Coumarin C−H Functionalization by Mn(I) Carbonyls: Mechanistic Insight by Ultra‐Fast IR Spectroscopic Analysis

Abstract Mn(I) C−H functionalization of coumarins provides a versatile and practical method for the rapid assembly of fused polycyclic pyridinium‐containing coumarins in a regioselective manner. The synthetic strategy enables application of bench‐stable organomanganese reagents in both photochemical‐ and thermal‐promoted reactions. The cyclomanganated intermediates, and global reaction system, provide an ideal testing ground for structural characterization of the active Mn(I) carbonyl‐containing species, including transient species observable by ultra‐fast time‐resolved spectroscopic methods. The thermodynamic reductive elimination product, solely encountered from reaction between alkynes and air‐stable organometallic cyclomanganated coumarins, has enabled characterization of a critical seven‐membered Mn(I) intermediate, detected by time‐resolved infrared spectroscopy, enabling the elucidation of the temporal profile of key steps in the reductive elimination pathway. Quantitative data are provided. Manganated polycyclic products are readily decomplexed by AgBF4, opening‐up an efficient route to the formation of π‐extended hybrid coumarin‐pyridinium compounds.


Introduction
CÀ H activation and functionalization of organic molecules can provide a convenient and efficient method for the direct reaction of ubiquitous CÀ H sites. Transition metals have been widely applied in the CÀ H bond functionalization of a myriad of O, S and N-containing heterocyclic compounds. [1]Less focus has been placed on the biologically and medicinally-relevant coumarin ring systems, and what has been achieved has been dominated by precious metals, for example Pd, [2,3] Ru, [4] Rh [5] and Ir. [6]The deployment of earth abundant metals for the functionalization of coumarins would represent an important step forward particularly from a from sustainability, environmental and metal-security perspectives, [7] with Satoh's recent work with precious and costly Rh highlighting the potential for diverse organic transformations. [5]Manganese, a highly abundant transition metal, has emerged as a distinctive costeffective source for directed [8] and non-directed [9] CÀ H bond functionalization reactions, exhibiting complementary and oftentimes differential reactivity to metals such as Pd. [10]There can be no stronger argument for the employment of an earth abundant metal for a synthetic transformation if it is able to dictate the reaction outcome.The Mn(I)-promoted CÀ H bond functionalization reactions of challenging heterocyclic substrates occur catalytically and stoichiometrically, as demonstrated by Ackermann, [11] Wang, [12] and Fairlamb and Lynam. [13]he Mn-catalyzed CÀ H bond functionalization reactions operate primarily via redox-neutral processes.Alternative pathways were exemplified in the reaction of a 2-pyrone derivative with [MnBr(CO) 5 ] to give either alkenylated 2-pyrone product 3 a' or pyridinium adduct 3 a, depending on the reaction conditions employed. [15]This process revealed important mechanistic details that was directly translatable to archetypical 2-aryl-pyridine CÀ H bond functionalization processes.We anticipated that the reductive coupling pathway could be harnessed into a general synthetic route to access highly functionalized π-conjugated chromeno-pyridines/pyridinium salts, such as 5 a' from coumarins such as 4-(7-NEt 2 ).Related products exhibit an eclectic array of biological and medicinal properties, [14] indeed coumarinpyridinium hybrid salts are active cholinesterase inhibitors. [15]rucially, we recognized that the coumarin ring system could confer stabilization of the transient intermediates formed en route to such products, enabling mechanistic details to be probed such as alkyne coordination, insertion and reductive elimination.

Results and Discussion
Ultra-fast time-resolved spectroscopy is a powerful tool to study these processes experimentally, [13a,c-f] and CO-photodissociation provides a useful pathway to activate the complexes to study their interactions with the solvent and substrates of the reaction.The central hypothesis that these manganese complexes may undergo light-induced CO dissociation was probed by time-dependent density functional theory (TD-DFT) calculations for novel cyclomanganated complex 6-(7-NEt 2 ) (which could be formed preparatively from 4-(7-NEt 2 ) by reaction with [BnMn(CO) 5 ], in a quantitative manner).A minimized structure for 6-(7-NEt 2 ) was computationally identified using DFT methods through optimization in Gaussian16 using the B3LYP global hybrid functional and flexible DGTZVP basis set, along with Grimme's dispersion-correction for improved long-range dispersion energies (GD3), and a conductor-like polarizable continuum model (CPCM) CH 2 Cl 2 solvent model.These calculations are in keeping with those which we have conducted on closely related cyclomanganated compounds. [16]For the subsequent TD-DFT calculations, we assessed several functionals (B3LYP, cam-B3LYP, BP86 and PBE0) employing the DGTZVP basis set, which allowed us to compare the experimental UV-vis spectrum for 6-(7-NEt 2 ) in CH 2 Cl 2 at 298 K with the computed UV-vis spectral data.Thus, the data set best showing a match with the experimental data was found by B3LYP/DGTZVP.We calculated 50 low-lying valence excited states.The TD-DFT predicted and experimental UV-vis spectra for 6-(7-NEt 2 ) are shown in Figure 1 (left-hand side).The latter shows a strong UV-vis absorption band λ max = 444 nm, with a slight shoulder at ca. 420 nm.A broader, but slightly higher energy calculated dominant band was determined by TD-DFT, the first lowest energy excited state being the HOMO 118 -LUMO 119 interaction with large oscillator strength (f = 0.7762) and of singlet character.The HOMO is primarily centered on the coumarin framework, connecting the electron-rich atom sites within the diethylamino-substituted coumarin system.The LUMO is found across the ligand and manganese(I) center.An axial CO ligand is weakened in the LUMO suggesting that structure 6-(7-NEt 2 ) could be excited by light to enable photodissociation in the presence of a suitable and reactive alkynyl substrate.
In order to investigate the ability of Mn-carbonyl complexes to functionalize the desired substrates, reaction of coumarin 4-(7-NEt 2 ) with phenylacetylene 2 a in the presence of 10 mol % [MnBr(CO) 5 ] (Scheme 1) was primarily found to not lead to the formation of alkenylated coumarins 5 a'.Analysis of the resulting reaction mixture indicated that 5 a-(7-NEt 2 ) as the only product, which is structurally related to 3 a (only trace amounts formed).The synthesis of 5 a-(7-NEt 2 ) was optimized so that it could also be prepared in 76 % yield from a reaction between [MnBn(CO) 5 ] in toluene at 95 °C giving cyclometalated 6-(7-NEt 2 ) in quantitative yield (Scheme 1).
It was anticipated that this reaction proceeded through an initial cyclomanganation reaction between be employed in this reaction giving the corresponding manganacycles 6-(X-R) in high yields (Scheme 2).
Consistent with the proposed mechanistic picture, reaction of 6-(7-NEt 2 ) with phenylacetylene 2 a, afforded 5 a-(7-NEt 2 ).As shown in Scheme 3, this method could be applied to a series of terminal and internal alkynes (2 a-h) to afford complexes 5(ah)-(7-NEt 2 ).The products were obtained in a highly selective manner, with the regiochemical outcome, with respect to the unsymmetrical alkyne, being corroborated by single crystal Xray diffraction (XRD).The formation of ferrocene-and pyridine-containing products (5 f-(7-NEt 2 ) and 5 g-(7-NEt 2 ), respectively) confirms the high tolerance of this protocol to potentially sensitive redox active and metal-coordinating moieties, respectively.
Attempts to extend this reaction to other cyclomanganated coumarin derivatives revealed the importance of the diethylamino substituent in 6-(7-NEt 2 ) in aiding the formation of 5 a-(7-NEt 2 ) under our standard reaction conditions.For example, when alterative substituents were introduced into the coumarin framework (Table 1 entries    coordination, [17] which is a finding that might be more broadly useful in Mn(I) CÀ H functionalization chemistry.Subsequent migratory insertion of the alkyne into the MnÀ C bond, and final reductive elimination, affords novel complexes 5 a-(X-R) (Table 1).As dissociation of CO from Mn(I) complexes such as 6 a-(X-R) can be promoted photochemically, [13] it was proposed that 5 a-(7-NEt 2 ) could form from 6-(7-NEt 2 ) and 2 a on exposure to light.This suggestion was supported by our TD-DFT calculations.A reaction between 6-(7-NEt 2 ) and 2 a was performed under irradiation (focussed LED, 355 nm) at room temperature in diethyl ether.This synthetic protocol afforded 5 a-(7-NEt 2 ) in 66 % yield, after 35 min of irradiation (see the Supporting Information for the reaction time course).
Analysis of the solid-state structures of complexes 5 a-(X-R), determined by single crystal X-ray diffraction, demonstrated that the manganese was η 4 -coordinated to the newly-formed fused ring system.This resulted in a deviation in anticipated sp 2 planarity of the fused 6-membered ring derived from the alkyne and 2-pyridyl fragments.This effect also underpins the observed upfield shifts in the 13 C{ 1 H} NMR spectrum of 5 a-(7-NEt 2 ), with the four dienyl carbon environments appearing at δ 108.4,102.8, 90.2 and 72.8 ppm.
Further experiments were performed to gain mechanistic insight into the steps which underpin the formation of 5 a-(7-NEt 2 ).Given that synthetic work had indicated that thermal (including TMNO-induced CO-loss) or photochemical loss (focused LED, 355 nm) of CO from 6-(7-NEt 2 ) was a key step in this reaction, the interaction between 6-(7-NEt 2 ) and 2 a was studied employing Time-Resolved Multiple-Probe Spectroscopy (TR M PS) with IR detection using the state-of-the-art LIFEtime facility at the Central Laser Facility (UK). [18]These experiments were conducted through irradiation of the reaction mixture with ultra-fast laser pulses (λ = 355 nm) to induce CO-loss.The nature and behavior of the resulting photoproducts were then studied by observing differences in the infrared spectrum between 1850-2100 cm À 1 , over pump-probe delays between 1 ps and 1 ms. [10]The resulting data are presented in Figure 2 as difference spectra, with negative peaks representing 6-(7-NEt 2 ) being consumed (bleached peaks) upon photolysis, and the positive peaks belonging to the new photoproducts.
Photoirradiation of 6-(7-NEt 2 ) and phenylacetylene 2 a in MeCN solution resulted in the detection of one photoproduct 6-(7-NEt 2 ) MeCN , from which no subsequent species were formed.Repeating the reaction with 6-(7-NEt 2 ) in neat 2 a showed that positive peaks appeared after 10 ps, which on the basis of our previous work, [13a,f] were assigned to arene-coordinated 6-(7-NEt 2 ) arene complex.Over the course of 50 ps the bands for 6-(7-NEt 2 ) arene decreased in intensity, being replaced by those for the π-bound η 2 -alkyne complex 6-(7-NEt 2 ) PhC�CH .The observed rate constant for this process was determined (k obs = (5.83� 0.44) × 10 10 s À 1 ).Remarkably, it was then possible to observe the subsequent CÀ C bond formation step corresponding to the migratory insertion of the alkyne into the MnÀ C bond.This resulted in the formation of 6-(7-NEt 2 ) -7-a in ca. 2 μs, which is evidenced by the red shift in the frequencies of the manganese(I) carbonyl stretches associated with loss of the πaccepting ligand to the 7-membered manganacycle. [10]The observed rate constant for the insertion process was determined (k obs = (5.62 � 1.91) × 10 5 s À 1 ).These data demonstrated that the reaction of alkyne and coumarin ligands is fast (< 10 μs), however, no evidence for any additional photoproducts was obtained for the remainder of the experiment (to a timescale of 1 ms), indicating that the reductive elimination reaction to form 5 a-(7-NEt 2 ) is significantly slower. [8c] 5 h-(7-NEt 2 ) was prepared in 82 % yield using the route in Scheme 4 to prove its viability.These data demonstrate that η 1 -alkynyl species, including 7 a and 7 h, are not active species in this reaction.13b] A one-pot experiment confirmed consumption of the coumarin 4-(7-NEt 2 ) in a reaction with [BnMn(CO) 5 ] to give 6-(7-NEt 2 ), as monitored by in situ IR using a Mettler-Toledo ReactIR instrument, fitted with a Si probe.The spectral changes could be easily monitored on the second timescale, which is a clean process (Figure 3).To support the mechanistic hypothesis, the steps leading up to the formation of 5 a-(7-NEt 2 ) were studied by computation using DFT calculations (Scheme 5, see Supporting Information for details of the methodology used).The spectroscopically observed arene-coordinated complex 6-(7-NEt 2 ) arene , formed through loss of a CO from 6-(7-NEt 2 ) and coordination of 2 a, was taken as the reference state for the calculations.Insertion of 2 a into the MnÀ C(coumarin) bond proceeds via a low energy transition state ( TS 6-(7-NEt 2 ) PhC�CH ) to give 6-(7-NEt 2 ) -7-a , a process corresponding to that observed by TR M PS see above.The DFT calculations also predict that the energetic span for the formation of 6-(7-NEt 2 )-iso from 6-(7-NEt 2 ) -7-a is much greater (78 kJ mol À 1 ) than for the alkyne insertion process through ( TS 6-(7-NEt 2 ) PhC�CH ) (30 kJ mol À 1 ), consistent with the former step being too slow to be observed by TR M PS (e. g., millisecond range).
With both the synthetic and mechanistic chemistry underpinning the formation of 6-(7-NEt 2 ) fully established, attempts were then made to remove the fused cationic ring system from the formally anionic tricarbonyl manganese.
Our first synthetic attempt involved treatment of 6-(7-NEt 2 ) with HCl in diethyl ether, which delivered 8 a as the tetrachloromanganate(II) salt, where the diethylamino group had been protonated (Scheme 6).Single crystal XRD analysis confirmed the structural connectivity of 8 a.
Treatment of 5 a-(H) under analogous conditions afforded the tetrachloromanganate(II) salt 9 a.An alternative reaction of 5 a-(H) with HBF 4 in diethyl ether afforded the tetrafluoroborate salt 10 a, which was confirmed by single crystal XRD analysis (Scheme 6).A simplified procedure enabled direct formation of compounds 10 a-13 a on reaction of 5 a-(X-R) with AgBF 4 . [19]This delivered the corresponding BF 4 salts of these cycloadducts (Table 2).
A literature survey revealed that there are scant reports on cyclorhenated structural variants, both in terms of the CÀ H  bond activation step and an assessment of the feasibility of subsequent migratory insertion/reductive elimination steps.A reaction of benzyl rhenium(I) pentacarbonyl with 4-(H) afforded cyclorhenated complex 6'-(H), using a higher reaction temperature of 110 °C (Scheme 2). [20]No product was detected from the reaction of 6'-(H) with phenylacetylene 2 a, under a variety of conditions (thermal-TMNO aided and photochemical).The result is in keeping with the reduced propensity for rhenium to act as an effective catalyst, highlighting the uniqueness of manganese in facilitating these transformations. [21,22]

Conclusion
A useful synthetic methodology for accessing hybrid coumarinpyridinium compounds has been devised using Mn(I) carbonyl chemistry.It was possible to selectively remove the functionalized coumarin from the coordination sphere of the manganese(I) center, providing facile and efficient access to structurally diverse and potentially high-value π-conjugated hybrid coumarin-pyridinium compounds in high yields.The coumarin scaffold was found suitable for forensic examination of the key steps underpinning CO loss, alkyne coordination and insertion, and subsequent reductive elimination at a manganese(I) carbonyl moiety.All steps were quantifiable by time-resolved IR spectroscopic analysis.Furthermore, we have detected and characterized a coumarin-based 7-membered manganacycle which acts as an integral intermediate towards the reductive elimination products characterized in this work 5 a-(X-R) and 5(a-h)-(7-NEt 2 ).Our results, taken together, show the potential for developing new synthetic methodologies in concert with physical organometallic chemistry-based measurements.Arguably, our approach provides a useful platform for directly comparing experimental results with theoretical data.

Experimental Section
General experimental details and instrumentation used are included in the Supporting Information document.Included below are generalized experimental procedures for the cyclometallation of the coumarin ligands 6-(X-R).

Cyclometalation of coumarin ligands 6-(X-R):
To a flame-dried Schlenk tube under N 2 , equipped with a magnetic stirrer bar, was added coumarin 4-(X-R) (1.0 equiv.)and benzyl manganese pentacarbonyl (1.0 equiv.),followed by dry toluene (50 mL mmol À 1 ).The solution was heated to 95 °C with stirring, which was left to continue stirring for a further 2.5 h.Upon completion, the reaction was cooled to room temperature and the mixture concentrated in vacuo to yield the cyclometalated coumarins.
Thermally-induced alkyne insertion and reductive elimination 5-(X-R): To a flame-dried Schlenk tube under N 2 , equipped with a magnetic stirrer bar, was added cyclometalated coumarin 6-(X-R) (1.0 equiv.) in dry Bu 2 O or Et 2 O (60 mL mmol À 1 ).To the solution, alkyne a-h (1.5 equiv.)and (trimethylamine N-oxide (1.0 equiv.)if required) were added.The solution was heated to 80 °C, with the aid of a water-filled cold-finger and solution left stirring for a further 18 h.Upon cooling to room temperature, the reaction mixture was diluted with dichloromethane (15 mL) and the solution concen-trated in vacuo (by rotavaporator) to give crude product.Crude product was dissolved in minimal dichloromethane and precipitated out with excess hexane (ca.10-fold.excess to dichloromethane).The precipitate was filtered and dried in vacuo (on a standard Schlenk line) to yield the reductive elimination products.

Figure 2 .
Figure 2. TR M PS-IR data for the reaction between 6-(7-NEt 2 ) and 2 a in neat alkyne 2 a.
Further experiments were conducted to shed light on the reaction mechanism.A one-pot reaction with coumarin 4-(7-NEt 2 ) in the presence of both an equivalent amount of [BnMn(CO) 5 ] and 2 a, delivered 5 a-(7-NEt 2 ) in quantitative yield (Scheme 4).This demonstrated that formation of 6-(7-NEt 2 ) was not competitive compared to the η 1 -alkynyl species [Mn-(CO) 5 (C�CPh)] 7 a, which was not detected under the reaction conditions.An analogous reaction using deuterated phenylacetylene [D 1 ]-2 a delivered [D 1 ]-5 a-(7-NEt 2 ) in quantitative yield (Scheme 5).The observation that [D 1 ]-5 a-(7-NEt 2 ) featured 98 % preservation of deuteration at the carbon atom derived from the terminal alkyne indicates that no [Mn(CO) 5 (C�CPh)] was formed from 2 a and [BnMn(CO) 5 ] under these reaction conditions.Initial CÀ H bond activation is therefore proposed to occur at the electrophilic CÀ H site within the coumarin ring.Consistent with this observation is the reaction of [Mn-(CO) 5 (C�CC 6 H 5 -4-Me) with 4-(7-NEt 2 ) which did not afford any 5 h-(7-NEt 2 ).

Table 1 .
Effect on yields due to coumarin functional group and in the presence of TMNO additive.

Table 2 .
Yields of BF 4 salts from reaction with AgBF 4 .