Improved Access to Chiral Tetranaphthoazepinium-Based Organocatalysts Using Aqueous Ammonia as Nitrogen Source

The class of 3,3′-diaryl substituted tetranaphthobisazepinium bromides has found wide application as highly efficient C2-symmetrical phase-transfer catalysts (PTCs, Maruoka type catalysts). Unfortunately, the synthesis requires a large number of steps and hampers the build-up of catalyst libraries which are often desired for screening experiments. Here, we present a more economic strategy using dinaphthoazepine 7 as the common key intermediate. Only at this stage various aryl substituents are introduced, and only two individual steps are required to access target structures. This protocol was applied to synthesize ten tetranaphthobisazepinium compounds 1a–1j. Their efficiency as PTCs was tested in the asymmetric substitution of tert-butyl 2-((diphenylmethylene)amino)acetate. Enantioselectivities up to 92% have been observed with new catalysts.


Introduction
In many asymmetric transformations the atropisomeric 1,1'-binaphthyl moiety serves as a highly efficient chiral backbone [1]. The possibility to introduce suitable functional groups, particularly at C-2 and C-2', as well as at C-3 and C-3' has made the binaphthyl group an indispensable chiral modifyer in stoichiometric and catalytic asymmetric synthesis [2][3][4][5][6][7][8]. While functionalities at C-2 and C-2 serve particularly as sites for primary interaction in organocatalysis or coordination in transition metal catalysis, substituents at C-3 and C-3 are introduced for secondary substrate or reagent activation or to tune steric interaction.
With the aim to facilitate the built-up of ligand libraries [27,28] we recently developed a strategy introducing 3,3′-aryl substituents at a late stage of the synthesis, and replacing allylamine with ammonia. Among others spiro-ammonium salts 1a-1c containing electron donating aryl groups were synthesized in good yield (Scheme 1) [29].
In the present paper we further extend this protocol to the introduction of electron withdrawing or more bulky aryl substituents to give amines 9d-9j. Subsequent cyclisation with 2,2'-bis(bromo)methyl-1,1'-binaphthyl afforded seven tetranaphthobisazepinium bromides 1d-1j which were tested in prototypical organocatalyzed reactions. Moreover, an optical resolution procedure of key intermediate 7 is provided.

Results and Discussion
With the intention to introduce various aryl groups at C-3/3′ of the binaphthyl core but at a late stage of the synthesis, diiodoazepine 7 was chosen as key intermediate. Its synthesis started from 1,1'-binaphthyl-2,2'-dicarboxylic acid 2 which is accessible in enantiomerically pure form on the multigram scale from non-racemic 2,2′-dimethyl-1,1′-binaphthyl [30] or 2,2'-binaphthol in two to four steps [31][32][33], respectively or as a racemate applying various procedures from the literature [34]. As we noticed that 1 (R = H, X = Br) can be conveniently prepared from non-racemic 2,2'-bis(bromomethyl)-1,1'-binaphthyl and aqueous ammonia in 94% yield [35], it was easy to apply similar conditions for 3,3'-substituted substrates with the expectation that steric hindrance might stop the reaction at the stage of the secondary amine. (For use of ammonia in the synthesis of symmetrical N-spiroazepine compounds, see [36][37][38].) As precursors 3 and 6 were prepared from 2 in three or four steps (overall yield: 47% for 3, 48% for 6) according to literature [39]. Treatment of dibromides 3 and 6 with aqueous ammonia (25%) in acetonitrile at 60 °C overnight afforded 4 and 7 in 85% and 88% [29], respectively. An alternative approach to 7 via the bis(trimethylsilyl)azepine 4 gave lower yield. An attempted spiro-cyclisation of 4 to 5 failed; only minor amounts of rearranged products could be identified (see Supplementary Materials). Non-racemic 4 and 7 can be prepared from enantiomerically pure precursors (R)-and (S)-2 but requires optical resolution of the diacid. Applying published procedures, this step is quite time consuming and uses toxic or expensive amines as resolving agents, which are often difficult to recycle. Consequently alternatives were With the aim to facilitate the built-up of ligand libraries [27,28] we recently developed a strategy introducing 3,3 -aryl substituents at a late stage of the synthesis, and replacing allylamine with ammonia. Among others spiro-ammonium salts 1a-1c containing electron donating aryl groups were synthesized in good yield (Scheme 1) [29].
In the present paper we further extend this protocol to the introduction of electron withdrawing or more bulky aryl substituents to give amines 9d-9j. Subsequent cyclisation with 2,2'-bis(bromo)methyl-1,1'-binaphthyl afforded seven tetranaphthobisazepinium bromides 1d-1j which were tested in prototypical organocatalyzed reactions. Moreover, an optical resolution procedure of key intermediate 7 is provided.
Optical resolution of 7: A procedure similar to that published for the enantioseparation of the unsubstituted analogue 3,5-dihydro-4H-dinaphth[2,1-c:1',2'-e]azepine was applied [40]. A 2:1 mixture of rac-7 and (S,S)-di-O-benzoyltartaric acid in dichloromethane (DCM)/MeOH-deposited colourless needles upon standing for several hours at r.t. An X-ray structure analyses of a single crystal showed a di-(R)-azepium tartrate with local C 2 -symmetry and one molecule of DCM in the asymmetric unit ( Figure 2). After removal of the auxiliary and one recrystallisation (R)-and (S)-7 were isolated in approximately 40% yield. The enantiomeric purity was determined by chiral high-performance liquid chromatography (HPLC) to be ≥99% (see Experimental section) [41]. basic conditions. Two side products with correct high-resolution mass spectrometry (HRMS) and 1 H-nuclear magnetic resonance (NMR) multiplets (two AB and one ABX systems corresponding to Ar-CH2-N and Ar-CH2-CH(N)-Ar) have been detected in the product mixture (See Supplementary Materials). To lower steric repulsion within the target molecule we cyclized azepines 4 and 9e with 2,2′-bisbromomethyl-1,1′-biphenyl yielding 5′ and 1e' in good yield. Organocatalysis: With this catalyst library in hand we next wanted to create a reactivity/selectivity profile for application in phase-transfer reactions (PTC) under strictly standardised conditions using the well known α-benzylation of tert-butyl 2-((diphenylmethylene)amino)acetate 12 with benzylbromide 13A, one of the most popular For the synthesis of azepinium compounds 1 arylation of 7 followed by reaction with non-racemic 2,2 -bisbromomethyl-1,1 -binaphthyl worked well [42]. A practicable route to target structures 1 might be therefore 2 → 6 → 7 (optical resolution) → 9 → 1. The general feasability was previously testet in the synthesis of 1a-1c [29] and was now extended to spiro ammonium compounds 1d-1j from enantiomerically pure 7. Suzuki-Miyaura reactions with appropriate boronic acids or tetramethyldioxaborolanes afforded 9d, 9e, 9i, and 9j in fair to good yield. Only for electron-withdrawing aryl groups the yields for Suzuki coupling were low (49% and 29% for 9f and 9h, respectively). In those cases N-Boc protected diiodide 8 was a proper intermediate (60%-80% for 9f-9h).
In contrast, coupling of 7 or 8 with ferroceneboronic acid or bisborolane 11 with 1-bromoferrocene failed to give 9k. The final step, the formation of spiro-ammonium compounds 1, proceeded under standard conditions with satisfying yields except for 1e which gave a bad mixture. The target compound was isolated by only 7% after repeated chromatography. We speculate that, similarly to 4, steric strain in 1e might facilitate subsequent Stevens rearrangement under slightly basic conditions. Two side products with correct high-resolution mass spectrometry (HRMS) and 1 H-nuclear magnetic resonance (NMR) multiplets (two AB and one ABX systems corresponding to Ar-CH 2 -N and Ar-CH 2 -CH(N)-Ar) have been detected in the product mixture (See Supplementary Materials). To lower steric repulsion within the target molecule we cyclized azepines 4 and 9e with 2,2 -bisbromomethyl-1,1 -biphenyl yielding 5 and 1e' in good yield.
Organocatalysis: With this catalyst library in hand we next wanted to create a reactivity/selectivity profile for application in phase-transfer reactions (PTC) under strictly standardised conditions using the well known α-benzylation of tert-butyl 2-((diphenylmethylene)amino)acetate 12 with benzylbromide 13A, one of the most popular conversions to test a new PTC (Scheme 2). Further on, with promising catalysts we were also interested to introduce substituents with functional groups to extend the scope for application.
conditions (Table 1) [43,44]. For all experiments we compared "expected yields" based on integration and "isolated yields" after chromatography to demonstrate equivalence of methods (Table 1, row 3 and 4). Typically a loss of 1%-5% of product after chromatography was observed. We attribute this to variing quality of column packing, incomplete separation, and changes in adsorbent activity. In addition, interaction of the substrate and product with silicagel resulting in partial cleavage of the imino group might be responsible for reduced isolated yields after purification [45]. Therefore, we considered reporting of the "chemical yield" by comparison of NMR signals of product and added internal standard more reliable and moreover, time-saving than an "isolated yield". In agreement with previous findings 1b, 1f, and 1g were most efficient [25]. Literature results could be reproduced in most cases, although reported conditions were in part different from our's (see notes in Table 1). Elongation of 3,3′-substituents (1c, 1d, see also X-ray structure Figure 3) did not further increase selectivity or activity, in several cases even reduced the asymmetric induction (entry 3,4). Particularly remarkable was the failure of 1e' yielding an almost racemic product in moderate yield (entry 5). This is also in line with the low ee obtained with 5′ (entry 11). We suspect that two effects might be responsible for destroying the asymmetric induction: (1) the strong predominance of the biphenyl atropomer with a configuration opposite to that of the binaphthyl as it was also found in the X-ray structure of 1e' (Figure 4), overriding eventually even a higher reactivity of the minor species with "homo-chirality"; and (2) the presence of more spherical and not distinctly directed substituents which might disguise the C2 symmetry of the catalyst. Also, the introduction of hetero aromates (1i, 1j) gave lower ee (entry 9,10). In a preliminary study new ammonium salts (S,S)-1a-1j, (S)-1e', and (S)-5 were tested in the asymmetric benzylation of tert-butyl 2-((diphenylmethylene)amino)acetate 12 under standard PTC conditions (Table 1) [43,44]. For all experiments we compared "expected yields" based on integration and "isolated yields" after chromatography to demonstrate equivalence of methods (Table 1, row 3 and 4). Typically a loss of 1%-5% of product after chromatography was observed. We attribute this to variing quality of column packing, incomplete separation, and changes in adsorbent activity. In addition, interaction of the substrate and product with silicagel resulting in partial cleavage of the imino group might be responsible for reduced isolated yields after purification [45]. Therefore, we considered reporting of the "chemical yield" by comparison of NMR signals of product and added internal standard more reliable and moreover, time-saving than an "isolated yield". In agreement with previous findings 1b, 1f, and 1g were most efficient [25]. Literature results could be reproduced in most cases, although reported conditions were in part different from our's (see notes in Table 1). Elongation of 3,3 -substituents (1c, 1d, see also X-ray structure Figure 3) did not further increase selectivity or activity, in several cases even reduced the asymmetric induction (entry 3,4). Particularly remarkable was the failure of 1e' yielding an almost racemic product in moderate yield (entry 5). This is also in line with the low ee obtained with 5 (entry 11). We suspect that two effects might be responsible for destroying the asymmetric induction: (1) the strong predominance of the biphenyl atropomer with a configuration opposite to that of the binaphthyl as it was also found in the X-ray structure of 1e' (Figure 4), overriding eventually even a higher reactivity of the minor species with "homo-chirality"; and (2) the presence of more spherical and not distinctly directed substituents which might disguise the C 2 symmetry of the catalyst. Also, the introduction of hetero aromates (1i, 1j) gave lower ee (entry 9,10).   After this preliminary estimation of reactivity and enantioselectivity of new spiro-ammonium catalysts we choose the more promising candidates 1c and 1d for next investigations and compared their efficiency with known catalysts with different reactivity 1g [25] and 1h [47]. Activated benzylic bromides (13B-F, Scheme 2) were tested under same conditions as in Table 1 to make results comparable (0 • C, 20 h, 1 mol % of catalyst in toluene with 50% KOH, 0.25 mmol scale). In addition, also electrophiles 13G-I with functional groups, eventually of interest for subsequent transformations, were included. As before, in all cases chemical yields based on NMR integration with IS were slightly higher (1%-4%) than those calculated from weighted products, isolated after chromatography (Table 2). Products 14B-F were formed in good to excellent yield and enantioselectivity particularly with 1g. Merely, 1h showed pronounced low reactivity and in two cases unusual poor ee which can be attributed to a significant degree of background reaction. For sterically less demanding electrophiles 13B, 13C, and 13F new ligands 1c and 1d were also effective, very similar to each other and also to the known 3,3 -bis(2-naphthyl) substituted analogue. Using more bulky/less reactive bromides 13D, 13E yields and/or enantioselectivity were lower in some cases. Also a pyridyl substituent could be introduced (14G) with use of catalysts 1c, 1d, 1g, and 1h. In all cases the reaction proceded smoothly giving comparable yield (81%-91%) and up to 83% ee. A more challenging electrophile was 13H which formed only 13% of 14H with 62% ee using the most reactive catalyst 1g under standard conditions. As a side product cyclopropane 15 was produced as a single stereoisomer from a Michael addition followed by cyclisation. To accelerate the reaction KOH was replaced with CsOH. With these conditions complete conversion was achieved yielding 61% of 14H (79% ee) and only 8% of 15. While nearly the same result was obtained with 1h, the enantioselectivtity with 1c and 1d remained low. Also the use of 4-iodocrotonate did not improve the results (for details see Supplementary Materials, Table S2). Finally, we aimed to introduce a N-protected alkenylamino substituent to obtain 14I. In pre-experiments we noticed that use of strongly alkaline media resulted in considerable loss of product through ring opening of the phthalimide. Therefore, solid Cs 2 CO 3 was used instead (33% yield, 88% ee). But even with the more reactive iodide (13I with I replacing Br) the yield was still low (20%-33%) and asymmetric induction moderate (51%-83% ee). Summarising, we presented an alternative route to non-racemic 3,3′-aryl substituted tetranaphtho-spiro-bisazepinium bromides as demonstrated in the synthesis of 1a-1j in seven to eight steps via 3,3′-substituted dinaphthoazepines 9a-9j [48] using 1,1′-binaphthyl-2,2′-dicarboxylic acid as starting material and 3,3′-diiodo-dinaphthoazepine 7 as the common key intermediate. This scalable synthesis required only two chromatographic purification steps and target structures have been isolated in 23%-25% overall yield. A preliminary study on their use in PTC was conducted using a simplified screening protocol. Moderate to good asymmetric induction of 41%-92% ee has been observed with new catalysts.   Summarising, we presented an alternative route to non-racemic 3,3 -aryl substituted tetranaphtho-spiro-bisazepinium bromides as demonstrated in the synthesis of 1a-1j in seven to eight steps via 3,3 -substituted dinaphthoazepines 9a-9j [48] using 1,1 -binaphthyl-2,2 -dicarboxylic acid as starting material and 3,3 -diiodo-dinaphthoazepine 7 as the common key intermediate. This scalable synthesis required only two chromatographic purification steps and target structures have been isolated in 23%-25% overall yield. A preliminary study on their use in PTC was conducted using a simplified screening protocol. Moderate to good asymmetric induction of 41%-92% ee has been observed with new catalysts.
Heptane, dichloromethane (DCM), and ethyl acetate (EtOAc) were distilled, absolute THF from sodium benzophenone ketyl, Et 2 O from LiAlH 4 ; acetonitrile, DCM, and triethylamine from CaH 2 ; n-BuLi was used as 1.6 M solution in n-hexane (Aldrich). All the other chemicals were analytical grade and used without further purification.
Suzuki-Miyaura coupling of (S)-8yielding (S)-9 (General Procedure C): A Schlenk tube, equipped with magnetic stirring bar, was charged with a solution of boc-protected diiodoazepine 8 (97 mg, 0.15 mmol) and tri-ortho-tolylphosphine (9.1 mg, 20 mol %) in toluene (3 mL) and a Na 2 CO 3 solution (2 M in H 2 O, 2 mL). Then, arylboronic acid (4 equiv.) was added and the mixture was degassed. After the addition of Pd(OAc) 2 (3.4 mg, 10 mol %), the mixture was left stirring at 80 • C for 12-48 h. The reaction was monitored by TLC (EtOAc/heptane, 30:70). After cooling to r.t., DCM (5 mL) and H 2 O (3 mL) were added and the phases were separated. The aqueous phase was extracted with DCM (3 × 3 mL) and the combined organic phases were washed with KOH solution (10%, 5 mL) and sat. NaCl solution and dried (Na 2 SO 4 ). After evaporation of solvents DCM (1 mL) and trifluoroacetic acid (1 mL) was added and the solution stirred at r.t. for 2 h. The reaction was carfully neutralized with solid NaHCO 3 . Extractive work-up was followed by MPLC.