This "tar" file contains the UNIX (Linux/AIX) version of the MNDO-MS
program system performing "quasi-Koopmans" calculations for predicting
primary electron impact mass spectrometric cleavages, as described in the
paper I.Mayer and A.Gomory, J.Mol.Struct. (THEOCHEM) 311, 331-341 (1994).
It is based on the QCPE program of Thiele and Dewar. Some auxiliary
programs are also included, automatizing completely the use.


A) After unpacking the file (command "tar -xvf mndoms-unix") one has to
issue the command "make all", and the "Makefile" will govern the
compilation of the four Fortran and C programs:

1. mndoms: the MNDO program with energy partitioning included. Source:

   mndo1.f, mndo2.f, mndo3.f, mndo4.f, mndo5.f, mndo6.f, givno.f, param.f,
   partic.f, pm.f, punio.f

2. bond2: calculates bond orders (Wiberg indices), free and total valences.


3. mnd: a utility program producing the MNDO input files necessary for the
   calculation of the neutral molecule and the ions with one electron
   deleted from one of the four highest occupied orbitals based on the
   output file generated by the geometry optimization run. Source:


4. atir: a utility program generating a temporary file with full card
   images (i.e. ends of lines padded by blanks), as required by the
   mndoms program. Source:


B) Make sure that the procedures (run or run-aix and mscript) are marked
executable (command "chmod +x filename").


1. One has to generate a standard MNDO input file for the molecule
studied, as required by the QCPE program of Thiele and Dewar. There are
software around doing this task - for instance PCMODEL 4.0 of Serena
Software definitely makes the job. (Unfortunately, the format is similar,
but not identical with that of MOPAC.) For an example see the file
"leucin.inp" included in the "tar".

The input file should have the name "molname.inp" where "molname" is an
arbitrary name. (This name will be used thorough.)

2. After the input file is generated, one has only to issue the command

        "run molname" ("run-aix molname")

and the script governs a geometry optimization run followed with 5 single
SCF calculations for the comparison of the neutral molecule and the ions
with one electron deleted from the orbitals HOMO, HOMO-1, HOMO-2 and
HOMO-3, respectively.

A number of input (extension "inp") files and the following output files
should be generated:

molnameopt.out -  a summary of the geometry optimization run. The
                   final geometry is not included - it can be found
                   in all the following "molname*.inp" files. (For an
                   example see the file "leucinopt.out" included in the

molnameneut.out -  a summary of the single point SCF calculation for
                   the neutral molecule at the optimized geometry.
                   In addition to the standard MNDO results, the matrices
                   of energy partitioning and bond orders - valences
                   is included. The format of the latter is discussed
                   below. (For an example see the file "leucinneut.out"
                   included in the "tar".)

molname(N).out  -  a summary of the single point "quasi-Koopmans" runs
Molname(N-1).out   with one electron deleted from the HOMO (orbital with
Molname(N-2).out   number N) and from some  next-to-homo orbitals (orbitals
Molname(N-3).out   N-1, N-2, N-3). In addition to the standard MNDO
                   results these contain some data about the orbital in
                   question, and the table of the "Appreciably changing
                   bond energies" which can be used to predict the
                   places of the primary bond cleavages. (As examples,
                   see files leucin27.out to leucin24.out.)

Predicting bond cleavages:

The primary bond cleavages are expected at the places where the bond
weakenings are significant (larger than, say, 0.7 or 1.0 eV) for the
ionization from HOMO (sometimes - e.g. for aromatic molecules - for
ionization from one of the next-to-HOMO orbitals). However, weakenings of
double and aromatic bonds and - in most cases - bonds formed by hydrogen
atoms should not be considered. (Sometimes weakenings of the bond between
the parent ion and a methyl group does not show up on the MS spectrum,
either.) These "static" data may be used for predicting the _places_ of
the primary bond cleavages, but usually not the relative ion intensities,
governed by a great number of different (mostly "dynamic") factors.

In the example of leucine molecule the HOMO is the 27-th occupied valence
orbital. Deleting an electron from it (see file "leucin27.out") one gets a
large weakening of the bond 1-2, which explains the leading peak with
m/z=86 in the MS spectrum, and a moderate weakening of the bond 2-3,
explaining the peaks with m/z=74 and 57. (The experimental spectrum also
contains lines with m/z=30 and 44 which cannot be attributed to simple
bond cleavages; they most probably originate from reorganization reactions
or from the products of the thermal decomposition of leucine.)

Matrices of bond orders and valences:

The bond orders, free and total valences give a further insight in the
electronic structure of the molecule and its ions. Actually they are
computed by the C-program "bond2" which extracts the necessary density
matrix elements from the original mndo output file MNDOC.OUT (which is
deleted by the script).

The format was determined by "historical factors". The numbering of the
atoms is given in the _second_ column, while the first column gives the
"free valences" as defined, e.g. in I. Mayer, Theor. Chim. Acta, 67, 315
(1985). (In the versions of MOPAC known to us another definition of free
valence is adapted, which we consider inadequate.) Then follows the matrix
of Wiberg indices in triangle form, with total atomic valences placed in
the diagonals.