All the reagents have been commercially acquired and have been used without further treatment:

・ Thiosemicarbazide. Sigma-Aldrich, St. Louis, MO, USA, 99%;

・ Acetophenone. Sigma-Aldrich, St. Louis, MO, USA, 99%;

・ 4-chloroacetophenone. Sigma-Aldrich, St. Louis, MO, USA, 97%;

・ 4-nitroacetophenone. Sigma-Aldrich, St. Louis, MO, USA, 98%;

・ 4-methylacetophenone. Sigma-Aldrich, St. Louis, MO, USA, ≥95%.

Figure 1 shows the total ion chromatogram (TIC) after injection of Compound IV.

As it can be seen, there are two substances. Peak purity allowed discarding co-elution in any of the observed peaks.

The mass spectrum in Figure 2(a) corresponds to the chromatographic peak at 5.2 minutes. According to the fragmentation routes analysis (Scheme 1), the mentioned spectrum is due to the thiosemicarbazone of 4-methy- lacetophenone (ammonia loss has very low energy requirement thus precluding the observation of the molecular ion in the spectrum). Schemes notations for fragmentation mechanisms are: α (alpha rupture), i (inductive cleavage), rH (Hydrogen rearrangement), rd (displacement reaction) and Hom. (homolityc cleavage). This last fragmentation pathway is somehow unusual and not favored for even electron ions. This work gives support to these mechanisms not only from the mass spectrometric data but also from theoretical calculations discussed later.

The inductive cleavage forms an unstable ion (low abundance) that easily decomposes inductively to form the tropylium ion. A rather unusual fragmentation pathway is observed to form the radical ion at m/z 133 from the rearranged molecular ion. The occurrence of the homolytic cleavage is a consequence of the relatively low energy requirement to break the N-N bond. The fragmentation pathways have been confirmed by GC-MS² (Ion trap, see experimental part), with the exception of those coming from the molecular ion (m/z 207), that has not been observed.

For thiosemicarbazones I and III the easy loss of ammonia is also the reason for the molecular ion absence in their spectra. For the thiosemicarbazone of 4-chloroacetophenone (II), the observation of the ion at m/z 168 in the highest mass region of the mass spectrum would be due to the easy loss of S=C=NH from the corresponding molecular ion.

The mass spectrum in Figure 2(b) corresponds to the chromatographic peak at retention time 10.2 minutes. It does match neither the predicted product spectrum nor any of the reactants spectra. Considering the fact that the substance is pure, it is clear that a reaction takes place in the injection port of the chromatograph.

An exhaustive analysis of the ions present in the spectrum and of the fragmentation routes, made possible to propose the “dimer” structure shown in Figure 3, which is consistent with experimental data (Scheme 2). The second largest peak at m/z 249 is assigned to the methyl loss and the base peak (m/z 91) is the tropylium ion.

The considerations already commented for the thiosemicarbazone spectrum also apply to the dimer spectrum although the homolytic cleavage takes place in an even electron ion at m/z 249 to render a stable 4-methylben-

zonitrile radical ion (m/z 117). Additionally, transposition-like reactions are observed with formation of nitriles (4-methylbenzonitrile and acetonitrile, to form m/z 132 in both cases). A four centers transition state should be considered to explain these displacement reactions. The fragmentation pathways have been confirmed by GC- MS² (Ion trap, see experimental part).

This mass spectral behavior is observed for every compound under study. The fragments shown in Scheme 2 are present in all spectra. In addition, the ratio [Fragment Dimer]/[Thiosem] is always much higher than 1 and it increases while injection temperature increases. Direct injection to the mass spectrometer resulted in mass spectra superposition with similar injection temperature dependence.

On the other hand, aliphatic semicarbazones and thiosemicarbazones, that have been previously prepared and studied, have not shown a similar behavior.

These observations have allowed postulating that the considered substances react in the injection port, breaking out and undergoing dimerization of one of the resultant fragments, as it is shown in Scheme 3.

Scheme 1. Fragmentation routes of the thiosemicarbazone of 4-methylacetophenone.

Small volatile molecules could be formed by radical reactions in the injection port of the GC (HN=C=S, H₂NNH₂), not detected under the experimental conditions (in very early eluates).

In order to get additional information that may support or reject this hypothesis, a series of combined substrates (binary samples) have been injected in the chromatograph (“cross experiment”).

These solutions have been prepared using equimolar quantities of each compound (2.2 × 10⁻⁵ moles in 2 mL of DMSO).

Table 2 shows the composition of the mentioned mixtures.

Scheme 2. Fragmentation routes of the proposed structure.

Scheme 3. Dimer structure formation.

Figure 4(a) shows the chromatogram of mixture E (4-chloroacetophenone thiosemicarbazone and 4-methy- lacetophenone thiosemicarbazone).

As it can be seen, there are three neutral species eluting from the chromatographic column.

A comparative analysis with the chromatograms of the individual solutions (see Figure 1 and Figure 2(a)), reveals that peak at 10 min corresponds to dimer species of neutral fragment from II and the other one at 10.6 min is assigned to the analogous fragment dimer from IV. Retention times and spectra are independent of injection temperature although this parameter modifies the relative intensities of the signals.

The mass spectrum of the chromatographic peak at 11.2 min is shown in Figure 4(b).

The molecular ion is at m/z 284. The ion at m/z 269 is the methyl loss and that one at m/z 91 should be assigned to the tropylium ion. The base peak, at m/z 117, can be assigned to the ionic structure of p-methyl-benzo- nitrile.

In the mass spectral analysis of the Compound II and its neutral fragment dimer, the signal at m/z 111 has been assigned to the ion C₆H₄Cl]⁺, and the peak at m/z 137 has been assigned to the structure Cl(C₆H₄)CN]⁺.

The experimental data and their analysis have allowed proposing the formation of a crossed dimer species (“heterodimers”) in the injection port, whose molecular weight is an intermediate between the molecular weights of the respective “homodimers”, as it is shown in Figure 5.

The structure shown above is consistent with all the spectral assignments.

The fragmentation routes and the respective ions are shown in Scheme 4, for this particular case.

Mass spectral considerations are similar to those from previous comments for the thiosemicarbazone and its corresponding homodimer although the presence of the chlorine atom would be the explanation for the very low abundances observed for m/z 152 and 111.

Additionally, displacement reactions that render ion products with aliphatic moieties attached to the charge nitrogen, have lower abundances (e.g. m/z 152 and 132), compared with ions having aromatic moieties instead (e.g. m/z 228).

The behavior previously described is observed for every mixture and the tendency involving the intensities ratios of the chromatographic peaks is followed in all cases. The fact that the highest intensity is shown by the heterodimer in each mixture, is caused by a statistical effect: in these experimental conditions, it is more likely the bonding between two fragments produced by the thermolysis of molecules of different nature than the formation of the homodimers.

Table 3 shows m/z values of the analogous ions to those presented in Scheme 4 for each pair of compounds. These data not only support the dimer formation in the injection port of the gas chromatograph but also the proposed fragmentation pathways which have been additionally proved by GC/MS in an ion trap mass spectrometer (see experimental part).

The bond dissociation energy for the homolytic fragmentation of the N-N bond of the ion shown in Scheme 5 was calculated to be 3.55 eV. For comparison purposes, the bond dissociation energy for the homolysis of an even electron ion in which the nitrogen atom is bound to an sp² carbon atom (see the ion produced by displacement reaction in Scheme 5), was also calculated. In this case, the bond dissociation energy is considerably higher,

Scheme 4. Fragmentation routes of the heterodimer of Compounds II and IV.

^aEach column corresponds to a specific fragmentation pathway. ^bIon particularly unstable (low abundance) due to the electro-withdraw- ing nitro group.

namely 5.36 eV, rendering also the m/z 117 ion. So that, this ion would be mainly formed by the first homolytic route. Respect to the inductive cleavage, the calculation renders an energy value of 1.35 eV (driven by the loss of the stable 4-methylbenzonitrile molecule) which is even lower than the first homolytic rupture. This pathway leads to the production of a highly unstable ion (m/z 132) which rapidly forms the very stable ion at m/z 91 (tropylium ion).