Introduction

Non-innocent ligands (NILs)1 have long attracted attention because of their unusual redox properties and their apparent, often deceptive, ability to stabilize metals in unusual oxidation states. The NILs have been employed not only for spectroscopic studies2,3 but also in different chemical transformations4,5. It may be postulated that due to the presence of easily accessible frontier orbitals in the NILs as compared to those of typical ligands, they facilitate electronic transfer through an inverted bonding process6. These ligands are now emerging as synthetically useful and attractive scaffolds with broad and exciting perspectives7. Nevertheless, fully exploring the potential of these NILs in group–transfer chemistry, energy storage and conversion, biological applications still remains in its early stage and in particular the influence of NILs in a catalytic reaction has remained in its infancy8,9. Though the area is dominated by coordination chemists10 and spectroscopists looking at influence of NIL in metals oxidation state, this concept is gradually entering into the area of catalysis8. In particular, in the field of organometallic chemistry, the NILs are emerging with ample opportunities in the realm of catalysis5.

In this context, recently we introduced phenalenyl (PLY) unit as a potential ligand backbone to study their influence on the catalytic activity in ring opening polymerization (ROP) reactions11. Phenalenyl is a well-known odd alternant hydrocarbon with three fused benzene rings12 with ability to form three redox species: cation, radical and anion (Figure 1a) using its readily avaialable nonbonding orbital (NBO)13. The concept of designing neutral free radical based conductors using this nonbonding orbital of phenalenyl was first proposed by Haddon14,15,16,17. Subsequently, this idea led to the discovery of the best neutral organic conductor at room temperature15. Recent articles by Morita, Takui and Hicks described the present status of phenalenyl based material chemistry emphasizing the radical electronic state of phenalenyl molecule12,18. On the other hand, we posed a question whether the presence of the readily available nonbonding orbital in the cationic state of phenalenyl unit19 (State 1, Figure 1a) can be utilized to design well-defined molecular catalysts which will influence the catalytic activity in a homogeneous organic transformation. It was postulated that on coordination to a metal ion, the phenalenyl unit will generate the closed shell cationic state and the empty NBO will function as electron acceptor. Recently, we have shown that the cationic state of phenalenyl unit when created with zinc metal coordination can be used as non-innocent building block for construction of spin memory device which functions based on the ability of electron acceptance of the phenalenyl unit20.

Figure 1
figure 1

(a) The three redox species of phenalenyl moiety: cation when the NBO is empty, radical when the NBO is singly occupied and anion when the NBO is completely filled19; (b) Organoaluminum complexes (1–5) based on phenalenyl ligand.

Full size image

In this study, we have chosen a homogeneous catalytic process, namely guanylation or hydroamination of carbodiimides with aromatic amines. Earlier study suggests that the hydroamination of carbodiimide proceeds via the methane liberation and formation of the metal-amide complex. The reaction proceeds via a transition state formed by interaction of amine and catalyst, presumably involving the transfer of electron density from amine to catalyst21. We became interested to study this particular homogeneous process to unravel the exact role of non-innocent behavior of phenalenyl ligand in the catalytic outcome; since the influence of the NIL in catalytic reaction is not very well understood8. The catalytic addition of amine to carbodiimides (also known as guanylation or hydroamination) offers an efficient and atom-economical route to produce substituted guanidines, which is of great interest to academic and industrial researchers22. As a part of our ongoing interest in the development of catalytic systems for hydroamination reaction23,24,25, herein we report the catalytic activity of a series of phenalenyl ligand based organoaluminum complexes towards hydroamination of carbodiimides. In the present study, five organoaluminum complexes, namely [(O,O–PLY)]AlMe2 (1), [N(Me),O–PLY]AlMe2 (2), [N(Cy),O–PLY]AlMe2 (3), [N(Me),N(Me)–PLY]AlMe2 (4) and [N(Cy),N(Cy)–PLY]AlMe2 (5) (Figure 1b) were used for catalytic hydroamination of carbodiimides to primary aromatic amines.

Results

Syntheses and characterization

This study utilized organoaluminum complexes of phenalenyl ligands based on O,O– (HO,O–PLY), N,O– [HN(Me),O–PLY and HN(Cy),O–PLY)] and N,N– [HN(Me),N(Me)–PLY and HN(Cy),N(Cy)–PLY] donor combinations. The organoaluminum complexes 1, 2 and 4 were synthesized according to the literature procedure11. Synthesis of the new organoaluminum complexes, 3 and 5 were accomplished by reacting [HN(Cy),O–PLY] and [HN(Cy),N(Cy)–PLY] ligands, respectively with AlMe3 under the evolution of methane (Figure 2). A solution of [HN(Cy),O–PLY] or [HN(Cy),N(Cy)–PLY] ligand in toluene was added drop-by-drop to the solution of AlMe3 in toluene in a 1:1.2 stoichiometric ratio at −78°C and finally heated at 90°C or at 110°C for 6 h or 24 h to yield the complex 3 or 5 (Figure 2). Absence of any characteristic N–H resonance at δ 12.5 or 14.0 ppm in C6D6 as revealed by the 1H NMR spectrum of the reaction mixture establishes almost quantitative conversion of the reactants into products in case of 3 and 5. Both complexes 3 and 5 were characterized by 1H NMR, 13C NMR spectroscopy, elemental analysis and single crystal X-ray diffraction studies. 1H NMR spectra of 3 and 5 in C6D6 exhibit a singlet at δ 0.01 and −0.10 ppm, respectively attributing to the proton resonance arising from the methyl group bound to the aluminum ion. The 13C NMR spectra of 3 and 5 in C6D6 exhibit a singlet at δ −4.5 and −3.5 ppm, respectively assigned to the 13C resonance arising from the methyl group bound to the aluminum ion. The structures of complexes 3 and 5 were determined unambiguously by single crystal X-ray diffraction technique (Figure 2). Suitable crystals of 3 and 5 were obtained from cold toluene solution at −20°C and 0°C, respectively. Complexes 3 and 5 were crystallized in the monoclinic and triclinic space group C2/c and P ī, respectively with one molecule in the asymmetric unit in both the cases. The X-ray structures of 3 and 5 reveal a distorted tetrahedral geometry around the aluminium ion. The observed Al–C bond distances in 3 are Al(1)–C(20), 1.9590 (19) Å and Al(1)–C(21) 1.9659(18) Å. Almost identical Al–C bond lengths are also observed in 5 [Al(1)–C(22), 1.981(3) Å; Al(1)–C(23), 1.976 (3) Å]. The Al–N (1.9429 (15) Å) bond distance found in 3 is longer than that observed in 2 (1.91 Å), while the Al–O bond distance (1.7820(13) Å) found in 3 is shorter than that observed in 2 (1.79 Å)11. The Al–N bond distances in 5 [Al(1)–N(1), 1.900(2) Å; Al(1)–N(2), 1.899(2) Å] are slightly shorter than that of the Al–N bond distances found in complex [(2,6–iPr2C6H3)NC(NiPr2)N(2,6–iPr2C6H3)]AlMe2 (av. 1.9263 Å)26.

Figure 2
figure 2

Syntheses and molecular structures of the organoaluminum complexes 3 and 5 bearing phenalenyl ligands as determined by single crystal X-ray diffraction studies.Thermal ellipsoids are drawn with 50% probability level. Selected bond distances (Å) and angles (°) of (a) 3: Al(1)–N(1) 1.9429 (15), Al(1)–O(1) 1.7820 (13), Al(1)–C(20) 1.9590 (19), Al(1)–C(21) 1.9659 (18), N(1)–C(13) 1.335 (2), O(1)–C(3) 1.3071 (19); N(1)–Al(1)–O(1) 95.19(6), N(1)–Al(1)–C(20) 112.02(7), O(1)–Al(1)–C(21) 107.30(7), C(20)–Al(1)–C(21)115.19(8); (b) 5: Al(1)–N(1), 1.900(2)), Al(1)–N(2) 1.899(2), Al(1)–C(22), 1.981(3), AI(1)–C(23), 1.976(3), N(2)–AI(1) 1.896(9), N(1)–AI(1) 1.918(9), N(1)–Al(1)–N(2) 97.5(6), N(1)–Al(1)–C(27) 109.7(9), N(2)–Al(1)–C(26) 107.9(9), C(27)–Al(1)–C(26) 115.4(9).

Full size image

Catalytic activity and the origin of non-innocence behaviour

Earlier, Bergman and co-workers26 as well as Zhang and Xi21 and co-workers have demonstrated that organoaluminum complexes can be used as a catalyst for the hydroamination/guanylation reaction of carbodiimides to produce substituted guanidines. These serve as building blocks for many biologically relevant compounds27 and also used as suitable ligand system for stabilization of a wide range of metals28. Over the last decade, several catalytic systems have been developed based on main group metals29,30,31, rare earth metals and also by transition metals32,33,34,35,36,37.

In the present study, the catalytic activity of the complexes 1–5 for hydroamination of carbodiimides with aromatic amines has been tested. The reaction of p-toluidine (CH3–C6H4–NH2, 6) with N,N′-dicyclohexylcarbodiimide (CyN = C = NCy) was first examined under various conditions (Table 1). As a control experiment, CyN = C = NCy was mixed with p-toluidine in C6D6 at 25°C, but no reaction was observed even after 30 h (Table 1, entry 1). In contrast, addition of 5 mol% of 1 gave the desired N,N′,N″-trisubstituted guanidine (7) within 30 h at 25°C (Table 1, entry 2). Subsequently, the amount of catalyst loading was varied to see the most effective catalytic condition. It was found that 5 mol% of catalyst loading works at its best (Table 1, entries 2–4). Change of polarity of solvents showed influence on catalytic activity of 1 (Table 1, entries 2, 5 and 6). Increase of temperature from 25°C to 60°C showed significant enhancement in the reaction rate (Table 1, entries 2 and 7). Furthermore, the catalytic activity of 1 in different solvents (C6D6, THF-d8 and CDCl3) was tested by kinetic measurement through 1H NMR spectroscopy to check the solvent effect on the catalytic reaction and the study reveals that C6D6 is the most effective solvent for the catalytic reaction (see Figure S2 of Supporting Information). Other organoaluminum complexes (2–5) were also found effective for this catalytic reaction under different reaction conditions (Table 1, entries 8–24). Complexes 2 and 3 displayed comparable catalytic activity only at higher temperature (80°C) leading to the formation of 7 in 97% and 95% yield, respectively (Table 1, entries 9 and 13). Similarly, complexes 4 and 5 exhibited the catalytic activity at 110°C producing 7 in 95% and 96% yield, respectively (Table 1, entries 18 and 22). This result clearly points out a distinct catalytic activity ordering, displaying the activity order 1 > 2 ~ 3 > 4 ~ 5 when the catalytic reactions were performed under identical condition.

Table 1 Standardization of reaction condition using p-toluidine and N, N′-dicyclohexyl carbodiimide as substratesa
Full size table

In order to check the versatility of this catalytic process, we tested the catalytic activity of the complexes 1–5 with a series of aromatic amines and reacted with different carbodiimides (8–17) under the standardized catalytic condition (see Supplementary Information for details). The catalytic results reveal that complexes 1–5 are active catalysts for a wide range of substituted anilines having electron-donating as well as electron-withdrawing substituents (see Supplementary Information, Tables S1–S3). Complex 1 exhibited catalytic activity at 25°C and 60°C, while 2 and 3 exhibited comparable activity only at 80°C. Complexes 4 and 5 displayed the catalytic activity only at 110°C. Moreover, the catalytic table suggests that carbodiimides having different substituents are tolerable under the catalytic condition (see Supplementary Information, Tables S1–S3). The catalytic activity of these organoaluminum complexes were found to be moderate as compared to the earlier reported catalytic systems22,26,27,28,29,30. The detailed catalytic study carried out with complexes 1–5 clearly reveals that the catalytic activity of complex 1 (O,O-system) is highest and that of 4 and 5 (N,N-systems) are lowest, while the catalytic activity of complexes 2 and 3 (N, O-systems) falls in between the catalytic activity of O,O-donor based catalyst and N,N-donor based catalysts. In order to understand this difference in the catalytic activity between the organoaluminum complexes, we carried out a DFT calculation. Organoaluminum complexes of phenalenyl based ligands11 as well as spirobiphenalenyl cationic boron systems38 can be represented in different canonical forms in which most of the positive charge remains on the phenalenyl unit (see Figure S3 of the Supplementary Information). The nonbonding orbital of the cationic phenalenyl moiety remains empty in these dimethylaluminum complexes (1–5) which largely contributes to the LUMO of the molecule (also see Figure 1a). A DFT calculation at the 6–311 ++ G (d,p) level was carried out on organoaluminum complexes (1, 3 and 5) and the result reveals that in these complexes (1, 3 and 5), the LUMO predominantly resides over the phenalenyl part of the molecule (Figure 3a). Interestingly, the calculation reveals that the relative energy of the LUMO is largely determined by the nature of the phenalenyl ligands, in other words the phenalenyl ligands can tune the LUMO energy levels in these complexes. The energy of the LUMO (Figure 3b) gradually increases from 1 (−72.6 kcalmol−1) to 3 (−63.3 kcalmol−1) to 5 (−55.6 kcalmol−1). This result points out that the existence of the lowest energy LUMO in 1 might lead to the best electron acceptance property of 1 among 1, 3 and 5 (Figure 3b). Additionally, the calculation reveals that the energy differences between LUMO and LUMO + 1 in these organoaluminum complexes are quite substantial. The energy differences between LUMO and LUMO + 1 are calculated as 45.1 kcalmol−1, 42.0 kcalmol−1 and 39.5 kcalmol−1 in 1, 3 and 5, respectively. Thus, the DFT calculation supports the presence of energetically accessible empty molecular orbital for electron acceptance in these phenalenyl-based organoaluminum complexes. In order to understand the electron acceptance property of these complexes, we carried out cyclic voltammetric study and these experiments indeed revealed the presence of two one-electron acceptance processes in case of 5 (Figure 4a), which can be explained by considering the generation of an anionic radical and a dianionic species (Figure 4b). The E1/2 values (E11/2 = −1.49 V and E21/2 = −1.80 V) indicate that electron acceptance from the first to the second one becomes progressively difficult as a result of increasing Coulomb repulsion in the system. Early EPR studies on spirobiphenalenyl boron salt had shown38 that the first reduction leads to the formation of an EPR active radical species and the second reduction results in the EPR silent anionic species supporting our result that the LUMO of a phenalenyl system can accommodate two electrons. Comparing the previous electrochemical data11, it reveals that the reduction of 1 is much more facile and that of 4 or 5 is the most difficult among the series (the first half-wave reduction potential (E11/2) values for 1; E11/2 = −0.87 V, 2; E11/2 = −1.14 V, 4; E11/2 = −1.45 V and 5; E11/2 = −1.49 V, respectively). This result points out that the LUMO of 1 is most accessible for electron acceptance supporting our DFT calculation (see Figure 3b). However, this catalytic activity ordering may also have direct contribution from the Lewis accepting ability of the aluminum center as the attachment of more electronegative O,O-phenalenyl ligand (1) can make the aluminum center more Lewis acidic as compared to those with N,O-phenalenyl (2 and 3) and N,N-phenalenyl ligands (4 and 5). The charge calculation on aluminum center supports this fact as well which shows that the charge on aluminum is highest in catalyst 1 (1.80 a.u.), as compared to that in 3 (1.77 a.u.) and 5 (1.75 a.u.).

Figure 3
figure 3

(a) The computed LUMO of the complexes 1, 3 and 5; (b) Molecular orbital energy profile diagram of the organoaluminum complexes (1, 3 and 5) showing that the nature of phenalenyl ligand can tune the energy of the LUMO in these complexes.

Full size image
Figure 4
figure 4

(a) Cyclic voltammetry of 5 in acetonitrile, referenced to Ag/AgCl via internal ferrocene revealing two successive one electron reduction processes; (b) Successive electron acceptance by 5 generates an anionic radical and a di-anionic species.

Full size image

Mechanistic study

The catalytic reaction was subjected to further study to gain more insight on the mechanistic process of the hydroamination reaction of carbodiimides and to correlate the catalytic results with the non-innocence behaviour of phenalenyl ligands. We adopted a combined approach using the kinetic studies, in situ NMR monitoring experiments, UV-vis absorption spectroscopy, Hammet study as well as DFT calculation. The study started with the detailed kinetic analysis to gather further insight into the hydroamination process of carbodiimide catalyzed by 1 using p-toluidine and N,N′-dicyclohexyl carbodiimide in C6D6. Kinetic studies for the formation of 7 were carried out with catalyst 1 and equi-molar amount of p-toluidine and N,N′-dicyclohexylcarbodiimide via 1H NMR spectroscopy. The evolution of the specific resonances of the N,N′,N″-trisubstituted guanidine (7) was monitored by 1H NMR spectroscopy relative to an internal standard (hexamethyl benzene) over the course of the first 7 h (see Figure S4 of Supporting Information for details). We first determined the order of the reaction with respect to amine concentration by keeping the concentration of other components virtually unaltered. Initially, we began our study with 21.71 mM of catalyst (1) concentration and the carbodiimide to amine molar ratio was maintained at more than 10:1. A plot of concentration of amine against time followed an exponential decay of amine concentration (Figure 5a) which is consistent with the first-order kinetics. The first order rate of the reaction with respect to the amine concentration was further confirmed from the plot of ln(C/C0) versus time, providing a linear plot with negative slope (Figure 5b)39. Next, we determined the order of the reaction with respect to the carbodiimide concentration by gradually varying concentration of carbodiimide. This study provided evidence of a first-order dependence of the reaction rate on carbodiimide concentration as well. The rate of reaction increases in a linear fashion with increasing concentration of carbodiimide (Figure 5c). The similar observation was noted by Marks and co-workers as well as by Hill and co-workers in case of intermolecular hydroamination reaction39. The individual rate constant values at different concentrations of carbodiimide were obtained after plotting ln(C/C0) versus time. These plots provide a straight line with negative slope and from this first order plot, the individual rate constant values were obtained. The dependence of the reaction rate with respect to catalyst concentration was established by gradually changing the concentration of catalyst (from 7.5 mM to 15.1 mM) and holding the amine (151.6 mM) and carbodiimide (151.6 mM) concentrations fixed. A plot of reaction rate versus catalyst concentration reveals a linear increase of the reaction rate with increase of the catalyst concentration (Figure 5d). This observation suggests a first order dependence of the rate of the reaction with respect to the catalyst concentration. The first order rate of the reaction with respect to the catalyst concentration was further confirmed from van't Hoff plot40. A plot of lnkobs versus ln[catalyst] provided a linear graph (Figure 5e) and the value of the slope was determined to be 1.42 (slope = order of the reaction)40. Related van't Hoff plot for intramolecular hydroalkoxylation mediated by lanthanide catalyst resulted in a slope of 1.43 as reported earlier40, which compares well with the present observation.

Figure 5
figure 5

Kinetic studies on the hydroamination of the carbodiimide for the formation of the guanidine (7) monitored by 1H NMR spectroscopy in C6D6 at 25°C.

a) plot showing the change of amine concentration with time; (b) plot of ln(C/C0) versus time; (c) plot of kobs and concentration of carbodiimide for the formation of 7; (d) plot of reaction rate versus concentration of catalyst and (e) van't Hoff plot for the formation of 7.

Full size image

Thus, from the present study, the overall rate law for the intermolecular hydroamination reaction of carbodiimides and amine catalyzed by 1 can be summarized as shown in eq. 1.

Consequently, the kinetic study of the catalytic reaction suggests that the reaction rate depends on concentration of amine, carbodiimide and catalyst. Subsequently, we carried out an in situ NMR study to understand the mechanistic aspect of the catalytic reaction in more details.

The in situ NMR study was recorded after mixing the organoaluminum complex 1 and p-toluidine in a 1:1 stoichiometric ratio and it revealed no detectable resonance of the CH4 formation even after 5 h at 25 °C (Figure S7, Supporting Information). Earlier work by Zhang, Xi and co-workers have postulated the formation of aluminium-amide bond by activation of amine with concomitant elimination of methane21. Nevertheless, a close look on the 1H NMR spectrum of 1:1 mixture of 1 and p-toluidine did indicate a shift in the aluminium bound methyl resonance as compared to that observed for pure catalyst. Furthermore, we observed that this shift is not specific to only 1:1 mixture of 1 and p-toluidine, rather it is common to the 1:1 stoichiometric reaction mixture of catalyst 2 or 4 and p-toluidine. Compound 1 exhibits the resonance of Al-Me group at δ 0.02 ppm, whereas in the 1:1 mixture of 1 and p-toluidine it shifts to more shielded region at δ −0.003 ppm (Figure 6a). However, this shift is smaller in case of the 1:1 mixture of 2 or 4 and p-toluidine (Figures 6b and 6c). This NMR shift indicates that the 1:1 mixture of catalyst and p-toluidine forms a donor acceptor adduct which is in equilibrium with the catalyst and p-toluidine. In this donor- acceptor interaction, the catalyst acts as an acceptor. The formation of such donor-acceptor aluminium amine adduct has been reported earlier by Wehmschulte and co-workers and others41. The larger NMR shift in case of 1:1 mixture of 1 and p-toluidine indicates stronger binding with 1 as compared to 2 or 4. To further investigate this donor-acceptor interaction, we estimated association constant of this equilibrium process between the catalyst and p-toluidine by UV-vis absorption spectroscopic measurements of the reaction mixture of these organoaluminum complexes (1, 2 and 4) with p-toluidine (also see Supporting Information). Complex 1 shows the absorption maxima at λ 356, 426 and 451 nm, complex 2 shows the absorption maxima at λ 349, 445 and 471 nm, whilst 4 shows absorption maxima at λ 368, 501 and 541 nm. Next, the effect of the p-toluidine to the solution of the three complexes (1, 2 and 4) was checked in the UV-vis absorption spectra (Figure 6d–f) after a careful stepwise addition p-toluidine. Figure 6d suggests that with the increase of p-toluidine concentration, the absorptions of 1 at λ 356, 426 and 451 nm increases. Similar observation was also observed in case of 2 as well as 4 (Figure 6e and Figure 6f), although the change in absorption is less as compared to that observed with 1. The association constant between the organoaluminum complexes and the p-toluidine was calculated from the UV-vis absorption spectroscopy using Benesi–Hildebrand equation (Figure 6g–i)42,43. A plot of 1/ΔA (A = absorbance) versus 1/[amine] gives a straight line with positive slope in all the three cases (Figure 6g–i). The values of the association constants calculated from these plots for 1, 2 and 4 are 7.02 × 104 M−1 (λ = 426 nm), 2.0 × 104 M−1 (λ = 445 nm) and 2.8 × 103 M−1 (λ = 541 nm), respectively (also see Supporting Information, Figure S9). These values suggest that the binding between complex 1 and p-toluidine is stronger than that with complexes 2 and 4. The similar observation was also indicated from the 1H NMR shift of Al-Me resonance of 1:1 catalyst-amine mixture. The binding affinities between these organoaluminum complexes and the amines may be attributed due to the lowest HOMO-LUMO energy gap between p-toluidine and 1 as compared to that calculated for 2 and 4 from the DFT study (Figure 7). Comparatively shorter HOMO-LUMO gap in case of 1 results in stronger donor-acceptor interaction. Furthermore, the addition of one equivalent carbodiimide into the 1:1 reaction mixture of 1 and p-toluidine triggered the formation of CH4 at 25 °C as revealed by its characteristic resonance at δ 0.15 ppm44 in 1H NMR (Figure S8, see the Supporting Information). The resonance at δ 0.15 ppm becomes more prominent with the gradual progress of the reaction. This result points out that the presence of carbodiimide plays a crucial role in the methane evolution process and subsequent product formation. The formation of CH4 on addition of one equivalent carbodiimide to the 1:1 stoichiometric reaction mixture of 1 and p-toluidine may be explained by considering that the carbodiimide shifts the initial equilibrium towards right and promotes the formation of aluminium-amide bond via methane evolution and amine activation process. Furthermore, in order to shed light on the cleavage of the amine N–H bond, we assessed the kinetic isotope effect. Determination of the kinetic isotope effect (KIE) has proven to be a very useful mechanistic tool, which has already been applied in organometallic systems45. The H/D kinetic isotope effect (KIE) measurement was performed using 6 and 6-d2 in the presence of catalyst 1 in C6D6 under the identical reaction condition. The first order plots of the hydroamination of 6 and 6-d2 under the reaction condition indicate kobs = 0.07131 h−1 and kobs = 0.02246 h−1, respectively, which translates into a primary KIE of 3.17 (Figure 8). The KIE effect of 3.17 suggests that H/D derived from the –NH2/–ND2 moiety of 6/6-d2 is involved in the slowest step (rate determining step) of the hydroamination of carbodiimides.

Figure 6
figure 6

The shift of Al-Me resonances in 1H NMR spectra recorded in C6D6 (all 1H NMR spectra are shown in identical scale).

(a) 1:1 mixture of aluminium catalyst 1 and p-toluidine, (b) 1:1 mixture of aluminium catalyst 2 and p-toluidine, (c) 1:1 mixture of aluminium catalyst 4 and p-toluidine.Spectral key: C = pure catalyst 1 or 2 or 4 and M = 1:1 mixture of aluminium catalyst and p-toluidine. Electronic absorption spectra of (d) organoaluminum complex 1 with different concentration of p-toluidine in dichloromethane (e) organoaluminum complex 2 with different concentration of p-toluidine in dichloromethane; (f) organoaluminum complex 4 with different concentration of p-toluidine in dichloromethane. Benesi–Hildebrand plots (Plot of 1/ΔA (A = absorbance) vs.1/[amine]) for determination of binding constant of (g) p-toluidine with the complex 1 (λ = 426 nm) in dichloromethane; (h) p-toluidine with the complex 2 (λ = 445 nm) in dichloromethane and (i) p-toluidine with the complex 4 (λ = 501 nm) in dichloromethane.

Full size image
Figure 7
figure 7

Molecular orbital energy profile diagram indicating that the HOMO–LUMO energy gap between aromatic amine (6) and catalyst is the lowest in case of 1.

Full size image
Figure 8
figure 8

H/D KIE for the formation of 7 and 7-d2 using catalyst 1 at 25°C.

Full size image

Discussion

Considering all these facts, the following mechanism was proposed for the organoaluminum complex catalyzed hydroamination of carbodiimide and primary aromatic amine. The dimethylaluminum complex 1 acts as a precatalyst and on addition of amine, it generates 18 which remains in an equilibrium with reactants. The equilibrium shifts to the forward direction after the addition of carbodiimide to the reaction mixture. This step is irreversible and occurs with elimination of methane gas leading to the formation of aluminium-amide bond by amine activation mechanism. The coordination of the nitrogen atom of the carbodiimide to aluminium ion facilitates the elimination of CH4 and formation of intermediate 19. This step is the rate determining step of the reaction as evidenced from the kinetic isotope effect study (vide supra). Additionally, the involvement of amine, carbodiimide and catalyst in the rate limiting step as proposed by this mechanism is also supported by our rate equation revealing the dependence of amine, carbodiimide and catalyst concentration (see eq. 1). Subsequently, the intermediate 19 rearranges via the insertion of carbodiimide into the Al-N bond and generates 20. Earlier work by Zhang, Xi and co-workers have demonstrated such insertion of carbodiimide into the Al-N bond21. We eliminated another alternative ligand dissociation mechanistic pathway reported by Bergman and co-workers26 since we did not observe any ligand dissociation during the catalytic cycle by NMR spectroscopy. Protonolysis of 20 by another molecule of primary amine and carbodiimide can regenerate 19 and can release the guanidine 21 (Figure 9) via path a. Also, there are two other pathways possible for aluminum catalyzed guanylation reaction as reported by Zhang, Xi and co-workers which cannot be ruled out. For example, intermediate 20 can rearrange into the guanidinate species 22 via path b through the intramolecular 1,3-hydrogen shift to form intermediate 22. Protonolysis of 22 by another molecule of primary amine and carbodiimide can regenerate 19 and release the guanidine 21. Moreover, intermediate 19 could also be rearranged to 22 directly to give the final desired product 21, after protonolysis with primary amine and carbodiimide.

Figure 9
figure 9

Proposed mechanism for addition of primary aromatic amines to carbodiimides.

Full size image

Furthermore, we have taken a closer look of the rate limiting methane elimination process with the help of DFT calculation which indicates that charge transfer takes place from the substrate to the phenalenyl part of the catalyst (see Figure S12, Supporting Information). Additionally, the Hammet study carried out to understand the effect of electron transfer. A Hammett plot of the resulting data provided a negative ρ value of −2.04 (Figure S13, Supporting Information). The negative ρ value supports that electron density is gained over the formation of transition state, a conclusion achieved by our DFT calculation on the transition state.

This study establishes that the cationic state of phenalenyl unit is not only useful for construction of the spin memory device by external spin injection20 but also they can act as organic Lewis acceptor unit to influence the catalytic outcome of a homogeneous reaction. The non-innocent behaviour of a ligand in group-transfer chemistry, energy storage and conversion, biological applications is one of the most discussed topics in the contemporary literature. However, the influence of non-innocent behaviour of a ligand in catalytic processes has not been studied with full potential until now and this study is a step towards this direction. This study establishes that the cationic state of phenalenyl unit posing an energetically accessible LUMO can act as a Lewis acceptor organic unit which can influence the outcome of a catalytic reaction. The DFT calculation establishes that the energy of the LUMO of catalyst is mainly controlled by the phenalenyl ligand which in turn determines the activity of the catalyst for the catalytic process.