Molybdenum-Catalyzed Enantioselective Ring-Closing Metathesis/Kinetic Resolution of Racemic Planar-Chiral 1,1′-Diallylferrocenes

Abstract

The molybdenum-catalyzed enantioselective ring-closing metathesis/kinetic resolution of a series of racemic planar-chiral 1,1′-diallylferrocene derivatives was reinvestigated utilizing the method of generating catalytically active chiral molybdenum-alkylidene species in situ, which allowed us to examine a variety of chiral molybdenum-alkylidene metathesis precatalysts in the present asymmetric reaction. With the catalyst screening experiments conducted in this study, the more practical reaction conditions, including a choice of a proper chiral molybdenum precatalyst, giving planar-chiral ferrocenes of higher enantiomeric purity and better chemoselectivity could be optimized.

1. Introduction

The introduction of two (or more) different substituents in a single η⁵-cyclopentadienyl ligand in ferrocene breaks the symmetry of the molecule, and so-called planar chirality is induced in it. Planar-chiral ferrocene derivatives have been recognized as being useful chiral scaffolds in organic and organometallic chemistry (Figure 1) [1,2,3,4,5,6,7,8], and various such complexes have been utilized in a wide range of asymmetric reactions as chiral ligands [9,10,11,12,13,14,15,16,17,18,19] or chiral catalysts [19,20,21,22,23,24,25,26]. In spite of their usefulness, enantioselective preparation of planar-chiral ferrocenes is still a challenging problem. Classical methods of obtaining enantiomerically pure (or enantiomerically enriched) planar-chiral ferrocene derivatives are the enantiomeric resolution of preformed racemates [27], including enzymatic resolution [28,29,30], and diastereoselective metalation utilizing chiral ortho-directing groups [31,32,33,34,35,36,37,38,39,40]. To the best of our knowledge, the first catalytic enantioselective synthesis of planar-chiral ferrocenes was reported in 1997 by Schmaltz and Siegel [41]. In 2006, three research groups (O’Brien [42], Moyano [43], and ourselves [44]) independently reported the catalytic asymmetric reactions giving enantiomerically enriched planar-chiral ferrocene derivatives. Since then, nearly 100 research works related to this topic have been published worldwide [45,46,47,48].

Since 2002, our research group has been interested in utilizing the ring-closing metathesis (RCM) reactions for the modulation of various transition-metal complexes [49,50,51,52,53,54,55] using the Schrock’s Mo- [56,57] and the Grubbs’ Ru-alkylidene [58,59,60,61] complexes. The RCM protocols were extended to the asymmetric counterparts to show the excellent enantioselectivity in the asymmetric synthesis of diverse planar-chiral transition-metal complexes either by the kinetic resolution (KR) of the racemic substrates [44,48,62,63,64] or by the desymmetrization of the C_s-symmetric substrates [48,65,66,67].

The chiral catalysts employed in our asymmetric reactions, which produced enantiomerically enriched planar-chiral transition-metal complexes, were chiral Schrock–Hoveyda molybdenum-alkylidene precatalysts [68,69,70,71] (Figure 2). At the beginning of the development of these chiral precatalysts, each chiral molybdenum complex needed to be prepared and isolated one by one prior to the catalytic applications, which made the screening of the chiral precatalysts/diolate ligands tedious and time-consuming. In 2006, Schrock and Hoveyda reported molybdenum complex A, which served as a universal precursor to generate a variety of chiral molybdenum-alkylidene precatalysts in situ by means of a reaction with an appropriate chiral diol (Scheme 1) [72]. The development of this method enabled rapid and operationally simple screening of the various chiral molybdenum-alkylidene precatalysts in asymmetric olefin metathesis reactions.

Whereas our very first contribution to catalytic asymmetric synthesis of planar-chiral transition-metal complexes, which was the enantioselective ring-closing metathesis/kinetic resolution of racemic planar-chiral 1,1′-diallylferrocene derivatives 1 (Scheme 2), was reported prior to the development of complex A, only one molybdenum-alkylidene species, Mo*/(R)-L1, was examined as a chiral precatalyst in the original publication [44]. In this article, various chiral Mo-alkylidene species were generated in situ, as shown in Scheme 1, and applied in the enantioselective RCM/KR reaction of racemic 1. After the extensive screening of the chiral Schrock–Hoveyda metathesis precatalysts, the more practical conditions giving planar-chiral ferrocenes of higher enantiomeric purity and better chemoselectivity could be determined. Here, we would like to describe the details of our observations.

2. Results and Discussion

2.1. Design and Preparation of Racemic Planar-Chiral 1,1′-Diallylferrocene Substrates 1a-c for the Molybdenum-Catalyzed Enantioselective Ring-Closing Metathesis/Kinetic Resolution

Our preliminary studies on the enantioselective RCM/KR of racemic planar-chiral 1,1′-diallylferrocene derivatives 1 [44] postulated that a couple of structural factors in the substrates were crucial to achieve the high enantioselectivity in the molybdenum-catalyzed reactions: (i) steric discrimination of the two allylic groups at the 1- and 1’-positions of the ferrocene core with a methyl substituent in the 2-allylic position of one of the two allylic substituents, and (ii) a bulky substituent R’ in the position adjacent to the unsubstituted (i.e., the more reactive) allyl group (Figure 3).

The preparation of the designed unsymmetric ferrocene derivatives rac-1a-c was achieved by the method developed by Manriquez et al., as shown in Scheme 3 [73]. Allylation of lithium 1,3-R’₂-cyclopentadienide with allyl bromide, followed by a deprotonation reaction with butyllithium, provided a THF solution of lithium 1-allyl-2,4-R’₂-cyclopentadienide (B). The allylation was highly regioselective, and the formation of a regioisomer, lithium 2-allyl-1,3-R’₂-cyclopentadienide, was negligible. A reaction of Fe(acac)₂ with stoichiometric B at –78 °C generated metastable intermediate C as a THF solution, and a subsequent reaction with sodium methallylcyclopentadienide (D) afforded rac-1a-c in the yields ranging 20–45%. The reaction conditions were not optimized. The reaction giving rac-1a-c was not completely hetero-selective, and the formation of homoleptic ferrocene byproducts, (η⁵-methallyl-C₅H₄)₂Fe and (η⁵-1-allyl-2,4-R’₂-C₅H₂)₂Fe, could not be eliminated. The target compounds could be separated from the byproducts by the standard column chromatography on alumina.

The ferrocene substrates for this study, rac-1a-c, possess a trisubstituted cyclopentadienide, η⁵-(C₅H₂-1-allyl-2,4-R’₂), which is responsible for inducing planar chirality in ferrocene compounds, and the monosubstituted η⁵-cyclopentadienyl ligand η⁵-(C₅H₄-methallyl). In the presence of an appropriate chiral (R)-molybdenum-alkylidene metathesis catalyst, one of the two planar-chiral enantiomers in 1 is preferentially cyclized to give enantiomerically enriched bridged ferrocene (ferrocenophane) (R)-2, and antipodal (S)-1 is left intact. The major side reaction of this RCM/KR process is the formation of 3, which is a product of the metathesis dimerization at the unsubstituted allyl group in the trisubstituted cyclopentadienide (Scheme 2).

For the highly enantioselective kinetic resolution of rac-1a-c, the allyl group in a trisubstituted cyclopentadienyl needs to be more reactive than the methallyl (2-methylallyl) substituent in a monosubstituted cyclopentadienyl. In general, less substituted olefins are more reactive than more substituted ones in olefin metathesis. When a chiral molybdenum-alkylidene species, (R)-Mo*, approaches rac-1, an initial reaction takes place preferentially at the allyl group in the planar-chiral trisubstituted Cp to give intermediates (R_a,R_p)- and (R_a,S_p)-E as a diastereomeric mixture (“R_a” represents the absolute configuration of the axially chiral biaryl moiety in Mo^*), and the formation of F is unfavorable. While (R_a,R_p)-E is transformed into (R)-2 smoothly via the RCM reaction, the epimeric intermediate, (R_a,S_p)-E, is forced to take an unfavorable conformation due to the steric repulsion between an R’ group in η⁵-C₅H₂R’₂(allyl) and an R group in the chiral biaryloxide ligand in (R)-Mo* to liberate (S)-1 intact (Scheme 4). On the other hand, the unfavorable formation of intermediate F is crucial for the high enantioselectivity in the RCM/KR process. Whereas the methallyl group in η⁵-C₅H₄-methallyl is remote from the planar-chiral η⁵-C₅H₂R’₂(allyl) moiety, its reaction with (R)-Mo* takes place with low diastereoselectivity. Following the RCM step in F is an intramolecular process, and, thus, both (R_a,R_p)- and (R_a,S_p)-F shall be transformed into the corresponding ferrocenophanes to give 2 with low enantioselection.

2.2. Molybdenum-Catalyzed Enantioselective Ring-Closing Metathesis/Kinetic Resolution of Racemic Planar-Chiral 1,1′-Diallylferroce Substrates 1a-c: Catalyst Screening Studies

At the outset, the screening of precatalysts/reaction conditions was examined for the enantioselective RCM/KR reaction of rac-1a. Precatalyst Mo*/(R)-L1 (10 mol % to rac-1a), which was generated in situ as outlined in Scheme 1, showed a nearly identical performance (selectivity and reactivity) to the preformed catalyst [44] in the reaction (

Table 1. Molybdenum-catalyzed enantioselective ring-closing metathesis/kinetic resolution of racemic planar-chiral ferrocenes rac-1a-c ^a.

Entry	Substrate b	Ligand	Conditions	Yields (%) of 1/2/3 c	% ee of (S)-1/(R)-2 d,e	krel f
1	1a (0.005)	(R)-L1	50 °C, 24 h	46/52/2	99/91	109
2	1a (0.005)	(R)-L2	50 °C, 24 h	45/55/0	97/88	65
3	1a (0.005)	(R)-L2	25 °C, 24 h	53/47/0	67/96	95
4 g	1a (0.05)	(R)-L2	25 °C, 1 h	55/45/0	75/96	110
5	1a (0.005)	(R)-L3	50 °C, 24 h	35/45/20	94/91	75
6	1a (0.005)	(R)-L4	50 °C, 24 h	0/100/0	--- h/--- h	--- h
7	1a (0.005)	(R)-L4	25 °C, 24 h	12/88/0	99/18	5.5
8	1b (0.005)	(R)-L1	50 °C, 24 h	51/38/11	89/95	117
9	1b (0.005)	(R)-L2	50 °C, 24 h	45/55/0	92/88	51
10	1b (0.005)	(R)-L2	25 °C, 24 h	45/55/0	97/91	89
11 g	1b (0.05)	(R)-L2	25 °C, 2 h	53/47/0	74/97	146
12	1b (0.005)	(R)-L3	50 °C, 24 h	44/52/4	91/92	76
13	1b (0.005)	(R)-L4	50 °C, 24 h	0/100/0	--- h/--- h	--- h
14	1b (0.005)	(R)-L4	25 °C, 24 h	20/80/0	90/26	4.4
15	1c (0.005)	(R)-L1	50 °C, 24 h	28/52/20	98/54	15
16	1c (0.005)	(R)-L2	50 °C, 24 h	1/99/0	--- h/5	--- h
17 g	1c (0.05)	(R)-L2	25 °C, 1 h	45/55/0	84/59	9.9

^a The reaction was carried out with rac-1 (0.10 mmol) in benzene using a molybdenum catalyst generated in situ (10 mol %), unless otherwise noted. ^b Initial concentration of substrate 1 in parentheses. ^c Determined via the ¹H-NMR analysis of the crude reaction mixture. ^d Determined by chiral HPLC analysis (see Materials and Methods Section for detail). ^e Enantiomeric excess of recovered 1a-c was determined after converting them into the corresponding 2a-c via the RCM reaction using the Grubbs-II catalyst. ^f Calculated based on a first-order equation [74,75]. ^g The reaction was carried out in C₆D₆. ^h Not determined.

, Entry 1). Although the enantioselectivity of this reaction was quite high, with a k_rel value of 109 (k_rel = ([reaction rate of the fast-reacting enantiomer]/[reaction rate of the slow-reacting enantiomer]; selectivity factor) [74,75], the observed drawbacks of using Mo*/(R)-L1 were (i) the competition between the RCM reaction giving the desired ferrocenophane 2a and the bimolecular metathesis giving side-product 3a, (ii) the necessity of the high-dilution conditions for minimizing the formation of metathesis dimer 3a, and (iii) the relatively high reaction temperature (50 °C) to retain the reasonable catalytic activity of the molybdenum species under the diluted conditions. It was found that the reaction using Mo*/(R)-L2 (10 mol % to rac-1a) promoted the intramolecular RCM reaction giving (R)-2 preferentially, and the formation of 3 via the intermolecular reaction was not observed (Entry 2). The enantioselectivity was reasonably high (k_rel = 65) but slightly lower than that in Entry 1. Precatalyst Mo*/(R)-L2 was catalytically more active than Mo*/(R)-L1 in the reaction of rac-1a, and, thus, the reaction could be conducted at a lower temperature. Enantioselectivity was improved to k_rel = 95 at 25 °C (Entry 3). Since Mo*/(R)-L2 did not drive the bimolecular reaction giving 3, the RCM/KR reaction catalyzed by Mo*/(R)-L2 could be conducted under the more concentrated conditions, which realized the shorter reaction time whilst retaining a high enantioselectivity. Under these conditions, RCM product (R)-2a of 96% ee was obtained in a 45% yield, and unreacted (S)-1a of 75% ee was recovered in 55%, of which k_rel was 110 (Entry 4). It is worth mentioning that the “concentrated” reaction could be carried out in an NMR sample tube using C₆D₆ as a solvent, which allowed for the direct monitoring of the reaction progress through ¹H-NMR measurements. The catalytic performance of Mo*/(R)-L3 was similar to that of Mo*/(R)-L1 (Entry 5). Precatalyst Mo*/(R)-L4 was the most reactive among the precatalysts examined but far less enantioselective. The reaction of rac-1a catalyzed by Mo*/(R)-L4 at 50 °C for 24 h was leading to the complete consumption of the substrate to provide nearly racemic 2a quantitatively (Entry 6). The reaction at 25 °C realized the kinetic resolution of rac-1a, but its enantioselectivity was far less satisfactory (k_rel = 5.5; Entry 7).

The trends of the molybdenum-catalyzed enantioselective RCM/KR reactions of rac-1b were similar to those of rac-1a. Precatalysts Mo*/(R)-L1 and Mo*/(R)-L3 showed fairly high enantioselectivity (k_rel = 117 and 76, respectively) but with unsatisfactory chemoselection, producing a considerable amount of dimeric 3b (Entries 8 and 12). On the other hand, Mo*/(R)-L2 was more reactive in the RCM/KR of rac-1b and did not catalyze the dimerization reaction (Entries 9-11). The reactivity of Mo*/(R)-L4 was too high, showing low enantioselectivity (Entries 13 and 14). The best result in the reaction of rac-1b was obtained using Mo*/(R)-L2 at the lower temperature (25 °C) under the concentrated conditions in C₆D₆ to give (R)-2b (97% ee, 47% yield) and recover (S)-1b (74% ee, 53%). The k_rel value of this reaction was estimated to be 146 (Entry 11).

Due to the sterically less-demanding cyclohexyl substituents in 1c (compared to ^tBu in 1a and SiMe₃ in 1b), the RCM/KR reaction of rac-1c was generally less enantioselective (Entries 15–17). The optimized reaction conditions, as in Entries 4 and 11, were applied to the reaction of rac-1c, and the reasonably good enantioselectivity of k_rel = 9.9 was achieved as well (Entry 17). It should be noted that the reactions of rac-1c under Mo*/(R)-L2 catalysis did not produce the undesirable metathesis dimer 3c (Entries 16 and 17), as expected.

Next, the conditions optimized for the reaction of rac-1a-c were applied in the enantioselective RCM/KR reaction of substrate rac-1d, in which the methallyl group in rac-1a was replaced with a prenyl (3,3-dimethylallyl) substituent. It had been reported previously that the reaction of rac-1d in the presence of catalytic Mo^*/(R)-L1 provided dimeric 3d in a 38% yield as a sole metathesis product and that bridged 2d was not detected. The unreacted substrate was recovered in 54%, of which the enantiopurity was as low as (S)-7% ee (Scheme 5, top) [44]. On the other hand, the reaction of rac-1d, as in Entry 4 in Table 1, afforded ferrocenophane (R)-2d of 58% ee in a 49% yield together with recovered (S)-1d of 56% ee in a 50% yield. The k_rel value for this reaction was determined to be 6.5 (Scheme 5, bottom).

3. Materials and Methods

3.1. General Information

All air- and/or moisture-sensitive reactions were conducted with standard Schlenk techniques under pre-dried nitrogen or with glovebox techniques under pre-purified argon. ¹H NMR (at 400 MHz) and ¹³C NMR (at 100 MHz) chemical shifts were reported in ppm downfield of internal tetramethylsilane. Tetrahydrofuran was distilled from sodium benzophenone-ketyl under dry nitrogen prior to use. Benzene and C₆D₆ were dried over a Na/K alloy and distilled and deoxygenated under high-vacuum conditions prior to use. Chloroform-d was distilled and deoxygenated from P₂O₅ under high-vacuum conditions and stored in a glovebox. C₅H₄^tBu₂ [76], C₅H₄(SiMe₃)₂ [77], C₅H₄Cy₂ [78], (pyrrolyl)₂Mo(=CHCMe₂Ph)(=N-C₆H₃-2,6-ⁱPr₂) [72], (R)-L1 [69], (S)-L3 [68], (R)-L4 [71], and the Grubbs-II catalyst [79,80] were prepared as reported. All the other chemicals were purchased from commercial suppliers and used without further purification, unless otherwise noted.

3.2. Preparation of Racemic Diallylferrocene Substrates rac-1a-d [44]

A typical procedure is hereby given for the synthesis of rac-1a. To a THF (8 mL) solution of Fe(acac)₂ (2.54 g, 10.0 mmol) was added a solution of lithium 1-allyl-2,4-^tBu₂-cyclopentadienide, which was prepared from C₅H₃(allyl)^tBu₂ (2.18 g, 10.0 mmol) and ⁿBuLi (1.60 M hexane solution, 6.3 mL, 10.1 mmol) in THF (25 mL), at –78 °C, and the mixture was stirred at 0 °C for 1h. After cooling the mixture to −78 °C, to this was added a solution of sodium methallylcyclopentadienide, which was prepared from C₅H₅-methallyl (1.06 g, 10.0 mmol) and NaH (240 mg, 10.0 mmol) in THF (10 mL). The resulting mixture was stirred at room temperature for 3 h. The mixture was diluted with hexane and filtered through a pad of Celite. After removal of the solvent under reduced pressure, the remaining dark-red oil was purified by column chromatography on alumina using hexane as an eluent, and following vacuum transfer gave rac-1a as a dark-red oil. The reaction conditions were not optimized. The characterization data of the diallylferrocene substrates rac-1a-d are given below.

3.3. Characterization Data of Racemic 1,1′-Diallylferrocene Substrates 1a-d [44]

rac-1-Allyl-1′-(2-methylallyl)-2,4-di(tert-butyl)ferrocene (1a). Yield: 45%. ¹H NMR (CDCl₃): δ 5.88–5.97 (m, 1H), 5.03 (d, J = 4.8 Hz, 1H), 4.99 (s, 1H), 4.62 (s, 1H), 4.59 (s, 1H), 4.13 (br, 1H), 4.08 (br, 1H), 3.93 (br, 2H), 3.76 (br, 1H), 3.72 (br, 1H), 3.22 (dd, J = 15.8 and 6.7 Hz, 1H), 3.07–3.12 (m, 1H), 3.01 (br, 2H), 1.65 (s, 3H), 1.27 (s, 9H), 1.18 (s, 9H). ³C{¹H} NMR (CDCl₃): δ 146.5, 138.2, 115.0, 110.8, 98.8, 96.8, 86.0, 81.7, 71.2, 70.1, 68.8, 68.1, 67.7, 64.1, 38.5, 34.3, 32.4, 31.9, 31.5, 30.5, 22.2. Anal. Calcd for C₂₅H₃₆Fe: C, 76.52; H, 9.25. Found: C, 76.30; H, 9.03. HRMS Calcd for C₂₅H₃₆Fe: 392.2165. Found: 392.2165.

rac-1-Allyl-1′-(2-methylallyl)-2,4-bis(trimethylsilyl)ferrocene (1b). Yield: 20%. ¹H NMR (CDCl₃): δ 5.87–5.97 (m, 1H), 5.01 (d, J = 3.9 Hz, 1H), 4.98 (s, 1H), 4.62 (s, 1H), 4.58 (s, 1H), 4.07 (s, 1H), 4.02 (s, 1H), 3.96 (s, 1H), 3.90 (s, 1H), 3.88 (s, 1H), 3.79 (s, 1H), 3.13 (d, J = 6.2 Hz, 2H), 3.00 (br, 2H), 1.64 (s, 3H), 0.26 (s, 9H), 0.22 (s, 9H). ¹³C{¹H} NMR (CDCl₃): δ 146.2, 138.2, 115.0, 110.3, 94.4, 86.5, 79.4, 78.2, 74.0, 73.3, 70.8, 69.7, 68.4, 68.1, 38.4, 34.2, 22.2, 0.6, 0.1. Anal. Calcd for C₂₃H₃₆FeSi₂: C, 65.07; H, 8.55. Found: C, 65.13; H, 8.43. HRMS Calcd for C₂₃H₃₆FeSi₂: 424.1703. Found: 424.1702.

rac-1-Allyl-1′-(2-methylallyl)-2,4-dicyclohexylferrocene (1c). Yield: 30%. ¹H NMR (CDCl₃): δ 5.91–6.01 (m, 1H), 4.99–5.06 (m, 2H), 4.63 (br, 1H), 4.58 (br, 1H), 3.77–3.97 (m, 6H), 2.94–3.09 (m, 4H), 2.13–2.25 (m, 3H), 1.84-1.93 (m, 3H), 1.66 (s, 3H), 1.61–1.78 (m, 6H), 1.15–1.40 (m, 9H), 0.91 (qd, J = 12.3 and 2.4 Hz, 1H). ¹³C{¹H} NMR (CDCl₃): δ 146.5, 138.0, 115.0, 110.1, 92.9, 92.5, 86.2, 82.9, 71.0, 70.5, 69.1, 69.0, 66.9, 64.0, 38.1, 37.4, 36.4, 36.3, 34.4, 34.3, 32.29, 32.26, 27.0, 26.9, 26.8 (2C), 26.6, 26.5, 22.3. Anal. Calcd for C₂₉H₄₀Fe: C, 78.36; H, 9.07. Found: C, 78.47; H, 9.20. HRMS Calcd for C₂₉H₄₀Fe: 444.2477. Found: 444.2484.

rac-1-Allyl-1′-(3-methyl-2-butenyl)-2,4-di(tert-butyl)ferrocene (1d). Yield: 33%. ¹H NMR (CDCl₃): δ 5.96–5.86 (m, 1H), 5.24–5.21 (m, 1H), 5.01 (d, J = 4.8 Hz, 1H), 4.98 (br, 1H), 4.15 (br, 1H), 4.09 (br, 1H), 3.95 (br, 1H), 3.92 (br, 1H), 3.83 (br, 1H), 3.79 (br, 1H), 3.19 (dd, J = 16.4 and 7.2 Hz, 1H), 3.07–2.99 (m, 3H), 1.68 (s, 3H), 1.65 (s, 3H), 1.25 (s, 9H), 1.16 (s, 9H). ¹³C{¹H} NMR (CDCl₃): δ 138.4, 131.1, 124.3, 114.9, 98.7, 96.7, 88.4, 81.7, 70.2, 69.1, 68.4, 67.9, 67.6, 63.8, 34.3, 32.4, 31.9, 31.5, 30.5, 27.8, 25.7, 17.8. Anal. Calcd for C₂₆H₃₈Fe: C, 76.84; H, 9.42. Found: C, 76.70; H, 9.59. HRMS Calcd for C₂₆H₃₈Fe: 406.2321. Found: 406.2328.

3.4. General Procedure for Molybdenum-Catalyzed Enantioselective RCM/Kinetic Resolution of rac-1

The detailed reaction conditions are summarized in Table 1. A mixture of Mo(=NC₆H₃-2,6-ⁱPr₂)(=CHCMe₂Ph)(NC₄H₄)₂ (5.4 mg, 10 μmol) and an appropriate chiral ligand L (11 μmol) were placed in a test tube (with a Teflon-sealed screw cap) and dissolved in dry benzene (3.0 mL) in a glovebox under pre-purified argon. The mixture was stirred for 15 min at room temperature, and then to this was added a solution of substrate rac-1 (0.10 mmol) in benzene (17.0 mL). The sealed test tube was taken out of the glovebox and was immersed in an oil bath maintained at 50 °C or 25 °C. After stirring the mixture for 24 h, the reaction was quenched by the addition of acetone (ca. 100 μL). The reaction mixture was passed through a short pad of silica gel (eluent: hexane/Et₂O = 9/1). The volatiles were removed under reduced pressure, and the conversion of the reaction was determined by the ¹H-NMR measurement of the crude residue. The residue was purified by preparative HPLC [LC-908 recycle HPLC system (Japan Analytical Industry Co., Ltd., Tokyo, Japan) with a GPC column (JAIGEL-H, chloroform, 3.5 mL/min)] to provide RCM product 2 and recovered substrate 1, respectively. Recovered unreacted 1 was treated with Grubbs-II catalyst (5 mol %) in benzene to give the corresponding 2 quantitatively, which was used for a chiral HPLC analysis. The absolute configurations of ferrocenophanes 2 as well as recovered unreacted substrates 1 were determined through the comparison of their signs of the specific rotations with those of the known compounds [44]. The characterization data of the RCM products and the conditions for the chiral HPLC analysis are listed below.

3.5. Characterization Data of Ferrocenophanes 2a-d [44]

1,1′-(3-Methyl-2-buten-1,4-diyl)-2,4-di(tert-butyl)ferrocene (2a). ¹H NMR (CDCl₃): δ 5.67 (t, J = 7.4 Hz, 1H), 4.14 (s, 1H), 4.02 (s, 1H), 3.99 (s, 1H), 3.89 (s, 1H), 3.82 (s, 1H), 3.73 (s, 1H), 3.28 (dd, J = 14.7 and 7.4 Hz, 1H), 3.12 (d, J = 14.7 Hz, 1H), 2.59-2.67 (m, 2H), 1.93 (s, 3H), 1.28 (s, 9H), 1.14 (s, 9H). ¹³C{¹H} NMR (CDCl₃): δ 138.1, 124.1, 99.9, 98.3, 85.3, 83.1, 70.6, 70.3, 66.8, 66.1, 66.0, 63.6, 32.6, 32.1, 31.4, 30.4, 29.4, 26.7, 24.8. Anal. Calcd for C₂₃H₃₂Fe: C, 75.82; H, 8.85. Found: C, 76.08; H, 8.91. HRMS Calcd for C₂₃H₃₂Fe: 364.1852. Found: 364.1856. [α]³⁰_D = +22 (c 0.50, CHCl₃ for the sample of (R)-96% ee). Chiral HPLC Analysis Conditions: Chiralcel OD-H; eluent: hexane/ⁱPrOH = 2000/1; flow rate: 1.0 mL/min; t₁ = 17.7 min (R-isomer), t₂ = 20.6 min (S-isomer).

1,1′-(3-Methyl-2-buten-1,4-diyl)-2,4-bis(trimethylsilyl)ferrocene (2b). ¹H NMR (CDCl₃): δ 5.76 (t, J = 7.8 Hz, 1H), 4.10 (s, 1H), 4.02 (s, 1H), 3.97 (s, 1H), 3.92 (s, 1H), 3.87 (s, 1H), 3.84 (s, 1H), 3.10 (d, J = 14.4 Hz, 1H), 3.02 (dd, J = 14.8 and 7.3 Hz, 1H), 2.75 (dd, J = 14.8 and 8.2 Hz, 1H), 2.69 (d, J = 14.4 Hz, 1H), 1.94 (s, 3H), 0.27 (s, 9H), 0.17 (s, 9H). ¹³C{¹H} NMR (CDCl₃): δ 138.1, 124.1, 96.3, 86.1, 78.5, 76.2, 74.8, 74.7, 70.1, 69.7, 66.5, 66.4, 29.4, 26.7, 25.0, 0.9, −0.1. Anal. Calcd for C₂₁H₃₂FeSi₂: C, 63.61; H, 8.13. Found: C, 63.46; H, 8.04. HRMS Calcd for C₂₁H₃₂FeSi₂: 396.1390. Found: 396.1395. [α]³¹_D = +50 (c 2.0, CHCl₃ for the sample of (R)-97% ee). Chiral HPLC Analysis Conditions: Chiralcel OD-H; eluent: hexane/ⁱPrOH = 2000/1; flow rate: 1.0 mL/min; t₁ = 15.5 min (R-isomer), t₂ = 16.5 min (S-isomer).

1,1′-(3-Methyl-2-buten-1,4-diyl)-2,4-dicyclohexylferrocene (2c). ¹H NMR (CDCl₃): δ 5.74 (t, J = 8.0 Hz, 1H), 4.09–4.07 (m, 1H), 3.94–3.93 (m, 1H), 3.86–3.85 (m, 1H), 3.85–3.84 (m, 1H), 3.76–3.74 (m, 1H), 3.53-3.52 (m, 1H), 3.04 (d, J = 14.4 Hz, 1H), 2.94 (dd, J = 15.2 and 7.8 Hz, 1H), 2.81 (d, J = 14.4 Hz, 1H), 2.70 (dd, J = 14.8 and 7.8 Hz, 1H), 2.42–2.35 (m, 1H), 2.11-2.02 (m, 2H), 1.98-1.83 (m, 3H), 1.94 (s, 3H), 1.75-1.66 (m, 6H), 1.43–1.08 (m, 9H), 0.96–0.86 (m, 1H). ¹³C{¹H} NMR (CDCl₃): δ 138.1, 123.6, 93.8, 93.0, 86.3, 84.9, 71.9, 70.3, 67.6, 66.6, 66.5, 63.2, 37.33, 37.25, 36.3, 34.5, 34.4, 31.7, 31.5, 29.6, 27.1, 26.8, 26.74, 26.65, 23.6, 22.8, 14.3. Anal. Calcd for C₂₇H₃₆Fe: C, 77.88; H, 8.71. Found: C, 77.70; H, 8.92. HRMS Calcd for C₂₇H₃₆Fe: 416.2165. Found: 416.2166. [α]³⁰_D = +23 (c 1.3, CHCl₃ for the sample of (S)-84% ee derived from recovered (S)-1c). Chiral HPLC Analysis Conditions: Chiralcel OD-H; eluent: hexane; flow rate: 0.5 mL/min; t₁ = 78.3 min (R-isomer), t₂ = 86.4 min (S-isomer).

1,1′-(2-Buten-1,4-diyl)-2,4-di(tert-butyl)ferrocene (2d). ¹H NMR (CDCl₃): δ 5.93–6.02 (m, 1H), 4.13 (s, 1H), 3.98–3.99 (m, 2H), 3.93 (s, 1H), 3.80 (s, 1H), 3.79 (s, 1H), 3.48 (dd, J = 15.1 and 6.0 Hz, 1H), 3.02 (dd, J = 15.1 and 6.0 Hz, 1H), 2.88 (dd, J = 14.6 and 7.3 Hz, 1H), 2.74 (dd, J = 14.6 and 7.3 Hz, 1H), 1.31 (s, 9H), 1.15 (s, 9H). ¹³C{¹H} NMR (CDCl₃): δ 131.6, 130.2, 99.6, 97.6, 86.2, 82.0, 70.6, 70.1, 66.9, 66.3, 65.7, 63.3, 32.6, 32.0, 31.4, 30.4, 24.3, 24.1. Anal. Calcd for C₂₂H₃₀Fe: C, 75.43; H, 8.63. Found: C, 75.29; H, 8.77. HRMS Calcd for C₂₂H₃₀Fe: 350.1695. Found: 350.1697. [α]³¹_D = −8.2 (c 0.56, CHCl₃ for the sample of (R)-58% ee). Chiral HPLC Analysis Conditions: Chiralcel OD-H × 2; eluent: hexane/ⁱPrOH = 3000/1; flow rate: 0.1 mL/min; t₁ = 94.2 min (R-isomer), t₂ = 101.4 min (S-isomer).

3.6. Calculation of Selectivity Factors “k_rel” in Table 1 and in Scheme 5

Selectivity factors (k_rel, k_rel’) of the first-order KR reaction are calculated by equations 1 or 2 [74,75], where c (0 ≤ c ≤ 1) stands for the conversion of the reaction, and ee_sub and ee_pro (0 ≤ ee ≤ 1) are the enantiomeric excesses of recovered (S)-1 and RCM product (R)-2, respectively.

$$k_{\mathrm{rel}}=\frac{\ln \left[\left(1-c\right)\left(1-{\mathrm{ee}}_{\mathrm{sub}}\right)\right]}{\ln \left[\left(1-c\right)\left(1+{\mathrm{ee}}_{\mathrm{sub}}\right)\right]}$$

(1)

$$k_{\mathrm{rel}}’=\frac{\ln \left[1-c\left(1+{\mathrm{ee}}_{\mathrm{pro}}\right)\right]}{\ln \left[1-c\left(1-{\mathrm{ee}}_{\mathrm{pro}}\right)\right]}$$

(2)

Among the three variables in Equations (1) and (2) (c, ee_sub, and ee_pro), conversion “c”, which was determined by the ¹H-NMR measurement of the unpurified reaction mixture, contained up to 5% experimental errors. On the other hand, the ee_sub and ee_pro values were much more accurate. The %ee values of recovered substrate (S)-1 and RCM product (R)-2 in Table 1 and in Scheme 5 were determined by chiral HPLC analysis, which is usually reproducible within 1% of errors, if the two enantiomers are clearly separated in the chromatograms. The k_rel value (determined from eq. 1) and the k_rel’ value (determined from eq. 2) in a single reaction should be identical, in theory. Accordingly, logical conversion of the reaction could be determined from ee_sub (%ee of recovered (S)-1) and ee_pro (%ee of (R)-2) using Equations (1) and (2). The logical conversion values thus obtained showed reasonable agreement with the experimental observations. The k_rel values in Table 1 and in Scheme 5 were calculated using logical conversion, ee_sub, and ee_pro.

4. Conclusions

The molybdenum-catalyzed enantioselective ring-closing metathesis/kinetic resolution of planar-chiral 1,1′-diallylferrocene derivatives 1a-d was reinvestigated utilizing the method of generating various catalytically active molybdenum species in situ. Among the molybdenum catalysts screened, Mo*/(R)-L2 showed the best overall performance with good enantioselectivity and excellent chemoselectivity. Since Mo*/(R)-L2 did not drive the undesirable bimolecular reaction giving 3, the RCM/KR reaction could be conducted under the concentrated conditions using this catalyst, which realized a shorter reaction time whilst retaining excellent enantioselectivity.