Crystal Fragmentation Inducing Euhedral Crystal Habits in Volcanic Rocks: Fracture Histories of Crystals from Various Tectonomagmatic Settings and Implications for Plumbing System Processes Open Access

Obtaining accurate constraints on the chronologies of pre-eruptive volcanic processes, such as magma recharge and mixing, is one of the key strategies in assessing future volcanic hazards (Martin et al. 2008). Studies of crystal size distributions (e.g., Cashman & Marsh 1988, Marsh 1988, Higgins 1996, Marsh 1998, Higgins 2000, Bindeman 2003, Turner et al. 2003, Morgan et al. 2007, Jerram & Martin 2008, Lormand et al. 2020) and of chemical diffusion across crystal growth zones (diffusion chronometry) (e.g., Gerlach & Grove 1982, Kohn et al. 1989, Nakamura 1995, Zellmer et al. 1999, Bindeman 2003, Costa et al. 2003, Zellmer et al. 2003, Morgan et al. 2004, Costa & Dungan 2005, Morgan & Blake 2006, Martin et al. 2008, Coulthard Jr. et al. 2024) are two approaches to obtain quantitative estimates of the timing of subvolcanic processes. Detailed reviews of these methods and their key considerations and challenges have been provided recently (Cashman 2020, Costa 2021), and reiterating those are beyond the scope of the present study. However, we would like to stress here that the quality of chronological estimates yielded by crystal size distribution and diffusion chronometric studies is principally limited by the level of detail with which a crystal’s growth record can be resolved. Using elemental mapping by electron probe micro-analysis (EPMA) and/or stacked CMOS-type active pixel sensor (SCAPS) imaging by ion microprobe allows minute chemical variations within crystals to be decoded, providing access to details within volcanic crystals that are cryptic to more commonly used methods such as backscattered electron (BSE) imaging.

The topic of the present contribution is crystal fracturing (or fragmentation, these terms are used synonymously here), which is now a commonly recognized process to operate in volcanic systems (Taddeucci et al. 2021). Crystals may fracture during syneruptive shock wave propagation in the conduit or due to melt inclusion decrepitation by overheating or decompression (Bindeman 2005). Fragmentation affects crystal size, and thus crystal size distributions (Bindeman 2005). Identification of crystal fragmentation based on the discontinuity in chemical zonation has already been demonstrated (Higgins et al. 2021). These previous studies have focused on syneruptive fracturing of the crystal cargo as a result of strain imposed on crystals carried by the magma during the eruptive process.

Crystal fragmentation has, however, not been studied in much detail, although it is now becoming clear that it occurs in volcanic materials ranging from lava flows to ignimbrites, with the latter generally carrying more fragments than unbroken crystals (Allen & McPhie 2003). Crystal fragment morphologies previously described in igneous systems include (i) irregularly shaped fragments with cuspate, embayed outlines, ascribed to melt inclusion decrepitation due to overheating and decompression (Best & Christiansen 1997); (ii) angular grains with subhedral crystal faces, ascribed to disaggregation of polygranular crystal aggregates (Best & Christiansen 1997); and (iii) regularly shaped, almost parallelepipedic, crystal fragments that may be ascribed to shear flow within crystal mushes (Smith 2002, Arbaret et al. 2007) or syneruptive processes such as vertical decompression or conduit implosion (Kennedy et al. 2005).

Here, we will show that fracturing of crystals may also be a common pre-eruptive process that can be recorded by volcanic crystals through distinct features in their internal zonation pattern. Repeated episodes of crystal fracturing, annealing and resorption, which are cryptic to commonly used imaging techniques, are recorded by the crystals. Fracturing along cleavage planes is frequently observed. Our findings provide insights into a process leading to euhedrality of crystals not typically recognized, and suggests that crystal fracturing is a key process in the formation of the crystal cargo in volcanic rocks.

We have studied crystals from samples that have been previously investigated, including from hot-spot settings (Cran Canaria, Troll & Schmincke 2002), volcanic arcs (southern Taupo Volcanic Zone, Lormand et al. 2021), back-arcs (D’Mello et al. 2024), and mid-ocean ridge settings (Zellmer et al. 2012). Here we briefly reiterate the analytical techniques employed.

Rock samples were mounted in resin, polished, and carbon coated for electron microprobe analysis at Institute of Earth Sciences, Academia Sinica, in Taipei using a field emission and tungsten electron probe microanalyzer (JEOL JXA-8500F, JXA-8900R) each equipped with five wavelength-dispersive spectrometers. Chemical distribution (mapping) analysis was performed under conditions of 15 kV and 50–80 nA for the acceleration voltage and beam current, respectively, with focused beam. X-ray intensities of SiKα, KKα, CaKα, NaKα, and FeKα were counted for 0.05 s at an interval of 1–5 µm with X-Y stage driving.

Semi-quantitative concentration maps for plagioclase crystals were obtained at the Isotope Imaging Laboratory at Hokkaido University using a Cameca IMS-1270 SIMS instrument equipped with a stacked CMOS-type active pixel sensor (SCAPS) to visualize the elemental distribution on the sample surface at high magnification (Yurimoto et al. 2003) for Al, Ca, Fe, Mg, Na, K, and Si. An O⁻ primary beam of 23 keV was irradiated on the sample surface of 80–100 µm in diameter using a 7 nA beam current. The sample surface was sputtered with a 20 nA beam current for ten minutes before the analysis. The positive secondary ion images of ²³Na, ²⁴Mg, ²⁷Al, ²⁸Si, ³⁹K, ⁴⁰Ca, and ⁵⁶Fe on the sample surface were collected by the SCAPS detector, with exposure times of 50, 500, 50, 25, 50, and 250 s, respectively. A submicron spatial resolution was achieved on individual images. Concentration ratio images normalized to Si were used to discriminate elemental zoning and textures of plagioclase crystals.

Figure 1a shows a typical anorthoclase crystal with distinct orthogonal cleavage planes, for which previous work has postulated a resorbed core, an overgrowth zone, and a semi-euhedral rim (Troll & Schmincke 2002). However, secondary electron imaging (Fig. 1b) discloses internal fractures, partly curved, partly following cleavage plains, which seem inconsistent with the interpretation of an overgrowth zone, but do not allow conclusive interpretations with regard to the crystal’s growth history. The backscattered image (Fig. 1c) does not provide significant additional insights. However, elemental mapping by FE-EPMA (Fig. 1d) finally reveals cryptic crystal zonation with internal zoning in Fe (and other elements, not shown) entirely unrelated to the previously inferred crystal morphology. Fracturing (labeled ‘recent edge fractures’ in Fig. 1d) is unambiguous along at least three sides of the crystal, where internal and rim growth zones are truncated. Note the variable distance of an Fe-poor growth zone (dark blue, delineated by thin dashed lines) from the crystal rim and the disappearance of this zone on the left and right sides of the crystal. Note also the slight variation in thickness of the outermost growth zone above this Fe-poor growth zone, which may normally be regarded as variable dissolution. However, the straightness of the upper side of the crystal argues that this may instead also present a recent edge fracture. Earlier, now annealed fractures cut an Fe-rich growth zone (light blue in Fig. 1d). Cryptic zoning, internal and edge fracturing, and fracture annealing may be very common phenomena in crystals from the studied ignimbrite, but they are not revealed by conventional imaging techniques.

Fracturing of microantecrysts (Zellmer 2021) has recently been reported from tephras erupted from the Tongariro Volcanic Centre in the southern Taupo Volcanic Zone (Lormand et al. 2021) inboard of the Hikurangi margin, New Zealand. The authors revealed cryptic zoning in Ca within plagioclase crystals by SCAPS-SIMS. We reproduce three of the obtained images in Figure 2 and indicate straight internal truncation of growth zones, which we consider as historic fractures, as well as dominantly straight edges that either cut zoning or are subparallel to outer growth zones of variable width, which we consider recent edge fractures.

D’Mello et al. (2024) studied the petrology of recent lava flows from Taranaki volcano in the Hikurangi back-arc, New Zealand. We here report clear recent edge fracturing parallel to cleavage planes in some plagioclase crystals (Fig. 3), recent edge fractures and historic annealed internal fractures in pyroxenes (Figs. 4, 5), and one evident internal fracture in an amphibole crystal (Fig. 6). These fractures would be cryptic to BSE imaging and but are revealed by EPMA mapping, as conducted here.

In samples from ocean ridges, which are typically considered comparatively simple petrogenetic settings without magma plumbing in thick overriding crustal lids, both internal historic and recent edge fractures can be seen in glomerocrysts carried by axial samples from the East Pacific Rise (Fig. 7) and by off-axis samples from the Juan da Fuca ridge (Fig. 8). The latter is particularly revealing by displaying the differences between crystals that are partially anhedral due to curved dissolution surfaces and partially euhedral due to recent edge fractures.

Our finding of cryptic fracturing, evidently exploiting cleavage planes (cf. Fig. 1), implies that the number of crystal fragments previously identified in volcanic rocks may be significantly underestimated and that crystal fragmentation therefore is a fundamental and, given the evidence for historic internal fractures, recurrent process operating in magmatic systems both prior to and at the onset of volcanic eruptions in a range of tectonomagmatic settings, irrespective of eruption style. Quantification of the percentage of crystals affected by historical and syneruptive fracturing is difficult, because (i) the high-resolution imaging work to reveal this process is time-consuming, therefore limiting the number of crystals typically studied to this level of detail, and (ii) there are presently too few studies that engage in such detailed analytical work. However, when analyzing the chemical stratigraphy of more than 40 crystals from ocean ridge basalts studied in detail by Zellmer et al. (2012), the majority of them display internal and/or edge fracturing. And ocean ridge magmatic plumbing systems are arguably the simplest. Further, we had no difficulty finding the examples of crystal fragmentation provided in the present study from the whole range of tectono-magmatic settings investigated. Therefore, we argue that pre- and syneruptive crystal fragmentation are very common, ubiquitous, and typically overlooked by standard analytical methods.

While historical internal fractures speak to the recurrence of crystal fragmentation processes during crystal growth, recent edge fractures provide key insights into the processes operating in magmatic systems just prior to and at the onset of volcanic eruptions. In the examples we have provided here, there is no evidence for thin euhedral overgrowth rims beyond the fracture surfaces. This implies that the erupted samples did not experience any cooling- or degassing-induced crystal growth on preexisting crystal fragments. The timescales between crystal fragmentation and eruption must therefore be very short (on the order of hours), too short for the kinetic process of crystallization to operate. This is consistent with the recently proposed model of the dominantly subsolidus nature of transcrustal plumbing systems yielding mafic to intermediate magmas in subduction zone settings (Coulthard Jr. et al. 2024, D’Mello et al. 2024, Zellmer et al. 2024). Here, crystal-free melts ascend rapidly through the crust, scavenging their entirely antecrystic (Zellmer 2021) crystal cargo from preexisting subsolidus plutonic rocks, which provide igneous xenoliths that fracture into glomerocrysts and ultimately to individual phenocrysts (sensu Iddings 1892). The present work suggests that similar processes may be operating in other tectonomagmatic settings, including at hot spots and ocean ridges. The disintegration of glomerocrysts into individual crystals is a process that may be considered opposite to synneusis (Vance & Gilreath 1967, Vance 1969, Dowty 1980), where individual crystals ‘swim together’ and merge to form glomerocysts. However, detailed observations of glomerocrysts and their crystals (Zellmer et al. 2016) indicates that disaggregation of glomerocrysts may be much more common.

Cryptic crystal fracture fragments pose considerable challenges in the application of chronological methods. Studies that use one-dimensional core–rim profiles to assess chemical variations within a crystal become inherently unreliable in the light of the present study. Geospeedometric work employing one-dimensional profiles may only be considered to provide accurate constraints if symmetric rim-to-rim profiles are used and the number of crystals studied is large. However, cryptic edge fractures may also affect geospeedometric constraints on pre-eruptive magmatic processes (e.g., Costa et al. 2003, Morgan et al. 2004, Martin et al. 2008): if substantial portions of the rims of studied crystals are removed by crystal fracture, the record of processes occurring just prior to eruption is annihilated, and the timescales derived will only constrain previous magmatic events. Again, studying many crystals will be necessary to allow confident assessments of the magmatic processes that ultimately trigger a volcanic eruption.

Further, it has been shown that crystal fracturing in volcanic rocks may produce curved, concave-down crystal size distributions (Bindeman 2005) that are often attributed to other processes, such as size-dependent growth or annealing of smaller crystals by Ostwald ripening (Higgins 1998, Cabane et al. 2005), or to temperature oscillations (Simakin & Bindeman 2008). While crystal nucleation, growth, and resorption are the primary parameters that control the shape and evolution of crystal size distributions in magmatic systems (Marsh 1988, 1998), the presence of cryptic crystal fracture fragments introduces some degree of uncertainty when interpreting such data. Therefore, an adequate characterization and imaging of crystals and crystal fragments in the light of cryptic fracturing will be paramount in future crystal size distribution studies aimed at deciphering the crystallization processes operating in subvolcanic magma reservoirs.

1. Crystal fragmentation is an important process operating in volcanic systems, irrespective of tectonomagmatic setting or eruption style, both prior to and at the onset of volcanic eruptions.

2. Cryptic crystal fracture fragments may be produced due to fragmentation along cleavage planes. They cannot be identified by conventional imaging techniques, suggesting that previous estimates of the proportion of crystal fragments in volcanic rocks are minima.

3. Historic internal fractures, sometimes annealed, are indicative of the fracturing processes operating throughout the history of crystal growth. They can be distinguished from recent edge fractures. The latter show no overgrowth, suggesting that the time between crystal fragmentation and eruption is too short to allow cooling- or degassing-induced crystallization. This is consistent with recent models of transcrustal plutonic system architectures in subduction zone settings and may indicate that similar plumbing system processes are ubiquitous, irrespective of tectonomagmatic setting or eruption style.

4. The data presented here pose considerable challenges to geospeedometric and crystal size distribution studies. Adequate chemical imaging of many crystals is necessary to preclude misinterpretations of such geochronological data.

We thank Anja Moebis, Nessa D’Mello, Michael Perfit, and Valentin Troll for access to samples for analysis, and Charline Lormand and Valentin Troll for several discussions that provided the impetus to write this paper. The constructive comments of Ilya Bindeman and an anonymous reviewer improved this script, and we thank Gregory Shellnutt for inviting this contribution and for editorial handling. This work was funded in part by the Natural Science Council of Taiwan (97-2628M001027-MY2 to GFZ and 97-2116M001008 to YI).