Discovery and Development of Organic Super-Electron-Donors

Based on simple ideas of electron-rich alkenes, exemplified by tetrakis(dimethylamino)ethene, TDAE, and on additional driving force associated with aromatization, families of very powerful neutral organic super-electron-donors (SEDs) have been developed. In the ground state, they carry out metal-free reductions of a range of functional groups. Iodoarenes are reduced either to aryl radicals or, with stronger donors, to aryl anions. Reduction to aryl radicals allows the initiation of very efficient transition-metal-free coupling of haloarenes to arenes. The donors also reduce alkyl halides, arenesulfonamides, triflates, and triflamdes, Weinreb amides, and acyloin derivatives. Under photoactivation at 365 nm, they are even more powerful and reductively cleave aryl chlorides. They reduce unactivated benzenes to the corresponding radical anions and display original selectivities in preferentially reducing benzenes over malonates or cyanoacetates. Additionally, they reductively cleave ArC–X, ArX–C (X = N or O) and ArC–C bonds, provided that the two resulting fragments are somewhat stabilized.

R ecently, families of highly reactive organic reducing agents, the "super-electron-donors", have been discovered and developed, based on very simple molecular design. 1 This Perspective charts their emergence, their preparation, and their applications to date. In line with the personal nature of Perspectives, the article surveys the particular role that my research group has enjoyed in these developments. Aside from the intellectual challenge of designing such reagents and uncovering their reactivity, these compounds may play important roles in cases where contamination of products with traces of redox-active transition metals needs to be avoided, and they may provide economic alternatives to metalbased reagents. 2, 3 Impressive early work on neutral organic reducing agents arose with the discovery of tetrakis(dimethylamino)ethene (TDAE, 1) in industry. 4 This compound showed its ability to act as a good reducing agent (Scheme 1) by reducing electronpoor perfluoro substrates, such as 2. Here, a likely pathway is that 2 accepts an electron to become a radical anion that fragments with loss of a fluoride anion. 5 The remaining radical is then easily reduced to the corresponding anion, leading to loss of the second fluoride ion in forming 3. Concomitantly, TDAE is oxidized to its radical cation 4 and/or its dication 5, where extensive delocalization of charge and/or radical character are made possible by the nitrogen heteroatoms. The development of this tetraazaalkene 1 as a reagent for a broader range of organic chemistry followed later (see below), but its essential skeletal characteristic, an electron-rich alkene, acts as the blueprint for the host of other organic electron donors now available. Some 20 years later, a sulfur analogue of this tetraazaalkene, tetrathiafulvalene 6, was prepared by Wudl and co-workers 6 and heralded the birth of organic electronics; research on this and related molecules has increased exponentially ever since.
In the 1990s, we were interested in developing TTF as a reagent for synthetic chemistry. Researchers in Latvia 7 had announced that tetrasubstituted derivatives of TTF reacted with arenediazonium salts through single-electron transfer and liberation of nitrogen, although no isolation of the organic products derived from the arenediazonium unit was reported, and indeed, when we started we were not aware of their complementary work, but we noted that nitrogen gas was liberated when TTF itself and arenediazonium salts were mixed at room temperature. This did not happen when simple dialkyl sulfides were mixed with arenediazonium salts, and so this was consistent with an electron-transfer reaction that was particular to TTF and that should lead to formation of aryl radicals. Beckwith and others had studied the trapping of aryl radicals through 5-exo-trig cyclization onto alkenes, 8 and so we probed for the aryl radicals in this way using arenediazonium salts 7 (Scheme 2). Electron transfer followed by loss of dinitrogen afforded aryl radical 9. Cyclization to afford radical 10 was followed by trapping of the radical cation of TTF (11) to give the sulfonium salt 12, marking the end of the radical steps. Loss of TTF was encouraged by neighboring group participation by the aryloxy moiety. Intermediate 13 was not detected but underwent rapid attack (i) by water present in the acetone solvent to afford alcohol products 14, (ii) by methanol as solvent to afford methyl ether 15, or (iii) by acetonitrile as solvent to afford a nitrilium salt in a Ritter process, which was hydrated to amide 16 on workup. This type of ionic/polar termination to a radical process was novel, and we termed these reactions "radical-polar crossover reactions". 9 This type of crossover is widely seen in organic chemistry but with different reagents and reactions, and so this title has since been adopted for a much wider variety of examples in the current literature. An unusual feature was that the intermediate sulfonium salts 12 did not undergo solvolysis when they were attached to primary carbon atoms (i.e., R = H) but did solvolyse in secondary and  tertiary substrates. This work with TTF and internal trapping with nucleophiles developed rapidly, leading ultimately to a synthesis of (±)-aspidospermidine (19). 10 At the end of that campaign, we reflected on the synthesis of aspidospermidine; we had not been able to use aryl halides as starting materials, since they did not react with TTF, but instead needed the much more easily reducible arenediazonium salts. Looking at the potential scale ( Figure 1) shows the magnitude of the problem. TTF and its radical cation are in equilibrium at 0.3 V. It makes sense that it can reduce arenediazonium salts for which the reduction potential is approximately E p = 0 V. 11 However, iodobenzenes have much more negative reduction potentials E 0 = −2.2 V, 12 and so their reduction is a daunting task. If a neutral organic molecule could be found that would reduce iodobenzene, we resolved to call it a "super-electron-donor" (SED). If such molecules could be made, they might address a number of other challenges too (Scheme 3), such as formation of aryl anions. This should be more difficult than the analogous formation of aryl Grignard reagents or aryllithium reagents, since those compounds feature a polarized carbon−metal bond rather than a naked carbanion. In addition, major challenges for these new reducing agents would be the reductive cleavage of arenesulfonamides, reductive cleavage of geminal bis-sulfones, and reductions of arenes, reactions that had until then been the preserve of highly reactive metals. These challenges led our thinking about whether such reactions might be achievable by neutral organic electron donors.
To understand how to design such strong neutral organic donors, we needed to learn two lessons. The first of these came from some earlier failed experiments. Thus, TTF 6 reduced arenediazonium salts, but under the same conditions, dibenzoTTF 20 did not (Scheme 4) (see, however, ref 7), and we attributed this difference to the different driving forces for aromatization in the two cases. The TTF radical cation 21, the product of single-electron transfer from TTF, contains an aromatic dithiolium ring; oxidation of the dibenzo derivative to 22 also affords a new 5-membered aromatic ring, but this ring is fused onto a pre-existing benzene ring, and this leads to less driving force for the oxidation of the fused molecule 20. Thus, the extent of the aromatic driving force in organic donor molecules is important.
The other lesson learned was that nitrogen plays a much more helpful role than sulfur in similar compounds. It can stabilize an adjacent carbocation better than sulfur because of the better overlap between similarly sized carbon and nitrogen orbitals. This plays through into aromatic stabilization energy also, where more effective overlap of an aromatic sextet that incorporates nitrogen can be expected than for the corresponding sulfur case. TTF 6 is a relatively weak electron donor, and to increase its strength, substitution of sulfur by nitrogen is required. Initial efforts with diazadithiafulvalenes, e.g., 23, showed that they were not sufficiently strong donors to react with iodoarenes. 13 This brought us to focus on the reactivity of TDAE (1), (CH 3 CN, E 1 1/2 −0.78 V, E 2 1/2 −0.61 V in MeCN), the discovery of which was mentioned earlier; here, four nitrogen atoms stabilize the loss of one or two electrons, and so this can be expected to be a very good electron donor. As an organic reagent, this has been extensively developed 14−18 and, inter alia, activates benzylic halides, e.g., 24, converting them into benzylic anions that can be used to attack carbonyl electrophiles, specifically aldehydes and ketones to give alcohol products 25 in this case (Scheme 5). 14a,c However, we were unable to activate aryl iodides with this donor. Despite this, we took on board the beneficial role of an alkene substituted by four nitrogens, as well as the importance of developing aromaticity in the transition states of the electron-transfer steps, and identified the benzimidazolederived compound 29 that had previously been prepared in order to test its redox potential 19 but that had not been used in synthetic transformations previously. The preparation of this compound is shown below (Scheme 6). N-Methylbenzimidazole is treated with diiodopropane, and the resulting disalt 26 is treated with base to afford the tetraazafulvalene 29.
This compound is a beautiful vibrant yellow solid (any trace of orange or red indicates oxidation), and its oxidation potential (CH 3 CN, E 1 1/2 −0.82 V, E 2 1/2 −0.76 V vs SCE) shows that it is a relatively strong electron donor; as for TDAE, it reacts spontaneously in air. Treating both aryl iodides, e.g., 30, and alkyl iodides, e.g., 33, led to formation of the corresponding aryl and alkyl radicals, as indicated by high-yielding cyclizations shown in Scheme 6. 20 Considering how similar its first oxidation potential is to that of TDAE (1), it had remarkably different reactivity toward iodoarenes. This may relate to better π-stacking of 29 with the iodoarene, giving its reactions a kinetic advantage over those of 1, since we learned later that our polycylic donors have a special affinity for arene substrates (see below).
This was the first time that reductive deiodination of an iodobenzene had been achieved with a neutral organic electron donor, acting in its ground state. A question arose about the source of the abstracted hydrogen atom in product 32. When the reaction with 30 was repeated, replacing DMF by DMF-d 7 and excluding the toluene cosolvent, no labeling of the product was seen, suggesting that abstraction had occurred from the donor 29 or from its oxidized radical cation or dication forms, following electron transfer.
Taking on board the lessons with respect to aromaticity, we then set our sights on the corresponding imidazole-derived donor, 35 (Scheme 7), which should be a stronger donor. However, Taton and Chen had already shown that this compound was not accessible. 21 Their efforts to isolate this compound, following deprotonation of the disalt 36, led to formation of a bis-carbene 39. This illustrated that the product of the first deprotonation, more routinely represented as the carbene 38, rather than the ylide 37, did not undergo a rapid cyclization onto the remaining imidazolium salt but instead underwent a second deprotonation to afford 39. Constraining the two imidazole-derived rings by a second trimethylene strap led to synthesis of 43. (Macrocycle 44 was produced simultaneously and found separate use in the presence of base as a complexing agent for metals. 22 ) Disalt 43 gave a faster cyclization following a deprotonation; the resulting monosalt was deprotonated again to form the doubly bridged donor 45. 21,23 Donor 45 could be used in situ if its preparation was conducted in DMF or it could be isolated pure if the preparation was carried out in liquid ammonia. After evaporation of the ammonia, the solid residue is extracted with diethyl ether. Evaporation of the ether provides pure 45 as a yellow organic powder.
The reactivity of this donor was then tested against the same substrate (30) that had reacted with donor 29 (Scheme 6). 23 This time, the deiodination again went smoothly, but instead of isolating the cyclized product that had previously arisen from radical cyclization, this time an uncyclized product 31 was almost the exclusive product. Since the aryl radical cyclizes rapidly, this meant that an alternative intermediate was formed. Our proposal was that an aryl anion formed, where the counterion would likely be the radical cation or the dication of the donor. As previously mentioned, this would be very different from forming a Grignard or organolithium species, since this new anion would not be stabilized by bonding to a metal. To probe for the formation of an aryl anion, substrate 48 was prepared where an aryl anion should cyclize rapidly onto an ester. When the experiment was performed, the indanone 50 (51%) was isolated, together with the deiodinated uncyclized product 49 (21%). Although the latter might have arisen from aryl radical formation followed by hydrogen atom abstraction, a separate reaction of this donor with substrate 30, which probes for aryl radical intermediates, had given almost no cyclized product, leading us to conclude that in the reaction with substrate 48, product 49 is more likely to have arisen from an aryl anion carrying out deprotonation of the radical cation 46 of the donor or the corresponding dication 47. Here, the sp 2 C−H protons are likely to be relatively acidic. 24 Scheme 5. Reductive Activation of 24 Leads to Nucleophilic Attack on Aldehydes

Scheme 6. Preparation and Reactivity of Donor 29
The Journal of Organic Chemistry Testing the reactivity of this donor with a range of organic substrates was then undertaken. Aryl iodides were reduced to arenes. Moving to deprotection of arenesulfonamides like 51, the arenesulfonyl unit was the site of the LUMO. Fragmentation to an amine-related product and to a sulfinate product would depend on the ease of fragmentation, which in turn would depend on the stability of the leaving group.
Resonance-stabilized leaving groups should afford easier cleavage, and this was indeed the case as seen for substrate 51. However, no cleavage was seen for the corresponding N,Ndialkylarenesulfonamides (but see below for later successful cleavages). Bis-sulfones, e.g., 54, were reduced to a sulfonestabilized anion and a sulfinate salt. On workup, these were protonated to sulfone 55 and sulfinic acid 56. For the especially favorable monosulfone, 57, desulfonation was also successful. 25 In this case, fragmentation of the radical anion of 57 should afford a benzenesulfinate anion and a substituted cinnamyl radical. The delocalization available to this radical was crucial for its formation by fragmentation, as when the cinnamyl double bond was not present; i.e., in substrate 59, no fragmentation was seen.
At this time, a bottleneck for our work was the synthesis of the "doubly bridged" (i.e., with two trimethylene bridges) donor 45. Its synthesis required a separation of disalt 43 from the macrocylic tetrasalt 44 that was both time-consuming and required great skill. This inconvenience stemmed from the requirement for the second trimethylene group in the formation of macrocycle 43. It was mentioned above that the prospective monotrimethylene-linked donor 35 had not proved accessible to synthesis, but despite this, we began to perform

Scheme 7. Formation and Reactivity of Imidazole-Derived Super-Electron-Donors
The Journal of Organic Chemistry This led to our quest to isolate the two donors 35 and 41. We succeeded, but the unbridged donor 41 was exquisitely reactive (and donor 35 was only slightly less so), and we concluded that its decomposition was catalyzed by a proton abstracted from the ultradry flask surface, affording 68 that underwent fragmentation to give carbene 69 and imidazolium salt 40. 27,28 This salt, 40, then provides a proton for another molecule of 41, showing the catalytic nature of the destruction. Thus these compounds, 35 and 41, are so reactive that they cannot be preserved in glass.
The issue of a more convenient electron donor was solved with the preparation of the bipyridinylidene 71 (Scheme 9), derived from 4-DMAP. 24,27 This deep-purple compound is easily prepared by treating the precursor disalt 70 with base, either in DMF for in situ preparation or in liquid ammonia, from which the pure solid product 71 can be isolated.
Cyclic voltammetry comparing the three types of donor 29, 45, and 71 was illustrative (Figure 2). The benzimidazole donor 29 is shown in blue, the doubly bridged donor 45 in green, and the DMAP-derived donor 71 in red. It is seen that all of the redox processes are reversible, i.e., that decomposition does not occur during the cycling processes. The further to the left the peaks appear on this voltammogram the more reducing is the electron donor responsible for that peak. Hence, the benzimidazole-derived donor is much weaker than the other two donors. For the DMAP-derived donor 71, a single peak, calibrated as a two-electron peak, is seen, and this donor is as Donor 71 is not only easier to prepare but also more robust. Indeed, an even simpler analogue, where the N(CH 2 ) 3 N of 71 is replaced by two NCH 3 groups, was also prepared and isolated by treating N-methyl-4-(dimethylamino)pyridinium iodide with base.
With donor 71 available easily in multigram quantities, its scope was now studied. It converted iodoarenes to aryl anions at room temperature, as seen in the reaction of substrate 48 (Scheme 9). Here, very efficient cyclization to the indanone 43 is seen, although a little protonation of the aryl anion also occurred to afford a product that, on hydrolysis, provided the acid 50 (8%). The higher conversion to indanone here (compared to what was seen with donor 45) likely results from the lower acidity of the oxidized forms of 71.
We now explored the chemistry of alkyl halides. While the milder donor 29 had converted alkyl halides to the corresponding alkyl radicals, the possibility of conversion to alkyl anions with the (inorganic salt-free) stronger donors 45 and 71 was explored. Once again, it needs to be stated that such naked organic anions would be quite different from organometallic counterparts (Grignard reagents and organolithium reagents) and also different from alkyl anions that might be produced during electrochemistry in solution in the presence of metal salt electrolytes. Hence, any predictions as to their stability that arise from electrochemistry experiments are likely to be challenged in our experiments to make naked anions. Hence, it was uncertain whether alkyl anions could be prepared here from alkyl halides.
Of the two strong donors, donor 45 was studied first. 29 Small amounts of homologated aldehyde products were produced in the initial reactions. When the reactions were worked up with dilute acid instead of with water, improved yields of these aldehydes were formed, suggesting that the aldehydes were being liberated from a protected form during workup. An example is alkyl bromide 72 that, under optimized conditions, afforded aldehyde 73 in 61% yield (Scheme 10). As the reactions had been carried out in dimethylformamide (DMF), at first this looked consistent with reaction between alkyl anions and DMF. However, when DMF was replaced as solvent by dimethylacetamide (DMA), the aldehydes were still formed, indicating that they were not dependent on DMF as solvent and showing that the extra carbon atom had been extruded from the donor. The mechanism for formation of the aldehyde some special alkyl halides were designed as probes. Initial plans were to probe for two-electron transfer events in the substrates using alkyl iodide 85 (Scheme 11). Thus, direct formation of an anion 86 by transfer of two electrons would lead to rapid elimination of the alkoxide ion before any intermolecular coupling could be achieved by the anion to an imidazolium salt. Hence the alcohol 88 would be isolated on workup. However the transfer of two electrons to the substrate 85 would not necessarily involve stepwise reaction via 86; a concerted E2 process is a likely alternative, and this would also lead to isolation of the same alcohol on workup. But the alcohol could also be liberated if radical 89 were formed and if coupling to radical cation 46 occurred to form adduct 90, followed by formation of carbene 91. Proton transfer would afford the enediamine 92, from which the alkoxide would be expelled, leading to alcohol 88. So to distinguish between the mechanisms for formation of 90 requires a little more sophistication in design, and this is addressed in substrates 94. Here, two-electron transfer would lead to the homoallylic ether 95 through elimination of the alkoxide; however, ether 95 should not react further under the reaction conditions (this was verified in blank reactions). By contrast, the radical coupling product 96, by proceeding through intermediate 97, could afford salt 99 with expulsion of alkoxide. That alkoxide could deprotonate the imidazolium product to afford dienediamine 100, from which the second alkoxide would then be easily expelled. When the reactions were conducted, very good yields of both alcohols ROH and R′′OH were isolated, providing the additional evidence in favor of radical coupling. (This coupling of alkyl radicals to a radical cation intermediate was not confined to this system but also occurred for the DMAPderived donor 71.) 30 Donor 71 and donor 45 performed almost identically in their reactions. Donor 71 was now used to expand the scope of the electron transfer studies, initially through reaction with Weinreb amides (Scheme 12). 31 Here, reduction of the N−O bond was seen. However, an interesting observation was that the ease of the transformation depended on what was present in the side chain. Thus, the substrate 102 underwent reduction to afford the secondary amide 103 in good yield (77%) using 1.5 equiv of donor 71. However, the simpler Weinreb amide 106 required 5 equiv of donor and prolonged reaction time to achieve a lower yield of product 107 (43%). Computational analysis showed that the LUMO of 102 is associated with the arene and not with the Weinreb amide functional group. Hence, initial electron transfer to the arene could be followed by intramolecular transfer to the Weinreb amide group. In the absence of an arene, the reaction is more difficult. This fact later brought us to study the interactions of the donors specifically with arene functional groups.
At this stage, we also studied the interaction of this donor with acyloin derivatives. 32 The reaction depended on the nature of the potential leaving group α to the carbonyl group. When this was a simple alkoxy group, this did not undergo efficient reductive cleavage. However, with better anionic leaving groups, e.g., 110 and 112, the reactions went efficiently. We will return to the cleavage of benzylic alkoxy groups later.
The power of donor 71 was also seen in the reduction of alkyl triflates. 33 Alkyl triflates are excellent substrates for S N 2 reactions at carbon, but in this case, reduction of the alkyl triflates to the parent alcohol was seen. The formation of the alcohols was first attributed to intervention by the solvent, DMF. Nucleophilic attack by DMF on alkyl triflates ROTf can occur, but isolation of the resulting alcohol ROH on workup should show incorporation of the oxygen atom from DMF. However, use of 18 O-labeled DMF led to unlabeled alcohol in our hands, and so the alcohol did not arise by attack by the solvent. Instead, electron transfer to the triflate group is occurring and leads to S−O bond scission. To our knowledge, this "deprotection of alkyl triflates" is unique in the literature.
The donor 71 had the advantages of being a strong donor that was easy to prepare and more robust than the donors based on imidazole. It was also relatively easy to alter its periphery to determine the effects of changes in its substituents and in the size of the oligomethylene linker. 34 Three analogues, 119−121, were prepared and tested by cyclic voltammetry (Scheme 13). As with 71, all of the cases showed reversible redox characteristics, i.e., no signs of decomposition under the conditions of the experiments. Three of these were extremely similar to each other in showing a single two-electron wave at essentially the same potential. The exception was the dimethylene case 119. Here, two one-electron waves were seen. The first oxidation potential was consistent with the oxidation potentials of the other donors, but the second occurred at less negative potential, indicating a relative reluctance to be oxidized to a dication. The likely reason is that in this case, the product dication 122 is constrained to be essentially planar, with full interaction between the two positive charges, leading to less stability than in the twisted conformations of related dications.
With the three motifs now studied, 29 derived from benzimidazole, 45 derived from imidazole, and 71 derived from 4-dimethylaminopyridine, it was clear that the benzimidazole donor is a very good one-electron donor to iodoarenes but that it cannot react with bromobenzenes or chlorobenzenes. While the imidazole-based donor 45 and the DMAPbased donor 71 are stronger donors, they mediate two-electron transfers to form aryl anions. To achieve more powerful one-

Scheme 12. Substrates Reduced by Donor 71
The Journal of Organic Chemistry electron donors to haloarenes, hybrid donors were prepared 35 that combine one ring derived from DMAP or from imidazole together with the other ring derived from benzimidazole. In fact, all three hybrid donors 123, 126, and 127 were prepared. Hybrid 126 showed redox chemistry similar to its "parents"; here, two one-electron redox steps were seen at the expected average potential for 45 and 71. The DMAP-benzimidazolederived hybrid donor 127 showed two one-electron reversible peaks at potentials intermediate between the values seen for its two "parents" 29 and 71. The imidazole−benzimidazole donor 123 showed a single peak, from which it was clear that this compound was anomalous, since here the oxidative current was not as strong as the reductive current in the cyclic voltammogram, a feature that was more pronounced at slow sweep rates. (For all of the cyclic voltammetry studies, the experiment starts with the stable oxidized disalts, which are reduced first to form the electron donors and then reoxidized; in the absence of decomposition, the reductive and oxidative currents should be equal in size). This indicated that decomposition of 123 itself was occurring, and proton transfer from the oxidized form of the donor to the donor itself was identified as the probable source. This protonation would lead to 124 and then to crucial rupture of the central bond to form imidazolylidene 125 in the first instance. This was not a problem with donor 127. Here, protonation of the donor can occur to form 128, but rupture of the central bond would lead to a pyridinylidene 129 which, from computational studies, is a much less stabilized carbene than the imidazolylidene mentioned above, and so, the cleavage is unlikely to happen.
Thinking about even greater challenges for electron transfer, we contemplated making stronger electron donors. Since the driving force for oxidation correlates with the aromatic stabilization energy gained on oxidation, the donor 130 was prepared. Here, oxidation of the donor through loss of two electrons would lead to dication 131 in which three rings had become aromatic, and this should be associated with a greater driving force for oxidation. 36 Cyclic voltammetry confirmed this, with a record redox potential (E 1/2 = −1.5 V vs SCE) being recorded for this neutral organic electron donor. Although 130 was readily accessible, the drive toward more powerful donors now took a different direction.
One feature of these organic electron donors is that they are strongly colored, either vibrant yellow (29,45) or deep purple (71). Accordingly, excitation by visible light or by near-UV should be possible to promote an electron from HOMO to LUMO. The promoted electron would then be strongly reducing toward substrates. This strategy could be employed for reducing different classes of molecules. In our case, we had found that iodoarenes were easily reduced by ground-state donors under moderate conditions. Bromides were much less reactive, while aryl chlorides were just unreactive. To test the effect of photoexcitation of the donors, the chlorobenzene 132 was subjected to donor 71 under photoactivated conditions and gave rise to 133, the product of clean reductive dechlorination (Scheme 14). 37 We were keen to give our photoactivated donor a sterner test, the reduction of a benzene ring that had no electronegative elements attached. 37 Reduction to an arene radical anion, followed by proton transfer, would give rise to a Birch reduction, but our donors are quite basic and therefore might be incompatible with proton transfer to such mild bases as arene radical anions. Under photoactivation conditions, back-electron transfer is always a possibility, and so we planned that even reversible electron transfer to the arene should be logged. Accordingly, the diphenylcyclopropane 134-cis was chosen. Conversion to the radical-anion 135-cis led to opening to afford the distal radical-anion 136. Reversible ring closure of 136 would afford a mixture of cis and trans isomers of 135 and workup to cis-and trans-diphenylcyclopropane 134. This was exactly the outcome of this experiment. Reductive trapping of the intermediate distal radical anion 136 was also observed to afford 1,3-diphenylpropane 137 in experiments of extended duration. Accordingly, electron transfer to benzene rings without highly activating electronegative substituents is achieved by donor 71.
The scope of the photoactivation reactions was then extended by looking at deprotection of benzylic esters and ethers. 38 Photoexcited donor 71 deprotected benzylic esters to carboxylic acids in high yields; benzylic ethers were also deprotected to alcohols, but the results were intriguing (Scheme 15). The deprotection of the esters was achievable under milder conditions than for the ethers. However mechanistic differences were apparent as highlighted for ester substrate 142 and ether substrate 143, for which identical deprotection conditions were selected. The ester 142 was converted to its radical anion, 144, and this underwent cleavage to afford pivalic acid (85%) on workup. No products were isolated from the benzylic moiety of this substrate and this was entirely consistent with radicals 145 and/or 146 being trapped by the radical cation of donor 71 as described earlier.
In contrast, the ether 143 underwent slower cleavage leading to recovery of 143 (45%) at the end of the experiment. 38 However, the reduced cyclopropane 149 (29%) was also isolated, showing a different mechanistic pathway than for the ester cleavage. The precursor of the cyclopropane 149 must be the benzylic anion 148, rather than the radical 145, and this highlights that cleavage of the benzylic ethers involves twoelectron reduction. In terms of the timing of the events, it is most likely that the radical anion 147 is slow to lose methoxide anion but that this fragmentation is triggered as another electron is received. This concerted process would avoid the formation of an antiaromatic dianion prior to loss of methoxide.
Scheme 7 showed that the ground-state donor 45 had cleaved arenesulfonamides, where the nitrogen leaving group, whether it be a radical or an anion, was stabilized by resonance, but N,N-dialkylarenesulfonamides were completely untouched by the donor. The effect of photoactivation was now explored using donor 71, and this proved to be highly effective at cleaving dialkyl arenesulfonamides, e.g., 150 and 152 (Scheme 16). 39 Again, this illustrates the significant boost to reactivity brought about by photoactivation.
Having seen effective cleavages of benzylic C−O bonds, the reactivity of benzylic and related C−N bonds was now investigated. The benzylic C−N bond in substrates such as 154 and the allylic C−N groups, as in 156, underwent efficient cleavage. In addition to cleavage of ArC−N bonds, cleavage of ArN−C bonds was also seen, e.g., in substrates 158 and 160. Activation of the systems through incorporation of the pivaloyl group in 156 and the N−CO 2 Et group in 158 significantly assisted these cleavages, probably by lowering the LUMO energies of the substrates. 39 One of these examples featured the remarkable transformation of the N-phenylproline 160 to Nphenylpyridone 161. Although the yield was low (30%), significant amounts of unchanged starting ester 160 were also recovered (62%). These donors are performing at the limit of their effectiveness, but modified versions of the donors that are slightly more powerful may be able to further facilitate these intriguing transformations and to extend reduction to even less activated substrates.
Even more remarkable chemistry was seen with C-benzyl malonates and C-benzyl cyanoacetates (Scheme 17). 40 The diethyl dibenzylmalonate 162 has long been known to undergo selective reductive cleavage with alkali metals (Na, K) to afford the ethyl dibenzylacetate 164. 41 In that transformation, electron transfer from the alkali metal to the ester group affords a ketyl radical anion, which undergoes fragmentation to afford the anion 169 and the alkoxyacyl radical 170. Neutralization affords the isolated product 171. Although this is exactly what we expect based on known reactivities, what is not known is to what extent the energetics of this transformation depend on the complexation of metal species with the ketyl during the overall process. The novel feature of our reagents is that no metal ions are present, thereby removing that stabilization. In addition, our donors all feature extended π-systems, and these would expect to associate preferentially, by π-stacking, with the arene rings of substrates. This could alter the regioselectivity of the reactions. When the reactions were performed, this was indeed borne out. No substrate showed cleavage of an ester group, as had been seen with K and Na as the reducing agents. Instead, selective debenzylation reactions occurred. For substrate 162, this proceeded through arene radical anion 165 that fragmented to benzyl radical 167 and malonate anion 168, affording the diethyl (mono)benzylmalonate 163 in excellent yield (75%) upon workup. As expected, across a range of substituted substrates, no products were isolated that derived from the benzyl radicals, which would be expected to couple with the donor radical cations to afford water-soluble products that would be easily separated from the desired products. So, using the photoactivated organic electron donor 71, electron transfer had occurred to an unactivated benzene ring, in preference to a malonate moiety. This overturns our perceptions of relative reactivities and, when developed further, may have important implications for the field of synthesis. Nor was the reactivity confined to malonate examples. 40 It had previously been reported that reaction of the ethyl dibenzylcyanoacetate 172 with samarium diiodide had afforded the ethyl dibenzylacetate 164. 42 Again, this can be expected to benefit energetically from association of the samarium ions with lone pairs on the substrate. When reacted with 71, no decyanation was detected; instead, debenzylation had occurred giving ethyl benzylcyanoacetate 173 (75%).
Most recently, opportunities to recognize the role of organic electron donors have expanded further, this time in relation to transition-metal-free coupling of haloarenes with arenes or styrenes. Itami, 43 Shi, 44 and Hayashi 45 announced the coupling of iodoarenes to arenes in the presence of potassium tertbutoxide but in the absence of transition-metal species. The proposal was that these reactions should proceed through aryl radicals. In an essay, Studer and Curran 46 described the radical chemistry as in Scheme 18A. Here, aryl radicals 176 add to benzene to afford a cyclohexadienyl radical 177, deprotonation of which gives 178, an arene radical anion. This transfers an electron to another molecule of halobenzene to begin another cycle. However, the reactions depend on a viable mechanism for generating aryl radical initiators 176. The reactions were found to proceed even better in the presence of certain additives, although the breadth of structures that facilitated the reactions was quite surprising. These included phenanthroline 183, 44,45 N-heterocyclic carbenes 184, 47 and pyridine 48 or related heterocycles 43 (Scheme 18B).
For the case of a phenanthroline, 183, the literature had proposed as a working hypothesis that a complex between phenanthroline and potassium tert-butoxide would allow electron transfer from tert-butoxide to occur. Similarly, complexation of potassium tert-butoxide by N-heterocyclic carbenes 184 was proposed to lead to electron transfer. However, examining the phenanthroline case, computational calculations in our hands suggested that the thermodynamic energy difference between educts and products would be enormous, and so the kinetic barrier for the transfer will be at least as high. 49 This pressed us to look for an alternative. Since our electron donors were adept at reductive cleavage of iodide from iodobenzenes, we investigated whether they could initiate the coupling reaction of iodobenzenes with benzenes. The answer was a resounding "yes". Traces of our donors, or their precursors that could be transformed into the donors upon treatment with base, were sufficient to give high-yielding coupling reactions as seen in Scheme 19 in the coupling of substrates 190 and 180 to benzene.
Since our donors are formed from "dimerization" of Nheterocyclic carbenes, this suggests the active component in the reaction where carbene complex 193 was used could be the tetraazafulvalene 195. The ability of N-heterocyclic carbenes like 194 to "dimerize" in the presence of a proton donor suggested that this might be the mechanism here. 28 The source of the proton would be tert-butyl alcohol, and this would arise in trace quantities from reaction of potassium tert-butoxide as a base with the iodoarene substrates.
This looks unrelated to the cases where phenanthroline was used as an additive, but it is related. In our hands, dark-green precipitates were formed when we repeated literature reactions using phenanthroline. Dark-green precipitates also formed when blank reactions between phenanthroline and potassium tert-butoxide were performed (Scheme 20). Analysis showed that these precipitates were highly sensitive to air when worked up. A more controlled workup involved reaction of the precipitates with iodine, an excellent electron acceptor. From this reaction, biphenanthroline 196 had formed. This shows that a phenanthroline anion, formed on treating phenathroline with potassium tert-butoxide, has added as a nucleophile to a second phenanthroline (which may also be complexed to potassium ion to enhance its electrophilicity), and that provides an excellent rationale for electron transfer. Either the monoanion 197 or, more likely, the dianion 198 arising from further deprotonation could act as an electron donor triggering the formation of aryl radicals to initiate the reaction.
The case of pyridine was similar. Here, a pair of isomeric bipyridines 202 and 203 was formed when a mixture of pyridine and tert-butoxide was heated, followed by quenching with molecular iodine. The precursor electron donors to compound 202 would be the monopotassium salt 200 or, possibly, the dianion 201 as shown. Interestingly, the isolation of the isomeric bipyridine 203 must start with deprotonation of pyridine in the 4-position. 50 Thus, although these cases appear at first glance very different from our SED reactions, in fact, a common mechanism can apply. This Perspective has looked at strong organic electron donors and their applications in synthesis. A great deal of additional research in the preparation of organic electron donors has been conducted, and key compounds 203−206 are represented in Figure 3. These compounds represent beautiful molecular architectures, but they have generally not yet been applied to synthetic transformations. Figure 3 lists the organic donors with their oxidation potentials. More than one oxidation potential has been noted in the literature for sequential electron loss events, and these are included.
In summary, based on the first discovery of the electron donor TDAE in 1950 in industry, we have recently seen the development of simple organic systems that are extremely powerful electron donors both in the ground state and upon photoexcitation. That such molecules can selectively reduce benzenes to their radical anions while leaving recognized electrophiles like malonates and cyanoacetates untouched is truly amazing. These developments have taken place with organic super-electron-donors, but complementary developments in electron transfer chemistry with both metal-free agents 1 and with transition-metal-containing complexes 2 make redox chemistry through electron transfer a fast moving and exciting area for research. We look forward to the next five years and the discoveries that they will bring.