Terpenes are the largest class of natural products with hundreds of structures reported each year, totaling over 80,000 examples as of 20191. Many terpenes exhibit interesting biological activities, e.g., to protect the producing organism from predators and/or pathogens and/or to enable intra- and inter-species chemical communication2,3. By utilizing a single enzyme class, the terpene cyclases, nature generates a remarkable number of structurally most diverse and often highly complex (poly)cyclic carbon frameworks. Even more impressive, this is accomplished from simple, achiral linear C5-isoprene derived precursors in a single step, as, for example, in the biosynthesis of the cyclic drimane terpenes (Fig. 1, top). Examples of these include the perfume ingredient and synthetic building block sclareolide (1)4,5,6,7, the antibacterials warburganal (2)8 and totarol (4)9, the antifungal drimenol (3)10 and the polycyclic anti-tumoral triterpene hopene (5, Fig. 1, top)11. Chemists have been fascinated for many decades by these remarkably complex, biosynthetic cyclization cascades. Mimicking the natural cyclization phase12,13 has been a major theme in biomimetic synthesis, even dating back to Eschenmoser’s and Stork’s seminal work on terpene assembly14,15. Since then, multiple publications have appeared offering access to a plethora of sophisticated structural terpene motifs, many of them arising from a protonation-assisted cyclization. However, when following the natural pathway, linear polyenes have to be treated with strong organic or inorganic Brønsted or Lewis acids such as trifluoroacetic acid16, fluorosulfonic acid17,18,19,20, SnCl421,22,23,24,25, BF3·OEt226,27,28, RuCl329,30,31, and In(III) salts30,32, mostly under cryogenic conditions, to allow for productive ring closure. These synthetic methodologies suffer from several general drawbacks, including moderate diastereoselectivities and low functional group tolerance that overall lead to a narrow substrate scope33,34,35. Enantioselective polyene cyclizations have likewise been explored, with all known examples employing either Lewis or Brønsted super acids equipped with bulky, BINOL-derived ligands. These methods are not generally applicable as each catalytic system needs to be tailored for a specific and small set of substrates to effectively convey stereoinduction during cyclization36.

Fig. 1: Examples of terpene natural products, their biosynthesis, and conceptual approaches for biomimetic polyene cyclizations.
figure 1

a Examples of versatile cyclic terpenes 15. b Enzymatic terpene synthesis: the cyclase phase gives rise to structurally diverse hydrocarbon skeletons that are further elaborated during the oxidase phase. c Supramolecular approaches mimicking enzymatic confinement by employing macromolecular cages for selective cyclizations44,45. d Concept for macromolecular solvent clusters facilitating effective cyclizations through pre-arrangement of the apolar polyene chain in the dipolar solvent environment.

The harsh and strongly acidic conditions needed to conduct terpene cyclizations are in clear contrast to biosynthetic terpene formation, which nature performs with high precision under mild conditions at a physiological pH. Structural pre-organization of the substrate carbon chain within the enzyme’s active site is key to success. Depending on how the carbon chain of the polyprenol is folded inside the confined, hydrophobic pocket of the cyclase (Fig. 1b), various head-to-tail (HTT) or tail-to-head (TTH) cyclized hydrocarbons are delivered from identical precursors, utilizing the same reaction pathway in a shape-selective transformation. For example, geranylgeraniolpyrophosphate (6) is the sole precursor for the taxadiene 7, that upon further enzymatic oxidations and tailoring give rise to taxol (10)37, to primaradiene (8, → diaporthein B, 11)38 and to casbene (9, → japodagrin, 12)39. Such a predictable control of molecular conformation in solution thereby steering the selectivity of the reaction14 has been a long-standing problem for organic chemists and is the ultimate motivation for the field of polyene cyclizations.

In the late 1960s, Breslow had already identified supramolecular frameworks exhibiting precise host-guest interactions on a molecular level as potential catalysts with enzyme-like properties and thus coined the term “artificial enzymes”40. Since then, multiple successful examples have been reported of creating molecular containers in the form of nanoscale cavities, pores, pockets, or channels to accelerate reaction performance and alter or enhance selectivity41,42,43. Only recently has this biomimetic concept entered the field of C,C-cyclizations. Self-assembled, polyanionic gallium catecholamide cages 14 can increase the chemoselectivity of the Prins cyclization of citronellal (13) to isopulegol (15) and its diastereomers by forming a hydrophobic pocket, thereby excluding water from the ring-closing step (Fig. 1c)44. The only terpene cyclization conducted in a host cavity so far was reported by Tiefenbacher and colleagues. The self-assembled, resorcinol-based hexameric capsule 17 efficiently promoted cyclization of linear precursors, such as nerol (16), to cyclic monoterpenes 18 with the key step being the cisoid-carbocation formation45,46. These examples constituted landmark conceptual discoveries. However, they still only convert a small set of compounds, because their predefined and rigid structures prevent them from reaching the broad applicability that small-molecule catalysis offers. Thus, for selective reactions, a confined space is necessary. While the first approaches have been made in this direction, methods to achieve this challenge in solution are vastly underexploited to date.

Inspired by these supramolecular approaches, in this work we aim to exploit and extend the concept of artificial enzyme catalysis in polyene cyclizations. The goal is to create a more flexible, building-block-oriented procedure. Emulating the mechanism as well as the conformation-determining properties of terpene cyclases, we create dynamic but structurally defined macromolecular assemblies in-situ. This is achieved by forming discrete, catalytically active Lewis acid-Lewis base complexes with a defined supramolecular structure assembled from fluorinated alcohols, such as 1,1,1,3,3,3-hexafluoroisopropanol (HFIP)47,48,49 or perfluoro-tert-butanol (PFTB), and catalytic amounts of ammonium or pyridinium salts. These are subsequently employed in a plethora of protonation-induced polyene cyclizations. We thereby succeed in transforming a broad variety of structurally most diverse alkenes into the corresponding ring-closed products with a huge functional group tolerance and high efficiency under complete stereocontrol. This catalytic method only requires cheap, off-the-shelf components to achieve conversion under mild and practical reaction conditions. Finally, we have a closer look at the mutual catalyst-solvent and substrate-solvent interplay. NMR investigations reveal structural folding of the substrate and the formation of 3D-self-assembled structures in solution, the latter being corroborated by molecular dynamics (MD) simulations.

Results and discussions

Reaction design and optimization

The starting point for our investigations was the halogenation-induced polyene cyclization we recently developed that provided unprecedented access to a manifold of iodinated, brominated, and even chlorinated carbocycles with excellent yields (up to 90%) and diastereoselectivities (up to >95:5)50,51. The fluorinated alcohol HFIP, together with a tailored Lewis basic electrophilic halogenation reagent, were decisive for a successful conversion. N-halo morpholines showed optimal performance in these halogenation-assisted polyene cyclizations. However, by design this approach always delivers halogenated products and is thus not suitable for the direct synthesis of non-halogenated terpene backbones. For the protonation-induced cation-π cyclization targeted within the current work, we thus began our studies by employing catalytic amounts of morpholinium salts with varying anions (20 mol% catalyst, Fig. 2 and Supplementary Information (SI), Fig. S1). Homogeranyl benzene (19) served as model substrate. Compound 19 possesses two C,C-double bonds of similar electrophilicity, of which the C7,C8-alkene must be chemoselectively addressed to achieve a selective transformation. The trans-decalin product 21 was obtained with excellent diastereoselectivity (>95:5), but reaction times (5h–1d) and chemical yields (8–72%) strongly depended on the anion of the morpholine salt (see SI). Morpholine hydrochloride (20a) gave the best results (72% yield after 1d, see entry 1, Fig. 2). Further extensive optimization of the reaction conditions including, i.a., the proton donor, solvent, and temperature (see SI for further information), identified HFIP, together with 20 mol% of the weakly acidic pyridinium hydrobromide (20b), as the ideal combination to furnish the desired tricyclic product 21 in 71% yield after 13 h at rt (Fig. 2, entry 2). Increasing the reaction temperature to reflux did significantly speed up conversion but also led to decomposition. Diverse side products were likewise obtained under all the above-mentioned conditions, mainly arising from protonation of the internal C3,C4 alkene, which were difficult to separate from the desired 21. Importantly, reactions conducted in non-fluorinated solvents under these conditions did not result in any conversion at all (see entries 4 and 5, Fig. 2 and Supplementary Table 3, SI).

Fig. 2: Optimization and scope of the proton-induced cyclization.
figure 2

Optimized reaction conditions: Substrate (1.0 eq) and DABCO(TfOH)2 (20c, 20 mol%) were dissolved in 0.1 m PFTB at rt under air and stirred until completion. All yields refer to isolated yields. aDetermined by GC-FID from the crude reaction mixture using an internal standard. Isolated yields are in parentheses. bThe diastereomeric ratio (d. r.) was determined by GC-FID or GC-MS. d.r. at C1. cr. r. refers to C7:C5 CO2Me. HFIP  1,1,1,3,3,3-hexafluoroisopropanol, PTFB perfluoro-tert-butanol.

Only by switching the solvent to perfluoro-tert-butanol (PFTB) complete cyclization of 19 was observed, giving 21 as the only product (94% isolated yield) with perfect diastereoselectivity (d.r. >95:5). To the best of our knowledge, PFTB has only been used in mechanistic studies until now and has not been investigated as a solvent in organic chemistry52,53,54,55. While PFTB is already a weak acid (pKa = 5.4), conversion of 19 without a catalyst afforded 21 only in trace amounts after 13 h (see SI). After 7 days less then 50% conversion of 19 was detectable by GC-MS. It is noteworthy that the purity of the fluorinated alcohols needs to be checked carefully. Purification prior to use, as stated in the SI, is necessary as acidic contaminants from the production process can interfere with the catalytic transformation. Further investigations of the catalyst showed that DABCO ditriflate (20c) was even superior to pyridine HBr (20b), as it furnished the desired product 21 in excellent selectivity (d.r. >95:5) and again almost quantitative yield (96%; 94% isolated yield) in less than one hour (cf. entry 7, Fig. 2 and SI). Hence, employing DABCO(TfOH)2 (20c) in PFTB in the transformation offers the opportunity to significantly expand the substrate scope of polyene cyclizations as it likewise addresses electron-poor and thus slowly reacting substrates in a reasonable time (vide infra). When exchanging 20c with strong organic or inorganic acids, many undesired products were produced, reducing the overall yield of 21, e.g., to 63% for TFA (see entry 8, Fig. 2, Supplementary Table 2 and Supplementary Fig. 1, SI).

Synthetic scope

To assess the synthetic utility of our cyclization cascade, we subjected a wide range of structurally and electronically diverse substrates to our reaction conditions (Fig. 2). Ring closure proceeded smoothly with p-methoxy substituted arene 22 (88% yield) or cyclohexyl derivative 23 (91% yield), both giving the cyclic material with excellent diastereoselection (d.r. = 91:9 to > 95:5). Geranylphenols afforded products 2425 in very good yields (64−85%) with diastereoselectivities of up to >95:5. Switching the terminating nucleophile to a carboxylic acid furnished the natural product (±)-tetrahydroactinidiolide (26) in 98% isolated yield. Next, we set out to test the functional group tolerance of our method. Different geranyl benzoates and heterocycles containing esters were smoothly converted into the monocyclic products 2729 (73–99%). Because of the absence of suitable nucleophiles, the trimethyl cyclohexene derivatives were isolated as the major or even only products (69:31 to >95:5 r.r). Acetates and ferrocene esters were also suitable, selectively giving the endo-elimination products 30 and 31. The mild reaction conditions even allowed the conversion of chemically sensitive substrates, such as epoxides, which formed 32 in 67% yield. Notably, the acid labile MOM protecting group in 33 was also tolerated, which is exceptional for acid-catalyzed reactions. Protonation occurred selectively at the prenyl portion, as allyl, propargyl, benzyl, and crotyl moieties were left untouched and gave products 3437 in quantitative isolated yields and fully retained E/Z ratios. Only if two prenyl units were present, the reaction became sluggish, furnishing 38 only in 14% yield. In general, all tested geranyl ethers afforded the corresponding cyclohexenyl products 3237 with >95:5 regioselectivity, independent of the workup (see Supplementary Table 4, SI for further information). This suggests that the deprotonation step trapping the transient carbocation occurs immediately after the cyclization step (cf. Fig. 1, bottom).

More complex starting materials, such as compounds bearing biotin, santonic acid, or amino acids equipped with Fmoc, Cbz, and even the Boc protecting groups, were successfully converted to the monocyclic products 3942. Electron deficient substrates leading to products 4345 were likewise produced in up to 90% isolated yield. The regioisomeric ratio corresponded to that previously observed for halocyclizations50,56. The tetracyclic product 46 and the antimicrobial and anticancer natural product drimenol acetate (47), together with its corresponding dimethoxy ester 48, were accessible from farnesol derivatives with complete diastereoselectivity, while homofarnesic acid provided the fragrance sclareolide (49) in 57% yield. The bicyclic monoterpene eucalyptol (50) and the DNA topoisomerase I inhibitor ß-lapachone (51) were smoothly obtained from α-terpineol and lapachol, respectively. Nerol-derived compounds 52 were likewise cyclized with almost unchanged regioselectivity and yield (21, 27b, and 37). Interestingly, when homoneryl benzene was used, the more stable trans-decalin product 21 was solely obtained with perfect diastereoselectivity. Kinetic studies for the conversions of 19 and 52a showed that the neryl derivative 52a was converted at a higher rate. In both cases, a stepwise process was obvious from the GC-MS data, suggesting equilibration of the alkene confirmation for 52a before ring closure (see Supplementary Figs. 710, SI).

Mechanistic studies

To investigate the reaction mechanism of this approach to polyene cyclizations, we subjected pyridinium deuterobromide (d-20b, 20 and 120 mol% loading) to the reaction mixture. In both cases, deuterium incorporation, at all sp2-carbon atoms that are susceptible to protonation and subsequent re-elimination, was detected in the final products d-21. This is consistent with the stepwise process seen in the kinetic investigations (cf. SI). From these results, we hypothesize:

  1. 1.

    that the active proton fluctuates in the H-bonding network of PFTB in a Grotthuß-like mechanism with a rapid exchange of the acidic protons in solution.

  2. 2.

    that a rapid and dynamic proton exchange between solvent and the catalyst occurs.

The determined kinetic isotope effect (KIE) value of 2.8 ± 0.1 (see SI) obtained by comparing the consumption rate of 19 in PFTB-OH and PFTB-OD further corroborated this assumption, as it revealed the importance of the PTFB-Lewis base H-bonding network in the rate-determining step57.

The role of the Lewis base in our system was examined by titrating DABCO 53 with increasing amounts of PFTB, which resulted in a significant downfield shift of the alcohol proton from 3.6 ppm to 10.7 ppm. This indicates strong H-bonding interactions between the Lewis base and PFTB (Fig. 3a and SI). The change of 1H-chemical shifts of the methylene groups in DABCO was less distinct upon the addition of PTFB, yet visible in the 13C-NMR spectra. This suggests that strong H-bonding interactions indeed take place between the Lewis base and the solvent, as previously observed for N-bromo morpholine and HFIP50. Further increasing the amount of PFTB to >4 equivalents resulted in precipitation of a PFTB-DABCO complex 54 (X-ray structure Fig. 3a, insert and SI). Similar to reported HFIP-amine salts50,58, two PTFB molecules share one proton with the ammonium cation nearby within this complex. In the case of DABCO, however, the pKa of the second amine is 3.0 (vs. 5.4 for PFTB) and thus attracts another molecule of PFTB through strong H-bonding interactions. This results in a final stoichiometry of DABCO/PFTB of 25:75.

Fig. 3: Mechanistic Investigations.
figure 3

a NMR titration curve for DABCO (53) in CDCl3 with increasing amounts of PFTB shows the interaction of the alcohol with the Lewis base. Insert: Crystal structure of DABCO(PFTB)3 salt (54). b Relative rates for conversion of 19 with different catalysts. For more experimental details see the SI. c 1D-NOESY correlation of aromatic protons with the methyl groups in 19. Spectrum was recorded by selective pulsing on HAr (7.25 ppm as band center) using a 0.1 m solution of 19 in the respective solvent. d 13C-NMR titration curves of 19 in CDCl3 with increasing amounts of PFTB, toluene, or tBuOH show that the strongest activation is observed in PFTB at C2 over C6, explaining the selective protonation of the terminal olefin. e Mechanistic summary of proton-induced polyene cyclization.

We then investigated the structure-reactivity relationship of the catalyst regarding both the cation and the anion (Fig. 3b). While without catalyst no significant conversion of 19 was visible, an increase in reaction rate was observed by applying the protonated N-compounds morpholine, pyridine, and DABCO in the form of their hydrobromide salts. The krel increased with decreasing pKa value. Changing the anion further accelerated the transformation. Applying DABCO bistriflate (20c) showed an almost doubled reaction rate (krel = 2.05 × 102) when compared to its hydrobromide derivative (krel = 1.24 ×102) and an even 20 fold increased rate constant relative to that of the DABCO TFA salt (20f, krel = 1.1 × 101). Gutmann-Becket59,60 experiments revealed that besides the Brønsted acidity the catalyst’s Lewis acidity, and thus the strength of the mutual interaction of amine salts and F-alcohol within the H-bonding network, plays an important role in the cyclization event (see SI). In general, the acceptor number (AN) of fluorinated alcohols does not correlate with their pKa value but rather with their ability to coordinate Lewis basic molecules. Upon addition of our catalysts 20 to PTFB the Lewis acidity further increased (AN (PTFB) = 53.8 vs. AN (PTFB + 20b) = 54.1 vs. AN (PTFB + 20c) = 59.8) surpassing even the Lewis acidity of SnCl4 (AN = 59.0) if PTFB is activated by DABCO ditriflate 20c.

Next, we turned our attention to the influence of the extended H-bonding network on reaction selectivity. Within the scope of this work, we have shown that fluorinated alcohols, particularly PFTB play an important role in the selective cyclization of mono- and sesquiterpenes. As hypothesized earlier by Denmark61 and our group50, fluorinated alcohols are capable of controlling the conformation of apolar linear materials by forming cyclic, linear, or helical supramolecular clusters in solution62,63,64,65,66. These domains force the substrates to prearrange in their thermodynamically favored chair-like conformation. This shape-induced control enables the transformation to proceed with excellent yields and selectivities. To test this hypothesis, we recorded 1D-1H-NOESY spectra of model substrate 19 in various solvents by selectively resonating 19 at the aromatic protons (Fig. 3c and SI). Only in PFTB did the protons of the terminal C1 Me-groups show significant NOE contacts to aromatic H-atoms, which can only be explained by the folding of 19 and thus reduction of the surface area of this lipophilic polyene substrate. Subjecting 19 to 13C-NMR titration with increasing amounts of PFTB, toluene, and tert-butanol in CDCl3 revealed that only in PFTB did the sp2-carbons significantly shift downfield, doing so by up to 3.5 ppm for C2 (Fig. 3d and SI). This observation was due to fluorine-π interactions stemming from the hydrophobic CF3 groups clustering around the apolar starting material62 and thus activating the more exposed head alkene (C2/C3) the most. This finding nicely explains the superior selectivity of terminal vs. internal protonation in 19 in PFTB. In other solvents, no such activation was observed, thus resulting in the production of significant amounts of apolar side products.

Altogether, these investigations clearly reveal that the strong H-bonding network formed by non-covalent interactions of catalyst 20c and PTFB plays a decisive role in controlling the solution phase conformation and thus steers the outcome of the reaction. The microstructured solution forces the linear polyene 19 into a distinct chair,chair-like conformation simultaneously activating selectively the terminal alkene moiety and stabilizing cationic intermediates and transition states. These aspects of structural pre-organization with their impact on reactivity and selectivity in chemical transformations are key to explaining the biogenic isoprene rule formulated by Eschenmoser14 and Stork15.

Molecular dynamics (MD) studies

The postulated role of the solvent was further explored by extensive MD studies (simulation details are given in the SI). Similar to the structuring induced by HFIP62, a clear structuring of the polar and apolar structural elements by the polar F-alcohol was observed. The most prominent result was the formation of supramolecular, linear ionic aggregates of the catalyst (pyridinium hydrobromide (20b), Fig. 4a and DABCO bistriflate (20c) see SI) in PFTB. This was facilitated by the polar hydroxy groups of PFTB, which preferentially face the charged ions and the apolar CF3 groups pointing towards the bulk solution, thereby creating distinct and structured catalytic domains in the solution by a different hydrogen bonding network compared to the bulk. This gives rise to a percolating global microstructure because the linear aggregates of pyridinium bromide 20b are not limited to a local inhomogeneity as in cyclohexane (Fig. 4b). The latter causes a depletion of ions in the bulk and allows contact with the ions only at the ion-cyclohexane interface. In PTFB, however, a new preferential orientation of hydrogen bonds arises through the inclusion of 20b, which is not present in pure PTFB. This prearrangement could potentially support proton transfer along the aggregates and follow a Grotthus-like diffusion, as similarly observed in the cyclase II enzyme ent-copy diphosphate synthase67. The non-point-like pyridinium cation is likewise important, as such structuring could not be observed in control simulations where 20b was substituted by NaBr.

Fig. 4: Solvent dependent change of ionic aggregation observed in MD simulations.
figure 4

Formation of supramolecular aggregates in a PFTB (first layer as sticks) and b cyclohexane (blue sticks). In a the polar hydroxyl groups of the immediate PFTB molecules (stick) are facing the pyridinium bromide aggregates and CF3 groups the bulk. c Aggregates observed in MeOH are less well defined (left side), or well defined (right side). The brightness of the atoms reflects their distance to the observer in the direction perpendicular to the focal plane; VdW representation is used for cations (silver), anions (orange), and 19 (turquoise).

Aggregates like those occurring in PTFB were likewise observed in control simulations using other polar, organic solvents such as MeOH (Fig. 4c), but were overall less well-defined. The question of whether the solvent itself facilitates prearrangement in the coiled state of 19 was tackled by comparing potentials of mean force (PMFs) obtained from umbrella sampling68 and the weighted histogram analysis method69 for the bond-forming carbon atoms of 19. For all observed solvents, the folded form of 19 was less favorable. However, with the addition of the ions, the coiled conformation of 19 was stabilized in PTFB as opposed to destabilized in cyclohexane (see SI). As the reaction can only happen from the coiled form, the preaggregation gives an additional favorable contribution to the reaction.


By pre-organizing substrate conformation, nature controls and directs the outcome of its chemical reactions with high efficiency. Terpene cyclizations are impressive examples of such shape-selective transformations, providing a manifold of different and structurally complex carbon frameworks depending on how the carbon chain gets folded inside the enzymatic pocket. Controlling substrate conformation in solution, as terpene cyclases do, and thus steering the selectivity of a certain transformation predictably is yet an unmatched challenge for organic chemists. With our studies, we introduced a broadly applicable method for the head-to-tail proton cyclization of structurally different linear polyenes to the organic synthetic toolbox. This was accomplished by mimicking the decisive enzymatic features of (a) adjusting the reagent and/or substrate’s reactivity, (b) stabilizing transition-states and intermediates, and (c) controlling substrate conformation within our in-situ formed, self-assembled, cooperative supramolecular network consisting of PTFB and catalytic amounts of ammonium or pyridinium salts. These dynamic PTFB-amine-supramolecular clusters show all decisive properties and are perfectly suited to be employed in carbocation cascade reactions, and thus, in the total synthesis of terpenes and their artificial analogs. Our approach is very mild, tolerating many functional groups including acid-labile functionalities. The procedure is operationally simple and robust, does not require inert or cryogenic conditions, and needs only affordable, commercially available catalysts and solvents. It provides polycyclic products with generally high to excellent yields and simultaneously assembles multiple stereogenic centers. The excellent diastereoselectivity (up to d.r > 95:5) is caused by the controlled conformational pre-folding of the substrate chain inside the supramolecular assemblies. Due to the size-adaptive nature of our supramolecular networks, the reaction offers a broad substrate scope. Applying this methodology, we have efficiently synthesized a broad range of terpene-type structural scaffolds, including the natural products tetrahydroactinidiolide (26), cyclogeranylacetate (30), drimenol acetate (47), sclareolide (49), eucalyptol (50), and β-lapachone (51). Through our mechanistic investigations, important insights into the underlying principles of H-bonding networks and their influence on reactivity and selectivity are now available. These principles will help to further advance the field of fluorous alcohol-triggered reactions from pure serendipity to a more rational reaction design. In addition, our studies impressively revealed that PTFB, which had not yet found its way into organic synthesis, can indeed serve as a versatile solvent, pushing distinct transformations even in a superior fashion to its more popular relative HFIP. PFTB constitutes an excellent reaction medium for proton-induced polyene cyclization. However, at the very least, it will also be effective for any transformation needing cationic species to be stabilized, in a dipolar medium, and mildly acidic conditions. We hope that PFTB will rise from a laboratory curiosity to a valuable solvent like HFIP.


General procedure for the cyclization of linear polyenes

The corresponding linear polyene substrate was dissolved in 0.1 m PFTB at rt in a 4 mL screw cap vial with a magnetic stir bar. In total 20 mol% of DABCO(TfOH)2 (20c) was added and the mixture was stirred until completion (determined by GC or TLC). Then, 2 mL CHCl3 was added to terminate the reaction and the solvent was removed under reduced pressure. The crude mixture was then directly purified by column chromatography on silica gel to obtain the corresponding product.