C-C bond-forming reactions of ground-state aryl halides under reductive activation

Under basic conditions aryl halides can undergo SRN1 reactions, BHAS reactions and benzyne formations. Appropriate complex substrates afford an opportunity to study inherent selectivities. SRN1 reactions are usually favoured under photoactivated conditions, but this paper reports their success using ground-state and transition metal-free conditions. In benzene, the enolate salt 12, derived by deprotonation of diketopiperazine 11, behaves as an electron donor, and assists the initiation of the reactions, but in DMSO, it is not required. The outcomes are compared and contrasted with a recent photochemical study on similar substrates. A particular difference is the prevalence of hydride shuttle reactions under relatively mild thermal conditions.


Introduction
AreC and Areheteroatom bond formations have long been key reactions in organic synthesis. Pd salts and related complexes have played an important role in this field, but other metals, notably copper, are central to some of these coupling processes. The 2010 Nobel Prize in Chemistry 1 was awarded to Heck, 2 Negishi 3 and Suzuki 4 for their pioneering studies on the former; whereas the Sonogashira reaction 5 is one of a host of other AreC coupling reactions that are now indispensable to organic chemistry. Furthermore, Buchwald's and Hartwig's studies of the synthesis of Areheteroatom bonds 6 have become extremely important, adding significantly to the much earlier discovery of the Ullmann synthesis. 7 Alongside these pioneering reactions, transition metal-free couplings have emerged and are now witnessing spectacular growth. 8e10 With the increasing cost of transition metal catalysts, and the requirement for costly separation and recovery of heavy metals from waste-streams, the attractions of the transition metalfree couplings are clear. We consider below three types of reactions (Scheme 1) that can take place with aryl halides, all of which occur under basic conditions (other types of reaction that occur for restricted classes of aryl iodides, such as S N Ar, are not considered here): (i) the S RN 1 reactions developed by Bunnett and coworkers. 11 In this case, electron transfer to aryl halide 1 affords an aryl radical 3, which couples with an anionic nucleophile to form radical anion 4. This species transfers an electron to another molecule of aryl halide to form another aryl radical 3 and thereby propagates the chain reaction. In doing so, 4 is converted into product 5. This family of reactions has shown expanded scope in recent years, mirroring the greater mechanistic understanding of the processes. 12e15 It includes both carbonecarbon and carboneheteroatom bond formations. Rossi has been a pioneer of much of the recent development of the S RN 1 reaction. 13,14 His studies generally focus on reactions that are conducted under photoactivation conditions. 14 (ii) benzyne coupling reactions, an area of rapid recent developments. 10, 16 Base-induced elimination of hydrogen halide from a halobenzene 1 forms the benzyne 6, which is attacked by a nucleophile to form aryl anion 7. On protonation, this affords the product 5.
(iii) the BHAS (base-promoted homolytic aromatic substitution) reaction. 8,9 Here, an aryl radical 3 adds to an arene; the new radical 8 is deprotonated to form an electron-rich radical anion 9. Transfer of an electron from this species leads to the biaryl product 10; the electron is transferred to another molecule of aryl halide 1 to form a new aryl radical 3, continuing the chain.
For transition metal-free coupling chemistry to advance, complex substrates need to be studied, including those where different types of coupling reactions are in competition. Rossi and coworkers have conducted a recent study where aryl radicals have more than one option from the classes of reactions listed above. 13 The studies were conducted (i) under transition metal-free photoactivation conditions, or (ii) without photoactivation but in the presence of iron (II) salts and pinacolone. Our interest in coupling reactions that are conducted in the ground-state and under transition metal-free conditions attracted us to this area. We recently studied BHAS coupling reactions of arenes with aryl halides under more routine thermal conditions and using organic electron donors to initiate the reaction of the aryl halides. 9i,9m,9s, 17 In this paper, we now report on a series of competing reactions for a selection of substrates that are carried out under thermal reaction conditions.

Results and discussion
Recent reports of BHAS coupling reactions use KOtBu together with one of a wide variety of organic additives, to facilitate coupling of haloarenes with arenes under transition metal-free conditions. 8, 9 We have previously shown that one of the most successful additives at initiating the BHAS pathway was N,N 0 -dipropyldiketopiperazine (DKP) 11. 9m,9s The enolate anion, 12, derived by deprotonation of DKP 11 (Scheme 2), acts as an electron donor to the aryl halides in the initiation step. Most reactions reported in the literature that proceed via the BHAS mechanism are performed using benzene as the solvent, since benzene is the coupling partner in these reactions. In the few examples where aryl radicals couple to enolate anions via the S RN 1 pathway, the solvents used are DMSO or liquid ammonia or DMF. 14a, 15 Given that the polarity of the medium is expected to play an important role in facilitating electron transfer reactions, the reaction outcomes in both benzene and DMSO were investigated and the results are compared throughout our studies.
Our initial studies exposed substrates 13aec to these transition metal-free reaction conditions (Table 1). When 13a was stirred at 120 C for 1 h in the presence of KOtBu and DKP 11, and in either anhydrous benzene or anhydrous DMSO, two products were isolated. The major product in both benzene and DMSO was 5H-benzo [e]-pyrrolo[1,2-a]azepin-11(10H)-one 16, which was isolated in 55% and 79% yields, respectively, and the minor product isolated was 3acetyl-5H-pyrrolo[2,1-a]isoindole 22 in yields of 7% and 10%, respectively [ Table 1: entries 1 and 2]. These products were isolated in yields that compared favourably to previous experiments under light irradiation or when iron salts were used as a catalyst [ Table 1: entries lit 1 and lit 2], and with the same preference for the formation of 16 as major product. 13 Blank reactions were performed without the DKP additive 11, to observe how the enolate anion of DKP, 12,i n fluences the yields of the reaction. We note that the choice of solvent has a dramatic effect on the yields. A blank reaction in benzene, without the DKP additive 11, afforded much lower yields of 16 compared to when DKP was present [ Table 1: entry 3]. However, a blank reaction in DMSO showed the opposite trend, and 16 was achieved with higher yields of 90% when the DKP additive was omitted [ Table 1: entry 4]. The difference in DMSO may not be due purely to polarity effects, as evidence mounts of the dimsyl anion as an electron donor. 18 When using DMSO as a solvent (but not benzenedsee Table 1: entry 5) the reactions could be achieved at lower temperatures. When the reaction was performed at room temperature (RT) in DMSO moderate to high yields of cyclized products were isolated both with and without the DKP 11 additive, respectively [ Table 1: entries 6 and 7]. A reaction that was conducted in the dark provided confirmation that these reactions proceed via a ground-state SET pathway, with yields matching those carried out in the presence of ambient light [ Table 1: entry 8]. So, even at RT, photochemical assistance is not needed for these S RN 1 reactions. When LDA was used as the base in the reaction, only the product 22 was isolated in low yields; the yield of 22 was comparable with the yield of 22 obtained in the presence of KOtBu [ Table 1: entry 9]. To test the effect of the halogen substituent, the substrates 13b,c were also treated under transition metal-free reaction conditions. When 13b was stirred at 120 C for 1 h in the presence of KOtBu and DKP 11, in anhydrous benzene, 13b did not react [ The proposed mechanism involves initial deprotonation of the substrates and additives by KOtBu (Scheme 2). Under the basic conditions, 13aec will exist in part as enolate 14aec, and similarly the DKP additive 11 will be partly converted to its enolate anion 12.
The enolate anion 12 acts as an electron donor, and will intermolecularly donate a single electron to 14aec to afford the aryl radical 17. The aryl radical formed could undergo intramolecular cyclisation onto the enolate anion to form the seven-membered cyclic radical anion 18. This radical anion 18 is electron-rich and can propagate the radical chain mechanism by donating an electron, to either 14aec or 13aec, and thus yield the seven-membered cyclic product 16. The formation of the minor by-product 22 likely arises when an electron is donated intermolecularly to the substrate 13aec. SET (single electron transfer) to 13aec forms the aryl radical 19 which cyclises onto the pyrrole ring to form the intermediate radical 20. Deprotonation of 20 affords the radical anion 21 and, after the transfer of an electron, the product 22 is formed (Scheme 2).
It was demonstrated that in DMSO, and to a slight extent in benzene, the substrate 13a cyclises in the absence of the DKP additive 11 (Table 1: entries 3 and 4). The enolate anion of DKP is proposed to act as the electron donor, but the substrate also forms an electron-rich enolate species, 14a, in the presence of KOtBu. The relative effectiveness of the two potential electron donors, in this case the enolate anion of DKP 12 or the enolate anion present within 14a, is often an important aspect in S RN 1 chemistry. Although two enolates are present, they are not necessarily of equal strength as electron donors. S RN 1 reactions are frequently dependent on traces of a more active electron donor (here 12)t o initiate the radical chaindthis is known as 'entrainment '. 11,13,14,19 In non-polar solvents like benzene, where electron transfer from neutral species to form charged species is not facilitated, a stronger donor like 12 can play a powerful role in assisting the initiation of the chains. Polar solvents, like DMSO, favour SET reactions in which charged species are formed from neutral starting materials, and so, donors like 12 are not needed to initiate the reaction. 18 If the enolate anion within 14aec donated an electron intramolecularly to the LUMO in the haloaryl moiety, the diradical 15 would form upon the loss of an iodide anion. The diradical 15 would undergo radical coupling in the cyclisation to form product 16. This route does not involve a radical chain. When substrate 13a was treated with LDA, the lithium enolate salt analogous to 14a formed. Based on the experiments above with KOtBu as base, the expected product was 16. However only 22 and unreacted 13a were observed. In this case, we expect full deprotonation of the substrate and hence a different rationalization than above for formation of 22. Its formation may suggest that the tightly bound lithium enolate anions are less effective electron donors than the corresponding potassium enolates (differences between potassium salts and salts of lower alkali metals are pronounced in BHAS and other coupling reactions. 15,9v ) Intramolecular electron transfer from the lithium analogue of 14a would have produced the (metal-free) diradical 15 that should behave as described above. Intermolecular electron transfer would create the lithium analogue of radical 17 ; here, coupling of the aryl radical to the lithium enolate must again be harder than for the potassium enolate, whether for conformational or electronic reasons, so that coupling to the pyrrole occurs, finally on workup forming 22.
The substrate 23 should follow the same type of mechanism as 13aec (Scheme 2) in the formation of products 24 and 25. This substrate was important for mechanistic reasons. Our previous studies had shown that coupling of haloarenes to arenes can be initiated through formation of benzynes. 9i,9m Since this substrate cannot form a benzyne, it confirms that the current transformations are not dependent on a benzyne route. Compared to the cyclisation of 13a, lower yields of products were achieved in the cyclisation of 23. The low yields may reflect the increased steric hindrance around the aryl radical that would arise from the methoxy substituents on the benzene ring. 9m Alternatively, electron-rich substrate 23 should also have a higher LUMO, hence the initiation by electron transfer to the haloarene would be harder, and require more energy. Table 2 illustrates the powerful effects that solvent and temperature can have on the reaction pathways that are followed. Substrates 26a,b were synthesized and reacted under various conditions in the presence of KOtBu and with either DMSO or benzene as the solvent.
Under the thermal activation conditions in benzene solvent, 26a afforded three products in low yields: the 6-membered cyclised product, 4-acetylphenanthridine, 27, its dihydro derivative, 28 and the deiodinated but uncyclised product 29 [ Surprisingly, when 26a was exposed to the thermal electron transfer conditions in DMSO, a completely different product, 30, was isolated in moderate yield [ Table 2: entry 2]. The product structure was confirmed by X-ray crystallography (Fig. 1). The formation of 30 also occurs in similar yields in the absence of the additive 11 [  Table 2: entry 5] at RT generated a mixture of products, 27, 29 and 30 in low yields and a new product identified as 42 was obtained in moderate yields. Product 42 was only seen at RT, but its desaturation product, 30, was seen at higher temperature [ Table 2: entry 2]. When substrate 29, which is the non-halogenated analogue of 26a,b was subjected to the reaction conditions in DMSO, the starting material was recovered [ Table 2: compare entries 7 and 5]. This substrate was used to provide added information about the mechanism of formation of products from 26a (Scheme 3). Particularly interesting is the fact that by isolating pure 29 and subjecting it to these reaction conditions, this does not lead to isolation or detection of 30 or 42, implying a role for the halogen of 26a,b in the reactions of these substrates.
In these reactions, we recognize that either deprotonation of 26a,b of the N-H or of the C-H of the ketone can occur to form the enolate; in addition, an equilibrium is likely to exist between the neutral and deprotonated forms. Scheme 3A proposes a possible route for the formation of 27, starting with an electron transfer from the DKP enolate 12 to 26a,b. The resulting aryl radical, 32, cyclises onto the neighbouring aryl ring via the BHAS mechanism and generates the intermediate 33. The rate of cyclisation of radical 32 must be competitive with hydrogen atom transfer (H.A.T.); the  hydrogen abstraction will quench 32 to generate the observed product 29. Deprotonation of 33 yields the radical anion 34, which will form 35, by donating an electron to the starting substrate 26a, which goes on to propagate the radical chain. Deprotonation of the NeH proton of 35, followed by hydride loss forms the product 27, which was the only product reported under photoactivation. 13 Hydride loss has been observed from alkoxides under the conditions of the coupling reactions; 9m in this case, the aromaticity of the nitrogen-containing ring in the product 27 provides additional driving force. The proposal that a hydride is eliminated from 36 to form 27 is supported by the isolation of 28; this compound can form when 27 behaves as the hydride acceptor for another molecule of 36. When the reaction is performed in DMSO, the major product is either 30 or 42, depending on the temperature of the reaction. It is proposed that in DMSO the substrate 26a will be present to some extent in its enolate anionic form, 31 (Scheme 3B). Since the isolated yields of 30 were similar both with and without the DKP additive 11, we proposed that this enolate anion 31 is capable of undergoing intramolecular SET, whereby an electron is donated from the enolate anion moiety to its haloaryl moiety. This would yield the diradical species 37. An 8-membered ring could form by the combination of the radicals, however it appears that this process must be slower than an alternative process. One possibility is that a [1,4]-hydrogen atom transfer by the aryl radical would form intermediate 38. 20 Instead of radical combination between the two electrophilic species in 38, hydrogen atom transfer occurs to form a conjugated imine 39. In the presence of KOtBu the enolate anion formation will occur to give 40,6-endo-cyclisation would afford salt 41, which would yield 42 by proton transfer either during the reaction or on work up. On the other hand, exposure of 41 to heat could lead to loss of hydride, consistent with the conversion of 36 to 27 discussed earlier, and leads to 43. Tautomerism would then afford 30. In addition, trace amounts of 28 were also observed in DMSO, which proceeds via the same mechanism as proposed above (Scheme 3A). Recently, Long et al. published an alternative synthesis of analogues of 30 using TEMPO and KOtBu in DMSO, but our route must occur by a different mechanism. 21 The final substrate tested was substrate 44 (Scheme 4), which is analogous to 26a except that the nitrogen of the tether is methylated to favour selective deprotonation to form the enolate anion. Under UV irradiation conditions, this substrate led, in Rossi's hands, to the formation of the 8-membered ring 47 (78%), in preference to a 6-membered cyclisation, that would have been analogous to that seen in the formation of 27 from 26a. We wanted to probe the effect of this methylation under our ground-state conditions.
When 44 was treated with KOtBu and 11 in benzene at 120 C, tetracycle 45 was isolated as a novel compound, the structure of which was confirmed using X-ray single crystal structure determination (Fig. 2), and additionally 46 was isolated in low yield [Scheme 4: reaction A]. However when the reaction was performed in DMSO at RT, this tetracycle 45 was not observed [Scheme 4: reaction B]. The products isolated were 46, similar to when the reaction was performed in benzene, 47 (which was the only product observed when the reaction was performed using photoirradiation 13 ) and 48.
Possible mechanisms for forming the products are shown in Scheme 5. Electron transfer within enolate 49 affords the diradical 50. This diradical could directly close to form the 8-membered ring in 47. Alternatively, hydrogen atom transfer through a favoured 6centred transition state would give diradical 51, that cyclises rapidly to form ketone 46. Of course, more than one mechanistic pathway may have relevance. Intermolecular electron transfer to enolate 49 would yield radical anion 52. Coupling of the radical to the anion would give 53, which could transfer an electron (to another molecule of 44) to give product 47.
Two possible routes are shown (and one of them is discussed below) for the formation of tetracycle 45. We took purified 47 and converted this to 45 in the presence of KOtBu. The 5-membered ring product 45 was not observed in DMSO at room temperature, which suggests that the application of heat is required to convert the 8-membered product 47 to the cyclic product 45 (Scheme 5).
Formation of 48 was not observed in the photochemical study. This compound can arise by cleavage of the bond between the benzylic N and the benzylic C atoms. This cleavage can be triggered from the radical anions 56 and/or 57, which can form by electron transfer to either of the arene rings of 44. (The radical anions 56 and 57 are shown, where substituent X]I or H; in our experience, deiodination of arenes is an easier reaction than cleavage of benzylic CeN bonds, and so we suspect that deiodination to X]H occurs first). Such reactions have recently been observed using powerful organic electron donors, 22 but this is the first observation of such reactions under KOtBu-facilitated coupling conditions.

Conclusion
In conclusion, the present work reports the results of our studies on thermally initiated reactions of various substrates using the additive DKP 11 and KOtBu, in DMSO or benzene. Entrainment with enolate 12 is of benefit in the less polar benzene solvent, but generally not so in the more polar DMSO. Although S RN 1 reactions are normally conducted under photochemical activation, the substrates represented here show that activation also occurs efficiently when ground-state conditions are used. 15

Experimental section
The electron transfer reactions were carried out within a glove box (Innovative Technology Inc., U.S.A.) under nitrogen atmosphere, and performed in oven-dried or flame-dried apparatus. All the reagents were obtained from commercial suppliers and used without further purification unless stated otherwise. A B€ uchi rotary evaporator was used to concentrate the reaction mixtures. Column chromatography was performed using Prolabo 35e70 mm particle size silica gel 60 (200e400 mesh). Thin Layer Chromatography (TLC) was performed using Merck silica gel 60 F 254 pre-coated aluminium plates. Visualisation was achieved under UVP mineralight UVG-11 lamp and by using methanolic vanillin or phosphomolybdic acid to develop them. Melting points were determined using a GallenKamp 'Griffen Melting Point Apparatus'. 1 H NMR spectra were obtained at 400 MHz (Bruker AV400) or 500 MHz (Bruker AV500) and 13 C NMR spectra were obtained at 100 MHz (Bruker AV400) or 125 MHz (Bruker AV500) using broadband decoupled mode. Spectra were recorded in either deuterated chloroform (CDCl 3 ) or deuterated dimethyl sulfoxide (DMSO-d 6 ), depending on the solubility of the compounds. Chemical shifts are reported in parts per million (ppm) calibrated on the residual non-deuterated solvent signal, and the coupling constants, J, are reported in Hertz (Hz). The peak multiplicities are denoted using the following abbreviations: s, singlet; d, doublet; t, triplet; q, quartet; sx, sextet; m, multiplet; br s, broad singlet; dd, doublet of doublets; td, triplet of doublets. GCeMS data were recorded using an Agilent Technologies 7890A GC system coupled to a 5975C inert XL EI/CI MSD detector. Separation was performed using the DB5MS-UI column (30 mÂ0.25 mmÂ0.25 mm) at a temperature of 320 C, using helium as the carrier gas. Positive Chemical Ionisation (PCI þ ) was used with methane as the ionization gas and a voltage of 952.941 V. High-resolution mass spectrometry (HRMS) was performed at the University of Wales, Swansea, in the EPSRC National Mass Spectrometry Centre. Accurate mass was obtained using nanospray ionization (NSI) or atmospheric pressure chemical ionisation (APCI) with a LTQ Orbitrap XL mass spectrometer. Infra-Red spectra were recorded using Shimadzu FTIR Spectrophotometer (Model IRAffinity-1) with a MIRacle Single Reflection Horizontal ATR Accessory.

Experimental procedures and characterisation of substrates
In a round-bottomed flask with anhydrous dichloromethane (30 mL) under argon at 0 C, chloroacetyl chloride (4 mL, 50 mmol) and propylamine (8.64 mL, 105 mmol, 2.1 equiv) were simultaneously added dropwise, and the reaction mixture was stirred at 0 C for 15 min. The reaction mixture was diluted with diethyl ether (200 mL) and a solid precipitated out of solution. The reaction mixture was filtered and the solid was washed with diethyl ether. The filtered solution was concentrated in vacuo and again diluted with diethyl ether (200 mL) and filtered. The filtered solution was concentrated in vacuo to yield the crude product 2-chloro-N-propylacetamide (6.72 g, 99%) as a pale yellow oil which was used without further purification. In a round-bottomed flask, 2-chloro-N-propylacetamide (6.604 g, 48.7 mmol) was diluted with anhydrous tetrahydrofuran (30 mL). At 0 C, a suspension of sodium hydride in mineral oil (60%, 1.95 g, 48.7 mmol, 1 equiv) in anhydrous tetrahydrofuran (20 mL) was added dropwise to the reaction mixture via cannula. The reaction mixture was stirred at 0 C for 10 min, then at RT for 3.5 h. The reaction mixture was quenched by dropwise addition of water and diluted with diethyl ether (150 mL). The reaction mixture was dried over anhydrous sodium sulfate. The filtered solution was concentrated in vacuo and purified by column chromatography (30%e100% ethyl acetate in petroleum ether) to yield 1,4-dipropylpiperazine-2,5-dione 11 (0.696 g, 14%) as pale yellow crystals, and recovered 2chloro-N-propylacetamide (3.26 g, 49%).