Highly Diastereoselective Chelation-Controlled 1,3- anti -Allylation of ( S )-3-(Methoxymethyl)hexanal Enabled by Hydrate of Scandium Triﬂate

: En route to the total synthesis of (+)-Neopeltolide, we explored Lewis acid-assisted diastereoselective allylation of MOM-protected 3-hydroxylhexanal with β -(2,2-diethoxyethyl)-substituted (allyl)tributylstannane. The hydrated form of scandium triﬂate was found to be essential for attain-ing high 1,3- anti -diastereoselectivity (d.r. 94:6), while the use of anhydrous catalyst resulted in a modest diastereocontrol (d.r. 76:24). The preferred 1,3- anti -selectivity in this transformation can be rationalized in the framework of the Reetz chelate model of asymmetric induction. The 1,3- anti conﬁguration of the product was conﬁrmed by its conversion into the known C 7 -C 16 building block of (+)-Neopeltolide. We also report an improved protocol for the synthesis of β -(2,2-diethoxyethyl)-substituted (allyl)tributylstannane, which can be utilized as a cost-efﬁcient bipolar isoprenoid-type C 5 -building block in the synthesis of natural compounds.


Introduction
Stereoselective allylation of carbonyl compounds allows to assemble a carbon-carbon bond along with installation of a new stereocenter [1][2][3][4]. The produced homoallylic alcohols provide multiple opportunities for the subsequent modifications and therefore are widely used in the target-oriented synthesis of natural and bioactive compounds [2][3][4][5].
During the implementation of our research programs devoted to the synthesis of natural and bioactive compounds from cyclopropanols [49][50][51][52] and cyclopropanol-derived build-ing blocks [53][54][55][56], we expected to develop a bifunctional allylation reagent A (Scheme 1) based on metalation of easily available allyl bromide 1 [53]. The reagent A can act as a synthetic equivalent of a bipolar isopentane synthon, as it was previously demonstrated by the synthesis of retinoid compounds via the Barbier-type chemistry [53,56]. We envisioned that besides the assembly of polyene scaffolds, organometallic derivatives of 1, especially its organotin derivative 2, could also be suitable for the stereoselective allylation of carbonyl compounds and therefore applied in the asymmetric synthesis of natural products. Our preliminary tests revealed that organotin compound 2 [57], along with its carboxymethyl analogue [54,55], are suitable for highly enantioselective Keck allylations. However, the substrate-controlled stereoselective coupling of 2 with oxy-functionalized aldehydes has not been examined. Moreover, we required to develop an expedient synthetic protocol for the preparation of 2 in multigram amount. As a result of our endeavors, here we report a convenient and cost-efficient procedure for multigram preparation of 2, and its application in the Lewis acid-mediated diastereoselective 1,3-anti-allylation of (S)-3-(methoxymethyl)hexanal 3. The stereochemical outcome of the reaction was further validated by the synthesis of known C 7 -C 16 bulding block of (+)-Neopeltolide [55], containing three stereocenters. y 2021, 13, x FOR PEER REVIEW 2 of 17 functionalized analogues [8,14,[36][37][38][39]. The development of more complex allyl-transfer reagents is appealing in view of their evidently high synthetic value [40][41][42][43][44][45][46][47][48].
During the implementation of our research programs devoted to the synthesis of natural and bioactive compounds from cyclopropanols [49][50][51][52] and cyclopropanol-derived building blocks [53][54][55][56], we expected to develop a bifunctional allylation reagent A (Scheme 1) based on metalation of easily available allyl bromide 1 [53]. The reagent A can act as a synthetic equivalent of a bipolar isopentane synthon, as it was previously demonstrated by the synthesis of retinoid compounds via the Barbier-type chemistry [53,56]. We envisioned that besides the assembly of polyene scaffolds, organometallic derivatives of 1, especially its organotin derivative 2, could also be suitable for the stereoselective allylation of carbonyl compounds and therefore applied in the asymmetric synthesis of natural products. Our preliminary tests revealed that organotin compound 2 [57], along with its carboxymethyl analogue [54,55], are suitable for highly enantioselective Keck allylations. However, the substrate-controlled stereoselective coupling of 2 with oxy-functionalized aldehydes has not been examined. Moreover, we required to develop an expedient synthetic protocol for the preparation of 2 in multigram amount. As a result of our endeavors, here we report a convenient and cost-efficient procedure for multigram preparation of 2, and its application in the Lewis acid-mediated diastereoselective 1,3-anti-allylation of (S)-3-(methoxymethyl)hexanal 3. The stereochemical outcome of the reaction was further validated by the synthesis of known C 7 -C 16 bulding block of (+)-Neopeltolide [55], containing three stereocenters. Scheme 1. Preparation of a bifunctional allylation reagent A and outline of the current work.

Improved Protocol for the Preparation of Functionalized (Allyl)Tributylstannane 2
Multistep synthesis commonly requires substantial quantity of starting materials at the initial stages. Therefore, a facile and cost-efficient access to large quantities of 2 was of primary importance. Using the advantages of cyclopropanol chemistry [58,59], allyl bromide 1 was readily prepared in multigram amounts and 94% overall yield from cheap and easily available ethyl 3,3-diethoxypropionate (4) via the consequent Kulinkovich cyclopropanation [60], mesylation, and MgBr2-mediated cyclopropyl-allyl rearrangement steps (Scheme 2) [53,[61][62][63]. The reaction sequence was flawlessly performed in a single run starting from 20 g of ester 4 (see the experimental part). No purification was required for the cyclopropane intermediates 5 and 6, which were obtained in nearly quantitative yields. Scheme 1. Preparation of a bifunctional allylation reagent A and outline of the current work.

Improved Protocol for the Preparation of Functionalized (Allyl)Tributylstannane 2
Multistep synthesis commonly requires substantial quantity of starting materials at the initial stages. Therefore, a facile and cost-efficient access to large quantities of 2 was of primary importance. Using the advantages of cyclopropanol chemistry [58,59], allyl bromide 1 was readily prepared in multigram amounts and 94% overall yield from cheap and easily available ethyl 3,3-diethoxypropionate (4) via the consequent Kulinkovich cyclopropanation [60], mesylation, and MgBr 2 -mediated cyclopropyl-allyl rearrangement steps (Scheme 2) [53,[61][62][63]. The reaction sequence was flawlessly performed in a single run starting from 20 g of ester 4 (see the experimental part). No purification was required for the cyclopropane intermediates 5 and 6, which were obtained in nearly quantitative yields.
The previously reported method [57] for the preparation of organotin compound 2 via Barbier-type coupling of 1 and Bu 3 SnCl was found impractical for multigram preparation. Following the previously reported approach, organotin compound 2 was obtained in a moderate 66% yield due to accompanying homo-coupling of 1 leading to 7. The purity of 2 was also unsatisfactory because of the contaminant inorganic salts. Incompatibility of acid-sensitive 2 with silica gel made its chromatographic purification impossible [57]. The previously reported method [57] for the preparation of organotin compound 2 via Barbier-type coupling of 1 and Bu3SnCl was found impractical for multigram preparation. Following the previously reported approach, organotin compound 2 was obtained in a moderate 66% yield due to accompanying homo-coupling of 1 leading to 7. The purity of 2 was also unsatisfactory because of the contaminant inorganic salts. Incompatibility of acid-sensitive 2 with silica gel made its chromatographic purification impossible [57].

Scheme 2.
Synthesis of allyl bromide 1 and its conversion into organotin derivative 2.
Therefore, we tested an alternative procedure of halogen substitution in 1 with Bu3SnLi (Scheme 2) [14]. To our delight, the new approach delivered the target organotin compound 2 in affordable 78% yield and noticeably better purity. According to 1 H NMR analysis, the content of homo-coupled product 7 was reduced to 8 mol.%, along with the presence of 15 mol.% of (Bu3Sn)2 dimer. These impurities do not interfere the reactivity of 2 and can be removed after the performing the allylation reaction. The same transformation was also convenient for multigram preparation (up to 7 g in a single run, see the experimental part). Hence, high yields and utilization of cheap chemicals have provided a convenient, scalable, and cost-effective access to 1 and 2 in the sufficient amounts.

Diastereoselective Allylation of Aldehyde 3 with (Allyl)Tributylstannane 2
While examining the potential routes towards the synthesis of (+)-Neopeltolide and its analogues [64][65][66], we attempted to perform the stereoselective allylation of MOMprotected 3-hydroxylhexanal 3 with (allyl)tributylstannane 2. The aldehyde 3 (ee > 99%) was prepared by following the known procedure [55] (see the Supplementary Materials). Initially, we planned to apply the venerable Keck asymmetric allylation, by using a catalytic system based on titanium tetraisopropoxide and a chiral BINOL ligand [57,[67][68][69]. Although being a well-developed approach, the Keck reaction has several restrictions, such as allylation of unsaturated or sterically hindered aldehydes [46,[67][68][69][70]. Moreover, the presence of multiple oxygen-containing functionalities in both 2 and 3 could interfere the reaction outcome due to the highly oxophilic nature of the titanium catalyst. In our hands, allylation of 3 with 2 by following the Keck reaction protocol has led to only trace amounts of the desired homoallylic alcohol 8 (Scheme 3) after an exhausting search for the optimal reaction conditions and even in the presence of trifluoroacetic acid [67,69]  Therefore, we tested an alternative procedure of halogen substitution in 1 with Bu 3 SnLi (Scheme 2) [14]. To our delight, the new approach delivered the target organotin compound 2 in affordable 78% yield and noticeably better purity. According to 1 H NMR analysis, the content of homo-coupled product 7 was reduced to 8 mol.%, along with the presence of 15 mol.% of (Bu 3 Sn) 2 dimer. These impurities do not interfere the reactivity of 2 and can be removed after the performing the allylation reaction. The same transformation was also convenient for multigram preparation (up to 7 g in a single run, see the experimental part). Hence, high yields and utilization of cheap chemicals have provided a convenient, scalable, and cost-effective access to 1 and 2 in the sufficient amounts.

Diastereoselective Allylation of Aldehyde 3 with (Allyl)Tributylstannane 2
While examining the potential routes towards the synthesis of (+)-Neopeltolide and its analogues [64][65][66], we attempted to perform the stereoselective allylation of MOMprotected 3-hydroxylhexanal 3 with (allyl)tributylstannane 2. The aldehyde 3 (ee > 99%) was prepared by following the known procedure [55] (see the Supplementary Materials). Initially, we planned to apply the venerable Keck asymmetric allylation, by using a catalytic system based on titanium tetraisopropoxide and a chiral BINOL ligand [57,[67][68][69]. Although being a well-developed approach, the Keck reaction has several restrictions, such as allylation of unsaturated or sterically hindered aldehydes [46,[67][68][69][70]. Moreover, the presence of multiple oxygen-containing functionalities in both 2 and 3 could interfere the reaction outcome due to the highly oxophilic nature of the titanium catalyst. In our hands, allylation of 3 with 2 by following the Keck reaction protocol has led to only trace amounts of the desired homoallylic alcohol 8 (Scheme 3) after an exhausting search for the optimal reaction conditions and even in the presence of trifluoroacetic acid [67,69] or B(OMe) 3 [71] as activating additives (Table 1, entry 1).  3 pre-dried in vacuum at heating afforded the same d.r. f The reaction was performed with 11.5 mmol of aldehyde 3 and 1.7 equiv. of 2 in the presence of 1.1 equiv. of the hydrated Sc(OTf) 3 . Isolated yield of anti-8 is given in parentheses.
First, we tried to carry out the reaction with TiCl 4 , which is known as an effective catalyst for the chelation-controlled Reetz-Keck-type allylation [28][29][30][31][32]. Unfortunately, a complex mixture of products was formed (Table 1, entry 2). Tin(IV) chloride behaved similarly (entry 3), while the less reactive titanium catalysts failed to furnish any products at all (entries 4 and 5). Magnesium bromide as another prominent catalyst [31,32] delivered the desired homoallylic alcohol 8, albeit with moderate diastereoselectivity (entry 6). The ratio of diastereomers was determined by 1 H NMR analysis of the crude reaction mixture, by integration of signals at δ 2.98 (d, J = 2.9 Hz) and 3.09 (d, J = 2.1 Hz) ppm, which correspond to hydroxyl protons of antiand syn-8, respectively. Analogously to MgBr 2 , zinc and zirconium(IV) chlorides also produced 8 with unsatisfactory diasteroisomeric ratios (entries 7 and 8).
While testing different metal triflates (entries 9-13), we found that allylation of 3 occurred with promising yield and diastereoslectivity in the presence of scandium(III) triflate [72][73][74][75]. The reaction mediated by Sc(OTf) 3 was especially successful in toluene as solvent (entry 10), while triflates of indium, ytterbium, and hafnium were noticeably less efficient (entries [11][12][13]. During these preliminary tests we also noticed that the stereochemical outcome of the reaction with Sc(OTf) 3 and the reactivity of the catalyst were strongly depended on the catalyst batch. While the allylation with an old reagent did not occur at −70 • C and required higher temperature (−25 • C), a fresh sample of commercial Sc(OTf) 3 , as well as the catalyst dried in vacuum at heating, were much more reactive and We surmised that the difference in reactivity between the batches can be explained by hydration of the old reagent with atmospheric moisture since Sc(OTf) 3 is hygroscopic and eventually forms octahydrate upon storage. Indeed, powder X-ray diffraction (PXRD) analysis of the old and new reagent confirmed our hypothesis and showed that the old reagent contained Sc(OTf) 3 ·8H 2 O as the main phase ( Figure 1). solvent (entry 10), while triflates of indium, ytterbium, and hafnium were noticeably less efficient (entries [11][12][13]. During these preliminary tests we also noticed that the stereochemical outcome of the reaction with Sc(OTf)3 and the reactivity of the catalyst were strongly depended on the catalyst batch. While the allylation with an old reagent did not occur at −70 °C and required higher temperature (−25 °C), a fresh sample of commercial Sc(OTf)3, as well as the catalyst dried in vacuum at heating, were much more reactive and delivered the target alcohol 8 already at −70 °C but with noticeably lower 76:24 d.r. (entry 10 vs. 14).
We surmised that the difference in reactivity between the batches can be explained by hydration of the old reagent with atmospheric moisture since Sc(OTf)3 is hygroscopic and eventually forms octahydrate upon storage. Indeed, powder X-ray diffraction (PXRD) analysis of the old and new reagent confirmed our hypothesis and showed that the old reagent contained Sc(OTf)3•8H2O as the main phase (Figure 1).  [76]. Slight hydration occurred since the sample was exposed to atmospheric moisture during the measurement. (c) PXRD patterns of the hydrated reagent, which contains Sc(OTf)3•8H2O as the main phase and trace amount of Sc(OTf)3•xH2O phase.
To our delight, controlled addition of water to the anhydrous Sc(OTf)3 allowed to prepare a catalyst with reproducible performance (see the experimental part), similar to those of the old batches (entries 15 and 16). The best outcome and the highest diastereoselectivity was observed when 2 equiv. of water was added. Moreover, we found that addition of diethyl ether as a co-solvent to toluene (ca. 20% v/v) allows to decrease the reaction temperature to −70 °C and therefore further improve the diastereoselectivity (up to 91:9, entry 17). Finally, anti-alcohol 8 was prepared in 72% isolated yield and with excellent 94:6 diastereomeric purity in a preparative reaction run starting from 11.5 mmol of aldehyde 3 (entry 18). It is important to note that at least 1.1 equiv. of hydrated Sc(OTf)3 must be used to attain high yields, probably due to the presence of several oxygen-containing functionalities in 8 and formation of a stable chelate complex with scandium.  3 . The main phase corresponds to Sc(OTf) 3 ·xH 2 O (x < 8) [76]. Slight hydration occurred since the sample was exposed to atmospheric moisture during the measurement. (c) PXRD patterns of the hydrated reagent, which contains Sc(OTf) 3 ·8H 2 O as the main phase and trace amount of Sc(OTf) 3 ·xH 2 O phase.
To our delight, controlled addition of water to the anhydrous Sc(OTf) 3 allowed to prepare a catalyst with reproducible performance (see the experimental part), similar to those of the old batches (entries 15 and 16). The best outcome and the highest diastereoselectivity was observed when 2 equiv. of water was added. Moreover, we found that addition of diethyl ether as a co-solvent to toluene (ca. 20% v/v) allows to decrease the reaction temperature to −70 • C and therefore further improve the diastereoselectivity (up to 91:9, entry 17). Finally, anti-alcohol 8 was prepared in 72% isolated yield and with excellent 94:6 diastereomeric purity in a preparative reaction run starting from 11.5 mmol of aldehyde 3 (entry 18). It is important to note that at least 1.1 equiv. of hydrated Sc(OTf) 3 must be used to attain high yields, probably due to the presence of several oxygen-containing functionalities in 8 and formation of a stable chelate complex with scandium.
Our results indicate that controlled hydration of Sc(OTf) 3 can be considered as a tool to attenuate the reactivity of Sc(OTf) 3 in allylation of carbonyl compounds, and perhaps in other transformations mediated by the same Lewis acid. Scandium(III) triflate has multiple catalytic uses in organic synthesis and can operate even in aqueous media [72][73][74][75]. However, the influence of small amounts of water on the catalytic performance of Sc(OTf) 3 , especially in stereoselective transformations, has been only scarcely reported, to the best of our knowledge [77][78][79].
The preferred 1,3-anti-selectivity in this transformation can be rationalized in the framework of the Reetz chelate model [16,17,[28][29][30][31][32]. We assume that the reaction could proceed through the formation of six-membered chelate intermediates I and II (Scheme 4) [9]. Intermediate II is less preferred than I, since n-propyl substituent and MOM-protecting group are both occupy axial positions in the former. Moreover, the reaction of I with allylstannane 2 leading to anti-alcohol 8 should proceed faster since the reagent 2 approaches from the least hindered side of the carbonyl group, as shown on Scheme 4. On the contrary, in complex II, both sides of the carbonyl group are sterically shielded with axial n-propyl and MOM substituents, which should result in higher activation barrier for the reaction of II with 2 in comparison with those of I. Although a mechanistic rationale for the improved d.r. in the case of hydrated catalyst is not fully clear, aqua ligands coordinated to scandium should introduce additional steric hindrances thus further enhancing the difference in reactivity between I and II and shifting the equilibrium towards the less sterically hindered intermediate I.
Having alcohol anti-8 in hand, we confirmed its stereochemical configuration by conversion into the known C 7 −C 16 building block of (+)-Neopeltolide (Scheme 5) [55]. Thus, cyclization of 8 in the presence of catalytic amount of p-toluenesulfonic acid (p-TsOH) and subsequent mild hydrolysis led to the formation of lactol 9. Oxidation of 9 with pyridinium chlorochromate (PCC) followed by base-catalyzed isomerization of the double bond yielded unsaturated lactone 10. Hydrogenation of 10 resulted in the formation of saturated lactone 11 as a single diastereoisomer and in a quantitative yield. Reduction of 11 with LiAlH4 afforded diol 12. To confirm 1,3-anti configuration of the stereocenters, diol 12 was transformed into the corresponding acetonide 13. The values of chemical shifts in 13 C NMR spectrum of 13 indicated the 1,3-anti-configuration of hydroxyl groups (Scheme 5), according to a known configuration assignment method [80,81]. The selective TBS-protection of the primary hydroxyl group in compound 12 and the subsequent methylation of the secondary hydroxyl produced ether 14 in 92% yield over two steps. Treatment of 14 with catalytic amount of pyridinium p-toluenesulfonate (PPTS) in methanol Intermediate II is less preferred than I, since n-propyl substituent and MOM-protecting group are both occupy axial positions in the former. Moreover, the reaction of I with allylstannane 2 leading to anti-alcohol 8 should proceed faster since the reagent 2 approaches from the least hindered side of the carbonyl group, as shown on Scheme 4. On the contrary, in complex II, both sides of the carbonyl group are sterically shielded with axial n-propyl and MOM substituents, which should result in higher activation barrier for the reaction of II with 2 in comparison with those of I. Although a mechanistic rationale for the improved d.r. in the case of hydrated catalyst is not fully clear, aqua ligands coordinated to scandium should introduce additional steric hindrances thus further enhancing the difference in reactivity between I and II and shifting the equilibrium towards the less sterically hindered intermediate I.
Having alcohol anti-8 in hand, we confirmed its stereochemical configuration by conversion into the known C 7 -C 16 building block of (+)-Neopeltolide (Scheme 5) [55]. Thus, cyclization of 8 in the presence of catalytic amount of p-toluenesulfonic acid (p-TsOH) and subsequent mild hydrolysis led to the formation of lactol 9. Oxidation of 9 with pyridinium chlorochromate (PCC) followed by base-catalyzed isomerization of the double bond yielded unsaturated lactone 10. Hydrogenation of 10 resulted in the formation of saturated lactone 11 as a single diastereoisomer and in a quantitative yield. Reduction of 11 with LiAlH 4 afforded diol 12. To confirm 1,3-anti configuration of the stereocenters, diol 12 was transformed into the corresponding acetonide 13. The values of chemical shifts in 13 C NMR spectrum of 13 indicated the 1,3-anti-configuration of hydroxyl groups (Scheme 5), according to a known configuration assignment method [80,81]. The selective TBS-protection of the primary hydroxyl group in compound 12 and the subsequent methylation of the secondary hydroxyl produced ether 14 in 92% yield over two steps. Treatment of 14 with catalytic amount of pyridinium p-toluenesulfonate (PPTS) in methanol and subsequent Swern oxidation furnished aldehyde 15 in 92% yield as a final C 7 -C 16 building block of (+)-Neopeltolide.

General Experimental Methods
Solvents were used as obtained from commercial sources without any further purification or dried if required over 4 Å molecular sieves. Chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA), Fluorochem (London, UK) and Alfa Aesar (Ward Hill, Mass, USA) and used as received unless other indicated. (S)-3-(Methoxymethoxy)hexanal (3) (ee > 99%) was prepared as described before [55] (see also the Supplementary Materials). Silica gel 40-100 μm was used for column chromatography; silica gel 60F254 plates were used for TLC. 1 H-NMR (400 MHz), 13 C-NMR (100.6 MHz) spectra were recorded on Avance III spectrometer (Bruker, Billerica, Mass, USA). Chemical shifts are given in δ value with CHCl3 (δ = 7.26 ppm) and CDCl3 (δ = 77.0 ppm) as internal standards for 1 H-NMR and 13 C-NMR spectra, respectively. FT-IR spectra were recorded on a Bruker Tensor 27 FT spectrometer. Only selected characteristic IR absorption bands are given. Specific rotations were measured by using an Anton Paar MCP 500 polarimeter. HRMS data were obtained on a HPLC/Q-TOF G6540A Mass Spectrometer (Agilent, Santa Clara, CA, USA) using AJS ESI method in positive ion detection modes or a LTQ Orbitrap Discovery spectrometer (Thermo Fisher Scientific, Waltham, Mass, USA) using electrospray ionization (ESI). Powder X-ray diffraction (PXRD) patterns for the samples of Sc(OTf)3 were recorded with an EMPYREAN diffractometer (PANalytical, Netherlands) using Cu-Kα radiation (Ni-filter) at 296 K with area detector 2θ range of ca. 0°−40°. Samples of Sc(OTf)3 were not protected from atmospheric moisture during the measurements. Crystallographic data for Sc(OTf)3•8H2O are available from the Cambridge Structural Database (CSD 415177) [82].

General Experimental Methods
Solvents were used as obtained from commercial sources without any further purification or dried if required over 4 Å molecular sieves. Chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA), Fluorochem (London, UK) and Alfa Aesar (Ward Hill, MA, USA) and used as received unless other indicated. (S)-3-(Methoxymethoxy)hexanal (3) (ee > 99%) was prepared as described before [55] (see also the Supplementary Materials). Silica gel 40-100 µm was used for column chromatography; silica gel 60F 254 plates were used for TLC. 1 H-NMR (400 MHz), 13 C-NMR (100.6 MHz) spectra were recorded on Avance III spectrometer (Bruker, Billerica, MA, USA). Chemical shifts are given in δ value with CHCl 3 (δ = 7.26 ppm) and CDCl 3 (δ = 77.0 ppm) as internal standards for 1 H-NMR and 13 C-NMR spectra, respectively. FT-IR spectra were recorded on a Bruker Tensor 27 FT spectrometer. Only selected characteristic IR absorption bands are given. Specific rotations were measured by using an Anton Paar MCP 500 polarimeter. HRMS data were obtained on a HPLC/Q-TOF G6540A Mass Spectrometer (Agilent, Santa Clara, CA, USA) using AJS ESI method in positive ion detection modes or a LTQ Orbitrap Discovery spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) using electrospray ionization (ESI). Powder X-ray diffraction (PXRD) patterns for the samples of Sc(OTf) 3 were recorded with an EMPYREAN diffractometer (PANalytical, Netherlands) using Cu-Kα radiation (Ni-filter) at 296 K with area detector 2θ range of ca. 0 • -40 • . Samples of Sc(OTf) 3 were not protected from atmospheric moisture during the measurements. Crystallographic data for Sc(OTf) 3 ·8H 2 O are available from the Cambridge Structural Database (CSD 415177) [82]. aqueous solution of NH 4 Cl (38 mL) was added by small portions at vigorous stirring. The mixture was additionally stirred at r.t. for 20 min and filtered. The precipitate was washed with CH 2 Cl 2 (3 × 150 mL), and the combined organic phases were washed with a saturated aqueous solution of NaCl (150 mL) and dried over Na 2 SO 4 . The solvent was removed under reduced pressure to afford cyclopropanol 5 (18.3 g, 100%) as a yellowish oil. The obtained compound was used directly in the next step; additional purification was not required.

4,4-Diethoxy-2-methylenebutyl Bromide (1)
A solution of 1,2-dibromoethane (19.0 mL, 220 mmol) in anhydrous diethyl ether (50 mL) was added slowly in a dropwise manner to magnesium turnings (4.8 g, 200 mmol) in anhydrous diethyl ether (100 mL). The reaction mixture was additionally stirred until the complete dissolution of magnesium occurred. Then, a solution of the crude mesylate 6 (26.5 g, 105 mmol) in anhydrous diethyl ether (120 mL) was added dropwise over 15 min at room temperature to the obtained solution of MgBr 2 and the resulting mixture was vigorously stirred for 2 h. Afterwards, water (200 mL) was cautiously added by small portions at external cooling (ice-water bath, 0 • C). The organic layer was separated, and the aqueous layer was extracted with diethyl ether (3 × 100 mL). The combined organic phases were washed with saturated NaHCO 3 solution (150 mL) and dried over Na 2 SO 4 . The solvent was removed under reduced pressure, and allyl bromide 1 was purified by using a short chromatographic column (SiO 2 , eluent PE:EtOAc, 50:1). Yellowish oil (23.4 g, 94%).

Preparation of Tributyl(4,4-diethoxy-2-methylenebutyl)stannane (2)
A solution of naphthalene (0.16 g, 1.25 mmol) in THF (2 mL) was added to lithium chipping (0.60 g, 86 mmol) in THF (48 mL) under inert atmosphere (argon). The mixture turned green and was stirred at room temperature for 1 h. Then, tributyltin chloride (6.50 mL, 24.0 mmol) was added dropwise, and the mixture was stirred at room temperature for 12 h. The resulting dark-green solution of Bu 3 SnLi was transferred into another reaction vessel and cooled to −78 • C (acetone-dry ice bath). A solution of allylbromide 1 (4.74 g, 20.0 mmol) in THF (40 mL) was added dropwise and the resulting reaction mixture was stirred at −78 • C for 1 h. The reaction was quenched by addition of saturated aqueous solution of NH 4 Cl (150 mL) and stirred while warming to room temperature for 1h. The organic layer was separated, the aqueous layer was extracted with Et 2 O (3 × 50 mL), and the combined organic phases were washed with saturated NaHCO 3 solution (50 mL) and dried over Na 2 SO 4 . The solvent was removed under reduced pressure to afford the crude stannane 2 (9.3 g). According to 1 H NMR analysis, the crude product contains 7 g (77 mol.%, 78% yield) of 2, contaminated with self-coupling products 7 (0.5 g, 8 mol.%) and Bu 3 SnSnBu 3 (1.8 g, 15 mol.%). However, the crude stannane 2 can be used for the allylation of aldehydes without any further purification.