|Home | About | Journals | Submit | Contact Us | Français|
Herein we describe our efforts to elucidate the key mechanistic aspects of the previously reported enantioselective photochemical α-alkylation of aldehydes with electron-poor organic halides. The chemistry exploits the potential of chiral enamines, key organocatalytic intermediates in thermal asymmetric processes, to directly participate in the photoexcitation of substrates either by forming a photoactive electron donor–acceptor complex or by directly reaching an electronically excited state upon light absorption. These photochemical mechanisms generate radicals from closed-shell precursors under mild conditions. At the same time, the ground-state chiral enamines provide effective stereochemical control over the enantioselective radical-trapping process. We use a combination of conventional photophysical investigations, nuclear magnetic resonance spectroscopy, and kinetic studies to gain a better understanding of the factors governing these enantioselective photochemical catalytic processes. Measurements of the quantum yield reveal that a radical chain mechanism is operative, while reaction-profile analysis and rate-order assessment indicate the trapping of the carbon-centered radical by the enamine, to form the carbon–carbon bond, as rate-determining. Our kinetic studies unveil the existence of a delicate interplay between the light-triggered initiation step and the radical chain propagation manifold, both mediated by the chiral enamines.
Ground-state enamine chemistry has been extensively explored since the 1950s. Following pioneering studies by Gilbert Stork, organic chemists have exploited enamines’ nucleophilic character to trap electrophiles and develop useful two-electron polar processes.1 Successively, chiral enamines I, generated in situ upon condensation of aldehydes 1 with secondary amine catalysts, have been recognized as key intermediates of organocatalytic enantioselective reactions (Figure Figure11a).2,3 Single-electron oxidation of ground-state chiral enamines by a chemical oxidant has also been found to render 3π-electron radical cation intermediates amenable to a range of unique open-shell reaction manifolds (singly occupied molecular orbital (SOMO) activation, Figure Figure11b).4 Overall, the past 15 years have witnessed the extensive use of the ground-state reactivity of enamines for the stereoselective functionalization of carbonyl compounds.5
Recently, our research laboratories demonstrated that the synthetic potential of chiral enamines is not limited to the ground-state domain, but can be further expanded by exploiting their photochemical activity. We revealed the previously hidden ability of enamines to actively participate in the photoexcitation of substrates and trigger the formation of reactive open-shell species from organic halides.6 At the same time, ground-state chiral enamines can provide effective stereochemical control over the enantioselective radical-trapping process. This strategy, where stereoinduction and photoactivation merge in a sole chiral organocatalyst, enables light-driven enantioselective transformations that cannot be realized using the thermal reactivity of enamines. Specifically, we used this approach to develop the α-alkylation of aldehydes7 with electron-deficient benzyl and phenacyl bromides (Figure Figure11c)6a and bromomalonates 2c (Figure Figure11d).6c The reactions were conducted at ambient temperature using household compact fluorescence light (CFL) bulbs as the light source.
At first glance, both processes depicted in Figure Figure11c,d seem to be classical substitution reactions of enamines proceeding through an SN2 manifold. However, they do not proceed at all without light illumination. Crucial for reactivity was the ability of enamines to trigger the photochemical formation of radicals from the alkyl halides 2 under mild conditions. Despite the superficial similarities between the two chemical transformations, they profoundly diverge in the radical generation mechanism. The first strategy (Figure Figure11c) relied on the formation of photon-absorbing electron donor–acceptor (EDA) complexes,8 generated in the ground state upon association of the electron-rich enamine I with electron-deficient benzyl and phenacyl bromides. Visible light irradiation of the colored EDA complex II induced a single-electron transfer (SET), allowing access to the reactive open-shell intermediates. In the second approach (Figure Figure11d), we used the capability of the chiral enamine I to directly reach an electronically excited state (I*) upon light absorption and then to act as an effective photoinitiator. SET reduction of the bromomalonate 2c induced the formation of the carbon-centered radical.
In this paper, we detail how a combination of photophysical investigations, nuclear magnetic resonance (NMR) spectroscopy, kinetic studies, and quantum yield measurements revealed further mechanistic analogies and striking differences for these enamine-mediated photochemical enantioselective alkylations of aldehydes with electron-poor alkyl halides. From a broader perspective, these studies explain how it is possible to translate the effective tools governing the success of ground-state asymmetric enamine catalysis into the realm of photochemical reactivity,9 thus providing novel reactivity frameworks for conceiving light-driven enantioselective catalytic processes.10
Our recent studies6 established that enamines I can interact with visible light in two different ways, serving either as donors in photoactive EDA complex formation (Figure Figure11c) or as photoinitiators upon direct excitation (Figure Figure11d). As the prototypical reactions for mechanistic analysis, we selected the alkylations of butanal (1a) with 2,4-dinitrobenzyl bromide (2a; Figure Figure22a), phenacyl bromide (2b; Figure Figure22b), and diethyl bromomalonate (2c; Figure Figure22c), all promoted by the commercially available diarylprolinol silyl ether catalyst A(11) (20 mol %).12 The reactions with 2a and 2b are representative of the EDA complex activation strategy,6a,6b while the chemistry in Figure Figure22c is triggered by the direct photoexcitation of the enamine.6c For all the processes, and in accordance with the original reports, we confirmed that irradiation by a household 23 W CFL bulb was needed to achieve the alkylation products 3a–3c in high yield and enantioselectivity.13 The careful exclusion of light completely suppressed the reactions, confirming their photochemical nature. The inhibition of the reactivity was also observed under an aerobic atmosphere or in the presence of TEMPO (1 equiv), the latter experiment indicating a radical mechanism.
Along with these similarities, the light-triggered reactions in Figure Figure22 showed striking differences too. When the experiments were conducted under illumination by a 300 W xenon lamp equipped with a cutoff filter at 385 nm and a band-pass filter at 400 nm (irradiation at λ ≥ 385 nm and λ = 400 nm, respectively), the reactivity of the three processes remained unaltered. However, the use of a band-pass filter at 450 nm or a blue light-emitting diode (LED) (λmax at 450 nm) completely inhibited the reaction with diethyl bromomalonate (2c). In sharp contrast, the enamine-mediated alkylations with 2a and 2b were not affected. We decided to conduct spectroscopic investigations to rationalize the different light-wavelength/reactivity correlation profiles while elucidating the origins of the enamine’s photochemical activity.
Immediately after mixing a methyl tert-butyl ether (MTBE) solution of the enamine, generated in situ upon condensation of butanal (1a) (3 equiv) with 20 mol % catalyst A, with 2,4-dinitrobenzyl bromide (2a) (1 equiv), we observed that the achromatic solution turned to a marked yellow color (Figure Figure33a). This observation raised the question of how the color developed.
The appearance of strong color on bringing together two colorless organic compounds is not uncommon. In 1952, this phenomenon inspired Robert Mulliken to formulate the charge-transfer theory.8c According to this theory, the association of an electron-rich substrate with a low ionization potential (such as an enamine)14 with an electron-accepting molecule with a high electronic affinity15 (such as electron-deficient benzyl and phenacyl bromides) can bring about the formation of a new molecular aggregation in the ground state: the electron donor–acceptor complex. EDA complexes are characterized by physical properties that differ from those of the separated substrates. This is because new molecular orbitals form, emerging from the electronic coupling of the donor and acceptor frontier orbitals (highest occupied molecular orbital (HOMO)/lowest unoccupied molecular orbital (LUMO)). EDA formation is accompanied by the appearance of a new absorption band, the charge-transfer band (hνCT), associated with an intracomplex transfer of a single electron (SET) from the donor to the acceptor. In many cases, the energy of this transition lies within the visible range.16 This is what happened when the enamine, generated in situ upon condensation of catalyst A and 1a, was mixed with both 2,4-dinitrobenzyl bromide (2a) (Epred = −0.66 V vs Ag/Ag+ in CH3CN) and phenacyl bromide (2b) (Epred = −1.35 V vs Ag/Ag+ in CH3CN). Indeed, the optical absorption spectra showed a bathochromic displacement in the visible spectral region, where none of the substrates absorb (red lines, Figures Figures33a,b). The new absorption bands, which in the case of 2a can reach the green region of the visible range (550 nm), cannot be accounted for by the addition of the absorption of the separate compounds, which can barely absorb visible light.
To further examine the implication of the enamine in the formation of photoactive EDA complexes, we synthesized the enamine 4 (Epox = +0.60 V vs Ag/Ag+ in CH3CN), prepared by condensation of catalyst A and 2-phenylacetaldehyde17 in the presence of molecular sieves. Upon isolation, 4 was mixed with electron acceptors 2a and 2b (Figure Figure33c). Using Job’s method18 of continuous variations, we readily established a molar donor:acceptor ratio of 1:1 in solution for both colored EDA complexes IIa and IIb, respectively (details in section D of the Supporting Information). Concomitantly, an association constant (KEDA) of 11.56 ± 0.02 M–1 for the complex IIa and 4.9 ± 0.1 M–1 for IIb in MTBE was determined by spectrophotometric analysis using the Benesi–Hildebrand method.19
The light-wavelength/reactivity correlation for the photochemical alkylations of butanal with 2a and 2b (parts a and b, respectively, of Figure Figure22) can be rationalized on the basis of the photoactivity of the enamine-based EDA complexes IIa and IIb (their absorption spectra, which are similar to the EDA absorption in Figure Figure33a,b, are reported in Figure S6 in the Supporting Information). Absorption of low-energy photons, including visible light, can induce an electron transfer to occur, leading to the chiral ion pair III (Figure Figure33d). Critical to reaction development is the presence of the bromide anion within the radical anion partner in III. The bromide, acting as a suitable leaving group, triggers an irreversible fragmentation event20 rapid enough to compete with a possible back electron transfer (BET), which would unproductively restore the ground-state EDA complex II instead.21 This fragmentation productively renders two reactive radical intermediates (the electrophilic carbon-centered radical IV and the α-iminyl radical cation V) which can initiate synthetically useful transformations, i.e., the alkylation of aldehydes. The enamine-based EDA complex activation strategy thus provides ready access to open-shell reactive species under very mild conditions and without the need for any external photoredox catalyst.
The enantioselective photochemical alkylation of butanal (1a) with diethyl bromomalonate (2c) showed profoundly different behavior. In addition to the distinct effect the light frequency had on the reactivity (as discussed in Figure Figure22), we did not observe any color change in the solution, which remained achromatic during the reaction progression. The absence of any photoabsorbing ground-state EDA complex was further confirmed by the optical absorption spectrum of the reaction mixture (red line in Figure Figure44), which perfectly overlaid the absorption of the enamine, generated upon condensation of the catalyst A with 1a (green line in Figure Figure44). In a separate experiment, we observed that the addition of a large excess of 2c to a solution of enamines did not change the absorption spectra, further excluding any EDA association in the ground state (Figure S13 in the Supporting Information). Closer inspection of the absorption spectrum indicated that the only photoabsorbing compound at 400 nm (a wavelength suitable for triggering the reaction) was the enamine22 (green line in Figure Figure44, absorption band until 415 nm). This observation prompted us to evaluate the possibility that the direct photoexcitation of the enamine could trigger the radical generation from 2c. This mechanistic scenario was consonant with the experiment performed using a band-pass filter at 450 nm (a wavelength that could not be absorbed by the enamine), since a complete inhibition of the reaction was observed (Figure Figure22c). The implication of the enamine within the photochemical regime was unambiguously established by Stern–Volmer quenching studies. As detailed in our original study,6c we recorded the emission spectra of enamine 4 upon excitation at 365 nm. The excited state of 4 and its emission were effectively quenched by bromomalonate 2c (see section E2 in the Supporting Information for details).
These observations indicate that the photochemical activity of chiral enamines and their potential for light-induced radical generation are not limited to the formation of ground-state EDA complexes. As detailed in Figure Figure55, the enamine I, upon light absorption, can reach an electronically excited state (I*) and act as a photoinitiator, triggering the formation of the electron-deficient radical IVc through the reductive cleavage of the bromomalonate C–Br bond via an SET mechanism23 (Epred(2c) = −1.69 V vs Ag/Ag+ in CH3CN). The reduction potential of the excited enamine was estimated as <−2.0 V (vs Ag/Ag+ in CH3CN) on the basis of electrochemical and spectroscopic measurements (see section E3 in the Supporting Information for details).24 In analogy with the EDA complex activation (Figure Figure33d), here too the SET event leads to both an electrophilic radical, IV, and the α-iminyl radical cation V.
Photophysical investigations established that in situ generated chiral enamines can use two different photochemical mechanisms to provide open-shell species from organic halides 2a–2c while avoiding the need for any external photoredox catalyst. We then focused on the nonphotochemical steps inherent to the enantioselective alkylation of butanal (1a). As depicted in Figures Figures33d and and5,5, the enamine-mediated photochemical pathways bring about the formation of two radical species: the chiral radical cation V and the electrophilic radicals IV. A stereocontrolled radical–radical coupling of IV and V can be invoked to account for the formation of the new carbon–carbon bond and the α-carbonyl stereogenic center within the final products 3a–3c (Figure Figure66a). This mechanistic framework would require an enamine-mediated photochemical event for every molecule of product generated.
It must be noted, however, that many radical reactions generally proceed through self-propagating radical chain pathways.25 In chain processes, product formation occurs through propagation steps that convert the open-shell intermediate (originating from the substrate precursor) into the final product while regenerating the chain-propagating radical. Reactions will occur if the propagation sequence is rapid enough in comparison with possible termination pathways, and if there is a suitable mode of initiation (that is, effective radical formation from a closed-shell substrate). In our case (Figure Figure66b), a chain propagation sequence can be envisaged such that the nucleophilic ground-state enamine I would trap the photochemically generated electrophilic radical IV to form the α-amino radical VI. Since α-aminoalkyl radicals are known to be strong reducing agents,26VI would induce the reductive cleavage of the electron-poor alkyl bromide 2 through an outer-sphere SET process, thereby regenerating the radical IV while releasing the product 3 and the amino catalyst A (more mechanistic details are discussed in Figure Figure77). In this scenario, the enamine-based photochemical radical generation strategies, which afford radicals IV and V, would serve only to initiate a radical self-propagating chain process.
To help distinguish between the two mechanisms, we determined the quantum yield (Φ)27 of the model reactions, which defines the moles of product formed per moles of photons absorbed by the system.28 Using potassium ferrioxalate as the actinometer, we measured quantum yields of 25, 20, and 20 for the reactions in CH3CN29 with 2a, 2b, and 2c,30 respectively (λ = 450 nm for 2a and 2b and 400 nm for 2c). These results are consonant with a self-propagating radical chain mechanism as the main reaction pathways for the three enamine-mediated photochemical alkylations of butanal under study. The measured quantum yields (Φmeasured) refer to the overall reactions. As such, these values do not take into account any possible nonproductive energy-wasting processes,31 including parasitic quenching by energy or electron transfer as well as unimolecular decay processes, which do not lead to product formation but which affect the efficiency of photoinitiation. To better estimate the actual chain length of the reactions, we measured the quantum yield of the initiation step,32 determining a Φinitiation of 0.77, 0.68, and 0.11 for 2a, 2b, and 2c, respectively (λ = 450 nm for 2a and 2b and 400 nm for 2c, details in sections G2 and G4 of the Supporting Information). Taking these data into account, the actual chain lengths of the model reactions (Φestimated = Φmeasured/Φinitiation) are considerably longer, with a lower limit of 32, 29, and 182 for 2a, 2b, and 2c, respectively.
Figure Figure77 details the general mechanism proposed for the alkylation of butanal with 2,4-dinitrobenzyl bromide (2a), phenacyl bromide (2b), and diethyl bromomalonate (2c). They differ in the nature of the light-triggered initiation step, but are characterized by a similar propagation cycle in which the ground-state enamine I traps the photogenerated electrophilic radical IV. Overall, the mechanism exploits the dichotomous reactivity profile of enamines in the ground and excited states. The photochemical activity of the enamines, either by EDA complex activation or by direct excitation, generates radicals IV from the closed-shell intermediates 2a–2c (Figure Figure77a).33 This event, by feeding in radicals from outside the chain, serves as the initiation of self-propagating radical chains. The radical trap from the ground-state chiral enamine I forms the new carbon–carbon bond while forging the stereogenic center (Figure Figure77b). Considering the consolidated ability of catalyst A to infer high stereoselectivity in enamine-mediated polar reactions,11 it is no surprise that the addition of the radical IV to I proceeds in a stereocontrolled fashion. Two pathways are feasible for the propagation step (Figure Figure77c): the α-aminoalkyl radicals VI, resulting from the radical trap, can transfer an electron to the starting alkyl halides 2. This SET process regenerates the chain-propagating radical IV while giving the bromide–iminium ion pair VII, which eventually hydrolyzes to release the product 3 and the amino catalyst A. The outer-sphere SET process is facilitated by the formation of the stable bromide and iminium ions. Alternatively, an atom-transfer mechanism can be envisaged, where the α-aminoalkyl radical VI would abstract a bromine atom from 2, regenerating the radical IV while affording an unstable α-bromo amine adduct, VIII,34 which would eventually evolve to the iminium ion pair VII. This pathway would provide a rare example of enantioselective catalytic atom-transfer radical addition (ATRA),35 a historical methodology useful for functionalizing olefins with organic halides.
To discriminate between the possible propagation manifolds, we prepared and isolated the iminium ion IX (Figure Figure77d), derived from the condensation of pyrrolidine and isobutyraldehyde, which mimics the actual iminium ion intermediate VII involved in the catalytic cycle. VII could not be synthesized because of the steric hindrance of catalyst A hampering a facile condensation with the aldehydic product 3. Evaluating the redox properties of IX is pertinent since its electrochemical reduction provides access to an α-aminoalkyl radical of type VI, the key intermediate of the chain propagation. We measured by cyclic voltammetry a reduction potential (Epred of IX) of −0.95 V vs Ag/Ag+ in CH3CN (irreversible reduction to give the α-aminoalkyl radical X). This value means that the α-amino radical of type VI is incapable of reducing either 2b or 2c (Epred(2b) = −1.35 V vs Ag/Ag+ in CH3CN; Epred(2c) = −1.69 V vs Ag/Ag+ in CH3CN), indicating that a bromine-transfer mechanism is likely operative with phenacyl bromide and bromomalonate substrates. In contrast, an SET reduction is the most likely pathway when using 2a, since its potential (Epred(2a) = −0.66 V vs Ag/Ag+ in CH3CN) makes an SET reduction from intermediate VI feasible.
Several aspects of the mechanism proposed in Figure Figure77 deserve comment. The underlying radical chain pathway is not surprising when considering that the transformations closely resemble atom-transfer radical addition (ATRA) processes35 or a Kornblum–Russell SRN1-type alkylation.36 The SRN1 is a process through which nucleophilic substitution is achieved on aromatic and aliphatic compounds that bear a suitable leaving group and that do not react through polar nucleophilic mechanisms. This class of transformations is characterized by an innate chain mechanism involving electron-transfer steps with radical ions as intermediates. In some examples of SRN1-type reactions, electron-rich olefins, including enamines,37 efficiently trap electrophilic radicals. In addition, electron-poor benzyl37 bromides are suitable substrates for the SRN1 reaction manifold. On the other side, bromomalonates and phenacyl bromides34 are suitable substrates for ATRA processes, which classically proceed via radical chain mechanisms.35
Another aspect to consider is the central role of the chiral amino catalyst A. Although the process is characterized by an innate radical chain, the organic catalyst plays a direct role in product formation. Indeed, A is essential for the propagation mechanism since it transforms an inactive substrate (the aldehyde 1), which is unsuitable for participating in the radical chain, into the electron-rich chiral enamine I, a key intermediate of the propagation cycle. In addition, the enamine is directly involved in both the stereodefining event and the photochemical initiation. As for the initiation, the fate of the chiral α-iminyl radical cation V, emerging from the photoinduced SET to 2 (Figures Figures33d and and5),5), deserves further comment. Intermediate V is an unproductive species, since it lies outside of the chain propagation manifold which converts substrates into products. We have obtained evidence that V is an unstable intermediate which cannot be reduced back to the progenitor enamine I. Instead, the α-iminyl radical cation V collapses to give a variety of degradation products that, despite our efforts, have remained unidentified so far.38 Thus, the enamine I serves as a sacrificial initiator of the chain mechanism39 since, for any photoinduced SET event, a propagating radical IV is generated while a molecule of the chiral catalyst A is destroyed via decomposition of the intermediate V. By using both gas chromatography (GC-FID; FID = flame ionization detection) and NMR analyses,40 we established that the amount of catalyst A decreases constantly during the photochemical alkylation in correlation with the number of initiation events (further discussions in the following sections). The irreversible cyclic voltammogram of the preformed enamine 4 (Figure S16 in the Supporting Information) is also congruent with the proposed enamine degradation pathway.
With a clearer mechanistic picture in mind, we decided to perform kinetic studies to better understand the relative importance of the initiation step and the propagation cycle for the overall rate, while establishing the turnover-limiting step of the model photochemical catalytic alkylations. However, before this, we investigated whether the different photochemical pathways available to enamines for initiating the chain process (EDA complex formation vs direct photoexcitation) might have an influence on the enamine formation and its concentration in solution. This matters because the amount of enamine in solution has a direct effect on the kinetic profiles of the reactions, since the enamine is involved in both initiation and chain propagation (Figure Figure77a,b).
The catalytically active enamine intermediate I is generated via the reversible condensation of the chiral amino catalyst A with butanal (1a) (Figure Figure88a). This reversible process is characterized by an equilibrium constant (Kenamine = [I][H2O]/[1a][A]). As with all chemical equilibria, the system follows Le Châtelier’s principle. As a consequence, any perturbation of the equilibrium (as induced by a change in concentration, for example) will shift the position of equilibrium to the side that opposes the perturbation. As discussed above (Figure Figure33c), the formation of the enamine-based EDA complex is also an equilibrium, where KEDA identifies the association constant. For example, the EDA complex IIa (formed by the association of the preformed enamine 4 with 2,4-dinitrobenzyl bromide (2a)) has a KEDA of 11.6 M–1 in MTBE. This scenario suggests that the presence of acceptor 2a can alter the original state of equilibrium for enamine formation. In other words, it can directly influence the relative concentration of free catalyst A and enamine I in solution (Figure Figure88a).
To verify this possibility, we used 1H NMR spectroscopic analysis to investigate the equilibrium of enamine formation under the reaction conditions (Figure Figure88b). Upon mixing 0.3 mmol of 1a and 0.02 mmol of the amino catalyst A in 0.5 mL of anhydrous CD3CN, both enamine I and free catalyst A were detected in a ratio of 1.2:1. An equilibrium constant (Kenamine) of 0.155 ± 0.002 was determined (see section H1 in the Supporting Information for details). The addition of 0.1 mmol of 2a induced a shift in the position of the equilibrium toward the enamine I, as demonstrated by the 1.8:1 ratio of I and free catalyst A. This is congruent with the fact that the formation of the EDA complex, by sequestering I, shifts the dynamic equilibrium of enamine formation to the side that reduces the perturbation (in this case, the forward reaction).41 Since these studies were made in the absence of light, we then studied the effect of illumination on the dynamic equilibrium system (Figure Figure88c). We used a xenon lamp coupled with a monochromator, which, by bringing the light in close contact with the NMR tube through an optical fiber,42 allowed for the in situ illumination of the samples. When the EDA complex mixture, originally kept in the dark, was irradiated in situ in the NMR spectrometer (λ = 470 ± 5 nm, irradiance 28.8 mW/cm2), a large shift in the position of the enamine equilibrium was immediately observed (3.8:1 ratio of I to A after 30 s of irradiation). After 60 s of irradiation, the signals of the free catalyst A could no longer be detected, meaning that the system dramatically shifted toward the enamine I. This observation can be reconciled with the photochemical activity of the enamine-based EDA complex II, which, upon excitation, induces the irreversible formation of the electrophilic radical IV (upon fragmentation of the C–Br bond within the ion pair III; see Figure Figure33d) and the unstable α-iminyl radical cation V.40 These light-triggered events decrease the concentration of both the enamine and 2a, further favoring the forward reactions of the multiple equilibrium systems depicted in Figure Figure88c.
The importance of the irreversible events that follow the photoinduced SET is corroborated by a similar experiment where 2a was replaced by 2,4-dinitrotoluene (5) (Figure Figure88d). 5 can act as an acceptor partner in EDA complex formation with the enamine I (KEDA = 4.6 ± 0.1 M–1 in MTBE with enamine 4), but it cannot undergo an irreversible fragmentation, since it lacks a suitable leaving group (e.g., the bromine within 2a). In the dark, the addition of 1 equiv of 5 to a solution of catalyst A and butanal (1a) induced a displacement in the equilibrium of the enamine formation, changing the I:A ratio from 1.2:1 to 1.6:1. This is because an EDA complex, II, can be generated, which perturbs the equilibrium of enamine formation. In sharp contrast, illumination did not change the concentration of the enamine I to any extent. This observation is consonant with an unproductive photoinduced SET and a fast back electron transfer (BET) that, by restoring the ground-state EDA complex II, do not influence either the overall equilibrium of the system or the distribution of catalyst A, which is partitioned between the free state and the enamine I.
These experiments were then repeated with the bromomalonate 2c (results not shown in Figure Figure88). In this case, the equilibrium of the enamine formation (Kenamine = 0.155 as in Figure Figure88a) was not perturbed by the addition of 2c. This is because the mechanism of initiation is based on the direct photoexcitation of the enamine I and does not involve any preassociation with 2c. Thus, the presence of 2c does not influence the partitioning of the catalyst A between the free state and the enamine I.
We then performed kinetic studies to gain a better understanding of the factors governing the photochemical enamine-based alkylations of butanal (1a). In particular, we sought to assess whether the existence of two different initiation methods, but seemingly similar propagation cycles, would bring about distinct or analogous kinetic profiles. The amine-A-catalyzed alkylation of 1a with 2,4-dinitrobenzyl bromide (2a) was chosen as representative of the EDA complex activation strategy (Figure Figure99a),43 while the reaction with diethyl bromomalonate (2c) exploits the direct photoexcitation of the enamine (Figure Figure99b). Initial rate experiments were performed in acetonitrile as the solvent to avoid the precipitation of the lutidinium bromide, generated during the reaction.29 The progress of the two reactions was monitored by 1H NMR analysis using two different approaches (see section I in the Supporting Information for details). We used a xenon lamp with a band-pass filter at 450 nm (irradiance 4.7 mW/cm2) to illuminate the EDA-complex-mediated reaction with 2a (Figure Figure99a), while a cutoff filter at 385 nm (irradiation at λ ≥ 385 nm, irradiance 300 mW/cm2) was employed for the process with 2c (Figure Figure99b). This setup required an independent reaction to be performed for every data point at different times. The initial-rate kinetic studies were repeated using in situ 1H NMR spectroscopy to directly monitor the reaction progress.44 In this second case, we used a xenon lamp coupled with a monochromator, which allowed for the in situ illumination of the samples. The EDA complex-based reaction with 2a was irradiated at 470 nm (irradiance 28.8 mW/cm2), while 400 nm (irradiance 20.4 mW/cm2) was used for the alkylation chemistry with 2c. Both approaches gave similar and reproducible kinetic profiles.
Figure Figure99 details the results of our initial-rate kinetic investigations, performed across a range of concentrations for each reaction component. A first-order dependence on the catalyst A was inferred for both the EDA-complex-based process with 2a (Figure Figure99a) and the reaction with bromomalonate 2c (Figure Figure99b). However, striking discrepancies in rate orders were observed in the dependence on butanal (1a) and organic halides 2a and 2c. The EDA-complex-mediated alkylation showed a zeroth-order dependence on the 1a concentration and an unexpected negative fractional order in [2a]. In sharp contrast, the photochemical alkylation of 2c is characterized by a half-order dependence on both [1a] and [2c]. We also explored the effect of water on the kinetic profile of the two processes using both the independent measurement method and in situ NMR approach (details in Figures S31 and S39). No alteration of the kinetic profiles was observed after the addition of either 1 or 2 equiv of H2O. These results indicate that the iminium ion hydrolysis, which leads to the alkylation product 3 while liberating the catalyst A, is not turnover-limiting.
We then tried to reconcile the strikingly different kinetic behaviors of the two systems with our previous observations. The zeroth-order dependence on butanal (1a) for the EDA-complex-mediated alkylation with 2a implies that the enamine I, generated in situ upon condensation of A and 1a, is the resting state of catalyst A. This conclusion is consonant with the NMR spectroscopic studies reported in Figure Figure88b,c indicating that, under the reaction conditions—that is, when the EDA complex between the enamine I and 2a is formed and under illumination—the equilibrium position of the enamine formation is completely shifted toward the enamine I. This means that a negligible amount of catalyst A is available in its free state and, consequently, the concentration of 1a does not affect the formation of the reactive enamine catalytic intermediate. In sharp contrast, our NMR studies established that the equilibrium of the enamine formation is not perturbed by the addition of bromomalonate 2c. In the direct photoexcitation of the enamine I, the amine catalyst A is partitioned between the free state and the enamine intermediate I. Thus, a definitive resting state cannot be identified, with the catalyst concentration shared between different intermediates. This situation is congruent with the observed positive fractional order in [1a] (Figure Figure99b).
Concerning the reaction rate’s dependence on the alkyl halide 2, the negative fractional order in [2a] for the EDA-complex-driven process (Figure Figure1010a) deserves in-depth discussion. As previously mentioned, for any SET event taking place within the photoactive EDA complex (Figure Figure33d and initiation step in Figure Figure77), a propagating radical, IV, is generated while a molecule of the chiral catalyst A is destroyed via decomposition of the unstable α-iminyl radical cation V.40 To verify whether the disappearance of the catalyst was related to the number of initiation events, we followed the evolution of [A] over time across a range of concentrations of 2a, which is the acceptor partner in EDA complex formation. Since there is zeroth-order dependence on [1a] and due to the fact that we could not detect any trace of catalyst A in its free state by NMR analysis, we monitored the evolution of A in our experiments by determining the enamine concentration in solution.41 The initial-rate measurements in Figure Figure1010b suggest that the decrease in [A] correlates with the 2a concentration. If the rate of disappearance of A is proportional to [2a]n, then the data should fit eq 1. As a result, the kinetic data can be plotted as indicated in eq 2.
Equation 2 indicates that, for reactions with the same initial concentrations of amino catalyst A, plots of [A] versus [2a]nt should be superimposable.45Figure Figure1010c shows such a superimposition for three reactions that have comparable initial concentrations of A but different concentrations of 2a. In Figure Figure1010c, the overlay found for plots of [A] versus [2a]t (n = 1 in eq 2) establishes a first-order dependence on [2a] for the catalyst’s disappearance.
The unitary dependence was also observed using in situ NMR monitoring of the reaction progress (see sections F3 and I1 in the Supporting Information for details). Using this approach, we performed two sets of experiments under the same conditions, but using a different intensity of irradiation (λ = 470 nm for both sets of experiments, but an irradiance of 28.8 mW/cm2 vs 3.0 mW/cm2). In the latter set of experiments, a lower absolute rate of catalyst decomposition was determined, in consonance with a less effective initiation regime. This observation establishes a direct correlation between the disappearance of catalyst A and the number of photochemical initiation events, since both the concentration of 2a and the intensity of light influence the rate of degradation for catalyst A.32
We then wanted to measure the real effect of [2a] on the rate of alkylation leading to product 3a, discounting the effects of catalyst A degradation in the initiation regime. Considering the zeroth-order dependence on 1a, the rate equation should read as eq 3, with n being the rate order with respect to 2a, overlooking its effect on [A]. In eq 4, the product formation is normalized by [A], thus discounting the effect of the catalyst degradation.
The rate-order dependence on [2a] was calculated by plotting the data according to eq 5, derived from eq 4. The order (n) is obtained from the slope of the logarithmic plot displayed in Figure Figure1111a, which indicates a positive fractional-order dependence on [2a] (n ≈ 0.4), while c is a constant (c = −2.31) given by the x-intercept. Figure Figure1111b displays the fitting of the kinetic data to eq 4 for n = 0.4, showing a good overlay.
The rate law indicates a turnover-limiting step within the radical chain propagation cycle (see Figure Figure77 for the general mechanism). If the initiation step was rate-determining, a first-order dependence with respect to both EDA partners I and 2a would be expected instead. The first-order dependence on catalyst A (whose concentration is equal to the enamine I concentration) suggests that the rate-determining step is the trapping of the electrophilic carbon-centered radical IV from the ground-state chiral enamine I to form the new carbon–carbon bond. We would expect higher order dependence on [2a] if the turnover-limiting step were the SET process, which regenerates the chain-propagating radical IV from the α-aminoalkyl radicals VI.
In analogy with the preceding discussion, the rate-order assessment indicates that the rate-determining step is the enamine trapping the electrophilic radical IV, derived from 2c, to form the carbon–carbon bond (see Figure Figure77 for the general mechanism).
Notable, no significant degradation of catalyst A was observed during the alkylation with 2c within the time frame of interest for the initial-rate measurements using the method of independent experiments.46 When the time of irradiation of the photochemical alkylation with 2c was extended, the disappearance of catalyst A became significant. However, using the same 23 W CFL light source and considering the same time interval, the catalyst degradation was much higher in the alkylations of 2a and 2b than the alkylation of 2c (details in section F of the Supporting Information). This observation, along with the measured quantum yields of photoinitiation (Φinitiation = 0.77, 0.68, and 0.11 for 2a, 2b, and 2c, respectively), suggests that the direct excitation of the enamine I is a less effective radical generation strategy than the enamine-based EDA complex approach. This scenario can be rationalized on the basis of the bimolecular nature of the initiation mechanism with 2c, which requires the excited enamine to encounter 2c for an effective SET. These conditions make an unproductive relaxation of the excited intermediate, which restores the ground-state enamine, more likely. In contrast, the photochemistry underlying the processes with 2a and 2b is dominated by EDA complexes. These form in the ground state, holding together the two partners involved in the following photoinduced SET. In this case, the initiation mechanism is based on a more efficient unimolecular process.
In summary, we have used a combination of conventional photophysical investigations, NMR spectroscopy, and kinetic studies to elucidate the key mechanistic aspects of the enantioselective photochemical α-alkylation of aldehydes with electron-poor organic halides. Quantum yield measurements established that a radical chain propagation mechanism is operative, while reaction profile analysis and rate-order assessment indicated that the trapping of the carbon-centered radical by the enamine is the rate-determining event. Central to these processes is the unique and diverse reactivity of chiral enamines. Their photochemical activity, either by EDA complex activation or by direct excitation, generates radicals from the organic halides 2a–2c. This event, by feeding in radicals from outside the chain, serves as the initiation of self-propagating cycles. The enamine lies at the heart of the propagation cycle too, since it traps the radical to generate an intermediate (the α-amino radical VI) which is key for sustaining the chain sequence. We also uncovered how enamine formation and its concentration in solution are directly influenced by the different photochemical pathways available to enamines for initiating the chain process (EDA complex formation vs direct photoexcitation). Overall, the kinetic and spectroscopic investigations allowed us to understand the delicate interplay between the light-triggered initiation step and the radical propagation manifold, suggesting that this approach can be generally applied to the mechanistic elucidation of chain processes. From a broader perspective, this study demonstrates that the synthetic potential of chiral enamines is not limited to the ground-state domain, but can be further expanded by exploiting their photochemical activity, providing novel reactivity frameworks for conceiving light-driven enantioselective catalytic processes.
Financial support was provided by the ICIQ Foundation, MINECO (Project CTQ2013-45938-P and Severo Ochoa Excellence Accreditation 2014-2018, SEV-2013-0319), the AGAUR (Grant 2014 SGR 1059), and the European Research Council (Grant ERC 278541—ORGA-NAUT). A.B. is grateful to the MECD for an FPU fellowship (ref FPU13/02402). We are indebted to the team of the Research Support Area at ICIQ. We thank Dr. Elena Arceo and Dr. Mattia Silvi for preliminary investigations and useful discussions.
The authors declare no competing financial interest.