Introduction

Whereas organic alkenes are widely known to undergo a reaction sequence of protonation followed by nucleophilic addition (i.e., alkene electrophilic addition), such a process is generally not accessible to benzenes, owing to the highly stabilized aromatic ring. However, we recently demonstrated that this reaction sequence was not only possible for a dihapto-coordinated (η2) benzene complex (Fig. 1)1,2, but that the resulting η2-diene complex (3) could participate in a second protonation/nucleophilic addition sequence to form the corresponding cis-3,6-disubstituted cyclohexene complex (4)2. The range of nucleophiles that can be added in this manner includes cyanide, enolates, Grignard reagents, amines, and alkoxides2. However, the highly π-basic nature of the tungsten system required to coordinate and activate the benzene ring also stabilizes the η2-arenium intermediate (2), thereby diminishing its ability to react with so-called “π-nucleophiles” such as arenes and alkenes3. We posited that a different strategy, in which the benzene was first double-protonated, might be possible. The resulting dicationic species (5), if accessible, would be highly electrophilic, and could be capable of reacting with much milder nucleophiles than its monoprotonated precursor 2. The resulting π-allyl species (e.g., 68) would then be positioned to react with a second nucleophile to generate a 3,6-disubstitued cyclohexene complex (e.g., 911). Herein we show a new approach to dearomatization of benzenes in which coordination to a tungsten complex enables the double protonation of the bound arene and subsequent coupling to a second aromatic molecule. The resulting allyl species can then be combined with a “hard” nucleophile resulting in novel cis-3,6-disubstituted cyclohexenes.

Fig. 1: The tungsten-promoted double-protonation of benzene followed by electrophilic aromatic substitution (EAS), and addition of cyanide ion.
figure 1

The terms distal and proximal are relative to the PMe3 group. 6, 9: Ar = 4’-methoxyphenyl; 7, 10: Ar = 4’-hydroxyphenyl; 8, 11: Ar = thiophen-2’-yl. Yields: 6-8: 25–64%; 911: 50–70% dr ~4:1. OTf - = CF3SO3-. Inset: the stereoselective double-protonation of a deuterobenzene complex followed by EAS of anisole. Tp = (trispyrazolyl)borate.

Results

Double protonation of benzene

We initially focused on the parent benzene complex, WTp(NO)(PMe3)(η2-benzene) (1; Tp = (trispyrazolyl)borate), which we have previously shown could be protonated by diphenylammonium triflate (pKa ~1; DPhAT) to generate the η2-arenium complex 2 (Fig. 1)1. When the η2-arenium species 2 is treated with a CD2Cl2 solution of triflic acid (HOTf; T = 0 °C), the 1H NMR spectrum reveals that a second protonation occurs on the benzene ring to form 5; NOESY and HSQC NMR data indicate two adjacent diastereotopic methylene groups (Fig. 1). Repeating the reaction in neat DOTf at −78 °C and gradually warming the solution to 0 °C indicates that the initial reaction of 2 and acid generates a paramagnetic complex as indicated by three broad peaks from 5 − 7 ppm. These signals gradually give way to the doubly protonated complex 5 - d2 as the brown solution turns deep orange. In contrast, if the benzene complex 1 is subjected to HOTf in CD2Cl2 at −30 °C, only the monoprotonated complex 2 is formed; this solution evolves to form 5 only after warming to 0 °C. Attempts to isolate the dicationic complex 5 by precipitation with diethyl ether resulted in decomposition. However, when 5 was generated in situ and treated with anisole, phenol, or thiophene at −30 °C, an electrophilic aromatic substitution (EAS) reaction occurred between the free aromatic and the “carbenium” of 5 proximal to the PMe3. Addition of the arene occurred anti to the metal to form η2-allyl complexes 68D. As demonstrated below, anisole and phenol react at the para carbon (C4’), while thiophene reacts at the α carbon (C2’). This reactivity significantly differs from the precursor η2-benzenium complex 2, which shows no signs of reactivity with arenes, save for indole2. The resulting η2-allyl species 6 - 8D (also referred to as hyperdistorted η3-allyl4, or σ-π distorted5,6,7), are heavily weighted toward the conformer with the carbenium carbon distal to the PMe34. Subsequently, the addition of a second nucleophile (CN-)2,4 resulted in cis-3,6-disubstituted cyclohexene complexes 9 - 11D. Unfortunately, these products were all accompanied by roughly 20% of a second isomer, both for the η2-allyl intermediate (6 - 8P) and for the final cyclohexene complex (9 - 11P). These minor products were ultimately characterized (vide infra) as diastereomers of the major cyclohexene products in which the free arene added to the carbenium distal to the PMe3 and the second nucleophile added to the proximal allyl carbon. Attempts to improve the diastereoselectivity of this reaction through adjusting temperature, solvent or reaction time were unsuccessful.

When the reaction sequence to generate the anisole addition product 9D was repeated using the deuterated benzene complex 1-d6, the two protons incorporated in the double protonation sequence were identified by two signals appearing for 6D-d6 at 3.50 and 1.22 ppm (CD3CN), corresponding to the 4-endo and 5-exo positions (NOESY). These observations reveal two different mechanisms for protonation: consistent with our earlier studies of monoprotonated benzene, the first protonation occurs syn to the metal (H4 endo)1, while the second protonation occurs anti (H5 exo). The 1H NMR spectrum of 6D-d6 indicates that even at high acid concentrations, both protonations are highly regio- and stereoselective, with minimal amounts of proton signal (0-15%) appearing at other positions.

The first protonation of the benzene ring is observed to occur syn to the tungsten1. Previously we have demonstrated the ability to methylate the NO ligand of the cyclohexene complex WTp(NO)(PMe3)(C6H10)1, so it stands to reason that the nitrosyl can also be protonated by strong acid, forming 1H. In turn, DFT calculations demonstrate that a purported transfer of a proton from the NO ligand to the ring (1H2; Fig. 2; SI) can occur with a modest transition state energy of 8.2 kcal/mol and free energy change of −8.8 kcal/mol. It is tempting to ascribe the paramagnetic species (vide supra) observed at −78 °C to the nitrosyl-protonated benzene complex, 1H. Casting doubt on this hypothesis is the observation that the analogous methylated cyclohexene complex WTp(NOMe)(PMe3)(C6H10) is diamagnetic1. Calculations further show that an analogous NO-assisted second protonation is possible (TS = 6.1 kcal/mol, ΔG = −13.9 kcal/mol; Fig. 2); however, the double protonation of 1-d6 to form 6D-d6 unambiguously results in a trans arrangement of the two ring protons, indicating that the second protonation must occur mainly by an intermolecular pathway, anti to the metal. Thus, while we cannot rule out an intermolecular protonation of the η2-benzene for both first and second protonations of 1, given the high stereoselectivity observed for 6D-d6, it seems likely that the first protonation of benzene is intramolecular (NO assisted) while the second is intermolecular.

Fig. 2: Comparison of first (left) and second (right) intramolecular protonations of benzene.
figure 2

1H is the purported NO-protonated conjugate acid of 1. 2H is the purported NO-protonated conjugate acid of 2. Only the first protonation results in stereochemistry supported by experiment.

According to calculations, the double-protonated benzene complex 5 can be considered as a highly distorted η4-tungsten(II)-diene complex (Fig. 3), with elongated bond lengths between tungsten and the terminal diene carbons (W-C1 and W-C4: 2.66, 2.79 Å; cf. W-C2 and W-C3: 2.30, 2.37 Å). These distorted structural features are reminiscent of those seen for the η2-allyl species described earlier. A search of the Cambridge Structural Database8 failed to identify any analogously distorted η4-diene structures. However, the structure of 5 is reminiscent to those found in zirconium and hafnium complexes of η4-cyclooctatetrene9. The distal carbenium carbon of 5 has the longest bond to the metal (2.79 Å) and might be predicted to be the more reactive site of addition; however, nucleophilic attack occurs predominantly at the proximal carbenium. Such an addition generates η2-allyl species (68D) with the remaining carbenium distal to PMe3. The distal form (D) is known to be several kcal/mol more stable than the isomers resulting from distal addition of the arene (68P)2. Hence, we rationalize the kinetic preference for the addition of the arene to the proximal carbenium by invoking a transition state that resembles the product in which the carbenium is distal to the PMe3.

Fig. 3: DFT optimized structure of [WTp(NO)(PMe3)(C6H8)]2+ (5), the result of the double-protonation of the η2-benzene ligand of 1.
figure 3

The structure features a highly distorted η4-coordinated cyclohexadiene complex.

The observed 4:1 selectivity discouraged us from developing a synthetic method for enantioenriched cyclohexenes using this approach. Hints of analogous reactivity were observed for the molybdenum complex MoTp(NO)(DMAP)(η2-benzene)10, (1-Mo) including spectroscopic evidence for the Mo analog of 6D (6D-Mo). The large-scale preparation of 1-Mo and spectroscopic data for 6D-Mo can be found in the SI. However, the high sensitivity of these compounds to acid ultimately discouraged our further investigation.

Double protonation of anisole

We next considered a modified strategy (Fig. 4) in which an η2-anisole complex would be double-protonated. We reasoned that the methoxy substituent would not only facilitate the double protonation, but also could help direct the aryl addition to the para- carbon of the anisole, analogous to what we have previously observed for anilines11. In contrast to our aniline observations, we anticipated that the oxocarbenium could be easily reduced later in the reaction sequence. The tungsten anisole complex 12D exists in solution as a 3:1 equilibrium with its stereoisomer 12P12,13. However, the 2H-anisolium complex 13D has been shown to be thermodynamically favored over its proximal analog 13P ( >20:1), again favoring the oxocarbenium carbon in the distal position13. When 13D was subjected to highly acidic conditions (HOTf/acetonitrile), protonation occurred exclusively at the terminal carbon of the diene fragment (C3) to form the dication 14D (Fig. 4). We note that changes in the W-P coupling constants are a convenient way to monitor the progress of the reaction: the neutral anisole complex (12D) has JWP = 312 Hz, single protonation (13D) has JWP = 285 Hz, and double protonation (14D) has JWP = 240 Hz. Treating the dication 14D with phenol, anisole, and thiophene resulted exclusively in the enonium species 1517D.

Fig. 4: The tungsten-promoted double-protonation of anisole followed by electrophilic aromatic substitution (EAS), reduction, and addition of cyanide ion.
figure 4

HOTf = CF3SO3H; Tp = trispyrazolylborate. 6,9,15,18: Ar = 4-methoxyphenyl; 7,10,16,19: Ar = 4-hydroxyphenyl; 8,11,17,20: Ar = thiophen-2-yl. Yields for 911: 50–70%.

Although the anisole complex exists in solution as two interconverting diastereomers (12D, 12P), only 12P is present in the crystalline solid (prepared from solution precipitation)14. Adding a cold solution of HOTf in acetonitrile (−30 °C) to a powder of 12 (−30 °C) results exclusively (dr > 20: 1 P/D) in the 2H-anisolium complex 13P. This compound, like its benzene analogue, can be protonated a second time to form the dicationic complex 14P, which can then be elaborated into the η2-enonium complexes 15 -17P through the additions of anisole, phenol, or thiophene. Subsequent reduction of the η2-enonium complexes (1517) for either the P or D series generates the allyl ether complexes 1820, and this is followed by acid-induced loss of methanol to form the π-allyl complexes 6 - 8. Finally, treatment with NaCN produces the cis-3,6-substituted cyclohexene complexes 9 - 11, but in this case, each diastereomer can be generated completely free of the complementary diastereomer (dr > 20: 1; Fig. 5). We note that even though 6 - 8P favor a conformation in which the carbenium is distal to the PMe3, steric factors apparently favor addition to the proximal form (Fig. 4). The cyclohexene product is then liberated with the oxidant NOPF6 to generate the organic compounds 4′-hydroxy-1,2,3,4-tetrahydro-[1,1′-biphenyl]−4-carbonitrile (21; 70%), 4′-methoxy-1,2,3,4-tetrahydro-[1,1′-biphenyl]−4-carbonitrile (22; 34%), and 4-(thiophen-2-yl)cyclohex-2-ene-1-carbonitrile (23; 56%). Although beyond the scope of this preliminary study, we note that the arene complexes 1 and 12 can be prepared in enantioenriched form2,15,16. Therefore, while this preliminary report only describes racemic mixtures, either hand of the organic cis-3,6-cyclohexene would be available from an enantioenriched anisole complex.

Fig. 5: Molecular structure determinations (50% ellipsoids) for the cis-3,6-disubstituted cyclohexene complexes 10D and 11D.
figure 5

These structures feature the addition of phenol (10D) and thiophene (11D) to a double-protonated anisole precursor (14).

Scope of the reaction sequence

In the reaction sequences outlined in Fig. 4, a single regio- and stereoisomer of a cis-3,6-disubstituted cyclohexene complex is obtained (9 - 11). In a preliminary screening of arene scope, protonation of 13D (generated in situ) reacts satisfactorily with other electron-rich heterocycles (indoles and furans) as well as other π systems such as azulene and ferrocene (Fig. 6). Using 6-methoxytetralin as the arene, the resulting oxonium was elaborated into the corresponding π allyl complex analogous to 68D (SI; 41D). This species was then subjected to a range of preliminary screen of carbon nucleophiles including a silylated ester enolate, a silylated acetylide, Rupert’s reagent, and benzyl Grignard, as well as the heteronucleophiles morpholine and methoxide. In all cases a cis-3,6-disubstitited cyclohexene complex was produced as a single stereoisomer (29D34D; dr > 20:1; Fig. 7). Oxidative decomplexation with NOPF6 resulted in the free organics 3540 (Fig. 8).

Fig. 6: Preliminary scope of aromatic π-nucleophiles that react with the double protonated anisole complex 14D.
figure 6

Successful additions include benzene derivatives, indole, furans, azulene, and ferrocene. Inset: molecular structure determination of the azulene derivative 27D.

Fig. 7: Preliminary examples of cis-3,6-disubstituted cyclohexene complexes prepared from an η2-allyl complex and various nucleophiles.
figure 7

Nucleophilic reagents include TMSCF3 (29D), trimethylsilylacetylene/BuLi (30D), BnMgCl (32D), methyl trimethylsilyl dimethylketene acetal (31D), morpholine (33D), and MeOH/KOH (34D).

Fig. 8: Preliminary examples of cis-3,6-disubstituted cyclohexenes prepared by this method.
figure 8

These compounds result from the oxidative decomplexation of the corresponding tungsten complexes. dr > 20:1 for all complexes listed.

Discussion

The Scholl reaction (Friedel-Crafts Arylation) is the coupling of two aromatic rings with the aid of a strong Lewis or Brønsted acid followed by dehydrogenation and is closely related to oxidative aromatic coupling17. This historically significant C-C bond forming reaction17 normally leads to aromatic products, often as oligomeric mixtures, and is driven by the large stabilization gained upon their rearomatization. In contrast, the syntheses of 3,6-disubstituted cyclohexenes such as 2123 and 3540 is enabled by the tungsten, which inhibits rearomatization of the bound carbocycle. Of note, these compounds have not been reported previously, despite their relatively modest structures. The closest comparisons are 1,4-dihydronapthalene analogs prepared from a Diels-Alder reaction with benzyne18, or reaction sequences involving the coupling of aryl halides to cyclohexenes or cyclohexanones. More generally, methods employed to couple aromatics to cycloalkanes typically involve cross-coupling reactions such as Negishi19, Stille20, Suzuki21, and Hiyama couplings22, but such reactions are more difficult than sp2-sp2 coupling protocols and are often plagued by elimination byproducts. Furthermore, these reactions typically require the use of precious metal catalysts and aryl halides or other suitable aryl precursors. Corey-House23,24 and Kochi-Schlosser type couplings avoid precious metals but require aryl Grignard reagents25. For cases where an organic arene is utilized, strong Lewis acid activators are typically required (Friedel-Crafts)26. Lewis and Brønsted acids have successfully been used in Friedel-Crafts alkylations26, especially in the case of benzyl electrophiles, where rearrangements of the carbocation intermediate are less of an issue. However, examples carried out with high stereoselectivity are rare due to the fact that the electrophile typically passes through a planar sp2 intermediate26. The closest comparisons of EAS reactions related to the current study involve cyclohexadienyliumiron complexes combining with anilines or phenols to generate carbazoles27,28. In these studies, the iron complex does not control the stereochemistry of the reaction and cyclohexadienes or arenes are produced. Limited examples of EAS reactions have also appeared in our own work, in the synthesis of γ-substituted enones29 and tetrahydroindolines30. However, in no case previously were we able to couple these reactions to a second nucleophilic addition. A more comprehensive scope of cis-3,6-disubstituted cyclohexenes available by this new method, including enantioenriched variations, will be disclosed in due course.

Methods

Detailed synthetic procedures and characterizations of all new materials are provided in Supplementary Information document. NMR spectra were obtained on 500, 600, or 800 MHz spectrometers. Chemical shifts are referenced to tetramethylsilane (TMS) utilizing residual 1H or 13C signals of the deuterated solvents as internal standards. Phosphorus NMR signals are referenced to 85% H3PO4 (δ 0.00) using a triphenyl phosphate external standard (δ − 16.58). Chemical shifts are reported in ppm, and coupling constants (J) are reported in hertz (Hz). Infrared (IR) spectra were recorded on a spectrometer as a glaze on a diamond anvil ATR assembly, with peaks reported in cm−1. Electrochemical experiments were performed under a nitrogen atmosphere. Cyclic voltammetric data were recorded at ambient temperature at 100 mV/s, unless otherwise noted, with a standard three electrode cell from +1.8 to −1.8 V with a glassy carbon working electrode, tetrabutylammonium hexafluorophosphate (TBAH) electrolyte (∼1.0 M), and acetonitrile (MeCN) or tetrahydrofuran (THF) as the solvent. All potentials are reported versus the normal hydrogen electrode (NHE) using cobaltocenium hexafluorophosphate (E1/2 = − 0.78, −1.75 V) or ferrocene (E1/2 = 0.55 V) as an internal standard. Peak separation of all reversible couples was less than 100 mV. All synthetic reactions were performed in a glovebox under a dry nitrogen atmosphere unless otherwise noted. All solvents were purged with nitrogen prior to use. Deuterated solvents were used as received from Cambridge Isotopes. When possible, pyrazole (Pz) protons of the (trispyrazolyl)borate (Tp) ligand were assigned as “Pz3/5 or Pz4”. B-H peaks (around 4 − 5 ppm) in the 1H NMR spectra are not assigned due to their quadrupole broadening; however, confirmation of the BH group is provided by IR data (~2500 cm−1). Details of all compound syntheses are provided in the supporting information document.