k3/k4 ) [endo product]/[exo product]
(2)
The effects of solvents on the selectivity and rate of the
model reaction has been commonly explained by assuming
a transition-state model. The mechanism is illustrated in
Scheme 1, and the energy profiles are illustrated schemati-
cally in Figure 1. The selectivity changes can be related to
the free energy difference between the solvated transition
states, and the reaction rate changes can be related to the
difference in the free energy of the solvated reagents and
solvated transition states (minimum energy reaction path).
For example, consider the dipole moments of the assumed
transition states for compounds 3 and 4 (Scheme 2). As a
result of the differences in orientation of the methyl acrylate
molecule in the transition state (TS), 3-TS has a larger dipole
moment than 4-TS. If the dipole moments are taken in
isolation, this suggests that the selectivity will be greater in
polar solvents where the TS leading to 3 will have a lower
free energy. Similarly, the polarity of both TSs is greater
than the polarity of the reagents. This implies an overall rate
increase in polar solvents. These observations led Berson et
al3 to develop an empirical measure of solvent polarity that
was based on the selectivity of this reaction.
Figure 1. Free-energy profile for the Diels-Alder reaction
showing the free-energy difference (∆∆Gq) between the transi-
tion states for compounds 3 and 4. Here the free energy of 3
and 4 are shown for convenience and are not necessarily equal.
Scheme 2
consequence of running reactions in concentrated solutions,
it becomes very difficult to make generalised assumptions
for the parameter values to be used in these correlations.
For example, the determination of the thermodynamics of
mixtures requires some degree of experimental data. Alter-
natively, molecular simulation has been used to gain a better
understanding of the local environment around the TSand
has the potential to enable parameter estimation. Changes
in the solution composition will occur over the course of
the reaction (i.e., as the reagents are consumed and the
products are created). As such, experiments conducted under
concentrated conditions will be more representative of
industrial conditions and allow a more relevant choice of
solvent for better yields, selectivity, and ultimately recov-
erability.
Unfortunately, the dominant interactions leading to both
rate and selectivity effects are more complicated than those
represented by the differences in the TS dipole moments.
Influences such as solvent polarity (π), the solvent-induced
solvophobicity (S), dielectric properties (ꢀ,µ) and the hydrogen-
bonding characteristics of the solvent (R, acceptor; â, donor)
have been shown3-5 to be correlated with the overall rate of
reaction and the selectivity.
Empirical links between the properties of solvents and
the rate and product selectivity have been developed.
Experimental studies of the dilute model reaction by Cativiela
et al.4-6 culminated in regression equations written in terms
of these common solvent parameters. The regressions were
developed using the Abboud-Abraham-Kamlett-Taft model7
to alleviate problems arising from the parameters not being
independent variables (such as Et(30) and π). Here the
solvent influence on the overall rate of reaction was mostly
attributed to differences in the solvent-induced solvopho-
bicity. This is an effect that has been supported by studies
in water-based solutions8 where hydrophobic effects were
the dominant influence. However, regressions for the selec-
tivity were found to have a greater dependency on the effects
of solvent polarity than on the solvent-induced solvopho-
bicity. These contrasting influences illustrate the difficulty
in making predictions of the effects of solvents on kinetics.
Previous experimental studies of the Diels-Alder and
other reactions have been undertaken in dilute conditionss
clearly an unrealistic situation in chemical processing. As a
Results and Discussion
Two independent experimental methodologies were de-
veloped to determine the kinetic properties of the Diels-
Alder reaction system. Small laboratory-scale reactions in
the solvents, hexane and methanol, were undertaken (see
Experimental Section). GC was used as the predominant
technique to measure the concentrations of the reaction
solution components. The second method involved Fourier
transform infrared spectroscopy (FTIR) to measure the
overall reaction rates for comparison with those determined
by the first method.
Different techniques are needed to analyse the raw data
generated from each experimental method. Data analysis
using FTIR was based on the determination of the concentra-
tion of compound 2. This involved a multiparameter factor
analysis to determine profiles for the absorbance changes of
each of the species in solution. Simple extrapolation was
then used to determine the concentration of 2. Analysis by
GC required a comprehensive procedure involving data
reconciliation (outlined below). Unfortunately, the inability
of the FTIR technique to distinguish between the endo and
(3) Berson, J. A.; Hamlet, Z.; Mueller, W. A. J. Am. Chem. Soc. 1962, 84,
297.
(4) Cativiela, C.; Garcia, J. I.; Mayoral, J. A.; Avenoza, A.; Peregrina, J. M.;
Roy, M. A. J. Phys. Org. Chem. 1991, 4, 48-52.
(5) Cativiela, C.; Garcia, J. I.; Mayoral, J. A.; Salvatella, L. J. Chem. Soc.,
Perkin Trans. 2 1994, 847-851.
(6) Cativiela, C.; Garcia, J. I.; Mayoral, J. A.; Roy, M. A.; Salvatella, L.;
Assfeld, X.; Ruiz-Lopez, M. F. J. Phys. Org. Chem. 1992, 5, 230-238.
(7) Abraham, M. H. Pure Appl. Chem. 1985, 57, 1055-1064.
(8) Breslow, R. Acc. Chem. Res. 1991, 24, 159.
472
•
Vol. 3, No. 6, 1999 / Organic Process Research & Development