(1997-Present)

*Publication Numbers:* 25,29,36.

The formation of molecular hydrogen in the insterstellar

medium is a process of fundamental importance.

It was recognized long ago that H$_2$ cannot
form in the gas phase

efficiently enough to account for its abundance.

It was thus proposed that dust grains act as
catalysts and a theoretical

framework was developed for the calculation of
the hydrogen

recombination rate.

For a number of years, my colleagues

G. Vidali and V. Pirronello

have been working on experimental studies of

hydrogen recombination on astrophysically relevant
surfaces

in order to examine this proposal.

The aim of these experiments is to examine the

efficiency of such surfaces as catalysts for
hydrogen

recombination and to find out

whether such processes may account for the abundance
of

H$_2$ in interstellar clouds.

The desorption spectra of

HD (which was measured instead of H$_2$ to reduce
noise)

indicated that the kinetics was of the second
order at low

coverages of H and D.

This lead to an expression for the formation

rate of H$_2$ in clouds which is

quadratic in the density of H atoms,

differing

from the commonly used expression of

Hollenbach and Salpeter which is linear in the
density.

I became involved in this project after realizing
that the same

rate-equation analysis used in our studies of
thin film

growth is required in order to

interpret the experiments, and to

bridge the huge gap between the laboratory and
the astrophysical

time scales.

Using a rate equation model, we were

able to find an analytical solution of the problem
under steady state

conditions, and showed that both

the linear and quadratic expressions

are obtained as two limits

of the same model.

Later, we developed a more complete rate equation
model

for diffusion and recombination of hydrogen on
surfaces.

Using our model for quantitative analysis of
the experimental

results, we obtained excellent fits to the experimental

data which led to firm predictions for the values
of the relevant

activation energies. Using computer simulations
we managed

to connect between the experimental time scales
(minutes)

and the astrophysical time scales (millions of
years), and to

predict the recombination efficiency of the experimentally
used

surfaces.