Dynamics of degassing in evolved alkaline magmas: Petrological, experimental and theoretical insights

Abstract

In the last few decades, advanced monitoring networks have been extended to the main active volcanoes, providing warnings for variations in volcano dynamics. However, one of the main tasks of modern volcanology is the correct interpretation of surface-monitored signals in terms of magma transfer through the Earth's crust. In this frame, it is crucial to investigate decompression-induced magma degassing as it controls magma ascent towards the surface and, in case of eruption, the eruptive style and the atmospheric dispersal of tephra and gases. Understanding the degassing behaviour is particularly intriguing in the case of poorly explored evolved alkaline magmas. In fact, these melts frequently feed hazardous, highly explosive volcanoes (e.g., Campi Flegrei, Somma-Vesuvius, Colli Albani, Tambora, Azores and Canary Islands), despite their low viscosity that usually promotes effusive and/or weakly explosive eruptions. Decompression experiments, together with numerical models, are powerful tools to examine magma degassing behaviour and constrain field observations from natural eruptive products and monitoring signals. These approaches have been recently applied to evolved alkaline melts, yet numerous open questions remain.

To cast new light on the degassing dynamics of evolved alkaline magmas, in this study we present new results from decompression experiments, as well as a critical review of previous experimental works. We achieved a comprehensive dataset of key petrological parameters (i.e., 3D textural data for bubbles and microlites using X-ray computed microtomography, glass volatile contents and nanolite occurrence) from experimental samples obtained through high temperature-high pressure isothermal decompression experiments on trachytic alkaline melts at super-liquidus temperature. We explored systematically a range of final pressures (from 200 to 25 MPa), decompression rates (from 0.01 to 1 MPa s⁻¹), and volatile (H₂O and CO₂) contents. On these grounds, we integrated coherently literature data from decompression experiments on evolved alkaline (trachytic and phonolitic) melts under various conditions, with the aim to fully constrain the degassing mechanisms and timescales in these magmas. Finally, we simulated numerically the experimental conditions to evaluate strengths and weaknesses in decrypting degassing behaviour from field observations.

Our results highlight that bubble formation in evolved alkaline melts is primarily controlled by the initial volatile (H₂O and CO₂) content during magma storage. In these melts, bubble nucleation needs low supersaturation pressures (≤ 50–112 MPa for homogeneous nucleation, ≤ 13–25 MPa for heterogeneous nucleation), resulting in high bubble number density (~ 10¹²–10¹⁶ m⁻³), efficient volatile exsolution and thus in severe rheological changes. Moreover, the bubble number density is amplified in CO₂-rich melts (mole fraction X_CO2 ≥ 0.5), in which continuous bubble nucleation predominates on growth. These conditions typically lead to highly explosive eruptions. However, moving towards slower decompression rates (≤ 10⁻¹ MPa s⁻¹) and H₂O-rich melts, permeable outgassing and inertial fragmentation occur, promoting weakly explosive eruptions. Finally, our findings suggest that the exhaustion of CO₂ at deep levels, and the consequent transition to a H₂O-dominated degassing, can crucially enhance magma vesiculation and ascent. In a hazard perspective, these constraints allow to postulate that time-depth variations of unrest signals could be significantly weaker/shorter (e.g., minor gas emissions and short-term seismicity) during major eruptions than in small-scale events.

Introduction

Predicting the evolution of surface-monitored signals at active volcanoes in terms of magma transfer through the Earth's crust is challenging. Understanding the behaviour of magma degassing is thus fundamental. In fact, decompression during magma ascent leads to the exsolution of volatiles dissolved in the melt (H₂O > CO₂ > Cl, F, S) through the nucleation and growth of bubbles (i.e., the onset of the degassing process), which in turn allow for magma compressibility and buoyancy. Volatile exsolution also affects magma rheology by increasing melt polymerization and favouring microlite/nanolite crystallization, with consequent changes in viscosity and yield strength. Therefore, magma degassing and vesiculation (where vesicles refer to frozen bubbles within tephra; Cashman and Scheu, 2015) strongly control volcanic conduit dynamics and, thus, the eruptive style and the atmospheric dispersal of tephra and gases (e.g., Cashman and Mangan, 1994; Sparks et al., 1994; Gonnermann and Manga, 2012). These processes can be investigated using different approaches, including measurement of textural and chemical parameters in natural volcanic rocks (e.g., Gurioli et al., 2015 and references therein), high temperature-high pressure (HT-HP) decompression experiments (e.g., Shea, 2017 and references therein) and numerical modeling (e.g., Toramaru, 1989, Toramaru, 1995; Huber et al., 2014).

In particular, HT-HP decompression experiments, together with numerical models, allow to simulate magma degassing under controlled conditions and thus to better constrain field observations from natural eruptive products and monitoring signals.

In the last few decades, experimental and numerical approaches were widely applied to rhyolitic melts, leading to a well-recognized paradigm for their degassing behaviour (e.g., Hurwitz and Navon, 1994; Gardner et al., 1999; Mangan et al., 2004; Mourtada-Bonnefoi and Laporte, 2004; Toramaru, 2006; Cashman and Scheu, 2015 and references in these studies). Mainly, these studies suggest that homogeneous H₂O bubble nucleation occurs under very high (> 100 MPa) supersaturation pressures (∆P_HoN, see Table 1 for symbols). Under these conditions, the degassing style typically evolves in disequilibrium, although equilibrium can be promoted by low decompression rates and/or heterogeneous bubble nucleation (which abruptly reduce the supersaturation pressure up to a few MPa), so that the resulting bubble number density (BND) is strongly dependent on the decompression rate (dP/dt). Consequently, faster decompression rates are often associated with highly explosive eruptions, whereby the fragmentation of these highly viscous melts is thought to be driven by a violent vesiculation (due to one or two nucleation events) at shallow levels in the conduit.

Commonly, this paradigm is applied to describe a typical behaviour of magma degassing during explosive eruptions, in spite of the wide variability in composition, and then in physic-chemical properties, of magmas. Therefore, expanding the knowledge on kinetics and dynamics of magma degassing to a wider compositional spectrum would be crucial for volcanic hazard assessment. This is especially true for evolved alkaline melts that, despite their low viscosity (usually promoting effusive or weakly explosive eruptions; Lesher and Spera, 2015), are frequently able to feed highly explosive events at many active volcanoes (e.g., Neapolitan volcanoes: Campi Flegrei and Somma-Vesuvius; Roman volcanoes: Colli Albani Volcanic District; Tambora; Azores and Canary Islands).

Although high temperature-high pressure decompression experiments were recently performed on evolved alkaline (trachytic and phonolitic) magmas, their degassing behaviour is still poorly known. Overall, as shown below in this section, previous studies often lack mutual consistency in experimental settings and typically focused on specific issues, rather than fully constraining the degassing mechanisms and timescales. Previous decompression experiments, performed on H₂O-rich, CO₂-free alkaline melts, can be divided into two main groups, respectively at near-liquidus (Iacono-Marziano et al., 2007; Gardner, 2012; Gardner et al., 2013; Marxer et al., 2015; Preuss et al., 2016; Allabar and Nowak, 2018; Allabar et al., 2020a, Allabar et al., 2020b) and sub-liquidus (Larsen and Gardner, 2004; Mastrolorenzo and Pappalardo, 2006; Larsen, 2008; Mongrain et al., 2008; Shea et al., 2010) temperatures.

• Decompression experiments at near-liquidus temperatures (promoting homogeneous bubble nucleation) can be further distinguished based on the experimental variables, as follows:

  • Experiments on the Vesuvius AD 79 “white pumice” K-phonolite and on the Campi Flegrei Campanian Ignimbrite K-trachyte, at temperature ≥ 1050 °C, initial pressure (P_i) of 200 MPa and dP/dt between 0.0028 and 4.8 MPa s⁻¹. In detail, Iacono-Marziano et al. (2007) explored decompression-driven degassing of H₂O-saturated melts up to low final pressures (P_f ≥ 10 MPa), essentially limited to high dP/dt (1.7 and 4.8 MPa s⁻¹). Marxer et al. (2015) and Preuss et al. (2016) provided useful data for high P_f (≥ 60 MPa) and a wide range of dP/dt (from 0.0028 to 1.7 MPa s⁻¹) for both H₂O-saturated and undersaturated conditions, although their works mainly investigated experimental and analytical procedures (e.g., preparation of starting material, experimental decompression styles, bubble shrinkage correction). Allabar et al. (2020a) discussed in more detail bubble shrinkage in quenched glasses and its implication on glass textural and chemical features. Finally, Allabar and Nowak (2018) and Allabar et al. (2020b) explored the role of variable dP/dt (from 0.024 to 1.7 MPa s⁻¹) and initial H₂O contents (from 3.3 to 6.3 wt%) on bubble nucleation/phase separation. They also extended their analysis to the evolution of vesiculation, albeit poorly constrained at low P_f.

  • Experiments on different H₂O-rich evolved alkaline compositions (i.e., Na-phonolite, K-phonotephrite, Na-trachyte) at temperatures of 875 and 1150 °C, P_i of ~150 MPa and dP/dt of 2–2.5 MPa s⁻¹ (Gardner, 2012; Gardner et al., 2013), aimed at constraining the influence of melt composition on the surface tension and bubble nucleation.

  • In these studies, a wide range of ∆P_HoN (from 50 to 112 MPa) was estimated, as well as contradictory relationships between degassing styles (equilibrium vs. disequilibrium) and BND with dP/dt were found. Moreover, a few data exist at low P_f to carefully constrain the degassing regime (i.e., closed- vs. open-system) and the fragmentation mechanisms during magma ascent.

• Decompression experiments at sub-liquidus temperatures (promoting crystallization and, consequently, heterogeneous bubble nucleation) can be further distinguished in:
  

  • Experiments carried out on H₂O-saturated, microlite-rich, Vesuvius AD 79 “white pumice” K-phonolite at temperatures between 850 and 950 °C and variable dP/dt from 0.01 to 17 MPa s⁻¹ (Larsen, 2008; Mongrain et al., 2008; Shea et al., 2010). In this case, melts were systematically decompressed from a P_i between 200 and 100 MPa up to P_f as low as 10 MPa, allowing to study the evolution of degassing and vesiculation.

  • Pioneering experiments on evolved alkaline melts, which provided further, scattered, information on H₂O-saturated Laacher See Na- and K-phonolitic melts at 825–850 °C and dP/dt between 0.017 and 10 MPa s⁻¹ (Larsen and Gardner, 2004) and on H₂O-saturated Campi Flegrei K-trachytic melts at 900–950 °C and dP/dt ≤ 0.07 MPa s⁻¹ (Mastrolorenzo and Pappalardo, 2006).

  • Findings from these studies consistently indicate that the supersaturation pressure is substantially reduced up to 13–25 MPa in presence of mafic microlites, which act as nucleation sites, thus favouring equilibrium degassing, especially for slow dP/dt. However, these results were mainly used for comparisons with natural products of specific eruptions or eruptive phases, rather than being considered in the general context of the degassing dynamics of evolved alkaline magmas.

Finally, Fanara et al. (2017) and Sottili et al. (2017), through HT-HP decompression experiments on CO₂-saturated, H₂O-free, K-trachyte and trachybasalt, examined the effect of different crystal phases on heterogeneous bubble nucleation, showing that CO₂ bubbles can form indiscriminately on both sialic and mafic crystals, with various shapes.

We remark that a systematic experimental investigation of the influence of CO₂ on the degassing of evolved alkaline melts was never attempted so far, although carbon dioxide can attain high concentrations in alkali-rich liquids (e.g., Fanara et al., 2015; Schanofski et al., 2019). Also, carbon dioxide, the second most abundant volatile component in natural magmas, can strongly affect the physic-chemical properties of silicate melts (e.g., Wallace et al., 2015). Due to its relatively low solubility, CO₂ may control the (deep) early stages of magma degassing at high pressures and have a strong impact on volcanic explosivity, as a consequence of magma-crustal carbonate interaction (e.g., Freda et al., 2011; Pappalardo et al., 2018; Buono et al., 2020a, Buono et al., 2020b). Nevertheless, only a few studies focused on the CO₂ influence on magma vesiculation, such as the HT-HP decompression experiments on rhyolitic (Gardner and Webster, 2016 and references therein) and basaltic (Le Gall and Pichavant, 2016b and references therein) melt compositions with mixed H₂O-CO₂ fluids, and numerical modeling (Gonnermann and Manga, 2005; Su and Huber, 2017).

In summary, the available results from HT-HP decompression experiments on evolved alkaline melts were never combined to fully constrain magma degassing along the whole decompression path from magma chamber to surface. Moreover, the existing results constitute a heterogeneous dataset, resulting from an unsystematic combination of experimental variables and essentially lacking information for CO₂-rich melts. This is especially true for experiments at near-liquidus temperatures, able to provide general information for specific melt compositions, regardless peculiar mineralogical assemblages.

In this study we achieved a broad petrological dataset (including 3D textural data for bubbles and microlites, glass volatile contents and nanolite occurrence) on experimental samples obtained through HT-HP isothermal decompression experiments, systematically changing final pressures (from 200 to 25 MPa), decompression rates (dP/dt from 0.01 to 1 MPa s⁻¹), volatile species (H₂O and CO₂) and their relative contents. We used this new dataset, in light of a review of all the available petrological experimental data from the literature, in order to fully constrain degassing mechanisms and timescales in evolved alkaline magmas.

Here, HT-HP isothermal decompression experiments were carried out on K-trachytic melts at super-liquidus temperature (1200 °C), using an internally heated pressure vessel. Experimental samples were characterized both 3D texturally (i.e., bubble and microlite parameters) and chemically (i.e., glass volatile content and nanolite investigation). Particularly, 3D texture was examined using X-ray computed microtomography, which allows to quantitatively analyse a large sample volume in 3D at high resolution, thus avoiding mathematical corrections to convert 2D data in 3D and reducing the associated uncertainties (Baker et al., 2012).

Finally, we attempted to reproduce numerically the experimental data, using an approach based on the method of moments for bubble size distributions, as a correct prediction of the degassing behaviour under controlled conditions can be a crucial step to decrypt information from natural alkaline rocks and monitoring signals.