Experimental, Spectroscopic, and Computational Insights into the Reactivity of “Methanal” with 2-Naphthylamines

The reactions of 2-naphthylamine and methyl 6-amino-2-naphthoate with formalin and paraformaldehyde were studied experimentally, spectrally, and by quantum chemical calculations. It was found that neither the corresponding aminals nor imines were formed under the described conditions but could be prepared and spectrally characterized at least in situ under modified conditions. Several of the previously undescribed intermediates and by-products were isolated or at least spectrally characterized. First principle density functional theory (DFT) calculations were performed to shed light on the key aspects of the thermochemistry of decomposition and further condensation of the corresponding aminals and imines. The calculations also revealed that the electrophilicity of methanal was significantly greater than that of ordinary oxo-compounds, except for perfluorinated ones. In summary, methanal was not behaving as the simplest aldehyde but as a very electron-deficient oxo-compound.


Introduction
Although it is generally believed that the chemical reactions of methanal (1) with arylamines are already well studied, the opposite is true. The reactivity of methanal has been studied since the days of alchemy, and many of its reactions were described when the theory of bonds was born with the discovery of molecular structures. In those days, scientific methods for determining molecular structure were limited to elemental analyses (EA), which even nowadays have a rather limited accuracy (approx. ±0.3%). Indeed, these analyses are, therefore, incapable of distinguishing compounds with similar elemental compositions and are useless for recognizing isomers. Although the art of the early pioneers of chemistry was enthralling, contemporary scientific knowledge was insufficient to design the molecular structure of isolated products correctly.
Unfortunately, the inaccuracies that arose during these times have only occasionally been corrected, even at a time when nuclear magnetic resonance (NMR) spectroscopy was already a widely available technique for proving the molecular structure of organic compounds. For example, in our previous study [1] on the reactivity of 2-naphthylamine (2a) with a methanal equivalent under acidic conditions (Scheme 1), we isolated not only the known Tröger's base (TB) [2][3][4][5][6] derivative 3a and its methylated side products, but we also discovered its constitutional isomer 4a named spiro-Tröger's base (spiroTB). Through a literature survey, we found that TB 3a was prepared by Reed in 1886 [7][8][9], i.e., a year before Tröger published his base; both studies lacked a correct estimation of the molecular structure of the base. Indeed, SpiroTB 4a was not discovered by Farrar in 1964 [10] or Margitfalvi in 1998 [11]. In addition to the above-mentioned TB isomers, other described reaction products have included acridine 5a, quinazoline 6a, and dihydroquinazoline 7a, based on the reaction conditions used. already a widely available technique for proving the molecular structure of organic compounds. For example, in our previous study [1] on the reactivity of 2-naphthylamine (2a) with a methanal equivalent under acidic conditions (Scheme 1), we isolated not only the known Tröger's base (TB) [2][3][4][5][6] derivative 3a and its methylated side products, but we also discovered its constitutional isomer 4a named spiro-Tröger's base (spiroTB). Through a literature survey, we found that TB 3a was prepared by Reed in 1886 [7][8][9], i.e., a year before Tröger published his base; both studies lacked a correct estimation of the molecular structure of the base. Indeed, SpiroTB 4a was not discovered by Farrar in 1964 [10] or Margitfalvi in 1998 [11]. In addition to the above-mentioned TB isomers, other described reaction products have included acridine 5a, quinazoline 6a, and dihydroquinazoline 7a, based on the reaction conditions used.
Further ambiguities arise from the fact that the molecular structures of the reaction intermediates have not been accurately identified. Thus, if the true intermediate has a reactivity that is analogous to the expected one, the expected molecular structure of the intermediate may be mistaken for the correct one. This is the case of methanal itself. Scheme 1. Previously described products of the reaction of formaldehyde with 2-naphthylamines 2 under acidic conditions [1].
Methanal is a very reactive gas with a boiling point of 19.3 °C. It is usually considered the simplest organic aldehyde, but its reactivity differs significantly from ordinary aldehydes. Methanal is more similar to electron-deficient aldehydes such as trichloro-or trifluoroacetaldehyde, which form rather stable hydrates, hemiacetals, and hemiaminals in the presence of water, alcohols, and amines, respectively. The reactivity of methanal is even more unique since it contains no bulky groups attached to a carbonyl group; its low steric hindrance is probably the main cause of oligomerization, and thus, hydrates or hemiacetal of methanal dimer, trimer, etc., are well known. In addition, methanal can undergo a Cannizzaro reaction, i.e., it can act both as a reducing agent as well as an oxidizing one. The ability of methanal to act as a dehydrogenation agent could lead to the formation of dihydroquinazoline 7a from quinazoline 6a; such ability was observed for trifluoropyruvate [12].
Formalin is an aqueous solution of methanal, where methanal spontaneously forms the rather stable hydrate 10 (methanediol; equilibrium constant 2300) [13], followed by higher diols and O-methyl diols when methanol is used as a stabilizer. Since the concentration of free methanal is low [14,15] and the reaction products of both methanal and diols can be identical, there is no compelling reason to describe the formalin reaction as the reaction of methanal.
Similarly, paraformaldehyde is a mixture of higher diols with the general formula HO-(CH2O)n-H. Contrary to sources in the literature, boiling paraformaldehyde in methanol or ethanol does not produce methanal but mainly O-methyl or O-ethyl diols [16]. Scheme 1. Previously described products of the reaction of formaldehyde with 2-naphthylamines 2 under acidic conditions [1].
Further ambiguities arise from the fact that the molecular structures of the reaction intermediates have not been accurately identified. Thus, if the true intermediate has a reactivity that is analogous to the expected one, the expected molecular structure of the intermediate may be mistaken for the correct one. This is the case of methanal itself.
Methanal is a very reactive gas with a boiling point of 19.3 • C. It is usually considered the simplest organic aldehyde, but its reactivity differs significantly from ordinary aldehydes. Methanal is more similar to electron-deficient aldehydes such as trichloro-or trifluoroacetaldehyde, which form rather stable hydrates, hemiacetals, and hemiaminals in the presence of water, alcohols, and amines, respectively. The reactivity of methanal is even more unique since it contains no bulky groups attached to a carbonyl group; its low steric hindrance is probably the main cause of oligomerization, and thus, hydrates or hemiacetal of methanal dimer, trimer, etc., are well known. In addition, methanal can undergo a Cannizzaro reaction, i.e., it can act both as a reducing agent as well as an oxidizing one. The ability of methanal to act as a dehydrogenation agent could lead to the formation of dihydroquinazoline 7a from quinazoline 6a; such ability was observed for trifluoropyruvate [12].
Formalin is an aqueous solution of methanal, where methanal spontaneously forms the rather stable hydrate 10 (methanediol; equilibrium constant 2300) [13], followed by higher diols and O-methyl diols when methanol is used as a stabilizer. Since the concentration of free methanal is low [14,15] and the reaction products of both methanal and diols can be identical, there is no compelling reason to describe the formalin reaction as the reaction of methanal.
Similarly, paraformaldehyde is a mixture of higher diols with the general formula HO-(CH 2 O) n -H. Contrary to sources in the literature, boiling paraformaldehyde in methanol or ethanol does not produce methanal but mainly O-methyl or O-ethyl diols [16]. Pure methanal can only be generated through the dry decomposition of paraformaldehyde [14,15].
In this article, we focus on the conversion of naphthylamine 2a and methyl 6-amino-2naphthoate (2b) into the corresponding aminals 8a and 8b, and imines 9a and 9b, respectively (Scheme 2). Both aminal 8a and imine 9a have been described as the products of the treatment of naphthylamine 2a with formalin or paraformaldehyde, however, without any spectral evidence of their molecular structures. In light of the current knowledge, their formation under described conditions seems questionable. Since these compounds and their analogs are key intermediates for various syntheses, we have decided to re-examine their preparation.
2-naphthoate (2b) into the corresponding aminals 8a and 8b, and imines 9a and 9b, respectively (Scheme 2). Both aminal 8a and imine 9a have been described as the products of the treatment of naphthylamine 2a with formalin or paraformaldehyde, however, without any spectral evidence of their molecular structures. In light of the current knowledge, their formation under described conditions seems questionable. Since these compounds and their analogs are key intermediates for various syntheses, we have decided to re-examine their preparation.
The calculations presented here form part of our ongoing efforts to identify the best pathways for obtaining Tröger and spiro-Tröger species. A study of the mechanisms leading to the synthesis of these compounds, starting from anilines, has recently revealed important aspects of the mechanisms taking place under strong acid catalysis conditions [17]. The calculations here focus on neutral conditions with acetone as the model solvent. Scheme 2. The molecular structures of aminals 8a and 8b, and imines 9a and 9b.

Studies of Formalin and Paraformaldehyde
First, we studied the ability of formalin and paraformaldehyde to act as a methanal source in a solution using 1 H, 13 C, DQF-COSY, HSQC, and HMBC NMR spectra. It is worth noting that the true methanal content was greater than observed because part of it was probably in the gas phase above the solution in the NMR tube. The 1 H NMR spectra were not recorded under quantitative conditions so that determined contents may have varied within +/−10%.
In the solution of 5 μL of formalin in 0.5 mL of DMSO-d6 (dimethyl sulfoxide), we unambiguously identified (Scheme 3) methanal, methanol, water, diols 10, 11, and 12, hemiacetals 13 and 14, and acetal 15 in a molar ratio of 4:14:783:100:35:10:40:12:3, in addition to traces of formic acid and higher diols and hemiacetals. This corresponded to approximately 12% m/m of methanol in formalin (10-15% was declared by the supplier) and more than 33% m/m of methanal forms (37% was declared by the supplier). In accordance with a similar study in D2O [14,15], we did not observe any 1,3,5-trioxane (metaformaldehyde). On the other hand, small amounts of methanal and acetal 15 were observed, which were not mentioned in the study [14,15]. The relatively low concentrations of free methanal (>2%) increased dramatically at higher temperatures, as summarized in Table 1.  The calculations presented here form part of our ongoing efforts to identify the best pathways for obtaining Tröger and spiro-Tröger species. A study of the mechanisms leading to the synthesis of these compounds, starting from anilines, has recently revealed important aspects of the mechanisms taking place under strong acid catalysis conditions [17]. The calculations here focus on neutral conditions with acetone as the model solvent.

Studies of Formalin and Paraformaldehyde
First, we studied the ability of formalin and paraformaldehyde to act as a methanal source in a solution using 1 H, 13 C, DQF-COSY, HSQC, and HMBC NMR spectra. It is worth noting that the true methanal content was greater than observed because part of it was probably in the gas phase above the solution in the NMR tube. The 1 H NMR spectra were not recorded under quantitative conditions so that determined contents may have varied within +/−10%.
In the solution of 5 µL of formalin in 0.5 mL of DMSO-d 6 (dimethyl sulfoxide), we unambiguously identified (Scheme 3) methanal, methanol, water, diols 10, 11, and 12, hemiacetals 13 and 14, and acetal 15 in a molar ratio of 4:14:783:100:35:10:40:12:3, in addition to traces of formic acid and higher diols and hemiacetals. This corresponded to approximately 12% m/m of methanol in formalin (10-15% was declared by the supplier) and more than 33% m/m of methanal forms (37% was declared by the supplier). In accordance with a similar study in D 2 O [14,15], we did not observe any 1,3,5-trioxane (metaformaldehyde). On the other hand, small amounts of methanal and acetal 15 were observed, which were not mentioned in the study [14,15]. The relatively low concentrations of free methanal (>2%) increased dramatically at higher temperatures, as summarized in Table 1.
Pure methanal can only be generated through the dry decomposition of paraformaldehyde [14,15].
In this article, we focus on the conversion of naphthylamine 2a and methyl 6-amino-2-naphthoate (2b) into the corresponding aminals 8a and 8b, and imines 9a and 9b, respectively (Scheme 2). Both aminal 8a and imine 9a have been described as the products of the treatment of naphthylamine 2a with formalin or paraformaldehyde, however, without any spectral evidence of their molecular structures. In light of the current knowledge, their formation under described conditions seems questionable. Since these compounds and their analogs are key intermediates for various syntheses, we have decided to re-examine their preparation.
The calculations presented here form part of our ongoing efforts to identify the best pathways for obtaining Tröger and spiro-Tröger species. A study of the mechanisms leading to the synthesis of these compounds, starting from anilines, has recently revealed important aspects of the mechanisms taking place under strong acid catalysis conditions [17]. The calculations here focus on neutral conditions with acetone as the model solvent.

Studies of Formalin and Paraformaldehyde
First, we studied the ability of formalin and paraformaldehyde to act as a methanal source in a solution using 1 H, 13 C, DQF-COSY, HSQC, and HMBC NMR spectra. It is worth noting that the true methanal content was greater than observed because part of it was probably in the gas phase above the solution in the NMR tube. The 1 H NMR spectra were not recorded under quantitative conditions so that determined contents may have varied within +/−10%.
In the solution of 5 μL of formalin in 0.5 mL of DMSO-d6 (dimethyl sulfoxide), we unambiguously identified (Scheme 3) methanal, methanol, water, diols 10, 11, and 12, hemiacetals 13 and 14, and acetal 15 in a molar ratio of 4:14:783:100:35:10:40:12:3, in addition to traces of formic acid and higher diols and hemiacetals. This corresponded to approximately 12% m/m of methanol in formalin (10-15% was declared by the supplier) and more than 33% m/m of methanal forms (37% was declared by the supplier). In accordance with a similar study in D2O [14,15], we did not observe any 1,3,5-trioxane (metaformaldehyde). On the other hand, small amounts of methanal and acetal 15 were observed, which were not mentioned in the study [14,15]. The relatively low concentrations of free methanal (>2%) increased dramatically at higher temperatures, as summarized in Table 1.  The dissolution of 1 mg of paraformaldehyde in 0.5 mL DMSO-d 6 with 10 µL of water produced the equilibrated solution, which contained methanal, water, and diols 10, 11, and 12 in a molar ratio of 5:9494:441:51:9, alongside traces of higher diols. The methanal contents increased with increasing temperatures while the diols decomposed, as summarized in Table 2. Analogous results were obtained when methanol or ethanol was added to DMSO-d 6 instead of water. Similarly, high relative concentrations of methanal were identified when paraformaldehyde was treated with DMSO-d 6 without any added water, so the majority of the paraformaldehyde remained undissolved (both the solvent and the paraformaldehyde contained moisture). The equilibrated mixture in the solution contained methanal, water, and diols 10, 11, and 12 in a molar ratio of 61:9038:727:159:15 (the concentrations of higher diols were less than 1‰). As expected, the methanal contents increased (>40%) upon raising the temperature as the diols decomposed; however, a new methanal form appeared and became the major one at 115 • C. This form was characterized by the sole singlet at 4.81 ppm in the 1 H NMR spectrum ( 1 J HC = 165.5 Hz, from 13 C satellites), which was correlated in both the HSQC and the HMBC spectra to a 13 C singlet at 89.07 ppm. Unfortunately, the molecular structure was not determined; 1,3,5-trioxane has an identical HSQC/HMBC pattern; however, different chemical shifts as was confirmed by the 1,3,5-trioxane standard addition.
It is worth emphasizing that the composition of those samples strongly depended on the water contents, total concentrations, presence of formic acid (common impurity of aged paraformaldehyde), and the time required to reach equilibrium, and thus, reproducing of these experiments would be difficult.

Calculations on the Reactivity of Methanal
As a first task, it was worth quantitatively addressing the power of methanal as a nucleophile, which was also related to its strong tendency to form the hydrate 10. The reactivity of a species is traditionally related to its HOMO/LUMO (highest energy occupied molecular orbital/lowest energy unoccupied molecular orbital) gap. Thus, methanal was compared with ethanal, acetone, and the perfluorinated aldehyde and ketone parents. As shown in Table 3, methanal had a smaller HOMO/LUMO gap than the heavier parents, and it was close to the gaps found for the perfluorinated oxo-species. A more specific and well-established indicator of electrophilicity is available through the density functional theory (DFT) since it could be derived from the total density (i.e., the all-electron wavefunction) instead of the orbital energies (i.e., one-electron wavefunctions).
Thus, the electrophilicities were calculated at the most accurate level, using the total energies of the neutral, radical anion, and radical cation [18]: where µ and η stand for the chemical potential and the chemical hardness: VIP and VEA are the vertical ionization potential and electron affinities, respectively. Considering that negative electron affinities were involved, the VEA and VIP were obtained using the methodology proposed by Puiatti et al. [19,20]. According to the results summarized in Table 4, the calculated ω values suggest that methanal was a very strong electrophile. The higher electrophilicity of methanal also accounted for its ease of hydration, which is well-known in water. Table 5 reports the ∆G • hyd of methanal in a moderately polar solvent, using the ε of acetone as the IEFPCM model solvent [21]. Once again, methanal was found to be more similar to the electron poorer partner than to ethanal.

Attempts to Prepare Aminal 8a
Next, we revisited the preparation of aminal 8a, which was reported in 1902, as the product of the treatment of naphthylamine 2a with formalin in a molar ratio of 2:1 in acetone under reflux for five hours [22]. The obtained product (a yield was not given) was identified through elemental analyses (84.73 %C, 6.04 %H, and 9.25 %N) as aminal 8a (calcd. EA: 84.53 %C, 6.08 %H, 9.39 %N) and characterized as having a melting point of 104 • C (from ethanol).
The described procedure was then repeated; however, according to the 1D and 2D NMR spectra, the crude product mainly contained the starting naphthylamine 2a, quinazoline 6a, dinaphthylamine 17a, and bisquinazoline 16a in a molar ratio of 48:48:3:1, in addition to numerous trace products (Scheme 4). The presence of the expected aminal 8a could be deduced based on the 1 H NMR signal at 4.70 ppm, which was correlated in the HSQC spectrum with the 13 C signal at 53.16 ppm, and in the HMBC spectrum with the 13 C signal at 146.03 ppm. No clear signals of TB 3a, acridine 5a, or bisnapthylamine 18a were observed. The crystallization of the crude product from ethanol allowed us to obtain an insoluble part that contained quinazoline 6a (2% yield) and bisquinazoline 16a (3% yield) in addition to crystals of pure quinazoline 6a (35% yield). The calculated EA for both quinazoline 6a (85. 13  ethanol) was measured. Therefore, we conclude that the product published in 1902 [22] consisted of quinazoline 6a and not aminal 8a.
17a was surprising since it requires an attack of an R-CH2 moiety on the naphthalene which generally requires a process of acid catalysis. However, no acid was added t reaction, and the presence of acid would have led to the formation of TB 3a and/or spi 4a. However, these compounds were not observed in the crude product, but afte chromatography procedure, i.e., silica was acidic enough to catalyze the rearrange of 17a into 18a and to TB 3a formation.
An inspection of the molecular structure of quinazoline 6a revealed that it might formed through the cycloaddition of two molecules of imine 9a, which could have for during the reaction. However, the proposed mechanism was improbable, as suggeste our experiments on aminal 8b and imine 9b (vide infra). Scheme 4. Molecular structures of identified products with arbitrary numbering for NMR as ments.

Attempts to Prepare Aminal 8b
Scheme 4. Molecular structures of identified products with arbitrary numbering for NMR assignments.
The reaction was then repeated under the same condition; however, the crude product contained naphthylamine 2a, quinazoline 6a, and dinaphthylamine 17a in a slightly different molar ratio of 41:45:14, and surprisingly, no NMR signals of either bisquinazoline 16a or aminal 8a were identified. Owing to greater concentrations of dinaphthylamine 17a, we were able to assign all its 1 H and 13 C NMR signals and determine its molecular structure. Then, unlike in the original processing [22], we used column chromatography on silica instead of crystallization. Surprisingly, three fractions of various compositions were obtained. The 1 H NMR spectra were then used to calculate the total yields of the expected starting naphthylamine 2a (30% recovery) and quinazoline 6a (51% yield), and surprisingly only 1% yield of dinaphthylamine 17a. Moreover, acridine 5a (9% yield) and bisnapthylamine 18a (8% yield), which were present in all fractions, were also isolated alongside TB 3a (1% yield), which occurred in a single fraction.
Since neither bisnapthylamine 18a nor acridine 5a were present in the crude product, and dinaphthylamine 17a was isolated in a very low yield, a rearrangement of 17a into 18a during the chromatography could be considered, followed by the conversion of 18a into acridine 5a, since 18a was the expected intermediate product of 5a [1,23]. A similar rearrangement and formation of 18b when exposed to silica or to air and ambient light for prolonged periods have previously been described [24].
It should be emphasized that the formation of quinazoline 6a and dinaphthylamine 17a was surprising since it requires an attack of an R-CH 2 moiety on the naphthalene core, which generally requires a process of acid catalysis. However, no acid was added to the reaction, and the presence of acid would have led to the formation of TB 3a and/or spiroTB 4a. However, these compounds were not observed in the crude product, but after the chromatography procedure, i.e., silica was acidic enough to catalyze the rearrangement of 17a into 18a and to TB 3a formation.
An inspection of the molecular structure of quinazoline 6a revealed that it might have formed through the cycloaddition of two molecules of imine 9a, which could have formed during the reaction. However, the proposed mechanism was improbable, as suggested by our experiments on aminal 8b and imine 9b (vide infra).

Attempts to Prepare Aminal 8b
When the reaction of [22] was performed with methyl 6-amino-2-naphthoate (2b) instead of naphthylamine 2a, we observed a different behavior. While the reaction mixture was a homogenous solution for 2a, a white solid precipitated shortly after the addition of formalin to the solution of naphthylamine 2b, a white solid precipitated before slowly dissolving.
The reaction was repeated again, and the white solid intermediate was isolated by filtration (84% yield of crude aminal 8 after 2 h reflux; only 46% after 8 h reflux). The 1D and 2D NMR techniques allowed us to observe that the DMSO-d 6 solution of the white solid contained mainly aminal 8b, which was in equilibrium with hemiaminal 20b and starting naphthylamine 2b. When the sample was heated above 85 • C, the decomposition of both aminal 8b and hemiaminal 20b was observed, along with the formation of naphthylamine 2b, imine 9b, and traces of methanal ( Figure 1). Imine 9b was clearly recognized by the two doublets with 2 J HH = 16.2 Hz in the 1 H NMR spectra, which are typical for terminal N = CH 2 groups. However, the low contents (5% n/n) and low concentrations of imine 9b prevented us from identifying all of its NMR signals. When the heating was performed with the addition of water, only negligible amounts of imine 9b were formed, but a significant decomposition of aminal 8b and formation of methandiol (10) occurred (Scheme 5). When the sample was cooled back to 25 • C, the imine 9b was slowly converted back to aminal 8b and hemiaminal 20b, over a few hours. No formation of quinazoline 6b was observed.
by the two doublets with 2 JHH = 16.2 Hz in the 1 H NMR spectra, which are typical for terminal N=CH2 groups. However, the low contents (5% n/n) and low concentrations of imine 9b prevented us from identifying all of its NMR signals. When the heating was performed with the addition of water, only negligible amounts of imine 9b were formed, but a significant decomposition of aminal 8b and formation of methandiol (10) occurred (Scheme 5). When the sample was cooled back to 25 °C, the imine 9b was slowly converted back to aminal 8b and hemiaminal 20b, over a few hours. No formation of quinazoline 6b was observed.  When the reaction of [22] was performed with methyl 6-amino-2-naphthoate (2b) instead of naphthylamine 2a, we observed a different behavior. While the reaction mixture was a homogenous solution for 2a, a white solid precipitated shortly after the addition of formalin to the solution of naphthylamine 2b, a white solid precipitated before slowly dissolving.
The reaction was repeated again, and the white solid intermediate was isolated by filtration (84% yield of crude aminal 8 after 2 h reflux; only 46% after 8 h reflux). The 1D and 2D NMR techniques allowed us to observe that the DMSO-d6 solution of the white solid contained mainly aminal 8b, which was in equilibrium with hemiaminal 20b and starting naphthylamine 2b. When the sample was heated above 85 °C, the decomposition of both aminal 8b and hemiaminal 20b was observed, along with the formation of naphthylamine 2b, imine 9b, and traces of methanal ( Figure 1). Imine 9b was clearly recognized by the two doublets with 2 JHH = 16.2 Hz in the 1 H NMR spectra, which are typical for terminal N=CH2 groups. However, the low contents (5% n/n) and low concentrations of imine 9b prevented us from identifying all of its NMR signals. When the heating was performed with the addition of water, only negligible amounts of imine 9b were formed, but a significant decomposition of aminal 8b and formation of methandiol (10) occurred (Scheme 5). When the sample was cooled back to 25 °C, the imine 9b was slowly converted back to aminal 8b and hemiaminal 20b, over a few hours. No formation of quinazoline 6b was observed.

Preparation of Imine 9a under Acidic Conditions
The identification of imine 9b as an unstable compound in the presence of nucleophiles such as water or naphthylamine 2b (vide supra) prompted us to prepare imine 9a for comparison via known procedures.
The attempt to prepare imine 9a was likely first described in 1902 by Möhlau, who produced it as a product of the treatment of naphthylamine 2b with formalin in a molar ratio of 1:1 in ice acetic acid (a yield was not given) [22]. The molecular structure was suggested based on its EA (found: 85.36 %C, 6.07 %H, 9.17 %N; calcd. in 1902: 85.16 %C, 5.81 %H, 9.03 %N) and characterized by a melting point of 62-64 • C; no attempts were made to purify the compound.
However, when the procedure was reproduced, the obtained white solid contained quinazoline 6a, TB 3a, bisquinazoline 16a, and acridine 5a in a molar ratio of 71:23:3:3, contaminated with a few unidentified minor products. However, no 1 H NMR signals of imine 9a were observed.
In addition, after the aqueous acid filtrate had been left to stand overnight, a few milligrams of a greenish solid precipitated. The 1D and 2D NMR spectra of the product showed that the solid contained acridine 5a, bisnaphthylamine 18a (probably as salt 19a), TB 3a, dihydrogenquinazoline 7a, and quinazoline 6a in a molar ratio of 39:29:23:6:3; due to the high contents of acetic acid, all these compounds were at least partially protonated. The composition of the NMR sample (solution in DMSO-d 6 ) changed over one day. Bisnaphthylamine 19a disappeared while the acridine 5a contents increased, and the 1:1:1 triplet of a 14 N-ammonium cation appeared (7.12 ppm, 51.1 Hz); this confirmed the pathway for acridine 5a formation suggested by [9]. On the other hand, at least two unidentified compounds were formed.
However, when the procedure was reproduced, the obtained white solid contained quinazoline 6a, TB 3a, bisquinazoline 16a, and acridine 5a in a molar ratio of 71:23:3:3, contaminated with a few unidentified minor products. However, no 1 H NMR signals of imine 9a were observed.
In addition, after the aqueous acid filtrate had been left to stand overnight, a few milligrams of a greenish solid precipitated. The 1D and 2D NMR spectra of the product showed that the solid contained acridine 5a, bisnaphthylamine 18a (probably as salt 19a), TB 3a, dihydrogenquinazoline 7a, and quinazoline 6a in a molar ratio of 39:29:23:6:3; due to the high contents of acetic acid, all these compounds were at least partially protonated. The composition of the NMR sample (solution in DMSO-d6) changed over one day. Bisnaphthylamine 19a disappeared while the acridine 5a contents increased, and the 1:1:1 triplet of a 14 N-ammonium cation appeared (7.12 ppm, 51.1 Hz); this confirmed the pathway for acridine 5a formation suggested by [9]. On the other hand, at least two unidentified compounds were formed.
The formation of the oxo-compounds 21a and 22a as side products of a TB derivative preparation has never been reported. The only known oxo-TB analogs were previously prepared via formylation (via sBuLi and DMF) followed by aerial oxidation or through the direct oxidation of TB via KMnO4 (9 h reflux in CH2Cl2) [25]. Hence, the formation of the oxo-compounds 21a and 22a under our mild conditions over a two-minute reaction time was surprising. The low yields of the oxo-compounds 21a and 22a were relatively high compared to the yields of their possible precursors 3a and 7a, respectively. In addition, the most polar fraction resulting from the chromatographic separation was a complex mixture; however, the 1D and 2D NMR spectra revealed a set of signals which could be attributed to an exo-diastereomer of hydroxy-TB 23a formed as a possible intermediate product of oxo-TB 21a. Scheme 6. Molecular structures of oxo-TB 21a, oxo-quinazoline 22a, and an unidentified exo-diastereomer of hydroxy-TB 23a (racemic). Scheme 6. Molecular structures of oxo-TB 21a, oxo-quinazoline 22a, and an unidentified exodiastereomer of hydroxy-TB 23a (racemic).
The formation of the oxo-compounds 21a and 22a as side products of a TB derivative preparation has never been reported. The only known oxo-TB analogs were previously prepared via formylation (via sBuLi and DMF) followed by aerial oxidation or through the direct oxidation of TB via KMnO 4 (9 h reflux in CH 2 Cl 2 ) [25]. Hence, the formation of the oxo-compounds 21a and 22a under our mild conditions over a two-minute reaction time was surprising. The low yields of the oxo-compounds 21a and 22a were relatively high compared to the yields of their possible precursors 3a and 7a, respectively. In addition, the most polar fraction resulting from the chromatographic separation was a complex mixture; however, the 1D and 2D NMR spectra revealed a set of signals which could be attributed to an exo-diastereomer of hydroxy-TB 23a formed as a possible intermediate product of oxo-TB 21a.

Preparation of Imine 9a under Basic Conditions
Imine 9a formation was previously mentioned by Kadutskii in 2002Kadutskii in , 2006Kadutskii in , and 2012 through the treatment of naphthylamine 2a with paraformaldehyde (1:1) in ethanol in the presence of a catalytic amount of NaOH [26][27][28]. However, no spectroscopic data confirming the presence of imine 9a were reported. The formation of imine 9a in situ was only assumed based on the molecular structure of the isolated products, which could be considered the result of imine 9a reactivity.
Thus, the reaction was repeated and followed by NMR spectroscopy. Our analysis showed that the naphthylamine 2a was slowly consumed to reach an equilibrium with the majority of the imine-ethanol adduct 24a, followed by traces of the starting naphthylamine 2a and probably methanediol (10) and minor products (Scheme 7). Methanediol (10) was identified based on the singlet at 4.61 ppm in the 1 H NMR spectrum, which was correlated in the HSQC spectrum with a 13 C signal at 88.83 ppm, and exhibited no correlation in the HMBC spectrum (the chemical shifts may have been strongly affected by the presence of ethanol and sodium hydroxide).
presence of ethanol and sodium hydroxide).
When the sample was heated from 25 to 115 °C, no formation of imine 9a was observed, possibly due to the large excess of ethanol in the solution. This observation was similar to the heating of aminal 8b after the addition of water (vide supra).
Thus, a part of the reaction mixture was evaporated to dryness, and the residue was dissolved in 500 μL of DMSO-d6. A combination of 1D and 2D NMR spectra and mass spectrometry (MS) spectra revealed that the mixture contained not only adducts 24a and 25a but also its higher analog 26a. Newly, the mixture also contained aminal 8a and its higher analog 27a (Scheme 7). These compounds were not unambiguously proved through NMR spectra due to overlaid signals. Neither methanal nor imine 9a were observed. However, when the sample was heated to 115 °C, the formation of imine 9a was observed, and the 1 H NMR spectrum of imine 9a was obtained by subtracting the spectra obtained at 40 °C before and after heating (Figure 2). The formation of imine 9a at high temperatures was in accord with the formation of imine 9b and methanal (1) by heating aminal 8b and methanal equivalents, respectively (vide supra). It should also be emphasized that we observed a 1 H-13 C HMBC correlation of CH2 hydrogen atoms occurring exclusively with nitrogen-bearing carbon (well-separated signals around 145 ppm). This implies that there was no C-alkylation of the naphthalene core under these basic conditions, as is typical for reactions in the presence of an acid (the formation of TB derivatives) or in its absence (the formation of quinazoline). This means that the C-alkylation requires, at least not a too basic condition. In addition, since imine 9a was formed under these conditions in at least trace amounts, and quinazoline 6a was not observed as a product, the possible formation of quinazoline 6a by cycloaddition of imine 9a is unlikely. When the sample was heated from 25 to 115 • C, no formation of imine 9a was observed, possibly due to the large excess of ethanol in the solution. This observation was similar to the heating of aminal 8b after the addition of water (vide supra).
Thus, a part of the reaction mixture was evaporated to dryness, and the residue was dissolved in 500 µL of DMSO-d 6 . A combination of 1D and 2D NMR spectra and mass spectrometry (MS) spectra revealed that the mixture contained not only adducts 24a and 25a but also its higher analog 26a. Newly, the mixture also contained aminal 8a and its higher analog 27a (Scheme 7). These compounds were not unambiguously proved through NMR spectra due to overlaid signals. Neither methanal nor imine 9a were observed. However, when the sample was heated to 115 • C, the formation of imine 9a was observed, and the 1 H NMR spectrum of imine 9a was obtained by subtracting the spectra obtained at 40 • C before and after heating (Figure 2). The formation of imine 9a at high temperatures was in accord with the formation of imine 9b and methanal (1) by heating aminal 8b and methanal equivalents, respectively (vide supra).

Density Functional Theory Calculations
The processes studied here are summarized in Scheme 8, where the relative free energies of the main species refer to the energies of aminals 8a/b and two methanal mole- It should also be emphasized that we observed a 1 H-13 C HMBC correlation of CH 2 hydrogen atoms occurring exclusively with nitrogen-bearing carbon (well-separated signals around 145 ppm). This implies that there was no C-alkylation of the naphthalene core under these basic conditions, as is typical for reactions in the presence of an acid (the formation of TB derivatives) or in its absence (the formation of quinazoline). This means that the C-alkylation requires, at least not a too basic condition. In addition, since imine 9a was formed under these conditions in at least trace amounts, and quinazoline 6a was not observed as a product, the possible formation of quinazoline 6a by cycloaddition of imine 9a is unlikely.

Density Functional Theory Calculations
The processes studied here are summarized in Scheme 8, where the relative free energies of the main species refer to the energies of aminals 8a/b and two methanal molecules as ∆G • = 0. Views of the 3D structures of the main stationary points in Scheme 8 are available in the Supporting Information.
Molecules 2023, 28, x FOR PEER REVIEW 11 of 20 kcal/mol. Both reactions leading to the formation of imines 7a/b were more favored for 6b than for 6a. Another expected process for 6a/b (Scheme 8) was the entrance of another methanal molecule. At this stage, a partner of this quinazoline has already been proven as a key intermediate product formed in the last steps of the formation of TBs in the case of anilines as starting amines [17]. The attachment of the methylene unit involves the formation of an unstable i3tau intermediate that is rapidly tautomerized to i3. The activation barrier was found to be at 32.8 kcal/mol for 6a and was 2 kcal/mol higher for 6b. The intramolecular electrophilic attack of the alcohol carbon concerted with the water release (TS-i3-3) finally led to the formation of TB 3a/b. Once again, the process was easier for i3a than for i3b. The overall thermochemistry leading to the TB 3a/b starting from aminals 8a/b was found clearly exergodic by 30-31 kcal/mol. The thermal decomposition of aminals 8a/b yielding imines 9a/b and naphthylamines 2a/b was found to be slightly endergodic. The entrance of the first methanal to position 1 of the naphtylamine to yield i1a/b was found to be slightly exergodic, passing through transition states (TS) at 32.0 and 34.2 kcal/mol of the relative free energies for obtaining i1a and i1b, respectively. Then, the i1a/b species could react with the imine 9a/b to yield the adduct i2a/b, a process that was exergodic by about 1 kcal/mol and with a relatively small activation barrier. The intermediate products i2a/b could, in principle, close the quinazoline ring through an intramolecular Diels-Alder-like TS for yielding 6a/b. However, this process seems unlikely due to the high activation barrier of the TS-i2-6 compound (Scheme 8). A more likely pathway would have been the direct recombination of two imines 9a/b condensing to a tautomer of 6a/b through TS-9-6 tau , which had a relative free energy of about 20 kcal/mol lower than the former TS-i2-6. The intermediate product could easily tautomerize to 6a/b, which had remarkable stability of −14.1 and −14.7 kcal/mol for 6a and 6b, respectively.
Considering the stabilities of 6a/b, it is interesting to consider the different fates of this quinazoline. The thermochemistries of the formal dehydrogenation of 6a/b to yield 7a/b were found to be 2.3 and 4.7 kcal/mol endergodic in the cases of 6a and 6b, respectively. While the current calculations were performed considering a neutral acetone medium, the process has previously been described under different conditions, i.e., in an acid media [23]. Under those conditions, the overall thermochemistry of the process mediated by the reduction of imines 9a/b or reductive cleavage of 8a/b is shown separately in Scheme 9. Both processes had overall spontaneous thermochemistry by 18-19 and 13-15 kcal/mol. Both reactions leading to the formation of imines 7a/b were more favored for 6b than for 6a. Another expected process for 6a/b (Scheme 8) was the entrance of another methanal molecule. At this stage, a partner of this quinazoline has already been proven as a key intermediate product formed in the last steps of the formation of TBs in the case of anilines as starting amines [17]. The attachment of the methylene unit involves the formation of an unstable i3 tau intermediate that is rapidly tautomerized to i3. The activation barrier was found to be at 32.8 kcal/mol for 6a and was 2 kcal/mol higher for 6b. The intramolecular electrophilic attack of the alcohol carbon concerted with the water release (TS-i3-3) finally led to the formation of TB 3a/b. Once again, the process was easier for i3a than for i3b. The overall thermochemistry leading to the TB 3a/b starting from aminals 8a/b was found clearly exergodic by 30-31 kcal/mol.

Measurements and Materials
All chemicals and solvents were purchased from commercial suppliers and used without further purification. The NMR spectra were recorded on a 500 MHz instrument. The chemical shifts (δ) are indicated in ppm followed by their multiplicity, integral intensity, corresponding coupling constants (J) in Hz, and by the signal assignment, which is based on an analysis of ordinary 1 H-1 H COSY, 1 H-1 H NOESY, 1 H-13 C HSQC, and 1 H-13 C HMBC correlation spectra. The 2D spectra were recorded with high-resolution conditions utilizing the nonuniform sampling and processing with a linear prediction; hence the reproducibility of the chemical shifts was a few tens of ppb. The "cov." in the spectra description means that the signal is seriously covered by others, so a full description was impossible. The 1 H and 13 C APT chemical shifts are referenced to TMS (using the solvent signals CHCl 3 7.26 ppm, CDCl 3 77.0 ppm, CHD 2 SOCD 3 2.50 ppm, CD 3 SOCD 3 39.52 ppm). The MNOVA software was used for the processing, prediction, and simulation of NMR spectra. The interpretation of most spectra required a simultaneous analysis of various 1D and 2D spectra of two or more samples with different compositions. See also Supplementary Material. The mass spectra were obtained using electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI) with a linear quadrupole trap (LTQ) Orbitrap spectrometer. Silica (40-63 D, 60 Å) was used to separate the compounds by column chromatography.

Computational Procedure
The electronic structure calculations were all performed in the Gaussian 16. Rev. A03 package [21]. The structures of the reagents, transition states (TS), intermediaries, and products were all optimized at the level of theory CAM-B3LYP/6-311+G(d,p), with the solvent model SCRF-IEFPCM [29] (acetone). The CAM-B3LYP energies were corrected to consider the dispersion effects using the D3 version of the Grimme's dispersion with the Becke-Johnson damping, i.e., CAM-B3LYP-GD3BJ [30,31] (uncorrected energies available as Supplementary Material). The stationary points were characterized by their Hessian matrix, which was diagonalized to obtain the harmonic frequencies and then the zero point corrections to energy, enthalpy, and free energy. In relevant cases, an in vacuo intrinsic reaction coordinate (IRC) was determined using a mass-weighted step of 0.02 atomic units and by recalculating the Hessian every ten or 20 steps. The procedure was previously described and tested elsewhere [17,32]. For the relative free energies of Scheme 8, the ∆G • = 0 was taken as the energies of aminals 8a/b plus two methanal molecules to close the mass balance.
The level of theory and general procedure were previously described for the synthesis of symmetric and asymmetric TBs from anilines as starting reagents [17]. The procedure for obtaining the VEAs and VIPs to compute the electrophilicity using CAM-B3LYP was described in detail in [20].

Studies of Formalin and Paraformaldehyde
(a) An NMR tube was charged with 500 µL of DMSO-d 6 (standard quality) and 5 µL of formalin (ASC reagent, formaldehyde solution, 37% m/m in H 2 O, containing 10-15% methanol as a stabilizer) and closed with a gas-tight cup. The solution was monitored by 1 H NMR spectra at 25 • C. Equilibrium was reached within a few hours. The sample composition was determined by 1D and 2D NMR ( Table 1). The sample was heated to 50 • C, left to equilibrate (1-2 h), and analyzed by 1 H NMR. The same was done at temperatures of 75, 100, and 115 • C. After cooling back to 25 • C, the compositions were slowly returned to equilibrium (two days). (b) An NMR tube was charged with paraformaldehyde (1.0 mg, 33 µmol), DMSO-d 6 (0.5 mL), and water (10 µL, 555 µmol) and closed with a gas-tight cup. The mixture was shaken until the paraformaldehyde dissolved and then monitored by 1 H NMR at 25 • C. Equilibrium was reached within several hours. The sample composition was determined by 1D and 2D NMR (Table 2); the higher diols contents did not exceed 0.5‰ (n/n). The sample was heated following the procedure described in (a). (c) An NMR tube was charged with paraformaldehyde (1.0 mg, 33 µmol) and DMSO-d 6 (0.5 mL) and closed with a gas-tight cup. The mixture was shaken, but part of the paraformaldehyde remained undissolved. The sample was heated following the procedure described in (a).

Reaction of Naphthylamine 2b with Formalin under Neutral Conditions
Aqueous formaldehyde (37%, 0.1 mL, 1.23 mmol) was added to a solution of naphthylamine 2b (496 mg, 2.46 mmol) in acetone (20 mL). The mixture was refluxed for two hours (a white solid precipitated after 20 min). After cooling to room temperature, the solid was filtered off, washed with acetone, methanol, and diethyl ether, and dried to obtain 428 mg (84% yield) of crude aminal 8b. The filtrate was evaporated to dryness. According to the 1 H NMR and HRMS spectra, the obtained solid contained naphthylamine 2b, quinazoline 6b, and aminal 8b in a molar ratio of 81:15:4. When eight hours of reflux was, only a 46% yield of crude aminal 8b was obtained.
The NMR tube was charged with 1 mg of the crude aminal 8b and 500 µL of dry DMSO-d 6 and shaken until a solution was produced. The 1D and 2D spectra revealed the presence of aminal 8b, naphthylamine 2b, and hemiaminal 20b in a molar ratio of 71:23:6 at 25 • C. The sample was heated to 100 • C for an hour and monitored by 1 H NMR until it reached equilibrium. The 1D and 2D spectra revealed the presence of aminal 8b, naphthylamine 2b, hemiaminal 20b, and imine 9b in a molar ratio of 67:25:3:5, alongside traces of methanal.
Another NMR tube was charged with 0.64 mg of crude aminal 8b, 10 µL water, and 500 µL of dry DMSO-d 6 and shaken to get a clumsy solution. The 1D and 2D spectra revealed the presence of aminal 8b, naphthylamine 2b, and hemiaminal 20b in a molar ratio of 70:25:5 at 25 • C. The sample was heated to 100 • C for an hour and monitored by of a high excess of nucleophiles such as water, alcohols, or amines, we recommend calling them methanal equivalents instead of the usual methanal sources.
Finally, we have found that the formation of TB 3a and spiroTB 4a occurs rapidly even in acetic acid and that even silica can enable the formation of acridine 5a, dinaphthylamine 18a, TB 3a, and spiroTB 4a.
According to our experiments and DFT calculations, methanal is not behaving as the simplest aldehyde as is commonly assumed, but instead, its behavior resembled that of a very electron-deficient oxo-compound such as trifluoroethanal.
Computational calculations in the absence of acid catalysis showed plausible pathways for the formation of a stable intermediate product 6, which could evolve into imine 9 and even into TB 3. Under the conditions of the in silico study, both the imines and TBs are thermodynamically more stable than the aminals. Under these conditions, TB 3a was preferentially formed from 6a with respect to TB 3b.  Data Availability Statement: All data used to support the findings of this study are included within the article and Supplementary Materials. A few examples of simple NMR spectra are provided. The other spectra are available upon request as raw data files since their readability requires computer assistance due to their complexity.