Ke
Zhang‡
a,
Xi
Chen‡
b,
Zhoujie
Zhang
c,
Caijie
Bu
d,
Yong
Wu
*c and
Jiawei
Xu
*d
aCollege of Paediatric, Chongqing Medical University, Chongqing 401331, Chongqing, China
bDepartment of Chemistry and Materials Science, College of Science, Nanjing Forestry University, Nanjing 210037, Jiangsu, China
cJiangsu Key Laboratory of Numerical Simulation of Large-Scale Complex System (NSLSCS), School of Chemistry and Materials Science, Nanjing Normal University, Nanjing 210023, Jiangsu, China. E-mail: wuyong@njnu.edu.cn
dState Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002, Fujian, China. E-mail: xujiawei@fjirsm.ac.cn
First published on 14th October 2023
The present study investigates the standard model of [1,2]-fluorine migration and that triggered by the rearrangement of cyclopropyl-substituted fluoroepoxides. The [1,2]-fluorine migration reaction proceeds via a synchronous concerted, tight-ion-pair mechanism. When coupled with other reaction coordinates, the whole reaction follows an asynchronous mechanism, while the [1,2]-fluorince migration unit still retains its tight-ion-pair feature and the reaction coordinates of two C–F distances vary synchronously. A general reaction program for α-electron-deficiency-induced [1,2]-fluorine migration is proposed through an analysis of the intermediates generated from nucleophilic addition. The reaction mechanisms associated with α-electron deficiency and the rearrangement are scrutinized using computational chemistry. Two additional reaction programs for [1,2]-fluorine migration are identified. The Gibbs free energy change of [1,2]-fluorine migration exhibits a linear dependence on the value of the Fukui function of the substrate, which could lead to the production of the desired α-monofluoroketone and enhance the utilization of fluorine atoms.
From the perspective of green chemistry, the combination of both C–F bond formation and cleavage represents an approach that maximizes atom utilization, in contrast to conventional methods, which often result in the wastage of fluorine atoms through the formation of fluorine anions with cleavage of the C–F bond. Furthermore, some earlier methods require harsh conditions or specific substrates. Only a very few methods have been reported that apply usage of fluorine anions in C–F bond formation.11 Furthermore, the combination, which behaves as intramolecular fluorine migration, has rarely been reported either. The first true intramolecular [1,2]- and [1,5]-fluorine migration coupled with a ring-opening reaction of epoxides was reported by Hu and co-workers.12–14 Selectivity between the two types of migration is realized by controlling the acidity of the reaction system and solvent polarity. As shown in Scheme 1a, fluorine migration is enhanced by an electron-deficient carbon on the three-membered ring of the epoxides. Based on experimental evidence, Hu and colleagues noted that [1,2]-fluorine migration occurs via two mechanisms, which are intramolecularly concerted and tight-ion-pair mechanisms. On the other hand, [1,5]-fluorine migration follows a stepwise mechanism induced by another three-membered ring. However, it is important to note that only the reported [1,2]-fluorine migration effectively utilizes the fluorine atom in the substrate. This is because the fluorine in the product of [1,5]-fluorine migration can also originate from HF in the reaction system (as a result of fluorine leaving). Thus, it cannot be simply regarded as originating from the same substrate molecule, as emphasized by Hu et al. However, the [1,5]-migration is still of great significance, as it provides valuable evidence that during the formation of the tetrahedral intermediate, a considerable number of fluorine atoms in the reaction system can remain attached to the reaction centre and undergo migration pathways, rather than leaving to form the deprotonated product of α,β,γ,δ-unsaturated ketone. This “pause” at the (quasi-)tetrahedral intermediate can be used wisely by providing possible pathways for migration instead of direct fluorine leaving, which would result in a wastage of fluorine. The reaction conditions can be optimized based on Hu's work.
The practicality and convenience of [1,2]-fluorine migration may be the dawn of a new era in organic fluorine chemistry. The synthetic significance of [1,2]-fluorine migration, developed by Hu and co-workers, is the preparation α-monofluoroketone, but for epoxide compounds the procedure requires practical substrates. This procedure has already been applied in some recent work.15,16 Generally, α-monofluoroketone is an important building block with various downstream applications, such as β-fluoroamine and β-fluoroalcohol. Compared to other popular methods for the preparation of α-monofluoroketone, the reported [1,2]-fluorine migration gives a good yield and selectivity, without requiring specific reagents or complicated reaction conditions, which is no doubt a better choice for applications, especially in industrial production.
We gained inspiration from Hu's work. Theoretically, our approach involves designing an electron-deficient α-carbon to serve as a fluorine acceptor. This is the same case as in epoxide-induced [1,2]-fluorine migration, in which electron-deficient α-carbon is stabilized by an oxygen atom via neighboring group participation. In Hu's work, the ring-opening reaction is used to provide vacant valence for C–F bond formation. However, we choose to use a much more normal structure of α,β-unsaturated carbonyl compounds, which enables C–F activation through the formation of a tetrahedral intermediate. Additionally, we also noticed that in Hu's first work on [1,2]-fluorine migration, one of the reviewers also proposed another stepwise mechanism (Scheme 1b), which can be found in the reference section of Hu's work. Although we will later prove that Hu's reaction takes a concerted mechanism, the stepwise mechanism remains valuable as a typical example of α-electron-deficiency-induced [1,2]-fluorine migration, reflecting the basis of the reaction. This present study will involve a theoretical scrutiny of [1,2]-fluorine migration induced by α-electron deficiency. Subsequently, an in-depth analysis of Hu's reaction will be conducted.
Based on previous research and discussions, we present a generalized framework of α-electron-deficiency-induced [1,2]-fluorine migration (Scheme 1c), where the acceptor of the electron pair is denoted “A”, and the upstream substrates can be α,β-unsaturated acyl fluoride and additional nucleophilic reagent (Nu), or F− can simply be used in polar aprotic solvents. The product of fluorine migration can undergo further β-protonation to yield stable α-monofluoroketone or to conduct other functionalization procedures.
Based on quantum chemical calculations, we will analyze the nature of [1,2]-fluorine migration in the previously reported system and this newly predicted system, showing that the reaction system we designed is expected to show significant potential and broad applicability in various situations. We will first introduce the calculation method adopted in this work and explore the essence of [1,2]-fluorine migration from ‘α-electron-deficiency-induced’ and ‘epoxide-induced’ mechanisms. Later, a comprehensive investigation of a series of previously reported α,β-unsaturated acyl fluoride substrates will be undertaken from a thermodynamic standpoint, aiming to offer quantitative insights. Also, we designed and study several examples with potential applications based on our previous results. Additionally, for [1,2]-fluorine migration, two reaction programs are proposed, which are the β-carbonium anion mechanism and the α-SN2 coupled mechanism. In the last section, we summarize the reactions and mechanisms included in this study.
Solvation Gibbs free energy change (ΔGsolv) is calculated by dopting a solvation model based on density (SMD).27 Minnesota hybrid functional M05-2X and Pople's split valence basis sets of 6-31G(d) are used for calculation of the electronic energy of both gas (Egas) and solution (Esol) phases.28,29 The solvation Gibbs free energy change is therefore calculated as in eqn (1), in which Vm,g and Vm,solv represent the occupied molar volume under standard pressure and concentration at 298.15 K, respectively.
(1) |
The double-hybrid functional of rev-DSDPBEP86-D3(BJ) and basis sets of def2-TZVPP developed by Ahlrichs and co-workers are used for high-quality single point energy in gas phase for all structures.30,31 Therefore, Gibbs free energies are calculated as the sum of thermal correction for Gibbs free energy, solvation Gibbs free energy change and single point energy at a higher level. All detailed thermochemical data is listed in ESI section 2.†
Apart from the above-mentioned calculations, wavefunction analysis is carried out with the Multiwfn (Development version: 3.8) program developed by Tian Lu and co-worker.32 The geometries and visualization of various concepts are displayed by Visual Molecular Dynamics (VMD, version 1.9.3) and CYLview software.33
We then move on to determining which type of mechanism the reaction follows. Possible [1,2]-fluorine migration mechanisms proposed by Hu et al. (Scheme 1a and b), namely “concerted” and “tight-ion-pair”, are not actually opposed to each other. The “concerted” mechanism proposed by the referee of Hu's work is likely meant to describe the same covalent feature in two C–F interactions at a transition state. Therefore, our discussion here focuses on whether the transition state has the “covalent” or “tight-ion-pair” feature. From the two-dimensional electron density difference diagram presented in Fig. 1a, we can see that the area between carbon atom (C1 or C2) and fluorine atom is coloured blue, representing a decrease in electron density and indicating that covalent C–F interactions have not been established, and the fluorine atom is bound to reaction sites by electrostatic interaction. This is because covalent interaction will result in an increase in electron density compared with the sum of electron densities of the free atoms, due to the share of bonding electrons, where the red area between C1 and C2 is a good example. In the meantime, the area surrounding the fluorine atom shows a strong increase in electron density, in contrast with the decrease in electron density in areas surrounding the carbon atoms. This distribution is a typical feature of electrostatic interaction between the fluorine atom and both carbon atoms. We can also see this from the results of bond order and NPA charge. Fig. 1b shows that the fluorine atom interacts with both carbon atoms, but much more weakly than in common bonding interaction. Moreover, Fig. 1c shows that the fluorine atom experiences a decrease in atomic charge during the reaction process, being more negative than in both the tetrahedral intermediate and the product, behaving differently from covalent bonds. Besides, it should also be noted that the NPA charge of the fluorine atom will not reach the ideal −1.0 e of free F− due to electrostatic polarization, according to the electroneutrality principle.34,35 These pieces of evidence all lead to the conclusion that the [1,2]-fluorine migration reaction has a tight-ion-pair transition state.
Secondly, does another accessible transition state or stepwise pathway exist? This is not an easy question for computational chemistry since common optimization can only tell us that a steady or saddle point does exist; however, it can hardly give solid proof for the opposite. To give an answer to this question, we carried out a smooth scan for C1–F and C2–F bond lengths and plotted the potential energy surface, as displayed in Fig. 2. The x- and y-axes correspond to the difference between scan coordinates and C1–F and C2–F bond lengths in the tetrahedral intermediate, respectively. The lighter blue region ranging over the central bottom area corresponds to the tetrahedral intermediate, while the deeper blue region at the upper left corresponds to the migration product. The intrinsic reaction coordinate is projected onto the two-dimensional potential energy surface and a stream of the energy valley is displayed on the reaction pathway. Since the scan coordinate of C1–F bond length starts at the stable bonding distance in the tetrahedral intermediate, the area of shorter C1–F bond length is less meaningful for inclusion. Therefore, the potential energy surface area in Fig. 2 has already covered the major reaction region, yet there is no pathway other than what has been discussed before. This result reveals that the transition state in Fig. 1a does correspond to the true [1,2]-fluorine migration transition state and this system follows the tight-ion-pair mechanism.
Fig. 2 Potential energy surface obtained from a smooth scan for C1–F and C2–F bonds labelled in Fig. 1a. (Scan step size: 0.05 Å; 861 points in total; Level: M06-2X-D3/def2-SVP with IEF-PCM form of implicit solvent model.). The intrinsic reaction coordinate curve is projected onto the two-dimensional plane. |
Furthermore, the potential energy surface also tells us another story. As is well known in conventional organic chemistry, nucleophilic substitution on the carboxyl group usually follows a stepwise mechanism which includes addition and elimination steps. However, here the example shows us that the fluorine atom is not eliminated as a fluorine anion; instead, it makes use of the pause at the tetrahedral intermediate and migrates to the electron-deficient α-carbon to form α-monofluoroketone. This can be discovered from the potential energy surface since the right-hand part is an energy mountain (coloured in red), which corresponds to an increase in both C1–F and C2–F bond lengths and behaves as a fluorine-leaving process. Being high in energy, the fluorine-leaving process is difficult.
A reported example of Hu's reaction is presented in Fig. 3a. According to what has been introduced, the reaction is supposed to take three possible pathways: covalent concerted, tight-ion-pair concerted and stepwise pathways. Computational results support a mechanism with only one transition state. C–F distances are close to the transition state in Fig. 1a and correspond to the previously discussed tight-ion-pair concerted [1,2]-fluorine migration reaction. Mayer bond order gives the same idea that near the transition state structure of Hu's reaction, both C–F bond orders are close to zero. From Fig. 3b we can also see that the processes of C1–F bond cleavage and C2–F bond formation do not overlap, which shows that on reaching the transition state, the C1–F bond has almost been broken while the C2–F bond has not yet formed, i.e. the first half of fluorine migration has been completed to form an electrostatically bound fluoride anion, providing solid evidence for a tight-ion-pair transition state. This is even more thorough than what is observed in Fig. 1. The fluoride anion is closer to C1 with a distance of 1.663 Å, compared with 2.086 Å away from C2 at the transition state structure. The bond order of C–O bonds also gives a similar idea. The bond order of C1–O has reached 1.8 at the transition state; therefore, the ring-opening process of epoxides has almost been completed at this point.
The above-presented structural parameters strongly indicate an asynchronously concerted mechanism. In parallel, Fig. 3c shows the NPA charge variation along the intrinsic reaction coordinate. The atomic charge of fluorine reaches only −0.5 at its most negative value. The difference between this value and that of a free fluorine anion reveals that the fluorine atom is tightly bound along the reaction pathway, even though both C–F bonds are not formed near the transition state. At the same time, the atomic charge of oxygen has a notable minimum near the transition state, which is entirely different from and more negative than that of an oxygen atom in the ketone product. This shows that in the transition state structure, the C1–O bond is strongly polarized by the bound fluorine anion; although the C1–O π-bond has nearly formed, the C1 atom is not prepared to readjust the charge distribution with the O atom due to the existence of the bound fluoride anion.
Fig. 4 shows the potential energy surface of Hu's reaction obtained from a smooth scan. Since three bonds are significant in this process, we select the bond length of C1–O and the bond length difference between C2–F and C1–F as variants. This is due to the two selected C–F bonds showing different trends in variation: that is, the bond length of C1–F monotonically increases while the bond length of C2–F monotonically decreases in the process. Therefore, we can use the bond length difference to reflect both variants. This method is useful and widely applied. From the potential energy surface, we can see that before reaching the transition state, an increase in the C1–O bond length is the major variation in the structure, while the bond length difference of the C–F bonds does not vary a lot. In this part of the intrinsic reaction coordinate, the fluoride anion forms but does not migrate, while the epoxide structure has already been cleaved. Considering the previously discussed standard model in which the [1,2]-fluorine migration reaction unit follows a synchronous concerted, tight-ion-pair pathway on its reaction coordinates C1–F and C2–F, this is exactly consistent with a small change in the x-axis from the reactant to the transition state, demonstrating that [1,2]-fluorine migration still retains its reaction feature when coupled with other reaction units. Moreover, epoxide cleavage has already been completed in the transition state, which clarifies its role in induction. This is why we name this reaction example epoxide-induced [1,2]-fluorine migration.
Fig. 4 Potential energy surface obtained from a smooth scan for C1–O bond length and bond length difference between C2–F and C1–F labelled in Fig. 3a. (Scan step size: 0.10 Å; 273 points in total; Level: M06-2X-D3/def2-SVP with IEF-PCM form of implicit solvent model). The intrinsic reaction coordinate curve is projected onto the two-dimensional plane. |
The second half of the intrinsic reaction coordinate is the opposite, where the fluorine atom completes the migration induced by the positive C2 site generated by ring-opening of the epoxide structure. This reaction mechanism can be viewed as a combination of the concerted and stepwise mechanisms introduced in Scheme 1. With respect to energy, it behaves as a concerted mechanism since only one transition state is involved. Nevertheless, with respect to structural varieties, it behaves like a stepwise mechanism since the major structural varieties are asynchronized, as marked in Fig. 4 with a bent solid arrow. To partially conclude, by comparing the process of two coupled reactions together, we can see that the whole reaction follows an asynchronously concerted mechanism. The asynchronous feature is owing to a difference in progress between two coupled reaction units, which are [1,2]-fluorine migration and epoxide cleavage. Epoxide cleavage, which acts as the induction unit, has nearly completed its reaction coordinate (C1–O and C2–O) at the tight-ion-pair transition state. The [1,2]-fluorine migration unit still retains the concerted feature on its reaction coordinates (C1–F and C2–F) and the rebound process of C2–F happens after the transition state.
We also would like to ask whether there is another synchronized pathway for this reaction. The answer is obvious with the potential energy surface. As marked with a straight dotted arrow in Fig. 4, if a synchronized pathway exists, it will experience an extremely high energy increase in the region coloured white to red on the surface. This is because the C2 site does not have enough free valence to act as a fluorine acceptor in a synchronized pathway, where ring-opening and fluorine migration take place at the same time, leading to great displacement in the sp3-type hybrid orbitals of the C2 site. Moreover, as we have pointed out that fluorine migration is induced by electron deficiency of the C2 site, if the epoxide structure is not broken, there is not enough driving force for the migration reaction.
According to the above discussion of Hu's reaction, we may reach the following conclusions. Firstly, the epoxide-induced [1,2]-fluorine migration follows an asynchronously concerted pathway. The reaction process can be separated into two periods: (1) electron-deficient α-site generation and (2) α-electron-deficiency-induced fluorine migration. This is very consistent with our concept of designation. Secondly, when coupled with other reaction units, [1,2]-fluorine migration may retain its reaction feature in the isolated model discussed in the previous section. This is a suitable starting point for developing more complicated and synthetically meaningful reactions. Thirdly, since Hu's reaction reached an acceptable yield (12 h, 70%) and good selectivity (98:2, compared with the [1,5]-migration product), we believe that as long as the suitable reaction conditions are controlled, disadvantages brought about by fluorine-leaving process can be minimized. Therefore, we carried out more predictions based on the above investigation of a designed reaction and an experimentally reported reaction.
Scheme 2 Structures of α,β-unsaturated acyl fluorides as substrates, 3–8 are reported substrates and 9–20 are designed substrates. |
In Table 1, six reported α,β-unsaturated acyl fluorides (3–8) are used as substrates to combine with four representative nucleophiles, which are Grignard reagents of the benzene anion (Nu = C6H5−, 3a–8a), small inorganic anions of the cyanide anion (Nu = CN−, 3b–8b), carbonyl compounds with an active α-H of the 3,5-heptadione anion (Nu = −CH(COC2H5)2, 3c–8c) and the diethyl malonate anion (Nu = −CH(CO2C2H5)2, 3d–8d).
Tetrahedral IM structure | ΔG‡a | ΔE‡b | ΔGc | ΔEd | ΔqD/ee | ΔqA/ef | |
---|---|---|---|---|---|---|---|
a Gibbs free energy barrier. b Electronic energy barrier in vacuum (M06-2X-D3/def2-SVP level). c Gibbs free energy change. d Electronic energy change in vacuum (in kcal mol−1). e NPA charge variation of donor group (oxygen anion). f NPA charge variation of acceptor group (atomic charge of boron and oxygen in series 3–5 and sum of atomic charges for cyclopentadienyl and nitro group in series 6–8). | |||||||
3a, R = C6H5 | −2.1 | 9.3 | −6.5 | −6.6 | −0.907 → −0.589 | 0.441 → 0.013 | |
3b, R = CN | 6.6 | 15.5 | 0.7 | 0.7 | −0.853 → −0.546 | 0.463 → 0.087 | |
3c, R = CH(COC2H5)2 | 1.0 | 10.3 | −4.7 | −1.5 | −0.891 → −0.583 | 0.451 → 0.038 | |
3d, R = CH(CO2C2H5)2 | −0.7 | 2.4 | −8.1 | −3.1 | −0.877 → −0.584 | 0.448 → 0.023 | |
4a, R = C6H5 | −1.0 | 8.5 | −1.4 | −3.0 | −0.908 → −0.587 | −0.601 → −0.761 | |
4b, R = CN | 8.3 | 16.6 | 6.5 | 4.8 | −0.853 → −0.538 | −0.588 → −0.728 | |
4c, R = CH(COC2H5)2 | 2.0 | 11.0 | 0.6 | 1.9 | −0.891 → −0.577 | −0.595 → −0.752 | |
4d, R = CH(CO2C2H5)2 | 0.9 | 4.3 | −1.8 | 1.1 | −0.879 → −0.580 | −0.598 → −0.757 | |
5a, R = C6H5 | 2.9 | 10.8 | 7.6 | 5.0 | −0.911 → −0.591 | −0.659 → −0.792 | |
5b, R = CN | 13.4 | 20.5 | 13.9 | 11.9 | −0.856 → −0.560 | −0.649 → −0.758 | |
5c, R = CH(COC2H5)2 | 7.5 | 14.6 | 9.9 | 9.7 | −0.893 → −0.582 | −0.654 → −0.786 | |
5d, R = CH(CO2C2H5)2 | 6.2 | 5.9 | 10.1 | 8.4 | −0.877 → −0.570 | −0.655 → −0.777 | |
6a, R = C6H5 | −0.2 | 11.0 | −10.6 | −12.7 | −0.903 → −0.608 | −0.264 → −0.874 | |
6b, R = CN | 7.3 | 16.9 | −1.9 | −3.3 | −0.850 → −0.550 | −0.190 → −0.804 | |
6c, R = CH(COC2H5)2 | 1.5 | 15.2 | −12.6 | −13.4 | −0.896 → −0.606 | −0.215 → −0.843 | |
6d, R = CH(CO2C2H5)2 | −1.7 | 6.4 | −12.2 | −4.6 | −0.882 → −0.582 | −1.159 → −1.686 | |
7a, R = C6H5 | 4.7 | 12.0 | 6.0 | 1.3 | −0.915 → −0.581 | 0.128 → −0.020 | |
7b, R = CN | 15.8 | 22.8 | 17.7 | 14.2 | −0.861 → −0.497 | 0.133 → 0.001 | |
7c, R = CH(COC2H5)2 | 8.1 | 18.0 | 2.7 | 1.3 | −0.906 → −0.242 | 0.131 → 0.008 | |
7d, R = CH(CO2C2H5)2 | 5.3 | 3.0 | 5.5 | −0.1 | −0.898 → −0.568 | 0.130 → −0.013 | |
8a, R = C6H5 | 1.3 | 9.9 | −5.7 | −11.2 | −0.913 → −0.576 | 0.080 → −0.120 | |
8b, R = CN | 12.0 | 19.4 | 6.7 | 2.2 | −0.859 → −0.472 | 0.087 → −0.104 | |
8c, R = CH(COC2H5)2 | 5.2 | 13.2 | −2.6 | −4.0 | −0.895 → −0.564 | 0.083 → −0.108 | |
8d, R = CH(CO2C2H5)2 | 3.7 | 1.5 | −1.6 | −3.4 | −0.876 → −0.546 | 0.083 → −0.112 |
The substrate of series 3 was reported by Krygowski et al.37 When a benzene anion reacts with 3, [1,2]-fluorine migration proceeds via tetrahedral intermediate 3a, with the lowest Gibbs free energy barrier among series a of −2.1 kcal mol−1 and an electronic energy barrier of 9.3 kcal mol−1.
The preparation of 4 has been reported in many studies.38–40 Interestingly, the cause of the symmetry of 4, whichever carbonyl carbon combines with the nucleophile, results in the same tetrahedral intermediate. Intermediate 4a is generated by the combination of 4 and a benzene anion. When [1,2]-fluorine migration occurs, the corresponding Gibbs free energy barrier is −1.0 kcal mol−1. The reaction also has the lowest electronic energy barrier of 8.5 kcal mol−1 among series a.
Substrate 5 was reported by Corey et al., which is similar to 4 except the distal acyl fluoride group turns into an ester group.41 Comparing the Gibbs free energy barriers and electronic energy barriers of 4 and 5, all series (a–d) present the result that 4a–4d have lower barriers than the corresponding 5a–5d. The reason for this result is that the electron deficiency of 4 induced by the distal acyl fluoride group is more than that of 5 induced by the ester group and electron deficiency makes [1,2]-fluorine migration easier.
Substrate 6 is a kind of α,β-unsaturated acyl fluoride cyclopentadiene reported by Oziminski.42 The character of this substrate is that when the substrate combines with a nucleophile and undergoes [1,2]-fluorine migration via the tetrahedral intermediate, the distal cyclopentadiene group accepts a pair of electrons brought by the migrated fluorine atom, forming the stable cyclopentadienyl anion. Based on these results, this process also has advantages in thermochemistry. When the nucleophile is a benzene anion, this reaction via intermediate 6a has a Gibbs free energy barrier and electronic energy of −0.2 kcal mol−1 and 11.0 kcal mol−1, respectively.
The synthesis of 7 was reported by Liu et al.43 A similar substrate 8 was reported by Labidi et al.44 From the thermochemical data of 7 and 8, it is obvious that the Gibbs free energy barriers and the electronic energy barriers of series 8 (8a–8d) are all lower than the corresponding ones of series 7. Comparing the structures of 7 and 8, the reason is considered to be that the phenyl group of 7 reduces the electron deficiency caused by the distal nitro group, making [1,2]-fluorine migration more unfavourable. This finding suggests that the presence of a conjugated system in the substrate can have an impact on the reaction by modifying the electron deficiency caused by the electron acceptor.
In addition to the previously reported α,β-unsaturated acyl fluorides, we have also designed a series of α,β-unsaturated acyl fluorides (9–20) with similar electron-deficient α-sites caused by distal electron-acceptor groups through conjugated systems. According to the calculation results of series 3–8, it has been revealed that using a benzene anion as the nucleophile results in a comparatively lower Gibbs free energy barrier and electronic barrier of the reaction compared to the utilization of other nucleophiles. Therefore, the benzene anion is chosen to be the nucleophile used to predict the behaviour of designed substrates in the [1,2]-fluorine migration reactions. Table 2 presents the intermediate structures (9a–20a), NPA charge variations, Fukui functions, and Gibbs free energy changes of the [1,2]-fluorine migration reactions.
The distal electron-acceptor group of 9 can be considered as a carbonyl group protected by ethane-1,2-dithiol. 9a is the necessary intermediate in the reaction of the benzene anion with 9. The Gibbs free energy barrier of this reaction is 1.3 kcal mol−1 and the Gibbs free energy change is −23.5 kcal mol−1. The above data indicates that this process has a thermodynamic advantage. The reaction product of 9 also can remove the protection of thioacetal to produce a new acyl fluoride.
When substrate 10 undergoes nucleophilic attack, it experiences [1,2]-fluorine migration, wherein the -OTs group departs as the conjugated system transfer. This process is similar to a nucleophilic substitution reaction. Motivated by this reaction process, we have designed a series of substrates featuring α-substituted saturated acyl fluorides, in which its substituents are good leaving groups. We will provide detailed research into the [1,2]-fluorine migration process that occurs in these substrates when subjected to nucleophilic attack, in the following section, “α-SN2 Coupled Mechanism: Saturated Acyl Fluoride as Substrate”.
The electron-pair acceptor of substrates 13 and 15 is a nitrile group. However, the nitrile group of 13 connects with the α,β-unsaturated acyl fluoride directly, but the nitrile group of 15 is placed on the para position of a benzene ring connected with α,β-unsaturated acyl fluoride. By comparing the Gibbs free energy barriers between 13 and 15, the result indicates that the introduction of a phenyl group into the substrate increases the Gibbs free energy of [1,2]-fluorine migration. The same conclusion also can be reached by comparing 7a and 14.
The electron-pair acceptor of 16 is a trimethylsilyl group. Influenced by the phenyl group, it has no significant electronic effect on the substrate, which lead to a high Gibbs free energy barrier of 12.1 kcal mol−1. Substrates 17 and 18 are typical boranes and they have close Gibbs free energy barriers of 9.9 kcal mol−1. Moreover, the Gibbs free energy barrier and Gibbs free energy change of substrate 19 are also similar to those 17 and 18. Substrate 20 has an azo group. Influenced by the electronic effect of the functional group in the substrate, this substrate behaves well in [1,2]-fluorine migration, having a low Gibbs free energy barrier of 5.0 kcal mol−1.
All reaction examples show obvious charge redistribution from the oxygen anion to a distal electron-deficient group through a conjugated system. To figure out the essence of the reactions, a condensed Fukui function (CFF, fk+) is calculated for all atoms based on the 24 optimized structures. The Fukui function aims to reflect the variation in electron density when the system acquires or loses electron, as defined in eqn (2), where ρ(r) stands for the electron density on the specific point in three-dimensional real space and N stands for the electron number of the system. Since N is an integer and not a continuous variable, finite differential approximation is applied to give three commonly used forms (eqn (3)) instead of the partial derivative. Notice that N − 1, N and N + 1 are only notations for number of electrons and do not stand for potential. These definitions give real functions in three-dimensional space. Generally, the f+ form of the Fukui function shows the region in which the additional one electron will be distributed, giving a prediction of the reaction site for nucleophilic attack. It is also similar for f− and f0, giving predictions for electrophilic attack and radical attack, respectively. Based on the atomic population method, a condensed Fukui function (fk+, fk− and fk0; eqn (4)) can be generated from the Fukui function to specify a value for each atom. Since the reaction site of nucleophilic attack usually shows an electron-deficient character, fk+ can also be applied to compare the degree of electron deficiency between different molecules from a quantitative point of view.
(2) |
f+(r) = ρN+1(r) − ρN(r) | (3.1) |
f−(r) = ρN(r) − ρN−1(r) | (3.2) |
f0(r) = ½[f+(r) + f−(r)] = ½[ρN+1(r) − ρN−1(r)] | (3.3) |
fk+ = qkN+1 − qkN | (4.1) |
fk− = qkN − qkN−1 | (4.2) |
fk0 = ½(fk+ + fk−) = ½(qkN+1 − qkN−1) | (4.3) |
Fig. 5 plots the relationship between Gibbs free energy change (ΔG) and sum of the condensed Fukui functions of atoms in distal electron-deficient groups. In this work, the Hirshfeld population method is adopted to calculate population number of the atoms. Based on the results shown in Fig. 5, we can draw the conclusion that the Gibbs free energy change of [1,2]-fluorine migration exhibits linear dependence on the value of the condensed Fukui function of the substrate distal electron-deficient group, i.e., a higher condensed Fukui value corresponds to a more negative Gibbs free energy change. Fig. 5a shows the results of different nucleophile reactions with reported acyl fluorides (3–8) and Fig. 5b shows the results of benzene anion reaction with different acyl fluoride substrates (3–20). It all means that no matter the kind of nucleophile or acyl fluoride substrate, the above linear relationship is followed in all cases.
As shown in Fig. 5a, the relationship between Gibbs free energy change and condensed Fukui function value can be concluded. This shows that when acyl fluoride substrates combine with nucleophiles, for each kind of nucleophile, there is a good linear relationship between the Gibbs free energy change and Fukui function value. The highest coefficient of determination is 0.9591 for series a and the lowest is 0.9057 for series b. The coefficient of determination of the linear relationship shown in Fig. 5b is 0.9622. Moreover, the condensed Fukui function value of the –OTs group in substrate 10 is 0.6553, which is the highest among substrates 3–20, and substrate 10 has the lowest Gibbs free energy change of −26.1 kcal mol−1.
Substrates 11 and 12 exhibit significant differences in their behavior with respect to [1,2]-fluorine migration. Substrate 11 has a comparatively lower condensed Fukui function of 0.0162 and higher Gibbs free energy change of 21.3 kcal mol−1 in comparison to substrate 12. On the other hand, 11 and 12 are isomers, similar in structure, although the condensed Fukui function value of 12 is 0.3277, higher than that of 11. This difference is attributed to the presence of C–O sigma bond cleavage in the reaction of substrate 11, whereas the reaction of substrate 12 involves only charge redistribution.
Based on the above discussion, it can be inferred that when the substrate has a large Fukui function value, the stronger the electron-withdrawing capacity of the substituent, and the lower the Gibbs free energy change of the reaction tends to be. This principle can be utilized to predict the difficulty level of [1,2]-fluorine migration for various substrates and to expand the scope of its applications. Furthermore, we will delve into two kinds of distinctive substrate and their respective reaction mechanisms of [1,2]-fluorine migration based on the reaction in Scheme 1c.
According to Scheme 3, we have designed a series of reactions of a selected substrate (21) and external electrophiles (22a–30a). The Gibbs free energy barriers and Gibbs free energy changes of [1,2]-fluorine migration reactions 22–30 have been tabulated in Table 3. Substrate 21 is 1-fluoro-1-phenylprop-2-en-1-olate, which has been synthesized by the benzene anion undergoing addition to the acryloyl fluoride.
Tetrahedral IM | External electrophile | Producta | ΔG‡b/kcal mol−1 | ΔGc/kcal mol−1 | ||
---|---|---|---|---|---|---|
a Product of [1,2]-fluorine migration reaction. b Gibbs free energy barrier. c Gibbs free energy change. | ||||||
OCO (CO2) | 22a | 22b | 27.2 | −1.2 | ||
23a (24a) | 23b | 28.5 | 11.3 | |||
24b | 27.4 | −0.9 | ||||
25a | 25b | 34.6 | 4.9 | |||
26a | 26b | 39.1 | 10.2 | |||
27a | 27b | 31.3 | 4.4 | |||
28a | 28b | 34.7 | 7.9 | |||
29a | 29b | 31.5 | −20.9 | |||
30a | 30b | 27.5 | −8.4 |
The first external electrophile is carbon dioxide. In this reaction, the generated β-carbonium anion intermediate will attack the carbon atom in the carbon dioxide molecule to form fluorine-containing carboxylic acid. Computational results have shown that the steps of fluorine migration and β-carbonium anion intermediate addition to the external electrophile are concerted, as proved by vibrational analysis of the transition state and intrinsic reaction coordinate calculation based on the transition state structure. The Gibbs free energy barrier of reaction 22 is 27.2 kcal mol−1 and the Gibbs free energy change of reaction 22 is −1.2 kcal mol−1.
The external electrophiles (23a and 24a) are formaldehyde. In the reaction of 21 and 23a, as shown in Scheme 3, not only is 23b a possible product but also 24b. However, 23b is generated by the nucleophilic addition of the β-carbonium anion intermediate and external electrophile, while 24b is generated by cycloaddition. According to Table 4, the Gibbs free energy barrier and Gibbs free energy change of reaction 23 are 28.5 kcal mol−1 and 11.3 kcal mol−1, respectively. The Gibbs free energy barrier and Gibbs free energy change of reaction 24 are 27.4 kcal mol−1 and −0.9 kcal mol−1, respectively. Based on a comparison of the Gibbs free energy barriers and Gibbs free energy changes of these two reactions, we conclude that the cycloaddition process is favoured thermodynamically compared to the nucleophilic addition process when the β-carbonium anion intermediate reacts with an external electrophile.
Therefore, we further investigated the cycloaddition reactions 25–28. Notably, the β-carbonium anion intermediate is 1,3-dipolar, of which the positive centre is the carbonyl carbon atom and the negative centre is a terminal carbon atom. The external electrophiles 25a, 26a, 27a, and 28a are acetaldehyde, acetone, acetophenone, and benzophenone, respectively. The cycloaddition reaction of the β-carbonium anion intermediate and external electrophile is similar to a Diels–Alder reaction and undergoes a concerted mechanism. All of the reactions 25–28 are endothermic and have a high energy barrier.
29a is an excellent electrophilic reagent, after reacting with 21, resulting in the formation of products 29b and 4-dimethylaminopyridine (DMAP, 29c). The Gibbs free energy barrier for this reaction is determined to be 31.5 kcal mol−1, while the lowest Gibbs free energy change is recorded as −20.9 kcal mol−1.
Unlike previous external electrophiles, 30 is an olefin containing a potent electron-withdrawing group, which gives it 1,3-dipolar characteristics, thereby enabling it to undergo a [2 + 3]-cycloaddition reaction with a β-carbonium anion. Computational investigations have led to the initial conclusion that the reaction between 21 and 30a follows a stepwise mechanism. Specifically, [1,2]-fluorine migration occurs first to the substrate, generating a β-carbon anion that subsequently attacks 30a, yielding intermediate 30b. Finally, 30b undergoes cyclization to form the ultimate product. The Gibbs free energy barrier and the Gibbs free energy change for the addition reaction step are found to be 27.5 kcal mol−1 and −8.4 kcal mol−1, respectively. Subsequent comprehensive elucidation of this reaction will be undertaken in our forthcoming research.
In the preceding discussion, we described the reaction of α,β-unsaturated acyl fluoride with an external electrophile, which undergoes the β-carbonium anion mechanism. The reaction scheme for the external electrophile acting as the substrate, where it serves as an electron-pair acceptor, is similar to the scheme shown in Scheme 1c. However, in the absence of an electron-pair acceptor in the reaction system, when a nucleophile attacks the substrate of α,β-unsaturated acyl fluoride, the fluorine atom is released from the molecule instead of undergoing migration to the adjacent carbon atoms. Hence, this reaction scheme can also be classified as electron-deficiency-induced [1,2]-fluorine migration.
Notably, when a carbonyl compound is used as the substrate, a special reaction akin to the Diels–Alder cycloaddition reaction occurs. The α,β-unsaturated acyl fluoride is 1,3-dipolar and the carbonyl compound is 1,2-dipolar, and this reaction undergoes a concerted mechanism, completing the cycloaddition reaction with only one transition state.
We chose three distinct saturated acyl fluorides, which are 2-fluoro-2-oxoethyl 4-methylbenzenesulfonate (31), 2-fluoro-2-oxoethyl methanesulfonate (32) and 2-iodoacetyl fluoride (33), as our substrates. The leaving groups for these substrates are comprised of p-methylbenzenesulfonyl, methanesulfonyl, and iodine atoms. Throughout the course of the reaction, the benzene anion functions as a nucleophile and reacts with the aforementioned substrates. Each of these reactions leads to the production of the same α-monofluoroketone, 2-fluoro-1-phenylethan-1-one (34), albeit with different leaving groups. Of particular note, the Gibbs free energy barriers (5.2 kcal mol−1 for 31 and 5.1 kcal mol−1 for 32) and Gibbs free energy changes (−26.0 kcal mol−1 for 31 and −30.9 kcal mol−1 for 32) for reactions 31 and 32 are quite low. Our analysis of the Fukui function has revealed that the Gibbs free energy change of the reaction is related to the distal electron-deficient groups. In particular, substrates 31 and 32 share the common characteristic of having easily separable leaving groups. In the charge recombination of [1,2]-fluorine migration, a good leaving group improves the capacity of a pair of obtained electrons to be better received. Furthermore, for substrate 33, the Gibbs free energy barrier and Gibbs free energy change are 2.9 kcal mol−1 and −35.4 kcal mol−1, respectively, while the electronic energy barrier is 10.9 kcal mol−1 and the electronic energy change is −31.8 kcal mol−1.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ob01335a |
‡ These authors contributed the same to this work. |
This journal is © The Royal Society of Chemistry 2023 |