Part of my job is designing heatshields for planetary probes. I wrote this
atmospheric entry program which included a routine modeling the
thermodynamics of an arbitrary atmosphere (necessary for describing real
gas effects). I developed this routine using the Gordon-McBride equilibrium
thermochemistry code as my validation standard. I could not use Gordon-McBride
directly since it was a general purpose code (originally used as a solid
rocket combustion code) and way too slow. For my code, I started out with
air (nitrogen-oxygen-argon) and after much fiddling could reproduce the
Gordon-McBride result to machine accuracy. I then moved on to Mars (carbon
dioxide-nitrogen-argon) and got that to machine accuracy as well. The
atmosphere of Venus is very similar to Mars but must allow for a much higher
pressure. This lead to screwball species (ionized molecular nitrogen) that
had to be included in the model. After Venus I did Jupiter (hydrogen-helium)
which was relatively easy. At that point (feeling confident with success) I
decided to try the really tough one which is Titan (nitrogen-methane). When
I first ran Titan, my code failed instantly (zero division, segmentation
faults, you name it). I assumed it was a simple logic bug or garbage typed
into the data tables. However I could not find the error. I then started
playing around with temperature. I found that for very low temperature (room
temperature) the code worked fine for Titan. I then tried extremely high
temperature (8000 deg.K) and again the code worked fine. It was only for
mid-range temperature (around 3000 deg.K) that the code failed radically.
At that point I asked myself: "What's Gordon-McBride doing here?". So I
ran Gordon-McBride and noticed that one of the species was labeled as
"C(g)". When I first saw that I thought it meant "carbon gas" but then
I saw that carbon gas was simply labeled as "C". What "C(g)" really meant
was graphite dust. Then I realized that the methane in Titan's atmosphere
was decomposing into graphite dust and hydrogen gas if it was heated to the
right temperature. Dust of course is a solid and my atmosphere code did
not support phase change from gas to solid. This was why it was going nuts.
However at that point my curiosity was aroused. I played around with the
Gordon-McBride code some more and saw that some fairly complex species could
be generated, i.e. acytelene, ethane, propane, etc. There is no gaseous
carbon dioxide in Titan's atmosphere because it's so cold and all of the
original CO2 condensed out. However just for fun I cooked up a mythical
ancient Titan atmosphere that had 70% nitrogen, 20% methane and the rest
carbon dioxide. I then set it at a high temperature. The number of molecular
species that Gordon-McBride generated from this initial mixture was
phenomenal. However what was truly striking was among these species were
amino acid precursors. What the code showed was if one started with a warm
Titan atmosphere (allowing for CO2) then prebiological chemistry will occur
automaticly. I should emphasize that the mathematics used in both the
Gordon-McBride code and my code were very primitive. We both used
specific heat curve fits that were based upon partition functions. These
partition functions came straight from line spectra data. The final
solutions used classical Gibbs or Helmholtz Free energy methods.
Biological chemistry really does come from basic quantum chemistry. This
of course is well known. People have long said that Titan is extremely
interesting because it represents the pre-biological state of the Earth
billions of years ago. I just found it funky to see it happen right in
front of me.

Gary