CO2-Assisted asymmetric hydrogenation of prochiral allylamines

A new methodology for the asymmetric hydrogenation of allylamines takes advantage of a reversible reaction between amines and carbon dioxide (CO2) to suppress unwanted side reactions. The effects of various parameters (pressure, time, solvent, and base additives) on the enantioselectivity and conversion of the reaction were studied. The homogeneously-catalyzed asymmetric hydrogenation of 2-arylprop-2-en-1-amine resulted in complete conversion and up to 82% enantiomeric excess (ee). Added base, if chosen carefully, improves the enantioselectivity and chemoselectivity of the overall reaction.


Introduction
Optically active amines are used as pharmaceuticals, agrochemicals and resolving agents or chiral auxiliaries. [1][2][3][4][5][6] Many efforts have been directed towards the enantioselective hydrogenation of C]C and C]N double bonds for the synthesis of chiral amines. 1,3,4,[7][8][9][10][11][12][13] We sought an asymmetric hydrogenation of prochiral allylamines, in the expectation that it would represent a more direct, efficient and greener synthesis of chiral amines than current asymmetric hydrogenations of N-protected allylamines. Until today, the direct hydrogenation of unprotected allylamines has been largely ignored; the few examples are shown in Scheme 1. Botteghi et al. 14 reported that the hydrogenation gave low yields due to the unwanted hydrogenolysis of the C-N bond. Fahrang et al. 15 strategically hydrogenated the hydrogen chloride salt of their allylamine but did not comment on yield or purity of the product. However, both groups reported only moderate enantioselectivity. Yamashita and Yamano 16 screened multiple Josiphos ligands to nd one with good enantioselectivity for the hydrogenation of a precursor of Ramelteon, a melatonin receptor agonist.
We hypothesized that CO 2 could act as an in situ protecting group in a way that protects the amine from undesired side reactions. This could potentially replace the N-acetyl protecting group that is currently used for asymmetric hydrogenation of protected allylamines. The CO 2 would reversibly convert the allylamine substrate and/or the chiral amine product into a carbamate or carbamic acid (solid arrows in Scheme 2), which would circumvent additional steps of amine protection and deprotection, ultimately giving a more economical synthesis. In separate reports by Chatterjee et al. 17 and Xie et al. 18 CO 2 was used in the hydrogenation of nitriles and imines to protect the desired amine products, by the formation of a carbamic acid, from undergoing undesired further reactions. Thus the carbamic acid acts as an in situ protecting group for the kinetic product during the hydrogenation. 17,18 Fortunately, the reaction of allylamines with CO 2 to form carbamic acids or carbamate anions is known, but in the context of synthesizing cyclic carbamate esters. [19][20][21][22][23][24][25] In addition to serving as a protecting group, the carbamic acid or carbamate anion produced by the reaction of CO 2 with the amine might serve as a better metal-binding functional group (lower part of Scheme 2), allowing chelation in the hydrogenation transition state as occurs when unsaturated carboxylic acids are asymmetrically hydrogenated. 26 The olen binding step in the catalytic cycle, in which the C]C double bond is bound to the metal centre prior to insertion into the M-H bond, would thereby become an intramolecular rather than intermolecular step, which would quite feasibly enhance enantioselectivity.
To explore these two intriguing hypotheses, we chose to study the asymmetric hydrogenation of a primary allylamine with and without CO 2 , and with and without added base. The option of adding a base was included in order to promote the formation of carbamate anions rather than carbamic acids.
The results show that the conditions greatly affect the asymmetric hydrogenation of 6 (Table 1, 24 h). With solely H 2 , all catalysts produced 2-phenylpropan-1-amine, 7, in low to moderate yields and enantioselectivity (ee), consistent with the ndings of Botteghi et al. 14 Catalyst 1 gave the highest yield, 72%, and catalyst 5 gave the best ee, 74%.
The addition of DBU made minor improvements to the enantioselectivity of catalysts 1 and 2, but was otherwise unhelpful. The addition of CO 2 without base increased the yield (Table 1) and the purity of the product (by suppressing side products, Fig. S4 †) but not the enantioselectivity for catalysts 2-5. The addition of CO 2 with base dramatically increased the yield with all catalysts except 4. The enantioselectivity of the hydrogenation was adversely affected for all catalysts except 5. The best result, with high yield (96%) and reasonable ee (75%), was obtained with catalyst 5 in the presence of both CO 2 and base.
Following the positive results for the asymmetric hydrogenation at 24 h, the reaction time was investigated (Table 1, 14-Scheme 2 Upper route with solid arrows: the yield of the direct hydrogenation could be increased by the starting allylamine and/or the product being stabilized as the carbamate, even though the hydrogenation step itself involves the allylamine rather than the carbamate. Lower route with dotted arrows: alternatively, the yield and enantioselectivity could both be improved by the allylcarbamate binding to the metal centre, allowing chelation during the hydrogen transfer step.
Scheme 3 The catalysts initially tested for the asymmetric hydrogenation of prochiral allylamines.
15 h). With a decreased reaction time of 14-15, an increase in yield was observed for almost all catalysts and conditions, suggesting that extended reaction times allow the desired products to undergo further reactions giving unwanted products. However, enantioselectivities were not signicantly changed by the decrease in reaction time. Even shorter reaction times give lower enantioselectivity (Table S2 †). The success of catalyst 5 suggests that Rh-based catalysts may be more suitable than classical Ru BINAP catalysts for the asymmetric hydrogenation of 2-phenylprop-2-en-1-amine, 6. This seems surprising if one considers allylamines to be close analogues of allylic alcohols, for which the classical Ru BINAP catalysts are known to be excellent hydrogenation catalysts. 28 However, perhaps a better analogy would be to the b-ketoamines, for which cationic Rh complexes are better hydrogenation catalysts than the classical Ru BINAP complexes. 29 In the proposed transition state for those hydrogenations, the amine group binds to the Rh centre and the C]O double bond then undergoes Rh-H insertion leading to hydrogenation. 29 A similar mechanism may operate for the asymmetric hydrogenation of allylamines, although it is worth noting that a DFT study of the mechanism for asymmetric isomerization of allylamines (for which Rh catalysts are again superior to Ru) shows that the nitrogen is not coordinated during the hydrogen transfer step. 30 In light of the success of catalyst 5, three more Rh based catalysts were chosen from the catASium® family (Scheme 5) with the corresponding ligands (R,R)-Me-DUPHOS, 8, (R,R)-Me-BPE-Rh, 9, 8 and 3,4-bis-[(R,R)-(2,5-dimethylphospholan-1-yl] maleic anhydride 10. 8,31 Unfortunately, compared to catalyst 5, catalysts 8, 9 and 10 did not provide improved results, with catalyst 10 yielding similar results (Table S1 †). H 2 pressure is known to affect hydrogenation enantioselectivity. 26 To evaluate the effect of H 2 pressure in the present system, a lower pressure was tested. The reaction time was increased to 24 h to compensate for the anticipated lower rate of reaction. Unfortunately, the lowered H 2 pressure decreased the performance of catalyst 5 and caused no signicant improvements with 8 and 9 (Table S1 †).
Next, the effect of solvent on the reaction was examined ( Table 2). For catalyst 8 it was reported that the best solvents for the asymmetric hydrogenations of a-aminomethylacrylates, 26 ene-carbonates, 3 b-acylamido acrylates, 3 and enamides 4 were isopropanol (IPA), methanol (MeOH), and tetrahydrofuran (THF). For this reason, the asymmetric hydrogenation of 6 in the presence of CO 2 was tested in these solvents with catalysts 5 and 8 but no signicant improvement was obtained relative to the results with catalyst 5 in MeOH.
Despite the complete conversion of 6 in all runs, yields were low in many instances and unidentied peaks were observed in the 1 H NMR spectra. Even though the overall yields were found to be higher with the addition of DBU, we suspected that the use of DBU as the base may be leading to or assisting the decomposition of either the starting material or product. Therefore, we investigated the use of weaker bases (Table 2). N,N-Dimethylcyclohexylamine (CyNMe 2 ), and N,N-diisopropylethylamine (iPr 2 NEt) were tested with catalysts 5 and 8 in the solvent that provided the best results for each (catalyst 5 with MeOH and IPA, and catalyst 8 with IPA). The best condition for catalyst 5 a Experiments were done in triplicate and at RT in a 160 mL stainless steel vessel containing 10 mg 6 and 2 mL methanol in a 1 dram vial under 100 bar total pressure. Conversions for all reactions above were >95% and the experimental error for % yield and % ee were AE10 and AE4, respectively. Catalysts 2 and 5 produced (S)-7, while catalysts 1, 3, and 4 produced (R)-7. Yields are 1 H NMR values measured with an internal standard (1,3,5-trimethoxybenzene). Enantiomeric excess was determined by HPLC. b 100 bar H 2 . c 100 bar H 2 , 1 eq. DBU added (relative to 6). d 10 bar CO 2(g) added, followed by enough H 2(g) to bring the total pressure to 100 bar. e 10 bar CO 2(g) added, followed by enough H 2(g) to bring the total pressure to 100 bar, 1 eq. DBU added (relative to 6).
were found in the presence of H 2 , CO 2 , MeOH and DBU. However, CyNMe 2 produced the cleanest reaction by 1 H NMR spectroscopy with comparable enantioselectivity. For catalyst 8, the addition of CO 2 , IPA and iPr 2 NEt resulted in the highest yield (83%) and ee (72%). Catalyst 10 gave decent results in methanol with CO 2 and the weaker bases, but the best overall result is still with catalyst 5 in the presence of CO 2 and DBU. Using chiral bases (Scheme 6) caused modest improvements in the enantioselectivity with catalyst 8 but not catalyst 5 (Table  S3 †). The enantioselectivity was not affected by the chirality of the base, presumably because the chiral bases, in their cationic form, were not close enough to the catalytic centre to induce a chiral environment. Therefore, the success of these chiral bases at mildly improving the enantioselectivity is due to their weaker basicity rather than their chirality.
Using the best hydrogenation conditions (CO 2 + DBU in MeOH for 24 h) with catalyst 5, the reaction was scaled to 650 mg to obtain an isolated yield of 416 mg (64%, Fig. S5 †).
The asymmetric hydrogenation was also tested on three other allylamine substrates using catalysts 5, 8 and 10 (Scheme 7). For catalyst 5 the asymmetric hydrogenation was performed with CO 2 and CyNMe 2 , whereas catalyst 8 was utilized with CO 2 and (À)-11. For 10, both sets of conditions used for 5 and 8 were applied and found to be equally successful (Table 3).
Changing the phenyl ring of substrate 6 to a larger naphthyl ring in substrate 12 lowered the enantioselectivity by about 10% (Table 3). Adding electron-donating and electronwithdrawing groups to the para position of substrate 6 affected both the yield and the enantioselectivity of the allylamine. Adding an electron-donating ethoxy group (substrate 13) increased the ee to 81-82%. However, adding an electronwithdrawing triuoromethyl group on the para position (substrate 14) had the opposite effect where yields and enantioselectivity both decreased; the 1 H NMR spectrum of the reaction mixture aer the reaction appeared clean and showed that the reaction was incomplete aer 6 h. While the amount of product from substrate 14 might improve if the reaction time were longer, the enantioselectivity is clearly poor for this substrate.  b 100 bar H 2 . c 100 bar H 2 , 1 eq. of base added (relative to 6). d 10 bar CO 2(g) added, followed by enough H 2(g) to bring the total pressure to 100 bar. e 10 bar CO 2(g) added, followed by enough H 2(g) to bring the total pressure to 100 bar, 1 eq. of base added (relative to 6). f Conversion% 70-76%. g Conversion% 83%.
Scheme 6 Chiral bases used in the asymmetric hydrogenation of 6.

Conclusions
A new methodology has been developed for the asymmetric hydrogenation of allylamines. It was found that the Rh-based catASium® catalysts resulted in higher conversion and enantiomeric excess values than the Ru-binap based catalysts. Furthermore, by employing CO 2 and an added base in the asymmetric hydrogenation of 2-phenylprop-2-en-1-amine, 6, a clean reaction was obtained, probably due to CO 2 acting as a temporary protecting group for the amine functionality and ultimately increasing the yield of the reaction up to 94-96%. Nonetheless, the enantioselectivity of the reaction was not affected by the addition of CO 2 and DBU. This demonstrates that the CO 2 is not affecting the enantioselectivity-determining step and therefore the allylamine is not bound to the catalyst in the form of a carbamate ligand. The CO 2 helps by acting as a protecting group and not by causing the allylamines to bind as carbamates to the metal centre. Four 2-arylprop-2-en-1-amines were asymmetrically hydrogenated with the best catalysts from the above study at their optimal reaction conditions. Allylamine 2-(4-ethoxyphenyl) prop-2-en-1-amine, with an electron donating group, was hydrogenated with the greatest enantioselectivity (82% ee) and good yield (93%).
These ndings demonstrate that a direct asymmetric hydrogenation of prochiral allylamines, without prior derivatization or protection, is a viable strategy for preparing chiral amines. Further optimization of the catalyst and the conditions, including the benecial effect of CO 2 , should be able to bring the enantioselectivity to industrially useable levels.

Experimental methods
Solvents were dried by standard distillation procedures before use. All reagents were purchased from chemical suppliers, Alfa Aesar, Sigma Aldrich, and Strem, and used as received unless otherwise specied. The four allylamines were synthesized as described in the ESI. † Glassware was dried in an oven at 110 C before use. 1 H NMR and 13 C NMR spectra were recorded at 300 K on a Bruker AV-400 or AV-500 NMR spectrometer. Chemical shis (d) are expressed in ppm. Conversion and yield values were obtained through quantitative NMR spectroscopy, which was carried out using 1,3,5-trimethoxybenzene as the internal standard. Enantiomeric excess values were obtained by HPLC using Agilent Technologies 1260 Innity with Chiralpak IA chiral column (25 cm Â 0.46 cm i.d.) from Daicel. High resolution mass spectra (HRMS) ESI and EI were obtained on a Qstar XL QqTOF from Applied Biosystems/MDS Sciex.

General racemic hydrogenation for allylamines
The allylamines were rst hydrogenated with achiral catalysts in order to generate samples of the racemates for the development of instrumental methods capable of analyzing the enantiomeric mixture. The non-enantioselective hydrogenations were developed from a procedure by Hattori et al. 32 The procedure below is the same for all substrates: the hydrogenation of the allylamines, Pd(5%)/CaCO 3 was used. The hydrogenation of 2phenylprop-2-en-1-amine, 6, is presented below and can be regarded as a general protocol for the procedure regardless of minor changes in the substrates.
To a test tube, equipped with a magnetic stir bar, 2phenylprop-2-en-1-amine (10 mg, 75.1 mmol) and catalyst (0.5 wt% of the weight of the substrate, 2.1 mg) was added. The test tube was sealed with a rubber septum and then evacuated. THF (1 mL) was added and then hydrogen was added via a syringe needle from a hydrogen-lled rubber balloon (1 atm). The reaction mixture was stirred at room temperature for 5-6 h. Upon completion, the catalyst was removed by ltration through a diatomaceous earth plug, aer which the solvent was removed from the product by rotatory evaporation.
Stock solutions of the allylamine, catalyst, and the optional base were prepared in dry methanol the same day as the planned hydrogenation to ensure no decomposition of the chemicals occurred. In a 160 mL stainless steel high pressure vessel, containing up to a dozen 1 dram glass vials, each containing a magnetic stir bar, 2-phenylprop-2-en-1-amine, 6, (10 mg, 0.075 mmol), catalyst, (À)-4,5-bis[(2R,5R)-2,5-dimethylphospholanyl](1,2dimethyl-1,2-dihydropyridazine-3,6-dione)(1,5-cyclooctadiene) rhodium(I) tetrauoroborate, (1 mg, 0.0015 mmol) and, if desired, the optional base (ca. 0.075 mmol) was added under a nitrogen atmosphere. Additional dry methanol was added to each vial to obtain a total volume of 2 mL and then the vessel was sealed. The vessel was ushed 3 times with H 2 gas or (if CO 2 use in the experiment was planned) CO 2 gas, and pressurized at room temperature (22 C) to either 100 bar H 2 gas, or if the presence of CO 2 is desired, the vessel was pressurized to 10 bar with CO 2 gas and then H 2 gas was added until the total pressure was 100 bar. It is not correct to assume that the partial pressure of the H 2 gas was equal to the difference between the total pressure and the pressure of the CO 2 gas, because of signicant CO 2 -H 2 interactions. Once the vessel was pressurized, the reaction mixture was stirred for 6-12 h at room temperature. Once the reaction time was complete, the vessel was slowly depressurized, the solutions were ltered through diatomaceous earth, and concentrated by rotatory evaporation. Enantiomeric excess was determined by HPLC and yield was determined by 1 H-NMR spectroscopy using an internal standard, 1,3,5-trimethoxybenzene.
2-Phenylpropan-1-amine. The NMR yield was >95%. The 1 H and 13 C spectra matched those of the commercially available compound and those reported in the literature. 33

Conflicts of interest
There are no conicts to declare.