Analytical tools

X-ray µCT was performed at the Department of Geosciences and Environment at the University of Lausanne (UNIL) with a Bruker Skyscan© 1173 at an accelerating voltage of 40 kV, a current of 200 µA without a filter. Pixel size was varied between 5.5 and 6.0 µm, exposure time was set to 800 ms, the rotation step to 0.225° with accumulating 5–6 frames for each position. Image reconstruction was performed using the NRecon software and the Skyscan software package was used for 3D visualization. X-ray µCT was employed to locate the diopside crystal seed in the quenched melt and to minimize the risk of destroying it during polishing. Additionally, as a diffusional anisotropy was expected to occur (Zhang et al. 2010), we were able to orient the sample prior to the polishing. The final goal was to cut the polished section such that it contained the long axis of the crystal, presumably the c-axis, which was best recognized and, therefore, the easiest recovered axis during the polishing (and another, undefined orientation perpendicular to the c-axis). The final polishing step was performed with a 1–0.25 µm polycrystalline diamond suspension. A detailed description of the polishing procedure is provided in the Electronic Appendix (incl. Fig. A1).

EBSD allowed the determination of the crystallographic orientation of the diopside crystal seed to define a suitable place to take the concentration profiles (along a certain axis) and to calculate the exact crystallographic direction of each profile within the crystal seed (detailed description in the Electronic Appendix; Fig. A2). All measurements were performed at the Scientific Center for Optical and Electron Microscopy (ScopeM) at ETH Zürich with an EBSD system (OIM 7 Pegasus with Hikari EBSD detector and Octane Super EDX detector; all by Ametek-EDAX) attached to a FEI Quanta 200F FEG-SEM with the following setting: tilt of the sample by 70°, acceleration voltage of 20 kV and low vacuum conditions (60 Pa as samples were non-carbon coated). The working distance varied between 15 and 20 mm. For EBSD pattern acquisition, the program OIM Data Collection was used; automated runs were performed at a speed of around 40–50 frames per second using camera pixel binning of 4 × 4. Subsequently, the program OIM Analysis was used to process the data. Cleaning of the data was performed based on (1) grain confidence index standardization and (2) neighbor orientation correlation with a grain tolerance angle of 5°. Further filtering was performed on a confidence index of 0.2 and grain size (variable for each sample) to extract the orientation map for the diopside crystal seed without surrounding matrix (Fig. A3).

EPMA measurements were conducted with a JEOL JXA-8200 electron microprobe equipped with five wavelength-dispersive spectrometers (WDS) at the Institute of Geochemistry and Petrology at ETH Zürich. The following standards were employed for major element analysis: fayalite for Fe, periclase for Mg, albite or acmite for Na, albite or anorthite for Al and wollastonite for Si and Ca. An acceleration voltage of 15 kV, a beam current of 20 nA with a focused beam for the crystal phases and a defocused beam with a spot size of 20 µm for the residual melt/interstitial eutectic mixture were employed. The counting time was 20 s on the peak and 10 s on the backgrounds. For Na the counting time was set to 10 s on the peak and 5 s on the backgrounds to minimize alkali loss. Measurements were performed to determine the composition of the vitrified starting material and of the phase assemblage in all experiments. To obtain statistically significant measurements, each phase was measured on ≥ 2 representative areas (diopside crystal seed) or crystals (anorthite, Ca-rich cpx or wollastonite) consisting of ≥ 8 measuring points.

High resolution element diffusion profiles across the interface between the newly grown crystal rim and the diopside crystal seed were measured by quantitative EDS. The EDS system (TEAM 4.2 with Octane Super by Ametek-EDAX) was attached to a FEI Quanta 200F FEG-SEM at the ScopeM at ETH Zürich. The device was equipped with a Schottky type field emission source. The EDS “TEAM” software was used to define standards by re-analyzing the same areas/crystal that were previously measured by EPMA (no polishing after the EPMA measurement) and they were taken as ‘known’ concentrations: the diopside crystal seed was used for Si, Ca and Mg and anorthite crystals for Al (Fig. 1c, d). The reduced area mode was used to compensate for possible inhomogeneities in the crystal seed or the anorthite. Measuring time was 500 s for each area. By combining these two standards (cpx crystal seed and anorthite), one merged standard was created for each experiment. Both, standardization and diffusion profiles, were measured under identical operating conditions: 10 kV acceleration voltage, an EDS count rate of around 40,000 counts per second daily adjusted as no absolute beam current readings were available and an amp time of 1.92 µs resulting in a detector resolution of 135 eV. The profiles were always measured perpendicular to the rim/crystal interface (Fig. 1e, f). As the number of measurement points per line was limited to 20, 2–4 lines composed of 20 measurement points and a line length of 1–2 µm were placed consecutively as close as possible resulting in total profile lengths of 2–6 µm (i.e., a lateral distance of 50–100 nm between neighboring points). Each point was measured for 30 s. Evaluation of the results of the zero-time experiments and of a measurement through the interface between a crack within the crystal seed (Sect. 2.4 and Fig. 2), allowed us to quantify the spatial resolution of the instrument that amounted to < 550 nm (full width at half maximum; Table A1).

a BSE image of the location of the profile measured through the crack/crystal interface (profile shown in b which was used to parameterize the beam size in the Di/An system. c–f Zero-time experiment indicating similar profile length as in b