Several recent articles have drawn attention to the virtually boundless extent of chemical space; the domain that contains the totality of all possible compounds [1,2]. The total number of molecules that could be made from only 30 atoms is in the range 1020 to 1024 , “drug-like” chemical space comprises over 1060 molecules [4,5] and, of course, even these huge numbers are insignificant in comparison with the protein or nucleic acid spaces. The number of polypeptide chains of modest (250 unit) length, drawn from the 20 natural amino acids, exceeds the ‘trans-astronomical’ number of 10325 . The CAS registry currently contains about 108 chemical substances. Its present rate of growth is about 5 × 106 substances per year, so that at this rate more than 1054 years would be needed just to explore “drug-like” chemical space! It is abundantly evident that only a minute fraction of chemical space can ever be preparatively accessed. To address this problem, computational algorithms are being devised capable of virtual screening and/or for locating, within the total space, “islands” or “trees” of substances with potentially desirable properties such as bioactivity [1,2,7]. Ertl and co-workers developed self-organising neural networks which showed that a comparatively small number of ring structural units is associated with bioactivity. They listed 30 heterocyclic moieties as of crucial importance and 22 of these contained one or more N-atoms . Pyrrole, indole and related structures figured prominently, as did pyridine, quinoline, quinazoline and analogous 6-membered ring-containing aza-arenes. The immense size of chemical space presents exciting opportunities of discovering hitherto unknown and extraordinary substances with properties beneficial to human society. Its size also represents a huge challenge for preparative chemists such that it is imperative to open up every possible avenue that might facilitate the task. The exploration and exploitation of identified “islands” depends critically, of course, on the availability of practical preparative methods. Thus, the development of synthetic strategies for the ring systems associated with bioactivity that are fast, efficient and of low environmental impact, deserves special attention.
The advantages of free-radical based preparative methods include the usually neutral conditions, the tolerance for many unprotected functional groups and the availability of much kinetic, thermodynamic and mechanistic data to guide the design of experimental methodology. During the last two decades a great deal of research has been directed towards making radical-mediated synthetic methods safer, more efficient and more convenient [8,9,10,11,12,13]. New tactics have been devised for avoiding hazardous initiator peroxides or azo-compounds and for dispensing with toxic tin, mercury, copper and other metal reagents. For example, ‘pro-aromatic’ reagents, based on the cyclohexadiene structure, release many radical types without the need for metals . Murphy and co-workers’ development of organic super electron donors unlocked completely new ways of generating radicals and radical-ions and harnessing them synthetically [15,16,17]. The unique properties of organoboron compounds have led to the design of several different reagent types for radical release including B-alkylcatecholboranes [18,19] and N-heterocyclic carbene boranes [20,21,22,23]. The discovery of homogeneous photoredox catalysts (PCs) has had huge impact on radical-mediated preparations. The most popular are complexes of Ru or Ir [24,25,26,27] that re-introduce metals, albeit in small quantities. However, organic dyes and other donor molecules are also coming into use as PCs [28,29]. Heterogeneous photoredox catalysts, particularly titanium dioxide (titania, TiO2), possess the added convenience of easy removal after use by filtration or centrifugation. Their exploitation for radical mediated preparations is also developing rapidly [30,31,32].
The N–O bonds in oximes and in oxime derivatives are comparatively weak and break homolytically with production of a pair of N- and O-centered radicals. Aldehydes and ketones are available as starting materials in huge variety from natural and commercial sources. Oximes can be prepared essentially quantitatively from them simply by treatment with hydroxylamine hydrochloride. The consequence is that oxime esters of many types containing the >C=N–OC(O)– structural unit (carbonyl oximes) and oxime ethers containing the >C=N–O– unit are very readily accessible. Most members of both classes are stable to moderate heat and hydrolysis, are non-toxic and non-hazardous, are easily handled and have long shelf lives. Suitably functionalised, they have proved amazingly adaptable for radical generation by an unprecedentedly wide range of methods . These include conventional thermolyses, microwave irradiations, UV photolyses, sensitised UV photolyses and with several types of photoredox catalysis. Oxime derivatives are therefore particularly flexible, convenient and benign and stand as very attractive alternatives to other more hazardous radical precursors. This article reviews the use of both carbonyl oximes and oxime ethers in radical mediated organic syntheses.
When fittingly stimulated, both compound types initially yield N-centred iminyl radicals >C=N• (Im, Scheme 1). Those suitably accoutred with acceptor groups can, when appropriately manipulated, yield azaheterocycles. An O-centred radical [•OC(O)Z] is released from a carbonyl oxime together with the iminyl radical and can be chosen to end up as volatile or otherwise easily separable by-products. For the oxime ether precursors, best results are usually achieved with O-aryl substituents. In this case the by-product is usually a phenol (ArOH) which can readily be removed because of its mild acidity. Iminyl radicals with butene or butyne type side chains selectively undergo 5-exo cyclisation to produce 5-member ring containing dihydropyrrole type products. By way of contrast, iminyl radicals with aromatic or heteroaromatic acceptor substituents preferentially yield 6-membered ring pyridine, quinoline etc. products. In some instances this results from an initial 5-exo spiro cyclization followed by ring expansion via an aziridinyl type intermediate (see for example Section 4.1). Preparations of many different azaheterocycle types may therefore be achieved by careful choice of the acceptor substituent(s), and by tuning the reaction conditions and methodology.
This review also focuses on the iminyl radical based synthetic methodology developed for the sets of aza-arenes identified above as of importance in the bioactivity islands of chemical space.
2. Syntheses of Dihydropyrroles and Pyrroles
Organotin promoted radical methodology is justly famed because it works seamlessly in so many situations and has proved so dependable. Many ingenious syntheses of azaheterocycles have employed organotin hydrides or ditins for the generation of iminyl or aminyl radicals. Zard, for example, described tin hydride-mediated syntheses of dihydropyrroles, indolizidines and other aza-heterocycles from sulfenimines (PhS–N=C<), thionocarbazones and other derivatives [34,35]. Nanni and co-workers generated iminyl radicals by ring closures of C-centred radicals onto organic nitriles and hence prepared many heterocyclic systems. They also employed tin-free thermolytic and other processes [36,37,38]. Much of this earlier research has been reviewed by Bowman and Aldabbagh [39,40,41] and/or by Fallis and Brinza . Recently, Zhang and Studer have published an outstanding review of aza-arene syntheses flowing from radical additions to organic isonitriles . This methodology exploits the formation and ring closures of amidoyl radicals (Ar–N=C•–R).
2.1. Pyrrole and Dihydropyrrole Preparations from Carbonyl Oximes
Numerous alkaloids contain pyrrole, dihydropyrrole or related rings and many biological roles are associated with these structures [44,45,46]. Oxime esters 1 can easily be prepared from oximes reacting with either carboxylic acids or acyl halides [47,48]. On photolysis they release an iminyl radical together with an acyloxyl radical and the latter rapidly extrudes CO2 with production of a C-centred radical [48,49] (Scheme 2). Rodrigues, Sampedro and co-workers used acyl oximes such as 2 as efficient sources of iminyl radicals [50,51,52]. With this precursor type, the radical co-produced with the iminyl during UV photolyses was acyloxyl [CH3CO2•] that simply furnished volatile CH4 and CO2 as by-products. They reported that the iminyls could be conscripted into syntheses of many types of heterocycles including dihydropyrroles. UV photolysis of 2 through Pyrex released iminyl radical 3 with an alkenyl side chain. These selectively cyclised in the 5-exo mode with production of pyrolidinylmethyl radical 4 that subsequently abstracted an H-atom from co-reactant cyclohexa-1,4-diene (CHD) to yield 3,4-dihydropyrrole derivative 5 (Scheme 2).
Dioxime oxalates 8 are another type of oxime ester that proved convenient for clean generation of iminyl radicals. Symmetrical types were made by treatment of an oxime 6 with oxalyl chloride to yield oxime oxalyl chlorides 7 as intermediates (Scheme 3). Although these could be isolated, they hydrolysed and degraded quickly, so immediate treatment with another equivalent of either the same oxime 6, or a different oxime, yielded the symmetrical dioxime oxalates 8 or unsymmetrical types such as 11a–d respectively [53,54]. Dioxime oxalates 8 were “clean” and atom-efficient because, apart from CO2, they delivered only iminyl radicals 9 on UV photolysis. Photo-dissociations were aided by inclusion of 4-methoxyacetophenone (MAP) as photosensitizer and by the presence of an aryl substituent adjacent to the C=N bond.
3,4-Dihydropyrrole derivatives 10 were obtained in moderate yields with toluene acting as both solvent and H-atom donor (Scheme 3). Unsymmetrical dioxime oxalates such as 11 incorporated an aryl-oxime unit, to promote photo-dissociation, as well as an alkenic acceptor unit. These precursors enabled non-aromatic dihydropyrroles such as 12a,b to be accessed. Attempts to convert the latter to pyrolizidines 13 and indolizidines 14 were unsuccessful.
2.2. Pyrrole and Dihydropyrrole Preparations from Oxime Ethers
Thermal preparative methods are often superior because of their simplicity and non-hazardous nature. O-Phenyl oxime ethers 15 can easily be made by treatment of carbonyl compounds with the commercially available O-phenylhydroxylamine hydrochloride. Conventional thermolyses of appropriate derivatives were shown to provide dialkyl- or diaryl-iminyl radicals . Subsequently it was established that microwave heating (μwave) was a particularly efficient means of releasing iminyl radicals and mediating dihydropyrrole preparations [56,57].
The optimum procedure utilized toluene as both solvent and H-donor together with an equivalent of the ionic liquid (IL) 1-ethyl-3-methyl-1H-imidazol-3-ium hexafluorophosphate (emimPF6) to promote microwave absorbance. This method enabled ketones with but-3-enyl type side chains to be converted to dihydropyrroles 16 in good yields in two steps (Scheme 4). The phenoxyl radicals released from 15 also abstracted H-atoms from the solvent to afford phenol as an easily separable by-product. When oxime ether 17 with an alkyne side chain was microwave irradiated under similar conditions, pyrrole 19 was isolated in good yield. Evidently the first-formed methylene-dihydropyrrole 18 rearranged under the reaction conditions.
Castle and co-workers prepared a set of alkyne-substituted oxime ethers 20 and carried out microwave irradiations of mixtures with tetramethylpiperidine-N-oxide (TEMPO) in benzotrifluoride solvent . The ring closed radicals were trapped by the TEMPO with production of intermediates 21 (Scheme 4). These also rearranged, with loss of a piperidinyl radical, so providing 2-acylpyrroles 22 in good to excellent yields.
2.3. Photoredox Catalyzed Preparations of Pyrroles and Dihydropyrroles from Oxime Derivatives
Photoredox catalytic methodology [25,27,59] has also been developed for production of dihydropyrroles from O-aryl oxime ethers substituted with electron withdrawing groups (EWG) in their O-aryl units. Narasaka and co-workers prepared oxime ethers 23 containing 4-CN (or 2,4-di-NO2 or 4-CF3) aryl substituents. On inclusion of a catalytic amount of 1,5-dimethoxynaphthalene (DMN) and irradiation with UV light in 1,4-cyclohexadiene, dihydropyrroles 25 were isolated in good yields (Scheme 5) . The incident light raised the photocatalyst to an excited triplet state (PC*) that then transferred an electron to the oxime ethers with production of the radical anions 24 (Scheme 5). Loss of the stable phenolate type anions then occurred with release of the corresponding iminyl radicals that subsequently underwent 5-exo cyclisation and H-atom transfer with CHD to afford dihydropyrroles 25.
Furthermore, Leonori and co-workers reported recently that the dye Eosin Y (as PC) catalysed dihydropyrrole formation, simply with light of visible wavelength, when oxime ethers with O-2,4-dinitroaryl substitution 27 were employed as reactants . Remarkably, Et3N could replace Eosin Y: visible light irradiation of 27 (EWG = 2,4-di-NO2) with Et3N in CH3CN furnished imino-alcohols 31 in yields up to 85%. The complex of Et3N with the electron-poor ring of 27, on excitation with visible light, generated a radical anion that dissociated to give 2,4-dinitrophenoxide together with pyrrolidinylmethyl radical 28. The oxygen atom was believed to arise from an intermediate such as 29 that fragmented to nitrosophenoxide and a pyrrolidine-containing alkoxyl radical. The latter picked up an H-atom to deliver imino-alcohols 30 as the products (Scheme 5).
Weinreb and co-workers described an alternative strategy in which oximes 31 could be used directly when treated with 2,6-dimethylbenzenesulfinyl chloride in the presence of Hunig’s base (N,N-diisopropylethylamine, DIEA) and a radical trapping agent in dichloromethane (DCM) . The best radical traps (Z) were found to be TEMPO and diphenyl diselenide; but CHD could also be used when in great excess (Scheme 6). Probably sulfinate esters 32 were first formed that dissociated to a caged iminyl/sulfonyl radical pair 33. The reformed N-sulfonylimines 34 then released iminyl radicals that cyclised and were trapped to afford functionalised dihydropyrroles 35 in moderate to good yields (Scheme 6).
3. Preparations of Pyridine, Quinoline, Phenanthridine and Related Aza-Arenes
Compounds containing the quinoline unit are hugely important to the well-being of society because they play essential roles across the areas of medicine, pharmacology, nutrition, dyes, and even electronics . Bioactive natural products containing quinoline cores are widely distributed in many plants, marine plants, corals, sponges [63,64,65,66,67] and even in chestnut honey . Much the same can be said for compounds containing phenanthridine units. They are noted for their ability to bind to DNA  and have antimicrobial, anti-inflammatory, and anti-tumor activities [70,71,72,73,74]. Because of this importance to the field of drug development, synthetic methods for molecules containing these structural units, particularly short, mild and atom-efficient ones, have attracted great attention.
3.1. Preparations of Aza-Arenes from Carbonyl Oximes
Iminyl radicals 36 containing suitably sited aromatic acceptor units usually ring close in 6-endo mode with formation of a 6-membered N-containing ring 38 (Scheme 7). This rule holds good for cyclisations onto a wide range of aromatic acceptors including benzene, naphthalene, furan, thiophene, indole, pyridine and analogous moieties. As would be expected for less-favoured 6-endo processes, the rate constants for these iminyl ring closures are significantly smaller than for 5-exo cyclisations . Never-the-less practicable preparative protocols have been established for many mono- and poly-cyclic heterocycles.
It is usually necessary to work with precursors having the second imine substituent R1 either Me or Ar. If R1 is H (or a branched alkyl group) formation of a nitrile 37 is either the main process, or 37 is an important by-product (Scheme 7). Nitrile formation may entail an electrocyclic process of the oxime ester; in which case the proportion may be diminished by suitable solvent choice. With several types of oxime esters cyclisations onto aromatic rings initially create cyclohexadienyl type radicals 38. If a suitable H-donor is present as solvent, or otherwise, H-atom transfer to radicals 38 yields cyclohexadienes 39 as a mixture of isomers. In practice this route has rarely been developed. Instead restoration of aromaticity to the acceptor ring has usually been observed as in quinoline derivative 41. This can result when some radical in the system abstracts the highly labile tertiary H-atom of the cyclohexadienyl ring in radical 38. However, this is a disproportionation that requires two radicals to meet and so is not favored because reactive radical concentrations are always very low. The alternative is that an acceptor molecule A is either added, or is adventitiously present, and single electron transfer (SET) occurs with production of the corresponding cyclohexadienyl cations 40. These rapidly deprotonate to yield cyclised products 41 with restored aromaticity. This oxidative process is facilitated by non-H-atom donor solvents such as PhCF3 or t-BuOH.
Formyl and acyl derivatives of biphenyl, 4-phenylpyridine, 2-phenylthiophene, 3-phenylpyrazole, 2-phenylnaphthalene, 2-phenylindole and analogous aromatics are readily accessible and can be converted to the corresponding acyl oximes without difficulty (see Scheme 8). UV photolyses of these precursors through Pyrex in acetonitrile or t-butanol solutions enabled several 6-substituted phenanthridines 42 to be prepared [50,52]. Similarly, benzo[c][1,7]naphthyridines 43, and thieno[3,2-c]isoquinolines 44 were conveniently prepared from the corresponding acyloximes.
Analogous photochemical preparations were described for 2,4-dimethyl-2H-pyrazolo[4,3-c]quinoline 45 and for benzo[i]phenanthridine 46. However, the aldoxime precursor from 2-phenylindole gave only 2-phenyl-1H-indole-3-carbonitrile 47
3.1.4. Synthesis of 2-(4-Chloromethylphenoxy)-5-(trifluoromethyl)pyridine (4)
A solution of intermediate 3 (13.5 g, 50 mmol) in dichloromethane (80 mL) was cooled in ice-water bath followed by adding thionyl chloride (8.9 g, 74.8 mmol) dropwise over 20 min. Then a few drops of N,N-dimethylformamide was added thereto. The resulting solution was stirred at room temperature for 8 h. The reaction mixture was quenched by trash ice, then the organic phase was separated, washed by water and saturated NaHCO3, dried over anhydrous Na2SO4, and evaporated in vacuo to produce compound 4 (yield 82%) as a white solid; 1H-NMR (CDCl3): δ 8.36 (s, 1H, Py-H), 7.84 (dd, J1 = 8.8 Hz, J2 = 2.4 Hz, 1H, Py-H), 7.38 (d, J = 8.4 Hz, 2H, Ar-H), 7.07 (d, J = 8.4 Hz, 2H, Ar-H), 6.96 (d, J = 8.8 Hz, 1H, Py-H), 4.54 (s, 2H, CH2). Anal. Calcd for C13H9ClF3NO: C 54.28; H 3.15; N 4.87. Found: C 54.41; H 3.03; N 4.70.
3.1.6. General Procedure for the Preparation of 8a–8w
To a stirred solution of intermediate 4 (7.2 mmol), compound 7 (6 mmol) in anhydrous acetonitrile (30 mL) was added Cs2CO3 (7.2 mmol) at room temperature, the resulting mixture was heated to reflux for 10–18 h. After cooled to room temperature, the reaction mixture was filtered. After most of the solvent had been evaporated under reduced pressure, the residue was admixed with water (100 mL) and extracted with dichloromethane (3 × 30 mL). The combined organic layer was washed with water (3 × 30 mL), and dried over anhydrous Na2SO4. The solvent was removed using a rotary evaporator to give a residue, which was then separated by silica gel column chromatography using petroleum ether and ethyl acetate (v/v = 30:1) as eluent to afford the target compounds 8a–8w, with yields ranging from 44% to 63%. All 23 pyrazole oxime derivatives 8a–8w were novel and the physical and spectral data for these compounds are listed below.
1,3-Dimethyl-5-methyloxy-1H-pyrazole-4-carbaldehyde-O-[4-(5-trifluoromethylpyridin-2-yloxy)phenylmethyl]-oxime (8a): White oil, yield 52%. 1H-NMR (CDCl3): δ 8.44 (s, 1H, Py-H), 8.08 (s, 1H, CH=N), 7.90 (d, J = 8.8 Hz, 1H, Py-H), 7.49 (d, J = 7.6 Hz, 2H, Ar-H and Py-H), 7.15 (d, J = 7.2 Hz, 2H, Ar-H), 7.01 (d, J = 8.8 Hz, 1H, Ar-H), 5.16 (s, 2H, CH2), 3.94 (s, 3H, OCH3), 3.62 (s, 3H, N-CH3), 2.28 (s, 3H, CH3); 13C-NMR (CDCl3): δ 165.8, 153.0, 152.8, 146.8, 145.5, 145.4, 141.6, 136.7, 136.6, 135.2, 130.0, 121.4, 111.3, 97.4, 75.4, 61.7, 33.6, 14.0. Anal. Calcd for C20H19F3N4O3: C 57.14; H 4.56; N 13.33. Found: C 57.28; H 4.39; N 13.16.
1,3-Dimethyl-5-ethyloxy-1H-pyrazole-4-carbaldehyde-O-[4-(5-trifluoromethylpyridin-2-yloxy)phenylmethyl]-oxime (8b). White solid, yield 55%, mp 48–50 °C. 1H-NMR (CDCl3): δ 8.42 (s, 1H, Py-H), 8.04 (s, 1H, CH=N), 7.88 (d, J = 8.8 Hz, 1H, Py-H), 7.47 (d, J = 7.2 Hz, 2H, Ar-H and Py-H), 7.14 (d, J = 6.8 Hz, 2H, Ar-H), 6.99 (d, J = 8.4 Hz, 1H, Ar-H), 5.14 (s, 2H, CH2), 4.16 (q, J = 6.8 Hz, 2H, CH2), 3.60 (s, 3H, N-CH3), 2.26 (s, 3H, CH3), 1.32 (t, J = 6.0 Hz, 3H, CH3); 13C-NMR (CDCl3): δ 165.8, 152.8, 152.1, 146.8, 145.4, 145.3, 141.7, 136.7, 136.6, 135.3, 130.0, 121.4, 111.3, 97.8, 75.3, 70.6, 33.6, 15.3, 14.0. Anal. Calcd for C21H21F3N4O3: C 58.06; H 4.87; N 12.90. Found: C 58.23; H 4.69; N 12.76.
1,3-Dimethyl-5-tert-butyloxy-1H-pyrazole-4-carbaldehyde-O-[4-(5-trifluoromethylpyridin-2-yloxy)phenylmethyl]-oxime (8c). White solid, yield 50%, mp 64–65 °C. 1H-NMR (CDCl3): δ 8.45 (s, 1H, Py-H), 8.01 (s, 1H, CH=N), 7.90 (d, J