To verify the reliability of the above computational method, the gas-phase PBEPBE/BS1-Auto optimized structure of [(cyclohex-3-enyl)-η
6-benzene]manganese tricarbonyl complex
2, [(C
6H
9)(η
6-Ph)Mn(CO)
3+][BF
4], was compared with its reported X-ray structure (CSD entry: YUBXOI) [
12] (
Tables S1 and S2). A reasonable root-mean-square deviation (RMSD) of 0.2105 Å was obtained, with the crystal packing of CO groups causing the largest deviation. The additional comparisons from the gas-phase optimization of PBE-D3(BJ)/BS1-Auto and PBE-D3(BJ)/BS2-Auto were also performed, and slight improvements in RMSDs were observed. To further confirm the accuracy of the gas-phase optimization, the Gibbs free energies computed from gas-phase PBE/BS1-Auto and PBE-D3(BJ)/BS1-Auto were compared. An acceptable mean absolute deviation (MAD) of 1.64 (
Table S3) and an excellent linear fitting (R
2 = 0.9908,
Figure S1) were presented. To reasonably address the effect of polarization functions of hydrogen atoms on the geometry optimization, the PBE/BS4-Auto optimized structures were matched with the PBE/BS1-Auto optimized ones, and significantly small values of RMSD (in Å) were obtained (
Table S4). Additionally, no obvious differences in the electron density of bond critical point ρ
(BCP) from PBE/BS1-Auto optimized structures and PBE/BS4-Auto optimized ones could be observed (MSD = 0.004, MAD = 0.004) (
Figures S3 and S4). These observations clearly demonstrated the suitability of current method in the geometry optimization, which is also consistent with our previous studies and the other reported work [
20,
21,
22,
23].
2.1. Pathways for the Mn(CO)3 Migration
To gain insight into the unusual migration of the Mn(CO)
3 fragment from the cyclohexenyl ring to the phenyl group (
Scheme 2), the migration pathways were computationally investigated and are illustrated in
Figure 1 and
Figure 2. The protonation of (exo-phenyl)(η
3-cyclohexenyl)manganese tricarbonyl [(Ph)(η
3-C
6H
8)Mn(CO)
3] (complex
1,
Scheme 2) by HBF
4.Et
2O in dichloromethane (DCM) initially generated the (η
3-cyclohexenyl)Mn-hydride complex
3 (
Figure 1), which is favorable by −23.9 kcal mol
−1 (57.8 vs. 33.9 kcal mol
−1,
Figure 1). Once the (η
3-cyclohexenyl)Mn-hydride complex
3 was formed, the following migration of the introduced hydride atom formed the di-agostic (η
2-cyclohexenyl)manganese complex
4 with a Gibbs barrier of 1.3 kcal mol
−1 (
3 →
TS-3-4 →
4,
Figure 1). Compared to the mono-agostic (η
3-cyclohexenyl)Mn-hydride complex
3, the di-agostic (η
2-cyclohexenyl)manganese complex
4 is favorable by 9.1 kcal mol
−1 (33.9 vs. 24.8 kcal mol
−1,
Figure 1). Due to the initially existing and the later formed (two) Mn-H-C agostic bonds in the di-agostic (η
2-cyclohexenyl)manganese complex
4, the subsequent breaking of these two Mn-H-C agostic units in complex
4 leads to two different mono-agostic (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
5 (
4 →
5) and complex
6 (
4 →
6). The Gibbs barriers for the breaking of the initially existing Mn
…H
…C agostic bond in complex
4 to form complex
5 (
Figure 1) and the breaking of later formed Mn-H-C agostic bond to form complex
6 (
Figure 2) are 3.5 kcal mol
−1 (
4 →
TS-4-5 →
5,
Figure 1) and 2.4 kcal mol
−1 (
4 →
TS-4-6 →
6,
Figure 2), respectively. It is worth noting that the mono-agostic (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
6 (20.5 kcal mol
−1) is significantly lower in Gibbs free energy compared to its isomer complex
5 (27.4 kcal mol
−1,
Figure 1), and the steric effect from the methylene group is believed to be the main reason for the observed difference.
From the mono-agostic (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
5, the formation of the di-agostic (η
2-phenyl)manganese complex
8 was located with a Gibbs barrier of 15.4 kcal mol
−1 (24.8 kcal mol
−1 for
4 vs. 40.2 kcal mol
−1 for
TS-5-8,
Figure 1). Followed by the formation of the di-agostic (η
2-phenyl)manganese complex
8, the slip** of the η
2-phenyl group formed another di-agostic (η
2-phenyl)manganese complex
9 with a significant low Gibbs barrier of 2.0 kcal mol
−1 (
8 →
9,
Figure 1). The mono-agostic (η
2-phenyl)manganese complex
10 was then generated via the breaking of the endo-agostic bond in the di-agostic (η
2-phenyl)manganese complex
9, and the mono agostic η
2-phenyl complex
10 was 1.7 kcal mol
−1 more favorable compared to the di-agostic (η
2-phenyl)manganese complex
9 (34.5 vs. 36.2 kcal mol
−1,
Figure 1). The breaking of the exo-agostic bond in the mono-agostic η
2-phenyl complex
10 then generated the (η
6-phenyl)manganese complex
11 (
10 →
11,
Figure 1), which was favorable by 32.4 kcal mol
−1 (34.5 kcal mol
−1 for
10 vs. 2.1 kcal mol
−1 for
11,
Figure 1). To remain consistent with the reported X-ray structure of [(cyclohex-3-enyl)-η
6-benzene]manganese tricarbonyl cation complex
2 (CSD entry: YUBXOI) [
12], a two-step rotation of the cyclohexenyl groups was required from the (η
6-phenyl)manganese complex
11 (
11 →
12 →
2,
Figure 1). As a straightforward C
ph–C
cy single bond rotation, the expected low Gibbs barriers for the two-step rotation of 1.8 kcal mol
−1 (
11 →
12,
Figure 1) and 1.0 kcal mol
−1 (
12 →
2,
Figure 1) were obtained. The rate-limiting step in the first path (
4 →
5 →
8 →
9 →
10 →
11 →
12 →
2,
Figure 1) was the formation of the di-agostic (η
2-phenyl)manganese complex
8 (
TS-5-8,
Figure 1) with an overall Gibbs barrier of 15.4 kcal mol
−1 (24.8 kcal mol
−1 for
4 vs. 40.2 kcal mol
−1 for
TS-5-8). A net Gibbs reaction energy of 57.8 kcal mol
−1 in the exothermic migration of Mn(CO)
3 group from the (exo-phenyl)(η
3-cyclohexenyl)manganese tricarbonyl [(Ph)(η
3-C
6H
8)Mn(CO)
3] (complex
1), forming the η
6-benzene complex
2 [(C
6H
9)(η
6-Ph)Mn(CO)
3+][BF
4], was then obtained. It is worth noting that the protonation-induced migration of the Mn(CO)
3 group from (exo-phenyl)(η
3-cyclohexenyl)manganese tricarbonyl [(Ph)(η
3-C
6H
8)Mn(CO)
3] (complex
1) to form the η
6-benzene complex
2 [(C
6H
9)(η
6-Ph)Mn(CO)
3+][BF
4] was accomplished by a multi-step process, and a straightforward intramolecular inter-ring haptotropic rearrangement as presented in the Cr(CO)3 complexes [
15,
16,
17,
18,
19] was not observed in the current case. A series of agostic intermediates (
3,
4,
5,
7,
8,
9,
10) were involved in this protonation-induced migration of Mn(CO)
3 group, promoting the migration of Mn(CO)
3 group from the relatively electron-deficient cyclohexenyl group to the electron-rich phenyl group.
From the mono-agostic (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
6, two pathways were located leading to the formation of the di-agostic (η
2-phenyl)manganese complex
8 (
6 →
7 →
8,
Figure 2) and the η
6-benzene isomer complex
13, [(C
6H
9)(η
6-Ph)Mn(CO)
3+][BF
4] (
6 →
13,
Figure 1), respectively. In the first path, the vibration of the methylene group in the (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
6 broke the hydride atom and introduced the Mn
…H
…C agostic bond, but formed a new (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
7 (29.3 kcal mol
−1,
6 →
7,
Figure 2) with the methylene group involved with the Mn-H-C agostic bond. A low Gibbs barrier of 9.2 kcal mol
−1 was observed for the change in the Mn-H-C agostic bond (
6 →
7,
Figure 2). Next, the breaking of the Mn-C bonds between the η
2-cyclohexenyl and manganese in complex
7 generated the di-agostic (η
2-phenyl)manganese complex
8 with a Gibbs barrier of 10.9 kcal mol
−1 (
7 →
8,
Figure 2). A slightly higher Gibbs barrier of 15.5 kcal mol
−1 (24.8 kcal mol
−1 for complex
4 vs. 40.3 kcal mol
−1 for
TS-7-8), compared with the 15.4 kcal mol
−1 (24.8 kcal mol
−1 for complex
4 vs. 40.2 kcal mol
−1 for
TS-5-8,
Figure 1) for the rate-limiting step in the second path (
4 →
6 →
7 →
8,
Figure 2) was observed. Alternatively, in the second path (
6 →
13,
Figure 1), a single-step formation of another η
6-benzene isomer complex
13 from the mono-agostic (η
2-phenyl)(η
2-cyclohexenyl)manganese complex
6 was also successfully located with a Gibbs barrier of 15.9 kcal mol
−1 (
6 →
TS-6-13 →
13,
Figure 1). Once the η
6-benzene isomer complex
13 was formed, a simple rotation of the cyclohexenyl group led to the final η
6-benzene complex
2, [(C
6H
9)(η
6-Ph)Mn(CO)
3+][BF
4] with a low Gibbs barrier of 2.5 kcal mol
−1 (
TS-2-13,
Figure 1).
2.2. Characterization of Agostic Complexes
The above established pathways on the unusual protonation-induced migration of the Mn(CO)
3 fragment from the cyclohexenyl group to the phenyl group (
Figure 1 and
Figure 2) presented the following observations: (1) the overall rate-limiting step for this unusual migration is the formation of di-agostic (η
2-phenyl)manganese complex
8 (
4 →
TS-4-5 →
5 →
TS-5-8 →
8,
Figure 1) with a Gibbs barrier of 15.4 kcal mol
−1 (24.8 kcal mol
−1 for
4 vs. 40.2 kcal mol
−1 for
TS-5-8); (2) the exothermic reaction from complex
3 to complex
2 is overall favorable by 33.9 kcal mol
−1; and (3) the mono agostic complexes (
3,
5,
6,
7, and
10) and di-agostic complexes (
4,
8, and
9) served as the main intermediates in the above exothermic migration. To better understand the roles of these agostic complexes in the migration of Mn(CO)
3, the agostic bonds in these intermediates were further computationally characterized and well analyzed.
Several well-known geometry parameters and bonding characters in the agostic complex include the following: (
i) short TM-H distance (1.8–2.3 Å), (
ii) small TM
…H
…C bond angle (90–140°), (
iii) up field-shift agostic hydrogen, and (
iv) low spin coupling J
CH (50–100 Hz) [
7,
24]. Studies of the orbital interactions in the Mn
…H
…C agostic bonding indicated that both the σ
(C-H) →
d(TM) and
d(TM) → σ*
(C-H) π-back donation interactions contributed to the Mn
…H
…C agostic bond [
9,
25]. DFT-computed agostic parameters of the mono-agostic complexes (
3,
5,
6,
7 and
10) and di-agostic complexes (
4,
8 and
9) are summarized in
Table 1.
It is clear from
Table 1 that all these complexes (
3,
4,
5,
6,
7,
8,
9 and
10) fit the above well-known geometrical parameters of an agostic complex with the shortened Mn–H distance (1.813–2.085 Å), the prolonged C–H
(agostic) bond length (1.133–1.189 Å), and the small Mn
…H
…C bond angle (93.7–119.8°). The J
CH coupling constants in the Mn
…H
…C unit of these mono-agostic complexes (
3,
5,
6,
7 and
10) and di-agostic complexes (
4,
8 and
9) were about 53 Hz (averaged) lower than those J
CH couplings of the non-agostic ones. The high field agostic hydrogen atoms in the Mn
…H
…C agostic unit compared to the non-agostic hydrogen were also confirmed by the proton chemical shifts (by 6.9 ppm, averaged). The AIM (Atoms-In-Molecules) analyses of the Mn
…H
…C unit in the mono-agostic complexes (
3,
5,
6,
7 and
10) and di-agostic complexes (
4,
8 and
9) were presented in
Figure 3. The relative strength of a Mn–H bond and a H–C bond in the Mn
…H
…C agostic unit could be measured by the calculated electron densities of bond critical points [ρ
(BCP)] and the absolute value of the Laplacian of electron density (∇
2ρ) (
Figure 3), and a stronger chemical bond is characterized as a shorter bond distance and bigger Wiberg bond index, which could be demonstrated by the comparison of agostic complexes
3 and
5. The Mn–H distance in agostic complexes
3 and
5 are 1.823 and 1.977 Å (
Table 1), respectively, showing a stronger Mn–H bond in agostic complex
3 compared with complex
5. It was also confirmed by the computed Wiberg bond index of complexes
3 and
5 (0.16 vs. 0.11,
Table 1). The calculated electron densities of Mn–H bond critical points [ρ
(BCP)] for agostic complexes
3 and
5 are 0.0534 and 0.0461 a.u. (
Figure 3), and the related absolute value of the Laplacian of electron density (∇
2ρ) are 0.234 and 0.223 a.u., respectively (
Figure 3). It is worth noting that the endo Mn
…H
…C agostic unit in the di-agostic complex
9 (the
9*) had the longest Mn–H distance of 2.085 Å among all the agostic Mn–H bonds and had the shortest C–H bond of 1.133 Å (
Table 1) among all the agostic C–H bonds. Consequently, the smallest value of electron density of Mn–H bond critical points [ρ
(BCP)] (0.0268 a.u.,
Figure 3) and the smallest value of Laplacian of electron density (∇
2ρ) (0.113 a.u.,
Figure 3) were observed in endo-agostic Mn-H-C unit of the di-agostic complex
9 (the
9*). In contrast to the weak agostic Mn–H bond, the strongest agostic C–H bond of 1.133 Å in the di-agostic complex
9 (the
9*) was verified by the biggest value of electron density of agostic C–H bond critical points [ρ
(BCP)] (0.241 a.u.,
Figure 3) and the biggest absolute value of Laplacian of electron density (∇
2ρ) (0.657 a.u.,
Figure 3). The AIM analyses of other agostic intermediates were also obtained and presented in
Figure 3, confirming the existence of the Mn
…H
…C agostic interaction, which was consistent with previous reports [
20,
21].
To visually evaluate the agostic interactions in mono-agostic complexes (
3,
5,
6,
7 and
10) and di-agostic complexes (
4,
8 and
9), the NAdOs (natural adaptive orbitals) of the Mn
…H
…C agostic interaction were introduced (
Figure 4). Analyses of the NAdOs provided the following facts: (1) the eigenvalues of NAdOs of the Mn
…H
…C agostic unit in the complexes
3,
4 and
4* are significantly bigger than others (0.281,0.277 and 0.285, respectively,
Figure 4); (2) the smallest eigenvalue of 0.177 for agostic complex
9* is observed with the least contribution of 3
d(Mn) orbital into the NAdO of the Mn
…H
…C agostic unit (9.1%); (3) the highest contribution of 3d
(Mn) to the NAdO of the Mn
…H
…C agostic unit (20.9%) is observed in the mono agostic complex
3; and (4) the contribution of 2
p(C) orbital to the NAdO of the Mn
…H
…C agostic unit in agostic complex
9* is remarkably higher than that of agostic complex
3 (35.2% vs. 27.0%,
Figure 4). The higher contribution of the 3
d(Mn) orbital to the NAdOs of the Mn
…H
…C agostic unit in agostic complex
3 compared to complex
9* agrees with the relative stronger Mn–H bond in complex
3 than that of complex
9* (
Table 1,
Figure 3) [
20]. Meanwhile, the higher contribution of 2
p(C) orbital to the NAdOs of the Mn
…H
…C agostic unit in agostic complex
9* compared to complex
3 is entirely consistent with the stronger agostic C–H bond in complex
9* than that of complex
3 (
Table 1,
Figure 3). Additionally, to investigate the role of Mn
…H
…C agostic interactions in the stabilization of Mn agostic intermediates, the second order perturbative energy, E
(2), was obtained from the NBO computation. NBO analyses showed that the interaction of the σ
(C-H) donor with the 3
d*
(Mn) empty acceptor (σ
(C-H) → 3
d*
(Mn)) was the major contribution in Mn
…H
…C agostic interaction, and relative weak contribution from the back-donation of 3
d(Mn) donor to the σ*
(C-H) acceptor was also located (
Table S5). Not surprisingly, the lowest estimated stabilization energy of the Mn
…H
…C agostic interaction via the computed E
(2) from the dominant σ
(C-H) → 3
d*
(Mn) interaction in agostic complex
9* was observed (28.95 kcal mol
−1,
Table S5), which was notably lower than that of agostic complex
3 (63.63 kcal mol
−1,
Table S5).