Preparation of Enantiomerically Enriched P -Stereogenic Dialkyl-Arylphosphine Oxides via Coordination Mediated Optical Resolution

: Optical resolution of several dialkyl-arylphosphine oxides was elaborated using the Ca 2+ salt of ( − )- O , O ’-dibenzoyl-(2 R ,3 R )-tartaric acid as the resolving agent. The conditions of crystallization and purification of the enantiomerically enriched phosphine oxides were optimized. Ethyl-phenyl-propylphosphine oxide and butyl-methyl-phenylphosphine oxide were prepared with an enantiomeric excess higher than 93%, whereas, three other dialkyl-arylphosphine oxides were obtained with an enantiomeric excess of 37–85%. It was also found that the sterically demanding alkyl chains hinder the formation of stable diastereomeric complexes, which consequently led to less efficient resolution procedures.


Introduction
Chiral compounds form an important class in organophosphorus chemistry. These species may contain an asymmetric element either on the phosphorus atom or at other parts of the molecular scaffold [1][2][3]. Chiral organophosphourus compounds were traditionally used as ligands in transition metal-catalyzed transformations [4,5]. In the past two decades, application of phosphines [6][7][8], phosphine oxides [9,10] or phosphonium salts [11,12] as enantioselective organocatalysts became an emerging area. Many enantioselective (organo)catalytic transformations take place in the close proximity of the phosphorus atom, which signifies the application of the corresponding P-stereogenic species [13].
In this paper, the optical resolution of a series of dialkyl-phenylphosphine oxides was studied. For this compound class, only one resolution method applying BINOL was reported. Thus, our aim was to investigate, whether the significantly cheaper metal salts of tartaric acid derivatives can be used for the optical resolution of the selected acyclic phosphine oxides, as seen in Figure 1.

General
The details of chemicals, instruments used in this study, a few synthetic procedures, as well as the characterization of optically active P-oxides (1-7) can be found in the Supplementary Material.

Resolution of Ethyl-Phenyl-Propylphosphine Oxide (3) with in Situ Prepared Resolving Agents (Representative Procedure)
In total, 0.096 g (0.26 mmol) of DBTA·H2O and 7.1 mg (0.13 mmol) of CaO were refluxed in 0.31 mL of ethanol and 31 µL of water until it became clear. To this solution of in situ formed Ca(H-DBTA)2 [(R,R)-8], 0.050 g (0.26 mmol) of racemic ethyl-phenyl-propylphosphine oxide (3) in 0.31 mL of hot ethyl acetate was added. Colorless crystalline diastereomeric complex of (S)-3·Ca(H-DBTA)2 appeared after cooling the mixture to 25 °C. After standing at 25 °C for 24 h, the crystals were separated by filtration, washed with a mixture of 0.10 mL of ethanol and 0.10 mL of ethyl acetate, to give 0.075 g (62%) of (S)-3·Ca(H-DBTA)2 with a de of 69% ( When different solvent mixtures were used for resolution, the resolving agent [(R,R)-8] was prepared in situ in the corresponding alcohol and water, and phosphine oxide 3 was added in the other component of the solvent mixture. When only ethyl acetate was the solvent, diastereomeric complex (S)-3·Ca(H-DBTA)2 was prepared, as described above. Then the solvent was evaporated under reduced pressure, and diastereomer (S)-3·Ca(H-DBTA)2 was re-dissolved in boiling ethyl acetate. Crystallization and isolation of the diastereomer were carried out according to the representative procedure described above (Table 2, Entry 9).

Optimization of Optical Resolution of Ethyl-Phenyl-Propylphosphine Oxide (3)
For optimizing the conditions of the optical resolution, ethyl-phenyl-propylphosphine oxide (3) was selected as model racemate, as it has an average size within the dialkyl-arylphosphine oxides (1-7) investigated in this study.
As coordination of the chiral tartaric acid scaffold and phosphine oxide to the cation is one of the most important interactions in the diastereomers formed, (−)-O,O′-dibenzoyl-(2R,3R)-tartaric acid was converted to its calcium-, magnesium-, copper-, nickel-or cobalt-salts to test which derivative is the most suitable for enantiomeric separation of ethyl-phenyl-propylphosphine oxide (3). Interestingly, only the acidic Ca 2+ -salt of (−)-O,O′-dibenzoyl-(2R,3R)-tartaric acid [(R,R)-8] led to the formation of crystalline diastereomers, and to partial enantiomeric separation of phosphine oxide 3 (ee: 44%, yield 83%). Using other metal salts, no crystals were formed, indicating that coordination of either tartaric acid or phosphine oxide (3) is low towards magnesium-, copper-, nickel-or cobalt (See Supplementary Material for details).
With the most suitable metal salt in hand, we started our optimization studies. The corresponding resolving agent [(R,R)-8 -(R,R)-11] can either be prepared prior to reaction, or it can be formed in situ by reacting the given tartaric acid derivative with CaO in a mixture of ethanol and water. Besides (−)-O,O′-dibenzoyl-(2R,3R)-tartaric acid, (−)-O,O′-p-toluoyl-(2R,3R)-tartaric acid can also be used as chiral moiety. Moreover, Ca 2+ -salt of a given tartaric acid derivative can also be prepared in an acidic or neutral form [(R,R)-8 -(R,R)-11], in which the ratio of Ca 2+ and the given tartaric acid derivative is different. In our first few optimization experiments, we intended to select the most efficient resolution procedure, as well as the best tartaric acid derivative and its Ca 2+ -salt form [(R,R)-8 -(R,R)-11] (Scheme 1). In the optimization experiments, results obtained after the first crystallization were evaluated and compared, and the diastereomeric complexes were not purified further. Composition of the given diastereomer was determined by 1 H NMR, and enantiomeric excess of phosphine oxide 3 was measured by HPLC using a chiral stationary phase. Results are summarized in Table 1.  Results indicated, that higher enantiomeric excess and resolving capability values (i.e., higher efficiency) could be achieved when ethyl-phenyl-propylphosphine oxide (3) Table 1, Entry 2). Changing the chiral scaffold from dibenzoyl-tartaric acid to di-p-toluoyl-tartaric acid, or using neutral salt instead of acidic salts did not lead to improved results (Compare Table 1, Entry 2 with Entries 3-5). In general, better enantiomeric separation of phosphine oxide 3 was obtained with acidic salts [(R,R)-8 or (R,R)-9] than with neutral salts [(R,R)-10 or (R,R)-11] (Compare Table 1, Entries 2 and 3 or Entries 4 and 5). Resolving agents Ca(H-DBTA)2 and Ca(H-DPTTA)2 [(R,R)-8 or (R,R)-9] contain Ca 2+ and the corresponding tartaric acid derivative in a 1:2 ratio. Our previous studies revealed [57,58], that this stoichiometric ratio allows the formation of polymer-like catena structures, that stabilize the crystal structure of the diastereomer, and leads to improved enantiomeric separation of the given phosphine oxide coordinating to the Ca 2+ ion.
The experimental results summarized in Table 1  The diastereomeric complex that precipitated from the solution upon cooling was crystallized for 24 h, and then it was separated from the mother liquor by filtration (Scheme 2). Enantiomeric mixture of phosphine oxide 3 could be liberated from the diastereomer by extraction with aqueous NH4OH and dichloromethane. Results are summarized in Table 2.  Results indicated that ethanol is the most suitable alcohol for the resolution investigated, as the use of methanol or 2-propanol led to lower yields or ee values of phosphine oxide 3 (Compare Table  2, Entry 1 with Entries 2 and 3). Considering the other component of the solvent mixture, the best results were obtained with ethyl acetate (Table 2, Entry 1). Using butyl acetate, resolving capability value showed parity to the ethyl acetate case, but the ee of 3 was lower (Compare Table 2, Entries 1 and 4). Both yield and ee of phosphine oxide 3 were significantly lower when acetone, acetonitrile or toluene was used as solvent ( Table 2, Entries 5-7). Interestingly, the mixture of ethanol and ethyl acetate was necessary to obtain the best results, as lower ee and resolving capability values were obtained when ethanol or ethyl acetate was exclusively used for crystallization (Compare Table 2, Entry 1 with Entries 8 and 9). In the latter case, the resolution procedure was altered. Ethanol solution containing the in situ formed resolving agent [(R,R)-8] and racemic phosphine oxide 3 was evaporated to dryness, and the solid material was suspended in ethyl acetate.
The amount of resolving agent [(R,R)-8] and the composition of the solvent mixture were also optimized ( Table 2, Entries [10][11][12][13][14][15][16]. Results showed that optical resolution of ethyl-phenylpropylphosphine oxide (3) is more efficient when 1 equivalent of Ca(H-DBTA)2 is used (i.e., according to equivalent method), which indicates that the separation of two diastereomeric species [(S)-3·Ca(H-DBTA)2 and (R)-3·Ca(H-DBTA)2] by crystallization is easier than the separation of (S)-3·Ca(H-DBTA)2 diastereomer from the enantiomeric mixture containing (R)-3 in excess. Changing either concentration during crystallization ( Table 2, Entries 11 and 12), or the ratio of ethanol and ethyl acetate in the solvent mixture ( Table 2, Entries 13 and 14) did not lead to better ee and resolving capability values. Interestingly, the amount of water had a significant influence on the resolution efficiency. Throughout the optimization studies, water used was kept constant. When half amount of water was used, the yield of (S)-3·Ca(H-DBTA)2 increased to 97% from 79%, but the de of the diastereomer dropped to 56% from 74%, giving an overall resolving capability value of 0.54, which is comparable to the original results (S = 0.54). In contrast, double amount of water in the solvent mixture led to the formation of no diastereomeric complexes but the precipitation of the resolving agent [(R,R)-8] (Compare Table 2 Considering the results of the optimization studies (Tables 1 and 2), the most efficient enantiomeric separation of phosphine oxide 3 could be obtained with Ca(H-DBTA)2 in a 1:1 mixture of ethyl acetate and ethanol, and (S)-3 could be obtained with an ee of 74% after the first crystallization. Different purification strategies of enantiomeric mixtures of (S)-ethyl-phenylpropylphosphine oxide [(S)-3] were then evaluated, as it is also an integral part of every resolution procedure [59].
Three separate methods were tested, and the diastereomer of (S)-3·Ca(H-DBTA)2 prepared with a de of 74% under optimal reaction conditions were used in these experiments. Crystals of (S)-3·Ca(H-DBTA)2 were stirred at room temperature in a mixture of ethanol, ethyl acetate and water (digestion) (Scheme 3/II), or the diastereomer was recrystallized from the same solvent mixture (Scheme 3/III). The third purification method involved decomposition of (S)-3·Ca(H-DBTA)2 diastereomeric complex, followed by repeated resolution of the enantiomeric mixture of (S)-ethyl-phenylpropylphosphine oxide [(S)-3] with 1 equivalent of Ca(H-DBTA)2 [(R,R)-8] under optimal crystallization conditions (Scheme 3/IV and V). The results indicated that digestion was the least effective purification method, as de of (S)-3·Ca(H-DBTA)2 diastereomer increased from 74% to 77%. Using either recrystallization or repeated resolution, diastereomeric purity reached 94% or 93%, respectively. Repeated resolution gave somewhat better yields (48% vs. 43%). As ethyl-phenylpropylphosphine oxide (3) is an oily substrate at room temperature, the corresponding enantiomeric mixtures [(S)-3] could not be purified further by recrystallization without using any resolving agent or chiral additive.

Optical Resolution of Dialkyl-Arylphosphine Oxides (1, 2 and 4-7) with Ca(H-DBTA)2 [(R,R)-8] under Optimized Conditions
As the last step of this study, resolution of other dialkyl-arylphosphine oxides (1, 2 and 4-7) was also attempted with Ca(H-DBTA)2 [(R,R)-8] under optimal conditions. The resolving agent [(R,R)-8] was prepared in situ, a mixture of ethyl acetate, ethanol and water was used as solvent, crystallization time was 24h, and the enantiomeric mixtures of the corresponding phosphine oxides were purified by repeated resolution. In a few instances, the corresponding diastereomer was crystallized several times in order to reach high enantiomeric excess values. Resolutions were carried out according to the equivalent method (Scheme 4).   The results summarized in Table 3 show that the structure of phosphine oxide (1,2 and 4-7) had a significant influence on the efficiency of the resolution. (S)-Butyl-methyl-phenylphosphine oxide [(S)-4] could be prepared with a maximal ee of 96% after the second crystallization under optimized conditions, but yield of the corresponding phosphine oxide (S)-4, and consequently, efficiency of the resolution was rather low (Table 3, Entry 3). In order to increase the resolving capability values, resolution was carried out in half amount of solvent, and the diastereomer were crystallized three times. This change increased yield from 8% to 17%, and enantiomeric excess of (S)-4 reached also 96% after purifications (Table 3, Entry 4).
Four crystallizations were necessary, to obtain optically active ethyl-methyl-phenylphosphineoxide [(R)-1] and methyl-phenyl-i-propylphosphine oxide [(R)-5] with an ee of 85% or 80%, respectively (Table 3, Entries 1 and 5). However, yield of the corresponding diastereomers was lower than 32%, thus no additional purifications were used in these instances. Interestingly, MePrPhP(O) (2) could not be resolved efficiently (ee 37%) ( Table 3, Entry 2), despite the fact that this phosphine oxide was in close structural resemblance to other P-oxides used in this study.
c-Hexyl-and t-butyl-methyl-phenylphosphine oxides (6 and 7) could be resolved with the lowest efficiency, as no crystalline diastereomeric complexes were formed for phosphine oxide 6, whereas ee of the t-butyl-derivative (7) was only 3% (Table 3, Entry 6 and 7). These results indicated, that the increased steric demand cyclohexyl-or t-butyl-groups may hinder non-covalent interactions responsible for enantiomeric recognition, which leads to inefficient enantiomeric separation of these two phosphine oxides (6 and 7).

Conclusions
In conclusion, this study demonstrated that the coordination mediated optical resolution method originally developed for cyclic phosphine oxides has a more general scope, and this technique can be used for acyclic phosphine oxides. Enantiomeric separation of a few dialkylphenylphosphine oxides (1-5 and 7) was elaborated with metals salts of tartaric acid derivatives. The resolution procedure, crystallization and purification conditions were optimized, and the most efficient enantioseparation was obtained with acidic Ca 2+

Conflicts of Interest:
The authors declare no conflict of interest.