Lava balloons—peculiar products of basaltic submarine eruptions

Abstract

Between December 1998 and April 2001, a submarine basaltic eruption occurred west of Terceira Island, Azores (Portugal) in water depths between 300 and 1,000 m. Physical evidence for the eruption was provided by the periodic occurrence of hot lava “balloons” floating on the sea surface. The balloons consisted of a large gas-filled cavity surrounded by a thin shell (a few centimetres thick). The shells of the collected balloons are composed of two layers, termed the outer layer and the inner layer, defined by different bubble number density, bubble sizes and crystal content. The inner layer is further divided into three sublayers defined by more subtle differences in vesicularity. The outer layer is glassy, golden-coloured and highly porous. It shows signs of fluidal deformation and late-stage extension cracks. Interstitial glass contains 0.29 wt% H₂O and CO₂ is below detection. Melt inclusions contain up to 1.18 wt% H₂O and 1,500 ppm CO₂ (from different inclusions). Cooling rates of the outermost glass of the outer layer are found to be as high as 1,259 K/s. During ascent of low viscosity magma to the ocean floor, volatiles, dominated by CO₂, exsolved from the magma (melt + crystals). The buoyancy of the vapour phase that accumulated below a thin crust on lava ponded at the vent caused bulging and ultimately cracking of the crust. This allowed large bubbles (central cavity) surrounded by a film of vesicular magma (balloon shell) to leak into the water column. On contact with the seawater, the outermost part of the outer layer of the shell hyperquenched. If an entirely closed shell was produced during detachment, the trapped gas inside allowed buoyant rise. Only balloons with the right balance of physical properties (e.g. size and bulk density) rose all the way to the sea surface.

Appendices

Appendix 1

Whole-rock geochemistry

Rock powders were mixed with a flux of lithium metaborate and lithium tetraborate and fused in an induction furnace. The molten melt was poured into a solution of 5 % nitric acid containing an internal standard, and mixed continuously until completely dissolved (~30 min). The samples were analysed to determine major element compositions by ICP-MS at Activation Laboratories Ltd., Ontario (Canada). The relative standard deviation for major elements was less than 5 %. Three blanks and five controls were analysed per group of samples. Duplicates were fused and analysed every 15 samples. The instrument was recalibrated after 40 samples. Calibration was performed using BHVO, BR, GH and DRN international rock standards. Analytical precision (2σ) for major elements was about 1 %.

Appendix 2

Electron microprobe analysis

Glass, crystal phases, silicate melt inclusions (SMI) and their host crystals were analysed for major elements, plus S and Cl, by electron microprobe on a JEOL JXA-8800 Superprobe at the Institute for Research on Earth Evolution (IFREE), Japan Agency for Marine Earth Science and Technology (JAMSTEC). An accelerating voltage of 15 kV, a 15 nA beam current, a defocused 3 μm spot size and ZAF correction procedures were used. Na and K were analysed first on their respective spectrometers for 10 s on the peak and 5 s on the background. Peak and background analysis times for Si, Ti, Al, Fe, Mg and Ca were 20 and 10 s; for Mn, Ni, Cr, P were 30 and 10 s; for S were 50 and 25 s’ while for Cl, they were 240 and 120 s, respectively. Calibrations and spectrometer drift during measurements were checked by interspersing measurements on the secondary standards VG-2 glass, Saint John’s Island olivine, and scapolite amongst the sample measurements. To check for homogeneity at least 11 analyses were collected on the glass, at least three on selected crystals within the glassy samples, at least five on the SMI and at least six on the host crystal.

Appendix 3

FTIR spectroscopy

H₂O and CO₂ contents were determined by FTIR spectroscopy using a Varian FTS Stingray 700 Micro Image Analyser spectrometer at IFREE, JAMSTEC following the techniques of Nichols and Wysoczanski (2007). A UMA 600 microscope, attached to the spectrometer bench, was used to focus the beam on the area of interest. For the glass and fine particle spectra were collected over 512 scans with a beam size of 30 × 30 μm square, for the silicate melt inclusions (SMI) 1024 scans were collected mostly with a beam size of 20 × 20 μm, but some measurements were conducted at 15 × 15, 25 × 25 and 25 × 15 μm. On each glass sample, at least five spots were analysed; while on the fine particle, three spots were analysed to check for homogeneity while SMI were analysed from both sides. The concentrations of volatile species were calculated using the Beer–Lambert law (Stolper 1982). Peak heights (absorbance) above a linear baseline on spectra collected in transmitted light were used to calculate volatile concentrations. To determine total H₂O, the peak at ~3,500 cm⁻¹, caused by the O–H fundamental stretching vibration, was used with a molar absorptivity coefficient of 63 ± 5 l/mol cm (Dixon et al. 1988). CO₂ concentrations were calculated from the average height of the peaks at ~1,515 and ~1,435 cm⁻¹, caused by the asymmetric stretch of CO ²⁻₃ groups, with a molar absorptivity coefficient of 375 ± 20 l/mol cm (Fine and Stolper 1986). The bulk densities of glass, ash and SMI were calculated on the basis of their major element composition using the model of Lange and Carmichael (1987). Wafer thickness was determined using the frequency of interference fringes on spectra taken in reflected light through exactly the same area as was analysed in transmitted light. The SMI hosted in clinopyroxene was prepared so that it was exposed on both sides and thus the thickness of the inclusion equals the wafer thickness. However, some of the olivine-hosted SMI were prepared so that only one surface was exposed. In this case, the method described in Nichols and Wysoczanski (2007) was followed to obtain inclusion thickness.

Spectroscopic images of areas 350 × 350 μm were measured to compare the volatiles contained within the SMI with those in glass attached to the outside of the same crystal. The images were collected using a Varian Inc. Lancer focal plane array (FPA) camera attached to the UMA 600 microscope, set up to collect images across the mid-IR spectral region, 6,000–400 cm⁻¹. For more detailed discussions on imaging and the Lancer FPA camera, see Wysoczanski and Tani (2006) and Nichols and Wysoczanski (2007). For those SMI imaged, four spectra across the inclusion were used to add to the data collected by spot analysis and check for volatile homogeneity in the inclusion. The procedures to calculate the volatile data from these spectra are the same as those outlined above for the spot analysis.

Appendix 4

Differential scanning calorimetry

The structure of glass is mostly compositionally dependant and a careful characterization of its properties reveals the P-T-t conditions at which it formed. The variation of specific heat capacity (c _p ) with temperature was measured using a differential scanning calorimeter (DSC Netzsch® 404C at Corning Inc.). The c _p –curve, reflecting the natural cooling rate, was recorded during the first heating scan at 10 K min⁻¹. Subsequent repeated cooling and heating treatments, performed on the same sample, allowed further c _p –curves to be measured at known P-T-t conditions. A mathematical evaluation of the differences between the shape of the “natural” c _p –curve and those measured under known experimental conditions, allows the natural cooling rate across the glass transition region to be quantified. A detailed description of the measurement procedure and calibration, as well as the accuracy of the technique, can be found in Potuzak et al. (2008) and Guo et al. (2011).