Chemistry of Spontaneous Alkylation of Methimazole with 1,2-Dichloroethane

Spontaneous S-alkylation of methimazole (1) with 1,2-dichloroethane (DCE) into 1,2-bis[(1-methyl-1H-imidazole-2-yl)thio]ethane (2), that we have described recently, opened the question about its formation pathway(s). Results of the synthetic, NMR spectroscopic, crystallographic and computational studies suggest that, under given conditions, 2 is obtained by direct attack of 1 on the chloroethyl derivative 2-[(chloroethyl)thio]-1-methyl-1H-imidazole (3), rather than through the isolated stable thiiranium ion isomer, i.e., 7-methyl-2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium chloride (4a, orthorhombic, space group Pnma), or in analogy with similar reactions, through postulated, but unproven intermediate thiiranium ion 5. Furthermore, in the reaction with 1, 4a prefers isomerization to the N-chloroethyl derivative, 1-chloroethyl-2,3-dihydro-3-methyl-1H-imidazole-2-thione (7), rather than alkylation to 2, while 7 further reacts with 1 to form 3-methyl-1-[(1-methyl-imidazole-2-yl)thioethyl]-1H-imidazole-2-thione (8, monoclinic, space group P 21/c). Additionally, during the isomerization of 3, the postulated intermediate thiiranium ion 5 was not detected by chromatographic and spectroscopic methods, nor by trapping with AgBF4. However, trapping resulted in the formation of the silver complex of compound 3, i.e., bis-{2-[(chloroethyl)thio]-1-methyl-1H-imidazole}-silver(I)tetrafluoroborate (6, monoclinic, space group P 21/c), which cyclized upon heating at 80 °C to 7-methyl-2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium tetrafluoroborate (4b, monoclinic, space group P 21/c). Finally, we observed thermal isomerization of both 2 and 2,3-dihydro-3-methyl-1-[(1-methyl-1H-imidazole-2-yl)thioethyl]-1H-imidazole-2-thione (8), into 1,2-bis(2,3-dihydro-3-methyl-1H-imidazole-2-thione-1-yl)ethane (9), which confirmed their structures.


Introduction
Methimazole (thiamazole, 1-methyl-2,3-dihydro-1H-imidazole-2-thione (1) is a worldwide used thyrostatic drug [1]. Owing to the structural feature of the ambidentate heterocyclic [N-C-S] type anion, it has also found application as a terminal group in the synthesis of noncyclic crown ethers, as well as in various aspects of coordination chemistry [2,3].
In the course of our methimazole preliminary stability study, we recently reported S-alkylation of 1 by 1,2-dichloroethane (DCE) in solution [4]. This reaction occurs spontaneously, both in light and dark, at ambient humidity and room temperature within 15 days, In the course of our methimazole preliminary stability study, we recently reported Salkylation of 1 by 1,2-dichloroethane (DCE) in solution [4]. This reaction occurs spontaneously, both in light and dark, at ambient humidity and room temperature within 15 days, leading to the formation of 1,2-bis[(1-methyl-1H-imidazole-2-yl)thio]ethane (2), primarily isolated in the form of dihydrochloride tetrahydrate (2b). Depending on reaction/isolation conditions, the product can be isolated as an anhydrate (2a) or dihydrate (2c) (Figure 1). Their mutual interconversion was previously described [4]. In the context of the pharmaceutical purity profile, 2 is considered to be very interesting since it is a potential methimazole-related substance, and as such should not be present in the marketed product.
Therefore, aiming to get better insight into the formation pathways leading from 1 to 2, we carried out synthetic, NMR, crystallographic, and computational studies.

Reaction of Methimazole (1) with 1,2-Dichoroethane
The reaction of 1 and DCE, as previously reported, gave 1,2-bis[(1-methyl-1Himidazole-2-yl)thio]ethane dihydrochloride tetrahydrate (2b) in the form of colorless plate-shaped crystals. The melting point and the NMR data were identical to the data reported earlier [4]. The remaining mother liquor was analyzed and, after separation, yielded not only the unreacted methimazole, but also the colorless liquid, 2-[(chloroethyl)thio]-1-methyl-H-imidazole (3,12.1%). The HRMS and NMR analysis (see Experimental) confirmed the structure of 3. The reaction of 1 with boiling DCE in the presence of formic acid as a co-solvent gave 3 in a better yield than before (73%), along with 2a.
When the reaction of 1 and dry DCE was performed under reflux without acid (Scheme 1), the product, according to the NMR analysis, contained a mixture of 7-methyl-2H, 3H, 7H-imidazo [2,1-b]thiazol-4-ium chloride (4a) and bis[(1-methyl-1H-imidazole-2yl)thio]ethane dihydrochloride (2d). The latter was isolated by column chromatography in the form of pure colorless crystals. As in the previous case, the mother liquor contained 3, which was isolated in the form of a colorless oil. The question of 4a formation was answered when we discovered that the Schloroethyl derivative 3 quantitatively isomerizes into imidazothiazolium chloride 4a (Scheme 1) after 21 days at room temperature. 4a was isolated as needle-like crystals, its In the context of the pharmaceutical purity profile, 2 is considered to be very interesting since it is a potential methimazole-related substance, and as such should not be present in the marketed product.
Therefore, aiming to get better insight into the formation pathways leading from 1 to 2, we carried out synthetic, NMR, crystallographic, and computational studies.

Reaction of Methimazole (1) with 1,2-Dichoroethane
The reaction of 1 and DCE, as previously reported, gave 1,2-bis[(1-methyl-1H-imidazole-2-yl)thio]ethane dihydrochloride tetrahydrate (2b) in the form of colorless plate-shaped crystals. The melting point and the NMR data were identical to the data reported earlier [4]. The remaining mother liquor was analyzed and, after separation, yielded not only the unreacted methimazole, but also the colorless liquid, 2-[(chloroethyl)thio]-1-methyl-H-imidazole (3,12.1%). The HRMS and NMR analysis (see Experimental) confirmed the structure of 3. The reaction of 1 with boiling DCE in the presence of formic acid as a co-solvent gave 3 in a better yield than before (73%), along with 2a.
When the reaction of 1 and dry DCE was performed under reflux without acid (Scheme 1), the product, according to the NMR analysis, contained a mixture of 7-methyl-2H, 3H, 7Himidazo [2,1-b]thiazol-4-ium chloride (4a) and bis[(1-methyl-1H-imidazole-2-yl)thio]ethane dihydrochloride (2d). The latter was isolated by column chromatography in the form of pure colorless crystals. As in the previous case, the mother liquor contained 3, which was isolated in the form of a colorless oil.
In the course of our methimazole preliminary stability study, we recently report alkylation of 1 by 1,2-dichloroethane (DCE) in solution [4]. This reaction o spontaneously, both in light and dark, at ambient humidity and room temperature w 15 days, leading to the formation of 1,2-bis[(1-methyl-1H-imidazole-2-yl)thio]ethan primarily isolated in the form of dihydrochloride tetrahydrate (2b). Dependin reaction/isolation conditions, the product can be isolated as an anhydrate (2a) or dihy (2c) (Figure 1). Their mutual interconversion was previously described [4]. In the context of the pharmaceutical purity profile, 2 is considered to be interesting since it is a potential methimazole-related substance, and as such should be present in the marketed product.
Therefore, aiming to get better insight into the formation pathways leading from 2, we carried out synthetic, NMR, crystallographic, and computational studies.

Reaction of Methimazole (1) with 1,2-Dichoroethane
The reaction of 1 and DCE, as previously reported, gave 1,2-bis[(1-methy imidazole-2-yl)thio]ethane dihydrochloride tetrahydrate (2b) in the form of colo plate-shaped crystals. The melting point and the NMR data were identical to the reported earlier [4]. The remaining mother liquor was analyzed and, after separa yielded not only the unreacted methimazole, but also the colorless liquid [(chloroethyl)thio]-1-methyl-H-imidazole (3,12.1%). The HRMS and NMR analysis Experimental) confirmed the structure of 3. The reaction of 1 with boiling DCE i presence of formic acid as a co-solvent gave 3 in a better yield than before (73%), a with 2a.
When the reaction of 1 and dry DCE was performed under reflux without (Scheme 1), the product, according to the NMR analysis, contained a mixture of 7-me 2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium chloride (4a) and bis[(1-methyl-1H-imidazo yl)thio]ethane dihydrochloride (2d). The latter was isolated by column chromatogr in the form of pure colorless crystals. As in the previous case, the mother liquor conta 3, which was isolated in the form of a colorless oil. The question of 4a formation was answered when we discovered that th chloroethyl derivative 3 quantitatively isomerizes into imidazothiazolium chlorid (Scheme 1) after 21 days at room temperature. 4a was isolated as needle-like crysta The question of 4a formation was answered when we discovered that the S-chloroethyl derivative 3 quantitatively isomerizes into imidazothiazolium chloride 4a (Scheme 1) after 21 days at room temperature. 4a was isolated as needle-like crystals, its structure elucidated by NMR spectroscopy (Figure 2) and confirmed by single crystal X-ray diffraction analysis ( Figure 3 and Table S1). Indirectly, these results corroborated the structure of the S-chloroethyl derivative 3 as well. The imidazothiazolium salt 4a represents, to the best of our knowledge, the first room temperature stable thiiranium ion isomer reported so far. structure elucidated by NMR spectroscopy (Figure 2) and confirmed by single crystal Xray diffraction analysis ( Figure 3 and Table S1). Indirectly, these results corroborated the structure of the S-chloroethyl derivative 3 as well. The imidazothiazolium salt 4a represents, to the best of our knowledge, the first room temperature stable thiiranium ion isomer reported so far.  Figure 3a) suggest significant delocalization of the electron density and are in correspondence with literature data [5]. In the crystal structure of 4a, the planar heterobicyclic cations and chloride anions lie in the crystallographic mirror plane. Interatomic distance Cl − ···S of 3.2125(9) Å is attributed to a short contact suggesting an electrostatic interaction (Figure 3c). There are weak hydrogen bonds of the C-H···Cl type in the range 3.484(2) to 3.7655(12) Å interconnecting the ions into a 3D structure ( Figure  3c).
After isolation and the unambiguous structure determination of 3, a "sulfur mustard" type compound, and 4a, formed in the same reaction, we proceeded to elucidate the synthetic pathway to 2. The possibilities were: direct nucleophilic attack of 1 on the already formed intermediate 3 (pathway A), nucleophilic attack of 1 on the intermediate 4a (pathway B), or nucleophilic attack of 1 on the thiiranium (episulfonium) ion (5), a reactive (fleeting) intermediate in analogy with similar reactions [6,7] (pathway C) (Scheme 2).
Additionally,the highly reactive and, with a few exceptions [8], very unstable thiiranium ions are well described in the literature [6,7,9]. Recently, they were suggested as important synthetic intermediates in regio-and stereoselective sulfenoaminations of thioimidazoles with alkenes [10]. They can also be found as intermediates with a genotoxic potential [10,11] in nucleophilic reactions of "mustard" type compounds similar to our intermediate 3. structure elucidated by NMR spectroscopy (Figure 2) and confirmed by single crystal Xray diffraction analysis ( Figure 3 and Table S1). Indirectly, these results corroborated the structure of the S-chloroethyl derivative 3 as well. The imidazothiazolium salt 4a represents, to the best of our knowledge, the first room temperature stable thiiranium ion isomer reported so far.  Figure 3a) suggest significant delocalization of the electron density and are in correspondence with literature data [5]. In the crystal structure of 4a, the planar heterobicyclic cations and chloride anions lie in the crystallographic mirror plane. Interatomic distance Cl − ···S of 3.2125(9) Å is attributed to a short contact suggesting an electrostatic interaction (Figure 3c). There are weak hydrogen bonds of the C-H···Cl type in the range 3.484(2) to 3.7655(12) Å interconnecting the ions into a 3D structure ( Figure  3c).
After isolation and the unambiguous structure determination of 3, a "sulfur mustard" type compound, and 4a, formed in the same reaction, we proceeded to elucidate the synthetic pathway to 2. The possibilities were: direct nucleophilic attack of 1 on the already formed intermediate 3 (pathway A), nucleophilic attack of 1 on the intermediate 4a (pathway B), or nucleophilic attack of 1 on the thiiranium (episulfonium) ion (5), a reactive (fleeting) intermediate in analogy with similar reactions [6,7] (pathway C) (Scheme 2).
Additionally,the highly reactive and, with a few exceptions [8], very unstable thiiranium ions are well described in the literature [6,7,9]. Recently, they were suggested as important synthetic intermediates in regio-and stereoselective sulfenoaminations of thioimidazoles with alkenes [10]. They can also be found as intermediates with a genotoxic potential [10,11] in nucleophilic reactions of "mustard" type compounds similar to our intermediate 3.
After isolation and the unambiguous structure determination of 3, a "sulfur mustard" type compound, and 4a, formed in the same reaction, we proceeded to elucidate the synthetic pathway to 2. The possibilities were: direct nucleophilic attack of 1 on the already formed intermediate 3 (pathway A), nucleophilic attack of 1 on the intermediate 4a (pathway B), or nucleophilic attack of 1 on the thiiranium (episulfonium) ion (5), a reactive (fleeting) intermediate in analogy with similar reactions [6,7] (pathway C) (Scheme 2).
Additionally, the highly reactive and, with a few exceptions [8], very unstable thiiranium ions are well described in the literature [6,7,9]. Recently, they were suggested as important synthetic intermediates in regio-and stereoselective sulfenoaminations of thioimidazoles with alkenes [10]. They can also be found as intermediates with a genotoxic potential [10,11] in nucleophilic reactions of "mustard" type compounds similar to our intermediate 3.
In order to determine which reaction pathway is the most likely, and if the unstable tiiranium ion 5 is present in our reaction, the subsequent research continued both experimentally and theoretically.
Molecules 2021, 26, x FOR PEER REVIEW Scheme 2. Possible synthetic pathways from 1 to 2 (pathway A-presenting direct attack of pathway B-presenting attack of 1 to 4a, pathway C presenting isomerization of 3 to 5 and 4 and attack of 1).
In order to determine which reaction pathway is the most likely, and if the un tiiranium ion 5 is present in our reaction, the subsequent research continued both mentally and theoretically.

Kinetics of the S-Chloroethyl Derivative 3 Degradation
To explore if transformation of the S-chloroethyl derivative 3 to the stable im othiazolium salt 4a occurs through the thiiranium intermediate 5, the degradation DMSO-d6 at room temperature was monitored by accumulation of time-depend NMR spectra until the reaction completion.
Spontaneous transformation of 3 to 4a that occurs during the period of 3 mont confirmed, while no signals attributable to the proposed thiiranium derivative 5 w tected in any of the 1 H spectra. Pilot 13 C and HSQC spectra also revealed no signals region around 40 ppm where Dohn and Casida [12] reported the thiiranium ion f from cysteine derivatives under superacid conditions. It was concluded that eith mechanism does not involve the thiiranium ion as an intermediate, or, alternativ decomposes too quickly to be detected by NMR spectroscopy. However, after 3 days, an additional set of signals was detected at very low c tration belonging to a so far unknown isomer of 3, later synthesized, and its str confirmed as 1-chloroethyl-2,3-dihydro-3-methyl-1H-imidazole-2-thione (7).
Reaction kinetics of 3 into 4a and 7 was quantified using integrals of well sep imidazole proton NMR signals for all three species (Figure 4), while the rate con were calculated with Bruker Dynamics Center V2.6.3 software ( Figure 5). Conversi into compounds 4a and 7 proceeds to completion after approximately 75 days. Th tion ends in an apparent equilibrium with relative concentrations of 99.1% and 0.9 4a and 7, respectively. Therefore, to describe the observed decline in the signal in of 3 and the rise in signals belonging to 4a and 7, we used the model of consecuti order reactions with a reversible second step (Scheme 3) [13]. Scheme 2. Possible synthetic pathways from 1 to 2 (pathway A-presenting direct attack of 1 at 3, pathway B-presenting attack of 1 to 4a, pathway C presenting isomerization of 3 to 5 and 4a to 5 and attack of 1).

Kinetics of the S-Chloroethyl Derivative 3 Degradation
To explore if transformation of the S-chloroethyl derivative 3 to the stable imidazothiazolium salt 4a occurs through the thiiranium intermediate 5, the degradation of 3 in DMSO-d 6 at room temperature was monitored by accumulation of time-dependent 1 H NMR spectra until the reaction completion.
Spontaneous transformation of 3 to 4a that occurs during the period of 3 months was confirmed, while no signals attributable to the proposed thiiranium derivative 5 were detected in any of the 1 H spectra. Pilot 13 C and HSQC spectra also revealed no signals in the region around 40 ppm where Dohn and Casida [12] reported the thiiranium ion formed from cysteine derivatives under superacid conditions. It was concluded that either the mechanism does not involve the thiiranium ion as an intermediate, or, alternatively, it decomposes too quickly to be detected by NMR spectroscopy. However, after 3 days, an additional set of signals was detected at very low concentration belonging to a so far unknown isomer of 3, later synthesized, and its structure confirmed as 1-chloroethyl-2,3-dihydro-3-methyl-1H-imidazole-2-thione (7).
Reaction kinetics of 3 into 4a and 7 was quantified using integrals of well separated imidazole proton NMR signals for all three species (Figure 4), while the rate constants were calculated with Bruker Dynamics Center V2.6.3 software ( Figure 5). Conversion of 3 into compounds 4a and 7 proceeds to completion after approximately 75 days. The reaction ends in an apparent equilibrium with relative concentrations of 99.1% and 0.94%, for 4a and 7, respectively. Therefore, to describe the observed decline in the signal intensity of 3 and the rise in signals belonging to 4a and 7, we used the model of consecutive first order reactions with a reversible second step (Scheme 3) [13].  The obtained rate constant was calculated as an average value from two selected imidazole peaks resulting in k1 = 9.31 × 10 −7 s −1 with an R 2 value of 0.998 for the first reaction step, and k2 = 4.45 × 10 −7 s −1 with an R 2 value of 0.972. The final low level of 7 suggests that we actually observed the first order conversion from 3 to 4a.     The obtained rate constant was calculated as an average value from two selected imidazole peaks resulting in k1 = 9.31 × 10 −7 s −1 with an R 2 value of 0.998 for the first reaction step, and k2 = 4.45 × 10 −7 s −1 with an R 2 value of 0.972. The final low level of 7 suggests that we actually observed the first order conversion from 3 to 4a.    The obtained rate constant was calculated as an average value from two selected imidazole peaks resulting in k1 = 9.31 × 10 −7 s −1 with an R 2 value of 0.998 for the first reaction step, and k2 = 4.45 × 10 −7 s −1 with an R 2 value of 0.972. The final low level of 7 suggests that we actually observed the first order conversion from 3 to 4a.   The obtained rate constant was calculated as an average value from two selected imidazole peaks resulting in k 1 = 9.31 × 10 −7 s −1 with an R 2 value of 0.998 for the first reaction step, and k 2 = 4.45 × 10 −7 s −1 with an R 2 value of 0.972. The final low level of 7 suggests that we actually observed the first order conversion from 3 to 4a.

Trapping the Thiiranium Intermediate 5 with AgBF 4
Since no evidence of the thiiranium intermediate 5 during NMR reaction monitoring was found, an often-used silver salt method for thiiranium ion generation/trapping was employed.
One of the examples for this method is the use of silver tetrafluoroborate at temperatures below 0 • C [14]. To enable NMR acquisition below zero, toluene-d 8 was used to dissolve the "sulfur mustard" compound 3 at the temperature of −20 • C. An initial proton spectrum was recorded and then AgBF 4 was added. Immediately after addition of the Ag-salt, a white precipitate appeared and in all subsequent proton spectra (3 h and a gradual temperature increase to 25 • C), no peaks belonging to the starting 3, or any other unknown compounds, were detected, indicating that the formed precipitate is not soluble in toluene-d 8 . A small amount of DMSO-d 6 added at the end of the experiment dissolved the precipitate without problems.
The procedure was repeated, using DMSO-d 6 as solvent, at 25 • C. The addition of the Ag-salt to 3 immediately gave a noticeable change in chemical shifts in the proton spectrum ( Figure 6, red spectrum). Subsequently, in the preparative part of the study, this substance was synthesized and characterized as the Ag-complex of 3, i.e., bis-{2-[(chloroethyl)thio]-1methyl-1H-imidazole}-silver(I)tetrafluoroborate (6). However, no further change occurred during the following 40 min ( Figure 6, green spectrum). By increasing the temperature to 80 • C, the reaction started and 7 h later, the signals of 6 completely disappeared, while a new set of signals indicated the formation of a new species, which was later, during the preparative procedure, discovered to be 7-methyl-2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium tetrafluoroborate (4b). No signals attributable to the thiiranium 5 were detected. Since no evidence of the thiiranium intermediate 5 during NMR reaction monitoring was found, an often-used silver salt method for thiiranium ion generation/trapping was employed.
One of the examples for this method is the use of silver tetrafluoroborate at temperatures below 0 °C [14]. To enable NMR acquisition below zero, toluene-d8 was used to dissolve the "sulfur mustard" compound 3 at the temperature of −20 °C. An initial proton spectrum was recorded and then AgBF4 was added. Immediately after addition of the Agsalt, a white precipitate appeared and in all subsequent proton spectra (3 h and a gradual temperature increase to 25 °C), no peaks belonging to the starting 3, or any other unknown compounds, were detected, indicating that the formed precipitate is not soluble in toluene-d8. A small amount of DMSO-d6 added at the end of the experiment dissolved the precipitate without problems.
The procedure was repeated, using DMSO-d6 as solvent, at 25 °C. The addition of the Ag-salt to 3 immediately gave a noticeable change in chemical shifts in the proton spectrum ( Figure 6, red spectrum). Subsequently, in the preparative part of the study, this substance was synthesized and characterized as the Ag-complex of 3, i.e., bis-{2-[(chloroethyl)thio]-1-methyl-1H-imidazole}-silver(I)tetrafluoroborate (6). However, no further change occurred during the following 40 min ( Figure 6, green spectrum). By increasing the temperature to 80 °C, the reaction started and 7 h later, the signals of 6 completely disappeared, while a new set of signals indicated the formation of a new species, which was later, during the preparative procedure, discovered to be 7-methyl-2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium tetrafluoroborate (4b). No signals attributable to the thiiranium 5 were detected.

Further Synthetic Transformation
To unambiguously elucidate the structures of different forms discovered through changes in the NMR chemical shifts, 4a was treated with silver tetrafluoborate in methanol at room temperature. After 1 hour, pure 7-methyl-2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium tetrafluoroborate (4b) was obtained (Scheme 4). Its structure was elucidated by NMR spectroscopy, showing similar 1 H and 13 C chemical shifts to the corresponding shifts of 4a ( Figure 2). Its structure was confirmed by the single crystal X-ray structure analysis (Figure 7 and Table S1).

Further Synthetic Transformation
To unambiguously elucidate the structures of different forms discovered through changes in the NMR chemical shifts, 4a was treated with silver tetrafluoborate in methanol at room temperature. After 1 hour, pure 7-methyl-2H, 3H, 7H-imidazo[2,1-b]thiazol-4-ium tetrafluoroborate (4b) was obtained (Scheme 4). Its structure was elucidated by NMR spectroscopy, showing similar 1 H and 13 C chemical shifts to the corresponding shifts of 4a ( Figure 2). Its structure was confirmed by the single crystal X-ray structure analysis ( Figure 7 and Table S1).
The crystal structure of 4b consists of planar heterobicyclic cations and tetrafluoroborate anions.
As in 4a, the imidazole ring in 4b exhibits significant delocalization of the electron density, while the thiazole rings are more of aliphatic character. The differences in packing of 4a and 4b arise from the anion coordination preferences, which are size and geometry dependable. In the absence of good hydrogen bond donors, chloride anions and tetrafluoroborate are weakly coordinated with neighboring C-H groups. Following the same principle, treating the S-chloroethyl derivative 3 with silver tetrafluoborate in methanol at room temperature (Scheme 4), bis-{2-[(chloroethyl)thio]-1-methyl-1H-imidazole}-silver(I)tetrafluoroborate (6) was obtained in the form of colorless prism-shaped crystals (Scheme 4). This structure was elucidated by NMR spectroscopy and compared with the previous results from NMR measurements where heating of the Ag-complex 6 in DMSO-d6 at 80 °C for 7 h led to formation of the tetrafluoroborate salt 4b.
The structure of 6 was subsequently confirmed by the single crystal X-ray structure analysis ( Figure 8 and Table S1), its main feature being a silver ion bridge between two molecules. The crystal structure of 4b consists of planar heterobicyclic cations and tetrafluoroborate anions.
As in 4a, the imidazole ring in 4b exhibits significant delocalization of the electron density, while the thiazole rings are more of aliphatic character. The differences in packing of 4a and 4b arise from the anion coordination preferences, which are size and geometry dependable. In the absence of good hydrogen bond donors, chloride anions and tetrafluoroborate are weakly coordinated with neighboring C-H groups. The crystal structure of 4b consists of planar heterobicyclic cations and tetrafluoroborate anions.
As in 4a, the imidazole ring in 4b exhibits significant delocalization of the electron density, while the thiazole rings are more of aliphatic character. The differences in packing of 4a and 4b arise from the anion coordination preferences, which are size and geometry dependable. In the absence of good hydrogen bond donors, chloride anions and tetrafluoroborate are weakly coordinated with neighboring C-H groups.
Following the same principle, treating the S-chloroethyl derivative 3 with silver tetrafluoborate in methanol at room temperature (Scheme 4), bis-{2-[(chloroethyl)thio]-1methyl-1H-imidazole}-silver(I)tetrafluoroborate (6) was obtained in the form of colorless prism-shaped crystals (Scheme 4). This structure was elucidated by NMR spectroscopy and compared with the previous results from NMR measurements where heating of the Ag-complex 6 in DMSO-d 6 at 80 • C for 7 h led to formation of the tetrafluoroborate salt 4b.
The structure of 6 was subsequently confirmed by the single crystal X-ray structure analysis ( Figure 8 and Table S1), its main feature being a silver ion bridge between two molecules. Interestingly, Ag1···S1 and Ag1···S2 contact lengths are shorter than the sum of the van der Waals radii, while Ag1···Cl1 (from the neighboring molecule, 2 − x, 1 − y, 1 − z) and S1···Cl1 (1 + x, y, z) are at lengths less than this sum minus 0.17 and 0.32 Å, respectively. As in 4b, the weakly coordinating tetrafluoroborate anions are surrounded by C-H groups ( Figure 8c).
Therefore, it can be concluded that 3 is not in equilibrium with the thiiranium ion 5 (pathway C, Scheme 2), but that it spontaneously transforms into the very stable compound 4a.
Therefore, it can be concluded that 3 is not in equilibrium with the thiiranium ion 5 (pathway C, Scheme 2), but that it spontaneously transforms into the very stable compound 4a.
Therefore, it can be concluded that 3 is not in equilibrium with the thiiranium ion 5 (pathway C, Scheme 2), but that it spontaneously transforms into the very stable compound 4a.
Additionally, our attempt to directly prepare 2 by reaction of 4a with methimazole (1) (pathway B, Scheme 2) failed. By refluxing 4a with 1 in acetonitrile followed by column chromatography, pure 2, (8) was obtained in a 19.4% yield (Scheme 5). Its structure was elucidated by NMR spectroscopy and confirmed by the single crystal X-ray structure analysis ( Figure 9 and Table S1). The low yield is attributed to isomerization of 7 to 4a. In the crystal structure of 8, the molecule consists of two methyl-imidazole rings separated by a thioether -CH2-CH2-Sspacer. The two rings are planar and are inclined at 6.8° (Figure 9). The bond lengths and angles are comparable to those found in the CSD database. The molecules are stacked along the crystallographic b axis and held together by weak hydrogen bonds of the C-H···N type and π-interactions. In the other two directions there are only weak C-H···S and H···H contacts (Figure 9b). Furthermore, according to TLC of the reaction mixture (dichloromethane: methanol: formic acid = 8:1:0.5), no traces of 2 at Rf = 0.49 were detected. However, by using dichloromethane:acetone = 8:2 mixture as the mobile phase, beside the spot of 8 at Rf = 0.23, traces of an unknown substance at the front line (Rf = 0.98) were detected, indicating thermal instability of 4a.
The reaction performed under the same conditions, but without 1, led to the formation of the previously mentioned 7 (Scheme 5). Its structure was elucidated by NMR spectroscopy, with signals identical to the signals of 7 observed during kinetic decomposition of 3. The low yield of 7 (25.0%) was attributed to the fact that no changes in the chromatogram were observed after 4-10 h of the reaction time, indicating an equilibrium between 4a and 7. Additionally, the attempts to prepare suitable single crystals of 7 for Xray analysis failed because 7 isomerized back to 4a soon after isolation at room temperature, both in the solid state and in the solution. This corroborates our conclusions drawn from the reaction kinetics data.
Finally, by the reaction of 3 with 1 in formic acid and boiling MeCN (pathway A, Scheme 2), 2a was obtained along with unreacted 3. NMR and MS data of both isolated products are in line with those previously reported in the literature for 2 [4], and for 3 in the text above. In this reaction, no traces of 4a due to the protonation of nitrogen in 3 inhibiting intramolecular cyclisation and formation of 4a were observed.
The synthetic and NMR studies proved that the bis-derivative 2a is formed by direct reaction of the S-chloroethyl derivative 3 with methimazole (1) (pathway A, Scheme 2).  Similarly, the thione 8 was also obtained directly, in high yield (98.0%), by the reaction of 7 with 1 (Scheme 4) under reflux in MeCN.

Computational Search for the Thiiranium Ion Intermediate
In the crystal structure of 8, the molecule consists of two methyl-imidazole rings separated by a thioether -CH 2 -CH 2 -S-spacer. The two rings are planar and are inclined at 6.8 • (Figure 9). The bond lengths and angles are comparable to those found in the CSD database. The molecules are stacked along the crystallographic b axis and held together by weak hydrogen bonds of the C-H···N type and π-interactions. In the other two directions there are only weak C-H···S and H···H contacts (Figure 9b).
Furthermore, according to TLC of the reaction mixture (dichloromethane: methanol: formic acid = 8:1:0.5), no traces of 2 at Rf = 0.49 were detected. However, by using dichloromethane:acetone = 8:2 mixture as the mobile phase, beside the spot of 8 at Rf = 0.23, traces of an unknown substance at the front line (Rf = 0.98) were detected, indicating thermal instability of 4a.
The reaction performed under the same conditions, but without 1, led to the formation of the previously mentioned 7 (Scheme 5). Its structure was elucidated by NMR spectroscopy, with signals identical to the signals of 7 observed during kinetic decomposition of 3. The low yield of 7 (25.0%) was attributed to the fact that no changes in the chromatogram were observed after 4-10 h of the reaction time, indicating an equilibrium between 4a and 7. Additionally, the attempts to prepare suitable single crystals of 7 for X-ray analysis failed because 7 isomerized back to 4a soon after isolation at room temperature, both in the solid state and in the solution. This corroborates our conclusions drawn from the reaction kinetics data.
Finally, by the reaction of 3 with 1 in formic acid and boiling MeCN (pathway A, Scheme 2), 2a was obtained along with unreacted 3. NMR and MS data of both isolated products are in line with those previously reported in the literature for 2 [4], and for 3 in the text above. In this reaction, no traces of 4a due to the protonation of nitrogen in 3 inhibiting intramolecular cyclisation and formation of 4a were observed.
The synthetic and NMR studies proved that the bis-derivative 2a is formed by direct reaction of the S-chloroethyl derivative 3 with methimazole (1) (pathway A, Scheme 2).

Computational Search for the Thiiranium Ion Intermediate 5 2.4.1. Reaction without External Nucleophiles
S-Chloroethyl derivative 3 consists of a rigid N-methylimidazole ring and a flexible chloroethyl unit linked through the thioether bond ( Figure 10). Its reactivity is based on the fact that the terminal C-Cl is the weakest bond in the system, which makes Cl − a good leaving group. Specifically, the bond dissociation energy (BDE) that gives R-C + and Cl − is 25.5 kcal mol −1 , while for the central ethyl C-C bond, it is much higher at BDE = 77.7 kcal mol −1 corresponding to the homolytic cleavage and offering two carboncentered radicals. Analogously, both bonds to sulfur, C(imidazole)-S and C(alkyl)-S, are higher in energy, 73.6 and 45.4 kcal mol −1 , respectively, and most favorably proceed via the homolytic radical pathway. S-Chloroethyl derivative 3 consists of a rigid N-methylimidazole ring and a flexible chloroethyl unit linked through the thioether bond ( Figure 10). Its reactivity is based on the fact that the terminal C-Cl is the weakest bond in the system, which makes Cla good leaving group. Specifically, the bond dissociation energy (BDE) that gives R-C + and Clis 25.5 kcal mol -1 , while for the central ethyl C-C bond, it is much higher at BDE = 77.7 kcal mol -1 corresponding to the homolytic cleavage and offering two carbon-centered radicals. Analogously, both bonds to sulfur, C(imidazole)-S and C(alkyl)-S, are higher in energy, 73.6 and 45.4 kcal mol -1 , respectively, and most favorably proceed via the homolytic radical pathway. The structure of 3 contains two nucleophilic centers, the S-atom and the unsaturated imidazole N-atom, and both of these can undergo an internal SN2 rearrangement that liberates Cl - (Figure 11). The reaction with the S-atom gives the anticipated thiiranium ion 5 (Figure 11), yet the process has a very high kinetic requirement (ΔG ‡ = 29.8 kcal mol -1 ), and is, following the Cldeparture, thermodynamically very endergonic (ΔGR = 25.5 kcal mol -1 ), which makes 5 an unlikely end product. From there, it can further rearrange to 4a, yet this increases the reaction barrier to ΔG ‡ = 51.5 kcal mol -1 , despite gaining -14.3 kcal mol -1 in the reaction free energy, thereby rendering it unfeasible. Such a reaction profile confirms the unstable, high-energy, short-lived nature of 5, and the inability of NMR and MS techniques to detect it. In addition, following its generation, the reaction would preferably return to 3, rather than proceed to 4a, which disagrees with the formation of the latter. Hence, this rules out the option that the 3 → 4a conversion involves the thiiranium ion intermediate 5. Alternatively, the latter process can proceed through a direct nucleophilic attack of the unsaturated imidazole N-atom onto the C-atom bearing chlorine. In this way (Figure 11), the reaction involves a single step with the activation free energy of ΔG ‡ = 26.8 kcal mol -1 , being 3.0 kcal mol -1 lower than required to generate 5. This indicates a much more favorable route without the need for any intermediates, while a reasonable kinetic barrier and a significant exergonicity jointly confirm the spontaneous nature of this process under experimental conditions. The observed differences in the reactivity of N-and S-centers can be explained by considering the atomic charges in the initial 3. The Cl-atom is well prepared to act as an anionic leaving group, seen in its negative atomic charge of -0.11 |e|. The low nucleophilicity of the S-atom is well evidenced in its positive atomic charge, +0.27 |e|, which underlines a very limited tendency to react with electrophiles. On the other hand, both imidazole nitrogens bear negative charges, -0.42 |e| on the amine, and -0.59 |e| on the imine. While the nucleophilicity of the former is hindered by steric limitations, an even higher anionic charge on the latter rationalizes the demonstrated feasibility of the 3 → 4a conversion involving this site. The structure of 3 contains two nucleophilic centers, the S-atom and the unsaturated imidazole N-atom, and both of these can undergo an internal S N 2 rearrangement that liberates Cl − (Figure 11). The reaction with the S-atom gives the anticipated thiiranium ion 5 (Figure 11), yet the process has a very high kinetic requirement (∆G ‡ = 29.8 kcal mol −1 ), and is, following the Cl − departure, thermodynamically very endergonic (∆G R = 25.5 kcal mol −1 ), which makes 5 an unlikely end product. From there, it can further rearrange to 4a, yet this increases the reaction barrier to ∆G ‡ = 51.5 kcal mol −1 , despite gaining -14.3 kcal mol −1 in the reaction free energy, thereby rendering it unfeasible. Such a reaction profile confirms the unstable, high-energy, short-lived nature of 5, and the inability of NMR and MS techniques to detect it. In addition, following its generation, the reaction would preferably return to 3, rather than proceed to 4a, which disagrees with the formation of the latter. Hence, this rules out the option that the 3 → 4a conversion involves the thiiranium ion intermediate 5. Alternatively, the latter process can proceed through a direct nucleophilic attack of the unsaturated imidazole N-atom onto the C-atom bearing chlorine. In this way (Figure 11), the reaction involves a single step with the activation free energy of ∆G ‡ = 26.8 kcal mol −1 , being 3.0 kcal mol −1 lower than required to generate 5. This indicates a much more favorable route without the need for any intermediates, while a reasonable kinetic barrier and a significant exergonicity jointly confirm the spontaneous nature of this process under experimental conditions. The observed differences in the reactivity of N-and S-centers can be explained by considering the atomic charges in the initial 3. The Cl-atom is well prepared to act as an anionic leaving group, seen in its negative atomic charge of −0.11 |e|. The low nucleophilicity of the S-atom is well evidenced in its positive atomic charge, +0.27 |e|, which underlines a very limited tendency to react with electrophiles. On the other hand, both imidazole nitrogens bear negative charges, −0.42 |e| on the amine, and −0.59 |e| on the imine. While the nucleophilicity of the former is hindered by steric limitations, an even higher anionic charge on the latter rationalizes the demonstrated feasibility of the 3 → 4a conversion involving this site. In addition, experiments reveal a very slow further isomerization of 4a into the N-(chloroethyl) derivative 7 in low concentrations. This reaction is mediated by the Clanions that open the formed five-membered ring and offer the N-alkylated product. In doing so, Clattacks the C(alkyl)-S bond in 4a through the activation barrier of ΔG ‡ = 26.1 kcal mol -1 , which is, interestingly, 0.7 kcal mol -1 lower than needed to generate 4a from 3. However, a more significant progress of this reaction is hindered by the unfavorable reaction free energy, which for the 4a → 7 process is endergonic at ΔGR = 1.4 kcal mol -1 . This nicely explains why this reaction occurs very slowly at room temperature and confirms that higher yields of 7 can be achieved in this way by increasing the reaction temperature. In other words, the conversion of 3 into compounds 4a and 7 results in a 10-fold predominance of 4a, because both reactions in the 3 → 4a → 7 sequence are linked with similar kinetic requirements, yet the process leading to 4a is significantly more favorable through a highly exergonic thermodynamic character.
The presented results lead us to conclude that the anticipated intermediate 5 makes a reasonable assumption, yet its eventual formation strongly depends on the structure of the initial reactant. Namely, systems where there are no nucleophilic centers other than the S-atom that can initiate the β-chloride elimination, make such a pathway more likely. As an illustrative example, Finn and co-workers [17] reported the nucleophilic substitution of Cl-atoms in 2,6-dichloro-9-thiabicyclo[3.3.1]nonane with external nucleophiles that proceed through a highly reactive thiiranium ion, bearing a lot of steric strain that further facilitates the reaction progress. Yet, in 3, the presence of much nucleophilic imidazole nitrogen directs the rearrangement reaction to this site without the need for additional intermediates.

Reaction with the Introduction of External Nucleophile
Once methimazole 1 is added to the solution of 3, it changes the reaction outcomes by leading the conversion of the latter into 2, isolated in the form of the HCl salt (Scheme 1). There, 1 acts as an external nucleophile, whose structure is dominated by the N-H tautomer, while its S-H analogue is 14.1 kcal mol -1 less stable. During the course of the reaction, 1 has the option to offer 2 either by reacting with (i) the initial 3, (ii) the fivememebered derivative 4a, and (iii) the three-membered derivative 5 (Scheme 2). Given the demonstrated feasibility and the spontaneous character of the 3 → 4a conversion, it is reasonable to expect that the formation of 2 will involve 4a as an intermediate. Indeed, following the formation of 4a (Figure 12), it takes 7.3 kcal mol -1 to add 1 into the reactive complex, and an additional 39.1 kcal mol -1 to reach the transition state that describes the N-C cleavage within the five-membered ring and the formation of a new C-S bond with 1 that allows 2. This becomes the rate-limiting step with an extensive activation barrier of ΔG ‡ = 46.4 kcal mol -1 , making such a pathway as highly unlikely. The reason for such an In addition, experiments reveal a very slow further isomerization of 4a into the N-(chloroethyl) derivative 7 in low concentrations. This reaction is mediated by the Cl − anions that open the formed five-membered ring and offer the N-alkylated product. In doing so, Cl − attacks the C(alkyl)-S bond in 4a through the activation barrier of ∆G ‡ = 26.1 kcal mol −1 , which is, interestingly, 0.7 kcal mol −1 lower than needed to generate 4a from 3. However, a more significant progress of this reaction is hindered by the unfavorable reaction free energy, which for the 4a → 7 process is endergonic at ∆G R = 1.4 kcal mol −1 . This nicely explains why this reaction occurs very slowly at room temperature and confirms that higher yields of 7 can be achieved in this way by increasing the reaction temperature. In other words, the conversion of 3 into compounds 4a and 7 results in a 10-fold predominance of 4a, because both reactions in the 3 → 4a → 7 sequence are linked with similar kinetic requirements, yet the process leading to 4a is significantly more favorable through a highly exergonic thermodynamic character.
The presented results lead us to conclude that the anticipated intermediate 5 makes a reasonable assumption, yet its eventual formation strongly depends on the structure of the initial reactant. Namely, systems where there are no nucleophilic centers other than the S-atom that can initiate the β-chloride elimination, make such a pathway more likely. As an illustrative example, Finn and co-workers [17] reported the nucleophilic substitution of Cl-atoms in 2,6-dichloro-9-thiabicyclo[3.3.1]nonane with external nucleophiles that proceed through a highly reactive thiiranium ion, bearing a lot of steric strain that further facilitates the reaction progress. Yet, in 3, the presence of much nucleophilic imidazole nitrogen directs the rearrangement reaction to this site without the need for additional intermediates.

Reaction with the Introduction of External Nucleophile
Once methimazole 1 is added to the solution of 3, it changes the reaction outcomes by leading the conversion of the latter into 2, isolated in the form of the HCl salt (Scheme 1). There, 1 acts as an external nucleophile, whose structure is dominated by the N-H tautomer, while its S-H analogue is 14.1 kcal mol −1 less stable. During the course of the reaction, 1 has the option to offer 2 either by reacting with (i) the initial 3, (ii) the five-memebered derivative 4a, and (iii) the three-membered derivative 5 (Scheme 2). Given the demonstrated feasibility and the spontaneous character of the 3 → 4a conversion, it is reasonable to expect that the formation of 2 will involve 4a as an intermediate. Indeed, following the formation of 4a (Figure 12), it takes 7.3 kcal mol −1 to add 1 into the reactive complex, and an additional 39.1 kcal mol −1 to reach the transition state that describes the N-C cleavage within the fivemembered ring and the formation of a new C-S bond with 1 that allows 2. This becomes the rate-limiting step with an extensive activation barrier of ∆G ‡ = 46.4 kcal mol −1 , making such a pathway as highly unlikely. The reason for such an unfavorable reaction profile resides in the extensive stability of 4a, -14.3 kcal mol −1 from the initial 3, which disfavors its further conversions. Therefore, we can confidently state that 2 does not form from the preceding five-membered isomer 4a. the acidic conditions (formic acid employed here), such an environment will convert 3 into its protonated form 3H + , with excess protons residing at the unsaturated imidazole nitrogen. To support that, the basicity of this position is, on relative scales, around 10 pKa units higher than the basicity of the S-atom in 1. The latter quantitatively agrees with the fact that, for example, the related N-methylimidazole (pKa = 6.95) [18,19] is almost 9 pKa units more basic than N-methylthiourea (pKa = -1.75) [20,21], which validates these calculations. This process will reduce the nucleophilicity of the former, which will disfavor its rearrangement to 4a, thus allowing the only possible 3 + 1 → 2 conversion without byproducts 4a, 7, and 8, thereby placing computational results in excellent agreement with experiments. In concluding this discussion, we note that the reaction between 1 and 3 gives monocationic 2 that is ΔGR = 1.5 kcal mol -1 higher in energy than initial reactants ( Figure 12). This prompted us to inspect the possibility that the formed 2 spontaneously deprotonates into the neutral system (Scheme 6), while the liberated proton joins with the chloride anion depart as HCl. Yet, such an outcome is by as much as 10.5 kcal mol -1 less stable, which ties As an alternative route, we again considered the possibility for the generation of the short-lived intermediate 5, which could offer 2 through a reaction with 1 ( Figure 12). As shown, once 2 is formed, it resides 25.5 kcal mol −1 higher in energy than 3, from which it takes 6 kcal mol −1 to introduce 1 and further 5.2 kcal mol −1 to arrive at the transition state for the formation of 2. This leads to the overall activation barrier of ∆G ‡ = 36.7 kcal mol −1 , which is, interestingly, 9.7 kcal mol −1 lower than when the reaction occurs through 4a. Although more favorable, such a reaction profile still indicates that 5 will prefer going back to the initial 3, rather than progressing to 2, as the former pathway is both kinetically and thermodynamically more favorable. Therefore, this scenario also cannot explain the formation of 2 as experimentally demonstrated.
Lastly, 1 can result in 2 through a direct reaction with 3, a process where the S-atom in 1 acts as a nucleophile that cleaves the C-Cl bond in 3 and allows 2 with the liberation of Cl − . This is a one-step process, which requires 8.6 kcal mol −1 to bring both reactants into a reactive complex for a total activation barrier of ∆G ‡ = 33.9 kcal mol −1 , and a reaction free energy of ∆G R = 1.5 kcal mol −1 . The calculated parameters indicate the most feasible route to generate 2, yet the obtained kinetic and thermodynamic features make it less favorable than the internal cyclization 3 → 4a occurring without 1. This suggests that the generation of 2 will be slower than the reorganization to 4a, and its higher yields will be hindered by the excess generation of the latter. However, if the reaction between 1 and 3 occurs under the acidic conditions (formic acid employed here), such an environment will convert 3 into its protonated form 3H + , with excess protons residing at the unsaturated imidazole nitrogen. To support that, the basicity of this position is, on relative scales, around 10 pK a units higher than the basicity of the S-atom in 1. The latter quantitatively agrees with the fact that, for example, the related N-methylimidazole (pK a = 6.95) [18,19] is almost 9 pK a units more basic than N-methylthiourea (pK a = −1.75) [20,21], which validates these calculations. This process will reduce the nucleophilicity of the former, which will disfavor its rearrangement to 4a, thus allowing the only possible 3 + 1 → 2 conversion without byproducts 4a, 7, and 8, thereby placing computational results in excellent agreement with experiments.
In concluding this discussion, we note that the reaction between 1 and 3 gives monocationic 2 that is ∆G R = 1.5 kcal mol −1 higher in energy than initial reactants ( Figure 12). This prompted us to inspect the possibility that the formed 2 spontaneously deprotonates into the neutral system (Scheme 6), while the liberated proton joins with the chloride anion depart as HCl. Yet, such an outcome is by as much as 10.5 kcal mol −1 less stable, which ties in with experiments and confirms why 2 crystallizes as a hydrochloride salt, and not as, for example, an uncharged system. Finally, we investigated whether the available Cl − anions can interact with 1, thereby allowing HCl and the anionic 1 − , the latter potentially more nucleophilic and likely further promoting any of the studied processes. However, our calculations show that a reaction 1 + Cl − → 1 − + HCl is exceedingly endergonic, ∆G R = 28.5 kcal mol −1 , making this possibility as highly unfeasible. in with experiments and confirms why 2 crystallizes as a hydrochloride salt, and not as, for example, an uncharged system. Finally, we investigated whether the available Cl anions can interact with 1, thereby allowing HCl and the anionic 1 -, the latter potentially more nucleophilic and likely further promoting any of the studied processes. However, our calculations show that a reaction 1 + Cl -→ 1 -+ HCl is exceedingly endergonic, ΔGR = 28.5 kcal mol -1 , making this possibility as highly unfeasible. Scheme 6. Chemical reaction between 3 and 1 liberates the Clanion and offers monocationic 2, which can potentially deprotonate at the secondary imidazole amine position (in blue).

General
All solvents were distilled before use. Methimazole 99% was purchased from CU Chemie Ueticon, Lahr, Germany (water content 0.4%), 1,2-dichloroethane (Fisher Scientific, Bishop Meadow Road, Loughborough, UK), triethylamine (Sigma Aldrich, St. Louis, MO, USA, SAD), silver tetrafluoroborate, 99% (Alfa Aesar, Kendal. Germany), acetone (Carlo Erba Reagents, Val de Reuil Cedex, France, FRA), while formic acid, sodium sulfate anhydrous, acetonitrile, dimethylformamide, sodium chloride, sodium hydroxide, sodium hydrogen carbonate, and potassium hydroxide were purchased from Merck (Merck KGaA, Darmstadt, Germany). Solvents were distilled prior to use and dried using standard techniques. All reactions were done in flame-dried glassware. Water was purified by in-house system (Thornton 2000 CRS, Mettler Toledo, Colombus, OH, USA). pH measurements were performed using Mettler Toledo (Columbus, OH, USA) Seven Multi pH meter, and prior to measurement, it was calibrated using six points. All other used chemicals were of analytical grade. 1 NMR and 13 C NMR were recorded on a Bruker Avance AV600 spectrometer. High resolution mass spectra (HRMS) were recorded on an Agilent 6550 I Funnel quadrupole time-of flight mass spectrometer (QTOF) equipped with dual AJS ESI source (Agilent Technologies, Palo Alto, CA, USA). MS spectra were acquired on an Agilent 6460 triple quad (QQQ) mass spectrometer equipped with AJS ESI source (Agilent Technologies, Palo Alto, CA, USA). Ionic chromatography measurements were performed on a Thermo (Waltham, MA, USA) ionic chromatography using LC chlorine standard. HPLC analysis was performed on an Agilent Technologies (Santa Clara, CA, USA) HPLC instrument under gradient elution at a flow rate of 0.6 mL/min using mobile phase A (ammonium acetate, Merck KGaA, Darmstadt, Germany buffer) and mobile phase B (acetonitrile, Merck KGaA, Darmstadt, Germany) on a Zorbax C18 column. The effluent was monitored using the Agilent DAD/UV detector. Thermal analysis was performed using a Mettler DSC 1 instrument (Mettler Toledo, Greifensee, Swizerland) in a aluminium pans with a pierced lid at a heating rate of 10 °C/min under inert atmosphere with a flow rate of 55 mL/min. Temperature calibration was performed using the indium metal standard. Reaction progress was monitored using thin layer chromatography (NH3: isopropanol:toluene = 0.5:2.5:7) or (dichloromethane:acetone = 8:2) or (dichloromethane: methanol: formic acid = 8:1:0.5) on silica gel plates (Kieselgel G60 F254, Merck KGaA, Darmstadt, Germany). All weighing operations were carried using Mettler Toledo (Greifensee, Switzerland) balance, daily calibrated according to the internal program.

NMR Spectroscopy
Scheme 6. Chemical reaction between 3 and 1 liberates the Cl − anion and offers monocationic 2, which can potentially deprotonate at the secondary imidazole amine position (in blue).

General
All solvents were distilled before use. Methimazole 99% was purchased from CU Chemie Ueticon, Lahr, Germany (water content 0.4%), 1,2-dichloroethane (Fisher Scientific, Bishop Meadow Road, Loughborough, UK), triethylamine (Sigma Aldrich, St. Louis, MO, USA, SAD), silver tetrafluoroborate, 99% (Alfa Aesar, Kendal. Germany), acetone (Carlo Erba Reagents, Val de Reuil Cedex, France, FRA), while formic acid, sodium sulfate anhydrous, acetonitrile, dimethylformamide, sodium chloride, sodium hydroxide, sodium hydrogen carbonate, and potassium hydroxide were purchased from Merck (Merck KGaA, Darmstadt, Germany). Solvents were distilled prior to use and dried using standard techniques. All reactions were done in flame-dried glassware. Water was purified by in-house system (Thornton 2000 CRS, Mettler Toledo, Colombus, OH, USA). pH measurements were performed using Mettler Toledo (Columbus, OH, USA) Seven Multi pH meter, and prior to measurement, it was calibrated using six points. All other used chemicals were of analytical grade. 1 NMR and 13 C NMR were recorded on a Bruker Avance AV600 spectrometer. High resolution mass spectra (HRMS) were recorded on an Agilent 6550 I Funnel quadrupole time-of flight mass spectrometer (QTOF) equipped with dual AJS ESI source (Agilent Technologies, Palo Alto, CA, USA). MS spectra were acquired on an Agilent 6460 triple quad (QQQ) mass spectrometer equipped with AJS ESI source (Agilent Technologies, Palo Alto, CA, USA). Ionic chromatography measurements were performed on a Thermo (Waltham, MA, USA) ionic chromatography using LC chlorine standard. HPLC analysis was performed on an Agilent Technologies (Santa Clara, CA, USA) HPLC instrument under gradient elution at a flow rate of 0.6 mL/min using mobile phase A (ammonium acetate, Merck KGaA, Darmstadt, Germany buffer) and mobile phase B (acetonitrile, Merck KGaA, Darmstadt, Germany) on a Zorbax C18 column. The effluent was monitored using the Agilent DAD/UV detector. Thermal analysis was performed using a Mettler DSC 1 instrument (Mettler Toledo, Greifensee, Swizerland) in a aluminium pans with a pierced lid at a heating rate of 10 • C/min under inert atmosphere with a flow rate of 55 mL/min. Temperature calibration was performed using the indium metal standard. Reaction progress was monitored using thin layer chromatography (NH 3 : isopropanol:toluene = 0.5:2.5:7) or (dichloromethane:acetone = 8:2) or (dichloromethane: methanol: formic acid = 8:1:0.5) on silica gel plates (Kieselgel G60 F 254, Merck KGaA, Darmstadt, Germany). All weighing operations were carried using Mettler Toledo (Greifensee, Switzerland) balance, daily calibrated according to the internal program.
Method (B): 1-Chloroethyl-2,3-dihydro-3-methyl-1H-imidazole-2-thione (7, 100 mg, 0.56 mmol) with 1 (64 mg, 0.56 mmol) was refluxed in MeCN for 8 h. The mixture was evaporated under reduced pressure to dryness, the residue dissolved in water (10 mL), neutralized with 1 M NaOH, and extracted with dichloromethane. The organic layer was dried under anhydrous sodium sulphate, concentrated to dryness, and the residue purified by column chromatography on silica gel (mesh size 0.063-0.2) using dichloromethane: acetone = 8:2 as eluent. After evaporation of the selected fraction under reduced pressure, TLC pure 2,3-dihydro-3-methyl-1-[(1-methyl-1H-imidazole-2-yl)thioethyl]-1H-imidazole-2thione (8, 139 mg, 98.0 %) was obtained in the form of white crystalline product. Obtained melting point, Rf value, and NMR data are in line with the data reported previously in the text for 8.   600 µL), the solution cooled down to −20 • C, and the proton NMR spectrum of the solution was recorded at the same temperature, showing the signals of 3 accompanied by solvent signals. Then, silver tetrafluoroborate (6.0 mg, 0.04 mmol) was added and a series of consecutive proton NMR spectra were recorded for a total duration of 3 h while the temperature was gradually increased from −20 • C to 25 • C. Immediately after the addition of the salt, a white precipitate appeared and in all proton spectra no peaks belonging to the starting three were noticed, indicating that the formed precipitate is not soluble in toluene-d 8 . After no new signals appeared in a period of 3 h and the temperature reached 25 • C, DMSO-d 6 (600 µL) was added, the precipitate dissolved, and some new peaks appeared. After the addition of silver tetrafluoroborate (6.0 mg, 0.04 mmol), the recorded 1 H spectrum at the same temperature showed the signals belonging to Agcomplex 6. No other signals were noticed. After that, the temperature was increased up to 80 • C and the NMR spectra were consecutively recorded for a total duration of 7 h. After 1 h at 80 • C, the recorded proton spectrum exhibited low intensity signals belonging to tetrafluoroborate salt 4b in addition to the resonance lines of 6. Four hours at 80 • C later, the signals of 6 completely disappeared while signals of 4b dominated. Transformation from 3 to 4b was expected as the equimolar quantities of 3 and silver tetrafluoroborate were used, and the performed experiment clearly confirmed transformation of 3 via 6 to 4b ( Figure 6).

Single Crystal X-ray Diffraction
Single crystals were obtained by the following methods: 4a by spontaneous isomerization of 3 followed by crystallization, 4b and 6 by slow evaporation from methanol, and 8 by slow evaporation from water.
The crystal and molecular structures of 4a, 4b, 6, and 8, were determined by singlecrystal X-ray diffraction at the following temperatures: 293 K for 4a, 170 K for 4b, 6 and 8 (Table S1). Single-crystal diffraction experiments were performed on an Oxford Diffraction Xcalibur diffractometer (4a) and on a Rigaku Oxford Diffraction XtaLAB Synergy-S diffractometer (4b, 6 and 8). Program package CrysAlisPro [22] was used for data collection, cell refinement, and data reduction. Structures were solved by direct methods and refined using the SHELXT [23] and SHELXL [24] programs, respectively. The refinement procedure by the full-matrix least squares methods based on F 2 values against all reflections included anisotropic displacement parameters for all non-H atoms. The positions of H-atoms riding on carbon atoms were determined on stereochemical grounds. The anion moiety in 4b was modelled using various restraints (DFIX, SADI, SIMU and ISOR). The SHELX programs operated within the Olex2 crystallographic suite. Geometrical calculations and molecular graphics were done with MERCURY [25]. Supplementary crystallographic data sets for the structures are available through the Cambridge Structural Data base with deposition numbers 2,105,605-2,105,608. A copy of this information may be obtained free of charge from the director, CCDC, 12 Union Road, Cambridge, CB2 1EZ, UK (fax: +44-1223-336-033; e-mail: deposit@ccdc.cam.ac.uk or http://www.ccdc.cam.ac.uk.

Computational Methods
All molecular geometries were optimized with a very efficient M06-2X/6-31+G(d) model, which was designed to provide highly accurate thermodynamic and kinetic parameters for various organic systems. To account for the solvent effects, during geometry optimization, the implicit SMD solvation model corresponding to pure DMSO was included. Thermal corrections were extracted from the corresponding frequency calculations, so that all of the presented results correspond to differences in the Gibbs free energies at room temperature and normal pressure. The choice of such computational setup was prompted by its success in reproducing various features of different organic [26,27], organometallic [28,29], and enzymatic systems [30,31], being particularly accurate for relative trends among similar reactants, which is the focus here. All transition state structures were located using the scan procedure, employing both 1D and 2D scans, the latter specifically utilized to exclude the possibility for concerted mechanisms. Apart from the visualization of the obtained negative frequencies, the validity of all transition state structures was confirmed by performing IRC calculations in both directions and identifying the matching reactant and product structures connected by the inspected transition state. All the calculations were performed using the Gaussian 16 software [32].

Conclusions
We established that spontaneous transformation of methimazole (1) with 1,2-dichloroethane to 2 under mild conditions proceeds by the direct attack of 1 at the chloroethyl derivative 3. Although the isomerization of 3 into the stable thiiranium isomer 4a was observed (Scheme 7), the 1 to 2 conversion does not follow that pathway, nor does it go through the thiiranium ion 5, postulated in analogy to similar reactions reported in literature.
tions and molecular graphics were done with MERCURY [25]. Supplementary crystallo-graphic data sets for the structures are available through the Cambridge Structural Data base with deposition numbers 2,105,605-2,105,608. A copy of this information may be obtained free of charge from the director, CCDC, 12 Union Road, Cambridge, CB2 1EZ, UK (fax: +44-1223-336-033; e-mail: deposit@ccdc.cam.ac.uk or http://www.ccdc.cam.ac.uk

Computational Methods
All molecular geometries were optimized with a very efficient M06-2X/6-31+G(d) model, which was designed to provide highly accurate thermodynamic and kinetic parameters for various organic systems. To account for the solvent effects, during geometry optimization, the implicit SMD solvation model corresponding to pure DMSO was included. Thermal corrections were extracted from the corresponding frequency calculations, so that all of the presented results correspond to differences in the Gibbs free energies at room temperature and normal pressure. The choice of such computational setup was prompted by its success in reproducing various features of different organic [26,27], organometallic [28,29], and enzymatic systems [30,31], being particularly accurate for relative trends among similar reactants, which is the focus here. All transition state structures were located using the scan procedure, employing both 1D and 2D scans, the latter specifically utilized to exclude the possibility for concerted mechanisms. Apart from the visualization of the obtained negative frequencies, the validity of all transition state structures was confirmed by performing IRC calculations in both directions and identifying the matching reactant and product structures connected by the inspected transition state. All the calculations were performed using the Gaussian 16 software [32].

Conclusions
We established that spontaneous transformation of methimazole (1) with 1,2-dichloroethane to 2 under mild conditions proceeds by the direct attack of 1 at the chloroethyl derivative 3. Although the isomerization of 3 into the stable thiiranium isomer 4a was observed (Scheme 7), the 1 to 2 conversion does not follow that pathway, nor does it go through the thiiranium ion 5, postulatedin analogy to similar reactions reported in literature.

Scheme 7. Synthetic interconversion and isomerisation involving 4a.
The imidazothiazolium salt 4a in reaction with 1 prefers isomerization to the N-chloroethyl derivative 7, rather than alkylation to 2. The 7 further reacts with 1 and forms dimeric thione 8 in low yield. The structures of both 2 and 8 were confirmed by their thermal isomerization to 9. The intermediate thiiranium ion 5 was not detected by chromatographic and spectroscopic methods, nor caught by trapping it with AgBF 4 . However, during the trapping reaction, the silver complex of 3, i.e., 6, was isolated. When heated to 80 • C, 6 cyclized into the tetrafluoroborate salt 4b.
Conclusions from the computational studies, are in agreement with our experimental results, showed that the bis-derivative 2 is formed by direct reaction of 3 with 1, rather than through the postulated, but unlikely, thiiranium ion 5 or via the stable imidazotiazolium salt 4a.
Caution is needed when handling methimazole, due to its high reactivity in 1,2dichoroethane solution, even under mild conditions, and we discussed this previously. Our current results additionally revealed that in the 1 to 2 conversion, a "sulfur mustard" type