Understanding Precatalyst Activation and Speciation in Manganese-Catalyzed C–H Bond Functionalization Reactions

An investigation into species formed following precatalyst activation in Mn-catalyzed C–H bond functionalization reactions is reported. Time-resolved infrared spectroscopy demonstrates that light-induced CO dissociation from precatalysts [Mn(C^N)(CO)4] (C^N = cyclometalated 2-phenylpyridine (1a), cyclometalated 1,1-bis(4-methoxyphenyl)methanimine (1b)) in a toluene solution of 2-phenylpyridine (2a) or 1,1-bis(4-methoxyphenyl)methanimine (2b) results in the initial formation of solvent complexes fac-[Mn(C^N)(CO)3(toluene)]. Subsequent solvent substitution on a nanosecond time scale then yields fac-[Mn(C^N)(CO)3(κ1-(N)-2a)] and fac-[Mn(C^N)(CO)3(κ1-(N)-2b)], respectively. When the experiments are performed in the presence of phenylacetylene, the initial formation of fac-[Mn(C^N)(CO)3(toluene)] is followed by a competitive substitution reaction to give fac-[Mn(C^N)(CO)3(2)] and fac-[Mn(C^N)(CO)3(η2-PhC2H)]. The fate of the reaction mixture depends on the nature of the nitrogen-containing substrate used. In the case of 2-phenylpyridine, migratory insertion of the alkyne into the Mn–C bond occurs, and fac-[Mn(C^N)(CO)3(κ1-(N)-2a)] remains unchanged. In contrast, when 2b is used, substitution of the η2-bound phenylacetylene by 2b occurs on a microsecond time scale, and fac-[Mn(C^N)(CO)3(κ1-(N)-2b)] is the sole product from the reaction. Calculations with density functional theory indicate that this difference in behavior may be correlated with the different affinities of 2a and 2b for the manganese. This study therefore demonstrates that speciation immediately following precatalyst activation is a kinetically controlled event. The most dominant species in the reaction mixture (the solvent) initially binds to the metal. The subsequent substitution of the metal-bound solvent is also kinetically controlled (on a ns time scale) prior to the thermodynamic distribution of products being obtained.


■ INTRODUCTION
One of the most important steps in transition-metal-catalyzed reactions is the activation of the precatalyst. In most reactions catalyzed by organometallic complexes, the bench-stable reagent added to a reaction is a precatalyst that requires activation before it can participate in the bond activation and formation events that constitute the catalytic reaction coordinate. This activation process may arise from a number of different pathways. For example, a saturated precatalyst may undergo the loss of a coordinated ligand to permit substrate binding, or alternatively, a change in the coordination mode of an already coordinated ligand may promote the same phenomenon. Other examples of precatalyst activation include a change in oxidation state (e.g., Pd(II) → Pd(0)) or the use of a hemilabile ligand to reveal a vacant coordination site.
In all cases, the resulting activated metal complex is more reactive than the precatalyst and thus may initially interact with many of the different reaction components rather than the desired substrate alone. Gaining insight into the immediate fate of an activated precatalyst and its interaction with the different components of the reaction mixture is challenging, primarily due to its anticipated high reactivity and commensurately short lifetime.
This problem is exacerbated when studying the chemistry of 3d transition metal catalysts. Due to the smaller radial extent of the 3d orbitals, metal−ligand bonds are generally weaker compared to their 4d and 5d cogeners. 1 Furthermore, metal complexes based on 3d metals are more prone to one-electron, rather than two-electron, transfer reactions, 2 and different preferential coordination numbers entail that different mechanistic pathways with higher rates of substitution may occur. 3 Although care must be taken with such generalities (for example, the rate constant of substitution of M(CO) 5

(THF) by donor
Special Issue: Advances and Applications in Catalysis with Earth-Abundant Metals ligands increases in the order Mo > Cr > W with a shift in mechanism from dissociative to associative as the periodic table is descended), 4 it is clear that the mechanistic processes underpinning catalysis by 3d metals have the potential for increased complexity. 5 In a series of recent studies, we have demonstrated how timeresolved infrared (TRIR) spectroscopy may be used to directly observe the key steps underpinning Mn-catalyzed C−H bond functionalization reactions. 6−12 Central to the success of this approach has been the light-induced loss of a carbonyl ligand from a precatalyst, [Mn(C^N)(CO) 4 ] (C^N = cyclomanganated ligand). The loss of CO correlates with the activation pathway observed under thermal conditions to give the catalytically active tricarbonyl complexes, fac-[Mn(C^N)-(CO) 3 (L)]. 13−15 In the TRIR experiments, the photochemical pump simulates the activation process occurring thermally, and the subsequent IR probe pulse allows the fate of the manganese complex to be monitored through the vibrational modes of the remaining carbonyl ligands. The experiments have harnessed the ability of the time-resolved multiple-probe spectroscopy (TR M PS) method 16,17 to acquire spectra with pump−probe delays ranging from picoseconds to milliseconds. This has allowed for the observation of (i) solvent-coordinated complexes following CO loss; 7 (ii) coordination and subsequent insertion of alkynes, alkenes, and isocyanates into the Mn−C bond 6,11,12,18 ( Figure  1a); (iii) proton-shuttling events with the coordination sphere of the metal, including the microscopic reverse of the concerted metalation−deprotonation (CMD) mechanism; 10 (iv) competition between water and N 2 ligands for the manganese; 8 and (v) the intermediates involved in the borylation of aryl and heteroaryl diazonium salts. 9 It was anticipated that this approach could be used to gain an understanding of the speciation that occurs following the activation of the [Mn(C^N)(CO) 4 ] precatalyst in catalytic reaction mixtures. The band positions of the photoproducts arising from CO loss from [Mn(C^N)(CO) 3 (L)] are highly sensitive to the nature of the newly coordinated ligand "L" and the spectroscopic resolution of the LIFEtime spectrometer at the ULTRA facility used for the TR M PS experiments (ca. 2 cm −1 ), meaning that complex spectra containing multiple species may be deconvoluted. A strategy was envisaged in which the interactions of light-activated [Mn(C^N)(CO) 4 ] with the different components of a catalytic reaction could be investigated separately and the nature and dynamics of the resulting photoproducts investigated in isolation. Experiments would then be performed on catalytic reaction mixtures, and insight into the speciation of the activated complex would be obtained by comparison to the reference spectrum of each component. These data would therefore enable the immediate fate of the manganese complex upon activation to be determined. Such insight is especially important considering the recent report by Larrosa and co-workers which demonstrated that fac-[MnBr(CO) 3 (NCMe) 2 ] is a viable catalyst for room-temperature Mn-catalyzed C−H bond functionalization. 19 This demonstrates the importance of complexes containing a "Mn(CO) 3 " moiety in these reactions: our lightinduced strategy allows for direct access to this structural unit.
The successful demonstration of this strategy is now reported. These experiments demonstrate that in all cases initial coordination of the solvent occurs, followed by substitution in an essentially statistical manner by the other components of the reaction. The subsequent fate of the complexes then depends on the nature of the N-donor ligand employed.
Our procedure to explore the chemistry of these systems is based on a TRIR spectroscopy experiment. Here solutions of the complexes are continuously flowed through an IR cell that is held in the path of overlapped pump and probe beams. The solution is continuously replenished, as the photochemistry described in this work is irreversible and hence fresh sample is   The subsequent changes to speciation are then followed by a probe pulse that interrogates the IR spectrum of the sample between ca. 1850 and 2100 cm −1 . The probe pulse arrives at a defined time following activation by the pump, and this interval between pulses is referred to as the pump−probe delay, t. The synchronization provided by TR M PS means that this delay is repeated every 10 μs, and therefore, for every pump pulse spectra are recorded at t, t + 10 μs, t + 20 μs, t + 30 μs, etc. for 990 μs. The resulting data are presented as difference spectra with negative peaks corresponding to material consumed upon photolysis (in all cases, these correspond to the ground-state IR spectrum of the appropriate complex 1) and positive peaks representing the newly formed photoproducts.
It is important to highlight that irradiation of complex 1a results in competitive formation of 3  In previous studies, the interaction of the light-activated complexes 1a and 1b with toluene solutions of PhC 2 H, 3, was reported. 6,12 In these cases, the initial formation of toluene complexes (e.g., 6a; Figure 1c) was followed by substitution by PhC 2 H to give alkyne complexes (e.g., 7a) and finally C−C bond formation to give the seven-membered metallacycles (e.g., 8a), which are a key intermediates in Mn-catalyzed reactions. 15 To understand the speciation of the precatalyst following activation via CO loss, the interaction between light-activated 1a and 2phenylpyridine in toluene solution was explored using TR M PS. The resulting spectra (Figure 2c) demonstrated that at short pump−probe delays (<1 ns) a single species was formed with positive bands at 2005 and 1909 (br) cm −1 . These were identical to the previously reported toluene complex fac-[Mn(C^N)-(CO) 3 (toluene)]. 7,12 The coordination geometry at the metal was confirmed by the appearance of the sharp high-energy band and broader feature at lower energy, indicating the pseudo-C 3v symmetry of the complex. 21 Over the course of ca. 500 ns (Figure 2e), the bands for 6a were replaced by three new features at 1996, 1904, and 1887 cm −1 , assigned to complex 9a, a process that obeyed pseudofirst-order kinetics. 22 On the basis of the shift in CO bands to lower energy, 9a was assigned as the N-bound 2-phenylpyridine complex fac-[Mn(C^N)(CO) 3 (κ 1 -(N)-2a)] (Figure 2a).
The experiment was repeated but with phenylacetylene added to the reaction in equimolar amounts to 2a. As before, the toluene complex 6a was the initially formed species, which was then replaced over the course of 100 ns by bands corresponding to both 9a and alkyne complex 7a (Figure 2d). The observed rate constants for the formation of 9a, (2.01 ± 0.39) × 10 7 s −1 , and 7a, (2.51 ± 0.77) × 10 7 s −1 , were similar, as should be the case for competitive (pseudo-)first-order reactions.
At longer pump−probe delays, the bands for alkyne complex 7a were observed to decrease in intensity and to be replaced by those for metallacycle 8a, formed by migratory insertion into the Mn−C bond. The rate constant, (2.01 ± 0.28) × 10 5 s −1 , is similar to those reported previously (Figure 2f). 6,12 The reaction was performed again, but the ratio of the reagents mirrored those used in a catalytic reaction (1a = 0.1 equiv, 2a = 2 equiv, 3 = 1 equiv). The resulting spectra (see the Supporting Information) demonstrated that the same course of events occurred, with the formation of 8a (through alkyne complex 7a) and 9a. However, in this case, a greater degree of 9a was formed, as might be expected because 2a was present in a greater concentration than in the previous experiment. Consequently, 7a and 8a were formed in smaller amounts, which entailed that the resulting kinetic information was of lower quality, but the C−C bond formation step was still observed.
Having identified the compounds formed upon activation of a precatalyst relevant to the Mn-catalyzed alkenylation reaction (Figure 1a), a similar series of experiments were performed to determine the fate of 1b, the precatalyst for the [4 + 2] annulation reaction (Figure 1b).
In the first instance, the interaction between light-activated 1b and 1,1-bis(4-methoxyphenyl)methanimine (2b) was explored. Irradiation of 1b in a toluene solution of 2b resulted in the expected photodissociation of a CO ligand and the initial formation of the toluene complex fac-[Mn(C^N)-(CO) 3 (toluene)] (6b) (Figure 3c). Over the course of ca. 3 μs, the bands for 6b were observed to decrease in intensity and to be replaced by three highly red-shifted bands with frequencies of 2000, 1907, and 1888 cm −1 , which were assigned to 9b. The band intensities again confirmed the formation of a complex with a pseudo-C 3v coordination geometry. The observed rate constant for the growth of these new features was statistically identical to that for the loss of the band for 6b (Figure 3), indicating that the new species formed from 6b. The structure of 9b was assigned as fac-[Mn(C^N)(CO) 3 (κ 1 -(N)-2b)] on the basis that the substantial red shift in the energies of the three carbonyl bands in 9b indicated that a strongly donating ligand had been incorporated into the coordination sphere of the metal and that the new transient bands did not correspond to those of the previously identified water complex fac-[Mn(C^N)-(CO) 3 (OH 2 )]. 8 Simultaneously with the formation of 9b a small band was observed at 2025 cm −1 , consistent with the formation of fac-[Mn(C^N)(CO) 3 (N 2 )] (10), 8 which then depleted at longer pump−probe delays. This indicates that although the binding of N 2 to the manganese was kinetically competitive with 2b, the imine was the thermodynamically preferred ligand.
To understand the speciation when precatalyst 1b is activated, phenylacetylene was then added to the sample to mirror the 1:1.5 ratio of 2b and 3 used in the catalytic reactions. Once again, toluene complex 6b was the initially formed photoproduct, but over the course of ca. 500 ns, bands corresponding to both 9b and the previously observed 12 alkyne complex fac-[Mn(C^N)(CO) 3 (η 2 -HC�CPh)] (7b) appeared.
The reaction mixture changed at longer pump−probe delays. Over the course of 16 μs the bands for alkyne complex 7b were observed to decrease in intensity, but no evidence for the previously observed product of alkyne migration into the Mn−C bond of the cyclomanganated imine complex, 8b, was obtained. Instead, the bands for 9b continued to grow in intensity (k = (3.11 ± 0.21) × 10 5 s −1 ) with essentially the same rate constant as for the loss of 7b (k = (3.47 ± 0.19) × 10 5 s −1 ). At long pump−probe delays, 9b was the only photoproduct observed. Therefore, under these conditions, no evidence of the C−C bond formation step through migratory insertion of the alkyne into the Mn−C bond was obtained. Instead, the coordinated alkyne was substituted by free 2b. Experiments performed with 1a and 2a under identical concentrations (see the Supporting Information) demonstrated that the migratory insertion Organometallics pubs.acs.org/Organometallics Article reaction was still observed, demonstrating that the differences observed in the case of 1b and 2b were due to the nature of the substrates used rather an artifact of the conditions. A series of calculations using density functional theory (DFT) were performed in order to understand the difference in behavior between the 2-phenylpyridine-and 1,1-bis(4methoxyphenyl)methanimine-based systems. Details of the methodology used are provided in the Supporting Information. A series of isodesmic reactions were calculated in order to model the change in Gibbs energy for the conversion of 6 to 7, 8, and 9 as a function of the nitrogen donor ligand (Figure 4). The resulting data showed a significant change between the two nitrogen donor ligands. In the case of the 2-phenylpyridine substrate, the substitution of the coordinated toluene ligand in 6a by either 2a or PhC 2 H was shown to be exergonic by 35 and 31 kJ mol −1 , respectively. As shown in our previous work, the complex arising from the subsequent alkyne insertion into the metal−carbon bond lies at lower energy than the corresponding alkyne complex. 6,12 Identical calculations on the 1,1-bis(4-methoxyphenyl)methanimine system provided insight into the difference between the two systems. In this case, the substitution of toluene from 6b by either 2b or PhC 2 H was shown to be exergonic by 80 and 36 kJ mol −1 , respectively. The complex arising from alkyne insertion, 8b, was still located at −121 kJ mol −1 relative to 6b, although this was not observed when experiments were performed in the presence of 2b.
It is therefore proposed that the difference in behavior between the two systems reflects the change in relative energy of the alkyne complexes 7 and the N-donor complexes 9. In the case of the imine-based substrate 2b, the resulting N-bound complex 9a is at a substantially lower energy than the alkyne complex 7b. Therefore, after the initial kinetically controlled competitive substitution of the toluene in 6b to give a mixture of 7b and 9b, a second ligand substitution occurs in which 2b replaces the coordinated alkyne to give 9b as the sole product of the reaction. At the concentrations employed, this is faster than the migratory insertion reaction, as 8b is the lowest-energy species calculated. Indeed, the observed rate constant for the formation of 9b from 7b, (5.1 ± 0.2) × 10 5 s −1 , is twice the reported first-order rate constant for the migratory insertion reaction 7b → 8b, (2.25 ± 0.16) × 10 5 s −1 . 12 In the case of the 2-phenylpyridine-based system, there is no energetic driving force for 2b to substitute the phenylacetylene in 9a: their calculated Gibbs energies are very similar. Hence, no ligand substitution would be expected (as observed experimentally), and migratory insertion occurs.

■ CONCLUSIONS
The results from this study have demonstrated that the formation of a solvent complex immediately follows the key step in precatalyst activation: ligand dissociation. This is presumably a kinetically controlled event, as the solvent is the dominant component of the reaction mixture. The toluene is a weakly bound ligand, and the second-order rate constant for this substitution is ca. 10 7 mol −1 dm 3 s −1 . It is therefore proposed that the near-diffusion-controlled substitution of the coordinated solvent is essentially governed by the relative concentrations of the two substrates in the reaction. In these aspects the photoproducts derived from both 1a and 1b behave in an essentially identical manner, but there is a difference in behavior at longer times between the interactions involving 2-phenylpyridine and 1,1-bis(4-methoxyphenyl)methanimine. In the former case, the alkyne complex undergoes the expected migratory insertion reaction to give 8a. In the latter, the alkyne is substituted by uncoordinated 2b, and 9b is the sole product from the reaction. The calculations indicate that this is an artifact of the different affinities of 2a and 2b for the manganese, which reflects the thermodynamic preference for the imine-based ligand to act as an effective N-donor ligand.
Although at this stage it is not possible to extrapolate these results to demonstrate exactly why the outcome of the Mnmediated reactions of 2a and 2b with alkynes give different outcomes (alkenylation versus annulation in Figure 1a and Figure 1b), the data do demonstrate that the nature of the substrate containing the heteroatom directing group plays a key role in determining catalyst speciation. These results also indicate that the formation of species such as 9 may represent off-cycle catalyst sinks, as loss of the N-bound heteroatom-based substrates is required prior to the coordination of the unsaturated substrate and should be considered as part of the mechanistic processes underpinning Mn(I)-catalyzed reactions.
Cartesian coordinates (XYZ) Additional TRIR spectra, details of computational chemistry methods, collated rate constants, and calculated energies (PDF) ■ AUTHOR INFORMATION