In order to test our hypothesis, we accessed (E)‐6‐(3,5‐dimethoxyphenyl)hex‐2‐enal 7 a in a 4‐step synthesis from 3,5‐dimethoxybenzyl bromide (see supporting information). A variety of catalysts were screened (Table 1) including primary amine I (Entry 1), a bifunctional thiourea II (Entry 2), – a system with which we have much experience, a selection of chiral phosphoric acids III (Entry 3) and a range of secondary amines IV–X (Entries 4–10). Interestingly, the cinchona alkaloid derived primary amine I, used by Takao and co‐workers in the only other successful organocatalytic resorcinol derived Friedel−Crafts process known to date,^[6b] gave very poor selectivity (entry 9). Pleasingly, however the variously silylated Jørgensen‐Hayashi type catalysts^[8] did give the desired cyclized product. Although the unoptimized yields for these systems were low, we found that the selectivities were very encouraging. As described by Hayashi, Seebach and co‐workers, the bulkier protecting group led to higher selectivity in such 1,4‐additions,^[9] although interestingly the non‐trifluoromethyl system VI and VIII did not work at all (Entry 3). In contrast, the MacMillan catalysts IX and X gave better yields but poorer enantioselectivities.

Entry	Catalyst	Time, d[a]	Yield[b]	ee (%)[c,d]
1	I	1	69	9
2	II	nr	–	–
3	IIIa‐c	nr	–	–
4	IV	1	15	72
5	V	1	29	81
6	VI	nr	–	–
7	VII	nr	–	–
8	VIII	nr	–	–
9	IX	7	68	−68
10	X	1	58	−43
11e	V	7	26	92

We also found that removal of the co‐catalyst led to much improved selectivity (entry 11). This was confirmed insomuch that a screen of additives including acid co‐catalysts and bases did not significantly improve the reaction (see supporting information for complete study), leading us to conclude that such additives might be catalyzing the non‐enantioselective background process.

In view of this, and because of its superior selectivity, we decided our best option was to optimize the yield of the conditions described in entry 10. To begin with, we performed a solvent screen (Table 2) which showed that chlorinated solvents worked best with our reaction, with chloroform being optimal and giving improved yield and further improved enantioselectivity (96%, entry 2). Fascinatingly, in other solvents, both of these variables suffered ‐ with toluene, THF and ether giving no reactivity whatsoever (entries 4 and 5). Pleasingly however, modulation of concentration (Table 2, entries 7–9) showed that yield could be further improved with no detriment to the excellent enantioselectivity, whilst at the same time improving reaction time. Our last attempt to improve the yield was to vary the temperature (entries 10–11), where intriguingly not only did the yield and reaction rate improve with increased temperature, but the enantioselectivity remained the same, even at the optimal conditions of 50 °C (entry 11).

Entry	Solvent (conc, M)	T, °C	t	Yield: (%)[b]	ee (%)[c]
1	CH2Cl2 (0.2)	rt	7 d	26	92
2	CH2Cl2 (0.2)	rt	6 d	35	96
3	MeOH (0.2)	rt	2 d	38	47
4	THF (0.2)	rt	nr	–	–
5	Et2O (0.2)	rt	nr	–	–
6	MeCN (0.2)	rt	6 d	6	69
7	CHCl3 (0.05)	rt	2 d	31	86
8	CHCl3 (0.5)	rt	3 d	48	96
9	CHCl3 (1.0)	rt	2 d	31	95
10	CHCl3 (0.5)	40	2 d	50	94
11	CHCl3 (0.5)	50	7 h	68	94

To expand the scope of this reaction, we explored the use of various substrates (Table 3). Compounds varied from the nature of the saturated ring to the substitution of the resorcinol. Heterocyclic compounds worked well, including oxygen and various N‐substituted systems. All gave excellent enantioselectivities, including 8 c and 8 d, which can open the door to constrained δ‐amino acids which are of potential use within foldamer systems.^[10] Interestingly, catechol 8 k whilst suffering from poorer conversion gave a reasonable enantioselectivity. Substrates containing electron‐withdrawing substituents, unfortunately, failed to convert (see supporting information). However, post‐cyclization modification to introduce these is possible as demonstrated in compound 14 discussed vide infra.

Based on the observed stereochemical outcome (determined by X‐ray crystallographic analysis of the corresponding camphorsulfonyl system of compound 8 a – see supporting information), we propose the transition state shown in figure 2, whereby the bulky silyl group of the expected E‐iminium ion blocks the approach of the resorcinol nucleophile from the re‐face. Cyclization via S_NAr thus leads to the (S)‐tetrahydronaphthalene system reported.

We then demonstrated the utility of our resorcinol systems by using the Friedel–Crafts type 1,4‐addition in the synthesis of a CBD derivative, one of the major cannabinoids found in cannabis both in its ‘normal’ and ‘abnormal’ forms (Scheme 2). The medicinal utility of cannabinoids has led to the development of multiple cannabinoid‐derived medications such as Nabilone and Sativex.^[11] Cannabinoid analogues have also attracted widespread medical attention because of their interesting pharmacological properties including analgesic, antiemetic and as an appetite stimulant.[¹², ¹³, ¹⁴] Since the discovery and characterization of the CB1 and CB2 receptors and the human endocannabinoid system in the early 90s, the pursuit for small molecules capable of modulating these systems has accelerated astonishingly.[¹⁵, ¹⁶, ¹⁷] Interestingly, although it is known that CBD itself does not interact with this system, its clearly beneficial pharmacological properties have in turn established a need for creating new, useful and easily accessible routes to enantiopure unnatural analogues.

In our first synthesis, removal of the primary alcohol of 8 a by simple mesylation and treatment with lithium aluminium hydride gave deoxygenated system 9 in 97% overall yield. Subsequent O‐demethylation with boron tribromide at 0 °C afforded deprotected chiral resorcinol 10 in 99% yield. This was then reacted with cis‐isolimonenol 6 in either one of two conditions. The first was in the presence of boron trifluoride diethyl etherate ‐ conditions first used by Petrzilka, Haefliger and Sikemeier^[18] to access abnormal CBD derivative 11. The second set of conditions, however, allowed access to the normal CBD analogue, through the use of a substoichiometric quantity of PTSA. Not only did we obtain the desired target 12, but also the abnormal derivative 11 and a small quantity of what is speculated to be the bis‐limonene system 13 identified by LC‐MS. Exclusive access to the ‘normal’ system, however, was achieved through bromination of chiral bicyclic system 9, to give us the arylbromide system 14 upon demethylation. Coupling with the same terpene gave the ‘normal’ regioisomer of the bromo‐CBD compound 15 in 31% yield. The structure and absolute configuration of this was also unambiguously determined by X‐ray crystallographic analysis (using a molybdenum X‐ray source, Figure 3) owing to the known configuration of the limonene (1S,4R), and further confirmed the previously assigned absolute stereochemistry. Additionally, not only did this system allow us to obtain more crystalline material for XRD, but it also has two further clear advantages – first of course, that only the ‘normal’ regioisomer is obtained, and second that the aryl bromide can be derivatized to further systems via aryl coupling methodologies.