S. Hurwitz et O. Navon, BUBBLE NUCLEATION IN RHYOLITIC MELTS - EXPERIMENTS AT HIGH-PRESSURE, TEMPERATURE, AND WATER-CONTENT, Earth and planetary science letters, 122(3-4), 1994, pp. 267-280
We report the first measurement of bubble nucleation in hydrated rhyol
itic melts in response to pressure release. Two rhyolitic obsidians, o
ne containing less than 1% of microlites of Fe-Ti oxides and the other
about 20% of various crystals were hydrated at 150 MPa and 780-850-de
grees-C. After saturation was reached (5.3-5.5 wt% water), pressure wa
s lowered and the samples were allowed to nucleate and grow bubbles fo
r various amounts of time, before the final, rapid quenching of the ex
periments. The results demonstrate the importance of heterogeneous nuc
leation. Microlites of Fe-Ti oxides are very efficient as sites for bu
bble nucleation. In their presence, modest nucleation was observed eve
n after decompression by < 1 MPa, and decompression of more than 5 MPa
produced extensive nucleation (10(6)-10(8) bubbles cm-3). In the abse
nce of microlites, no nucleation occurred at DELTAP < 10 MPa. At DELTA
P > 10 MPa, bubbles also nucleated on crystals of biotite, zircon and
apatite. Modest nucleation (10(3) - 10(5) cm-3) took place even in cry
stal-free samples, but it was still heterogeneous. When DELTAP exceede
d 80 MPa, nucleation in crystal-free samples became extensive (10(5)-1
0(7) cm-3). The lack of correlation of bubble density with either time
or decompression suggests that nucleation was still heterogeneous. Nu
cleation rates were controlled mainly by the availability of sites. Ra
tes were faster than 10(6) cm-3 s-1 when microlites were present, and
faster than 10(5) cm-3 s-1 in the absence of microlites at DELTAP > 70
MPa. Narrow size distributions in most samples suggests that nucleati
on took place immediately after the pressure drop. The experimental da
ta we present here indicate that the presence or absence of efficient
nucleation sites can lead to two distinct modes of bubble formation. W
hen a large number of efficient sites (e.g., Fe-Ti oxide) are present,
bubble nucleation requires very little supersaturation, and to a good
approximation, gas and magma are in equilibrium. In magmas that are c
rystal-free or contain crystals that are inefficient at nucleating bub
bles, very high degrees of supersaturation are required in order to in
itiate nucleation. These two modes of exsolution may lead to contrasti
ng styles of convection, pressure build up and eruption.