We began our studies by preparing the fluorinated carbene precursor, diethyl 2-diazo-1,1,3,3,3-pentafluoropropylphosphonate (5). This compound was synthesized smoothly within three steps in 16.6% overall yield. The synthetic route commenced from the preparation of N-protected amine 3, followed by the deprotection of benzylamine 3 to furnish 4 and ended with the diazotization of 4 using tert-butyl nitrite. As a result, the target diazo reagent 5, which contains both a trifluoromethyl and a difluoromethylphosphonate moiety, was isolated as a stable, non-volatile liquid (Scheme 2).

Scheme 2: Synthesis of diethyl 2-diazo-1,1,3,3,3-pentafluoropropylphosphonate (5).

To highlight the synthetic utility of the novel carbene precursor 5, subsequent [2 + 1] cycloaddition reactions with selected terminal aromatic and aliphatic olefins were examined. The reaction conditions were first optimized using 4-methylstyrene as a model substrate. The results are summarized in Table 1.

Table 1: Optimization of the reaction conditionsa.

Entry Catalyst t (h) T (°C) Solvent Conversion rateb (%) 1 Rh2(OAc)4 5 111 toluene n.r. 2 Rh2(OAc)4 3 40 DCM n.r. 3 CuI 3 40 DCM n.r. 4 CuI 1 111 toluene 42 5 CuI 2 111 toluene 59 6 CuI 3.5 111 toluene 100c 7d – 24 40 DCM n.r.

aReaction conditions: alkene (0.15 mmol), diazo compound 5 (0.1 mmol), CuI (1 mol %), dry toluene, 111 °C, Ar atmosphere; bDetermined by 19F NMR spectroscopy; cIsolated yield 74%. dUnder UV irradiation.

Initially, dirhodium tetraacetate (Rh2(OAc)4), the most common catalyst for the preparation of cyclopropanes, was applied. To our surprise, the application of Rh2(OAc)4 did not lead to the desired product neither in dichloromethane nor in toluene (Table 1, entries 1 and 2). Switching the catalyst to copper(I) iodide in refluxing DCM, did not result in the formation of product 6a, as well (Table 1, entry 3). However, when CuI was used in boiling toluene, 42% of the diazo reagent 5 was converted to 6a after only 1 hour of stirring. Thus, to increase the conversion rate, the reaction time was prolonged up to 3.5 h. Indeed, after this time, complete conversion of the diazo reagent 5 was observed and cyclopropane 6a was isolated in 74% yield (Table 1, entry 6). In addition, in a catalyst-free reaction under UV irradiation, no reaction occurred and the starting diazo compound 5 was recovered (Table 1, entry 7).

Having identified the optimal catalyst and conditions, we then examined the scope of the cyclopropanation reaction by reacting selected styrenes with the diazo reagent 5. As evident from Scheme 3, this method is widely applicable to a range of styrenes that are either not activated or substituted with electron-donating (Me, OMe) or electron-withdrawing (Cl) groups on the aromatic ring. All reactions were carried out successfully and the corresponding cyclopropanes 6a–i were isolated in moderate to good yields. As expected, the best results were obtained with styrenes bearing electron-donating groups in the para-position furnishing cyclopropanes 6a and 6d in 74% and 67% yield, respectively. The presence of substituents in ortho and meta-positions on the phenyl ring (6e, 6f, 6h, 6i), decelerated the process. In the case of α-methylstyrene, complete conversion of diazo reagent 5 to compound 6b was observed only after 48 hours of heating, which might result from the steric hindrance around the reaction centre.

Scheme 3: Scope of the cyclopropanation. Reaction conditions: alkene (0.15 mmol), diazo compound 5 (0.1 mmol), CuI (1 mol %), dry toluene, 111 °C, Ar atmosphere. aYields refer to isolated products; bdr ratio determined by 19F NMR spectroscopy.

In almost all cases, a mixture of two diastereoisomers was obtained, with a slight preference for one diastereoisomer for compounds 6a, 6c–i. Furthermore, we were able to isolate both the main isomer as well as a mixture of two isomers for cyclopropanes 6a, 6c, 6d, 6g, 6h and 6i. To determine the relative stereochemistry of the main isomer, the 19F,1H-HOESY NMR spectrum of compound 6c was recorded. The spectrum shows direct correlation of one fluorine nucleus of the difluoromethyl phosphonate group with two cyclopropane protons at 1.75 and 2.91 ppm, respectively. Thus, it is conceivable that the major diastereoisomer adopts trans configuration (Figure 1).

Figure 1: 19F,1H-HOESY spectrum of compound 6c.

Furthermore, terminal aliphatic alkenes as well as N-Boc-allylamine were subjected to [2 + 1] cycloaddition reactions with the carbene precursor 5. The results are summarized in Scheme 4. As observed, the procedure was successful with a series of terminal olefins with different aliphatic chains. The total conversion of the diazo compound 5 was achieved after 2.5–3.5 hours of heating and the corresponding cyclopropanes 7a–g were obtained in moderate to good yield (28–53%). The highest yield (53%) as well as the best diastereoselectivity were recorded for the reaction with tert-butyl-N-allyl carbamate 7g (1:6) while the reaction with allylcyclohexane resulted with the lowest yield (7f, 28%), probably due to the steric hindrance of the reagents. In the case of N-Boc-allylamine, improved diastereoselectivity might result from the coordination of the amino group to the copper centre of an intermediately produced metallocarbene, thus favouring the formation of one diastereoisomer of 7g [50]. In addition, the cyclopropanes 7a–g were always obtained as mixtures of two diastereoisomers in different ratios and all attempts to separate them using column chromatography were unsuccessful. In addition, alkenes such as allylpentafluorobenzene, diethyl allylmalonate, allylbenzene and ethyl acrylate did not react with 5. Instead, the diazo compound 5 decomposed immediately under the reaction conditions described in Scheme 4.

Scheme 4: Scope of the cyclopropanation. Reaction conditions: alkene (0.15 mmol), diazo compound 5 (0.1 mmol), CuI (1 mol %), dry toluene, 111 °C, Ar atmosphere. aYields refer to isolated products; bdr ratio determined by 19F NMR spectroscopy.

To better understand the mechanism and to confirm the lack of selectivity during the cyclopropanation process with terminal alkenes, the reaction mechanism between the diazo reagent 5 and styrene as a model substrate in the presence of CuI catalyst was investigated by density functional theory (DFT) calculations (Table 2).

Table 2: Change in Gibbs free energy ΔG (kcal∙mol−1) from the CuI-catalyzed cyclopropanation of the diazo compound with styrene for possible stereoisomers Pr1 to Pr4.

Isomer Int1 TS1 Int2 TS2a Int3 TS3 Int4 TS4 Pr 1 5.2 16.4 −8.7   −26.4 −22.1 −52.3 −41.7 −44.6 2 8.5 −27.6 −22.5 −50.5 −42.0 −42.5 3   – – −51.1 −36.0 −44.1 4   – – −51.3 −35.9 −43.3

aExemplarily calculated for one option due to flat potential.

In the first step, CuI adds to the diazo compound 5 under formation of zwitterionic Int1 (ΔG = 5.2 kcal/mol). Afterwards, copper carbene complex Int2 is formed after extrusion of nitrogen. The transition state of this metal carbene formation TS1 was calculated with an activation free energy of 16.4 kcal/mol (Scheme 5).

Scheme 5: Addition of CuI to the diazo compound 5.

Int2 adds to styrene to the carbon atom in 2-position of the ethenyl group. At this point, four different orientations of the phenyl group are conceivable (Scheme 6 and Scheme 7, TS2_1 to TS2_4).

Scheme 6: Possible addition of styrene to Int2 yielding Int4_1 and Int4_2 through Int3_1 and Int3_2.

Scheme 7: Possible addition of styrene to Int2 yielding Int4_3 and Int4_4 without further intermediates.

After an early transition state with hardly any activation barrier, the addition of styrene to Int2 proceeds in the case of TS2_1 and TS2_2 to the intermediates Int3_1 and Int3_2. Afterwards, the Cu–C bond is broken, and a Cu–O bond is formed. CuI is transferred to the oxygen atom from the phosphonate group, yielding Int4_1 and Int4_2. In the case of TS2_3 and TS2_4, the addition proceeds smoothly towards Int4_3 and Int4_4 without further intermediates (Scheme 7 and Scheme 8). Extrusion of a catalyst yields Pr1, Pr2, Pr3 and Pr4 as presented in Scheme 8.

Scheme 8: Formation of the products Pr1 to Pr4.

As shown in Table 2, the formation of the enantiomeric set Pr1 and Pr3 (ΔG = −44.6 and −44.1 kcal·mol−1) is slightly more favorable than the formation of Pr2 and Pr4 (ΔG = −42.5 and −43.3 kcal·mol−1; cis-6c), respectively. This is consistent with the experimental results, since two sets of signals corresponding to the diastereoisomers trans-6c and cis-6c are always observed in the 19F NMR spectrum of the crude mixture, with a slight preference for the trans isomer (diastereomeric ratio 1.4:1). Moreover, although Int4 is the overall thermodynamic minimum of the reaction for all stereoisomers, it is still unsurprising that the reaction proceeds towards the product since the abstraction of the catalyst is the last step. Once abstracted, the catalyst is then involved in the next catalytic cycle and thus removed from the equilibrium between Int4 and Pr + CuI. In addition, since TS2 is an early transition state and the potential is concomitantly very flat, only TS2_2 was found by means of regular optimization towards a first order saddle point. For the other three possible reaction pathways, this step was examined by relaxed potential energy scans along the indicated C–C bond (see Supporting Information File 1 for full experimental data).