Recalescence in Silicate Melts: More Than Just a Flash in the Pan?

Abstract:

Rapid cooling of silicate melts facilitates undercooling, where nucleation is delayed and crystal growth begins at temperatures below equilibrium. In such cases, rapid crystal growth and release of latent heat of crystallization can produce net heating (recalescence). We have documented recalescence in Fe-Mg pyroxene and komatiite melts, using a thermal imaging (FLIR) camera for melt volumes on the order of cm³, and using differential scanning calorimetry for melt volumes on the order of mm³. On cooling Fe_0.8Mg_1.2Si₂O₆ liquid from ~1600˚C in air, at ~30˚C/s, crystallization begins at ~1110˚C. Averaging over the whole base of the crucible (~10cm²), the observed temperature increase is ~100˚C and it takes ~2.5 seconds to attain the thermal peak. Crystallization and heating can be seen migrating across the melt volume together. When looking at a 3x3 pixel spot (~1mm² in our setup), reheating to >1270˚C occurred in ~1 s. About 30 mg of the same melt was cooled in a differential scanning calorimeter. Cooled at ~1˚C/s, two distinct crystallization peaks were seen at ~1450 and ~1270˚C. Cooled at ~2˚C/s, the first peak was delayed to ~1315˚C and merged with the second. Examination of recovered samples indicates crystallization of enstatite, followed by Fe-oxides and tridymite in a silica-rich glass matrix. On cooling of komatiite liquid from ~1600˚C at ~50˚C/s, crystallization begins at ~1080˚C. The average temperature plateaus for ~2s and then continues cooling. When looking at a 3x3 pixel spot, heating of ~10˚C could be detected only by comparing different video frames. We conclude that (i) thermal imaging of lava flows needs to be conducted with mm-scale spatial resolution to assess true temperature fluctuations, and (ii) thermal models that incorporate latent heat of crystallization as an “effective heat capacity” term do not allow recalescence, and may inaccurately capture the thermal history of rapidly cooled lava.