Application of threo‐2‐amino‐1,2‐diphenylethanol as a Resolving Agent for Racemic Carboxylic Acids via Diastereomeric Salt Formation Authors: Koichi Kodama, Kosei Katayama, Takuji Hirose Published: 2026 ABSTRACT ABSTRACTThe ability of threo‐2‐amino‐1,2‐diphenylethanol (1t) as a resolving agent for the enantioseparation of carboxylic acids via diastereomeric salt formation was investigated. Although (−)‐1t was not applicable to the enantioseparation of racemic 2‐arylalkanoic acids and mandelic acid, enantiomers of tropic acid and 2‐halomandelic acids were efficiently separated by salt formation with (−)‐1t. This resolving ability was complementary to that of the corresponding erythro‐isomer (1e). The hydrogen‐bonding patterns and molecular packing mode of the less‐soluble diastereomeric salts of (−)‐1t were revealed based on the crystallographic analysis. Introduction Chiral 1,2‐aminoalcohols are versatile scaffolds for chiral drugs and catalysts in asymmetric synthesis [1, 2]. Among these, erythro‐2‐amino‐1,2‐diphenylethanol (1e) is commercially available in an enantiopure form from various manufacturers and it has been frequently used as a chiral auxiliary in the fields of asymmetric synthesis and chiral recognition [3, 4]. On the other hand, despite its similar chiral skeleton, the corresponding “threo” isomer, threo‐2‐amino‐1,2‐diphenylethanol (1t), has been less studied than 1e probably due to its limited accessibility (Figure 1) [5, 6, 7, 8]. Several studies have reported that chiral organocatalysts derived from 1t exhibited superior chiral induction ability to those derived from 1e [9, 10]. These results indicate that 1t is a potent compound for chiral recognition. Structures of (1R,2S)‐(−)‐1e and (1S,2S)‐(−)‐1t. Enantiomer separation via diastereomeric salt formation is a powerful and reliable method to access enantiopure acidic/basic compounds from their racemates [11, 12]. For separation of a racemic chiral carboxylic acid, enantiopure amines are used to form a diastereomeric salt pair. The 1,2‐aminoalcohol 1e is a frequently used resolving agent which is applicable to the enantioseparation of various racemic carboxylic acids via diastereomeric salt formation [13, 14, 15]. Furthermore, we have studied solvent‐induced chirality switching during the resolution of hydroxycarboxylic acids with 1e [16, 17, 18, 19]. Despite their structural similarity, the potential of 1t as a resolving agent has been overlooked. In this study we examined the utility of 1t as a resolving agent for the separation of several chiral carboxylic acids via diastereomeric salt formation. Furthermore, we compared the result with that of 1e and discussed the chiral recognition mechanism based on the crystal structures of the diastereomeric salts. Experimental Materials and Methods All the reagents and solvents were purchased and used as received unless noted. (1S,2S)‐(−)‐2‐Amino‐1,2‐diphenylethanol (1t) was synthesized from (1R,2S)‐(−)‐2‐amino‐1,2‐diphenylethanol (1e) according to literature (Scheme S1) [20]. Copies of the 1H NMR and IR spectra for the synthesized compounds are provided in Supporting Information (Figures S1–S4). Racemic halogen‐substituted mandelic acids (3d–3g) were reported in our previous literature [21]. The 1H NMR spectra were recorded by using a Bruker Avance 300 MHz NMR spectrometer. IR spectra were measured on a JASCO FT/IR‐4600 spectrometer as KBr pellets. Melting points were recorded on a MEL‐TEMP apparatus and reported uncorrected. The enantiomeric excess (ee) was determined by chiral HPLC analysis on a Daicel Chiralcel OD‐3 or OJ‐3 column with detection at 254 nm. General Procedure for the Enantiomer Separation of Carboxylic Acids (2 and 3) With (−)‐1t Equimolar amounts (0.50 or 0.35 mmol) of racemic carboxylic acids (2 or 3) and (−)‐1t were dissolved in methanol. After concentration, the resulting white solids were dissolved in an appropriate solvent by heating and allowed to stand at room temperature to obtain the salt crystals. The solids were filtered and dried under reduced pressure. The yield was calculated based on half the amount of the initial racemic acid. A portion of the salt (approximately 20 mg) was dissolved in 1‐M aqueous hydrochloric acid solution, and the aqueous solution was extracted with chloroform. After drying over anhydrous Na2SO4, the organic layer was concentrated to obtain the carboxylic acid. The resulting carboxylic acid was dissolved in methanol (1 mL) and toluene (2 mL) and trimethylsilyldiazomethane (hexane solution) was added until the solution remained yellow. The mixture was stirred for 30 min at room temperature, and the solvent was removed under reduced pressure. The residue was purified by silica gel PTLC (hexane/ethyl acetate = 2:1) to produce the corresponding methyl ester. The enantiomeric excess (ee) value of the ester was determined by chiral HPLC analysis. Copies of the HPLC charts are provided in Supporting Information (Figures S5–S19). Single Crystal X‐Ray Analysis of the Diastereomeric Salts Single crystals suitable for X‐ray diffraction analysis were prepared by slow evaporation of a saturated solution of the salts. X‐ray crystallographic data were collected on a Bruker Smart APEX II or D8 Quest ECO diffractometer using graphite monochromated Mo Kα radiation. Data collections were carried out at 150 K. The structures were determined via a direct method (SIR 2014) and refined using the SHELXL‐2018 program [22]. The crystallographic data reported in this study are summarized in Table S1. Crystallographic data (excluding structure factors) for the structures in this article have been deposited with the Cambridge Crystallographic Data Centre as supplementary publication nos. shown above. Copies of the data can be obtained, free of charge, on application to CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax: þ44‐(0) 1223e336033 or e‐mail: deposit@ccdc.cam.ac.uk). Results and Discussion The compound (1S,2S)‐(−)‐1t was synthesized from commercially available (1R,2S)‐(−)‐1e according to literature and characterized by NMR and IR spectra (see Supporting Information) [20]. First, we investigated the enantioseparation of three chiral 2‐phenylalkanoic acids (2a–2c) with (−)‐1t. The enantiomers of 2a and 2b are known to be separated by salt formation with 1e while 2c is not [13]. Equimolar amounts of each racemic acid 2 and (−)‐1t were mixed and recrystallized from suitable solvents. The results are shown in Table 1. The yield of the precipitated salt was calculated based on half the amount of the initial racemate. A portion of the precipitated salt was decomposed and the enantiomeric excess (ee) of the separated acid 2 was determined by chiral HPLC analysis after derivatization to the corresponding methyl ester. The resolution efficiency was defined as the product of the yield and ee values, reflecting the possibility of the separation. TABLE 1Enantiomer separation of racemic 2a–2c by diastereomeric salt formation with (−)‐1t. EntryRacemateSolvent (L/mol)Yield a (%)Ee b (%)Eff. c 1 2a Water (7)3916(R)0.062 2a Toluene (0.5)3222(R)0.073 2b Water (5)134 Rac.04 2b TolueneNot crystallized—5 2c WaterNot crystallized—6 2c TolueneNot crystallized— a Yield of the precipitated salt is calculated based on the half amount of the initial salt. b Ee was determined by HPLC analysis after derivatization to the corresponding methyl esters. Absolute configuration of the major enantiomer is shown in the parentheses. c Resolution efficiency = yield (%) × ee (%)/10,000. Enantiomer separation of racemic 2a–2c by diastereomeric salt formation with (−)‐1t. Yield of the precipitated salt is calculated based on the half amount of the initial salt. Ee was determined by HPLC analysis after derivatization to the corresponding methyl esters. Absolute configuration of the major enantiomer is shown in the parentheses. Resolution efficiency = yield (%) × ee (%)/10,000. After a preliminary investigation of recrystallization solvents, the salt of 2‐phenylpropionic acid (2a) obtained from water and toluene both afforded (R)‐2a with low resolution efficiencies (Entries 1 and 2). The salt of 2‐phenylbutanoic acid (2b) and (−)‐1t was crystallized from water but exhibited no enantiomeric preference (Entry 3). Finally, the salt of 3‐methyl‐2‐phenylbutanoic acid (2c) failed to crystallize from these solvents (Entries 5 and 6). These results suggest that a bulkier alkyl chain on the stereogenic center of the carboxylic acid 2 increases the solubility of the salts, thereby preventing crystallization. Consequently, in contrast to 1e, 1t was found to be unsuitable as a resolving agent for these 2‐phenylalkanoic acids (2). Next, (−)‐1t was applied to the enantioseparation of chiral hydroxycarboxylic acids (3a–3c). The enantiomers of these acids are important building blocks for bioactive compounds such as hyoscyamine [23, 24]. The enantioseparation results are summarized in Table 2. In contrast to the salts formed with 2, the salts of (−)‐1t with mandelic acid (3a) and 3‐phenyllactic acid (3b) were insoluble in water or toluene due to their low solubility (Entries 1–4). Although 1e was successfully applied to the resolution of racemic 3a and 3b, 1t was ineffective for these hydroxy acids [16, 18]. TABLE 2Enantiomer separation of racemic 3a–3c by diastereomeric salt formation with (−)‐1t. EntryRacemateSolvent (L/mol)Yield a (%)Ee b (%)Eff. c 1 3a WaterNot dissolved2 3a TolueneNot dissolved3 3b WaterNot dissolved4 3b TolueneNot dissolved5 3c WaterNot crystallized6 3c TolueneNot dissolved7 3c Ethanol (8)7294(R)0.688 3c 2‐Propanol (6)5080(R)0.409 3c Acetonitrile (20)8182(R)0.67 a Yield of the precipitated salt is calculated based on the half amount of the initial salt. b Ee was determined by HPLC analysis after derivatization to the corresponding methyl esters. Absolute configuration of the major enantiomer is shown in the parentheses. c Resolution efficiency = yield (%) × ee (%)/10,000. Enantiomer separation of racemic 3a–3c by diastereomeric salt formation with (−)‐1t. Yield of the precipitated salt is calculated based on the half amount of the initial salt. Ee was determined by HPLC analysis after derivatization to the corresponding methyl esters. Absolute configuration of the major enantiomer is shown in the parentheses. Resolution efficiency = yield (%) × ee (%)/10,000. On the other hand, the salt of (−)‐1t with tropic acid (3c), a structural isomer of 3b, was successfully crystallized using alcoholic solvents. Recrystallization from ethanol afforded the salt of (R)‐3c in high yield (72%) and enantiomeric excess (94% ee) (Entry 7). The resolution efficiency was as high as 0.68. Although crystallization from 2‐propanol afforded (R)‐3c salt with lower efficiency (Entry 8), acetonitrile proved to be an equally suitable solvent, yielding the (−)‐1t·(R)‐3c salt crystal with a high efficiency of 0.67 (Entry 9). This exceptionally high resolution efficiency achieved by the resolving agent 1t was comparable to that obtained with 1e [17]. To investigate the origin of the high efficiency in the enantioseparation of 3c with (−)‐1t, a single‐crystal X‐ray diffraction analysis of the diastereomeric salt was conducted. Colorless triangle‐shaped platelet crystals were grown by the recrystallization of the (−)‐1t·3c salt from ethanol, and the structure is shown in Figure 2a. The crystal was composed of (−)‐1t and 3c and they formed 1:1 salt. The stereochemistry of 3c was determined to be (R)‐isomer, which was consistent with the result of enantioseparation shown in Table 2. The polar ammonium and hydroxy groups of (−)‐1t adopted a gauche position and two bulky phenyl groups also occupied a gauche position. Such a gauche conformation has been similarly observed for (−)‐1e in its carboxylate salts reported previously (Figure 2b) [13]. In contrast to (−)‐1e, both phenyl groups in (−)‐1t occupied the antipositions relative to the polar functional groups, placing them further away from the polar groups. (a) Crystal structure of (−)‐1t·(R)‐3c salt viewed from the a axis. Oxygen and nitrogen atoms are represented with red and blue balls, respectively. The dotted lines show hydrogen bonds. (b) The molecular conformation of (−)‐1t and (−)‐1e in their less‐soluble salts with 3c. (c) Hydrogen‐bonding network of (−)‐1t·(R)‐3c salt viewed perpendicular to the b axis. The dotted lines show hydrogen bonds. Two of the three ammonium hydrogen atoms of (−)‐1t were involved in hydrogen bonds with the carboxylate oxygen atoms of (R)‐3c and remaining ammonium hydrogen atom did not participate in hydrogen bond formation (Figure 2c). The hydroxy hydrogen atom of both (−)‐1t and (R)‐3c formed hydrogen bonds with the carboxylate oxygen atoms of (R)‐3c. This additional hydrogen bond involving the OH group of 3c likely resulted in crystalline salt formation compared with 2. Through these hydrogen bonds, two‐dimensional network was constructed along the ab plane. Although the surface of this layer‐like structure was not flat due to the two phenyl groups of (−)‐1t oriented in a gauche position, neighboring layers stacked in an interpenetrating manner to achieve efficient molecular packing. The successful chiral recognition of (R)‐isomer of 3c was attributable to a CH/O interaction between the hydrogen atom of (R)‐3c on the stereogenic center and the hydroxy oxygen atom of (−)‐1t (with a CH···O distance of 2.44 Å and a C–H···O angle of 148°) (Figure 2a). We previously reported that in the less‐soluble diastereomeric salt of (−)‐1e and (S)‐3c, one‐dimensional columnar hydrogen‐bonding network was constructed with the aid of incorporated alcohol molecules [17]. These studies revealed that one important factor enabling 1e to act as an effective resolving agent is its ability to form stable salts with carboxylic acids by commonly constructing one‐dimensional columnar hydrogen‐bonding network. In contrast to (−)‐1e, the formation of the two‐dimensional hydrogen‐bonding network by the (−)‐1t salt was due to the less crowded conformation of (−)‐1t around its hydrogen‐bonding groups. Next, we focused on the enantioseparation of chlorine‐substituted mandelic acids, 4‐chloromandelic acid (3d) and 2‐chloromandelic acid (3e). The enantiomers of 3d and 3e serve as important chiral synthetic intermediates for pharmaceuticals such as clopidogrel [25, 26]. We previously reported that 1e was applicable to the enantioseparation of racemic 3d, whereas 3e could not be separated because its salt with 1e failed to crystallize. The results of the enantiomer separation of 3d and 3e with (−)‐1t are summarized in Table 3. TABLE 3Enantiomer separation of racemic 3d–3g by diastereomeric salt formation with (−)‐1t. EntryRacemateSolvent (L/mol)Yield a (%)Ee b (%)Eff. c 1 3d EtOH (3.5)508(R)0.042 3d Acetonitrile (27)1116(R)0.063 3d Ethyl acetate (86)239(R)0.024 3e Acetonitrile (2.5)8172(R)0.585 3e Ethyl acetate (11)6877(R)0.526 3f Acetonitrile (4)7076(R)0.537 3f Ethyl acetate (14)9867(R)0.668 3g Acetonitrile (6)9464(R)0.609 3g Ethyl acetate (14)7970(R)0.55 a Yield of the precipitated salt is calculated based on the half amount of the initial salt. b Ee was determined by HPLC analysis after derivatization to the corresponding methyl esters. Absolute configuration of the major enantiomer is shown in the parentheses. c Resolution efficiency = yield (%) × ee (%)/10,000. Enantiomer separation of racemic 3d–3g by diastereomeric salt formation with (−)‐1t. Yield of the precipitated salt is calculated based on the half amount of the initial salt. Ee was determined by HPLC analysis after derivatization to the corresponding methyl esters. Absolute configuration of the major enantiomer is shown in the parentheses. Resolution efficiency = yield (%) × ee (%)/10,000. The salt with 3d was deposited from several solvents; however, the ee values were quite low (Entries 1–3). On the other hand, the salt with 3e was crystallized from ethanol or ethyl acetate and (R)‐3e was dominantly included in the salt. The resolution efficiency was as high as 0.58 (Entries 4 and 5). In contrast to the resolving agent 1e, 1t exhibited high performance for the resolution of the ortho‐chlorinated mandelic acid 3e. Prompted by this result, we further investigated two additional mandelic acid derivatives as target racemates, 2‐bromomandelic acid (3f) and 2‐iodomandelic acid (3g), bearing a halogen atom at the ortho‐position of the phenyl group (Table 3). Similar to the results obtained for 3e, the salts of (−)‐1t with 3f and 3g were obtained from acetonitrile or ethyl acetate solutions, both showing high enrichment of the (R)‐isomer (Entries 6–9). The resolution efficiencies exceeded 0.6, demonstrating that (−)‐1t can be commonly applied to the enantioseparation of ortho‐halo‐substituted mandelic acids. The single crystal of the less‐soluble diastereomeric salt of (−)‐1t with (R)‐3e was prepared from an aqueous acetonitrile solution and its crystallographic analysis was carried out (Figure 3). The crystal was composed of 1:1 mixture of (−)‐1t and (R)‐3e to form a salt. The molecular conformation of (−)‐1t was identical to that observed in the salt with (R)‐3c, wherein the two phenyl groups and two hydrogen‐bonding groups of (−)‐1t adopted a gauche conformation. Two independent molecules of (−)‐1t and (R)‐3e were present in a unit cell and a complicated one‐dimensional columnar hydrogen‐bonding network was constructed along the a axis (Figure 3a). The molecules (R)‐3e and (−)‐1t were alternately stacked along the columnar direction. In addition to the ammonium‐carboxylate hydrogen bonds, the hydroxy groups of both (−)‐1t and (R)‐3e participated in the hydrogen‐bonding network. It is probable that the chlorine substituent at the ortho‐position of (R)‐3e occupied the space around the polar groups, which permitted the formation of a one‐dimensional hydrogen‐bonding network similar to the case of the salt formed with (−)‐1e. The benzene rings of (−)‐1t were oriented parallel to the columnar axis, whereas that of (R)‐3e was perpendicular to the columnar axis. As a result, the phenyl group of (R)‐3e was fixed by dual CH/π interactions with the CH moieties of (−)‐1t (Figure 3b). Such a fixation of the position of the phenyl group resulted in the effective recognition of (R)‐3e during salt formation. The consistently high resolution efficiencies observed for the three ortho‐halomandelic acids imply that the successful chiral recognition of (R)‐3f and (R)‐3g by (−)‐1t can be attributed to effects similar to those of (R)‐3e stated above. Crystal structure of (−)‐1t·(R)‐3e salt (a) viewed from the a axis. Oxygen, nitrogen, and chlorine atoms are represented with red, blue, and yellow balls, respectively. The dotted lines show hydrogen bonds. (b) Viewed from the c axis. The arrows show CH/π interactions. Conclusion In conclusion, the ability of threo‐(1S,2S)‐2‐amino‐1,2‐diphenylethanol (1t) as a resolving agent for the enantioseparation of carboxylic acids via diastereomeric salt formation was investigated. Although (−)‐1t was less versatile than its erythro‐isomer (1e) and failed to separate 2‐phenylalkanoic acids (2a–2c), mandelic acid (3a), and 3‐phenyllactic acid (3b), the enantiomers of tropic acid (3c) and 2‐halomandelic acids (3e–3g) were efficiently separated. Remarkably, while 1e was unsuitable for the enantioseparation of 3e–3g, 1t exhibited high resolution efficiencies, demonstrating a complementary relationship between these two isomers. From the crystallographic analysis of the less‐soluble diastereomeric salts of (−)‐1t, the two phenyl groups of (−)‐1t are remote from the hydrogen‐bonding groups. The salt with 3c was stabilized by the construction of a layer‐like hydrogen‐bonding network. On the other hand, the halogen substituent at the ortho position of phenyl group of 3e compensated for the vacant space around hydrogen‐bonding network, leading to the construction of a stable columnar hydrogen‐bonding network. Supporting information Scheme S1 Synthesis of (1S,2S)‐2‐amino‐1,2‐diphenylethanol ((−)‐1t). Table S1: Summary of crystallographic data reported in this study. Figure S1: 1H NMR spectrum of N‐formyl‐(1R,2S)‐2‐amino‐1,2‐diphenylethanol. Figure S2: IR spectrum of N‐formyl‐(1R,2S)‐2‐amino‐1,2‐diphenylethanol. Figure S3. 1H NMR spectrum of (1S,2S)‐2‐amino‐1,2‐diphenylethanol ((−)‐1t). Figure S4. IR spectrum of (1S,2S)‐2‐amino‐1,2‐diphenylethanol ((−)‐1t). Figure S5. HPLC chart of methyl ester of 2a (Table 1, Entry 1). Figure S6. HPLC chart of methyl ester of 2a (Table 1, Entry 2). Figure S7. HPLC chart of methyl ester of 2b (Table 1, Entry 3). Figure S8. HPLC chart of methyl ester of 3c (Table 2, Entry 7). Figure S9. HPLC chart of methyl ester of 3c (Table 2, Entry 8). Figure S10. HPLC chart of methyl ester of 3c (Table 2, Entry 9). Figure S11. HPLC chart of methyl ester of 3d (Table 3, Entry 1). Figure S12. HPLC chart of methyl ester of 3d (Table 3, Entry 2). Figure S13. HPLC chart of methyl ester of 3d (Table 3, Entry 3). Figure S14. HPLC chart of methyl ester of 3e (Table 3, Entry 4). Figure S15. HPLC chart of methyl ester of 3e (Table 3, Entry 5). Figure S16. HPLC chart of methyl ester of 3f (Table 3, Entry 6). Figure S17. HPLC chart of methyl ester of 3f (Table 3, Entry 7). Figure S18. HPLC chart of methyl ester of 3g (Table 3, Entry 8). Figure S19. HPLC chart of methyl ester of 3g (Table 3, Entry 9).