The photochemical alkylation and reduction of heteroarenes

The functionalization of heteroarenes has been integral to the structural diversification of medicinally active molecules such as quinolines, pyridines, and phenanthridines.


Introduction
Over the past decade, the rapidly increasing development of photochemical based organic transformations has been driven by research in photoredox catalysis. Transition-metal photocatalysts and organic dyes have granted chemists access to ever increasing alternatives to classical radical chemistry transformations as well as unconventional reactivity in the discovery of new organic transformations. 1 Implementation of photochemical processes is preferential to the classical, thermal methods of radical generation, as they oen eliminate the need for radical initiators, stoichiometric additives and harsh reaction conditions.
The functionalization of heteroaromatic scaffolds is an important topic in organic chemistry for its applications in the synthesis of biologically active molecules and other medicinally relevant studies such as structure-activity relationships. 2 The Minisci reaction has been famously important in the alkyl/aryl functionalization of electron-decient heteroaromatic scaffolds. 3 In this regard, a relatively nucleophilic radical (alkyl or aryl) is generated by an initiation process, where upon addition to an electron-decient heteroaromatic (usually protonated), can undergo rearomatization by an oxidation event, leading to the functionalized product. In the original report, 3 the silver mediated decarboxylation or carboxylic acids in presence of persulfates was described, however this reaction suffers from byproduct formation, low yields in some cases and the need for relatively harsh conditions (Fig. 1, eqn (1)). Since then, a variety of mild methodological advancements have been made using many functional groups. 4 One such functionalization that has been particularly challenging to organic chemists is the methylation of heteroaromatic scaffolds. Only a few striking protocols exist for this ambitious disconnection, demonstrating the need for further methodological development (eqn (2)). 5 The use of methanol as a precursor for methyl functionality also presents a signicant challenge in organic synthesis as few methodologies have been developed (mostly with transitionmetals) for this difficult disconnection. 6 In light of these reports, a mild and waste-limiting organic-based protocol employing MeOH would provide ideal methylation conditions. 7 Herein, we report the photochemical activation of protonated heteroarenes for their methylation in methanol (eqn (3)) as well as studies with a variety of other alcohols and ethers. The use of catalytic quantities of 2,4-diphenylquinoline for the methylation of heteroarenes that do not absorb in the UVA region or that degrade upon direct excitation is also disclosed. Additionally, we also report the discovery of an iPrOH mediated reduction of heteroarenes (eqn (4)), which to the best of our knowledge is the rst organic mediated photochemical protocol for the reduction of heteroarenes. Finally, mechanistic studies were performed to aid in the elucidation of the overall reaction mechanism is also disclosed.
absence of a photoredox catalyst. Furthermore, the use of butylor tosyl-functionalized sulfonyl chlorides also gave rise to the methylated product rather than the corresponding sulfonyl alkyl/aryl functionalities that may have derived from the addition or fragmentation of RSO 2 radicals. Subsequently, it was found that the methylation proceeded to give 2a (76% yield) in the presence of concentrated hydrochloric acid (c ¼ 2.0 M) under UVA LED irradiation (Table 1, entry 1). On one hand, a screening of various co-solvents was performed to test the viability of extending this methodology to a variety of alkanols. It was found that MeCN can be a good co-solvent, yielding 2a in 77% (entry 5). On the other hand, the optimization of HCl equivalents and reaction concentration gave 2a in 80% yield using HCl (5 equiv.) in MeOH (0.5 M) (entry 11). Interestingly, the byproduct 4a was isolated in 40% yield aer 60 h of UVA irradiation (entry 12). Starting from 1a or 2a gave the same product distribution over the 60 h experiment, indicating that the methylation reaction occurs at a faster rate relative to the onset of photochemical degradation of 2a. The reaction was also amenable to scale up (10 mmol), producing 2a in 65% yield (entry 13). Control experiments veried that the transformation required acid as well as UVA LED irradiation for product formation (entries 14 and 15). The use of acids such as TFA, H 2 SO 4 , and HOTf gave little to no conversion of the starting material whereas PTSA lead mostly to degradation (entries [16][17][18][19]. Finally, the reaction conditions were not hampered when  exposed to air atmosphere, indicating that the reaction likely proceeds via the excited singlet state (entry 20). Notably, exposing the reaction to oxygen atmosphere resulted in the hydroxymethylated product 3a in 59% yield (56% isolated, entry 21). In our hands, this reaction only converted 1a to 2a when using 2 UVA LEDs at $1 mm away from the reaction vessel, resulting in the generation of heat ($70 C). Using a temperature controlled setup, the optimized conditions gave no conversion aer 16 h at 20 C, showing the need for both light and heat for the successful transformation of 1a. A survey of the literature reveals that previous photo-mediated methylation reactions were described in low yield (<10%). These examples employed intense sources of irradiation, possibly leading to the degradation of the methylated heteroarene products. 8 Given the simplicity of this transformation, we surmised an investigation of reaction scope and mechanistic studies would be benecial to the understanding of our results.
Extending this methodology to other alcohol and ether coupling partners with lepidine was also evaluated (Table 2B). The reaction run in ethanol gave the corresponding ethylated 2k in 54% yield whereas the use of diethyl ether gave only 15% of the product. Methyl tert-butyl ether (MTBE) was found to give 2a in 34% yield. Interestingly, THF, known to be the best H-atom donors among ethers, 9 gave the corresponding addition and fragmentation product 2l in 88% yield. Tetrahydrofurfuryl alcohol gave 2m in 77% yield, showing regio-and chemoselectivity towards functionalization at the 5-position of the furan. 1,4-Dioxane gave 2n in 28% yield, where products 2k and 2n 0 were also isolated in 26% and 20% yields.
To our surprise, when using iPrOH with lepidine, the tetrahydroquinoline product 5a was observed in 31% yield (Table 2C). This is likely a reection of the reducing characteristics of the ketyl radical derived from iPrOH. 10 We then extended this light-enabled reduction to other heteroarenes such as methyl and adamantyl functionalized quinolines to produce the desired heterocycles 5b-e in moderate to excellent yields. Notably, the bromo functionality in 5c remained intact. 11 Using 410 nm LEDs, phenanthridine and acridine also gave reduced heterocycles 5f and 5g in 53% and 32% (60% brsm), respectively.
A systematic approach to deuterium labeling experiments was performed to help distinguish plausible mechanistic pathways (Table 3A). As expected, under fully deuterated conditions using CD 3 OD and DCl in D 2 O gave d 3 -2a in 73% yield (85% brsm, entry 1). When using CD 3 OH with HCl in H 2 O, the d 2 -2a in 78% yield (entry 2). Implementation of CH 3 OD and DCl in D 2 O provided the d-2a in 77% yield (entry 3). Interestingly, the conditions do not generate ratios of deuterium incorporation where scrambling of benzylic protons is observed in other methodology. 12 These results likely indicate that the nal product derives from an enamine intermediate that is not under equilibrium and is determined by the nature of the broadly . In addition, a mixture of 1a and d-1a (51 : 49), was treated under optimal conditions to give a KIE of 1.77 (entry 8, see ESI † for details). To compare the corresponding data for ethers, THF and d 8 -THF (1 : 1) were submitted to the optimal conditions, where a mixture of 2l and d 7 -2l were obtained in 54% yield as a 55.5 : 44.5 ratio, providing an observed KIE of 1.20 (entry 9).
A similar approach was taken for further understanding of the photochemical reduction of heteroarenes using iPrOH (Table 3B). Using d 8 -iPrOH with DCl in D 2 O for the light mediated reduction of lepidine, d 8 -5a was isolated in 63% yield (entry 1). Interestingly, when using d 7 -iPrOH with HCl in H 2 O, deuterium incorporation is not observed in 5a, isolated in 73% yield (entry 2). Finally, when using iPrOD with DCl in D 2 O (entry 3, 61% yield), d 8 -5a was obtained with a similar deuterium distribution as in entry 1. These results are indicative of the excited state protonated lepidine and the following reactive intermediates during formation of 5a do not undergoing HAT reactions with iPrOH. However, it is possible that HAT may occur on the nitrogen atom, where erosion of deuterium incorporation by the broadly exchanging nature of the N-H bond would occur during purication. In order to gain insights into the underlying mechanism of these transformations, we performed a series of mechanistic studies. UV-vis absorption studies conrmed lepidine only absorbs at the irradiation wavelengths of our UVA LED in the presence of acid ( Fig. 2A). In order to prove that the reaction is initiated by quenching of the lepidine singlet state by MeOH, we performed Stern-Volmer quenching studies. As seen in Fig. 2, the emission of lepidine in 2 M HCl in MeCN was found to be quenched by the addition of MeOH, with a Stern-Volmer constant (K SV ) of 1.07 M À1 . During our quenching studies, we also observed a slight redshi in the emission maxima of lepidine, which could be due to the change in the environmental conditions upon addition of MeOH. In good agreement, we also observed quenching of lepidine by MeOH when employing H 2 O as the solvent with no change in the emission maximum (see ESI †). Furthermore, the decreased quenching efficiency in H 2 O (0.078 M À1 ) correlate with our previous experimental observations which showed that H 2 O is detrimental to the reactivity of this system (Table 1, entry 2). However, we cannot rule out that a portion of the observed increase in quenching in MeCN is due to the change in the environmental conditions upon addition of MeOH. Clearly a change in the environmental conditions is not the sole cause of the quenched emission, as no quenching would be observed in H 2 O if this were the case. Based on the results of our deuterated studies which showed a KIE of 1.86, we propose that this quenching event proceeds through a protoncoupled electron-transfer (PCeT), forming a protonated lepidine intermediate and a hydroxymethyl radical. Considering the observed KIE of 1.86, it is proposed that this reaction proceeds via a sequential PCeT, with electron-transfer occurring rst, followed by proton-transfer. 13 Furthermore, it is unlikely that this quenching event results in the single-electron oxidation of MeOH to form a MeOH radical-cation, as this would not be expected to produce a KIE. It is also possible that this reaction could proceed through a hydrogen atom transfer (HAT) mechanism, however this is also unlikely as we would expect larger KIE values than those observed in this work if a HAT was involved in the rate determining step. 14 We also examined the steady-state quenching of lepidine's emission by THF, one of the ethers employed in this work ( Table  2, 2l). Once again, quenching of the lepidine emission was observed upon addition of THF (see ESI †), with a K SV of 1.08 M À1 . Due to the observation of minimal KIE in this system (1.25, Table 3A, entry 9), it is proposed that this quenching event proceeds solely through eT, and not through a PCeT event such as the system employing MeOH.
In order to determine if chain propagation is present in the underlying mechanism of these transformations, we decided to perform intermittent illumination studies. It is evident that many in the eld of photoredox catalysis understand that one can probe a photo-initiated chain reaction through the use of intermittent illumination. However, as the lifetime of most chains are in the sub-second timescale, current attempts to establish whether or not a photoredox transformation involves a chain by testing the effect of switching the light source on and off on the time scale of minutes are futile. Therefore, a more appropriate test would be the "rotating sector" method, which has been employed in the past to produce intermittent illumination on the appropriate timescale for typical chain reactions. 15 Recently, the Scaiano group presented an updated version of the "rotating sector" method, employing current LED technologies to pulse LEDs down to the nanosecond timescale, and by using this technique were able to successfully characterize multiple chain reactions. 16 Using this technique, we were able to demonstrate that the temporal prole of irradiation does not have any effect on the yield of methylated lepidine (Fig. 2D). This indicates the absence of a propagating chain in the underlying mechanism, hinting that a sacricial amount of lepidine is used as a photocatalyst to promote the photochemical methylation (vide infra).
With the data obtained for the methylation/alkylation of a variety of heteroarenes taken with the mechanistic data, it is proposed that upon protonation, the heteroarene (I) is excited to the singlet state (I*), where it can be quenched by MeOH to give a protonated radical intermediate (I + ) and a hydroxymethyl radical (Fig. 3A). This quenching event likely proceeds via a sequential PCeT type mechanism stabilized by hydrogen bonding between the heteroarene and methanol, based on the KIEs observed in our deuterated studies. This relatively nucleophilic hydroxymethyl radical can then react readily with an electrophilic protonated heteroarene (I), leading to intermediate (II). Intermediate (II) may then be reduced by intermediate (I + ), giving amino alcohol (III) and regenerating the ground state heterocycle (I, catalyst or as stoichiometric reactant). This may also proceed via PCeT based on the observed KIE in Table 3, entry 8. Intermediate III leads to intermediate IV by elimination of water, which aer tautomerization, gives the methylated product (2). Under reducing conditions with iPrOH (Fig. 3B), it is likely that the intermediate (I + ) becomes reduced by the incumbent ketyl radical derived from quenching with iPrOH. Intermediate V has several resonance forms under equilibrium, which undergo exchange reactions with the broadly exchanging portion of the solvent as observed with product d 8 -5a (Table 3B, entries 1 and 3). Upon further photochemical reduction with iPrOH, the reduced heteroarene 5 is realized. This mechanism of action is distinct from the photomediated reduction of haloarenes and selected arenes HAT mechanism in presence of iPrOH. 17 Finally, with the results and mechanistic data obtained in this study, we gathered that the methylation of heteroarenes that do not absorb in the UVA may be catalyzed by a heteroarene such as 2,4-diphenylquinoline with 410 nm irradiation (Table 4). Gratifyingly, the transformation using 2,6-di-tertbutylpyridine was found to proceed to full conversion in 24 hours using 2.0 mol% of 2,4-diphenylquinoline, yielding 71% of 2o, whereas less than 10% conversion was observed in the absence of 2,4-diphenylquinoline. Other heteroarenes that were found to either undergo degradation or react poorly in the UVA mediated methodology such as quinaldines were found to be improved in this catalytic process, providing the methylated products 2a-c in 57%, 71% and 62% yields, respectively. This highlights the advantage of employing a photocatalyst where irradiation wavelengths that avoid direct excitation of the reaction products can be employed. Finally, the photocatalytic methylation of 2,6-diphenylpyridine proceeded to give 2f in 90% yield. The facile photocatalytic methylation described demonstrates the high potential for this system to have broad applicability in the alkylation of a variety of heteroarenes and will be investigated further.

Conclusions
In summary, the photochemical alkylation of quinolines, pyridines, and phenanthridine from alcohols and ethers was described. Deuterium labelling studies showed the ability to control the formation of CD 3 , CD 2 H, and CDH 2 products, indicating the reaction likely proceeds through an enamine intermediate. We also reported the rst photochemical organicmediated reduction of heteroarenes using simply iPrOH and HCl. This study also provides mechanistic insights to a complex  Table 4 Catalytic methylation of heteroarenes a a Procedure: 1 (0.4 mmol, 1 equiv.), 2,4-diphenylquinoline (0.008 mmol, reaction pathway and the possibility of a broadly applicable catalytic system that will be disclosed in due course. ‡

Conflicts of interest
There are no conicts to declare.