RuO 4 -Mediated oxidation of N -benzylated tertiary amines. 2. Regioselectivity for N , N -dimethyl- and N , N -diethylbenzylamine as substrates

.


Introduction
In a previous paper 1 we studied the RuO 4 -mediated oxidation 2 of some N-benzylated cycloalkylamines and found that the attack occurs at both types of N-α-methylene positions, 3 i.e., endocyclic and exocyclic (benzylic).Proof of the incursion of the corresponding iminium cations as intermediates came from their capture as nitriles in the presence of cyanide anion (cyano trapping).Deprotonation of the endocyclic iminium cation to the respective cyclic enamine was observed too in the absence of cyanide.The statistically corrected regioselectivity (endocyclic/exocyclic) experienced by the mentioned substrates varied from 0.8 (morpholine derivative) to 2.1 (piperidine compound).Our results were highly different from those found in the literature.For instance, Bettoni et al. 3a have claimed the unique formation of endocyclic attack-derived compounds when starting from N-benzylpiperidine.
In the six-membered cycloalkylamines the endocyclic hydrogens are of two types (i.e., axial and equatorial) and this could influence, almost in principle, the regioselectivity.No such stereoelectronic constraints exist in the similar acyclic derivatives.Consequently, we decided to study the RuO 4 -mediated oxidation of N,N-dimethyl-(1A) and N,N-diethylbenzylamine (1B) and the respective results are presented in this paper.By analogy with the previously studied compounds, 1 the tertiary amines 1A-B could follow the transformations depicted in Scheme 1.Thus, two types of iminium cation might result in the first step, that is 2A-B (alkyl attack) and 3A-B (benzyl attack).These species are trapped by water and the resulting hemiaminals (4A-B and 5A-B, resp.) could undergo oxidation to the corresponding amides (6A-B and 7A-B, resp.), but also cleavage to amine+aldehyde equimolecular mixtures.For instance, 4A-B would give the corresponding secondary benzylamines 8A-B and the aliphatic aldehydes 9A-B; similarly, 5A-B could be cleaved to the secondary aliphatic amines 10A-B and benzaldehyde (11).In the case of 2B, which possesses an (N-β)C-H bond, deprotonation to the enamine 12 could also occur; oxidative cleavage of the C=C double bond in 12 would give 4 an equimolecular mixture of formamide 13 and formaldehyde (9A).As observed previously, 1 small amounts of the N-oxides 14A-B might also result from 1A-B.Finally, partial oxidation of the aldehydes 9A-B and 11 to the corresponding acids 15A-B and 16, respectively, can be also envisaged.
Scheme 1 might be correct if the indicated reaction products are inert against further transformation.As will be shown in the following, this was not the case especially because the secondary amines 8A-B underwent oxidation and other reactions by themselves.

Results and Discussion
Oxidation of 1A-B by RuO 4 (generated in situ from catalytic RuO 2 and NaIO 4 in excess), either without or in the presence of NaCN, was performed in the same conditions as before. 1 The identified reaction products and the corresponding yields are shown in Table 1 (entries 1-4).To understand better the behaviour of 1A-B, several control experiments were performed also and the respective results are partly listed in Table 1 (entries 5-11).In all reactions, benzaldehyde (11) was accompanied by small amounts of benzoic acid (16), whose yield was added to that experimentally found for 11.Accordingly, the yield of 11 in Table 1 means actually that of the 11+16 sum.The identification of the various reaction products was achieved by 1 H-and 13 C-NMR and also by GLC, but only 1 H-NMR spectroscopy was used for quantification.The spectral NMR features of all compounds of interest are presented in Tables 2 and 3 (see Experimental Section).
Before going into details, we can rule out the intervention of 14A-B as reactive intermediates in entries 1 and 2, respectively.An example is offered in entry 5 for the oxidation of 14B itself and this should be compared with the results of entry 2. Analogously to our previous findings, 1 the N-oxide 14B was far more resistant than the corresponding amine (see the substrate conversions in column 2).Moreover, its reaction products covered only some of those shown by the oxidation of 1B and resulted in very different relative yields.The N-oxide 14A behaved similarly (reaction not shown in Table 1).The origin of 17A-B, 18, 19A-B, 20, and 21A-B in entries 1 and 2 should be the oxidation of 8A-B.Although the RuO 4 -oxidation mechanism of secondary amines 8A-B is unknown and its elucidation is out of the scope of this paper, some considerations on it are necessary.We imagined as a working hypothesis the upper part of Scheme 2. Thus, the functionalization at the alkyl site might give 17A-B and 18+9A-B.Analogous reaction at the benzyl site would yield 19A-B and 11+23A-B.It is well known that benzaldehyde reacts easily with primary amines to give the corresponding Schiff bases, thus explaining the formation of 20 and 21A-B.Obviously, other pathways are possible.For instance, we do not exclude the direct formation of 21A-B, followed by its partial hydrolysis to 11+23A-B (not depicted in Scheme 2).At the same time, the oxidation of benzylamine (18) itself might give a mixture of 11 and 20.Whatever would be the real steps involved, it is clear that some benzaldehyde could result from the oxidation of 8A-B, either directly or via 18 (or from 21A-B).In other words, the yields of 11 in entries 1 and 2 seems to be the sum of contributions due to both tertiary (1A-B) and secondary (8A-B) amines oxidation.
According to Scheme 1, some aliphatic aldehydes (9A from 1A; 9A-B from 1B) accompany the formation of 8A-B and 13 when starting from 1A-B.Along with them, discrete amounts of the corresponding acids 15A-B could be present also, as indicated by the partial oxidation of 11 to 16 found experimentally.This means that the secondary amines 8A-B, formed from 1A-B, were actually in the presence of all these aliphatic aldehydes and acids.It is known 5 that 8A reacts really with formaldehyde (9A) and formic acid (15A) to give mainly the tertiary amine 1A, by the consecutive reactions The reaction goes on even with a molar deficit of 9A vs. 8A, but 15A must be in excess.This sequence suggests that the inverse transformation of 8A into 1A could be possible in the reaction conditions of entry 1. Extension of [a]-[c] to the case of 1B (entry 2) seems logical only for the reactions 8B+9A-B ↔ 4A-B and 4A-B+H + ↔ 2A-B+H 2 O. Indeed, the subsequent transformation of 2B into 1B is unlikely because acetic acid (15B) can not be oxidized similarly to formic acid (15A) as in reaction [c].On the contrary, the [c]-like step 2B+15A could occur and generates an unsymmetrical amine, i.e., N-ethyl-N-methylbenzylamine (1C).However, compound 1C (and/or its oxidation products) 6 has been never detected as an outcome of 1B or 8B (entry 2 or 7, resp.).Therefore, 8B can not be a source for 1B (and/or 1C).On the other hand, because 6A and 13 have been really obtained 7 by direct formylation of 8A-B, respectively, with 15A, this new route might be also possible during the RuO 4 -oxidation of 8A-B.To test these suppositions several control experiments were performed and the respective results are presented below.
Oxidation of 8A was repeated in identical conditions as those of entry 6, but some formic acid (entry 8) or formaldehyde (entry 9) was added from the beginning of the reaction.In the former case, only the relative yield of 6A was raised (by a factor of two), which represents a disappointingly low molar consumption of about 6% of the extra 15A. 8The relative yields of other reaction products were little influenced, including that of 22.At the same time, the substrate conversion dropped from 75% (entry 6) to 55%, probably because 8A has been subtracted to oxidation by protonation.Consequently, in our conditions, little (if any) formamide 6A might result by direct formylation of 8A with 15A.In the case of entry 9, the yields of 6A and 22 were three times higher with respect to those of entry 6, counting for more than 80% of 9A introduced; 8 obviously, the yields of the other reaction products were reduced consequently, but their relative ratios remained unaffected.The substrate conversion was also the same.Because 9A was in excess, some raise in the 15A concentration could be envisaged too on passing from entry 6 to 9.This should cause a marked decrease of the substrate conversion, which does not fit our experimental results.This means that the aforementioned variation of yields can be ascribed mainly to the extra 9A influence.Accordingly, formaldehyde (9A) participated really in the formation of both 6A and 22.Moreover, because the yields of 6A and 22 varied in an identical manner, an intermediate giving both 6A and 22 seems to be involved.
We repeated also the oxidation of 8B in the presence of added acetaldehyde (entry 10) and found higher yields of 6B and 13 (each by a factor of 3) with respect to those of entry 7; this represents a molar consumption of 46% of the added 9B. 8 Consequently, acetaldehyde played for 8B a role similar to that of formaldehyde to 8A.To the difference of the effect of added formic acid (entry 8), initial addition of acetic acid in the reaction mixture of 8B (reaction not shown in Table 1) caused only a smaller substrate conversion; the yields of the various reaction products remained unchanged, within the experimental errors.Finally, we checked the oxidation of benzylamine (18) and found its transformation into an 11+20 mixture, as expected (entry 11).
We never detected 1A in the experiments of entries 6, 8, or 9. Taking into consideration the identical substrate conversion in entries 1, 6 and 9 and the identical relative yields of 6A and 22 in entries 6 and 9, a detectable amount of 1A should be present in entry 9 if hypothesis A was acting.Consequently, the absence of 1A and the aforementioned considerations favor the hypothesis B. The same seems to be true also for 8B.
With all these facts in mind we are now able to rationalize the transformation of 8A into 6A+22 and of 8B into 6B+13 as shown in the lower left corner of Scheme 2. Condensation of 8A-B with 9A-B gives 4A-B, which will suffer oxidation to 6A-B and formal dehydration to 2A-B.The latter reaction could be assisted by the acids 15A-B and/or 16.Cation 2A alkylates the starting amine 8A to yield the dimer 22, but 2B prefers to give an equimolecular mixture of formamide 13 and formaldehyde (9A), via deprotonation to 12.When 9A-B are in excess (entries 9 and 10, resp.), the equilibria 9A-B+8A-B ↔ 4A-B are pushed more to the right, thus explaining the identical increase of the yields belonging to 6A+22 and 6B+13, respectively.The steps showing the transformation of 8A-B into 4A-B and 2A-B, depicted in Scheme 2, are practically the inverse pathways invoked in Scheme 1. Actually, this was the reason for which these reactions have been written as equilibria.Even resulting from different reactions, the species 2A-B and 4A-B are common intermediates in the oxidation of both secondary (i.e., 8A-B) and tertiary amines (i.e., 1A-B).The formation of the same reaction products (i.e., 6A-B, 13, 22) in these two cases now finds an explanation.

B. Regioselectivity and cyano trapping
In order to calculate the alkyl/benzyl regioselectivity of the 1A-B oxidation we need to know the yields of all compounds derived from 2A-B and 3A-B (Schemes 1 and 2).According to Section A, the compounds 17A-B, 18, 19A-B, 20, and 21A-B originate all from 8A-B and therefore are 2A-B-derived species.At the same time, the yields of 6A-B, 13, and 22 quoted in entries 1 and 2 are the sums of those deriving from RuO 4 +1A-B (Scheme 1) and the ones originating from 8A-B+9A-B (Scheme 2).This does not influence the regioselectivity calculation because 6A-B, 13, 22 and 8A-B originate all from the initially formed 2A-B.However, benzaldehyde (11) results from both 3A-B (Scheme 1) and 2A-B, via 8A-B (Scheme 2).Separate contributions can not be calculated because the corresponding kinetic data are not known.Moreover, the total amount of 11 is also unknown, because, apart from the quantifiable consumption to yield 20, it is not clear if 21A-B are initial reaction products or the results of the 11+23A-B reaction.This means that the regioselectivity can not be calculated using the data from entries 1 and 2. On the contrary, the calculation became possible for the reactions performed in the presence of NaCN, as discussed below.

Scheme 3
Similarly to our previous paper, 1 the iminium intermediates 2A-B and 3A-B, generated in situ from 1A-B, were efficiently trapped by cyanide anion as the nitriles 24A-B and 25A-B, respectively (Scheme 3).Small amounts of benzylamides 6A-B (and 13 from 1B), benzaldehyde (11), and the respective secondary amine (8A-B) resulted also, but at least 94% of the reacted substrate was recovered as nitriles (Table 1, entries 3 and 4, resp.).After optimization, this reaction might be used to prepare the corresponding α-aminoacids. 9Consequently, the RuO 4 /NaCN oxidation of tertiary aliphatic amines could be viewed as a useful, non electrochemical one step-synthesis of α-aminonitriles. 10The N-oxides 14A-B did not react in these conditions, confirming their non-implication as reactive intermediates in the oxidation of 1A-B.
These results allowed us to estimate the alkyl/benzyl regioselectivity (RS) of the respective oxidation reactions.To calculate RS, we must divide the yields' sum of 6A-B+8A-B+24A-B (+13 for 1B) to that of 11+25A-B (entries 3 and 4).Obviously, these ratios must be corrected statistically, by dividing them by three for 1A-and by two for 1B-derived compounds.However, as shown in Section A, benzaldehyde (11) originated from 3A-B and from 2A-B, via 8A-B.Despite this uncertainty, the regioselectivities can be calculated because the yield of 11 is too small to affect significantly the results.The corresponding RS values are quoted in the last column of Table 1.
From the regioselectivity point of view, it emerged that in both 1A-B as substrates the alkyl group is the preferred attacked site.At the same time, the methylic C-H bond (as that in 1A) resulted to be two times more active than a methylenic one (as that in 1B).
Compound 1B is structurally more similar than 1A to the previously studied case 1 of Nbenzylpiperidine.As mentioned in the Introduction, the last compound presented a regioselectivity of 2.1, that is an identical value to that found now for its acyclic analog 1B.It results that RuO 4 is a too powerful oxidant to discriminate axial and equatorial C-H bonds in a piperidine ring.
For the amines 1A-B, the reaction course is well described by Schemes 1 and 2 (or 3), but the first step remains still unspecified.Several possibilities might be advanced for the formation of iminium cations, but we cite only three: (i) hydrogen-atom-transfer (HAT), (ii) electron-transfer (ET), or, by analogy with the RuO 4 -oxidation of esters, 11 (iii) a concerted mechanism with an S E 2-like transition state.We discovered previously 12 that 1B undergoes oxidation to 8B and 11 in bona fide HAT or ET conditions with regioselectivities (alkyl/benzyl) of 0.7 and 0.4, respectively.These values are significantly different from that of 2.1 found for 1B in the present paper.This seems disfavoring a HAT or ET mechanism for the RuO 4 -oxidation.However, because the rate-determining step is unknown, only a kinetic study might clarify the real nature of the involved mechanism. 9

Conclusions
Oxidation by RuO 4 of N,N-dimethyl-(1A) and N,N-diethylbenzylamine (1B) took place at both types of their (N-α)C-H bonds, that is alkyl and benzyl, giving initially, on one hand, benzylamides 6A-B (and 13 from 1B or 22 from 1A) and the corresponding monoalkylbenzylamines 8A-B and, on the other hand, benzamides 7A-B and benzaldehyde (11), respectively.Small amounts of the corresponding N-oxides 14A-B were formed too by a side, minor reaction.The first oxidative step was ascribed to the formation of the corresponding iminium cations 2A-B and 3A-B, trapped as nitriles by added NaCN.In these last reaction conditions, the alkyl/benzyl regioselectivity was 4.1 and 2.1 for 1A-B, respectively.Comparison of the regioselectivity values belonging to 1B and N-benzylpiperidine indicated RuO 4 as being a too powerful oxidant, unable to distinguish between axial and equatorial C-H bonds in the latter compound.In the absence of NaCN, the secondary amines 8A-B complicated the reaction outcome of 1A-B by their own oxidation to 11, benzylamine (18), Schiff bases (20, 21A-B), traces of N-monosubstituted amides (17A-B, 19A-B), and also to all other compounds written before as originating from 1A-B, unless 7A-B and 14A-B.Formation of these common products was attributed to the reaction of 8A-B with formaldehyde (9A) or acetaldehyde (9B), generated during the oxidation.Both oxidation of 1A-B (or 8A-B) and the reaction 8A+9A (or 8B+9B) occurred through some common intermediates (i.e., hemiaminals 4A-B and benzyliminium cations 2A-B).The N-α-C .carbon-centered radical or the amine cation radical, as requested by a HAT or ET mechanism, respectively, seemed not to be involved as precursors during the generation of 2A-B.

Experimental Section
General Procedures.The GLC and NMR apparatuses and procedures were already described. 1elting points were taken with a Boetius hot plate and are uncorrected.Materials.Hydrated ruthenium dioxide, 1A, 8A-B, 17A, 18, 20, 21A (all from Aldrich), and sodium periodate (Merck) were used as purchased.Carbon tetrachloride (Chimopar) was stored over anhydrous Na 2 CO 3 and filtered prior to use.Compounds 1B, 12 6A-B, 7 7A, 13 7B, 14 23 are all known and were synthesized according to the indicated procedures.

N,N-Diethylbenzylamine N-oxide (14B
).To a solution of 1.5 g (9.2 mmol) of 1B in 5 mL of methanol, heated at 50-55°C and stirred, aliquots of 0.15 mL each of hydrogen peroxide (30%) were added every 15 minutes.After the ninth addition (total H 2 O 2 : 1.35 mL; 13.2 mmol), the stirring was maintained for 3 hours at the same temperature.The reaction mixture was evaporated in vacuo and the resulting solid was triturated with ether in order to obtain  2 and 3. NMR Spectra.The 1 H-and 13 C-NMR features of all compounds of interest are collected in Tables 2 and 3, respectively, unless those of 1A, 24a 8A, 24b 8B, 24c 11, 24d 16, 24e 17A, 24f 18, 24g and 20, 24h as being easily accessible.The 1 H-and 13 C-NMR chemical shifts are expressed with respect to internal (CH 3 ) 4 Si (0 ppm) and CDCl 3 (77 ppm), respectively.
Oxidation by RuO 4 (+NaIO 4 ).The previous procedure 1 was slightly modified as concerning the work-up.To a mixture of CCl 4 (5 mL) and aqueous NaIO 4 solution (10 mL, 0.4M) hydrated RuO 2 (10 mg) was added, followed immediately by one mmol of substrate dissolved in 5 mL of CCl 4 .In the case of solid 14A-B, which are insoluble in CCl 4 , RuO 2 was added to a CCl 4 /aq.NaIO 4 (10/10; mL/mL) mixture, followed by the N-oxide added as such.In all cases the whole mixture was magnetically stirred for 4-7 hours at room temperature.Aqueous 2.5M NaOH solution (2 mL) was added, the mixture stirred for 15 minutes, filtered, and the layers separated.The filter cake was well triturated with fresh CCl 4 and water and the filtration and separation repeated.The CCl 4 -and aqueous layers were combined separately to yield organic (I) and aqueous mixture (II), respectively.A known aliquot of mixture I was freed from solvent (in vacuo, max.bath temperature of 50°C) to give the residue Ia.Mixture II was continuously extracted with CH 2 Cl 2 and the two layers separated.The organic phase was dried (Na 2 SO 4 ) and the solvent evaporated as before to leave the residue IIa.The remaining aqueous layer was acidified with concentrated HCl and the continuos CH 2 Cl 2 -extraction repeated.Evaporation of the dried organic layer gave the residue IIb.Aromatic ipso carbons are quoted in italics.c Two E/Z isomers are present; the values of the major one are underlined.
Identification of the various reaction products was mainly performed by 1 H-and 13 C-NMR spectroscopy using solutions in CDCl 3 of residues Ia, IIa, and IIb.Small amounts of unambiguously synthesized or commercial compounds were added into the analyzed sample and the spectra compared.Additionally, GLC was used too to identify the most volatile constituents of mixture I.For this purpose, the mixture I was extracted with aqueous 2.5 M HCl solution, washed with water until neutral, dried over Na 2 SO 4 (mixture Ib), and analyzed for non basic constituents.The acidic aqueous layer was basified with NaOH, well extracted with CH 2 Cl 2 , and the organic layer (mixture Ic) analyzed for basic compounds.Identification was achieved by GLC peak superposition in the presence of authentic materials.Because the acidic treatment of mixture I destroyed most of 20 and 21A-B leaving additional 11, the GLC analysis could not be used for quantitative measurements.Sometimes, the mixtures Ib and Ic were evaporated and the respective residues analyzed by NMR, as before.As an example, the distribution of the identified compounds derived from 1A was the following: 1A (unreacted), 6A, 7A, 8A, 11, 18, 20, 21A, and 22 in mixture I; 6A, 7A, 22 (all three in relatively small amounts), 14A, 17A, and 19A in residue IIa; 16 in residue IIb.
Quantification of the reaction products was achieved by 1 H-NMR on mixture I and residues Ia, IIa, and IIb (all in CDCl 3 as a solvent), in the presence of known amounts of an internal standard (cyclohexane or dichloromethane).Analysis of the more diluted mixture I was indicative only for the main constituents.The amounts of its minor constituents were estimated by the correlation with the analysis of residue Ia (Note).In the case of 1A-B or 8A-B the mixture I accounted for 75-95% of the recovered materials.Synthetic mixtures of all desired compounds were worked up as before in order to determine the corresponding losses.These results were then used to correct the experimentally found amounts.Note.The solvent evaporation (i.e., I→Ia) implied uncontrollable losses of 1A-B, 8A-B, 11, and 18 (due to partial evaporation) and partial consumption of 11 and 18 to give additional 20.Some hydrolysis of 21A-B occurred too.Correlation of I-and Ia-data was possible because the amounts of 22 (for 1A or 8A) and of 6B (for 1B or 8B) were not influenced by evaporation.The yields in Table 1 correspond to their initial amounts.Cyano trapping.The previously described procedure 1 was followed, but the work-up was identical to the newly proposed one.The acidification and the subsequent steps were performed carefully in a good hood.

Table 2 .
1H-NMR data a a Data useful in product identification are listed only.b Proton coupling constants (J) are given in Hz.Benzylic hydrogens are abbreviated as Bn.c Two E/Z isomers are present; the values of the major one are underlined.

Table 3 .
13C-NMR data a a Data useful in product identification are listed only.b Benzylic carbons are abbreviated as Bn.