grains with a size of up to 1 mm. They demonstrate a wide variety of zoning patterns, including normal-zoned with Mg-rich cores, reverse-zoned with Mg-poor cores, oscillatory-zoned with a Mg-poor core, Mg-rich mantle, and Mg-poor rim, and “homogenous” clinopyroxenes. These zoning patterns are similar to those described for clinopyroxene phenocrysts from the Yalguba variolites (VZ-type in the present study) by Svetov et al., 2020 [37]. The most abundant zoning patterns in the VZ-type variolites are normal-zoned and “homogenous” phenocrysts, whereas, in the S-type and VU-type variolites, the reverse zoning is dominant. In all samples, clinopyroxene microcrysts occur as euhedral, tabular or prismatic crystals with a size of ca. 10–100 μm, commonly forming glomeroporphyritic aggregates (Figure 3).

Feldspars are other abundant rock-forming minerals in the Yalguba variolites, but in contrast to clinopyroxenes, they exhibit considerable variability in composition and texture depending on the variolite type (Table S4, Supplementary Materials).

3.2.1. VZ-Type

The varioles in the VZ-type variolites are characterized by well pronounced zoning pattern with an inner zone composed of cryptocrystalline groundmass presented mainly by plagioclase with a minor amount of quartz and an outer rim composed of K-felspar (Figure 3A–D). Plagioclase compositions are in the range of andesine (An₄₂Ab₅₄Or₄) to pure albite with the Or content varying from 4% to 0% (Figure 4A). No regular variations of plagioclase composition from the centre to the margins of the variole core are observed. The K-felspar rim with a width of 300–500 µm is continuous and has a sharp interface with the variole core (Figure 3A,C). It has a composition close to the orthoclase end-member with significant enrichment in An (An₈Ab₀Or₉₂). Besides the clinopyroxene phenocrysts and microcrysts, the acicular microlites (ca. 10 μm length) along with dendritic crystals are embedded in the variole groundmass. They are mainly pseudomorphed by chlorite and titanite but rarely preserve the relicts of clinopyroxene. The varioles display a wide variety of dendritic patterns, including fernlike crystals in the centre and crystals with a comb texture in the outer parts of the variole core and the K-felspar rim (Figure 3B,C). The width of acicular or dendritic crystals decreases from 2–3 μm in the centre of the variole to 0.5–1.2 μm in the rim, whereas the length decreases from about 200 μm in the centre of the variole to 6–25 μm in the rim.

The matrix in the VZ-type variolites is composed of areas with fine-grained epidote and chlorite titanite, some of them are sporadically enriched in quartz. Titanite is presented by fine xenomorphic grains disseminated in the matrix (Figure 3D).

3.2.2. VU-Type

Varioles in the VU-type variolites are unzoned (Figure 3E,I) and composed of heterogeneous cryptocrystalline groundmass. In the VU-type variolites from massive lavas the variole groundmass has oligoclase-albite composition, but high-K patches close in composition to orthoclase (Ab_0–17Or_73–100An_0–8) occur near the variole–matrix margin (Figure 4B). Varioles also contain clinopyroxenes partially replaced by chlorite and titanite that occur as acicular crystals or fernlike crystals in the centre of the variole.

In VU-type variolites from the pillow lava, the variole groundmass is closed to andesine (Ab₄₀An₅₅Or₅) composition (Figure 4C) with irregular grains of secondary albite (Figure 3J). The variole groundmass contains acicular elongated chlorite and titanite grains presumably replacing clinopyroxenes. Microlites are randomly oriented and have a width of ca. 1–2 μm and a length of up to 200 μm (Figure 3F). The matrix in the VU-type variolites is similar for both massive and pillow lavas. It contains clinopyroxene phenocrysts and microcrysts embedded in chlorite–prehnite-epidote groundmass with a minor amount of quartz and titanite (Figure 3H,L). Near the varioles, it is enriched in chlorite, whereas moving outward the variole boundaries epidote becomes predominant (Figure 3E,I). The only difference in matrix composition between the VU-type variolites from massive and pillow lavas is that the matrix of VU-type variolites from pillow lava preserves optically isotropic areas that might be interpreted as devitrified volcanic glass.

3.2.3. S-Type

In the S-type variolites, the varioles are composed of spherulites formed by acicular plagioclase crystals with a width of 10–15 μm and length of up to 200 μm (Figure 3L,M). Plagioclase compositions range from andesine (An₄₈Ab₄₉Or₃) to oligoclase (An₂₇Ab₇₀Or₃) with decreasing of An component from the centre to the tips of spherulites (Figure 4D). The matrix is cryptocrystalline and has heterogeneous composition ranging from oligoclase (An₁₀Ab₈₇Or₃) to anorthoclase (An₁₃Ab₇₉Or₈). Noteworthy, is that Or content in matrix feldspars significantly scatters and reaches a high value of up to 20% (Figure 4D).

3.3. Variolite Rock Chemistry

The variolitic lavas in the Yalguba section are subalcaline rocks and range in composition from basalts (VU-type) to basaltic andesites (VZ- and S-type) and andesites (VZ-type) (Figure 5). VU-type variolites from the pillow lavas have the most primitive composition (9.4 wt % MgO), whereas MgO content decreases to 7.3 wt % in S-type, 6.3 wt % in VU-type from massive lava and 4.7–7.2 wt % in VZ-type. Bulk rock composition of all types of variolites has similar trends for most major and trace elements. They have high Al₂O₃ (Figure 6) and constant moderate TiO₂ content (1.3–1.4 wt %). The compositional similarity was also pronounced in the trace element distribution characterized by the depletion in high field strength elements (HFSEs) with slight Nb (Nb/Nb* = 0.71–1.04) and Sr (Sr/Sr* = 0.72–1.16) anomalies (Figure 7A). The Yalguba variolites have a REE distribution pattern with moderate LREE (light rare earth elements) enrichment ([La/Sm]_N = 2.05–2.5) and HREE (heavy rare earth elements) depletion ([Gd/Yb]_N = 6.15–8.69) along with the absence of obvious Eu anomaly (Eu/Eu* = 0.79–1.19) (Figure 7A,B). VU-type variolites from massive lava have some compositional characteristics that differ from the other variolites. They are characterized by evaluated CaO content, depletion in LILE (Rb and Ba) and higher positive (Sr/Sr* = 2.28) and Eu (Eu/Eu* = 1.57) anomalies (Figure 7A,B). These differences might be explained by local hydrothermal alteration.

The significant compositional differences are recognized between varioles and matrix from VZ- and VU-type variolites. The varioles have an andesite composition and the matrix corresponds to picrobasalt or basalt (Figure 5). In these variolites, the varioles have high SiO₂ and Na₂O + K₂O and low MgO and FeO_tot content compared with the matrix (Table S5, Supplementary Materials). The most noticeable differences in variole and matrix compositions are observed in VZ-type variolites. The VZ-type variole has high SiO₂ (61.3 wt %) and Na₂O + K₂O (5.9 wt %) content compared with the matrix with SiO₂ (51.7 wt %) and Na₂O + K₂O (0.3 wt %). In contrast, the VZ-type matrix is enriched in FeO_tot (14.9 wt %), CaO (11.8 wt %) and MgO (6.6 wt %) is approximately one-half or twice higher than in the VZ-type variole. VU-type variolites have a similar trend in the variole–matrix compositional variations. Likewise the bulk composition of variolites, the Al₂O₃ and TiO₂ contents do not show significant variations between the varioles and matrix (Figure 6, Table S5 in Supplementary Materials).

In contrast to VZ- and VU-type variolites, no significant difference between variole and matrix composition is recorded in S-type variolites. In these variolites, both varioles and matrix are andesibasaltic, similar to the bulk rock composition (Figure 5). Because the S-type variolites matrix–variole separation was complicated, the major element composition was calculated using SEM-ESD area estimation, which showed more apparent compositional differences (Table S6, Supplementary Materials). The varioles have low SiO₂ and Na₂O + K₂O and high CaO content. The matrix is enriched in SiO₂ and Na₂O + K₂O. No significant differences in MgO and FeO_tot are observed between the variole and matrix (Figure 6).

The trace element and REE distribution patterns of the variole and matrix are similar to that of the bulk samples, but there are some compositional differences between the varioles and matrix, depending on the variolite type. In VZ-type variolites the matrix has higher concentrations of REE compared to the varioles, although in VU-type variolites the difference was not recorded. Despite the varioles and matrix exhibiting similar REE distribution patterns with enrichment in LREE, in VZ-type variolite the LREE enrichment in the matrix is more apparent ([La/Yb]_N = 14.23) than in the varioles ([La/Yb]_N = 8.33) (Figure 7C). Additionally, the varioles have a minor negative Eu anomaly (Eu/Eu* = 0.86) similar to the bulk rock, whereas the matrix exhibits a minor positive Eu anomaly (Eu/Eu* = 1.19–1.34), that is not observed in the bulk rock. The differences between the varioles and matrix are also manifested in the trace element distribution. The varioles are enriched in LILEs (Rb and Ba) and show slight Sr anomaly (Sr/Sr* = 1.18–2.04), whereas the matrix is strongly depleted in LILEs and exhibits strong negative Sr anomaly (Sr/Sr* = 0.18–0.63) (Figure 7D).

In contrast to VZ-type and VU-type variolites, in S-type variolites, the varioles have higher REE content, than the matrix. The variole and matrix exhibit similar REE distribution patterns, but the varioles are more enriched in LREE ([La/Yb]_N = 11.49) than the matrix ([La/Yb]_N = 6.99). The S-type varioles show no Eu anomaly (Eu/Eu* = 1.05), whereas the matrix has a minor negative Eu anomaly (Eu/Eu* = 0.88) (Figure 7C). In S-type variolites, both variole and matrix resemble the trace element distribution pattern of bulk rock with slight depletion in HFSEs and a slight negative Sr anomaly (Sr/Sr* = 0.77–0.79) (Figure 7D).

Summarizing the rock chemistry data, the whole rock composition of the VZ-type and S-type variolites corresponds to the more evolved lava compared to those of VU-type variolites. It is evidenced by the low MgO and high SiO₂ content and enrichment in REE. VZ- and VU-type variolites show distinct compositional features between the variole and matrix. The varioles are enriched in SiO₂ and Na₂O + K₂O, whereas the matrix is characterized by a higher content of FeO_tot, CaO and MgO. Moreover, the matrix is enriched in REE with a minor positive Eu anomaly and strongly depleted in LILEs with a negative Sr anomaly. In the S-type variolites, the variole and matrix have similar compositions close to the bulk rock.

4. Discussion

In the SF sequence, the variole-bearing rocks are recognized in one of the five known volcanic cycles, namely in the lower part of the SF (Figure 1). The Yalguba Ridge is a typical section of the lower part (SF1) of the SF presented by a continuous sequence of interlayered massive and pillow lavas. Field relationships and compositional variations of the volcanic rocks in the Yalguba sequence argue that they were derived from a single parental magma. It is indicated by a similar distribution of trace and rare earth elements in the Yalguba volcanic rocks (Figure 7). An additional argument comes from the similarity in the trace element composition of clinopyroxene phenocrysts from the VZ-type variolitic and variole-free massive lavas recognized in [37]. Compositional variations in the Yalguba volcanic rocks from primitive (basalt, Mg# = 0.60) to evolved (andesite, Mg# = 0.49) lavas suggest various degrees of parental magma differentiation. Variolitic textures are recognized in both primitive (VU-type) and evolved (VZ- and S-type) lavas.

4.1. Variolite Types

Based on microtextures and chemical and mineralogical compositions, three variolite types are distinguished in the variole-bearing rocks from the Yalguba Ridge. Their summarized characteristics are presented in

Table 1. Classification of variolite types from the Yalguba Ridge.

Variolite Type	Variole Texture	Lava Flow	Rock			Mineral Composition			Origin
			Bulk	Variole	Matrix	Variole	Variole Rim	Matrix
VZ	zoned	massive	basaltic andesite, andesite	andesite	basalt	Ab, Q, Cpx	K-Fsp	Q, Chl, Ep	liquid immiscibility
						Cpx phenocrysts and microcrysts
VU	unzoned	massive	basalt	andesite	picrobasalt	Pl, Ab, K-Fsp, Q,	absent	Q, Chl, Ep, glass relicts	liquid immiscibility
		pillow		andesite	basalt	Cpx phenocrysts and microcrysts
S	spherulite	massive	basaltic andesite	basaltic andesite	andesite	Pl	absent	Pl, Q, Chl	undercooling
						Cpx phenocrysts and microcrysts

Notes: abbreviations: Ab—albite; Chl—chlorite; Cpx—clinopyroxene; Ep—epidote; K-Fsp—K-feldspar; Pl—plagioclase; Q—quartz.

. VZ- and VU-type variolites have similar features, including similar variole microtextures, and contrast chemical and mineralogical composition of globules and matrix. In contrast, S-type variolites have distinct variole microtextures compared to VZ- and VU-type variolites and no significant contrast between variole and matrix composition is observed.

Regardless of the microtexture and mineral distribution differences, all types of variolite have similar mineral assemblage mainly consisting of clinopyroxene and feldspars. In the variolites, clinopyroxenes are mainly presented as zoned and unzoned augite phenocrysts and microcrysts with Mg# = 0.64–0.89. Clinopyroxene morphology as well as their similar composition and homogeneous distribution between varioles and matrix suggest that most of them crystallized in a magmatic chamber before the eruption.

VZ- and VU-type variolites are characterized by contrast mineral composition of variole and matrix, since the matrix does not contain feldspars, but is enriched in clinopyroxene, chlorite and quartz compared to varioles (Table S1, Supplementary Materials). In contrast, in the S-type variolites, no significant differences in mineral distribution between varioles and matrix are revealed.

In contrast to clinopyroxenes, feldspars from the Yalguba variolites manifest a wide variety of textures and compositions. In S-type variolites, feldspars compose spherulites in the varioles along with matrix groundmass, whereas in the VZ-type and VU-type variolites, they form only the variole groundmass. These feldspar textures (devitrified glass or spherulite) suggest that crystallization of feldspars in the Yalguba variolites occurs during magma ascent and/or emplacement and further lava cooling. Despite the studied variolites experiencing albitization to various extents, the feldspar preserved the primary composition of plagioclase (andesine) and K-feldspar.

In the S-type variolites, feldspars occur as andesine spherulites embedded in the oligoclase–anorthoclase groundmass. Crystallization of spherulites needs intensive undercooling (>300 °C) [8,13,14,38]. The spherulite crystallization temperature might be roughly estimated to be around 700 °C with respect to the liquidus temperature of the andesine composition, which is about 1100 °C, using the plagioclase phase diagram for the P = 1 Atm [39]. The oligoclase–anorthoclase matrix groundmass is in disequilibrium with the andesine spherulites from the varioles and might crystallize at lower temperatures. However, the composition of matrix feldspar is highly variable and is significantly enriched in Or component up to 26% corresponding to the above solvus feldspar compositions (Figure 4A). So, it might be assumed that the feldspar matrix was solidified during the rapid cooling with further devitrification of glass. In summary, the textures and compositions of feldspars in S-type variolites suggest that their formation occurred under intensive undercooling. It is in agreement with the low thickness of the spherulitic lava flow that might provide rapid cooling.

In VU-type variolites, feldspars are recognized only in the varioles and demonstrate a wide variety of compositions, including (1) andesine variole groundmass; (2) patches with K-feldspar composition rarely recorded in the interfaces between the variole and the matrix; (3) irregular grains of albite embedded in andesine groundmass. The (1) and (2) feldspars might be interpreted as the antiperthites (the K-rich patches embedded in plagioclase groundmass) formed during the solvus decomposition during the lava cooling. The exsolution temperature was calculated using the Solvcalc program [40] to be around 650 °C (Table S7, Supplementary Materials). The formation of albite grains might be attributed to the later postmagmatic alteration of the andesine groundmass. The recognition of VU-type variolites in the chilled zones of the pillows suggests their rapid cooling. Additionally, rapid cooling of VU-type variolites is evidenced by the glass relicts recorded in the matrix. Moreover, the variole groundmass contains the acicular microlites of presumably clinopyroxenes partially or completely altered. Their rare occurrence and small size are likely caused by a short growth period terminated by quenching in the pillow margins.

In contrast to S-type and VU-type variolites, VZ-type variolites are characterized by zoned varioles with plagioclase core and K-felspar matrix. Cpx dendritic crystals are recognized both in the core and rim of the varioles indicating that their formation predated or was simultaneous with the variole zoning. Since the growth of dendritic crystal occurs at a high cooling rate [8], we might assume that the Cpx dendritic crystals in varioles of VZ-type variolites crystallized under the conditions of rapid cooling after emplacement at the initial stage of lava cooling.

The feldspars in the variole groundmass in the VZ-type variolites crystallize at the late stages of lava cooling. The groundmass of variole cores has andesine-albite composition with no regular variations of plagioclase composition from the centre to the margins of the variole. The plagioclase core is surrounded by the well-pronounced K-felspar rim composition close to the pure orthoclase. The K-feldspar from the rim has an unusual composition manifested in extremely high content of An component up to 10%, whereas the K-feldspar are reported to include not more than 5% [41]. The plausible explanation of this phenomenon might be the preservation of glass relicts or the presence of two feldspar exsolution phases with submicron size, which are hard to detect in SEM studies and are not recognized in the present study. Feldspar from the variole core and rim are not in equilibrium suggesting their separate crystallization. It might be assumed that the differentiation of varioles occurs due to the diffusion of K⁺ and Na⁺ ions when the varioles are still liquid. This process can occur during the relatively slow lava cooling, at least, compared to VU-type variolite.

The driving force of the diffusion might be the temperature gradient between the core and rim of the variole. Microlites and dendritic Cpx crystals recognized in the rim and outer part of the core have a low size compared to the dendritic Cpx crystal that occurred as fernflake in the centre of the variole. This is evidence that the outer part of the variole cooled faster than the interior. Noteworthy, the varioles in VU-type variolites have a similar trend in the distribution of feldspars manifested by the fact that varioles with mainly andesine-albite composition contain the K-feldspar patches at the variole margins. Though the zoning of the varioles in VU-type variolites is poorly resolved, the diffusion of K⁺ and Na⁺ ions might also contribute to its formation, but the process is limited because of rapid cooling).

4.2. Origin of Variolites

To explain the formation of variolitic textures four main hypotheses were suggested including (1) mixing of contrasting felsic and mafic melts [1,2]; (2) liquid immiscibility [2,3,4,5,6,7], (3) rapid cooling of superheated magmas [8,9] and (4) postmagmatic hydrothermal or metamorphic alteration of basaltic glass [10].

The alteration of volcanic rocks can produce variolitic texture by overprinting of a pre-existent spherical texture (roundish voids, vesicles or spherulites) during the metamorphic transformations [9,10]. The Yalguba variolites have almost unaltered varioles and lack vesicles, so metamorphic alteration can be ruled out to explain their origin.

In VZ- and VU-type variolites the varioles of andesite or basaltic andesite composition are scattered in the basaltic or picrobasaltic matrix (Figure 5). This distinct contrast in the major element composition of varioles and matrix might be produced by both liquid immiscibility and magma mingling. The previous studies of the Yalguba variolites (VZ-type in the present study) proposed that liquid immiscibility was likely to be the mechanism of their formation [19,21]. This assumption is supported by specific textural features typical for liquid immiscibility including coalescence of varioles and sharp interfaces between varioles and matrix. The present study shows, that the VU-type variolites have chemical and textural characteristics similar to the VZ-type variolites and meet the criteria of liquid immiscibility origin.

Liquid immiscibility should provide a significant difference in the distribution of not only major but also minor elements between the varioles and the matrix [42,43]. Complementary distribution of major elements, namely the enrichment of SiO₂, Na₂O, K₂O in the varioles and MgO, FeO and CaO in the matrix, has been observed in previous [19,21] and present studies of VZ-type variolites from the Yalguba massive lavas and in the present study of VU-type variolites.

Model experiments on liquid immiscibility [42,43,44,45] indicated that HFSE (Zr, Hf, Nb, Ta, Y, Ti, P) and transition metals (Ni, Cr, Co, Sc, V) are enriched in Fe-rich melt (matrix in the studied variolite), whereas LILE (Cs, Rb, K, Na) are enriched in the siliceous melt (varioles in the studied variolite). However, Gudin et al., 2012 [21] reported that trace element composition of varioles and matrix in Yalguba variolites (VZ-type in the present study) are similar and no redistribution of these elements between varioles and matrix has been observed, which is controversial with the liquid immiscibility hypothesis. The present study of the VZ- and VU-type variolites shows the apparent differences between the variole and matrix in LILE composition. The varioles are enriched in LILEs (K, Rb and Ba), whereas the matrix is strongly depleted in LILEs and exhibits a strong negative Sr anomaly (Figure 7D). Moreover, the matrix is enriched in REE and exhibits a minor positive Eu anomaly, whereas the varioles exhibit a minor positive Eu anomaly. The enrichment of LILE in varioles and REE enrichment in the matrix are constituents of the liquid immiscibility hypothesis [42,43,44,45]. However, our study did not reveal any differences in transition element (Cr and Ni) (Figure 6) and HFSE (Figure 7) content between the variole and matrix in VZ- and VU-type variolites. The similar indistinguishable Cr and Ni content in varioles and matrix were reported for variole-bearing rocks from Mount Ada Basalt of the Pilbara Supergroup, Western Australia, which were also proposed to originate from liquid immiscibility [7].

VZ- and VU-type variolites of the Yalguba Ridge are unlikely to originate from the mixing of contrasting felsic and mafic melts for the following reasons: (1) similar trace element distribution in varioles and matrix; (2) similar composition of clinopyroxene phenocrysts and microcrysts in varioles and the matrix indicate that their crystallization preceded the liquation processes; and (3) magma mingling should be reflected in gradual boundaries between the varioles and the matrix caused by reactions between non-equilibrium melts, that were not observed in the studied variolites.

Variolitic texture in S-type variolites is associated with andesine spherulites formed by the varioles, which are embedded in the albite matrix. So the spherulitic crystallization does not result in the establishment of compositional contrast between variole and matrix. Spherulites might be interpreted as characteristic supercooling textures, which form when a heated magma is emplaced due to eruption from depth and is rapidly cooled on or near the surface [8,14]. This process provides the lack of nucleation sites for mineral growth combined with comparatively high supersaturation and results in spherulite growth. Evidence of the rapid cooling governs the spherulite formation in S-type variolites of the Yalguba Ridge comes from (1) the low thickness of S-type variolite flow, which provides the rapid cooling of lava; and (2) the uniform distribution of varioles within the lava flow is constituent with the simultaneous nucleation of spherulites in the whole lava flow. However, spherulitic textures might also result from devitrification [8,46,47]. To distinguish spherulite crystallization from primary magmatic melt or devitrification is a challenge [48,49]. The commonly applied criterion of devitrification is the association of spherulite formation with fractures [14,48,49]. So the absence of fractures within the spherulites and matrix in the S-type Yalguba variolites contributes to the primary crystallization from a melt. It has been shown [50] that increasing cooling rate led to Na⁺ and K⁺ depletions in plagioclase. This trend is constituent with the low concentration of these elements in spherulites and Na⁺ and K⁺ enrichment in the matrix which solidificated after spherulite crystallization. Moreover, the regular decreasing of An component from the centre to the tips of plagioclase spherulites (Figure 6) is an additional argument of primary spherulite crystallization. Devitrification occurs below the glass transition temperature, so it is unlikely that the spherulites of high-temperature composition (andesine) can grow at a low-temperature matrix (albite). Therefore, these mineralogical and textural observations lead us to suggest that the spherulites have a primary magmatic origin due to the rapid cooling of superheated magma.

4.3. Liquid Immiscibility vs. Undercooling

Variolitic textures in the Yalguba variolites could be produced by liquid immiscibility (VZ- and VU-type) or rapid cooling (S-type). Since the variolitic texture of liquid immiscibility is recognized in the quenched zones of pillows and both in massive and pillow lavas it might be assumed that its formation is independent of eruption conditions and the melt separation caused by liquid immiscibility has occurred before the eruption, in the magmatic chamber or during magma ascent. In contrast, spherulite formation in S-type variolites is caused by the rapid cooling of superheated magma after eruption. The bulk rock geochemical data show that variolites with similar origins were recognized in the Yalguba volcanic rocks with different compositions, e.g., basalts and basaltic andesites (Table 1). However, the variolitic lavas with similar composition (basaltic andesite) produce distinct types of variolitic textures, produced by liquid immiscibility (VZ-type) or rapid cooling (S-type).

Based on these results, it might be assumed that the formation of variole-bearing rocks might be caused by various variations of P–T conditions of magma ascent and eruption, rather than by the compositional differences between variole- and non-variole-bearing rocks. The degree of H₂O saturation of the parental magmas might also be the factor to provide the textural differences in the Yalguba volcanic rocks. The Yalguba variolites are part of the lower unit (SF1) of the Suisaari Formation and are interpreted as the initial volcanic cycle [24,27]. It is associated with the explosive regime evidenced by the formation of the thick layers of pyroclastic rocks at the bottom of the Suisaari sequence suggesting that the parental magmas were enriched in H₂O and gasses. Further accumulation of the residual H₂O saturated melts in the magmatic chambers accompanied by the transformation of explosive volcanism to extrusive one provide the formation of variolitic lavas, including that of from the Yalguba Ridge. During the further evolution of volcanism, the water amount is exhausted which leads to the formation of only non-variole-bearing massive and pillow lavas recognized in the four upper volcanic sequences of the Suisaari Formation (SF2–SF5).

The question arises as to why the liquid immiscibility occurs to produce the VZ- and VU-type variolites and is not manifested in the S-type variolites. Since the trigger of liquid immiscibility might be the H₂O saturation of parental magma [51], we might assume that the VZ- and VU-type variolites originate from the hydrous magma, whereas the formation of the spherulitic variolites is caused by anhydrous magma. The evidence pointing to this assumption is the trend of mineral crystallization after magma ascent and/ or emplacement. Crystallization of plagioclase is suppressed at high H₂O content [52]. The VZ- and VU-type variolites meet this criterion since only clinopyroxene dendritic and acicular crystals are recognized to be the first phase crystallizing during the magma emplacement, so we might assume that the corresponding parental magma was H₂O-saturated. In contrast, S-type variolites contain andesine spherulites, as well as non-variole-bearing massive and pillow lavas in the Yalguba Ridge [37] contain acicular plagioclase crystals, which might indicate that the parental magma is H₂O-undersaturated.

However, no geochemical evidence was obtained to distinguish H₂O-saturated or -unsaturated magmas caused the formation of the Yalguba variolites. Hydrous magma has characteristic geochemical signatures, such as low Nb/Th < 7 [53]. Variolites of various origins VZ-type and S-type have low Nb/Th ratios of 4.7 and 5.0, respectively, whereas VU-type variolites assumed to originate from liquid immiscibility are characterized by higher Nb/Th value of 8.3 similar to massive lavas and pillow lavas (Nb/Th = 7.4–10.3).

Enrichment in H₂O might be caused by fractionation of the melt in the magmatic chamber. Since the UV- and S-type variolites have similar MgO content (Figure 6), this assumption can be excluded. H₂O saturation of parental magma might also occur due to contamination. All types of variolites from the Yalguba Ridge have similar chemical composition and manifested no geochemical evidence of significant crustal contamination because no obvious positive Zr-Hf anomaly was observed in these rocks (Figure 7), whereas felsic crustal materials is characterized by significant enrichment in Zr and Hf. Additionally, the La/Nb ratio (1.00–1.18) in Yalguba variolites is lower than in the continental crust (La/Nb = 1.6–2.6) [54]. The obtained data indicate insignificant felsic crustal contamination. Although Puchtel et al., 1998 [30] reported on contamination of Suisari lavas from the Onega Basin, the obtained data indicate insignificant felsic crustal contamination in the studied variolites. So, we assume that the possible source of H₂O enrichment in the parental magmas might be associated with the contaminants, whose contribution to the trace element composition is hardly to be recorded. One of the possible sources might be the assimilation of Paleoproterozoic salt sediments. The sediments compose the Jatulian (2.3–2.1 Ga) volcano-sedimentary succession and underlay the Suisari Formation, which includes the Yalguba volcanic rocks [55]. They are presented mainly by halite and anhydrite, so their contribution to the trace element composition is hardly to be recorded. Another possibility is the assimilation of organic carbon which could also cause liquid immiscibility [56]. The Paleoproterozoic succession of the Onega Basin includes carbon-rich rocks, which are interpreted as a giant Paleoproterozoic oilfield [57]. The assimilation of organic carbon in the Yalguba section was recognized in the basalt pillow lavas underlying the studied variolitic lavas. These rocks contain carbonaceous matter in the amygdales, inclusions in feldspar, and devitrified volcanic glass and are characterized by high vesicularity. No carbonaceous matter has been recorded in the studied variolitic lavas and they have low vesicularity. However, it might be assumed that after the assimilation the organic matter caused the liquid immiscibility of the melt in the magma chamber, but further was degassed during the magma ascent.

The lower H₂O content in parental magma of the spherulitic variolites could be explained in several ways. The first is a replenishment of an intermediate magma chamber by magmas different in H₂O content. It is constituent with the fact that crystallization of all types of variolites took place in an open system as has been indicated by analysis of clinopyroxene phenocrysts in [37] and the present study. Another possibility is that the parental magmas of the variolites of all types might have relatively high H₂O content at least compared with the non-variole-bearing lavas from the Yalguba Ridge. In this case, the precursor magma of S-type variolites is assumed to lose H₂O due to the magma degassing, which might occur in the magmatic chamber as well as during the magma ascent. Distinguishing between these possibilities is a challenge, but the first hypothesis seems to be likely since the sequence of the volcanic rocks from the Yalguba Ridge is mainly presented by non-variole-bearing massive and pillow lavas as well as S-type variolite that might be interpreted as originating from the anhydrous melt. However, the second assumption also cannot be excluded.

5. Conclusions

Variolites are widespread in the Paleoproterozoic (ca. 1.97 Ga) volcanic sequence of the Yalguba Ridge (Suisaari Formation, Onega Basin, Russia) and are worldwide known. The present detailed field and petrological studies reveal that variolites of the Yalguba Ridge display a wide variety of variolitic textures. In addition to previously reported liquid immiscibility variolites from the Yalguba Ridge [19,21], two new types of variolites were recognized in the volcanic sequence of the Yalguba Ridge including variolites with unzoned varioles (VU) and spherulitic variolites (S-type).

Similar to the previously described variolites (VZ-type in the present study), variolites with unzoned varioles have contrasting chemical and mineralogical compositions of varioles and matrix, that support an origin by liquid immiscibility. In these rocks, variolitic textures were recognized in chilled margins of the pillows suggesting that melt separation caused by liquid immiscibility occurred before magma emplacement. In variolites with liquid immiscibility origin the differences in variole microtextures namely the formation of zoned and unzoned varioles, are caused by the various cooling rate after the magma emplacement.

Spherulitic variolites have textural and compositional characteristics distinct from VZ- and VU-types. In S-type variolites, the varioles are composed of plagioclase spherulites with no significant contrast between variole and matrix composition. The mineralogical and textural features of S-type variolites suggest that the spherulites have a primary magmatic origin due to the rapid cooling of superheated magma.

Variolites with liquid immiscibility origin are most abundant in the Yalguba Ridge volcanic rocks. The variety of variolitic textures in the Yalguba section might be caused by the H₂O content in parental magma and cooling conditions. Hydrous parental melt provides the liquid immiscibility caused by the formation of VZ- and VU-type variolites, whereas anhydrous melt provides the formation of S-type variolites.