values of concentrations and Global Quantities and the Kinetic Parameters. Thus, we
can store and retrieve different candidate solutions to our problem.
5 Case Examples
Now we discuss some case examples to show how DFT-based microkinetic simulations can be used to get a deep understanding of a reaction mechanism that, in turn,
may help improve reaction conditions or the reaction mechanism itself (better
reaction steps, catalysts, or additives).
5.1 Example 1
Catalyst deactivation [17] is one of the key mechanisms that determine whether a
catalytic process is suitable for large-scale manufacturing. Finding deactivation
mechanisms is therefore one of the most relevant tasks in the kinetic modeling of
catalytic processes for industrial applications because, once the mechanism is
known, it is possible to devise methods to avoid or mitigate it. Deactivation mechanisms are usually sought by trying to figure out possible chemical reactions that
temporarily trap or permanently degrade the catalyst.
The case presented here is an example of a particular deactivation mechanism
(Fig. 11) in which, according to our proposed model [14], a solvent THF molecule
temporarily locks a substrate-catalyst intermediate (A) into an inactive state (LÁTHF)
by hindering the return from a reversible reaction (r8). It is experimentally known
that addition of collidine hydrochloride (collÁHCl) inhibits the deactivation mechanism. In our proposed model, based on DFT calculations, collÁHCl forms a complex
(LÁcollÁHCl) that prevents the formation of the L ÁTHF complex.
We calculated the DFT energies at B3LYP-D3/6-31 + G(d, p) level of theory. As
shown in Fig. 2 above, the results simulated with the as-calculated DFT values are
useless, especially for experiments 2, 3, and 4. Figure 11 shows experiments at two
different concentrations without (Exp1, Exp2) and with (Exp3, Exp4) collÁHCl.
Exp1 exhibits strong catalyst deactivation effects (slow conversion) after about
5 min. We used only three (plus the initial one) experimental data points (large
cross symbols in Fig. 11a) to fine-tune the raw DFT values within the calculation
error bars. The simulator (input file EventsSequence.cps from SI) closely reproduces
all the experimental data, having “seen” only three data points, used for fine-tuning
the DFT values. The DFT-based simulator, which has not seen any experiments with
collÁHCl, predicts the drastic suppression of catalyst deactivation, as shown by
comparing the Exp1, Exp3 curves. It also predicts other four experiments for
different concentrations of epoxide, catalyst, and additive. Considering that the
only guidance to optimize this 32-barrier simulator has been three experimental
DFT-Based Microkinetic Simulations: A Bridge Between Experiment and Theory in. . .
97
can store and retrieve different candidate solutions to our problem.
5 Case Examples
Now we discuss some case examples to show how DFT-based microkinetic simulations can be used to get a deep understanding of a reaction mechanism that, in turn,
may help improve reaction conditions or the reaction mechanism itself (better
reaction steps, catalysts, or additives).
5.1 Example 1
Catalyst deactivation [17] is one of the key mechanisms that determine whether a
catalytic process is suitable for large-scale manufacturing. Finding deactivation
mechanisms is therefore one of the most relevant tasks in the kinetic modeling of
catalytic processes for industrial applications because, once the mechanism is
known, it is possible to devise methods to avoid or mitigate it. Deactivation mechanisms are usually sought by trying to figure out possible chemical reactions that
temporarily trap or permanently degrade the catalyst.
The case presented here is an example of a particular deactivation mechanism
(Fig. 11) in which, according to our proposed model [14], a solvent THF molecule
temporarily locks a substrate-catalyst intermediate (A) into an inactive state (LÁTHF)
by hindering the return from a reversible reaction (r8). It is experimentally known
that addition of collidine hydrochloride (collÁHCl) inhibits the deactivation mechanism. In our proposed model, based on DFT calculations, collÁHCl forms a complex
(LÁcollÁHCl) that prevents the formation of the L ÁTHF complex.
We calculated the DFT energies at B3LYP-D3/6-31 + G(d, p) level of theory. As
shown in Fig. 2 above, the results simulated with the as-calculated DFT values are
useless, especially for experiments 2, 3, and 4. Figure 11 shows experiments at two
different concentrations without (Exp1, Exp2) and with (Exp3, Exp4) collÁHCl.
Exp1 exhibits strong catalyst deactivation effects (slow conversion) after about
5 min. We used only three (plus the initial one) experimental data points (large
cross symbols in Fig. 11a) to fine-tune the raw DFT values within the calculation
error bars. The simulator (input file EventsSequence.cps from SI) closely reproduces
all the experimental data, having “seen” only three data points, used for fine-tuning
the DFT values. The DFT-based simulator, which has not seen any experiments with
collÁHCl, predicts the drastic suppression of catalyst deactivation, as shown by
comparing the Exp1, Exp3 curves. It also predicts other four experiments for
different concentrations of epoxide, catalyst, and additive. Considering that the
only guidance to optimize this 32-barrier simulator has been three experimental
DFT-Based Microkinetic Simulations: A Bridge Between Experiment and Theory in. . .
97