Spherulites as Recorders of Multi-batch Felsic Magma Transport and Time Scales: A Case Study from Southeast of China

Abstract

Granite, formed by the consolidation of felsic magma, is widely exposed on the surface of the Earth’s continents. However, the physical processes behind the transport of felsic magma from the deep crust to the surface remain poorly understood. Recently, the fracture-dike transfer model has gained widespread acceptance as a comprehensive theory explaining the ascent, transport, and emplacement of felsic magma in the crustal magma plumbing. Observing the magma migration process has been challenging due to the hidden geological evidence. Currently, no geological feature accurately tracks the transport path, composition evolution, and time scale of magma during its ascent. In this paper, we report the discovery of numerous spherulites that developed in two spherulitic rhyolite porphyry dikes. Based on the petrographic and geochemical characteristics of spherulites, we established that the ascent and emplacement processes of the magma were caused by continuous supply from deep magma sources and repeated episodes of ascent. Finally, magma source depletion halted its ascent and stabilization of its location at a depth of about 3 km below the ancient surface. According to the growth rate of the spherulites when the magma resided in the magma chamber, we quantitatively calculated that the magma resided in the three magma chambers for a total residence time of about 90 days. These findings provide valuable insights into the accumulation, ascent, and emplacement of multiple batches of magma into large rock bodies. These insights advance the understanding of the felsic magma plumbing system and provide constrained quantitative data on the dynamics of magma migration in the subsurface magma plumbing systems of volcanoes. Natural spherulites usually occur in rhyolite and obsidian around volcanic institutions. The magma migration path recorded by spherulites in this case study and the established spherulite geochronological method should have general applicability to a larger range of volcanic rock areas with similar geological backgrounds.

1. INTRODUCTION

The transport of felsic magma (SiO₂ > 65 wt %) from the deep crust to shallow levels remains poorly understood. In recent decades, the application of modern geophysics, geochemistry, and volcanology (Peltier et al., 2009; Magee et al., 2018) has led to the development and refinement of the fracture-expanding felsic magma transport theory (Clemens & Mawer, 1992). This theory suggests that felsic magma rapidly ascends through propagating fractures, e.g. dikes, in extremely short periods (Clemens & Mawer, 1992). Dynamic models on time scales of months to centuries are replacing the once popular view that the generation of felsic magma is a slow, balanced process that takes millions of years to complete (Petford et al., 2000).

Large intrusions are formed via the emplacement and cumulative assembly of many small magma pulses (Coleman et al., 2004; Glazner et al., 2004; Annen, 2011). A common theme is that magma transport typically occurs via the frequent transport of small volumes rather than via the infrequent transport of large volumes (Glazner et al., 2004). Dense networks of fracture-spreading magmatic plumbing systems (Ma et al., 1998; Gudmundsson, 2012; Cashman et al., 2017; Ma & Li, 2017; Ma et al., 2020) exist beneath large granitic batholiths composed of aggregations of small dikes (Zellmer & Annen, 2008). Based on the consensus, a fracture-dike transport theoretical model of felsic magma transport has been established (Baker, 1998; Cruden, 1998; Petford et al., 2000; Cruden & Weinberg, 2018). Magmatic systems are thus considered to be vertically extending, transcrustal, interconnected networks of magma conduits and magma/paste reservoirs (Cashman et al., 2017; Cruden & Weinberg, 2018; Magee et al., 2018). However, this model of magma transport still lacks continuous transport details and quantitative time-scale evidence, and few examples have been reported so far.

Mesozoic continental volcanic rocks are exposed on a large scale along the southeastern coast of China, and a large number of dike groups often developed in the late stage of volcanic eruptions. These dike groups provide excellent examples of the connection of the deep magma reservoir to the magma plumbing system at the shallow depths. The spherulitic rhyolite porphyry dike groups are distributed around the secondary volcanic institutions in the Dongkeng volcanic depression, and a large number of dike groups developed during the Late Jura–Early Cretaceous large-scale volcanic eruption in this region (Fig. 1). Previous research has shown that the spherulites in the rhyolite porphyry formed at the beginning of the ascent stage of the immiscible evolution of the deep magma and its ascent until the magma reached near surface depths and consolidated (Zhu et al., 2022). Therefore, the spherulite growth trajectory may be the key to unlocking the channel connecting the deep magma reservoir to the shallow surface. The purpose of this study was to investigate how spherulites record the transport path of multiple magma batches, to determine the migration time scale of the felsic magma, to reconstruct the network structure and geometry of the magma plumbing system, and to determine the locations of economic deposits in the magma channel. The data acquired are of great significance for improving the theory of magma migration and the understanding of the mechanism of modern volcanic activity.

2. GEOLOGIC BACKGROUND

The study area is located in the Zhejiang, Fujian, and Guangdong volcanic activity belt (class II; Fig. 1a) along the southeast coast of China in the southwestern part of the giant volcanic eruption belt bordering the Pacific Ocean (class I; Fig. 1a). The Dongkeng volcanic depression (class IV; Fig. 1b and c) is located in the northwest corner of the Ningde ring volcanic eruption center (class III; Fig. 1b). The volcanic rock belt extends from Zhejiang in the north, through Fujian, Guangdong in the south, with a length of 1000 km from north to south and a width of 200–300 km from east to west. The late Yan-Shan volcanic eruption entered its peak period, and the ejections formed several giant volcanic basins oriented in the NE direction (Fig. 1b). A set of single peak andesite–dacite–rhyolite rock associations formed in the Late Jurassic and Early Cretaceous. The basalt and andesite–rhyolite formed in the middle and late Early Cretaceous and have characteristic bimodal association of mafic and felsic rocks.

A set of Late Jurassic (J₃) to Early Cretaceous (K₁) multi-cycle continental felsic volcanic rocks and pyroclastic rocks developed in this area, accompanied by multiple stages of sub-volcanic dike intrusion (Fig. 1c). According to the regional geology of Fujian Province (Fujian Geology Survey Institute, 2016), the volcanic rocks and pyroclastic rocks can be divided into the third member (J₃-K₁n³) and the fourth member (J₃-K₁n⁴) of the Nanyuan Formation in the Late Jura-Early Cretaceous, and the Huangkeng Formation (K₁h) and Xiazhai Formation (K₁z) in the Early Cretaceous (Fig. 1c). The K–Ar age of basalt in the Yongtai Standard section of the Huangkeng Formation (K₁h) is 113.2 Ma (Fujian Provincial Bureau of Geology and Mineral Resources, 1985). The K–Ar age of rhyolite in Yongtai standard section of the Xiazhai Formation (K₁z) is 106.8 Ma (Fujian Provincial Bureau of Geology and Mineral Resources, 1985).

Based on its Nd and Sr isotope compositions, the Mesozoic volcanic magma in Fujian has the characteristics of a mixed crust–mantle source (Huang et al., 1986). In addition, the results of seismic and magnetotelluric sounding suggest that the felsic magma was formed at the intracontinental crust–mantle boundary (depths of 28–31 km) (Wang et al., 1993) through the mantle bulge and via lower crust deep melting. Large-scale upwelling and eruption of Late Jurassic-Early Cretaceous magma formed a thick combination of continental volcanic rocks and intrusive rocks. After the large-scale volcanic eruption, the continuous tectonic extensional environment in the mantle ascent region led to expansion of the upward channel, creating a magma plumbing network in the hot crust; the magma migration in the hot crust was freed from the constraints of imposed by cooling (Weinberg, 1999). In the Late Cretaceous, the magmatic activity shifted from large-scale volcanic eruptions to transport through a magma plumbing system and ascended in multiple pulses to reach the middle and upper crust, forming a (spherulite-bearing) felsic dike swarm and a small group of stocks around the early volcanic structures. The average denudation rate of the volcanic rocks from the Jurassic (J) to Cretaceous (K) in Fujian was 2.5 km/100 Ma (Li, 2008). Therefore, after diagenesis of the spherulitic rhyolite porphyry (~120 Ma), approximately 3 km of the overlying country rock was denuded.

The dike groups in the study area are frequently exposed in concentrated areas and are typically located in or around the center of the early volcanic structure (Fig. 1c, e), forming multiple magma plumbing clusters. The main lithologies of the magma plumbing system components exposed at the surface are granite porphyry (γπ), syenite porphyry (ξπ), quartz rhyolite porphyry (Qπ), and spherulite rhyolite porphyry (λπ).

The granite porphyry (γπ) is distributed in a plum blossom pattern with four small stocks at the center of the Dongkeng volcanic depression. The stocks have diameters ranging from 0.5 to 1.5 km and are nearly vertically orientated. These stocks intruded into the K₁ volcanic rocks.

The syenite porphyry (ξπ) is distributed in two locations: the southern and northwestern parts of the Dongkeng volcanic depression (Fig. 1c). In the southern part of this depression, a group of syenite porphyry dikes occurs within an area approximately 1 km wide and 3 km long area, including up to 10 individual dikes. The individual dikes are 800 to 2000 m long and 5 to 10 m thick. Controlled by the NW-trending shear zone, the dikes are strictly parallel and equidistantly spaced. They trend northwest with a dip angle of 90 ± 10°. In the northwestern part of this depression, the syenite porphyry intruded into the K₁ volcanic rocks and cut through the quartz rhyolite porphyry dikes. The U–Pb isotopic age of a single zircon from this syenite porphyry is 96.8 ± 0.2 Ma (Xiao et al., 2020).

The quartz rhyolite porphyry (Qπ) is distributed around the secondary volcanic structures in the Dongkeng volcanic depression (Fig. 1c). Among the exposed quartz porphyry dikes, one of the largest scale trends NE 50°and has a dip angle of SE 40°. It is 650 m long and approximately 5–60 m wide. This quartz porphyry forms a magmatic plumbing system that strictly controls the distribution of the cryptoexplosive breccia-type gold deposits. Exploration in the mining area has verified that the porphyry and the gold ore body within it extend to a depth of more than 800m (Zhu et al., 2021).

The spherulitic rhyolite porphyry (λπ) is distributed around the secondary volcanic structure in the Dongkeng volcanic depression (Fig. 1c). It is not exposed at the surface. Two spherulitic rhyolite porphyry dikes were investi-gated using data from four drill holes (Fig. 1d). They intruded into the volcanic breccia lava (Fig. 2b). The controlled dikes are more than 200 m in length, extend to a depth greater than 100 m, and have an average thickness of about 25 m. The dike strikes nearly east–west, dips to the south, and has a dip angle of 30–40° (Fig. 1d). The lithological characteristics of the spherulitic rhyolite porphyry are described in detail in the next section. In terms of the compositional characteristics of the magma, studies (Zhu et al., 2022) have shown that the drastic change to supercooling at about 900 °C caused the magma to separate into two immiscible liquid phases: a silicon-rich spherulite phase enriched in SiO₂, K₂O, and Na₂O and an iron-rich matrix phase, which was enriched in ∑Fe₂O₃, Al₂O₃, TiO₂, and MgO. With the material exchange and migration between the two phases, the proportion of the spherulite phase increased, and its Si₂O content stabilized at 72.92 to 74.52 wt %, while the matrix phase decreased, and its Si₂O content gradually decreased from 73.30 to 67.62 wt %.

3. MATERIALS AND METHODS

3.1. Core samples

The analytical samples were collected from well ZK10005 at depth (Fig. 1d), and the rocks were not subjected to hydrothermal alteration and surface weathering, so they remained fresh and representative. We collected samples from three cores 51-11/17, 52-4/11, and 53-10/13 from depths of 137.10, 140.30, and 145.60 m, respectively. The samples underwent EMPA, X-ray diffraction, wavelength dispersive spectrometry (WDS), and transmission electron microscopy analyses (Table 1).

3.2. Electron microprobe analysis

Electron microprobe analysis was conducted to obtain the major element compositions of the different generations of spherulites and the light and dark concentric growth rings in the spherulites. EMPA samples were prepared from each core sample (Table 1) by cutting and grinding the rocks into 0.05 mm thick sections. A JEE 420 evaporative coater (JEOL, Tokyo, Japan) was used to apply a thin coating of carbon. Carbon coating for both the thin section and standard samples was performed simultaneously, and the thickness of the carbon coating was 20 nm. The EPMA and EDS were conducted at the Zijin School of Geology and Mining laboratory at Fuzhou University in Fuzhou. The analyses were performed using a JEOL JXA-8230 Electron Probe Microanalyzer (JEOL, Tokyo, Japan) with an acceleration voltage of 15 kV, a probe current of 20 nA, and a beam diameter of 5 μm. Spherulite samples from each generation were selected in the slides for the downstream analyses. The light and dark concentric growth rings were analyzed separately, and the different spherulite components were analyzed quantitatively. The standard sample used in the analysis was selected according to the General Principles for Quantitative Analysis using Electron Probes and X-ray Energy Spectra under a Scanning Electron Microscope (national standard GB/T17359-98) (State Bureau of Quality Supervision, 1998). The chemical formulas of the spherulites were calculated based on the oxygen method (O = 8) (Bulakh et al., 2014). The loss on ignition values of the analyzed whole-rock spherulites and matrix samples are 3 to 4 wt % (Zhu et al., 2022). Since EPMA cannot determine the loss on ignition, the sum of the measured EPMA oxide contents (Table S1) is little less than 100%. However, this does not affect the accuracy of the analysis of the other individual oxides.

To test the reliability and repeatability of the EPMA data for the samples, repeated tests were carried out at the original location of the same test point (that is, 100% internal test), and another spherulite sample (W-9) was selected and sent to the external laboratory for testing (that is, external testing). For samples W-7–2, W-8–1 and W-8–2, the results for the measured points of 88 EPMA sites yielded the weighted average chemical formula of K_1.025Na_0.017Fe_0.007Al_1.108Si_2.903O₈ (Tables S1 and S2). The formula based on the repeated test results for 88 sites is K_1.018Na_0.017Fe_0.008Al_1.108Si_2.910O₈. For sample W-9, the test results for 25 points yielded the average chemical formula of K_1.054Na_0.012Fe_0.002Al_1.075Si_2.926O₈. Comparison of the internal and external analysis results confirmed that the selected spherulite samples and the test results are reliable and reproducible.

3.3. X-ray powder diffraction analysis

X-ray powder diffraction analysis was conducted to determine the mineral unit cell parameters and the mineral compositions of the spherulites. The samples from group WX-1 were collected from core 53–10/13 (Table 1) and were coarsely crushed into particles smaller than 1 mm. Then, pure sample fragments were selected under a microscope and ground into fine powder. The X-ray diffraction analysis was conducted on a single spherulite (WX-1) at the X-ray laboratory of Zijin College of Geology and Mining, Fuzhou University using a D8 Advance A25 X-ray Powder Diffractometer (Bruker, Germany) and a 192-channel LynxEye detector. The analysis was performed using a Cu target, a tube pressure of 40 kV, a tube flow of 30 mA, and an X-ray beam diameter of 30 μm. The X-ray scattering was recorded between 15° and 65° with a 0.02° step size and 0.1 s per step.

3.4. TEM analysis

Using transmission electron microscopy, we conducted phase identification of the cryptocrystalline minerals that were not identifiable at the resolution of the optical microscope. We selected three TEM test samples (Table 1) based on the data from the rock and mineral identification, EMPA, and X-ray powder diffraction analyses, and then, we conducted TEM analysis after sample preparation. Samples IBT-2 and IBT-b were collected from cores 51-11/17 and 53-10/13, respectively (Table 1), and were prepared using the ion-beam-thinning (IBT) method. Sample FIB-c was collected from core 53-10/13 (Table 1) and was prepared using the focused-ion beam (FIB) method. The sample preparation and TEM analysis were performed at the Sinoma Institute of Materials Research in Guangzhou. Samples IBT-2 and IBT-b were prepared using a GATAN Precise Ion Polishing System (model 695, USA). The sample was thinned using the ion beam until it was perforated, with an ion beam energy of 5 keV and an ion gun angle of ±8°. The sample was then further thinned at 4.5 keV and an angle of ±6° for 5 min, at 4.0 keV and ± 4° for 5 min, and finally at 3.0 keV and 3° for 5 min. Sample FIB-c was prepared using a Helios 5 UX dual-focused ion beam scanning electron microscope (USA). The parameters were as follows: an ion gun, gallium source, acceleration voltage of 500 V to 30 kV, beam current of lpA-65 nA, resolution of 2.5/4.0 nm (30 kV), and electronic secondary resolution of 0.6 nm (30 kV). The prepared sample was approximately 6 μm × 4 μm × 0.05 μm (length × width × and thickness). The TEM analysis was conducted using a Talos F200X (FEI, USA) and the following parameters: acceleration voltage of 200 kV, TEM point resolution of 0.25 nm, line resolution of 0.14 nm, information resolution of 0.12 nm, and energy dispersive spectrometer (EDS) resolution of 136 eV.

3.5. Derivation of the relationship between spherulite growth and magma evolution

Based on petrographic observations and corroborated by chemical analysis data, we compared our results with previous experimental findings to infer the relationship between the spherulite growth and magma transport. If the petrological structure and chemical composition were identical, it was concluded that the material formed under the same environmental conditions. Specifically, if the spherulites exhibited uniform structures and identical chemical compositions, it was concluded that they grew in a stable environment and had sufficient time to reach chemical equilibrium. Such environmental conditions during magma transport can only be met in the resting periods within a magma chamber. Therefore, we conclude that the spherulites with uniform structures and identical chemical compositions grew during the resting periods in the magma chamber.

As magma ascends and cools, quenching inevitably occurs at the edges of spherulites, leaving features that may appear as dark quenched edges, similar to the dark layer formed when a red-hot piece of iron is suddenly immersed in water. These quenched edges and radial fiber structures can be observed under a high-power microscope. The growth of spherulites occurs during the intermittent periods of magma, and the growth time of each generation of spherulites corresponds to the magma chamber residence time. The growth of the generations of spherulites during rest periods in the magma chamber corresponds to the quenched edges and radial fiber structures form during cooling of the magma during its ascent. This allows us to trace the ascent and resting paths of the magma. Compositional changes that occur systematically during the growth of generations of spherulites should be regarded as a result of compositional evolution.

4. RESULTS

4.1. Petrological characteristics of spherulites

The spherulitic rhyolite porphyry represents an ultra-shallow intrusive rock body, characterized by three distinct lithological components: the spherulites, matrix, and rhyolite strips (Fig. 2a–c). The large spherulites with diameters of 8–10 mm is more abundant than the small spherulites with diameters of 1–4 mm. The spherulite content is 5–20% in the inner contact belt of the dike, and it increases to more than 70% in the center of the dike (Fig. 2a and c). The larger spherulites were compressed, elongated, and aligned into aggregation, while the smaller spherulites exhibit bean-shaped structures (Fig. 2a and c).

The matrix consists predominantly of cryptocrystalline material, and the proportion of the matrix decreases as the proportion of the spherulites increases. The rhyolite strips are mostly concentrated at the edges of the dikes, and their content is less than 5%. The growth of the spherulites is superimposed on the rhyolite strips, indicating that the spherulites formed later than the rhyolite strips (circles in Fig. 2a). Microscopically, the rock contained K-feldspar phenocrysts (1–2%), with a particle size of 0.2–1.2 mm, and some phenocrysts are surrounded by spherulites, forming peritectic structures (Fig. 2c). This indicates that the phenocrysts are the product of early magma crystallization and were formed before the spherulites.

Under the optical microscope, the spherulites were observed to be composed of felsic cryptocrystalline aggregates. According to the order of the formation of the spherulites, they can be divided into three generations, which are described below.

The first-generation of spherulites (S1) was often encapsulated by later generations, forming aggregates with diameters of 1–4 mm. They are composed of several small spherulites with diameters of 0.1–0.5 mm. The small spherulites are irregular polygons, and the contact boundaries between the spherulites are sub-circular suture structures, indicating that the spherulites were in a plastic aggregation state when they aggregated (Fig. 2e, f). The spherulites contain light-colored spots and stripes. The spots account for the majority of the spherulite and are less than 5 μm in diameter, while the stripes account for a minority of the spherulite and are less than 10 μm long (inset in Fig. 2f). The stripes are weakly oriented toward the center of the spherulite aggregates, forming radial fibers (Fig. 2f).

The second-generation of spherulites (S2) grew independently to diameters of 3–5 mm, and they primarily grew as new layers on the S1 (Fig. 2e). The spherulites contain light-colored dotted streaks (constituting a minority, with diameters of less than 5 μm), and the majority of the light-colored linear streaks have lengths of 10–20 μm (inset in Fig. 2i). The streaks form radial fibers (Fig. 2i). The contact boundaries between the S2 spherulites are rounded, and rigid gaps often formed as a result of their agglomeration (Fig. 2d). A quenched darkened edge structure with a width of approximately 20–50 μm and thick radial fibers occur at the edge of the S2 spherulites (Fig. 2h).

The third-generation of spherulites (S3) grew independently to diameters of 8–10 mm, but they usually grew on the previous generation of spherulites (Fig. 2h). The S3 spherulites can be divided into an inner ring (IR) and an outer ring (OR in Fig. 2h). The outer rings exhibit a concentric growth ring pattern, have widths of 0.6–0.8 mm, and consist of 10 alternating light rings (LRs, approximately 60 μm wide) and dark rings (DRs, approximately 15 μm wide) (Fig. 2h, k). The S3 spherulites contain radial fibers. These fibers consist of both light gray (light radial fibers, LRFs) and dark brown (dark radial fibers, DRFs) needle-like structures (Fig. 2k). The fibers are up to 50 μm long (inset in Fig. 2k). The needle-like fibers taper from the OR to the IR, and they transformed and cut the concentric growth rings (Fig. 2h and k), indicating that the concentric growth rings were formed before the radial fibers. The radial fibers in the S3 are thicker than those in the S1 and S2 spherulites.

The matrix has been completely devitrified into perlite (Fig. 2e, h), indicating that the magma ascended to near the surface, was fully vitrified by the last and most violent quenching stage, and consolidated into rock. The glass underwent long-term devitrification to form perlite. The devitrification process ends at the boundaries of the spherulites (Fig. 2e, h), and only a small part of the devitrified silica breaks through the spherulites in the form of veins or points (Fig. 2e). This indicates that the spherulites and the matrix simultaneously began the devitrification process after the diagenetic period, and the former’s ability to resist devitrification was much stronger than that of the latter.

4.2. Spherulite features under TEM

The TEM morphology image of an S1 spherulites shows that S1 spherulites are composed of a large number of fully crystalline minerals at the submicron scale (Fig. 2g). The TEM morphology of the S2 spherulites shows that they are also composed of a large number of fully crystalline minerals at the submicron scale (Fig. 2j). We observed lath-shaped oriented minerals with widths and lengths of 300–500 nm and 600–800nm, respectively, in the OR of an S3 spherulites (lower right corner in Fig. 2l). TEM image also reveal finer, non-directionally arranged minerals and the pronounced development of interstitial cavities between these mineral particles (upper left corner in Fig. 2l). The white dashed line in Fig. 2l separates the lath-shaped and fine minerals and may be the dividing line between the LRFs and DRFs (Fig. 2k). Figure 2m shows a more detailed image of a cavity. As the crystallization progressed, the residual magma and cavities formed via the exudation of volatile gases become more prevalent (Fig. 2m). Before finally solidifying into glass, this viscous residual magma was pulled into a dumbbell shape. As can be seen in the upper left corner of Fig. 2l, the residual magma was prevalent in the spaces between the crystals and around the cavities; that is, after the liquid spherulites were generated, crystallization occurred inside them, and the spherulites were formed from liquid magma and then crystallized into a large number of nanocrystals rather than a single crystal.

4.3. Spherulite EMPA main element compositions and WDS

EMPA was conducted on 88 test points (Fig. S1) along six transects (Fig. 3A–B, C–D, D–E, F–G, H–I, and J–K) in spherulites from the three generations of spherulites. The EMPA results are listed in Table S1, and the line charts of the percentage of major oxides at the test point is shown in Fig. 3. The weighted average chemical formula of the spherulites is K_1.025Na_0.017Fe_0.007Al_1.108Si_2.903O₈, which is very close to the theoretical formula of sanidine (KAlSi₃O₈). It closely resembles the composition of high-temperature sanidine spherulites (Smit & Klaver, 1981) in iridium-rich clay rocks formed by an asteroid impact at the K–T boundary in Caravaca, Spain (Table 2).

Based on structural differences, the spherulites can be divided into two distinct regions. The first region consists of S1, S2, and the IR region of S3, whose structure are uniform (Figs 2e, h and  3a–c); while the second region consists of the concentric growth ring region of the OR of S3 (Figs 2h, k and  3f, g, l, and m). Our analysis revealed that the major element compositions of these two parts exhibit four distinct characteristics.

First, the major element data for the entirety of the S1 and S2 spherulites and the IR region of S3 remain stable (as shown by the S1, S2, and S3/IR data in Table 2 and Table S1). This is represented by the nearly horizontal projection line (Table S1 and Fig. 3 data points 1–13 in h, data points 1–10 in i, data points 28–31 in k, data points 19–27 in n, and data points 1–5 in o) and the uniform color on the WDS spectra (Fig. 3p–r). The consistency of the major element compositions in this part shows that S1, S2, and the IR of S3 grew in a relatively stable environment. The magma in the magma chamber remained stagnant for a sufficient duration, enabling differentiation of the spherulitic phase from the matrix phase. The spherulitic phase then migrated and adhered to the surface of the spherulites, promoting growth of the spherulites.

Second, there are no significant changes in the major element composition in the light ring area in the OR of S3 (Table S1). Each light ring component plotted along a horizontal line on the projection line (indicated by the upward arrows in Fig. 3j, k, n, and o). The WDS results also show that the color of the light rings was uniform (Fig. 3s–u). The chemical formulas of the light rings in each ring only vary slightly (Table 2), which further indicates that the light rings grew in a stable environment during the intermittent magma period.

Third, there are distinct differences in the major element composition in the DR area of the outer ring of S3 (Table S1). The Al₂O₃ content gradually increases from the inside to the edge of the spherulite, while the SiO₂ and K₂O contents decrease (indicated by the downward arrows in Fig. 3k, n, and o). However, the opposite trend occurs along transect D–E (indicated by the downward arrows in Fig. 3j).

Fourth, the dark rings exhibit higher Al₂O₃ and ∑Fe₂O₃ contents than those of the light rings, while their SiO₂ and K₂O contents are lower than those of the light rings (Table 2 and Table S1). The difference in the compositions of the light and dark rings exhibits as a zigzag pattern on the projection lines (Fig. 3; data points 20–26 in j, data points 35–38 in k, data points 29–37 in n, and data points 6–17 in o) and as alternating bands on the WDS component (Fig. 3s–u). These results suggest that the dark rings have a more mafic composition, and the light rings have a more felsic composition. These findings are consistent with the finding that the spherulite composition is more felsic and the matrix composition is more mafic along the main magma immiscibility-driven evolution path (Zhu et al., 2022).

4.4. X-ray powder diffraction analysis

The X-ray powder diffraction analysis results revealed that the mineral assemblage in sample WX-1 consists of sanidine, quartz, and calcite (Fig. 4), which is in agreement with the results of the EMPA for the sanidine components (Table 2 and Table S1). The crystal cell parameters (Table 3) are highly similar to the data for sanidine standard PDF25-0168 and the composition of high-temperature sanidine (Smit & Klaver, 1981).

5. DISCUSSION

The consolidation of the magma into rock resulted in the final crystallization phenomenon, which almost completely covered all signs of the prior immiscibility-driven evolution process. Fortunately, after the studied spherulites formed, the matrix magma rapidly cooled into glass before it crystallized, freezing the spherulites and retaining the characteristics of and information about their growth process. This provides an opportunity to reconstruct the magma migration trajectory and the magma plumbing network system.

The discussion presented in this section takes spherulite growth as the main line to gradually restore the individual parts of the magma migration path.

5.1. Evidence of the immiscible origin of spherulites

The spherulites coexisted with the liquid matrix. The matrix was devitrified into perlite, indicating that the magma was in a liquid state before the matrix solidified. It is well known that if an object goes through a liquid phase in the initial stage of its formation, it will assume a spherical shape to minimize the surface energy (the most stable state) while in the liquid state. This implies that the spherically shaped material remained in a liquid state within another phase medium before solidification. Therefore, if the spherulites formed during a liquid phase, they would maintain a spherical shape while residing within the magma. All of the studied spherulites remained spherical from the time of their formation and were preserved in the matrix (Fig. 2a, e, and h). The phenomenon of immiscibility is common in silicate melts, such as glass and ceramic glaze (Chen et al., 1980; Li & Peng, 1997), and even multi-liquid phase separation has even been identified in iron red glaze (Chen et al., 1980).

We investigated the residual magma after crystallization of the spherulites (Fig. 2l and m), and obtained solid evidence that the liquid spherulites were formed prior to the crystallization of the magma. Similarly, residual magma in spherulites in obsidian has also been discovered via TEM observation (Castro et al., 2008). The existence of hundreds of nanocrystals in the spherulites also indicates that nanocrystals crystallized within the liquid spherulites. The crystallization of anhydrous minerals inside the spherulites expelled the water in the original spherulites (Castro et al., 2008), which also indicates that the liquid spherulites formed prior to crystallization of the magma.

The whole-rock composition analysis result for the spherulites and matrix suggest that as the amount of the spherulite phase increased, SiO₂, K₂O, and Na₂O in the magma tended to migrate and be enriched toward the spherulites, making the spherulite phase more felsic. In addition, the Al₂O₃, TiO₂, MgO, and ∑F₂O₃ become concentrated and enriched in the residual matrix phase, causing this phase to evolve toward a more mafic composition (Zhu et al., 2022). This is consistent with the results of laboratory and natural rock magma immiscible differentiation into two coexisting silicon-rich and iron-rich liquid states (Dixon & Rutherford, 1979; Philpotts, 1982; Jakobsen et al., 2005; Charlier & Grove, 2012).

In addition, the compositional evolution of the light and dark growth rings in the OR of S3 also suggests an immiscible origin for the spherulites. According to our EMPA results, the composition of the light rings is stable (silicon-rich phase) and more felsic, while the composition of the dark rings (iron-rich phase) toward the edges of the spherulites is more mafic. This also indicates that the trends of the immiscible magma differentiation and evolution was developed toward a more mafic composition, which is consistent with the results of the whole-rock analysis.

5.2. Estimates of temperature changes in the magma plumbing system

The temperature changes in a magma plumbing system can be estimated according to the spherulite growth temperature. The crystallization temperature of rhyolite phenocrysts from the adjacent region was 900–973 °C (Fujian Geology Survey Institute, 2016), which is considered to be roughly the initial temperature of the magma immiscibility and spherulite formation. The spherulites grew under a medium supercooling temperature ΔT of approximately 75–145 °C (Lofgren, 1971a, b; Swanson, 1977) and their growth stopped at a glass transition temperature Tg of 690 ± 20 °C (Castro et al., 2008). It is well known that sanidine is a marker of high-temperature magma crystallization, with a formation temperature of 801–880 °C (Chen et al., 2012). The studied spherulites are aggregates of sanidine microcrystalline minerals, which formed at a temperature lower than the K-feldspar phenocryst crystallization temperature and higher than the magma vitrification temperature. Therefore, we estimate that the growth temperature of the spherulites was 900–700 °C. In summary, the temperature of the deep magma reservoir was 900–1000 °C, and the temperature varied from 900 to 700 °C from the bottom to the top of the magma plumbing system. Additionally, the final residual magma was in a sharp cooling environment located approximately 3 km below the ancient surface. The magma then vitrified and solidified into rock.

5.3. Calculation of the magma chamber residence time based on the spherulite growth time

Over the past 40 years, many researchers have measured the growth rates of spherulites (Swanson, 1977; Stolper, 1982; Newman et al., 1986; Zhang et al., 1997; Castro et al., 2008). Castro et al. (2008) estimated the growth rate of a spherulite to be 10⁻⁸–10⁻⁷ cm/s, which is very close to the experimentally determined value for a model of orthoclase-quartz eutectic melt (~10⁻⁸–10⁻⁷ cm/s; Baker & Freda, 2001).

The temperature range of this growth rate was 850–700 °C, and the spherulites were found to grow up to several millimeters in diameter (Castro et al., 2008). This growth rate has been widely applied (Clay et al., 2012; Surour et al., 2015).

The petrographic and EPMA results presented in this paper suggest that the IR areas of the S1, S2, and S3 spherulites grew in the MC1, MC2, and MC3 magma chambers, respectively (Fig. 5). During these intermittent stages, the environment in the magma chambers was relatively stable, resulting in the formation of spherulites with a uniform structure (Figs 2 and 3) and stable chemical compositions (Fig. 3). This indicates that the spherulites and the matrix had sufficient time to reach chemical equilibrium. Therefore, we used the growth rates of spherulites of similar sizes at temperatures of 700, 800, and 850 °C (see Table 3 in Castro et al., 2008) to estimate the growth time of the IRs of the S1, S2, and S3 spherulites. The diameters of the S1 and S2 spherulites are 0.5 and 5.0 mm, respectively. The diameter of the IR of the S3 spherulite is 8 mm, and the width of each light ring in the OR of the S3 spherulite is 60 μm.

According to our calculations, the S1 spherulite grew for 1.21 days (Table 4), indicating that the magma stayed in the first magma chamber (MC1) for a short period of time (Fig. 5a3 and c3). The growth time of spherulite S2 is 14.54 days (Table 4), representing the intermittent stopping period of the magma in the second magma chamber (MC2) (Fig. 5a4 and c4). It took 74.42 days for the IR in spherulite to develop (Table 4). This was the period when the magma resided in the third magma chamber (MC3) (Fig. 5b6 and c6). The growth time of each light ring in the OR of the S3 spherulite was 1.1 days. This indicates that the magma, located in MC3, finally entered a high-frequency pulsed ascent phase. After each pulse, the magma ascent paused for 1.1 days, during which a light ring was formed. Then, another magma pulse occurred, producing a dark ring. The alternating growth cycle of the light and dark rings occurred 10 times, representing 10 intermittent pulses of the magma in the MC3 (Fig. 5b7 and b8) over a period of 11 days, similar to the eruption process of a geyser.

The growth times of the three generations of spherulites represent the duration of the magma pulses and the residence time in the magma chambers (the total growth time of S1 + S2 + S3 IR was 90 days). This indicates that during the transport in the magma plumbing system, the transport of the magma repeatedly paused and reactivated, i.e. resuming its ascent. Furthermore, the farther the magma traveled from the deep magma reservoir source, the longer it paused along the transport path episode. It is generally believed that the growth of mid-to-upper crustal magma reservoirs occurs through multiple pulses, meaning that is magma plumbing systems are repeatedly reactivated or reopened (Cruden & Weinberg, 2018).

5.4. Reconstructing the migration trajectories of multiple batches of magma based on the growth records of multiple generations of spherulites

Regarding the formation of the phenocrysts and rhyolitic bands, initially, the magma separated from the melting zone to form a magma reservoir at depth (Fig. 5a1 and c1). At this time, there was less nucleation and the crystallization rate was high. A small amount of potassium feldspar phenocrysts crystallized, indicating that low supercooling occurred in the magma chamber environment (ΔT < 40 °C) (Lofgren, 1971a; Swanson, 1977). Subsequently, the magma ascended, the temperature decreased, the nucleation rate increased, and the crystallization rate decreased, forming rhythmic stripes composed of microcrystals (Figs 2a and  5a1).

Regarding the formation of the S1 spherulites, as additional fractions of the magma left the melting zone, the pressure in the magma reservoir increased sharply, causing the first batch of magma to ascend and open up a fracture channel. The supercooling degree (△T1) of the ascending magma entering the channel increased, leading to magma immiscibility and the formation of silicon-rich microparticle droplets (Fig. 5a2 and c2). The silicon-rich microparticle droplets collided and aggregated under the influence of the magma flow and agglomerated in the first magma chamber (MC1) to form the plastic S1 spherulites (Figs 2e and  5a3, c3).

Regarding the formation of the S2 spherulites, the amount of magma that left the melting zone continued to increase, and the ascending magma was continuously supplied by the magma reservoir. This magma left MC1 and intruded upward again as the second batch of intrusions. Along the way, the temperature dropped and the supercooling degree changed even more (ΔT2 = ~75–145 °C) (Lofgren, 1971a; Swanson, 1977). The S1 spherulites were quenched, leading to the generation of weak radial fibers, and weak crystallization occurred. When the magma migrated into the second-level magma chamber (MC2) and remained there intermittently for about 14.54 days (Table 4), the magma immiscibility continued to evolve, and the differentiated materials attached to the S1 spherulites, which then transformed into S2 spherulites (Figs 2e and5a4, c4). The development of thicker radial fibers, more obvious quenched edges (Fig. 2h and i), and rigid gaps (Fig. 2d) in the S2 spherulites compared to the S1 spherulites indicates that the temperature in MC2 was lower. This also suggests that the magma was located farther away from the magma source and had migrated toward the surface, and the geothermal gradient decreased along this path (Fig. 5a4 and c4).

The formation of the S3 spherulites (Fig. 5a5, b6–9, and c6–9) can be subdivided into five steps. First, the magma ascended in a pulsed pattern. The temperature decreased further, and the supercooling (△T3) degree changed even more (ΔT1 < ΔT2 < ΔT3). Quenched edges and radial fibers formed in the S2 spherulites (Fig. 2h and i), and weak crystallization occurred. Second, the magma intermittently remained in (MC3) for about 74.42 days (Table 4), and the matrix-differentiated materials attached to (or grew independently) the spherulites generated in the early stage to grow the IR of S3 spherulites (Fig. 5b6 and c6). Third, MC3, located close to the ancient surface, may have been subjected to the addition and vaporization of deep groundwater. This may have resulted in a high pressure in the upper part of the magma chamber, causing explosions and magma eruptions (Decker & Christiansen, 1984; Xu et al., 2006). The magma pulses resumed but with a higher frequency (small scale, multiple batches, similar to the motion of a geyser pulses). Two to three small pulses preceded the rapid pulse mode, producing two to three weak concentric growth rings (Figs 2h, k and  5b7). Fourth, strong upward pulse intrusion occurred. The temperature of the magma continuously decreased, which increased the viscosity of the magma. When the magma pulse caused the spherulites to roll (similar to snowballing), the molten slurry adhered to the surface of the spherulites, resulting in the growth of narrow dark concentric ring. In contrast, a wider, lighter concentric growth ring developed during the interval (1.1 days) between the magma high-frequency magma pulses. We observed up to 10 alternating concentric light and dark growth rings (like the growth rings of trees) in the OR of the S3 spherulite, which record the frequent pulses of ascent and intermittent magma transport (Figs 2h, k and  5b8). The dark rings were formed when the undifferentiated magma directly adhered to the spherules. Toward the OR, the components of the dark rings became more mafic, which indicates that the matrix magma evolved toward a more mafic composition. Fifth, the last magma pulse reached a depth close to the paleosurface, and the low temperature at this depth was conducted from the outer surface to the interior of the spherulites. The radial fibers formed after quenching cut and modified the earlier-formed concentric growth rings (Figs 2h, k and  5b9). The final quenching and cooling stage fully solidified the matrix into glass, and it also enhanced the crystallization in the previously formed S1, S2, and S3 spherulites.

Notably, the contact boundaries between the three generations of spherulites, as well as between the spherulites and the matrix, are always clear and sharply defined circles (Fig. 2). There are no indications of the spherulite edges being altered by the matrix (magma), such as dissolved holes or harbor-shaped dissolved edges. Based on the X-ray analysis results, apart from secondary calcite and quartz developed along secondary fractures (Fig. 3 and Fig. S1), the mineral assemblage did not include thermal alteration minerals (Fig. 4). Therefore, we believe that significant mixing did not occur between the new magma batches that ascended and the older magma in the magma chamber. This could be due to the following possibilities: first, there was little to no compositional difference between the new and old magma, and no or very little mixing occurred. Second, due to the high viscosity of the felsic magma, the magma ascended through the magma plumbing system in a manner similar to squeezing toothpaste and pushed forward in segments without mixing of the new and old magma. Thus, we conclude that the concentric growth rings in the OR of the S3 spherulite were not formed by the high-frequency ascent of mafic magma pulses into MC3, i.e. near the surface, and mixing with the felsic magma.

5.5. Mechanism of the ascent and emplacement of multiple batches of magma

The partial melting process of rock is a process of volume expansion, which enables the magma to gain upward momentum (Takada, 1989; Petford et al., 1993; Dehls et al., 1998; Xu et al., 2006). The magma continues to accumulate and expand in the reservoir, and when enough energy is accumulated, it rapidly ascends and opens fracture channels (Cruden & Weinberg, 2018). As the magma in the channel cools and the supercooling decreases, the magma undergoes immiscible evolution and differentiates into two phases of immiscible liquid, namely, spherulites and matrix. The growth of the spherulites and the concentration of the matrix result in the evolution of the residual magma toward a more mafic composition (Zhu et al., 2022). In this study, the reconstruction of this process is also supported by the EMPA results for the dark growth lines in the OR of the S3 spherulite. The three generations of spherulite growth record three episodes of magma pulse intrusion, which remained in three secondary magma chambers, MC1, MC2, and MC3, until recharge from the magma source reservoir provided additional energy, allowing the magma to continue to ascend. Multiple batches of continuously molten magma reached their ascent destinations through repetition of the injection-rupture-re-injection-re-rupture expansion-intrusion process. Three kilometers below the ancient surface, the 10 concentric growth lines in the OR of the S3 spherulite formed. They record 10 high-frequency magma pulse intrusion events. The magma in MC3 ascended in high-frequency pulses and did not erupt at the surface. However, if the magma had erupted at the surface, this would result in a volcanic eruption. When the magma erupted at the surface, it produced extrusive volcanic rocks such as the spherulite rhyolite porphyry. For the intrusive rock and extrusive rocks in this example (if the magma erupted at the surface), the eruption and intrusion would have created an upper and lower relationship in the longitudinal section. In terms of the origin, both types of rocks were formed via consolidation of the same magma.

Cao et al. (2016) simulated that magma can ascend over considerable distances and intrude into the rigid upper crust, but this requires the formation of dikes and the injection of multiple magma pulses. These multiple pulses prevent the system from freezing too quickly and trigger dike intrusion ahead of the main magma body. This process thermally softens the overlying rock, facilitating the ascent of new magma pulses (Cruden & Weinberg, 2018). In the hot crust, magma migration is freed from the constraints of cooling and is generally widespread at the mesoscale. The associated thermal advection may shift the geothermal gradient to shallower depths (Weinberg, 1999). Our study area, which is situated in a thermally active region formed by large-scale volcanic eruptions, benefits from the thermal softening of the surrounding rocks, which has facilitated the ascent of magma and the formation of dike swarms (Fig. 1c).

5.6. Method for measuring the time scale of magma retention in the magma chamber

Quantitative analysis of felsic magma migration pathways in the upper crust and its storage state within the magma chamber is a frontier subject in petrology, volcanology and ore deposit research. The migration process of magma from the source reservoir to its final emplacement location is generally considered to include multiple stages of migration, that is, it pauses within multiple magma chambers before continuing its ascent. Scholars have studied the time scale of the magma chambers residence time for a long time and have concluded that magma can reside in a magma chamber for a long period of time (Halliday et al., 1989).

The results of isotope geochronology have revealed that the time scales of magma residence in magma chambers are different. For example, the ⁴⁰Ar/³⁹Ar method of dating quartz porphyry revealed that the felsic magma in the Grand Canyon in the United States resided in the magma chamber for 1.1 Ma (Vanden & Schirnick, 1995). Zircon U–Th determination results indicate that the residence time of Sakura Island volcanic magma in the relevant magma chamber in Japan was 0.3 to 15.7 Ka (Black et al., 2001). However, the accuracy of isotope dating technology is restricted by the contents of radioactive parent isotopes and their daughter products, the uncertainty of the decay constant, the closure of the isotope system, and the properties of the occurrence minerals, all of which can lead to large errors.

The mineral diffusion geochronology method is based on the diffusion model of an element or isotope in a mineral from a high concentration zone to a low concentration zone under certain temperature conditions. Using this model, the time scale of the diffusion of the element or isotope under the given thermal history and other conditions can be calculated (Turner & Costa, 2007). However, due to the uncertainty of the initial temperature estimation and the mineral diffusion coefficient (measured in the laboratory), as well as the error of mineral the profile analysis, the relative error of in diffusion geochronology estimates can exceed100% (Li et al., 2024). Taking Ti in quartz as an example, the difference in the diffusivity obtained in different experiments is more than three orders of magnitude, and the duration for which granitic magma remains on the solid phase line calculated based on this varies from several decades to millions of years, which significantly affects the understanding of the storage state of magma (Li et al., 2024).

The new geochronological method we have established, called ‘spherulite dating’, is based on the petrological characteristics of spherulites and records their growth process during their residence in the magma chamber. This involves estimating the temperature of the growth of different generations of spherulites, measuring the diameter of the spherulites and estimating the durations of the growth of the different generations of spherulites using the measured natural spherulite growth rate (Castro et al., 2008). This time is the time scale for the residence of the magma in the magma chamber. This method can not only track how many times the magma resides in the magma chamber during the migration process but can also estimate the time scale of the residence of the magma in the magma chamber. Our method is simple and clear, and it avoids the interference of geologically long processes, complex theoretical derivations, and the many errors caused by human analysis techniques. The 90-day time scale of the residence of the magma in the magma chamber estimated in our case study is consistent with the theoretical model of magma migration over short periods of time (a dynamic model with time scales of several months to centuries; Petford et al., 2000) and is also consistent with the assumption that the magma only resides in the magma chamber for tens of days to a few months (Druitt et al., 2012; Wotzlaw et al., 2014). Currently, due to differences in test materials, various dating methods cannot be directly compared quantitatively. However, we believe that compared to traditional isotopic and mineral diffusion methods, spherulitic dating offers significant advantages and higher temporal precision in quantifying magmatic processes. The accuracy and precision of different dating methods will be further evaluated in future studies.

However, the ‘spherulite dating method’ has certain limitations and constraints. This method is most suitable for specific volcanic eruption geological settings, particularly where spherulite-bearing rocks are produced during the effusive volcanic phases of a volcanic eruption center. In contrast, optimal results may not be achieved in other environments. The method requires careful petrological observations to confirm the origin of magmatic immiscibility in spherulites and to differentiate the growth generations of spherulites. Based on these growth generation divisions, the correlation between the magmatic evolution trajectory and spherulite growth generations is tracked through geochemical element analysis to enhance the reliability of the method. The selection of parameters for the timing model in the spherulite dating method relies on the quantitative measurement of natural spherulite growth rates (Castro et al., 2008). As a first-order approximation, the choice of growth rate parameters for the timing model must be approached with caution. It is essential to demonstrate the temperature conditions of spherulite growth and verify the applicability of the selected parameters.

5.7. Thermal field of magmatic plumbing and mineralization

Dikes are the most common channels for the ascent and intrusion of magma (Petford et al., 2000). The thermal fields in such magmatic plumbing systems are the best pathway for the migration of metallogenic fluids and also constitute a magma plumbing mineralization system (Su et al., 2014). Zhang et al. (2013) studied the magmatic thermal field and found that metallogenic fluids can only rise smoothly to the surface along the magmatic thermal field channel. Otherwise, they will consolidate and be located deep under the surface after encountering the cold country rocks (Fig. 5c). By studying the magmatic thermal field of a 7-km-wide stock, (Tang et al., 2013) found that after magma emplacement, the thermal field maintained a temperature of 700–800 °C for 0.2 Ma and 600–700 °C for 0.5 Ma, and the temperature remained above 300 °C for 1.0 Ma, providing enough time for a cataclysmic metallogenic event to occur (Zhang et al., 2013).

In the ore magmatic crypto-explosive breccia-type gold deposit in the study area (Fig. 1c), the ore-forming fluid rapidly ascended along the thermal channels left by the magmatic plumbing system opened by the quartz porphyry magma. It is estimated that the ore magma migrated through the magmatic plumbing system at a rate of 0.825 cm/s, and it ascended from a depth of 30 km to the shallow metallogenic location in 50 days (Fig. 5c and d) (Zhu et al., 2021). Therefore, according to our results, the mineralization in the thermal field of the magma plumbing system occurred as rapid, catastrophic event rather than a gradual process over extended periods.

5.8. Concentric growth rings of spherulites help to reveal the activity of ancient volcanoes

Modern volcanic eruption records indicate they exhibit both periodic and intermittent behaviors (Peltier et al., 2009). A periodic eruption occurs after a considerable period of dormancy, spanning months, years, decades, or even longer. Conversely, an intermittent eruption is characterized by multiple, frequent eruptions over a short period of time, such as several minutes, hours, or days. Although the ultra-shallow volcanic magma analyzed in this study did not reach the surface, the concentric growth pattern of the S3 spherulite recorded the occurrence of high-frequency pulses form the uppermost magma chamber and revealed that they were similar to the frequent intermittent eruptions of volcanoes. The trajectory of these intermittent magma pulses exhibited strong temporal characteristics, making the spherulite growth rings useful for studying the pattern of ancient volcanic eruptions and providing a reference for modern volcanic activity prediction models.

6. CONCLUSIONS

In summary, based on our case study, we established a spherulite recorder model, which can reconstruct the magma transfer path in the plumbing system in detail and can calculate the time scale of the pauses during the ascent process. We summarize several key steps in this model below.

1. The spherulites must originate form an immiscible magma. Spherulites with this origin can leave traces of the magmatic compositional evolution during the growth of the spherulites. For example, in our case study, the felsic magma immiscibility evolution not only left a record in the major element composition of the matrix but also imprinted the same record in the major element composition of the dark concentric growth lines in the OR of the S3 spherulites.

2. Records of multiple generations of spherulite growth can be identified based on the petrographic features and geochemical data. The magma stops in magma chamber along the ascent pathway, and these stops are the spherulite growth periods. The growth time of each generation of spherulites corresponds to the duration for which each batch of magma resides in the magma chamber, so the duration of the spherulite growth stage can be used to determine how long the magma resides in each magma chamber.

3. In our case study, we identified three generations of spherulite growth, which recorded the ascent and stagnation of three batches of magma through three magma chambers. The 10 concentric growth lines developed in the OR of the S3 spherulite record the transition of the magma from pulsed intrusion to high-frequency pulsed eruption, which is similar to the process of volcanic activity.

4. Natural spherulites are typically found in rhyolite and obsidian (Keith & Padden, 1963; Lofgren, 1971b) and are distributed in the volcanic zone around the west coast of the Pacific Ocean (class I; Fig. 1a), usually around volcanic institutions. Spherulitic rhyolite porphyry and rhyolite are typical rocks formed during effusive volcanic phases and often occur on a large scale around volcanic structures (Fujian Geology Survey Institute, 2016). Studying the spherulite growth process can help to determine the network structure of the magma plumbing system, the magma transportation path, the compositional evolution of the magma, and the magma chamber residence time in the crust, providing a magmatic emplacement model for the prediction of magmatic deposits, contributing to the study of ancient volcanic eruption patterns, and providing a reference for the establishment of modern volcanic eruption geological disaster warning models.

5. It should be noted that the mainstream view of spherulite formation is that magma crystallizes rapidly under high supercooling conditions after the emplacement of volcanic rocks at shallow depths in the crust (Lofgren, 1971b; Swanson, 1977; Baker & Freda, 2001; Smith et al., 2001; Monecke et al., 2004). The petrological significance of spherulites has not been explored deeply based on this hypothesis. We found that the origin of the magmatic immiscibility of spherulites can provide valuable information about magma ascent processes, and spherulites have profound petrological significance regarding research on magma ascent processes.

At this stage, we could only estimate the time scale of the magma chamber residence time during its ascent. We have not yet provided a method to estimate the entire time scale of magma ascent from a deep reservoir to a shallow depth. The participation of more researchers is needed to fill this research gap in the future.

ACKNOWLEDGEMENTS

We acknowledge Drs Xiaocheng Zhu and Dr. Bernie Dominiak (NSW Department of Primary Industries) for providing critical review of this manuscript. We are grateful to the anonymous reviewer for their thorough, constructive comments and suggestions, which greatly improved this paper. Zijin School of Geology and Mining Fuzhou University for providing electron probe laboratory and XRD laboratory equipment.

SUPPLEMENTARY DATA

Supplementary data are available at Journal of Petrology online.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest regarding this manuscript.

FUNDING

This work was funded by the first author.

DATA AVAILABILITY

The data presented in this manuscript is available within the manuscript itself. Specimens might be provided upon request.

Handling Editor: Prof. Takashi Mikouchi

REFERENCES