Re: CCL:MOLECULAR MECH QUESTIONS

[Per-Ola Norrby <peon # - at - # medchem.dfh.dk>]

 On 1997  Dec 17, E. Lewars wrote:
 >Molecular mechanics (MM) gives strain energies.  Questions:
 >
 >Are these statements correct ? :
 >
 >(1)  Strain energy differences are enthalpy differences (deltaH).
 >     (If true, then deltaH under what conditions--zero K?  room temp? I
 >suppose
 >     it depends on the parameterization  of the force field: k_stretch etc
 >      might have been for, say, 0 K...?)
 	Close enough for practical purposes.  If you're a purist, read on
 below.
 >(2)   MM programs that calculate frequencies can in principle calculate
 >      entropies and thus free energies.
 	True.  You'll have to be very careful though, using the harmonic
 approximation for a low mode might give serious errors.  This problem is
 not unique to MM, most QM programs also use the harmonic approximation.  A
 few programs (both QM and MM) are smart enough to treat free rotations
 separately, but I don't know of any simple approximations that gives
 accurate entropies when you deal with intermediate modes (say, 10-50 cm-1,
 maybe even wider).
 	For the purists, an elaboration on (1):
 	Molecular mechanics energies will in principle give back the type
 of energies they are parameterized with.  A few force fields, like CFF95
 and MMFF, were parameterized from ab initio data, and will thus yield
 potential energies at the bottom of the energy well.  The difference
 between this number and the enthalpy at zero K is the zero point energy,
 and in energy DIFFERENCES this number can frequently be neglected.  Many
 force fields (for example, the MM2/MM3/MM4 programs) instead uses
 enthalpies directly, including the vibrational contribution into the
 potential energy surface.  This gives some tricky theoretical problems when
 you add a vibrational analysis to something that already includes a part of
 the vibratinal contribution, especially when you move away from stationary
 points.  In practice, this is ignored, and all molecular mechanics force
 fields are treated as if they yield a potential energy.
 	Note that enthalpies (and free energies and entropies) in principle
 are properties that depend on all contributing geometries within a
 potential energy well.  Thus,  you cannot really talk about the enthalpy of
 a single point on the potential energy surface.  From a potential energy
 surface, you may calculate the enthalpy for each well, but you cannot go
 backwards and deduce a potential energy surface from the enthalpies (other
 than postulating a PES and see if it gies a correct enthalpy, of course).
 	One more note: the programs MM3 and MM4 have partially overcome the
 problem by an end-of-calculation analysis that converts the vibrationally
 averaged properties to what would be predicted from several different types
 of experiments (including, the "bottom-of-the-well" properties from ab
 initio).  This seems to work well at least for bond lengths.  I'd be
 interested to know if some better theoretician than me can find out if you
 can really go back to a true potential energy surface though.
 	Holiday greetings from
 	Per-Ola Norrby
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  *  Per-Ola Norrby, Associate Professor
  *  The Royal Danish School of Pharmacy, Dept. of Med. Chem.
  *  Universitetsparken 2, DK 2100 Copenhagen, Denmark
  *  tel. +45-35376777-506, +45-35370850    fax +45-35372209
  *  Internet: peon # - at - # medchem.dfh.dk, http://compchem.dfh.dk/