Conformation searching

Conformers are isomers which have the same connectivity, but differ in
bond rotation positions. This includes ring flipping

You may be trying to model the most stable conformer, or you may wish to
model several of the more stable, which you presume are in equilibrium
with each other

Generally, your knowledge as a chemist will allow you to hypothesize roughly
the structures of the conformer(s) you are seeking

Start by putting together your rough model as close to your hypothesized
structure as you have patience for, using the facilities for bond rotation
provided in the modelling package you are using. If you start geometry
optimisation from close to your hypothesized structure, this improves the
chance that it will converge to that conformer rather than a different
one.

You may find that however close your starting point is, the minimisation
routine always goes somewhere else. In that case, for the modelling
method
you are using, and the minimisation software you are using, your hypothesis
was wrong. If there is an energy minimum at that geometry, it is
too shallow to be found.

A much more important problem is that, after all this, there are one or
more important conformers which you have not thought of, and which the
minimisation routine has not found by accident

You must make a sensible search for other conformers, otherwise you have
not modelled the system

The more elaborate modelling programs can do this automatically:
one popular method is the Monte Carlo method, as follows

The program starts from an already found conformer

Random changes are made to bond rotations

Geometry is optimised, starting from each such new point

A list is kept of found conformers

When each conformer has been found at least twice, the program stops

Even with a powerful computer and a fast method, this can take hours or
days for a large molecule

If you do not have a program to do it for you, you should attempt a manual
conformer search: you need to consider

bond rotation

ring flipping (which involves several bond rotations at the same time)

inversion at atoms with lone pairs, usually N, P, S

Make a list of what you plan to do, before you start, to make sure you
cover all the combinations, however unreasonable they may seem to you.
Tick off and annotate the list as you proceed.
Bond rotation

You do not need to rotate methyl groups, because all rotamers they can
minimise to are the same

You do need to rotate:

any unsymmetrically substituted C, e.g. CH_{2}Me

all OH groups (there could be H bonding in some positions, as well as preferred
p
bonding orientations for any OR group)

all amine groups

If you have two independent phenyl groups in a molecule, they often prefer
to have their planes perpendicular to each other if possible

Amide NRR' groups are usually planar, with a delocalised lone pair.
If R is not equal to R', try turning the group through 180 degrees
Ring flipping

You have to break the ring, rotate some bonds to different rotamers, then
join it up again

You need only consider rotamers for which some further, different rotation
will allow the ring to be rejoined

Go round the ring systematically
Inversion

Unless there are big groups on N, inversion tends to be fast on the NMR
timescale. However, you still need to find out whether the two invertomers
have similar enough energies for both to be important models

If the N is nearly flat, the minimisation routine may always jump through
to the more stable invertomer, so you may not be able to find out about
both

Three coordinate P or S invert much more slowly, and both invertomers may
be isolatable on the laboratory timescale. You need to model both.
For a discussion of conformation searching, you should read Chapter 4 of
Goodman's 'Chemical Applications of Molecular Modelling'