Amines are very important targets in organic chemistry since they are key substrates in organic synthesis, as well as in pharmaceutical and agrochemical research fields. In addition, chiral amines have been extensively used as resolving agents [1,2,3,4], starting materials for the preparation of therapeutic drugs [5] and chiral auxiliaries in asymmetric synthesis [6,7,8]. Due to this, synthetic chemists have made great efforts in order to implement efficient procedures for the preparation of chiral amines [9,10] among which the reduction of iminic substrates plays a capital role. In recent years, the asymmetric transfer hydrogenation (ATH) has emerged as a powerful, convenient and wide-spread methodology to perform the reduction of C=X bonds (X = O or NR) [11] and it has become the second most used reduction methodology for this kind of functional group, just behind hydrogenations with molecular hydrogen. The ATH protocol presents several advantages: it is operationally simple, proceeds under mild reaction conditions, generally uses low catalyst loadings and avoids the handling of hazardous chemicals such as metallic hydrides or hydrogen gas. The ATH of prochiral ketones affords enantiomerically enriched secondary alcohols with excellent results [12,13,14,15,16,17,18,19]. Nevertheless, the asymmetric reduction of iminic compounds by transfer hydrogenation yielding chiral amines still remains underdeveloped and represents a challenging objective [20,21].

Due to the low electrophilic character of the iminic carbon, substrates in which the nitrogen atom is bonded to an activating group (aryl, alkoxy, amino, sulfinyl, sulfonyl or phosphinyl) are generally used in reduction methodologies leading to chiral amines. The phosphinyl group is very attractive since its removal from the reduction products can be conveniently achieved under mild acidic conditions, leading to the free primary amines [22,23]. N-phosphinyl imines have found a variety of synthetic applications in asymmetric processes [10,24,25]. For instance, the preparation of chiral amines can be achieved by stereoselective reduction of N-phosphinyl ketimines through different synthetic approaches such as hydrogenation [26,27,28], hydrosilylation [29,30,31,32,33], Meerwein-Schmidt-Ponndorf-Verley (MSPV) reaction [34] or the use of boron [35] and aluminium hydrides [36].

Although the transfer hydrogenation protocol has been already applied to the reduction of different iminic substrates bearing alkyl [18], benzyl [17,18], aryl [17,18], sulfonyl [13,18,37] or sulfinyl [38,39,40,41] groups attached to the nitrogen atom, there are only a few examples on the use of N-phosphinyl imines as substrates [18,42]. In the last years, we have been interested in the use of β-amino alcohols as ligands for ruthenium complexes used as catalysts for the ATH of imines, especially N-(tert-butylsulfinyl)imines [38,39,40,41]. Since, to the best of our knowledge, the use of β-amino alcohols as chiral ligands has not been reported so far for the ATH of N-(diphenylphosphinyl)imines, we became interested in trying to apply our methodology to those substrates. Thereby, herein we present our preliminary results on the reduction of phosphinyl imines in isopropyl alcohol with ruthenium complexes bearing chiral β-amino alcohols.

In recent years, one of our main research lines has focused on the asymmetric synthesis of chiral amines by enantioselective addition of dialkylzinc reagents to N-(diphenylphosphinyl)imines using β-amino alcohols as ligands [47,48,49]. Since those ligands have shown to give excellent results in the ruthenium-catalyzed asymmetric transfer hydrogenation of ketones [12,13,14,15,16,17,18,19] and N-sulfinylimines [38,39,40,41] in isopropyl alcohol, we decided to investigate if they could also be applied to the reduction of N-phosphinyl ketimines by the same methodology. One of the most noteworthy chiral β-amino alcohols in the ATH of both ketones and N-sulfinyl imines is cis-1-amino-2-indanol [50]. Due to this, we chose (1S,2R)-1-amino-2-indanol L1 as a test ligand for the ruthenium-catalyzed transfer hydrogenation of the N-phosphinyl imine derived from acetophenone 1a (Table 1). The ruthenium catalyst was prepared by refluxing a mixture of the ruthenium dimer [RuCl₂(p-cymene)]₂ and L1 in isopropyl alcohol.

Since a proportion Ru-dimer:aminoalcohol:base = 1:4:10 is common for the ATH of ketones, we chose that proportion as the starting point of our study, using 5 mol% of the Ru-dimer. Under those conditions, the Ru complex bearing L1 was able to reduce the imine 1a in isopropyl alcohol in the presence of potassium hydroxide, affording product 2a in a moderate yield and with 66% ee after stirring for 5 h at room temperature (Table 1, entry 1). This promising result encouraged us to try to optimize the reaction conditions. First, the amount of L1 was reduced keeping the loadings of Ru-dimer and base constant and we found that both yield and ee improved when the ratio Ru-dimer: L1 was 1:2 (compare entries 1–3 in Table 1). Hence, we decided to keep that ratio in the following tests. Next, the loading of base was evaluated and a decrease in the enantioselectivity, in comparison with the results of entry 2, was observed when the amount of potassium hydroxide was either increased to 75 mol% (Table 1, entry 4) or decreased to 25 mol% (Table 1, entry 5). Since the optimum ratio Ru-dimer: L1:base seemed to be 1:2:10, it was maintained in further experiments. A different metallic cation in the base did not cause any noticeable effect on the enantioselectivity: lithium and sodium hydroxides gave almost the same ee’s as potassium hydroxide, the yield being higher with the latter (compare entries 2, 6 and 7 in Table 1). We tried to reduce the amount of the Ru-dimer to 3 mol%, but this considerably slowed down the transfer hydrogenation process and resulted in a lower yield of 2a, although the enantioselectivity did not vary (compare entries 2 and 8 in Table 1).

The effect of the temperature was also studied. The ee decreased to 60% when the reaction was performed at 50 °C (Table 1, entry 9). A decrease of the temperature to 0 °C drastically lengthened the reaction time and, furthermore, the yield decreased to 48% (Table 1, entry 10). However, the enantioselection of the process slightly improved to 78% ee. In order to try to improve the yield, the reaction was repeated and an upward temperature gradient was applied: after stirring for 1 h at 0 °C, the reaction was allowed to gradually reach room temperature. We were delighted to see that, under these conditions, a 70% yield was obtained with no detriment to the enantioselectivity (Table 1, entry 11). A slight improvement of both yield and enantioselectivity were observed when the initial temperature was −20 °C instead of 0 °C (Table 1, entry 12). Since in our previous studies concerning the ATH of N-(tert-butylsulfinyl)imines [38,39,40,41] we found that potassium tert-butoxide was a more convenient base than potassium hydroxide, we decided to evaluate the former in the ATH of phosphinylimine 1a and we obtained a further improvement, giving 87% yield and 82% ee (Table 1, entry 13). A decrease of the initial temperature to −40 °C slightly reduced both the yield and the enantiomeric excess (Table 1, entry 14). After having performed this screening, we chose the conditions shown in entry 13 as the optimal ones.

The influence of the arene ligand in the ruthenium catalyst was also studied and we found that the use of [RuCl₂(benzene)]₂ instead of [RuCl₂(p-cymene)]₂ almost completely prevented any asymmetric induction, the ee of the product 2a being only 8%.

Once we had determined the optimum reaction conditions, we decided to test some other chiralβ-amino alcohols with the aim of evaluating the influence of the ligand structure on the success of the ATH process. Amino alcohols (1R,2S)-norephedrine L2, (1R,2S)-2-amino-1,2-diphenylethanol L3 and prolinols L4 and L5 were tested as potential ligands, and the results are summarized in Table 2. The ruthenium complex bearing norephedrine L2 was able to reduce imine 1a, though the results concerning both yield and enantioselectivity were rather poor (Table 2, entry 2). When the ATH reaction was carried out using ligand L3, the reduction product was obtained in only 39% yield and with a moderate ee. Therefore, ligands L2 and L3 have shown to be clearly inferior to L1 (compare entries 1 to 3 in Table 2). This drop in enantioselectivity could be due to the fact that (1S,2R)-1-amino-2-indanol L1 has a more rigid structure than ligands L2 or L3. Therefore, it seems that the rigidity of the β-amino alcohol seems to be very convenient for achieving good enantioselectivities in this ATH process. Finally, in agreement with our observations in the ATH of sulfinylimines [38,39,40,41], a ligand with a free primary amino group led to higher enantioselectivities than other ligands bearing secondary amino groups: when the ligands with a prolinol backbone L4 and L5 were tested, either no reduction product or only traces of it could be detected after working-up the reaction (Table 2, entries 4 and 5).

After having established that L1 was the ligand of choice, we tested two more imines as substrates. The asymmetric transfer hydrogenation of imines 1b and 1c, derived from propiophenone and 4-chloroacetophenone respectively, under the conditions of entry 13 in Table 1 gave the expected phosphinamides 2b and 2c in high yields and with ee’s of 82 and 80%, respectively (Table 2, entries 6 and 7).

In conclusion, we have developed a new efficient procedure for the reduction of prochiral N-(diphenylphosphinyl)ketimines through a ruthenium-catalyzed asymmetric transfer hydrogenation process in isopropyl alcohol using a β-amino alcohol as a chiral ligand. The rigid structure present in (1S,2R)-1-amino-2-indanol seems to be very convenient in order to achieve good enantiomeric excesses. The arene ligand on the ruthenium complex also has an influence on the stereochemical outcome of the reaction. Since the phosphinyl group can easily be removed under mild acidic conditions, this procedure represents an efficient method to synthesize optically enriched chiral primary amines. Further efforts to improve the enantioselectivity and to extend the substrate scope of this new ATH process are currently underway in our laboratory.