Scope and limitations of the solution and solid phase synthesis of homoallylic amines via N-acyliminium ion reactions

The full scope and limitations of solution and solid phase one-pot three component N- acyliminium ion reactions are detailed. After studying the scope in solution with respect to the carbamate, nucleophile and aldehyde component, a 'translation' was made to the solid phase. The solid phase reactions were eventually carried out using the so-called SEC linker system, which was previously developed in our group. In order to maximize the scope of the nucleophile component, additional studies were successfully conducted using two-step processes involving stable N-acyliminium ion precursors.


Introduction
In the past decade, numerous types of reactions have been 'translated' from the solution to the solid phase, so that solid phase combinatorial chemistry may currently enable the synthesis of various compound libraries of any size. 1 However, solid phase reactions should not simply deliver as many compounds as possible, but rather as 'diverse' compounds as possible. Therefore, in the process of 'translating' solution phase reactions to the solid phase, one should consider that establishing the scope and limitations on the solid phase in terms of variety in functionality, steric and electronic properties must be an integral aspect of this process.
In a series of publications, it was shown by our group 2,3,4,5 and by others 6 that N-acyliminium ion chemistry can be efficiently carried out on a solid phase, provided that an appropriate linker system is used. With the development of the novel polystyrene-based SEC and TEC linker systems, [3][4][5] an adequate immobilization approach was found in our group. By using these linkers, extensive and more systematic investigations in the scope and limitations of solid phase Nacyliminium ion chemistry can be performed. In conjunction with a preliminary publication on one-pot three component N-acyliminium ion reaction for the synthesis of homoallylic amines, 2 this paper will detail a full account of the scope and limitations of this reaction. In the solution phase one-pot three component synthesis of protected homoallylic amines introduced by Panek 7 and Veenstra, 8 studies on the scope of this N-acyliminium ion reaction with respect to the amide and aldehyde components were reported (Scheme 1). These studies showed that a wide variety of aromatic and aliphatic aldehydes could be used in combination with some carbamates and sulfonamides. Electron rich and poor aromatic aldehydes reacted equally well and steric bulk in the aldehyde component did not seem to play a significant role.

Scheme 1
In this contribution, studies to reach a broader scope of the corresponding solid phase three component N-acyliminium ion reaction will be presented to give the corresponding homoallylic amines. 9, For a better understanding of the solid phase reaction, these investigations started with a more detailed scope determination using solution phase model systems. A significant part of the research focused on the use of reactive species (generated in situ or isolated) that subsequently could be used as N-acyliminium ion precursors. During the course of our work, the group of Mioskowski reported a similar type of approach, 10 in which via a three-component reaction from amides, aldehydes and trimethyl orthoformate a stable N,O-acetal was isolated and used as an N-acyliminium ion precursor in a subsequent step. This concept may lead to a 'two- Allyl carbamate 10 and benzotriazole (Bt) derivative 11 were also prepared as model systems for solution phase N-acyliminium ion reactions. The two compounds were prepared according to known literature procedures. Allyl carbamate 10 was obtained in one step from allyl chloroformate using the procedure of Roos et al., 11 while Bt-derivative 11 was obtained in one step from benzyl carbamate using Katritzky technology. 12 Bt-compounds usually exist as a mixture of 1-yl and 2-yl isomers (viz. structures 1-yl-11 and 2-yl-11), both of which display a comparable leaving group ability. 13 However, Bt-derivative 11 was mainly isolated as the pure 1yl-11 isomer. 12 Evidently, allyl carbamate 10 was used as a model system for the allyl carbamate linker, while carbamates 8 and 9 were used as models for the corresponding SEC and TEC linker systems. Benzyl carbamate was used as a simple and readily available model system for a generally immobilized carbamate. To explore the two-step variant of the three component Nacyliminium ion reaction, Bt-derivative 11 was used.
First, the scope of the three component N-acyliminium ion reaction with allyl carbamate 11 was investigated. This carbamate was reacted with 1 equiv of aldehydes 12a-f, nucleophile 3A-F and BF 3 ⋅OEt 2 in CH 2 Cl 2 at rt. The reactions were quenched upon disappearance of the starting material or after a maximum reaction time of 20 h. Several aromatic and aliphatic aldehydes were used in combination with allyltrimethylsilane (3A). Furthermore, benzaldehyde (12a) was used in combination with several nucleophiles (Table 1). a The corresponding diethyl acetal was used.
As shown in the Table, electron rich aromatic aldehydes (entries 1-2) gave the best results, while the electron poor aromatic aldehydes (entries 3-4) resulted in the formation of the desired homoallylic carbamate in moderate yields and longer reaction times. The same was true for the aliphatic substituents (entries [5][6]. With respect to the nucleophile, the scope proved to be significantly narrower. Although the use of the substituted allylsilane 3B (entry 7) resulted in a high yield of the desired product (80%), the reaction was much slower than with allyltrimethylsilane (3A, entry 1). 1,2-Propadienyltributylstannane (3C) has been successfully used previously as a nucleophile in N-acyliminium ion chemistry. 14 However, in this case, the use of the stannane (entry 8) only resulted in the formation of the desired homopropargylic carbamate 13aC in a yield of 10%. The use of Me 3 SiCN (entry 9) and silyl enol ethers (entries 10-11) gave traces of the desired products, although these nucleophiles have previously shown their viability in N-acyliminium ion chemistry. 15,16 The low yields achieved with the nucleophiles 3C-F might be explained by the rather low stability of these compounds under the conditions needed for the formation of the N-acyliminium ion (BF 3 ⋅OEt 2 , rt). The scope of the solution phase one-pot three component N-acyliminium ion reaction with respect to the nucleophile was further investigated using benzyl carbamate (14) as a model system ( Table 2). The 1-substituted allylsilane 3G (entry 2) reacted smoothly and resulted in a 71% yield of homoallylic carbamate 15aG. The geometry of the newly formed double bond was assigned by 1 H NMR experiments and proved to be exclusively (E). Suprisingly, 2-and 3substituted allylsilanes 3H and 3I (entries 3-4) did not lead to any product formation at all. The steric hindrance of the chloride and methyl substituent on the reactive double bond of the allylsilane might explain the poor reactivity of these nucleophiles in this specific N-acyliminium ion reaction. Although the use of 2,3-butadienylsilane 3J in Sakurai-type additions to aldehydes and acetals has been reported by the groups of Hatakeyama and Takano, 17 the use of this nucleophile in N-acyliminium ion reactions has received only recent attention. 18 The application of allenylsilane 3J in the one-pot three component N-acyliminium reaction with benzyl carbamate (14) and benzaldehyde (12a) resulted in the formation of diene 15aJ in a yield of 62%. Analogous to the unsuccessful use of silyl enol ethers (Table 1), at this point the more stable vinyl acetate 3K was used as a nucleophile (entries 6-7). However, under the conditions that were previously used for the formation of the transient cation (1 equiv BF 3 ⋅OEt 2 , entry 6), no product was formed either. Recently, new developments to generate iminium and N-acyliminium ion intermediates by using catalytic amounts of metal triflates (Sc(OTf) 3 , Yb(OTf) 3 , Sn(OTf) 2 , Hf(OTf) 4 , etc.) have been reported by Kobayashi et al. 19 When the Lewis acid was changed to catalytic Sc(OTf) 3 (entry 7), the desired ketone 15aK was obtained, albeit in a low yield of 28%. To gain a better understanding of the 'translation' of the N-acyliminium ion reaction to the SEC and TEC linkers that we eventually would like to apply, the model reactions were extended to carbamates having the sulfonylethyl and thioethyl functionalities already present in the molecule. Thus, carbamates 8 and 9 were used for the one-pot three component reactions (Table 3). The yield of sulfonylethyl carbamate 17aA (89%, entry 1) was comparable with the yield earlier obtained in combination with allyl carbamate 11 (82%, Table 1, entry 1). However, the same components in combination with thioethyl carbamate 9 afforded product 18aA in a somewhat lower yield (65%, entry 3). This finding was comparable to solid phase results obtained with other N-acyliminium ion reactions on the SEC and TEC linker systems. 4 The lower yields with thioethyl carbamate 9 and the corresponding TEC resin might be explained by the presence of the nucleophilic sulfur functionality, which could interfere with the reaction conditions that are required for the N-acyliminium ion reaction. A benzylic substituent was introduced using the diethyl acetal 16f (entry 2), which was used because of the poor stability of the corresponding aldehyde. Logically, this diethyl acetal functionality also generated a more reactive precursor for the N-acyliminium ion formation (an N,O-acetal rather than an N,Ohemiacetal) and thus resulted in the smooth formation of product 17fA in a yield of 82%. The same positive effect of the diethyl acetal functionality was found with the introduction of an nhexyl substituent (entries 4-5). While n-heptanal (12g) produced the product 18gA in a yield of only 45%, the use of the corresponding diethyl acetal 16g improved the yield of 18gA to 66%.
To further investigate the scope with respect to the aliphatic diethyl acetal component, benzyl carbamate (14) and allyltrimethylsilane (3A) were used in combination with the functionalized aliphatic diethyl acetal 16h (Scheme 3). After a reaction time of 18 h, a 75% conversion of benzyl carbamate (14) was observed and during the reaction a white precipitate was formed, which after isolation was identified as bis-carbamate 21. The formation of similar bis-carbamates in this type of reaction was also reported by Veenstra. 8 However, the bis-carbamate precipitate was by them observed at the beginning of the reaction and was thereafter completely transformed into the desired product. In our hands, the bis-carbamate 21 and the desired protected diamine 15hA were obtained as a 1:1 mixture, which could be easily separated using column chromatography. Subsequently, bis-carbamate 21 itself was used as an N-acyliminium ion precursor, which after a reaction of 18 h resulted in a 6:4 mixture of compounds 21 and 15hA. Clearly, the scope of the three component N-acyliminium ion reaction in particular with respect to the nucleophile is still rather limited. Encouraged by the results of using more activated intermediates for the N-acyliminium ion formation, it was decided to further investigate the 'two-step' approach. In contrast to Mioskowski's results with aromatic amides, 10 in our hands the addition of aldehydes to benzyl carbamate in the presence of TFA and HC(OMe) 3 did not lead to the desired N,O-acetal, but resulted in the exclusive formation of the corresponding bis-carbamates. 20 Therefore, in addition to N,O-acetals, the Bt-derivative 11 was used as a readily available and versatile N-acyliminium ion precursor (Table 5). By using allyltrimethylsilane (3A) as the nucleophile, homoallylic carbamate 15aA was obtained in a good yield (80%, entry 1), thus proving the viability and efficiency of this Btderivative in N-acyliminium ion chemistry. 2,3-Butadienylsilane 3J (entry 2) was reacted with Bt-derivative 11 to provide diene 15aJ, the preparation of which was described earlier in Table 2 (entry 5). In the current case, diene 15aJ was obtained in a slightly lower yield than in the previous example (53% and 62%, respectively). The application of a vinylogous silyl enol ether as a nucleophile was demonstrated by the use of furan 3L (entries 3-4). 21,22 Sc(OTf) 3 proved to be unsuitable for N-acyliminium ion formation with Bt-derivative 11, since it led to no product formation at all. However, with the use of BF 3 ⋅OEt 2 , the desired α,β-unsaturated lactone 15aL was obtained in a moderate yield of 51%. The use of furan as an aromatic nucleophile in Nacyliminium ion chemistry has been shown before, 23 in which the best results were obtained using moderately strong protic acids in combination with furan as the solvent. Thus, Btderivative 11 was reacted with CSA in furan (16 M) to afford compound 15aM in a yield of 55% (entry 5). The introduction of the furan functionality may allow further functionalization by oxidation to the corresponding carboxylic acid functionality. Although a heterogeneous RuO 2 /NaIO 4 system is usually used for the oxidation of α-furyl carbamates, 24 some examples of homogeneous oxidations with ozone are also known. 25 Hence, furan 15aM was ozonolyzed to afford the N-protected phenylglycine derivative 26 in a yield of 79% (Scheme 4). Consequently, with this reaction sequence,the scope of products was extended to Cbz-protected α-amino acids. MeOH, -78 ºC, 30 min 79%

Scheme 4
Next, with the results of the solution phase N-acyliminium ion reactions in mind, the scope and limitations of the same reactions on a solid phase were determined. For this purpose, the earlier developed SEC linker system 23 was used. The introduction of aromatic side chains is presented in Table 6.
a Yields were determined over two steps, starting from resin 23.
In these cases, the yields of the homoallylic amines 25 were determined after cleavage from the resin using a NaOMe solution. Again, the same trend in reactivity of the aromatic aldehydes was observed: electron rich aldehydes afforded the desired products 25 in high yields (entries 1-2, 5), whereas electron poor ones produced a lower yield (entry 3), or no product at all (entry 4).
Furthermore, the yields of these products were around 20% higher than those of the same products produced on the acid labile Wang linker system. 2 This observation again substantiated the viability of the SEC linker for application to N-acyliminium ion chemistry.
Earlier, the introduction of aliphatic side chains in the solid phase one-pot three component N-acyliminium ion reaction on the Wang resin proved to be troublesome. 2 On the other hand, the introduction of aliphatic substituents in the solution phase reaction using acetals 16 did result in the efficient formation of the desired products. Therefore, the use of diethyl acetals in combination with SEC resin 23 was investigated (Table 7).
a Yields were determined over two steps, starting from resin 23.
By using the aliphatic diethyl acetals 16f-g, j-k (entries 1-4), the corresponding products could indeed be obtained, albeit in moderate yields. Surprisingly, with acetal 16h in the solid phase reaction (entry 5) no product was obtained at all, whereas in solution phase (Scheme 3) the desired product was formed together with a substantial amount of the corresponding biscarbamate. Nevertheless, the successful introduction of aliphatic substituents (entries 1-4) significantly extended the scope of the process. Encouraged by these results, the application of the corresponding diethyl acetal functionality in the introduction of an electron poor aromatic substituent was also tested (entry 6). Unfortunately, the desired homoallylic amine 25dA was only formed in a low yield of 7%.
In addition, the application of the different silyl nucleophiles 3G-J in combination with some aromatic aldehydes in the three component N-acyliminium ion reaction was investigated (Table  8).
a Yields were determined over two steps, starting from resin 23.
By using the substituted allylsilanes 3G-I(entries 1-3), the 1-substituted silane 3G again afforded the corresponding product (42%), while the 2-and 3-substituted allylsilanes 3H-I were not reactive at all. A diene functionality could be introduced by making use of allenylsilane 3J (entry 4). Furthermore, silanes 3G and 3J were successfully used in combination with the aromatic aldehydes 12b and 12i (entries 5-8) to afford the corresponding homoallylic amines 25 in reasonable yields.
Finally, to extend the scope of the N-acyliminium ion reaction even more, the 'two-step' approach, involving the immobilized Bt-derivative 26 was applied ( Table 9). The results show that the scope with respect to the nucleophile indeed could be extended in case of aromatic aldehydes (entries 2-3). Analogous to the solution phase reaction, CSA was used to generate the N-acyliminium ion intermediate, while the aromatic nucleophile was used as the solvent. Moreover, the application of the Bt-derivative improved the yield of product 25dA to 36% over three steps (entry 4). Thus, the 'two-step' approach also extended the scope of the reaction with respect to electron poor aromatic aldehydes.
a Yields were determined over three steps, starting from resin 23.
If one compares the results of the solution phase N-acyliminium ion reactions with those of the solid phase, a few interesting features appear. In general, the yields of the solid phase reactions are somewhat lower than the solution phase ones. For example, whereas the use of pnitrobenzaldehyde (12d) in solution phase afforded the desired product in a reasonable yield, the same aldehyde on solid phase only produced a trace of the desired homoallylic amine. Although in the solution phase a similar trend in reactivity was observed, the low reactivity of aliphatic and electron poor aromatic substituents in the solid phase reactions was much more pronounced. This difference might be explained by the polarity of the environment in which the reactions take place: in solution phase, the cationic intermediates can be to some degree stabilized by the relatively polar solvent, while on solid phase the extent of stabilization is reduced as a result of the more apolar environment of the polymer support.
Furthermore, the use of diethyl acetal 16h in solution phase led to a considerable amount of the undesired bis-carbamate (cf. 31 in Scheme 5), while on the solid phase no such product was formed at all. Veenstra

Scheme 5
In conclusion, the scope and limitations of the one-pot three component N-acyliminium ion reaction have been explored. In both the solution and solid phase reactions a similar trend can be observed. With respect to the aldehyde component, the best yields were obtained with electron rich aromatic ones. Whereas aliphatic side chains could be introduced in satisfactory yields by using the corresponding diethyl acetals, electron poor aromatic substituents could not efficiently be introduced by using the three component N-acyliminium ion reaction. However, a 'two-step' approach involving a Bt-derivative has led to an extension of the scope to also include electron poor aromatic groups. With respect to the nucleophile, the scope was mainly restricted to simple substituted allylsilanes. A useful addition of potential nucleophiles was found in the application of 2,3-butadienylsilane. Furthermore, aromatic nucleophiles were used in the 'two-step' approach. Overall, it can be concluded that a diverse range of products was made, although the introduction of different structural classes required its own optimization, which has led to a range of optimal reaction conditions.

Experimental Section
General Procedures. All reactions with air or moisture sensitive reagents were carried out under an inert atmosphere of dry nitrogen. Standard syringe techniques were applied for the transfer of air or moisture sensitive reagents and dry solvents. The solid phase reactions were gently stirred with a magnetic stirring bar. The resins were washed according to the indicated sequence and dried in vacuo (50 ºC) prior to use. The resin was allowed to swell/shrink for at least 1 minute before each filtration. Infrared (IR) spectra were obtained from KBr pellets or neat, using a Bruker IFS 28FT-spectrometer with wavelengths (ν) reported in cm -1 . IR spectra of resins were 2-Benzylthioethanol (6). 26 Benzyl bromide (7.21 mL, 42.7 mmol), and Cs 2 CO 3 (14.6 g, 44.9 mmol) were suspended in dry DMF (150 mL), mercaptoethanol (25.0 mL, 44.9 mmol) was added dropwise and the reaction mixture was stirred for 4 h at 60 ºC and then for 20 h at rt. Then, saturated aqueous NH 4 Cl (100 ml) was added and the layers were separated. The aqueous layer was extracted with CH 3 CCl 3 (2 × 75 mL), the combined organic layers were washed with aqueous saturated NaCl (75 mL), dried (MgSO 4 ) and concentrated in vacuo to obtain 6 (7.23 g, 40.2 mmol, 94%) as a colorless oil, which was used without any further purification. 1

General procedure A for the three component reaction with carbamates 8-10 and 14
The carbamate, aldehyde (1 equiv) and nucleophile (1 equiv) were dissolved in CH 2 Cl 2 . The Lewis acid (1 equiv) was added and the reaction mixture was stirred for the indicated time at rt. Then, saturated aqueous NaHCO 3 was added and the layers were separated. The aqueous layer was extracted with CH 2 Cl 2 (2 ×), the combined organic layers were washed with aqueous saturated NaCl, dried (MgSO 4 ) and concentrated in vacuo. The crude product was purified using column chromatography (EtOAc/PE).

General procedure B for the N-acyliminium ion reaction with carbamate 11
Carbamate 11 8 and the nucleophile were dissolved in CH 2 Cl 2 . BF 3 ⋅OEt 2 was added and the reaction mixture was stirred for the indicated time at rt. Then, aqueous Na 2 CO 3 (10%) was added and the layers were separated. The aqueous layer was extracted with CH 2 Cl 2 (2 ×), the combined organic layers were washed with aqueous saturated NaCl, dried (MgSO 4 ) and concentrated in vacuo. The crude product was purified using column chromatography (EtOAc/PE).