[3 + 2]-Cycloaddition reaction of sydnones with alkynes

This review covers all known examples of [3 + 2]-cycloaddition between sydnones and both terminal as well as internal alkynes/cycloalkynes taken from literature since its discovery by Huisgen in 1962 up to the current date. Except enumeration of synthetic applications it also covers mechanistic studies, catalysis, effects of substituents and reaction conditions influencing reaction rate and regioselectivity.


Introduction
Since Huisgen's discovery of the [3 + 2]-cycloaddition between 3-substituted sydnones and both terminal as well as internal alkynes [1,2] many researchers have tried to utilize this synthetic approach for the synthesis of polysubstituted 1,2diazoles (pyrazoles, indazoles). However, until 2013 when Taran's group introduced the regioselective Cu(I)-phenanthroline catalysis [3] this method was of limited value due to the harsh reaction conditions and sometimes also due to low regioselectivity in those cases when a non-symmetrical alkyne was employed as a reactant. Surprisingly, until the fall of 2017, no comprehensive work concerning this important topic was published. This encouraged us to write this review. During its completion a new feature article bridging this gap was published by Taran et al. [4]. In order to avoid duplication our review is therefore focused in more detail on thermal, photo-chemical as well as metal-catalyzed reactions of sydnones with alkynes and factors that influence the yield and ratio of both possible regioisomers and also the kinetics and mechanism of this cycloaddition reaction.
It was also observed, that formation of isomeric pyrazole-4,5dicarboxylates (B) can sometimes accompany the production of pyrazole-3,4-dicarboxylates (A) under thermal conditions [33] although their formation is not photoinduced (cf. next chapter) because the reaction also takes place in the absence of light. Depending on the temperature, a new reaction pathway involving benzylic group migration, CO 2 extrusion and final cycloaddition was proposed (Scheme 4). Kinetics and mechanism of thermal cycloaddition The kinetics and reaction mechanism of the thermal cycloaddition between 4-methyl-3-phenylsydnone and DMAD was first studied by Huisgen and Gotthardt [54] in p-cymene at 90-110 °C. They found the cycloaddition to be overall second order and its activation entropy ΔS ≠ = −130 J·mol −1 ·K −1 showed association character of the rate-limiting step with a rel-atively tight transition state. Moreover, for the cycloaddition of the structurally similar ethyl phenylpropiolate in various solvents only a small decrease of the bimolecular rate constant with increasing solvent polarity (in terms of relative permittivity) was observed excluding a transition state having a polarized character. Finally, substitution effects in the 3-(4-substituted phenyl) group of sydnones were studied and a relatively low Hammett reaction constant ρ ≈ +0.8 was estimated from Scheme 5: Mechanism of thermal cycloaddition between sydnones and alkynes. four derivatives (MeO, Me, H and Cl). An even smaller dependence of the rate constants on the solvent polarity and substituent effect sensitivity (ρ ≈ +0.3 to +0.4) was described [55] for reactions of 3-(4-substituted phenyl)sydnones with more reactive DMAD while the activation entropy (ΔS ≠ = −106 to −121 J·mol −1 ·K −1 ) remained similar. The reaction mechanism (Scheme 5) consistent with these kinetic measurements involves rate-limiting formation of a bicyclic intermediate via a concerted [3 + 2]-cycloaddition followed by its very fast decomposition (extrusion of CO 2 ) via a retro-Diels-Alder [4 + 2]-cycloaddition. The almost spontaneous extrusion of CO 2 is caused by an energetically favorable aromatization occurring in this step leading to the formation of the stable pyrazole ring. Both reaction steps are also compatible with Woodward-Hoffmann rules, taking into account orbital symmetry considerations [56].
Three types of [3 + 2]-cycloadditions (labelled I-III) are known from the literature [57] each differing in the frontier molecular orbital energies between the dipole and dipolarophile. While for type I (HOMO-controlled) combining a high-lying dipole HOMO with a dipolarophile LUMO the reaction is accelerated by electron-donating substituents on the dipole and electronwithdrawing substituents on the dipolarophile (both lowering the HOMO-LUMO energy gap), for type III (LUMO-controlled) combining a low-lying dipole LUMO and a dipolarophile HOMO where substituent effects are completely opposite. For type II cycloadditions in which two-way interac-tions between the dipole HOMO and the dipolarophile LUMO or the dipole LUMO and the dipolarophile HOMO are possible -due to similar energy gaps -both electron-rich as well as electron-poor dipolarophiles/dipoles react more quickly than parent (unsubstituted) ones. Using semi-empirical quantum calculations (CNDO/2), Houk et al. [58] calculated average HOMO/ LUMO energies for azomethine-imines (ε HOMO = −8.6 eV and ε LUMO = 0.3 eV) and predicted that the ε LUMO for structurally related sydnones containing an electron-withdrawing -COO-motif should be even much lower suggesting a LUMOcontrolled reaction (type III). Such a prediction seems to be correct for reaction of 4-(substituted phenyl)sydnones with DMAD for which positive Hammett ρ-values were observed [54,55]. On the other hand Huisgen and Gotthardt [54] measured bimolecular rate constants for the above-mentioned reaction of 4-methyl-3-phenylsydnone and various acetylenes in p-cymene at 140 °C ( Table 2) and found a reactivity sequence corresponding rather to type II or even type I cycloadditions.
The most reactive were electron-poor alkynes (acetylene(di)carboxylates, benzoyl phenylacetylene) while electron-rich alkynes (tetradec-1-yne, 1-phenylpropyne) were much less reactive. Unfortunately, the reaction rate constant was not measured for the reaction with acetylene itself. However, on the basis of the published [1,2] synthetic protocol (acetone, 170 °C, 25 h) it appears that this cycloaddition is very slow and requires a higher temperature. Recently [42][43][44], a kinetic investigation was performed for the cycloaddition of various sydnones with strained cycloalkynes such as bicyclo[6.1.0]non-4-yne-9-methanol (BCN) or 3,3,6,6tetramethylthiacyclohept-4-yne (TMTH). It was found that the reaction of BCN with 3-(4-substituted phenyl)sydnones roughly obeys a Hammett correlation with ρ ≈ +1.35 ± 0.25 [43] thus indicating a type III mechanism. However, the effect of substituent in position 4-of 3-phenylsydnone is ambiguous. While all halogens substantially accelerate the reaction rate (F > Cl > Br > I) other substituents cause up to tenfold deceleration (H > Me > CF 3 > CN) regardless of their polar effects [43,44]. Steric factors cannot explain the influence of 4-substituent because 4-phenylsydnone reacts equally as unsubstituted one. The most reactive 4-fluoro-3-phenylsydnones [44] were found to react with BCN and TMTH in two kinetically independent reaction steps corresponding to fast formation of the addition intermediate and its slow decomposition to pyrazole and CO 2 . Such ambiguous substitution effects are therefore worthy of further investigations.
The yields (Table 3) are generally lower than those of reactions performed under thermal conditions -most probably due to the lower stability of the key intermediate -N-phenylnitriliminewhich can undergo dimerization or reverse trapping of evolved CO 2 . Yields are always much better for 3,4-diarylsydnones for which the corresponding N-phenylnitrilimine is resonancestabilized. The yields also depend on the photoreactor construction [64]. For example 1,3-diphenylsydnone reacts with DMAD in a batch reactor (Rayonet) under 300 nm irradiation to give only 29% of dimethyl 1,3-diphenylpyrazole-4,5-dicarboxylate while in a wetted-wall photo reactor (Normag) the yield is increased up to 84% (at 17 °C in DCM).

Thermal reaction of sydnones with terminal alkynes
As early as in his first work [1] dealing with sydnone-alkyne cycloaddition Huisgen et al. found that some non-symmetrical alkynes (oct-1-yne, phenylacetylene and especially methyl propiolate) gave mixture of both pyrazole regioisomers. The following Table 4 summarizes all known examples [1,2,8,20,24,[32][33][34]36, where the ratio of both possible       regioisomers or at least chemical yield of the major regioisomer was given.
The first people who qualitatively discussed the regioselectivity on the basis of semi-empirical quantum calculations was Houk et al. [94] who (except of above-mentioned low-lying LUMO of sydnone [58]) calculated sydnone LUMO terminal orbital coefficients and found them to be almost identical thus indicating low selectivity in LUMO-controlled cycloadditions (type III). However, Gotthardt and Reiter [8] who were also dealing with regioselectivity of sydnone cycloadditions with methyl propiolate pointed out that the reason for the lower regioselectivity can also be attributed to the low-lying HOMO of this dipolarophile. While for the LUMO-controlled reaction (type III) only the 3-substituted pyrazole is expected to be the main product, for the HOMO-controlled reaction (type I) 4-substituted pyrazole should be formed preferentially (Scheme 7 adapted from reference [8]). The combination of both reaction pathways (type II) therefore gives a mixture of 3-and 4-substituted pyrazoles. This situation is typical especially for cycloadditions with alkyl propiolates (cf.  [8]. The presence of the nitro group(s) lower(s) the LUMO energy of the sydnone and a type III mechanism is slightly favored. The same trend [8] can be seen from the substitution effect in position 4 of the starting 3-phenylsydnone when reacted with methyl propiolate (Table 4, entries 6, 7, 23-26) but almost no influence is observed for reactions with phenylacetylene (Table 4, entries 3, 4, 69, 91). Generally, it can be concluded that any substituent in position 4 reduces the regioselectivity.
The last factor that influences the ratio of isomers involves the thermodynamic conditions -namely the temperature and pressure. A nice temperature/pressure-selectivity study of the cycloaddition of 3-phenylsydnone with methyl propiolate was undertaken by McGowin et al. [93] in supercritical CO 2 . At 7.6 MPa they found a linear dependence between the natural logarithm of selectivity (defined as the 3-/4-isomer ratio) and the reaction temperature. In accordance with the common reactivity-selectivity principle, the higher temperature lowers selectivity from 5.52 (i.e., 85:15) at 80 °C to 3.14 (i.e., 76:24) at 160 °C but increases sydnone conversion and pyrazole yield. On the other hand, a variation of the pressure from 7.6 to 30.4 MPa at constant temperature (80 °C) caused a decrease in the total yield by approximately 50%, with slightly increased selectivity (from 4.96 to 6.56). Lowering of the yield with increasing pressure confirms the reversibility of the first step (see Scheme 5) because of retardation of CO 2 cleavage from the bicyclic intermediate. Such reversibility was also suggested by Harrity et al. [92] on the basis of quantum calculations. While the formation of the bicyclic intermediate was calculated to be only slightly exergonic (−3.3 kcal·mol −1 ) the overall reaction is highly exothermic (−108.2 kcal·mol −1 ).
These results show that minor differences in selectivity published by various authors (e.g., entry 3 in Table 4) can be ascribed to changes in temperature (different boiling point of benzene, toluene, xylenes, DCB, …) used in synthesis. In several cases (e.g., entries 83-86 [84] and 141 and 142 [90] in Table 4) too high temperature (200 °C) can contribute to a substantial drop of selectivity. It is also known that some sydnones start to decompose at temperatures exceeding 180 °C [74] which can cause lowering of the pyrazole yield.
The Table 5 again summarizes all the examples found, including reaction conditions from which we have come to several conclusions.

Scheme 9:
Unsuccessful reaction with phenylpropiolic acid.    According to frontier molecular orbital theory both combinations, i.e., HOMO(dipole)-LUMO(dipolarophile) (type I) and HOMO(dipolarophile)-LUMO(dipole) (type III) should lead to the production of individual regioisomers (cf. Scheme 7). All substituents R 1 -R 4 have an influence on the HOMO-LUMO energy gaps and consequently on the ratio of both isomers especially in those cases when such energy gaps are similar. Again, the substituents on the alkyne (R 3 , R 4 ) have great influence on the outcome of the reactions. Strong electron-withdrawing substituents R 3 (COOR, COR, SO 2 Ar, CF 3 ) in combination with any aryl (R 4 : Ph, substituted Ph, heteroaryls) strongly prefer position 4 in the final pyrazole ring (see entries 2-6, 9, 10, 14, 21-28, 37-45, 64-83 in Table 5) when reacting with 4-unsubstituted 3-phenylsydnones or 3-alkylsydnones (see entries 6, 29-31 in Table 5). Both these substituents jointly lower the LUMO while their influence on energy of the HOMO is contradictory. Consequently, the type I mechanism is clearly preferred. If R 4 has also similar electron-withdrawing ability (e.g., CHO, CF 3 , see entries 17, 18, 46 in Table 5 or R 4 = halogen, see entries 12, 120-123 in Table 5 and even R 4 = CH 2 OH, see entries 19 and 20 in Table 5) then almost complete loss of selectivity occurs and the ratio of both regioisomers is close to 50:50. Markedly reversed regioselectivity is observed only for R 4 = CH(OMe) 2 which is probably connected with the higher steric demands of this group.
A substitution in position 3 of the sydnone has a much smaller influence on the regioselectivity which is in accordance with longer distance between the substituent and both dipole termini. Substitution of the 3-phenyl ring (e.g., entries 26-28, 37-45, 50, 53, 54 in Table 5) or even change of the whole 3-substituent (alkyls vs phenyl, see entries 29-31, 86, 87 in Table 5) cause no or only a minor change in the ratio of both regioisomers. In some cases the same conclusion can be drawn for changes of the substituent in 4-position of the sydnone (cf. entries 2, 7, 8, 89, 93, 94 or 88, 91, 92 in Table 5). On the other hand, the presence of a substituent can sometimes increase as well as decrease the ratio of both regioisomers (cf. entries 26 and 32-35) for no easily discernible reason.
The last type of non-symmetrical internal alkynes to be considered are cycloalkynes. Their strain-promoted reactions again proceed quickly under mild reaction conditions (cf. section concerning symmetrical internal alkynes) but their regioselectivity is generally low, which is in accordance with the re- activity-selectivity principle. The first example was described [44] by Taran's group in 2016 when they observed an ultrafast reaction of 6-[11,12-didehydrodibenzo[b,f]azocine-5(6H)-yl]-6oxohexanoic acid with 4-fluoro-3-(4-methylphenyl)sydnone (Scheme 12). Unfortunately the regioselectivity of the reaction was not specified.
An aryne generation (Scheme 13) was also used for the synthesis of a key intermediate of the potent antitumor PARP inhibitor -niraparib -containing an indazole core [113]. A substituted 2,3-aryne was generated in situ from (siloxy)benzocyclobutenes and CsF but the regioselectivity was poor: a ratio of both possible regioisomers of 45:55 was obtained.
The in situ generation of arynes or six-membered cycloalkynes from their corresponding trimethylsilyl triflates was recently used by Garg et al. [115] and Bräse et al. [116] in expanding the utility of oxygen-or nitrogen-containing strained heterocycloalkynes (Scheme 15) but the regioselectivity was poor in most cases.
Scheme 15: Reaction of sydnones with heterocyclic strained cycloalkynes.  [64] Photochemical reaction of sydnones with non-symmetrical alkynes As mentioned in the previous section, Gotthardt and Reiter [63,64] studied the photochemical reaction of sydnones with terminal alkynes. They have also studied the reaction with phenylacetylene, methyl propiolate and ethyl phenylpropiolate in a batch reactor under irradiation with 300 nm light ( Table 6).
The formation of both regioisomers a and b was observed when the most reactive methyl propiolate was used as a reactant. Moreover, the ratio (16:84) obtained from starting 3,4-diphenylsydnone is similar with those obtained from 1,3-diphenylnitrilimine independently generated either from 2,5-diphenyltetrazol or from N-phenylbenzenecarbohydrazonoyl chloride. This observation clearly supports the mechanism depicted in Scheme 6. The distribution of both regioisomers qualitatively agrees with  [3,120] the proposal of Houk et al. [94] combining the dipole HOMO with the dipolarophile LUMO (type-I mechanism).

Copper-catalyzed reaction of sydnones with terminal alkynes
A substantial breakthrough in the field of 3-arylsydnone-terminal alkyne cycloaddition was achieved by Taran's group in 2013 [3]. They developed a regioselective Cu(I)-phenanthroline-catalyzed variant of this reaction (i.e., copper-catalyzed sydnone alkyne cycloaddition; henceforth called CuSAC) enabling regioselective formation of 1,4-disubstituted pyrazoles under much milder reaction conditions (in various solvents including aqueous solution at 25-60 °C, Table 7) than previously used for its thermal-mediated counterpart. Such mild reaction conditions together with very high and reverse regioselectivity and efficiency (in most cases 85-99% yields) makes the CuSAC reaction a very good alternative to the well-established azide-alkyne click-reaction [117] useful not only in classical organic synthesis but also in bioconjugation applications. Moreover, a further improvement was later devised by the same authors, which avoids the highly toxic N-nitroso-N-phenylglycine, (precursor of sydnone) and involves a three-step onepot transformation of starting N-phenylglycine to the corresponding pyrazole [118].
There are several limitations of the CuSAC reaction. First, it apparently fails with 3-alkyl sydnones and also with almost all 4-substituted 3-phenylsydnones except 4-F [44], 4-Cl and 4-Br derivatives [119]. However, this fortunate exception gave the further possibility to exchange halogen (especially bromine) by either an aryl, alkyl or alkenyl group via Suzuki coupling reaction with boronic acids to give otherwise rarely available 1,4,5trisubstituted pyrazoles [119]. The second limitation is that the CuSAC reaction proceeds only with terminal alkynes. The latter fact clearly indicates some kind of participation of the alkyne's slightly acidic terminal hydrogen in the reaction mechanism. Indeed, as early as in his primary paper [3] Taran suggested formation of Cu(I) acetylide (for additional information concerning reactions involving Cu(I) acetylides see references [121,122]) as the key species coordinating N2 of the sydnone through the Cu atom in the transition state. This suggestion was supported by Gomez-Bengoa and Harrity et al. [92] who performed thorough quantum calculation of various transition states involving different modes of interaction between 3-phenylsydnone and Cu(I) phenylacetylide (Scheme 16) and found Taran's suggestion as the most plausible because of the lowest activation free energy (ΔG ‡ = 25.4 kcal·mol −1 ) and due to the observed 1,4-regiocontrol. Intrinsic reaction coordinate (IRC) calculations also showed concerted but asynchronous for-  cleavage of one of the two copper atoms. The copper triazolide formed in this way is then hydrolyzed to the final triazole. The same presumption (Scheme 17) concerning the role of the two Cu atoms was also adopted by Taran in his newer paper [119] but no experimental evidence for this mechanism has been given yet.
From previous studies it is known that for the spherically symmetric d 10 Cu(I) ion, the common geometries are two-coordinate linear, three-coordinate trigonal planar, and four-coordinate tetrahedral [124]. Phenanthrolines form with Cu(I) at 1:1 ratio three-coordinated trigonal planar complexes or at 2:1 ratio tetra-coordinated tetrahedral complexes [125]. If Cu 2 (CN) 2 (in which CN is isoelectronic with acetylide) is employed as a Cu(I) source then three-coordinated trigonal planar polymeric arrangement was observed [126]. From this observations it appears that mono-or dimeric three-coordinated trigonal planar Cu(I)-acetylide-phenanthroline complex should be the reactive species during CuSAC. This idea was supported by the fact that during the reaction of 4-bromosydnones with 4-phenylbut-1-yne [119] tridentate tris(benzimidazole) ligands completely failed and tris(triazole) ligands gave only poor to moderate yields (16-65%) even at 100 °C, whereas all the bidentate ligands (phenanthrolines L 1 , L 2 and diimidazo[1,2a:2',1'-c]quinoxalines L 3 -L 6 ) were found to be more efficient both in terms of the isolated yield as well as the regioselectivity (see entry 39 in Table 7). From the comparison of phenanthro-line (L 1 , L 2 ) and diimidazo[1,2-a:2',1'-c]quinoxaline (L 3 -L 6 ) complexes it appears that the higher angle between the two coordinative nitrogen atoms may have a positive impact on the catalytic efficiency.
Gomez-Bengoa and Harrity et al. [92] also inspected the role of Cu(I)/Cu(II) salts as well as other Lewis acids which could strengthen the electrophilicity of the starting sydnone under thermal reaction conditions. They found two competitive catalytic routes leading to different cycloaddition products. According to their original presumption some Lewis acids (TMSOTf < Zn(OAc) 2 < MgBr 2 < Cu(OTf) 2 ) catalyzed the thermal reaction of phenylsydnone with phenylacetylene to give the expected 1,3-diphenylpyrazole in a ratio >10:1 over the 1,4-diphenyl isomer. Quantum calculations and IR measurements performed for the most active Cu(OTf) 2 have shown that this salt coordinates to the sydnone oxygen carrying a negative charge which leads to an energy decrease of the sydnone LUMO and an increase of its electrophilicity. Also computed activation free energy (ΔG ‡ = 25.4 kcal·mol −1 ) for the rate-limitting [3 + 2]-cycloaddition step leading to the 1,3-isomer was substantially lower if compared to the uncatalyzed reaction pathway (ΔG ‡ = 32.5 kcal·mol −1 ).
If other Cu(II) salts were used as a catalyst then the ratio of 1,3-/1,4-isomers gradually changed from 90:10 to 3:97 (Table 8).  A completely different ratio of both isomers was observed when Cu(II) carboxylates and acetylacetonates were employed instead of Cu(OTf) 2 . This was explained by different operating mechanisms. While Cu(OTf) 2 , Cu(TFA) 2 and Cu(BF 4 ) 2 behave mainly as Lewis acids, other Cu(II) salts/complexes preferentially oxidize one equivalent of phenylacetylene to give 1,4diphenylbuta-1,3-diyne (isolated in 80% yield) and the evolved Cu(I) salt then forms Cu(I) acetylide with a second equivalent of phenylacetylene. Thus formed Cu(I) acetylide is then responsible for gradual increasing of 1,4-pyrazole occurrence. Quantum calculations [92] and IR measurements performed for Cu(OAc) 2 also show that the Lewis acid character of this salt is less pronounced and formation of the 1,3-diphenylpyrazole necessitates a much higher activation free energy (ΔG ‡ = 41.4 kcal·mol −1 ) than for the uncatalyzed reaction. Formation of 1,4-diphenylpyrazole through Cu(I)-acetylide addition is then the clearly preferred reaction pathway. Moreover, Cu(OAc) 2 acts as a very good catalyst not only in the reaction of parent phenylsydnone with phenylacetylene [92]. After appropriate prolongation of the reaction time it delivers the corresponding 1,4-disubstituted pyrazoles in good to excellent yields and with a regioselectivity ratio exceeding 95:5 (Table 9). It is worth noting that 3-benzyl sydnone (representative of otherwise unreactive 3-alkylsydnones) reacts with the highly reactive ethyl propiolate to give ethyl 1-benzylpyrazole-4-carboxylate in good yield. Copper(II) acetate anchored on a modified silica gel can also serve as an efficient catalyst in batch reactor or if housed in stainless steel cartridges [127] in continuous-flow conditions (Table 10). Again, the 4-substituted pyrazole is preferentially formed.

Conclusion
Since its discovery in the sixties of the last century, the thermal [3 + 2]-cycloaddition of sydnones with alkynes represents a valuable synthetic tool for the preparation of polysubstituted pyrazoles and indazoles despite the limitations: the need for high temperatures (90-170 °C) and sometimes poorer regioselectivity. These obstacles can be surpassed either by suitable substitution (activating electron-withdrawing groups, removable silyl or carboxylate groups or replaceable boronic esters) or by efficient catalysis using Lewis acids. Preferential formation of 1,3-di-or 1,3,5-trisubstituted pyrazoles (>90:10) is observed in most cases when a terminal alkyne was used as a reactant. On the other hand, the recent discovery of Cu(I) catalysis in the sydnone-alkyne cycloaddition (CuSAC) enables regioselective formation of complementary 1,4-disubstituted or 5-halogeno-1,4-disubstituted pyrazoles under very mild reaction conditions (aqueous t-BuOH solution at 60 °C) and can be considered as a good illustration of the click-reaction. Another important example of sydnone cycloaddition involves a very fast reaction with strained seven-or eight-membered cycloalkynes (strainpromoted sydnone alkyne cycloaddition; SPSAC) which takes place without any catalyst and at ambient temperature. Such mild reaction conditions, (ultra) fast and unambiguous product formation make SPSAC useful in bio-orthogonal applications and competitive in comparison with analogous strain-promoted azide-alkyne cycloaddition (SPAAC). The last possibility of how to influence the cycloaddition between sydnones and alkynes involves photochemical performance of this reaction. Under UV-irradiation sydnones form the corresponding unstable nitrilimines which then undergo [3 + 2]-cycloaddition to give pyrazoles carrying substituents originating from alkynes in positions 4 and 5 instead of 3 and 4. Yields of this photochemical reaction are mostly lower than 50% which makes this method less convenient.