Economy of Catalyst Synthesis—Convenient Access to Libraries of Di- and Tetranaphtho Azepinium Compounds

Efficient optimization procedures in chiral catalysis are usually linked to a straightforward strategy to access groups of structurally similar catalysts required for fine-tuning. The ease of building up such ligand libraries can be increased when the structure-modifying step (introduction of a substituent) is done at a later stage of the synthesis. This is demonstrated for the extended family of di- and tetranaphtho azepinium compounds, widely used as chiral phase transfer catalysts (PTC). Using 2,6-diiodo-4,5-dihydro-3H-dinaphtho[2,1-c:1′,2′-e]azepine and 4,8-diiodo-6,7-dihydro-5H-dibenzo[c,e]azepine, respectively, as key intermediates, 18 spiro-azepinium compounds were synthesized in a total yield of 25–42% over 6–7 steps from 1,1′-binaphthyl-2,2′-dicarboxylic acid or diphenic acid, respectively. The replacement of iodo groups with aryl substituents was performed as the last or the penultimate step of the synthesis.

S3 cleave diastereomeric amides. 2 In method g3, finally, the preferred formed 1:2 salt crystallizes and was separated from the mother liquor. The resolving agent was recovered in good yield but following this protocol only one enantiomer of 6 was obtained. 3 The yields of all methods are typically in a range of 40% for each enantiomer.
An interesting report was published which significantly shortens the synthesis. The oxidation of 1-bromo-2-methylnaphthalene (21) with O2 catalysed by Co(OAc)2 giving 24 (step m) is conducted in a steel autoclave 3 and substitutes three steps b-d. The apparently easy operation without purification and good yield (87%) on a large scale (482 mmol) makes this protocol very attractive saving time and man power (same overall yield as b-d within 1%). Merely, the requirement of a 1L-autoclave which might be not generally available is unfavourable.
Finally, a two step sequences from 29 to 6 might be considered as well. In an early report the bistriflate 30 was methoxycarbonylated under Pd(II)/dppp catalysis to afford the dimethylester of 6 in 83% yield. 18 The use of CO and requirement of noble metal catalysis obviously hampered upscaling and broad use of this protocol. Two other processes working on gram scale were recently reported. After transformation of 29 to 2,2'-diethylphosphate 35 (quant. yield) this was treated with Li-naphthalenide at -78 °C to give the di-lithio compound which reacted with CO2 to afford 6 in up to 89% yield (3.6 mmol scale). The need of a column chromatography to purify 6 makes up-scaling more difficult (5 mmol scale reported). 19 In an other report triflate 30 was converted to diphenylester 36 using phenylformiate as CO source and Pd(OAc)2/DPPP as catalyst which was followed by hydrolysis to afford 6. 20 It is worth noting that both processes can be performed stereoconservative, i.e. without racemisation.
Summarizing, for multigram synthesis of 1,1'-binaphthyl-2,2'-dicarboxylic acid (6) two comparable routes are available, starting from either 2-methylnaphthalene (20) (Route A, Scheme S1) or 2,2'-dihydroxy-1,1'-binaphthyl (29) (route B, Scheme S2). Preference will be given depending on the need of racemic or non-racemic material. In the first case Route A is more convenient and requires 4 steps (if reaction m can be performed) or 6 steps with an overall yield of 66-67%. If at this stage an optical resolution is performed the yield will drop to 25-26% for each enantiomer of 6. In this case Route B is superior yielding 63% of non-racemic 6. In contrast, the asymmetric biaryl coupling (Scheme S2, v) requiring expensive catalysts and long reaction time is less appropriate particularly for large scale

S5
preparations. An evaluation of both routes based on time and manpower requirement is rather difficult as the reported time for each step in Table S1 is a rough estimate on the published procedures and do not include preparation/drying/evaporation of solvents. Nevertheless, for the preparation of 5-10 g of 6 an approximate time frame with 10-12 days for Route A and 7-9 days more for optical resolution (g2), and 10-11 days for route B will be a valid approximation.
Comments on Table S1 a: While the bromination of 2-methylnaphthalene (20) with Br2 in CS2 yields up to 91% of 21 after distillation, we found the use of HBr/H2O2 a more convenient method which could be upscaled to 0.5 mol yielding 95% of the desired product without purification (> 98%, NMR) and sufficiently pure for the next step.  f: Many binaphthyl coupling methods are known but from the practical point of view the classical Ullmann coupling in DMF is still attractive due to simplicity of the procedure, easy work-up and good yields. Copper powder was activated by treatment with EDTA solution. 21 The crude dimethyl 1,1'binaphthalene-2,2'-dicarboxylate was immediately hydrolysed and after extractive purification is sufficiently pure. g: At this stage an optical resolution may be performed.
h-k: These steps were already published for enantiomerically pure substrates and largely omit chromatographic purification. Only for step k the mother liquor from the crystallisation was chromatographed. Repetition with racemic substrate gave comparable yields (±2%).
i: Reaction with aqueous ammonia yielded exclusively the secondary amine 8, provided the reaction temperature was kept at 60 °C. No tertiary amine or spiro-ammonium compound was detected.

X-ray Analysis
Experimental data and CCDC-Codes can be found in Table S2. Crystal data, data collection parameters, and structure refinement details are given in Tables S3 to S10. Crystal structures visualized in Figure S1 to S4.  Figure S1: Crystal structure of 3a, drawn with 50% displacement ellipsoids. The asymmetric unit is built up by 1 and 2*1/2 independent molecules of 3a. The 2*1/2 molecules, one counter ion and CHCl3 molecules are omitted for clarity. All three moieties form the same chiral arrangement. The centrosymmetric space group forces the inverse chiral form. Four voids with each 451.2 Å 3 (9.6% of unit cell) had to be excluded from refinement. The corresponding value of electrons is 109.5 each. We could not find satisfactory positions for solvent atoms.