Wenrui Zheng*,
Lanlan Ding,
Jiaoyang Wang and
Yingxing Wang
College of Chemistry and Chemical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China. E-mail: wrzheng@sues.edu.cn; Fax: +86 21 67791220; Tel: +86 21 67791216
First published on 4th March 2016
The C–O homolytic bond dissociation enthalpies (BDEs) were evaluated by composite high-level G4 and 14 density function theory (DFT) methods. By comparing these DFT methods, the wB97 provided the most accurate results and the root mean square error (RMSE) is 7.6 kJ mol−1 for C–O BDE calculations. Therefore, alkenyl and aryl C(sp2)–O cleavage and the substituent effect of carboxylates/carbamates were investigated by a wB97 method. Based on the mechanism of the Ni-catalyzed Suzuki–Miyaura cross-coupling reactions, the aryl C(sp2)–O BDEs and the substituent effect of Ni complexes of carboxylates/carbamates formed in the oxidative addition step were also discussed. In addition, the NBO analysis further disclosed the essence of the substituent effects on C–O BDEs.
For example, Shi et al. achieved the cross-coupling reaction of aryl carboxylates with boronic acid in good yield, which is also applied to modify natural products such as estrone and flavone.10 Chatani et al. developed the Ni-catalyzed amination of aryl carboxylate and various aminated products can be obtained.11,12 Snieckus et al. reported Kumada coupling reactions of O–carbamates using the catalyst of [Ni(acac)2], in which the substrate scope was greatly expanded, such as the carbamates with both electron-donating and electron-withdrawing groups (Scheme 1). As a result, the highly functionalized arenes could be produced.13 Shi et al. successfully developed the Suzuki–Miyaura reaction of alkenyl acetates and carbamates, in which [NiCl2(PCy3)2] was testified as the best pre-catalyst (Scheme 2).8,9 Martin et al. discovered a mild Ni/Cu-catalyzed silylation of pivalates via C–O cleavage, which can be a powerful alternative methodology for preparing organic silanes from readily available precursors.14 Chatani et al. developed a new method for the rhodium-catalyzed Suzuki–Miyaura cross-coupling reaction of aryl carbamates with organoboron reagents.15
For these, it can be found that the alkenyl/aryl C(sp2)–O cleavage of carboxylates and carbamates is significant for understanding the cross-coupling reactions. The alkenyl/aryl C(sp2)–O homolytic bond dissociation enthalpy (BDE) of carboxylates and carbamates, as one of the thermodynamic properties which measures the bond strength, can describe the C(sp2)–O cleavage to some extent in the reactions. So far, there are many interesting questions which have not been systematically examined. These include the following: (1) what are the alkenyl/aryl C(sp2)–O BDE values of a specific carboxylate and carbamate? (2) How does the different substituent influence the C(sp2)–O bond strength? (3) Combined with the mechanistic studies, how does the transition-metal catalysis (especially the most used Ni catalysis) affect the alkenyl/aryl C(sp2)–O homolytic cleavage of carboxylates and carbamates? (4) From a microscopic perspective, what is the essence determining the C(sp2)–O bond strength? Answers to these questions will be helpful for further study of C(sp2)–O activation of these compounds involved in cross-coupling reactions.
In the present study, the alkenyl/aryl C(sp2)–O cleavage and the substituent effect in both carboxylates/carbamates and corresponding Ni complexes were investigated in detail by theoretical methods, due to the incomparable advantage in chemical thermodynamics compared with experimental techniques.16–18
R1–OC(O)R2(g) → R1˙(g) + R2(O)CO˙(g) | (1) |
R1–OC(O)NR2R3(g) → R1˙(g) + R2R3N(O)CO˙(g) | (2) |
The enthalpy of each species can be calculated from the equation:
H(298 K) = E + ZPE + Htrans + Hrot + Hvib + RT | (3) |
In the C–O BDE calculations, the composite high-level ab initio methods G4 (ref. 22) was used, which are suitable for systems of less than 8 non-hydrogen atoms. Furthermore, a series of DFT methods including B3LYP,23 X3LYP,24 M06,25 M06-2X,26 M06-L,27 t-HTCH,28 wB97,29 wB97X,29 wB97X-D,30 MPWB1K,31 MPW1K,32 MPWKCIS1K,33 MPW1KCIS33 and B98 (ref. 34) were selected. For all DFT calculations, geometry optimization was conducted at the B3LYP/6-31+G(d) level, which is a good choice for structure optimization for its high accuracy and reasonable computing resource demands.35–39 Frequency calculations were performed at the same level, in order to confirm the correct nature of the stationary points and to extract the zero-point vibrational energies (ZPE). The basis set for the single-point energies calculation by all DFT methods is 6-311++G (2df,2p). In the Ni complexes, for Ni atom, the effective core potential LANL2DZ basis set was used for optimization and the SDD basis set for single-point calculation. It is worth mentioning that the bond dipoles, natural spin densities and natural charge were calculated at the wB97/6-31+G(d) level by natural bond orbital (NBO) analysis.
All above the calculations were performed by the Gaussian 09 programs.40
Entry | Compounds | Exp. | G4 |
---|---|---|---|
a MD (mean deviation) = Σ(xi − yi)/N; MAD (mean absolute deviation) = Σxi − yi|/N; RMSE (root mean square error) = [Σ(xi − yi)2/N]1/2 (N = 12, xi represents the calculated data for each species, and yi represents the experimental data accordingly). | |||
1 | ![]() |
383.7 ± 12.6 | 378.6 |
2 | ![]() |
457.7 ± 2.1 | 453.5 |
3 | ![]() |
459.4 ± 4.2 | 452.9 |
4 | ![]() |
448.5 ± 10.5 | 450.8 |
5 | ![]() |
460.2 ± 8.4 | 466.1 |
6 | ![]() |
368.2 ± 10.5 | 367.6 |
7 | ![]() |
380.3 ± 12.6 | 364.2 |
8 | ![]() |
423.8 ± 4.2 | 416.9 |
9 | ![]() |
422.2 ± 6.3 | 415.3 |
10 | ![]() |
413.4 ± 8.4 | 413.2 |
11 | ![]() |
382.8 ± 12.6 | 372.2 |
12 | ![]() |
389.5 ± 12.6 | 377.8 |
MD | — | −2.8 | |
MAD | — | 6.9 | |
RMSE | — | 8.3 |
It can be found that for the 12 C–O BDE calculations, the discrepancies between experimental and G4 values are small (less than 10 kJ mol−1), except for the entry 7, entry 11 and entry 12. For these three C–O BDEs, the experimental uncertainties are all 12.6 kJ mol−1. Therefore, we considered that the G4 method can predict the C–O BDEs with superior accuracy. On the other hand, we also believe that the 12 experimental C–O BDEs are reasonable. The MD, MAD and RMSE values of G4 are −2.8 kJ mol−1, 6.9 kJ mol−1 and 8.3 kJ mol−1 respectively. The linear correlation between experimental and G4 C–O BDEs is shown in Fig. 1. And the correlation coefficient squares (R2) is 0.968.
By comparing with the G4 method, the density functional theory (DFT) method is currently practicable for BDE calculation of large systems because of relatively low CPU-cost and reasonable precision.44,45
Afterwards, we aim to obtain an accurate DFT method for systematic calculation of alkenyl/aryl C(sp2)–O BDEs of carboxylates and carbamates. Besides the 12 C–O BDEs in Table 1, we extended our training set by adding 7 C–O BDEs with experimental values in larger systems. And the 14 DFT methods, such as the exchange–correlation functional like M06 and M06-2X, the long-range corrected hybrid density functional like wB97 and wB97X were selected. The detailed results of 19 C–O BDEs calculating by 14 DFT methods are listed in the ESI.† The mean deviation (MD), mean absolute deviation (MAD) and root mean square error (RMSE) values are listed in the Table 2, which can describe the correlations between theoretical and experimental BDEs.
DFT methods | MD | MAD | RMSE |
---|---|---|---|
a The values in parentheses are calculated at wB97/6-311++G(2df,2p)//wB97/6-31+G(d) level. | |||
B3LYP | −38.6 | 38.6 | 40.3 |
X3LYP | −33.8 | 34.0 | 37.8 |
M06 | −1.1 | 9.6 | 11.5 |
M06-2X | 12.1 | 12.8 | 14.8 |
M06-L | −27.7 | 28.0 | 37.1 |
t-HTCH | −36.9 | 36.9 | 42.8 |
wB97 | −2.4(−4.8) | 5.9(10.4) | 7.6(13.8) |
wB97X | −5.8 | 7.5 | 9.3 |
wB97X-D | −5.9 | 13.1 | 16.8 |
MPWB1K | −3.5 | 7.7 | 9.3 |
MPW1K | −17.2 | 22.0 | 24.7 |
MPWKCIS1K | −19.5 | 19.5 | 21.6 |
MPW1KCIS | −34.1 | 34.1 | 45.4 |
B98 | −24.9 | 24.9 | 26.4 |
Among these DFT methods, the wB97 yields the highest precision with the smallest RMSE value of 7.6 kJ mol−1, and the MD, MAD values are −2.4 kJ mol−1, 5.9 kJ mol−1 respectively. The wB97X and MPWB1K methods give the second superior accuracy with the RMSE values of both 9.3 kJ mol−1. The worst method for C–O BDE calculation is MPW1KCIS and the RMSE is the highest of 45.4 kJ mol−1. The values of MD and MAD are both 34.1 kJ mol−1. As one of the most popular method, B3LYP46,47 presents the conspicuous systematic error with the MD, MAD, RMSE values of −38.6 kJ mol−1, 38.6 kJ mol−1 and 40.3 kJ mol−1 separately. In addition, we calculated the 19 C–O BDEs at the wB97/6-311++G(2df,2p)//wB97/6-31+G(d) level, and the results are also listed in this table (in parentheses). It can be found that the RMSE is 13.8 kJ mol−1, which is larger than wB97/6-311++G(2df,2p)//B3LYP/6-31+G(d) level (7.6 kJ mol−1). The linear correlation between the wB97 and experimental C–O BDEs is depicted in Fig. 2. And the correlation coefficient squares (R2) is 0.978.
In summary, we recommended the wB97/6-311++G (2df,2p)//B3LYP/6-31+G(d) level method was used to predict alkenyl/aryl C(sp2)–O BDEs of carboxylates and carbamates in the following.
By comparison, the R1 group plays a remote effect on the C–O BDEs. The C(sp2)–O BDEs with the different R1 groups of carboxylates and carbamates are summarized in the Table 3. After the natural bond orbital (NBO) analysis48 conducted at wB97/6-31+G(d) level, the values of C–O bond dipole are also shown in the table.
It can be found that the C–O BDEs cover a range of over 40 kJ mol−1 that from 420.3 kJ mol−1 to 464.0 kJ mol−1 for carboxylates, and there is about 32 kJ mol−1 gaps (from 396.4 to 428.2 kJ mol−1) for carbamates. It can be seen that the C–O BDEs of carboxylates are larger than the carbamates, and for the same R1 group, the discrepancy between them is more than 20 kJ mol−1. The natural spin densities of O˙ radicals after the C–O breakage in carboxylates and carbamates are depicted in the Fig. 4. The BDE discrepancy can be interpreted as that the spin density of O˙ radical of carbamates mainly delocalized onto the N atom, which makes the radical more stable.
For substituted carboxylates, comparing with the –H, the EWGs of R1 such as –NO2, –CHO and –COOH can decrease the C–O BDEs. For example, when the R1 is –CF3, the C–O BDE is 420.3 kJ mol−1, which is much smaller than the 443.9 kJ mol−1 (R1 = H). However, the EDGs of R1 including –CH3, –NH2 can increase the C–O BDEs. For example, the largest C–O BDE is found in R1 = NH2 (464.0 kJ mol−1). When the R1 is –F, the C–O BDE is 439.8 kJ mol−1, which is slightly smaller than –H, which shows that the –F group exhibits the electron-withdrawing effect on C–O BDEs. By comparison, the CEGs of R1 have little influences on C–O BDEs in carboxylates, which may be because the π–π conjugate effect between the R1 and –CHCH– plays the similar role in the radicals and molecules. In addition, the same change pattern of C–O BDEs is found in the carbamates.
Herein, as shown in Table 3, the NBO C–O bond dipole values range from 1.37 to 1.58 for carboxylates, and from 1.35 to 1.68 for carbamates. The NBO analysis produced two good linear correlations between the C–O BDEs of carboxylates and carbamates and C–O bond dipole values. The relationships are depicted in the Fig. 5 and the correlation coefficients (R) are 0.867 and 0.866 respectively. The relationships show that the larger C–O BDEs would lead to larger NBO C–O bond dipole.
In addition, Hammett analysis49–53 can provide us a way to predict the BDE changing patterns. In the present study, two excellent linear relationships between C(sp2)–O BDEs of carboxylates and carbamates with substituent constants σp+ (listed in Table 3) were found, in which the correlation coefficients (R) are 0.946 and 0.960 respectively. The correlations are depicted in the Fig. 6.
Secondly, the R2 group has a direct influence on the C–O BDEs comparing with R1. The C–O BDEs of substituted carboxylates and carbamates with different R2 groups are listed in the Table 4, in which the R1 group is fixed as –CH3 for discussion. Simultaneously, the qC × qO values are listed in this table. It can be seen that the C–O BDEs are in the range of 418.4 kJ mol−1 to 476.8 kJ mol−1 for carboxylates, and there is a difference of about 60 kJ mol−1 between them. For carbamates, there is about 80 kJ mol−1 difference between the minimum 355.7 kJ mol−1 and the maximum 434.7 kJ mol−1. The results indicated that the R2 plays a larger effect on the C–O BDEs than the remote effect of R1.
R2 | Entry | Entry | ||||
---|---|---|---|---|---|---|
BDEs (kJ mol−1) | qC × qO | BDEs (kJ mol−1) | qC × qO | |||
–NO2 | 1 | 437.3 | −0.194 | 11 | 392.3 | −0.205 |
–CHO | 2 | 433.4 | −0.078 | 12 | 402.4 | −0.092 |
–COOH | 3 | 441.7 | −0.073 | 13 | 369.7 | −0.062 |
–CN | 4 | 422.9 | −0.070 | 14 | 389.2 | −0.090 |
–CH3 | 5 | 447.3 | −0.152 | 15 | 413.0 | −0.176 |
–NH2 | 6 | 447.2 | −0.263 | 16 | 410.8 | −0.274 |
–F | 7 | 476.8 | −0.376 | 17 | 434.7 | −0.403 |
–OH | 8 | 463.3 | −0.327 | 18 | 438.1 | −0.339 |
–OCH3 | 9 | 459.1 | −0.331 | 19 | 429.6 | −0.355 |
–Ph | 10 | 418.4 | −0.149 | 20 | 355.7 | −0.164 |
For carboxylates (entry 1–10), it can be found that the EWGs of R2 including –NO2, –CN, –CHO, and –COOH (entry 1–4) can reduce the C–O BDEs. By contrast, the EDGs of R2 including –CH3, –NH2, –OH, –F, and –OCH3 (entry 5–9) can obviously increase the BDEs. For example, when the R2 = OH, the C–O BDE is 463.3 kJ mol−1, which is larger than the R2 = CN of 422.9 kJ mol−1. The largest C–O BDE is found in the R2 = F (476.8 kJ mol−1), implying that the –F exhibits the strong electron-donating effect of R2 on C–O BDEs. For the CEG (R2 = Ph, entry 10), the C–O BDE is the smallest of 418.4 kJ mol−1. The same phenomena can be found in the carbamates (entry 11–20).
The natural spin densities of the radical center C˙ with different R2 groups such as –CN, –F and –Ph are listed in the Fig. 7. The BDE change pattern of R2 can be explained as following: when R2 is –F (EDG), the natural spin density is 0.968, which mainly concentrated on the C˙ radical center. However, for R2 = –CN (EWG), the natural spin densities of the radical center C˙ are 0.853, which are partly delocalized onto the –CN group. It can be concluded that the EWGs may directly disperse the electron density of the radical center and enhance the stability of radical. On the contrary, the EDGs are disadvantageous for the radical stabilities, which can make the C–O cleavage more difficult. For the –Ph (CEG), the natural spin densities of the radical center C˙ are the smallest of 0.703, in which the delocalization effect is the most obvious. It shows that the π–π conjugated effect between the –Ph and the CC bond, as well as the steric effect of –Ph both have large influences on C–O BDEs. And the stronger delocalization effect of the radical center can lead to the smaller C–O BDEs of carboxylates and carbamates.
After the NBO analysis conducted at wB97/6-31+G(d) level, the values of qC × qO are also shown in the table, where the qC and qO denote the natural charges of C and O atoms of C–O bond in carboxylates and carbamates respectively. In addition, the qC × qO values range from −0.376 to −0.070 for carboxylates and from −0.403 to −0.062 for carbamates. As shown in Fig. 8, the NBO analysis disclosed two good linear correlations between the C–O BDEs of carboxylates and carbamates and qC × qO values, and the correlation coefficients (R) are 0.880 and 0.873 respectively, in which two abnormal points marked in the figure are excluded (entry 10 in (a), entry 20 in (b)). Generally, the larger C–O BDEs would result in larger absolute qC × qO, which reveals the essence of C–O bond.
Overall, comparing the effect of R1 with R2, the same C–O BDEs change patterns of R1 and R2 are found in carboxylates and carbamates. Interestingly, in our past studies, the remote and direct substituent groups exhibited different C–O BDEs change patterns in alkenyl phosphates/sulfonates.54
On one hand, the aryl C–O BDEs of substituted carboxylates are in the range of 451.4–466.8 kJ mol−1 and there is only about 15 kJ mol−1 difference. Herein, the small difference indicated that the remote substituents on benzene play a minor role in C–O BDEs. That is to say, regardless of whether the groups are EWGs or EDGs, and regardless of whether the groups are located at the ortho-, meta-, or para-positions on benzene, the C–O BDEs changing is not obvious.
On the other hand, for substituted aryl carbamates, the BDEs range from 384.6 to 430.9 kJ mol−1. Different with carboxylates, the C–O BDEs with ortho-substituted groups on benzene are much smaller than meta- and para-positions except the –CN and –F. In addition, the C–O BDE differences are small in terms of meta- and para-substituted carbamates for both EWGs and EDGs.
The mechanism of the Ni-catalyzed Suzuki–Miyaura cross-coupling reaction of aryl acetates/carbamates were theoretically investigated by Li et al.55 and Houk et al.56 The oxidative addition,57–61 transmetalation62–65 and reductive elimination66–69 are the three key steps in the catalytic cycle, in which the C–O activation in the oxidative addition step plays the important role. The preferred mono-phosphine mechanism of oxidative addition step by Li and Houk are depicted in the Fig. 9 in detail. Based on the studies, we calculated the aryl C(sp2)–O BDEs of the Ni complex (a) for carboxylates and Ni complex (b) for carbamates (Fig. 9) in the oxidative addition step with different substituents, and the results are listed in the Table 6, in which the ligand L is selected the P(CH3)3 as the computational model for simplicity.
![]() | ||
Fig. 9 The oxidative addition step in the mechanism of Ni-catalyzed Suzuki–Miyaura cross-coupling with aryl carboxylates/carbamates. |
Comparing with the aryl C–O BDEs in carboxylates/carbamates (Table 5), we can find that the C–O BDEs in the Ni complex are greatly reduced (over 100 kJ mol−1). For example, the C(sp2)–O BDE with –F ortho-substituted in phenyl acetate (in Table 5) is 463.0 kJ mol−1, while in the corresponding Ni complex, the C–O BDE is greatly lowered to 260.2 kJ mol−1. By reference, in the hydrogen peroxide HO–OH, the O–O BDE is 210.66 kJ mol−1.70
The aryl C–O BDEs in Ni complex of carboxylates are in the range of 260.2–308.1 kJ mol−1 and there is over 40 kJ mol−1 difference. The large difference disclosed that the substituent effects become distinguished in the oxidative addition with the Ni catalysis. In addition, for EDGs like –OCH3, –F and EWGs like –CF3, –CHO, there exists evidence ortho-effect in the Ni complex. While, for two EDGs (–CH3 and –NH2), the ortho-effect is not obvious. For example, the C–O BDE with –F ortho-substituted is 260.2 kJ mol−1, and the C–O BDEs with –F meta- and para-substituted are 286.7 kJ mol−1 and 288.1 kJ mol−1 respectively, in which over 20 kJ mol−1 discrepancy is found.
The aryl C–O BDEs in Ni complex of carbamates range from 212.6 kJ mol−1 to 263.4 kJ mol−1, and there is over 50 kJ mol−1 difference. Similarly, for most of the substituted carbamates in Table 6, the C–O BDEs with ortho-substituted are much smaller than meta- and para-positions (only except –CH3), which shows that the substituents including EDGs and EWGs exhibit evident ortho-effect in the Ni complex. The theoretical results proved the well-known ability of aryl carbamates in directed ortho metalation (DoM), which are in accordance with the experimental studies.7,8,71–74 For example, in the experimental reports of Shi et al., the ortho-OMe substituted carbamates in the cross-coupling reaction had the good yields with 70%, which indicated the superior stabilities and good behavior in DoM.7 By comparison, in the experimental reports of Garg et al., the para-OMe substituted carbamates in the cross-coupling reaction had just 41% yield.75
By NBO analysis, the natural charges of C and O atoms of C–O bond in carboxylates/carbamates and the corresponding Ni complexes as well as the natural spin density of C˙ after C–O cleavage are listed in the Fig. 10. In the phenyl acetate with –CHO para-substituted, it can be found that the absolute qC × qO values are larger than the corresponding Ni complex. And the sum of natural charges of Ni and P atoms is as high as 1.188 in the Ni complex, which shows that the charge distribution is greatly influenced by the Ni catalysis in the oxidative addition. After the C–O cleavage, the natural spin densities of the radical center C˙ of carboxylate and Ni complex are 1.017 and −0.003 respectively, which indicated that there are significant delocalization effects onto Ni atom (the natural spin density of Ni is 1.017) in Ni complex radical. The similar results are found in the carbamate with –OCH3 ortho-substituted. These phenomena disclosed the essence of the C(sp2)–O activation by transition metal catalysis in the cross-coupling reactions.
More intuitively, a molecular orbital analysis76 has been conducted. The highest occupied molecular orbitals (HOMO) of substituted carboxylates/carbamates and the corresponding Ni complexes are shown in Fig. 11. In the substituted carboxylates/carbamates, the electron densities are almost concentrated on the benzene, while in the corresponding Ni complexes, the electron densities are greatly located at the Ni center.
(1) In alkenyl carboxylates/carbamates, for the R1 group (remote effect), the EWGs of R1 can decrease the C(sp2)–O BDEs. However, the EDGs of R1 can increase the C(sp2)–O BDEs. By comparison, the CEGs of R1 have little influences on C–O BDEs. The NBO analysis produced two good linear correlations between the C–O BDEs of alkenyl carboxylates and carbamates and C–O bond dipole values. In addition, two excellent linear relationships between C(sp2)–O BDEs of carboxylates and carbamates with substituent constants σp+ were also found.
(2) In alkenyl carboxylates/carbamates, comparing with R1, the R2 plays a larger effect on the C–O BDEs. The EWGs of R2 can reduce the C–O BDEs. By contrast, the EDGs of R2 can obviously increase the BDEs. For the CEG, the C–O BDE is the smallest. Overall, comparing the effect of R1 with R2, the same C–O BDEs change patterns of R1 and R2 are found in alkenyl carboxylates/carbamates, which is interestingly different with the C–O BDE change patterns in our previous studies.
(3) In aryl substituted carboxylates, the remote substituents on benzene play a minor role in C–O BDEs. While in substituted aryl carbamates, the C–O BDEs with ortho-substituted groups on benzene are almost much smaller than meta- and para-positions.
(4) In the Ni complexes of carboxylates/carbamates, the C–O BDEs are greatly reduced (over 100 kJ mol−1). Furthermore, for aryl carboxylates, the substituent effects become distinguished in the oxidative addition with the Ni catalysis. The theoretical results proved the well-known ability of aryl carbamates in directed ortho metalation (DoM), which are in accordance with the experimental studies.
Footnote |
† Electronic supplementary information (ESI) available: The 19 C–O BDEs calculated by 14 DFT methods. See DOI: 10.1039/c5ra27859g |
This journal is © The Royal Society of Chemistry 2016 |