Kiralj R., Ferreira. M. M. C., "Quantitative Structure-Property Relationship (QSPR) Study of ¹⁷O Carbonyl Chemical Shifts in Substituted Benzaldehydes". Águas de Lindóia, SP, Brazil, 10-15/09/2006: 10th International Conference on Chemometrics in Analytical Chemistry (CAC-2006, CAC-X), Book of Abstracts (2006) P078. Poster 078.

10th International Conference on Chemometrics in Analytical Chemistry P078

Quantitative Structure-Property Relationship (QSPR) Study of
¹⁷O Carbonyl Chemical Shifts in Substituted Benzaldehydes

Rudolf Kiralj*, Márcia M. C. Ferreira   rudolf@iqm.unicamp.br

Laboratório de Quimiometria Teórica e Aplicada, Instituto de Química, Universidade Estadual
de Campinas, Campinas – SP, 13083-970 BRAZIL

Keywords: chemical shifts, chemometrics, benzaldehydes
_____________________________________________________________________________________
 

   Li and Li (LL) have proposed recently1 an empirical equation for calculation of  ¹⁷O  NMR chemical shifts
in  benzaldehydes.   This  equation  uses   contribution  of  individual   o-,  o’-,  m-, m’-   and   p-positioned
substituents and a correction constant for polar solvents.  However,  in spite of high predictive ability of the
equation,  due to its empirical nature,  it  cannot be validated properly as  a regression model.  Besides,  its
application  is  limited  to  benzaldehydes   that   include  a  small  set  of  substituents.   These  facts  have
encouraged    the  authors  of  this   work  to  develop  a  simple  and  fast   Quantitative Structure-Property
Relationship (QSPR) methodology for prediction of   ¹⁷O  NMR  carbonyl chemical shifts  in benzaldehydes
and related systems.
   Geometry  of  fifty substituted benzaldehydes with known   ¹⁷O  NMR shifts¹   (prediction set)  and of  ten
without known shifts  (prediction set)  was optimized  at  semi-empirical  PM3  level  and various geometric
and steric descriptors  accounting for properties of the benzene ring,  aldehyde group and their connecting
carbon-carbon bond, were calculated. QSPR models were based on these descriptors and on partial least
squares regression,  principal component regression  and  multiple linear regression.  Principal component
analysis,  hierarchical cluster analysis and crystal structure data  retrieved from  the  Cambridge Structural
Database  (CSD)  were used in order to get more  insight  into  the  chemical  background  of  relationships
between chemical shifts and molecular descriptors. All data were autoscalled prior to analysis.
   Five molecular descriptors were selected: nuclear-nuclear repulsion energy for  benzaldehyde  (C1)  and
the first aromatic  (C2)  carbon atoms,  electrostatic potential-based partial atomic charge  of  the  carbonyl
oxygen O,  standard deviation  of  the six C-C bond  lengths in the benzene ring,  the  C1-C2  bond  length,
and Mulliken partial atomic charge of C2.  These descriptors exhibit high correlation with experimental ¹⁷O
shifts,  with  absolute  correlation  coefficients  being  0.89  to  0.93.   Unlike LL,  the  chemical  shifts  were
averaged whenever possible (for 13 samples).   The  obtained  PLS  and  PCR (two principal components)
and also MLR regression model have Q > 0.93, R > 0.94, SEV £ 11,  SEP < 10,  and less than 14 samples
with   relative    error    over    10%.   The    models    were    cross-validated    (one-leave-out)    and    also
externally validated (10 samples excluded).   The  models were  capable  to  predict  chemical shifts  for 10
arbitrarily selected benzaldehydes,  while  the  LL  model could be applied only to four samples. Due to the
simplicity and short consumed time per sample,  the proposed  QSPR models can be used for prediction of
¹⁷O carbonyl shifts in substituted benzaldehydes and related systems.
   The exploratory analysis  and  selected  structures  from  the  CSD  show  that  the  increase  in  electron
density  and  decrease  in  chemical shift of  the  carbonyl oxygen  is  caused by  electron  donation  of  the
benzene ring and  substituents,   and  of  the  hydrogen  bond  established  between  the  aldehyde and  o-
hydroxyl group. In fact,  all  these  fragments  may  be  considered as integral  parts of the benzaldehyde p-
delocalized system. The variations  in  the  ¹⁷O  shifts  are  well  described  by five selected local molecular
descriptors  at  semi-empirical  level,   accounting  for  geometric  and electronic properties of the aldehyde
group, the benzene ring and their connecting carbon-carbon bond.  The influence of the hydrogen bonding
to the chemical shifts was noticed also.   The  first principal component  is  related  to  cumulative  electron
withdrawal/donation effects felt by the carbonyl oxygen,  whilst the second principal component is probably
related to the variations in the benzaldehyde heteroaromatic character.
   Electron withdrawal/donation  effects  caused  directly  or  indirectly  by  substituents,  hydrogen bonding
and benzaldehyde non-planarity  are the major factors that influence variations in electron density  and ¹⁷O
shifts.   These  phenomena  have   been   well-described  by  molecular  descriptors   from  semi-eimpirical
calculations, and the corresponding  QSPR  have  shown  to  be   suitable  for  prediction  of these shifts in
benzaldehydes.
 

Acknowledgment: FAPESP
__________________________________________________________________________________________________________________________

References

¹ Li L.-D.; Li L.-S. Magn. Reson. Chem. 2004, 42, 977-982.