Synthesis of Imide and Amine Derivatives via Deoxyamination of Alcohols Using N-Haloimides and Triphenylphosphine

Abstract

A deoxyamination methodology of activated and unactivated alcohols is presented. The reaction is mediated by phosphonium intermediates generated in situ from N-haloimides and triphenylphosphine. The protocol allows for the synthesis of phthalimide and amine derivatives in moderate to good yields at room temperature. A series of NMR experiments have provided insight into the reactive intermediates involved and the mechanism of this deoxyamination reaction.

Introduction

Amines are ubiquitous in pharmaceuticals, and biologically active natural products.^[1,2] Of particular interest, primary amines are found in various FDA approved drugs^[2a] and other bioactive compounds (Scheme 1A). Traditionally, synthesis of primary amines is performed via substitution reactions between ammonia and alkyl halide electrophiles using excess ammonia to avoid poly-alkylation of the amine (Scheme 1B). The Gabriel synthesis uses phthalimide as the nitrogen source to avoid polyalkylations; deprotection of the phthalimide using hydrazine affords the primary amine of interest. Another way to avoid poly-alkylation, is the use of stable precursors such as alcohols (Scheme 1C); the Mitsunobu reaction can couple alcohols and phthalimide through the formation of phosphonium intermediates generated using azodicarboxylate reagents.^[3–8] A downside of the traditional Mitsunobu reaction is its limited substrate scope (nucleophiles with pKa≤10) and the toxic and potentially explosive nature of azodicarboxylate reagents. In previous work, Froyen showed that replacing azodicarboxylates with N-bromosuccinimide (NBS) as the activating agent enabled the coupling of alcohols to amines to afford the deoxyaminated products in moderate to good yields.^[9]

Scheme 1.

(A) Examples of pharmaceuticals containing primary amines, (B) Substitution reaction and Gabriel synthesis. (C) The Mitsunobu reaction and Froyen’s deoxyamination. (D) Proposed approach that enables incorporation of the activating agent into the product.

Unfortunately, both the Mitsunobu reaction and Froyen’s NBS transformation have poor atom economy as they use an activating agent solely for the purpose of generating the phosphonium species. As such, the azodicarboxylate and the succinimide byproducts of those reactions are discarded increasing the environmental footprint of the reaction. With this in mind, and given the importance of generating primary amine precursors in the form of phthalimide-containing compounds, we contemplated the possibility of modifying Froyen’s methodology so that the imide byproduct could be incorporated into the final product as the nitrogen source in lieu of adding an external amine as a coupling partner. Such an approach would make the N-haloimide agent play a dual role of i) activating agent and ii) pronucleophile.

Based on our previous work with phosphonium species reacting with carboxylic acids to form amides,^[10] we hypothesized that if a base additive was introduced in our reaction conditions we could force the phosphonium species, generated in situ, to first react with an alcohol to generate a leaving group, and then force the imide counter anion into performing a substitution reaction that generates the desired phthalimide product (Scheme 1D).

Results and Discussion

Initial investigations were performed with benzyl alcohol 1a as our model substrate, Ph₃P, and N-chlorophthalimide (NCphth) 2a as both the activating agent and the pronucleophile (Table 1). Cesium carbonate (Cs₂CO₃) was used as base and freshly distilled DMF was used as the solvent (Table 1, Entry 1). Gratifyingly, the desired product was obtained in 86% NMR yield, with an isolated yield of 78%. Increasing reaction temperature (Entries 2–3) resulted in negligible change in yield. However, the reaction was observed to be sensitive to changes in solvent (Entries 4 & 5). Ethereal solvents completely inhibit the desired reactivity and generate some of the aminated THF,^[11] but anhydrous toluene afforded the desired product albeit in lower yields. Using phosphines other than Ph₃P did not provide better yields, and the use of P(Cy)₃ led to a complete shutdown of reactivity (Entries 6 & 7). Similarly, using bases other than Cs₂CO₃ had a deleterious effect on yield, presumably due to poorer solubility and side reactivity (Entries 8–10). Bases that can react with NCphth, such as LiOtBu, were observed to compete with Ph₃P leading to decomposition of the N-haloimide.^[11] Lastly, a series of control screens show that the base and the phosphine reagents are required to effectuate the desired reactivity (Entries 11 & 12).

Table 1.

Optimization of Reaction Conditions^[a].

Entry	Base	Phosphine	Solvent	T [°C]	Yield [%][b]
1	Cs2CO3	Ph3P	DMF	r.t.	90 (86)[c]
2	Cs2CO3	Ph3P	DMF	60	88
3	Cs2CO3	Ph3P	DMF	80	79
4	Cs2CO3	Ph3P	THF	r.t.	-
5	Cs2CO3	Ph3P	Toluene	r.t.	65
6	Cs2CO3	P(o-toyl)3	DMF	r.t.	62
7	Cs2CO3	P(Cy)3	DMF	r.t.	-
8	K2CO3	Ph3P	DMF	r.t.	19
9	LiOtBu	Ph3P	DMF	r.t.	39
10	K3PO4	Ph3P	DMF	r.t.	63
11	-	Ph3P	DMF	r.t.	-
12	Cs2CO3	-	DMF	r.t.	-

^[a].Reaction conditions: 1a [0.185 mmol, 1 equiv], 2a [0.27 mmol, 1.5 equiv], base [0.27 mmol, 1.5 equiv], phosphine [0.27 mmol, 1.5 equiv], freshly distilled anhydrous solvent [1 mL], room temperature was 25°C.

^[b].1H-NMR yields using dibromomethane as internal standard.

^[c].Isolated yield.

With the optimized conditions in hand, we initiated an alcohol scope study using NCphth 2a as our aminating source (Scheme 2). Both electron donating and electron withdrawing groups in the para position were well tolerated for benzyl alcohols. Methyl, methoxy, and phenyl groups provided the desired products 3b, 3c, and 3d in good yields (70–76%) with several substrates showing comparable yields to traditional Mitsunobu approaches.^[12a–b] All aromatic halogens provided the desired products in good yields (61–75%), showing tolerance to mild electronic changes within the aromatic ring and thereby mirroring previous Mitsunobu reactivity (Scheme 2, products 3e–3h).^[12c–d] Piperonyl alcohol provided the desired product 3i in 72% yield, illustrating the protocol’s compatibility with acetals. 2-Naphthalenemethanol provided product 3j in 78% yield while and 1-naphthalenemethanol gave product 3k in 67% yield, both well within range of analogous Mitsunobu protocols.^[12e] Using a tertiary alcohol failed to provide the desired product 3l strongly suggesting that the reaction proceeds via an S_N2 mechanism and not a carbocation intermediate. 1-Phenyl-1-pentanol, provided product 3m in 50% with the remaining mass generating the styrene byproduct via a competitive elimination process which is less frequent in similar Mitsunobu reactions.^[12f] Low yields were obtained when 2-pyridinemethanol was used to generate products 3n (32%). The para acetoxy product 3o was obtained at 33%, presumably as a result of base induced hydrolysis of the ester motif. Other activated alcohol substrates such as allylic trans cinnamyl alcohol gave product 3p (20%) and propargylic alcohols provided products 3q–3s in 21%, 29%, and 50%, respectively. Initial attempts to make unactivated alcohols undergo the transformation failed to provide the desired product. But further optimization (see supporting information for full table of optimization) showed that switching base from Cs₂CO₃ to K₃PO₄ and increasing the temperature to 70°C enabled the formation of products 3t–3w (Scheme 2). As such, 2-phenylethanol generated the desired amination product 3t in 78% yield over its thermodynamically favored styrene elimination product. 4-Phenylbutanol also provided the desired product 3u (83%). These yields are comparable to the yields obtained using traditional Mitsunobu reaction conditions,^[12g–h] but with greater atom economy due to the dual role of the N-chlorophthalimide as activating agent and pronucleophile. Finally, product 3v was also obtained in excellent yield (85%), while secondary unactivated alcohols afforded product 3w in poor yield (11%) due to increased steric hindrance.

Scheme 2. Alcohol scope.

Reaction conditions: unless stated otherwise, reactions were performed with [100 mg] of alcohol, NCPhth [1.5 equiv], PPh₃ [1.5 equiv], Cs₂CO₃ [1.5 equiv], in freshly distilled anhydrous DMF [3 mL], stirred at room temperature for 12–16 h. All yields are isolated. [a]. Reaction was stirred at 60°C for 48 h. [b]. Unactivated alcohols were reacted using K₃PO₄ [2 equiv] as a base, stirred overnight at 70°C in freshly distilled anhydrous DMF, [0.4 M] concentration.

Various N-haloimides were then investigated as additional nitrogen sources (Scheme 3). N-halosuccinimides were observed to react in modest to good yields (44–73%) for compounds 4a, 4b, and 4c. N-bromoimides were less efficient nitrogen sources, presumably due to unwanted interactions with the solvent.^[13] Gratifyingly, N-chlorosuccinimide (NCS) gave the desired aminated product from primary aliphatic alcohols at a nearly quantitative yield (Scheme 3, 4d, 99%). Increased yields were obtained reacting benzyl alcohols with N-chloro-3,3-dimethyl-glutarimide, affording products 4e, 4f, and 4g (56%, 43%, and 80% respectively), providing potential access to molecular analogues of some stimulants of the central nervous system.^[14] Reaction of N-chloroglutarimide with unactivated alcohol afforded product 4h in lower yields (46%). Our protocol was capable of producing isatin derivative, which are known to be cytotoxic,^[15] in moderate to good yields starting from commercially available 1-chloro-2,3-indoledione (products 4i and 4j).

Scheme 3. N-haloimide Scope.

[a]. Reactions were performed with [100 mg, 1 equiv] of alcohol, NCPhth [1.5 equiv], PPh₃ [1.5 equiv], Cs₂CO₃ [1.5 equiv], in freshly distilled anhydrous DMF [3 mL], stirred at room temperature for 12–16 h. [b.] Unactivated alcohols were reacted using K₃PO₄ [2 equiv] as a base, stirred overnight at 70°C in freshly distilled anhydrous DMF, [0.4 M] concentration.

Inspired by Froyen’s work^[9] and based on our observation that N-chloroimides provided better yields than their bromo counterparts, we hypothesized that employing NCS as an activator in place of NBS might improve the efficiency of Froyen’s original methodology for the coupling of primary and secondary amines. After a brief optimization (see supporting information), we discovered that stirring NCS and PPh₃ in the presence of benzyl alcohol and 3 equivalents of amine in DMF at room temperature afforded product yields comparable to Froyen’s procedure but with a wider substrate scope of primary amines.

Using our NCS/PPh₃ procedure benzylamine was successfully coupled to benzyl alcohol (Scheme 4) to afford product 6a in 76% yield. Electron-rich and electron-poor benzyl alcohols afforded products 6b and 6c in 60% and 69% yield, respectively. Cyclohexylamine was also coupled to 4-methylbenzyl alcohol in good yield (6d, 80%). Other primary amines coupled to alcohols in moderate to good yields affording products 6e and 6f in 61% and 82% yields, respectively. Secondary cyclic amines were also well tolerated under the reaction conditions; both pyrrolidine and morpholine, common motifs in biologically active compounds,^[16] reacted efficiently and afforded products 6g and 6h in 78% yield, which were comparable to Froyen’s original NBS protocol. Secondary alcohols however showed similar limitations to the methodology presented above; coupling 1-phenylethan-1-ol to 4-methylpiperidine afforded the product 6i in a modest 23% yield. Overall our results illustrated that an NCS-based coupling system maintained the efficiency of the analogous NBS technology and also help expand the reaction scope. To illustrate the utility of this amine coupling protocol, we synthesized the anti-Parkinson’s agent Piribedil from piperonyl alcohol and 2-(piperazin-1-yl)pyrimidine (Scheme 4, 6j) at a modest 48% yield, providing a complementary synthetic method to previous approaches.^[17]

Scheme 4. Free Amine Coupling Via NCS Alternative Activator.

Reaction conditions: All were performed with [0.83 mmol, 1 equiv] of alcohol 1, NCS [1.5 equiv], PPh₃ [1.5 equiv], and amine 5 [3 equiv], all dissolved in DMF [3 mL], stirred overnight at room temperature.

We conducted several mechanistic investigations to better understand the reactive species and intermediated generated throughout the reaction. In our previous work,^[10] we showed that mixing N-haloimides in presence of Ph₃P generates a mixture of chloro-phosphonium (A) and imido-phosphonium (B) salts (Figure 1). We hypothesized that performing the reaction in DMF would lead to the formation of DMF adduct C as proposed in the literature.^[18a,b] Using ³¹P NMR studies, we started introducing DMF directly to our standard PPh₃/NCPhth mixture (Figure 1). After adding 1 equivalent of DMF a new peak at 16.9 ppm was observed, which was presumed to be a Vilsmier-Heck phosphonium C. As we increased the concentration of DMF (Figure 1; 3 equiv. to 5 equiv.) we observed increased formation of C with a proportional decrease in the imido-phosphonium B and complete removal of the chlorophosphonium A. This suggests that in DMF both phosphonium species A and B are completely consumed to form species C. Therefore, species C is the reactive species that reacts with alcohols to generate the alkoxy-phosphonium intermediate D that undergoes displacement to afford the desired product and generate triphenylphosphine oxide. In toluene, however, which was also shown to be a suitable solvent, species A and B would be the reactive species.

Figure 1. ³¹P NMR detection of phosphonium species.

DMF [1–5 equiv], PPh₃ [0.405 mmol, 1.5 equiv], NCPhth [0.405 mmol, 1.5 equiv]. All materials dissolved in CDCl₃.

Additional mechanistic investigations were guided by previous reports that NCS in presence of Ph₃P reacts with alcohols to provide the chlorinated products;^[19] reactivity similar to the Appel Reaction. Given the similarity of reagents, we hypothesized that our own deoxyamination might proceed, at least partially, through a chlorinated intermediate that is then displaced by the amide anion. To investigate these assumptions, ¹H NMR time studies of the model substrate were performed to see what products are generated throughout the course of the reaction (Figure 2). At the 30 minutes time frame three distinct products were observed: benzyl alcohol (4.61 ppm, BnOH) starting material, benzyl chloride (4.68 ppm, BnCl) intermediate, and benzyl phthalimide (4.85 ppm, BnNPhth). This clearly indicates that deoxychlorination partially occurs under the reaction conditions. At 90 minutes, a decrease in the peak intensity of BnOH as well as a proportional increase in signal intensity of BnCl and BnNPhth is observed. Once the reaction had reached completion, a complete consumption of BnOH is observed as well as a change in signal ratio between BnCl and BnNPhth from 1:1 to 1:5. These results suggest that both chlorination and amination compete as nucleophilic displacements steps of the oxyphosphonium intermediate, but the presence of base in our reaction conditions facilitates reaction of the phthalimide with the benzyl chloride and or alkyloxyphosphonium intermediates to generate the final desired product.

Figure 2. Detection of chlorination products.

Each reaction was run using BnOH [0.185 mmol, 1 equiv], NCPhth [0.278 mmol, 1.5 equiv], PPh₃ [0.278 mmol, 1.5 equiv], Cs₂CO₃ [0.278 mmol, 1.5 equiv], dissolved in anhydrous DMF [1 mL] and stopped and worked-up at the shown time intervals. CDCl₃ was used for ¹HNMR analysis.

Based on the experimental results above, we propose a mechanism (Scheme 5) beginning with the activation of the N-haloimides via PPh₃ to generate the chlorophosphonium A as well as imidophosphonium B. In DMF, A and B are consumed to generate intermediate C. Reaction of species C with alcohols in the presence of a base leads to the formation of alkoxy-phosphonium D. Thereafter, intermediate D can react through three pathways; 1) D directly reacts with phthalimide anion to generate desired final imide product F and generate OPPh₃; 2) D reacts with the amine substrate to generate desired final amine product G and generate OPPh₃; and 3) D first reacts with chloride to generate the chlorinated byproduct E, which can further react with phthalimide anion or amine substrate to give the final products F and G, respectively.

Scheme 5.

Proposed Reaction Mechanism.

Conclusion

In summary we have developed a diazo-free deoxyamination of alcohols using PPh₃ and N-haloimides in which the imide section of the activating agent is incorporated into the final product as the nitrogen source. Our method allows for the construction of new C–N bonds via the direct amination of activated and unactivated alcohols with imide coupling partners as well as free amines when NCS is used as the activating agent. Products containing the phthalimide moiety can be deprotected to the primary amine using hydrazine or methyl-hydrazine.^[20] We conducted ¹H/³¹PNMR studies of our reaction in order to identify noteworthy reactive species and intermediates to provide a detailed picture of the transformation and its mechanism.