Magma fragmentation is the defining feature of explosive volcanism. Yet of all the processes involved in explosively erupting systems (e.g. bubble nucleation, bubble growth, crystallisation, foaming, fragmentation, postfragmentational annealing) fragmentation is possibly the least understood. Not even the basic mechanism is agreed upon. This is unfortunate because knowledge of the mechanism of fragmentation is a neccessary prerequisite for any reliable quantitative modelling of the conditions necessary for explosive activity, for predictive modelling of the frequency, intensity, duration of eruptions and would undoubtedly enhance the interpretation of the physical characteristics of the pyroclastic products to serve as a guide in the interpretation of pyroclastic rocks, both modern and ancient.
Several detailed investigations, including theoretical and laboratory studies, on magma degassing and fragmentation have been developed and reviewed recently in attempts to constrain the possible scenarios for developing the conditions necessary for explosive fragmentation of magma and to try to create a mechanistic understanding of the fragmentation process itself. From this combination of field-based characterization, theoretical and numerical simulations and high pressure � high temperature experimental testing, a general picture of the sequence of events leading to the fragmentation of silicic magma is emerging but several important mechanistic details remain to be clarified. One of the most important of these, the fragmentation event, constitutes the endmember of a series of processes which involves the nucleation of bubbles in the oversaturated magma, their growth by mass transfer of volatiles and expansion due to progressive decompression during magma ascent, and the rupture of the melt phase. The debate on fragmentation is focussed mainly on two possible scenarios. In one scenario, magma fragmentation is seen to be a consequence of the textural evolution of the magma to a highly foamed state where disintegration of the walls separating bubbles becomes inevitable due to foam collapse (i.e., foam disintegration in an accelerating flow). In a mechanical sense we may view this foam disaggregation as a consequence of a cumulative strain involved in the bubble growth process with fragmentation occuring as the strain criterion is overcome. Except in an indirect way, fragmentation via this mechanism should not be influenced by the rate of bubble growth/foaming but rather by its cumulative extent. In another scenario, magma is fragmented by stresses that exceed the tensile strengh of the condensed phases in the magma (melt ± crystals). For melt dominated systems, this implies that the strain rate imposed on the magma during fragmentation is close to that which generates unrelaxed or viscoelastic response of the melt phase. This mechanism contains then a strain rate criterion for fragmentation and is not directly dependent on the extent of vesiculation to achieve this criterion. These two contrasting hypotheses can be expected to have quite different implications for the conditions and consequences of the fragmentation process.
An elegant way to evaluate fragmentation mechanisms is provided by the study of the chief product of fragmentation, the pyroclasts. Explosive eruptions of foamed silicic magmas produce one or both of two types of pumice clasts. These are (1) the so-called "woody" pumice or "tube" pumice where extreme bubble elongation is preserved and (2) pumices where the bubbles have a more or less equant or spherical shape, here "spherical pumices" . Pyroclastic deposits containing both types of pumices as well as those composed exclusively of spherical pumice appear to be common, whereas deposits exclusively containing tube pumices are rare. Perhaps for this reason, little attention has been focussed on the careful description of tube pumice. This is especially disappointing because the texture of tube pumices (see below) provide strong evidence to support the notion that such samples do indeed represent material effectively quenched (chemically and texturally) at the fragmentation event. Thus tube pumice may well represent the best strain marker of eruptive processes at fragmentation that is available to us.
To evaluate this possibility, we provide here,
for the first time, a chemical, physical and microstructural description
of tube pumice from a deposit composed exclusively of water-rich and virtually
crystal-free tube pumices. In particular, we draw attention to microdomains
where compressional strain has been incorporated as localised zones of
ductile shear deformation after vesiculation and prior to fragmentation
(Fig. 3.7-7). From a quantification
Fig. 3.7-7: Hand specimen (a) and thin section (b) of tube pumice. Sections parallel and normal to shear planes reveal two types of deformation, (1) sinistral shear deformations that are located on sinistral shear planes and (2) dextral shear deformations placed on dextral shear planes. Bending of bubble walls along sinistral shear planes varies from a few degrees to 45º to the tubes. Near 45º, the walls of adjacent bubbles start to contact, the bubble width is a minimum, and the shear plane is best defined, yielding a cleavage crenulation of several centimeters length.
of strain rates operative during foam elongation,
localized shear deformation and fragmentation, we infer that the highly
foamed magma was subjected to multiple rapid stress events that led to
widespread longitudinal compression of the tube foam via plastic or viscoelastic
melt deformation and longitudinal tension of the pre-compressed tube foam
leading to brittle fragmentation (Fig. 3.7-8). We further infer that these
processes were followed immediately by extremely rapid quenching to preserve
these features. Thus tube pumice records the late transition of magma deformation
from viscous, through viscoelastic to brittle response immediately prior
to and during the explosive fragmentation of several cubic kms of hydrous
Fig 3.7-8: Synthetic pattern of the deformational structures observed in pumice clasts from a pyroclastic flow deposit. The strain state and stress state ellipses for each specific structure is also shown. Bubble nucleation is assumed to occur under a hydrostatic tension regime, where 1=2=3 on an imaginary point corresponding to the center of the bubble. In the samples studied here other stress fields appear superimposed over the hydrostatic stress field causing stretching and the two types of crenulation of bubbles.