The extraordinary richness of the reaction between diazomethane and tetracyanoethylene: can computational calculations shed light on old papers?

In the quest of the structure of the intermediate between Dand D-pyrazolines, the reactivity of these molecules tetrasubstituted by cyano groups in adjacent positions (3,3,4,4 or 4,4,5,5) has been explored in their neutral and protonated forms. Many reactions reported in the literature for pyrazolines have been studied and quantified (energies and transition states). Thirty-three structures of pyrazolines, their open-ring counterparts and their complexes are described. Acid–base equilibria, rotations, electrocyclic reactions and sigmatropic transpositions are reported.


Introduction
The reaction of diazomethane 2 with tetracyanoethylene 1, both very common compounds, has only been studied two times. In 1962, Bastús and Castells reported the reactions of Fig. 1 (their numbering of formulae is different). 1 They indicated that 3 can be explosive and this probably prevented other authors from repeating its preparation. D 1 -pyrazoline 3 (4,5-dihydro-5H-pyrazole-3,3,4,4-tetracarbonitrile) was isolated and it spontaneously evolved nitrogen to yield 1,1,2,2-tetracyanocyclopropane 4 that was already known having been prepared by other methods. 2 Compound 3 was washed with benzene to eliminate all traces of 1 and was slowly dissolved in dry ether to yield a compound to which structure 5 (2,2,3,3-tetracyano-1,5-diaza-bicyclo[2.1.0]pentane) was assigned. Compound 5 when treated for about 90 min with a 5% solution of 1 in dry ether afforded D 2 -pyrazoline 6. The isomerization of 6 to 5 was performed using wet ether or dry ether containing traces of hydrogen chloride. Both substances can be kept for several weeks without alteration. According to Bastús and Castells, the 6 to 5 isomerization involves the pyrazolinium cation 6bH + . 1 The role of TCNE (1) in the 5 -6 isomerization was assigned to a 1 : 5 complex.
Huisgen et al. repeated the reaction. 3,4 They cited Banús and Castells but they isolated only 4 and 6. Leaving aside 4 (also 6bH + was not characterized), we have summarized in Table 1 all the available information on the compounds in Fig. 1.
Calatroni and Gandolfi reported a series of reactions that are related to the work of Banús and Castells (Fig. 2). 5 D 1 -pyrazoline I reacted in two ways with TCNE (1) to afford a charge-transfer complex II and adduct III that was not isolated nor identified. They assumed that the reaction I " III is fast and reversible. Besides, TCNE promotes the isomerization I -IV in agreement with Castells' results. A reaction product V was postulated corresponding to the reaction of IV with TCNE.
Calatroni and Gandolfi indicated in their paper that they wanted to determine the crystal structure of II (yellow crystals) but they probably failed because no structure like II was reported in the CSD. 6 The only one that bears resemblance to II is FEJDUT (VI), see  We decided to study theoretically structures 3, 5 and 6 and their corresponding protonated cations (Fig. 4) as well as some concerted reactions related to Woodward-Hoffmann rules. [8][9][10]

Computational details
The geometries of pyrazolines (Pz) and pyrazolinium cations (PzH + ) have been fully optimized using the functional B3LYP 11 and the 6-311++G(d,p) basis set 12 in the spin restricted formalism as implemented in the Gaussian 16 package (the coordinates of all the optimized geometries are gathered in the ESI †). 13 The minimum energy and transition state structures of all compounds were characterized using frequency analysis. The solvent effects have been evaluated by re-optimizing the structures at the B3LYP/6-311++G(d,p) level and using the self-consistency reaction field (SCRF) method 14 based on the polarized continuum model (PCM) of Tomasi and co-workers 15 in ethanol, diethylether and benzene as solvents using the standard parameters provided by the Gaussian-09 program. For the infrared spectra, the frequencies have been scaled by a factor of 0.9679. 16 Absolute chemical shieldings have been calculated with the GIAO approximation 17 and then transformed into chemical shifts using empirical equations. 18 The static intrinsic reaction coordinates (IRCs) 19 were analyzed in two cases. Diazomethane, a 1,3-dipole with octet stabilization, 10 reacts with CC double and triple bonds to afford pyrazolines and   pyrazoles. 20 For the reaction 1 + 2 -3, D 1 -pyrazoline 3 lies 17.1 kJ mol À1 higher than the potential surface minimum, the D 2 -pyrazoline 6 ( Table 2), and the barrier with regard to 3 is 73.8 kJ mol À1 . For the reaction between diazomethane and different olefins, Ess and Houk calculated barriers between 57 and 70 kJ mol À1 . 21 Calculated IR, the NMR properties of compound 5 are reported in Table 3.

Results and discussion
This compound has two conformations depending on the position of the NH bond, towards or out of the ring (Fig. 5).

4,5-Dihydro-1H-pyrazole-4,4,5,5-tetracarbonitrile (6)
D 2 -pyrazoline 6 is the most stable of the three isomers ( Table 2). The 3 -6 tautomerization involves a 1,3 CH to NH prototropy. A direct proton transfer has a high barrier that in similar fivemembered rings prevents this mechanism, while the following one involves assistance by solvent molecules (one or two) such as water or alcohols with a low barrier. 22 The calculate TS between 3 and 7 is 289.7 kJ mol À1 and between 7 and 8, it is 277.8 kJ mol À1 . The experimental IR spectrum (KBr) and the calculated bands (gas phase) agree exceptionally well assuming that there is an intercept. The corresponding linear regression line is: We have removed Castells' bands at 1592 and 1595 cm À1 assigned to NH bending because according to eqn (1) they should appear at 1269 and 1417 cm À1 . On the other hand, if the 1592 cm À1 band is assigned to NQNQC of 7, the agreement is good (fitted value 1544 cm À1 ). In conclusion, on IR grounds, structure 5b should be rejected while structure 7 is acceptable.
The available 13 C NMR data of compound 6 3 agree with the calculated values according to the linear regression, The most different signals of the three compounds are the 15 N chemical shifts of the ring nitrogen atoms: +91.3 and +105.4 ppm for 3; À236.8 and À29.5 for 6; À285.8 and À297.0 for 5b.

3.4.
A first attempt to find a more stable structure for intermediate 5 We have considered that the intermediate instead of being diaziridine 5b could be zwitterion 7 (Fig. 6). Compound 7 is much more stable than 5b, 88.3 vs. 151.0 kJ mol À1 , but still too high for allowing the 6 -5 backward reaction. Solvent effects (amongst them those used by Castells 1 ) calculated using the PCM model decrease the difference by a small amount: benzene, 79.7; diethylether, 75,4; and ethanol, 70.4 kJ mol À1 .
Calculation of the IR spectrum of compound 7 leads to scaled bands at 3419 cm À1 (nNH) and 1523 cm À1 [(NQNQC) AE ].  Using eqn (1), these values in KBr should be 3319 and 1520 cm À1 . Castells et al. did not report a CQN band but 3319 cm À1 is consistent with the experimental value of 3295 cm À1 .

Acid-base equilibria, rotations, electrocyclic reactions and sigmatropic transpositions
Two reviews provide an overview of the electrocyclic reactions and sigmatropic reactions involving pyrazolines. 23,24 We have already reported the 1 + 2 -3 reaction with a TS of 73.8 kJ mol À1 with regard to 3 (IRC, Fig. 7). Other reactions that we have studied theoretically are reported in Fig. 8 and 11 together with acidbase equilibria (protonation) and with rotations about single bonds. Note that the loss of dinitrogen to afford 4 is a very exergonic reaction. Regarding Fig. 8, as we have already commented, the proposed structure of intermediate 7 lies 88.3 kJ mol À1 higher than 6 (71.2 kJ mol À1 above 3). A 1,2 (C to N) transfer of hydrogen has a barrier of 289.7 kJ mol À1 ; the subsequent 1,2 (C to N) transfer of hydrogen to afford 6 has a barrier of 289.7 kJ mol À1 . These very high barriers do not correspond to real pathways because proton transfers assisted by solvent molecules have much lower barriers (see previous discussion).
Ring opening of 7 to 8a should occur thermally in a disrotatory way. 23,24 However, the geometry of 8a (Fig. 9) allows only a conrotatory mechanism with a barrier of 71.9 kJ mol À1 with regard to 7. A rotation about the single NC bond affords the more stable 8b while prototropy results in the less stable azine 8c. Note that according to the Woodward-Hoffmann rules, 8-10 a conrotatory mechanism is thermally forbidden.
We then considered that the complex 1 : 5 ( Fig. 1) could be the cycloaddition product of 7 (an azomethine imine) with 1 but the resulting 9 is a bicyclic compound very destabilized by the eight cyano groups (Fig. 8).
Loss of HCN from 6 affords 15. According to Rodríguez Morán, when they carried out the reaction of 1 + 2 to afford 6, 3 release of hydrogen cyanide was observed. 27 The last reaction we have studied is the cycloaddition of diazomethane 2 on D 2 -pyrazoline 6. It is known that D 2 -pyrazolines react with 2 to afford 1,2-diazabicyclo[3.1.0]hexanes, related to 14. 28 These compounds should result from the loss of dinitrogen of either 13a or 13b. According to the literature, the cycloaddition of diazomethane on imines (or Schiff bases) results in 1,2,3-triazolines related to 13a. 29 In our case, the reverse addition leads to a more stable compound, 1,3,4-triazoline 13b instead of 1,2,3-triazoline 13a; this is probably related to the fourth N atom, i.e. our compounds are hydrazones, not imines. The loss of dinitrogen to form 14 is strongly favored (À151.4 kJ mol À1 ).
We then decided to carry out calculations parallel to those of Calatroni and Gandolfi. 5 They are reported in Fig. 10.
Compound 16 is the complex of 1 and 3 (compared with the 1 : 5 complex of Fig. 1 and with complex II of Fig. 3); it is located 7.1 kJ mol À1 above 6 and at À10.0 kJ mol À1 from 3. The zwitterion 17 (compare with III of Fig. 3) is not stable and reverts to 16 both in the gas phase and in ethanol (PCM); this sheds doubt on the hypothetical III structure. 5 Structure 19 lies 66.0 kJ mol À1 above 6, also an indication that structure V in Fig. 3 was probably never formed. Charge-transfer complex 18a is not stable and evolves to the hydrogen-bonded complex 18b; it is located at À20.1 kJ mol À1 from 6. The stabilization results from the hydrogen-bond and from a CN/CN stacking between both molecules. Fig. 6 The equilibrium between 5a and 7. View Article Online Fig. 9 The open ring compounds 8a and 8b their cations 8aH + and 8bH + and compounds 9 and 15. It is also interesting to study the reactivity of the conjugated acids of structures of Fig. 8 because Castells et al. indicated that 5/6 equilibration was acid catalyzed (Fig. 11). 1 Usually, D 2 -pyrazolines protonate on N1 (type a) 30 but probably due to the presence of the four cyano groups, in this case, the protonation takes place preferably on N2 (6bH + ), which corresponds to the hypothesis of Castells et al.. However, the most important finding is that 5H + (even the most stable 5bH + , Table 2, 154.3 kJ mol À1 ) has an energy that makes it impossible to isomerize 6H + into 5bH + . The cations have relative energies similar to those of the neutral molecules, 6 o 3 { 5. Although the synthesis of D 2 -pyrazolines protonated on N1, 6aH + , from 1 and 10H + , is unknown, both the stability and the barrier made it a feasible possibility.
One of the reactions of D 2 -pyrazolines protonated on N1 is to open into compounds related to 20b; this has been proven  experimentally. 30b However, in the present case, the optimization led to a quaternary azetinium (2,3-dihydroazetium) cation 20a, still too high in energy. The loss of HCN from 6bH + generates a protonated isopyrazole 15H + (1H-isopyrazolium) that is slightly less stable. Protonation of the cyano groups leads to cations 6cH + (CN at position 4) and 6dH + (CN at position 5) of similar energies to 6aH + , i.e. the CN groups are similar to the amino group of a pyrazoline. This is a surprising result because the PA of CH 3 CN is 787 kJ mol À1 and that of CH 3 NH 2 is 899 kJ mol À1 , 31 but the role of the four CN groups and the D 2 -pyrazoline structure may modify the proton affinity; it is known that cyano derivatives can be superbases. 32 From the most stable of these cations, loss of protonated hydrogen cyanide results in the formation of 15 with an energy of 59.9 kJ mol À1 .

Conclusions
Castells intermediate 5 (Fig. 1) 1 is characterized by IR in KBr by a stretching NH at 3295 cm À1 and a bending NH at 1592 cm À1 as well as by the absence of a CQN band although these last bands appear in D 2 -pyrazolines at 1555-1570, 33 1557, 34 1560, 35 1580-1592, 36 1600 37 and 1618-1622 cm À1 . 38 Therefore, the compound has a NH group, i.e., it is not a D 1 -pyrazoline, but the presence or absence of a CQN band is dubious.
The reactivity of intermediate assumed to be 5 is that it never affords 3 but it is in equilibrium with 6. Light and TCNE isomerize 5 to 6; water, HCl and time isomerize 6 to 5. 1 If pure 6 can be transformed into 5, this latter must be an isomer not containing TCNE.
Of all the compounds that we have studied, which is the best candidate? Our calculations of IR spectra and its high energy show that 5 can be definitely excluded. That the intermediate could be a salt was possible when HCl was used but not with wet ether. If, by an experimental error, TCNE still remained in the reaction medium, 16 is a good candidate due to its energy and absence of barrier.
Finally, the best candidates because they are consistent with the IR data are 7, as already discussed, and 8b. The linear equation relating both variables is: Exp. = (115 AE 31) + (0.93 AE 0.1) Calc., n = 7, R 2 = 0.999, RMS residual = 27 cm À1 The problem is that the available information is too scarce, there is not even the complete IR spectrum of 5 let alone the 1 H NMR data. This added to the non-reproducibility of the preparation of 5 3,27 made it impossible to ascertain its structure. Charge transfer complexes are colored (yellow/red) but no indication of the color is reported. 1 On the other hand, a plethora of structures and reactions surrounding the chemistry of tetracyanoethylene (1) has been explored covering different aspects of the chemistry of pyrazolines 39 that would prove useful in related studies because several of them were not experimentally known.

Conflicts of interest
There are no conflicts to declare.