Acetophenone Mannich bases: study of ionic liquid catalysed synthesis and antioxidative potential of products

Three-component Mannich reaction of acetophenone or 4-iodoacetophenone with a variety of substituted anilines and benzaldehyde, catalysed with diethanolammonium chloroacetate, was performed under mild conditions. Mannich bases (MBs), of which five are new, were obtained in good to excellent yields. All compounds were characterized using elemental analysis, NMR and IR. In addition, detailed experimental and simulated UV–Vis spectral characterization of these compounds is presented here for the first time. In vitro antioxidative potential of synthetized MBs was evaluated using 2,2-diphenyl-1-picryl-hydrazyl radical and density functional theory (DFT) thermodynamical study. It was shown that compounds with anisidine moiety express moderate antioxidative activity. Mechanism of the organocatalysed Mannich reaction was thoroughly inspected by means of DFT. The reaction undergoes the hydrogen bonding-assisted mechanism. Moreover, the proposed rate determining step of the overall reaction is water elimination in the process of iminium ion formation. To the extent of our knowledge, this is the first detailed report on the influence of this type of catalyst on the formation of iminium ion, as a crucial intermediate for the whole reaction.


Introduction
The Mannich reaction is one of the most important methods for the formation of a new C-C bond and synthesis of aminocarbonyl molecules [1][2][3][4] (20 mol%) and 1 ml of ethanol were placed in flask and stirred at room temperature for 24 h. When the reaction was completed, the solid product was separated by filtration and washed with ethanol. The MBs were obtained by recrystallization from chloroform and propanol 2 : 1 (v/v) and identified by 1 H NMR, 13 C NMR, IR spectroscopy, melting point and elemental analysis. Upon the evaporation of the solvent, the remaining catalyst was used in new experiments. It is worth highlighting that there was no significant loss in the recycled catalyst activity, i.e. the yields were up to 5% lower. The characterizations of the new compounds are given in main part of the manuscript, while for known compounds in the electronic supplementary material, as well as 1 H and 13 C NMR spectra for all compounds.

DPPH free radical scavenging assay
The free radical scavenging activity of the examined compounds was performed using the DPPH method [45]. DPPH solution (1 ml, 0.05 mM) in methanol was mixed with the tested compound (20 ml of compound solution in DMSO and 980 ml of methanol). After incubation at room temperature for 20 and 60 min, the absorbance was determined spectrophotometrically at 517 nm. Methanol was used as a control solution. IC 50 values represent the concentration necessary to obtain 50% of a maximum scavenging capacity. Nordihydroguaiaretic acid (NDGA) was used as positive control. All measurements were performed on three replicates. The results are presented as mean values + standard deviation (s.d.) of three independent measurements (electronic supplementary material, tables S4 and S5).

Details of calculations
All calculations were performed with the Gaussian 09 software [46]. M052X functional in conjunction with 6-311þþG(d,p) basis set for all atoms except iodine, where 6-311G(d,p) was used [47,48]. The geometrical parameters of all stationary points were optimized in ethanol (1 ¼ 24.3) using the conductor-like solvation model (CPCM) [49,50]. The nature of all calculated structures was determined by frequency calculations: all positive eigenvalues for equilibrium structures, and one negative eigenvalue for transition states. The natural bond orbital analysis (Gaussian NBO version) was performed for all structures. Simulations of UV-Vis spectra were performed using TD-DFT and structures optimized in methanol, because experimental spectra were acquired using this solvent. For the simulation of UV-Vis spectra M052X and B3LYP methods were used. The thermodynamic parameters and reaction enthalpies were obtained by the optimization of all relevant species in methanol, because it was used for experimental in vitro radical scavenging determination, using M052X method, 6-311þþG(d,p) basis set for all atoms except iodine, where 6-311G(d,p) or def2-TZVP (Triple-Zeta-Valence basis set with polarization and effective core potentials) basis sets were used, and the CPCM [51].

Results and discussion
In an extension of our study of the Mannich coupling reaction, we conducted reactions of benzaldehyde, anilines and aromatic ketones (acetophenone or 4- It is important to emphasize that, under these optimal conditions, only formation of the Mannich product was observed, unlike other similar Mannich reactions where aldols were detected as side products [52]. Reactions were carried out with benzaldehyde and differently substituted anilines. The highest yields were obtained in cases where in the para position of aniline are electron withdrawing halogen substituents or with the aniline itself (table 2).

1 H NMR and UV -Vis spectral characterization of MBs
All obtained products were characterized by NMR, IR and UV-Vis spectra, elemental analysis and melting points. It is worth pointing out that UV-Vis spectral characterization of these compounds will be discussed here for the first time. The signals observed in the 1 H NMR spectra of the MBs are given in table 3, and corresponding atoms labelling in scheme 1. In the 1 H NMR spectra of all MBs, detected signals originating from C 2 methylene group appear in the region 3.30 -3.59 ppm as doublet of doublet or multiplets. Proton bonded for asymmetric C 3 carbon atom appears in the range of 4.89-5.03 ppm as triplet or multiplets. In the spectra of the MBs protons of the amino group appeared at about 4.50 ppm as broad singlets exchangeable with D 2 O. The protons attached to the  Table 3. Chemical shifts of protons (ppm) in the MBs skeleton of MB1 -MB10 ( 1 HNMR spectra). In all experimental UV-Vis spectra of MB1 -MB10, there are two major absorption bands around 205 and 260 nm, as well as a small band around 310 nm (figure 1 and electronic supplementary material, figure S3). To explore which transitions are responsible for the appearance of each absorption band, UV-Vis spectra of these compounds were simulated using TD-DFT (figure 1 and electronic supplementary material, figure S3), and Kohn -Sham orbitals were constructed (electronic supplementary material, figures S4 -S13), while electron transitions and corresponding orbital energies are provided in electronic supplementary material, tables S1 -S4. Spectra were simulated using M052X and B3LYP methods. Surprisingly, the B3LYP method failed in the reproduction of absorption spectra. Namely, in all spectra bands are either significantly redshifted or completely absent. Fleming et al. showed that the inaccuracy in l max can be caused by the solvation model used [53]. On the other hand, in their study, all experimental bands were present in simulated spectra. Bearing in mind that in our study some of the bands were completely absent when the B3LYP method was used, results obtained with the M052X method will be discussed here.
In all M052X simulated spectra, absorption bands are significantly redshifted. For this reason, simulated values were scaled using scaling factor obtained with the least square method. Here, details on the origin of absorption bands of MB1 will be presented, while all other relevant data regarding compounds MB2 -MB10 is provided in the electronic supplementary material. Experimental absorption band around 202 nm is a consequence of several electronic transitions, represented with two bands found at 204 and 206 nm. The first one is a consequence of HOMO-5 to LUMO þ 1 electron transition, while the second one of HOMO-4 to LUMO þ 2, HOMO-4 to LUMO þ 4, HOMO-3 to LUMO þ 3 electron transitions. Based on the localization of orbitals responsible for these electronic transitions, as well as on their relative energies (electronic supplementary material, figure  S4, tables S1 and S3) one can see that they are a consequence of large energetical but small spatial separation. Simulated spectra revealed that experimental band at 260 nm is also a consequence of several electronic transitions, appearing as two bands at 235.2 and 275 nm. Lower wavelength band is a consequence of relatively small spatial separation (HOMO to LUMO þ 4, HOMO to LUMO þ 5), while the higher wavelength band of relatively smaller energy gap (HOMO-2 to LUMO). Small absorption band at 314 nm appears owing to HOMO to LUMO electron transition, represented by the smallest energy gap, but also with a relatively small spatial gap.

Radical scavenging activity of MBs
In vitro antioxidant activity of synthetized MBs was determined on the basis of their interaction with the 1,1-diphenyl-2-picryl-hydrazyl (DPPH) stable free radical (table 4 and (table 4). Since all investigated compounds do not have phenolic OH group, it is reasonable to expect that the NH group is responsible for the scavenging of DPPH radical. Furthermore, it is obvious that the presence of the methoxy group in the para position of the ring A plays a decisive role in the antioxidative activity. Namely, only presence of this strong electron donating group in the aniline moiety of MBs induced increased antiradical activity.
To determine the most probable mechanism of radical scavenging of MB8 and MB10 (hydrogen atom transfer (HAT), single electron-proton transfer (SET-PT) and sequential proton-loss electron-transfer (SPLET)), DFT was employed, homolytic/heterolytic dissociation of N-H bond was investigated. For estimation of the mechanism in the absence of free radicals, bond dissociation enthalpy (BDE), ionization potential (IP) and proton abstraction (PA) energies were calculated [13,[54][55][56][57][58][59][60], while in the presence of free radicals, the reaction enthalpies (DrH ) were calculated for the reactions of MB8 and MB10 with each of the eight selected free radicals. Details on terms upon which radical selection was made (hydroxy ( † OH), hydroperoxy ( † OOH), methylperoxy (CH 3 -O-O † ), superoxide radical anion (O 2 †2 ), methoxy ( † OCH 3 ), tert-butoxy ( † OC(CH 3 ) 3 ), vinyl peroxy (CH 2 ¼CH-O-O † ) and DPPH), as well as details on the calculation of thermodynamic parameters in the absence and in the presence of free radicals are provided in our previous studies, and for the sake of clarity presented in the electronic supplementary material [13,61]. Nevertheless, thermodynamical parameters and reaction enthalpies indicate which of HAT, SET-PT and SPLET mechanism prevails.
At first glance, based on the obtained high positive values for IP and DH IP , SET-PT is not a preferred mechanism for the radical scavenging. In the absence of free radicals, for both compounds, significantly lower values for PA than for BDE undoubtedly show that preferred antiradical mechanism is SPLET. On the other hand, enthalpies obtained for the reactions with free radicals point out HAT as a dominant antioxidative mechanism. Namely, for both MB8 and MB10, values for DH BDE are considerably lower than those obtained for DH PA . Similarly, as in the case of phenolic antioxidants, reaction enthalpies for the investigated aromatic amino carbonyl compounds are strongly influenced by nature and reactivity of quenched radical. The values for DH BDE increase as reactivity of radical species decrease. The most exothermic reaction is in the case of † OH radical, which is the most reactive among tested radicals. Somewhat higher, but still low DH BDE values are obtained in the case of † OCH 3 and † OC(CH 3 ) 3 radicals. Low negative or low positive reaction enthalpies for peroxy radicals indicate much slower reactions, while considerably endothermic nature of the reaction with O 2 †2 clearly shows that this radical will be poorly scavenged.
In addition to reaction enthalpies, change in Gibbs free energies was also inspected (electronic supplementary material, table S6). Obtained free energies indicate that there is almost no difference in reactivity obtained based on reaction enthalpies. The only deviation is in the case of PA of both MB-8 and MB-10. Namely, as in the case of PA obtained from enthalpies (table 4), change in Gibbs free energy shows that SPLET will be preferred mechanism in the absence of free radicals and points out MB-10 as a slightly better radical scavenger. On the other hand, results obtained based on Gibbs free energies indicate MB-8 as a better radical scavenger.
Bearing in mind that for iodine atom relativistic effects are not negligible and using 6-311G(d,p) basis set might provide some inaccuracy, additional calculations with the larger def2-TZVP basis set were done (electronic supplementary material, table S7). The obtained results were almost the same as those calculated with 6-311G(d,p) basis set. This is most probably a consequence of the iodine atom being distanced from the reaction centre.

Mechanistic study for the formation of MBs in the presence of [HDEA][ClAc]
To explore the role of the catalyst in the reaction, the mechanistic study was performed using DFT. In our previous studies, we explored diastereoselectivity of the Mannich reaction catalysed with some ethanolamine-based ILs [62,63]. In those investigations, it was assumed that the catalyst takes active role only in dehydration step, while the C-C coupling is not catalysed. In this paper, mechanistic pathways of the organocatalysed reaction of acetophenone, benzaldehyde and aniline in the presence of [HDEA][ClAc] (figure 2, scheme 2) and pathways for the non-catalysed reaction (electronic supplementary material, figure S1) were investigated. A comparison of the structures and corresponding energies along these paths enables one to inspect the role of the catalyst in the first (C-N bond formation) and the second (C-C bond formation) steps. In the first step, the formation of iminium ion 6 via protonation of the substrates was inspected. Additionally, the influence of ethanol  as a solvent molecule on the formation of iminium ion was explored. As far for the second phase of the reaction, i.e. C-C bond formation, we followed up the nucleophilic attack of the enol form of acetophenone to iminium ion 6, in the presence of DEA and [ClAc], which are yielded during the reaction from [HDEA] [ClAc]. Reaction is initiated by the cationic part of the catalyst [HDEA], which contributes to the C-N bond formation via hydrogen bonding with the oxygen atom of the carbonyl group of benzaldehyde. Simultaneous nucleophilic attack of aniline nitrogen atom to the carbonyl group, along with the protonation of oxygen of carbonyl group of benzaldehyde by the catalyst (1-2TS) enables the formation of new C-N bond. We considered the possibility that C-N bond formation can be achieved with the support of [HDEA]. Namely, we assumed that [HDEA], as cationic part of the catalyst, and oxygen from carbonyl group of benzaldehyde can form a hydrogen bond, and as a consequence of N-H . . . O ¼ C hydrogen bonding, the electrophilicity of carbonyl carbon atom will be increased. This assumption was confirmed with the NBO analysis of 1. Comparison of the NBO charges in 1 and SB-1 revealed that there is a slight increase in positive charge on this atom in 1 (electronic supplementary material, figure S2). Because of this, interatomic distances between the interacting carbon and nitrogen atoms are shortened. This unlocked the door for the concerted transition state 1-2TS, where nitrogen of aniline performs a nucleophilic attack to the carbon atom of the carbonyl group. New C-N bond is being formed, and in the same time, a proton is being transferred from nitrogen atom of [HDEA] to the carbonyl oxygen atom. Carbonyl p bond is being broken and the electron pair is used for the O-H bond formation. Compared to the non-catalysed C-N bond formation (electronic supplementary material, figure S1), [HDEA] support resulted in significant decrease in the energetic demands for the C-N bond formation, with a necessary barrier of 35.0 kJ mol 21 . Transition state 1-2TS leads towards the formation of intermediate 2.
Here, C-N bond is completely formed, a proton is transferred from nitrogen atom of [HDEA] to the oxygen of carbonyl group. Consequentially, DEA is formed, and hydrogen bonded to the newly formed OH group (N. . .H-O hydrogen bond). When compared to the non-catalysed reaction, C-N coupling is energetically much more demanding, with an activation barrier of 149.6 kJ mol 21 (electronic supplementary material, figure S2). Intermediate 2 has two potential pathways for further transformations leading towards iminium ion 6. The first one is with the participation of DEA, while the second one takes place without this molecule. In the reaction route where DEA is still present, 2 isomerizes to 17.1 kJ mol 21 [64].
To test the influence of solvent along the reaction course for the formation of iminium ion 6, a discrete molecule of ethanol was included (figure 2, blue line). This way, the influence of solvent on proton transfer and dehydration process was examined. Based on the obtained energies for these processes, one can see that ethanol did not significantly influence the deprotonation of the nitrogen atom in 3-4TSb. On the other hand, isomerization of intermediate 4b to 5b is somewhat unfavoured, as well as dehydration process in 5-6TSb. Namely, activation barrier is increased to 98.2 kJ mol 21 . Additionally, the influence of solvent was explored on the formation of SB, in non-catalysed reaction (electronic supplementary material, figure S1). Here, the participation of solvent molecule is significantly lowering barriers for the formation of SB, making this reaction more competitive to the formation of iminium ion 6.
rsos.royalsocietypublishing.org R. Soc. open sci. 5: 181232 Influence of DEA formed during the dehydration process in 5-6TS, and [ClAc] anion on the C-C bond formation was also explored (figure 2 and scheme 2). First, we will examine C-C coupling without any assistance. It is worth pointing out that we examined similar C-C coupling in our previous studies, where we used cyclohexanone instead of acetophenone [62,63]. To keep the comparability of the non-catalysed and catalysed routes, we will briefly present the non-catalysed process. The first step of this phase of the reaction is a nucleophilic attack of enol form of acetophenone (methylene carbon) to the carbon atom of C ¼ N bond of iminium ion 6. This assumption was confirmed by revealing transition state 6-7TS, with the necessary activation barrier of 37.2 kJ mol 21 , respectively. In both cases, C ¼ NH þ and enol C¼C p bonds are being broken, electron pair from C¼C bond is being used for the formation of new C-C bond, while the p electron pair from C ¼ NH þ is being placed on the nitrogen atom. This way, intermediate 7, a protonated final product of the reaction, is formed. Unfortunately, we were unable to locate corresponding transition states for the deprotonation of these intermediates, either by DEA or ClAc. Nevertheless, we located intermediates 8-DEA and 8-ClAc, which would be yielded from such process.
To examine the influence of DEA and [ClAc] in this phase of the reaction, we included these species in the interaction which would produce a new C-C bond. To be precise, we assumed that bases DEA, with 'free' electron pair situated on the nitrogen atom, and [ClAc] anion, can establish hydrogen bonds with hydrogen from OH group of enol in 6. This was confirmed by revealing the structures of 6 DEA and 6 [ClAc]. Next reaction step takes place with the simultaneous formation of C-C bond, and proton transfer from enol oxygen to the hydrogen bond acceptor atom of each base. In the same time, p C¼N and C¼C (enol) bonds are being broken, electron pair from p C¼N bond is being placed on the nitrogen atom of C¼N, while p C¼C electron pair of enol is used for new C-C bond formation with 6. O-H bond is being broken, and proton transferred to the base. Necessary barriers for 6-8TS-DEA and 6-8TS-ClAc amount to 2.9 and 19.4 kJ mol 21 . For the difference with the non-catalysed process of C-C bond formation, there is no formation of protonated final product 7. Instead, in 8-DEA and 8-ClAc proton is completely transferred to the corresponding acceptor atoms of each base (nitrogen of DEA, or oxygen of [ClAc]). This way, the final product of the reaction MB1 is liberated, and cationic part of the catalyst [HDEA] or chloroacetic acid is formed. In such manner, in the case of DEA, the catalyst is regenerated, while in case of [ClAc] precursor of the catalyst is formed.
One can note that the presence of [ClAc] slightly lowered C-C coupling activation barriers, while DEA significantly decreased these activation energies. Obtained results clearly point out that the ratedetermining step of the overall reaction is water molecule elimination in 5-6TS transition state. It is worth pointing out that, especially in the case of the reactions with anilines substituted with methyl and methoxy group, beside the formation of Mannich product, formation of the SB was detected. It might be rationalized in the way that in SB formation process solvent is taking some part in the reaction by lowering the barrier for proton transfer and therefore making this reaction more competitive to the first phase of Mannich reaction, i.e. formation of the iminium ion.

Conclusion
Three-component Mannich reaction of benzaldehyde, anilines and acetophenone, or 4iodoacetophenone, was performed in the presence of [HDEA] [ClAc]. All products were obtained in moderate to high yields. Five of them (MB1 -MB3, MB9 and MB10) are newly synthetized compounds, firstly reported in this paper. In all experimental UV-Vis spectra of MB1 -MB10, there are two major absorption bands around 205 and 260 nm, as well as a small band around 310 nm. Unlike numerous studies, here B3LYP method failed in the reproduction of the experimental spectra. On the other hand, M052X method reproduced all experimental spectra.
In vitro DPPH radical assay showed that among all investigated compounds, only compounds with anisidine moiety (MB8 and MB10) express moderate activity. Results obtained for thermodynamic parameters point out SPLET as preferred radical quenching mechanism. On the other hand, enthalpies for the reactions are strongly dependent on the nature of radical species. Nevertheless, considerably lower values for DH BDE than those obtained for DH PA point out HAT as preferred mechanism to SPLET in the presence of free radicals, and in methanol as solvent.
Based on the results obtained from the mechanistic study, one can conclude that hydrogen-bonded organocatalysis plays a very important role in both phases of the reaction. Rate-limiting step of the entire reaction is water molecule elimination in 5-6TS transition state. The importance of the hydrogen bonding in this reaction is reflected in the findings that the presence of [HDEA] rsos.royalsocietypublishing.org R. Soc. open sci. 5: 181232 significantly lowered the barrier for the formation of the new C-N bond. DEA, formed during this process, enhances proton transfer from nitrogen to oxygen. That way, barriers for the formation of iminium ion are significantly lowered. In addition, the final phase of the reaction, i.e. simultaneous formation of the new C-C bond and deprotonation of the oxygen atom, is also promoted by either DEA or [ClAc]. On the other hand, the effect of ethanol, as solvent and co-catalyst in proton transfer, is more pronounced in the formation of SB, than in processes for the formation of iminium ion 6.
Data accessibility. The datasets supporting this article have been uploaded as part of the electronic supplementary material.
Authors' contributions. V.P.P. managed the research, performed calculations and wrote the manuscript. D.S. and V.M.M. performed experimental part of the work, assisted with data analysis and revising of the manuscript. Z.D.P. assisted with data analysis, writing and revising of the manuscript.