Kiralj R., Ferreira M. M. C., “A Simple Prediction of  Crystal Density and Related Properties of Condensed Polynuclear Aromatic Hydrocarbons”. São Carlos, SP, 20-22/11/1997: XIV Reunião da Sociedade Brasileira de Cristalografia [14th Meetings of Brazilian Crystallographic Society], Resumos [Abstracts], (1997) 6. Oral, Poster.

A SIMPLE PREDICTION OF CRYSTAL DENSITY AND RELATED
PROPERTIES OF CONDENSED POLYNUCLEAR AROMATIC
HYDROCARBONS
 
 

*Rudolf Kiralj and **Márcia M. C. Ferreira
*Rudjer Boskovic Institute, POB 1016, HR-10001 Zagreb, Croatia
**Instituto de Química - UNICAMP, Campinas - SP  13081-970
 

              Previous   studies   on   crystal   packing   of   condensed   polynuclear   aromatic
hydrocarbons  were based  on calculations  of  various energetic  and  structural parameters
and  analysis   of   their  one-parameter  correlations.    A  special  attention   was  given  to
prediction of  crystal packing properties of  the  hydrocarbons from molecular structure and
the classification of packing types to herringbone (H),  sandwich-herringbone (SH),  g-  and
b-types, was established. In this work, samples of 27 crystal structures of planar condensed
polynuclear aromatic hydrocarbons containing only benzene rings,  were extracted from  the
Cambridge  Structural  Database.   A pocket calculator  and  chemical graphs  were used  to
make  a set of  17  variables:   4  dependent crystal packing variables   (the shortest unit cell
axis,  the crystal  density,   the  average  atomic  density,  the  average  electron  density),  6
independent  variables  based  on  the  size  of  a molecule   (the number  of  benzene  rings,
relative  molecular mass,   the  mass  and  number  fractions  of  C  atoms,   C/H ratio,   the
number of C atoms)  and  7  variables on molecular shape (numbers of different types of  C
atoms which could appear in a  molecule).   The  shortest  crystallographic  axis  versus  any
variable shows grouping of  samples into  10  HB,  6  SH,  9  g-  and  2  b-types  structures.
             Principal component analysis on the independent set  of  variables showed  the  set
to be three dimensional  (three principal components describe  93.7%  of  the total variance)
and  the  samples   are  roughly  grouped  except  kekulene,   according   to  their  molecular
structures,   as crystals  of  catacondensed and  pericondensed aromatic hydrocarbons.  The
density  and  the  average  atomic  volume  variables  correlate  well  with   the  independent
variables:   PLS regression using  3 principal components gave correlation above  R = 0.9  in
all cases.   Some of the independent variables,   correlate well inside the group  of  molecular
size  and  molecular   shape  variables,   as   shown   by  hierarchical  cluster  analysis.   The
densities increased  and  the  average  atomic volume decreased  as  molecular size increases.
The  fourteen pericondensed  in   comparison  to  thirteen  catacondensed  samples show  in
average to contain large molecules and never to pack as  HB  type,  while the catacondensed
never pack  as  b-type.   A  slight  increase  of  densities  of  pericondensed  samples  as  the
number  of  inner C atoms increases,  is  observed.

FAPESP
 
 
 
 
 

6