We have used the Submillimeter Wave Astronomy Satellite (SWAS) to carry out
deep integrations on the N-j = 3(3) --> 1(2) transition of O-2 in a variet
y of Galactic molecular clouds. We here report no convincing detection in a
n initial set of observations of 20 sources. We compare O-2 integrated inte
nsities with those of (CO)-O-18 in a similarly sized beam and obtain 3 sigm
a upper limits for the O-2/(CO)-O-18 abundance ratio less than or equal to
2.3 in four clouds and less than or equal to 3.6 in five additional clouds.
Our lowest individual limit corresponds to N(O-2)/N(H-2) < 2.6 x 10(-7) (3
sigma). A combination of data from nine sources yields < N(O-2)/N(H-2)> =
[0.33 +/- 1.6 (3 sigma)] x 10(-7). These low limits, characterizing a varie
ty of clouds in different environments at different Galactocentric radii, i
ndicate that O-2 is not a major constituent of molecular clouds and is not
an important coolant. The abundance of O-2 is significantly lower than pred
icted by steady state single-component chemical models. The present results
are best understood in the context of cloud chemical and dynamical models
that include the interaction of gas-phase molecules and grain surfaces and/
or circulation of material between well-shielded and essentially unshielded
regions. This circulation may be powered by turbulence or other driving fo
rces that effectively keep molecular clouds chemically unevolved.