70 71 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 0 / 2 1 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 0 / 2 1 Watchingmagma on themove Related publication: Dobson K. J., Allabar A., Bretagne E., Coumans J., CassidyM., Cimarelli C., Coats R., ConnolleyT., Courtois L., Dingwell D. B., Di Genova D., Fernando B., Fife J. L., Fyfe F., Gehne S., JonesT., Kendrick J. E., Kinvig H., Kolzenburg S., LavalléeY., Liu E., Llewellin E.W., Madden- Nadeau A., Madi K., Marone F., Morgan C., Oppenheimer J., Ploszajski A., Reid G., Schauroth J., Schlepütz C. M., Sellick C.,Vasseur J., von Aulock F.W., Wadsworth F. B.,Wiesmaier S. &Wanelik K. QuantifyingMicrostructural Evolution inMovingMagma. Front. Earth Sci. 8 , 287 (2020). DOI: 10.3389/ feart.2020.00287 Publication keywords: Volcanology; Rheology; Synchrotron; in situ ; Magma; X-ray tomography T he magma stored beneath volcanoes is an evolving mixture of molten rock (liquid), crystals (solid) and bubbles (gas). The amount and distributionof these threephases control how, and if, themagmaflows.Whenmagmaerupts and cools, it has averydifferent crystal and bubble content. As we cannot take amicroscope down into themagma beneath a volcano, we need to develop laboratory techniques to understand howandwhenmagmamoves. Previous experiments showed that interactions between the solid particles and gas bubbles control magma’s ability to flow. In this study, researchers recreated flowing magma, using high-speed X-ray imaging on Diamond Light Source’s Joint Engineering Environmental and Processing(JEEP)beamline(I12)towatchtheseinteractions.Thisresearchdevelopedthetechnicaltoolstolookatmagmaathighmagnification duringdeformationandflow.The 4Ddata (3dimensions plus time) showed, for thefirst time, howmuch the distributionof bubbles, liquid and crystals changes duringflow, howmany bubbles coalesce into larger bubbles, andhowdifferent regions of the sample behave very differently. Thisresearchispartofalargerprojectinvestigatinghowmagmastructureatthemicroscopicscalecontrolsflow.Abetterunderstandingofhow magma behaves will improve our ability to predict how a volcano will erupt. More widely, magmas are just one example of a complex multi- phase fluid. Themethods developed here can be used to investigate other similar systems, such as concrete, ceramics and certain foodstuffs. Magma is a constantly evolvingmixture of crystals and bubbles suspended in silicate melt. Magmatic behaviour as it starts to move, as it flows, fragments and during any subsequent eruption is controlled by the rheological properties at that specific point in time and space. A complex rheological evolution is driven by the changing melt viscosity, volume, componentry, and size distribution of the suspended phase(s). In most magmas (where the crystal +/-bubblesare>25-30volume%),thespatialdistributionof,and interactions between crystals and bubbles is also critical. A key challenge in volcanology is to describe magmatic processes as rheologically controlled behavioural ‘tipping points’, and ultimately to identify how these are recorded in volcano monitoring signals. Recent work on understanding the rheology of magmatic suspensions have demonstrated that changes in magma viscosity and the transition from Newtonian to non- Newtonian are controlled by the crystal and bubble content, shape, surface texture, size distributions and strain rate 1,2 . However, characterisation of the sample textures (critical for deriving rheological laws) is, without exception, restricted to snapshots before and after an experiment. To date, experimental work typically captures only a small piece of that evolution and then only the effects on bulk rheological behaviour. Understanding and predicting the rheological properties requires knowledge of particle-particle particle-bubble and bubble- bubble interaction microphysics that we currently lack because we cannot observe magmatic flow in situ . The rocks we can recover post-eruption contain textural and chemical overprints acquired during the entire evolution from initial melt formation to cooling, and do not represent the magma as it was in motion. In this study we deploy a bespoke high temperature furnace and the XRheo rheological apparatus on I12, exploiting the high-speed X-ray tomography capability of the beamline to capture the evolution of magmatic microstructure during flow for the first time.The experimental set up that we developed and the analytical protocols illustrated in Fig. 1 now permits detailed, in situ , characterisation of sample textures, allowing accurate parameterisation of the rheological data: and opening up an entirely new field of study in magma rheology. The focus of this study was primarily the successful deployment of the XRheo.The published study shows how standard wide gap Couette rheological testing protocols can be modified to work in a rotating frame of reference. The rotating reference is provided by the beamline and defined by the image acquisition requirements of the sample material. Initial imaging of the sample with the spindle applying zero torque gives a 3D understanding of the initial microstructure, before deformation is initiated (during imaging) by increasing the rotation speed of the spindle to the appropriate differential strain rate. By imaging before deformation and continuously through acceleration and into steady state deformation we were able to capture the evolving microstructure alongside the standard rheological measurements. By using a ’gapped’ acquisition protocol (1 tomography collected every n/2 rotations we were able to get spatially registered data, while balancing the competing needs of high acquisition rates (to prevent motion blur), obtaining measureable but not excessive displacements between 3D frames, and keeping total data volumes manageable. To explore a large experimental space we integrated complementary data from high viscosity-low strain rate experiments (>2 hours per experiment, ~1 s per 3D dataset) collected at the TOMCAT beamline (Swiss Light Source), with data from the full range of viscosity - strain rate conditions collected using high-speed monochromatic tomography at I12 (0.5 to 0.0125 s per 3D dataset). This covered off liquid viscosities of 10 -3 -10 4 Pa.s, crystal and bubble concentrations of 0-50 volume %, and more than 2 two orders of magnitude in strain rate without reaching the practical limit of data quality. Focusing first on simple bubble-bearing crystal-free systems and then working towards increasing microstructural complexity we were also able to identify the complex range of microstructural changes that happen during the onset of deformation. The spatial and temporal heterogeneity in the displacements continue through the acceleration phase and still occur well into ’steady state’ deformation. The textural evolution can be clearly seen in the reconstructed data, even when considering only a few time points within the longer time series (illustrated in Fig. 2). The benefit of the tomographic time series is that it enables more quantitative analysis of the heterogeneity in the microstructure using Digital Volume Correlation (DVC) (Fig. 3, the data shown are for the same part of the bubble-bearing time series as in Fig. 2). Although only just beginning the process of quantitative analysis in this study, by applying a multiscale nested global DVC 3 approach we can already identify regions of higher (and lower) than average displacement. The non-Newtonian behaviour and influence of the microstructural changes is clear in the heterogeneity and magnitudes of the displacements. For example, localised coalescence events accelerate regions of the melt beyond than maximum applied strain rate. While the detailed work of developing the micro-mechanical understanding of phase interactions recorded here is the focus of ongoing work; this study showcases the potential of real-time tomography and in situ experiments to quantify the complex volcanic and magmatic processes of flow and deformation, and could provide new insight on a much wider variety of complex fluids. References: 1. Mader H. M. et al. The rheology of two-phase magmas: A review and analysis. J. Volcanol. Geotherm. Res. 257 , 135–158 (2013). DOI: 10.1016/j. jvolgeores.2013.02.014 2. Truby J. M. et al. The rheology of three-phase suspensions at low bubble capillary number. Proc. R. Soc. A Math. Phys. Eng. Sci. 471 , 20140557 (2015). DOI: 10.1098/rspa.2014.0557 3. Madi K. et al. Computation of full-field displacements in a scaffold implant using digital volume correlation and finite element analysis. Med. Eng. Phys. 35 , 1298–1312 (2013). DOI: 10.1016/j. medengphy.2013.02.001 Funding acknowledgement: This work was supported by NERC M018687/1 & M018687/2, ERC Starting Grant 406388 (SLiM), ERC ADV Grants 247076 (EVOKES) and 834225 (EAVESDROP) and the H2020 Marie Skłodowska-Curie fellowship DYNAVOLC – No.795044, and benefited from access to TOMCAT beamline X02DA of the SLS under proposal 20150413, as well as to Diamond Light Source for time on I12-JEEP under proposal EE15898. Corresponding author: Dr Katherine Dobson, University of Strathclyde,
[email protected] Imaging andMicroscopy Group Beamline I12 Figure 1: The XRheo system installed at I12 (a) Photo; (b) schematic. This system can be easily adapted to other synchrotron or laboratory imaging systems and used ex situ for laboratory bench testing. (c) single use cup and spindle configuration for high temperature set up (d) typical acquisition protocol used in this study, with the timing of the repeated multi-scan acquisitions adjusted to fit key sections of a deformation cycle. Figure 2: 2D greyscale images frommid-height slice perpendicular to the rotation axis (see insert) showing the microstructural evolution of two suspensions. A 1 mm grid has been overlain all images. (a) steady state deformation phase of a high temperature bubble bearing synthetic magmatic sample (900 ° C, 0.3 rpm differential spindle speed). Several bubbles are marked in frames #69-71 so you can track motion. (b) steady state deformation of a low temperature analogue system (N2700000 oil + 30% by volume olivine crystals, room temperature, 5 rpm differential spindle speed). Figure 3: DVC outputs from several sequential time steps during acceleration (top) and steady state (bottom) deformation (including those shown in Fig. 2a).