Large and multi scale mechanistic modeling of Diels-Alder reactions
- Authors: Isamura, Bienfait Kabuyaya
- Date: 2022-04-06
- Subjects: Computational chemistry , Diels-Alder reaction , Python (Computer program language) , Reaction force theory , Fullerenes , Diolefins , AMADAR (Automated workflow for Mechanistic Analysis of Diels-Alder Reactions , ONIOM
- Language: English
- Type: Master's thesis , text
- Identifier: http://hdl.handle.net/10962/232317 , vital:49981
- Description: The [4+2] cycloaddition reaction between conjugated dienes and substituted alkenes is known as the Diels-Alder (DA) reaction, in honor of two German chemists, Otto Diels and Kurt Alder, who first reported this marvelous chemical transformation. The DA reaction is one of the most popular reactions in organic chemistry, allowing for the regio- and stereospecific establishment of six-membered rings with up to four stereogenic centers. This pericyclic reaction has found many applications in areas as diverse as natural products chemistry, polymer chemistry, and agrochemistry. Over the past decades, the mechanism of the Diels-Alder (DA) reaction has been the subject of numerous studies, dealing with questions as diverse as the mechanistic pathway, the synchronicity, the use of catalysts, the effect of solvents and salts, etc. On the other hand, as an example, fullerenes (and particularly [60] fullerene) have been found to act as good dienophiles in DA reactions to the extent that many functionalized fullerenes with interesting applications are still synthesized by reacting C60 with dienes. However, despite the very abundant literature about the mechanism of the DA reaction, some pertinent questions have been still pending, including, without being restricted to, the prediction of transition state (TS) geometries and the modeling of DA reactions involving large systems, such as those of C60 fullerene. It must be emphasized that TSs are not easy to predict and the main reason is that many existing algorithms require that the search is initiated from a good starting point (guess TS), which must be very similar to the actual TS. This problem is even more difficult when many TSs are to be located as may be the case in large-scale studies. Moreover, due to the large size of the C60 molecule, the usage of accurate high-level computational methods in the investigation of its reactivity towards dienes is computationally costly, implying the need to find the best threshold between accuracy and computational cost. Therefore, the present study was carried out to contribute to solving the problems of large-scale prediction of DA transition state geometries and the multi-scale modeling of C60 fullerene DA reactions. To address the first problem (large-scale prediction of TSs), we have developed a python program named “AMADAR”, which predicts an unlimited number of DA transition states, using only the SMILES strings of the cycloadducts. AMADAR is customizable and allows for the description of intramolecular DA reactions as well as systems resulting in competing paths. In addition, The AMADAR tool contains two separate modules that perform reaction force analyses and atomic decomposition of energy derivatives from the predicted Intrinsic Reaction Coordinates (IRC) paths. The performance of AMADAR was assessed using 2000 DA cycloadducts and showed a success rate of ~ 95%. Most of the errors were due to basis set inconsistencies or convergence issues that we are still working on. Furthermore, a set of 150 IRC paths generated by the AMADAR program were analyzed to get insight into the (a)synchronicity of DA reactions. This investigation confirmed that the reaction force constant 𝜅 (second derivatives of the system energy with respect to the reaction coordinate) was a good indicator of synchronicity in DA reactions. A close inspection of the profile of 𝜅 has enabled us to propose an alternative classification of DA reactions based on their synchronicity degree, in terms of (quasi)-synchronous, moderate asynchronous, asynchronous, and likely two-steps DA reactions. Natural population analyses seemed to indicate that the global maximum of the reaction force constant could be identified with the formation of all the bonds in the reaction site. Finally, the atomic resolution of energy derivatives suggested that the mechanism of the DA reaction involves two inner elementary processes associated with the formation of each C-C bond. A striking mechanistic difference between synchronous and asynchronous DA reactions emerging from this study is that, in asynchronous reactions, the driving and retarding forces are mainly caused by the fast and slow-forming bonds (elementary process) respectively, while in the case of synchronous ones both elementary processes retard and drive the process concomitantly and equivalently. Regarding the DA reaction of C60 fullerene that was considered to illustrate the problem of multiscale modeling, we have constructed 12 ONIOM2 and 10 ONIOM3 models combining five semi-empirical methods (AM1, PM3, PM3MM, PDDG, PM6) and the LDA(SVWN) functional in conjunction with the B3LYP/6-31G(d) level. Then, their accuracy and efficiency were assessed in comparison with the pure B3LYP/6-31G(d) level considering first the DA reaction between C60 and cyclopentadiene whose experimental data are available. Further, different DFT functionals were employed in place of the B3LYP functional to describe the higher-layer of the best ONIOM partition, and the results obtained were compared to experimental data. At this step, the ONIOM2(M06-2X/6-31 G(d): SVWN/STO-3G) model, where the higher layer encompasses the diene and pyracyclene portion of C60, was found to provide the best tradeoff between accuracy and cost, with respect to experimental data. This model showed errors lower than 2.6 and 2.0 kcal/mol for the estimation of the activation and reaction enthalpies respectively. We have also demonstrated, by comparing several ONIOM2(DFT/6-31G(d): SVWN/STO-3G) models, the importance of dispersion corrections in the accurate estimation of reaction and activation energies. Finally, we have considered a set of 21 dienes, including anthracene, 1,3-butadiene, 1,3-cyclopentadiene, furan, thiophene, selenothiophene, pyrrole and their mono-cyano and hydroxyl derivatives to get insight into the DA reaction of C60 using the best ONIOM2(M06-2X/6-31 G(d): SVWN/STO-3G) model. For a given diene and its derivatives, the analysis of frontier molecular orbitals provides a consistent explanation for the substituent effect on the activation barrier. It revealed that electron-donating (withdrawing) groups such as -OH (–CN) cut down on the activation barrier of the reaction by lowering (extending) of the HOMOdiene – LUMOC60 gap and consequently enhancing (weakening) the interaction between the two reactants. Further, the decomposition of the activation energy into the strain and interaction components suggested that, for a given diene, electron-donating groups (here –OH) diminish the height of the activation barrier not only by favoring the attractive interaction between the diene and C60, but also by reducing the strain energy of the system; the opposite effect is observed for electron-withdrawing groups (here –CN). In contrast with some previous findings on typical DA reactions, we could not infer any general rule applicable to the entire dataset for the prediction of activation energies because the latter does not correlate well with either of the TS polarity, electrophilicity of the diene, or the reaction energy. , Thesis (MSc) -- Faculty of Science, Chemistry, 2022
- Full Text:
- Date Issued: 2022-04-06
- Authors: Isamura, Bienfait Kabuyaya
- Date: 2022-04-06
- Subjects: Computational chemistry , Diels-Alder reaction , Python (Computer program language) , Reaction force theory , Fullerenes , Diolefins , AMADAR (Automated workflow for Mechanistic Analysis of Diels-Alder Reactions , ONIOM
- Language: English
- Type: Master's thesis , text
- Identifier: http://hdl.handle.net/10962/232317 , vital:49981
- Description: The [4+2] cycloaddition reaction between conjugated dienes and substituted alkenes is known as the Diels-Alder (DA) reaction, in honor of two German chemists, Otto Diels and Kurt Alder, who first reported this marvelous chemical transformation. The DA reaction is one of the most popular reactions in organic chemistry, allowing for the regio- and stereospecific establishment of six-membered rings with up to four stereogenic centers. This pericyclic reaction has found many applications in areas as diverse as natural products chemistry, polymer chemistry, and agrochemistry. Over the past decades, the mechanism of the Diels-Alder (DA) reaction has been the subject of numerous studies, dealing with questions as diverse as the mechanistic pathway, the synchronicity, the use of catalysts, the effect of solvents and salts, etc. On the other hand, as an example, fullerenes (and particularly [60] fullerene) have been found to act as good dienophiles in DA reactions to the extent that many functionalized fullerenes with interesting applications are still synthesized by reacting C60 with dienes. However, despite the very abundant literature about the mechanism of the DA reaction, some pertinent questions have been still pending, including, without being restricted to, the prediction of transition state (TS) geometries and the modeling of DA reactions involving large systems, such as those of C60 fullerene. It must be emphasized that TSs are not easy to predict and the main reason is that many existing algorithms require that the search is initiated from a good starting point (guess TS), which must be very similar to the actual TS. This problem is even more difficult when many TSs are to be located as may be the case in large-scale studies. Moreover, due to the large size of the C60 molecule, the usage of accurate high-level computational methods in the investigation of its reactivity towards dienes is computationally costly, implying the need to find the best threshold between accuracy and computational cost. Therefore, the present study was carried out to contribute to solving the problems of large-scale prediction of DA transition state geometries and the multi-scale modeling of C60 fullerene DA reactions. To address the first problem (large-scale prediction of TSs), we have developed a python program named “AMADAR”, which predicts an unlimited number of DA transition states, using only the SMILES strings of the cycloadducts. AMADAR is customizable and allows for the description of intramolecular DA reactions as well as systems resulting in competing paths. In addition, The AMADAR tool contains two separate modules that perform reaction force analyses and atomic decomposition of energy derivatives from the predicted Intrinsic Reaction Coordinates (IRC) paths. The performance of AMADAR was assessed using 2000 DA cycloadducts and showed a success rate of ~ 95%. Most of the errors were due to basis set inconsistencies or convergence issues that we are still working on. Furthermore, a set of 150 IRC paths generated by the AMADAR program were analyzed to get insight into the (a)synchronicity of DA reactions. This investigation confirmed that the reaction force constant 𝜅 (second derivatives of the system energy with respect to the reaction coordinate) was a good indicator of synchronicity in DA reactions. A close inspection of the profile of 𝜅 has enabled us to propose an alternative classification of DA reactions based on their synchronicity degree, in terms of (quasi)-synchronous, moderate asynchronous, asynchronous, and likely two-steps DA reactions. Natural population analyses seemed to indicate that the global maximum of the reaction force constant could be identified with the formation of all the bonds in the reaction site. Finally, the atomic resolution of energy derivatives suggested that the mechanism of the DA reaction involves two inner elementary processes associated with the formation of each C-C bond. A striking mechanistic difference between synchronous and asynchronous DA reactions emerging from this study is that, in asynchronous reactions, the driving and retarding forces are mainly caused by the fast and slow-forming bonds (elementary process) respectively, while in the case of synchronous ones both elementary processes retard and drive the process concomitantly and equivalently. Regarding the DA reaction of C60 fullerene that was considered to illustrate the problem of multiscale modeling, we have constructed 12 ONIOM2 and 10 ONIOM3 models combining five semi-empirical methods (AM1, PM3, PM3MM, PDDG, PM6) and the LDA(SVWN) functional in conjunction with the B3LYP/6-31G(d) level. Then, their accuracy and efficiency were assessed in comparison with the pure B3LYP/6-31G(d) level considering first the DA reaction between C60 and cyclopentadiene whose experimental data are available. Further, different DFT functionals were employed in place of the B3LYP functional to describe the higher-layer of the best ONIOM partition, and the results obtained were compared to experimental data. At this step, the ONIOM2(M06-2X/6-31 G(d): SVWN/STO-3G) model, where the higher layer encompasses the diene and pyracyclene portion of C60, was found to provide the best tradeoff between accuracy and cost, with respect to experimental data. This model showed errors lower than 2.6 and 2.0 kcal/mol for the estimation of the activation and reaction enthalpies respectively. We have also demonstrated, by comparing several ONIOM2(DFT/6-31G(d): SVWN/STO-3G) models, the importance of dispersion corrections in the accurate estimation of reaction and activation energies. Finally, we have considered a set of 21 dienes, including anthracene, 1,3-butadiene, 1,3-cyclopentadiene, furan, thiophene, selenothiophene, pyrrole and their mono-cyano and hydroxyl derivatives to get insight into the DA reaction of C60 using the best ONIOM2(M06-2X/6-31 G(d): SVWN/STO-3G) model. For a given diene and its derivatives, the analysis of frontier molecular orbitals provides a consistent explanation for the substituent effect on the activation barrier. It revealed that electron-donating (withdrawing) groups such as -OH (–CN) cut down on the activation barrier of the reaction by lowering (extending) of the HOMOdiene – LUMOC60 gap and consequently enhancing (weakening) the interaction between the two reactants. Further, the decomposition of the activation energy into the strain and interaction components suggested that, for a given diene, electron-donating groups (here –OH) diminish the height of the activation barrier not only by favoring the attractive interaction between the diene and C60, but also by reducing the strain energy of the system; the opposite effect is observed for electron-withdrawing groups (here –CN). In contrast with some previous findings on typical DA reactions, we could not infer any general rule applicable to the entire dataset for the prediction of activation energies because the latter does not correlate well with either of the TS polarity, electrophilicity of the diene, or the reaction energy. , Thesis (MSc) -- Faculty of Science, Chemistry, 2022
- Full Text:
- Date Issued: 2022-04-06
Development of a computational chemistry scheme for testing the utility of synthetic bacteriochlorin in dye-sensitized solar cells
- Authors: Kota, Ntsika
- Date: 2018
- Subjects: Dye-sensitized solar cells , Computational chemistry , Density functionals , Electronic excitation , Molecular orbitals , Oscillator strengths , Bacteriochlorin
- Language: English
- Type: text , Thesis , Masters , MSc
- Identifier: http://hdl.handle.net/10962/62327 , vital:28155
- Description: A computational chemistry scheme, based on density functional theory, was developed for in silico testing of a few bacteriochlorin properties relevant to dye-sensitized solar cells. These properties included electronic excitation wavelengths, molecular orbital energy levels, and oscillator strengths among others. Comparisons were made among four species, using computational proxies for electron injection quantum yield and photo-induced current production. The proxy measures for current production (frontier orbital energy level and short circuit current) made consistent, though qualitative, predictions about the ranking of the four dyes. The proxy measures for electron injection quantum yield (change in planar dipole moment and density of states) made less categorical predictions about the ranking. Overall, the scheme singled out one dye as the worst, but made no conclusive predictions about the relative ranking of the other three. There was insufficient data for comparison of the ranking predictions with experiment.
- Full Text:
- Date Issued: 2018
- Authors: Kota, Ntsika
- Date: 2018
- Subjects: Dye-sensitized solar cells , Computational chemistry , Density functionals , Electronic excitation , Molecular orbitals , Oscillator strengths , Bacteriochlorin
- Language: English
- Type: text , Thesis , Masters , MSc
- Identifier: http://hdl.handle.net/10962/62327 , vital:28155
- Description: A computational chemistry scheme, based on density functional theory, was developed for in silico testing of a few bacteriochlorin properties relevant to dye-sensitized solar cells. These properties included electronic excitation wavelengths, molecular orbital energy levels, and oscillator strengths among others. Comparisons were made among four species, using computational proxies for electron injection quantum yield and photo-induced current production. The proxy measures for current production (frontier orbital energy level and short circuit current) made consistent, though qualitative, predictions about the ranking of the four dyes. The proxy measures for electron injection quantum yield (change in planar dipole moment and density of states) made less categorical predictions about the ranking. Overall, the scheme singled out one dye as the worst, but made no conclusive predictions about the relative ranking of the other three. There was insufficient data for comparison of the ranking predictions with experiment.
- Full Text:
- Date Issued: 2018
- «
- ‹
- 1
- ›
- »