Efficient Copper-Catalyzed Multicomponent Synthesis of N-Acyl Amidines via Acyl Nitrenes.

Direct synthetic routes to amidines are desired, as they are widely present in many biologically active compounds and organometallic complexes. N-Acyl amidines in particular can be used as a starting material for the synthesis of heterocycles and have several other applications. Here, we describe a fast and practical copper-catalyzed three-component reaction of aryl acetylenes, amines, and easily accessible 1,4,2-dioxazol-5-ones to N-acyl amidines, generating CO2 as the only byproduct. Transformation of the dioxazolones on the Cu catalyst generates acyl nitrenes that rapidly insert into the copper acetylide Cu-C bond rather than undergoing an undesired Curtius rearrangement. For nonaromatic dioxazolones, [Cu(OAc)(Xantphos)] is a superior catalyst for this transformation, leading to full substrate conversion within 10 min. For the direct synthesis of N-benzoyl amidine derivatives from aromatic dioxazolones, [Cu(OAc)(Xantphos)] proved to be inactive, but moderate to good yields were obtained when using simple copper(I) iodide (CuI) as the catalyst. Mechanistic studies revealed the aerobic instability of one of the intermediates at low catalyst loadings, but the reaction could still be performed in air for most substrates when using catalyst loadings of 5 mol %. The herein reported procedure not only provides a new, practical, and direct route to N-acyl amidines but also represents a new type of C-N bond formation.

Introduction

Amidines are useful substrates with a variety of applications. The n class="Chemical">amidine structure in 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) is used frequently as a non-nucleophilic base in synthetic chemistry,[1] and more complex amidines are even considered to be “superbases”.[2] Amidines also have been used as ligands for organometallic complexes[3] and as starting material for the synthesis of N-heterocycles.[4−10] The structure itself is present in various natural products and biologically active compounds such as janoxepin, pentamidine, and sildenafil (see Figure ).[11,12]

Figure 1

Several amidine compounds and their applications.

Multiple methods exist for the synthesis of amidines in general, including the Pinner reaction,[13] substitution of phenoxyn class="Chemical">imidates,[14] Beckman-type rearrangements,[15] and dehydrogenative coupling reactions,[16,17] or by catalysis with palladium[14,18−21] or nickel,[22] as well as some multicomponent reactions.[23−33] Chang developed strategies for the synthesis of sulfonated and phosphorylated amidines via traditional click chemistry with azides, followed by a thermal N2 exclusion, resulting in a net coupling of a nitrene to an alkyne.[29,34−45] However, the general (kinetic) stability of most triazoles puts severe limitations on the applicability of other azides, and this method also proved to be unsuitable for the synthesis of N-benzoyl or N-acyl amidines.[45,46] Nicasio and Pérez showed that a copper-catalyzed reaction of acetylenes with N-aroyl or N-acyl azides produces triazoles or oxazoles instead of N-acyl amidines.[47,48] In fact, development of direct methods for the synthesis of N-acyl amidines remains an unsolved challenge. Current synthetic routes to N-acyl amidines usually rely on reactions of premade primary amides with amino acetals[49] or acylation of amidines with stoichiometric coupling reagents.[50−52] A direct, atom-efficient synthetic strategy is thus desired.

N-Acyl n class="Chemical">amidines are a particularly interesting class of compounds. They are important substructures in a variety of medicines[53,54] and have been used as precursors for cyclization reactions,[50,53] as ligands for catalysts and other molecular assemblies,[54] or as substrates for reductive alkylation.[55] We therefore focused on developing a new method for the direct synthesis of N-acyl amidines via a one-pot multicomponent reaction.

Since our group is focusing on C–H activation using metallon class="Chemical">carbenes and metallonitrenes for the catalytic formation of new C–C and C–N bonds,[56−62] we considered that a related acyl nitrene transfer strategy could be useful for the synthesis of N-acyl amidines. However, realizing new strategies to convert acyl nitrenes in a selective manner is rather troublesome. While amides are among the most stable nitrogen-containing moieties, acyl nitrenes are intrinsically unstable. Even when bound to a metal, alkyl migration via the Curtius rearrangement generally rapidly leads to an isocyanate (Scheme , left).[63−65] Hence, in the context of seeking protocols to couple acyl nitrenes to C-nucleophiles, such Curtius rearrangements are undesired and thus represent a selectivity problem, in particular for aliphatic acyl nitrenes. The barrier for this rearrangement is lower for aliphatic acyl nitrenes than for aromatic derivatives, and the intrinsic instability of aliphatic acyl azides makes them challenging to isolate and unsuitable for prolonged storage.

Scheme 1

Acyl Nitrene Formation from 1,4,2-Dioxazol-5-ones, Followed by an Undesired Metal-Induced Curtius Rearrangement (Left) versus Desired Nitrene Insertion into an M–Cacetylide Bond (Right)

In addition to the intrinsic instability of acyl nitrenes, efficient formation of these reactive intermediates constitutes an additional challenge. Conventional pren class="Chemical">cursors used for metallonitrene chemistry are iminoiodinanes, haloamines, organic azides, and hydroxylamines,[66] most of which are difficult to prepare and display poor solubility, require elevated temperatures to activate, or are associated with selectivity and sustainability issues. Cyclic nitrene precursors were recently shown to be efficient nitrene precursors,[67] which are more practical. In particular, 1,4,2-dioxazol-5-ones are noteworthy acyl nitrene precursors in this perspective. They are stable until at least 100 °C, have a low activation barrier upon coordination to a metal, and produce only CO2 as a side product.[68] They can be synthesized by two practical steps from acids, esters, or acyl chlorides.[69] Although these properties have been known for over 50 years, this class of substrates only have become popular for applications in catalysis since the work of Bolm in 2014.[70]

Still, the intrinsic instability of N-acyl metallonitrenes generated from these substrates is a major issue. Successful apprn class="Chemical">oaches to use N-acyl metallonitrenes in catalysis typically rely on directed C–H activation or generation of the acylnitrene on a metal center already containing a preactivated co-substrate.[71−77]

Inspired by the work of Wang on n class="Chemical">copper-catalyzed carbene insertion into C–H bonds,[82] we anticipated that acyl nitrenes could be coupled to terminal alkynes as a new type of C–N bond formation. We envisioned that the activation of 1,4,2-dioxazol-5-ones on copper acetylides could readily form an acyl nitrene on copper that should rapidly insert into the Cu–C bond to furnish a C–N bond, avoiding decomposition by the Curtius rearrangement and thus providing a direct route to acylated amidines (Scheme , right).

Results and Discussion

In an initial attempt to couple acyl nitrenes to n class="Chemical">acetylenes, we added dioxazolone 1a to a mixture of copper bromide, chloroform, phenylacetylene (2a), and diisopropylamine (3a). The yellow color, typical for copper acetylide complexes, slowly turned dark green upon addition of the dioxazolone. Analysis of the crude reaction mixture by mass spectrometry suggested that a three-component reaction had taken place.

1H NMR analysis indeed confn class="Chemical">irmed the formation of the three-component reaction product 4aaa in 63% yield (Table ). The reaction thus readily gives access to this N-acyl amidine. To optimize the reaction conditions, we explored the efficiency of the reaction using different Cu(I) catalysts and different solvents (ligands used in this study are shown in Figure ). Most copper salts worked equally well (Table , entries 1–5). The [Cu(MeCN)4]+ salts appeared to work slightly better, which is commonly observed for copper-catalyzed reactions.[83] We also employed some well-defined Cu(I)-ligand complexes, including complexes containing the Xantphos ligand (Xantphos = 4,5-bis(diphenylphosphino)-9,9-dimethylxanthene) (Table , entries 6–11).[84−86] Surprisingly, while most complexes perform poorly, [Cu(OAc)(Xantphos)] gave a very high selectivity. The reaction works well in a variety of different solvents under aerobic conditions at room temperature (Table , entries 12–19). Results obtained in toluene and ethyl acetate show that the reaction is selective in both polar and apolar solvents. In methanol and acetonitrile, the desired product was not observed, which is suggestive of catalyst inhibition by coordination of these solvents to copper. In tetrahydrofuran (THF), however, full conversion toward the product is observed. Excellent yield is obtained with 5 mol % catalyst loading (or more), but the yield drops with lower catalyst loadings (Table , entries 20–24). Notably, when the reaction is diluted 10 times, formation of the product 4aaa was not observed (Table , entry 25). Instead, 1,1-diisopropyl-3-phenethylurea compound 5aa was formed, indicating a Curtius rearrangement of the acyl nitrene. Replacing the Xantphos ligand by 1,2-bis(diphenylphosphino)ethane (dppe) or 2 equiv of PPh3 results in a decreased yield of 4aaa (Table , entries 26 and 27).

Table 1

Optimization of Reaction Conditionsa

no.	catalyst (mol % Cu)	solvent	yield (%)
1	CuCl (10)	CHCl3	55
2	CuBr (10)	CHCl3	63
3	CuI (10)	CHCl3	62
4	[Cu(NCMe)4]BF4 (10)	CHCl3	65
5	[Cu(NCMe)4]PF6 (10)	CHCl3	75
6	[Cu(TpBr3)(NCMe)] (10)	CHCl3	5
7	[Cu(TpMs)(THF)] (10)	CHCl3	0
8	[CuCl(IPr)] (10)	CHCl3	37
9	[CuI(Xantphos)] (10)	CHCl3	0
10	[Cu(OAc)(Xantphos)] (10)	CHCl3	98
11	[Cu(NCMe)2(Xantphos)]BF4 (10)	CHCl3	72
12	[Cu(OAc)(Xantphos)] (10)	DCM	58 (97%b)
13	[Cu(OAc)(Xantphos)] (10)	THF	95
15	[Cu(OAc)(Xantphos)] (10)	EtOAC	92
17	[Cu(OAc)(Xantphos)] (10)	DCE	74
19	[Cu(OAc)(Xantphos)] (10)	toluene	93
20	[Cu(OAc)(Xantphos)] (5)	CHCl3	97
21	[Cu(OAc)(Xantphos)] (2)	CHCl3	43 (94%b)
22	[Cu(OAc)(Xantphos)] (1)	CHCl3	10 (19%b)
23	[Cu(OAc)(Xantphos)] (0.5)	CHCl3	0
24	[Cu(OAc)(Xantphos)] (0.1)	CHCl3	0
25c	[Cu(OAc)(Xantphos)] (5)	CHCl3	0 (98%b)
26	[Cu(OAc)] + dppe (5)	CHCl3	74%b
27	[Cu(OAc)] + 2 PPh3 (5)	CHCl3	82%b

Reaction of 1a (0.5 mmol), 2a (2 equiv), 3a (2 equiv), and catalyst in solvent (1 mL), stirred overnight in a closed 4 mL vial at room temperature in air. Yields are determined by 1H NMR spectroscopy with 1,3,5-trimethoxybenzene as external standard.

Reactions performed in a N2-filled glovebox.

10 mL of chloroform was used.

Figure 2

Ligands used in this work.

Reaction of 1a (0.5 mmol), 2a (2 equiv), 3a (2 equiv), and catalyst in solvent (1 mL), stirred overnight in a closed 4 mL vial at room temperature in an class="Chemical">ir. Yields are determined by 1H NMR spectroscopy with 1,3,5-trimethoxybenzene as external standard.

We hypothesized that aerobic oxidation of one or more reaction intermediates could inhibit catalysis. Indeed, performing the reactions under an N2 atmosphere led to high yields of the product (Table , entries 12, 21, and 25). Surprisingly, in the absence of a catalyst but in the presence of n class="Chemical">amine 3a full conversion of dioxazolone 1a to urea 5aa was observed at room temperature. Further screening (see the SI) demonstrated that secondary amines are able to activate and decompose 1,4,2-dioxazol-5-ones, resulting in the urea (Scheme ). This side reaction and the Curtius rearrangement catalyzed by an oxidized copper complex likely explain the decreased yields when using lower catalyst concentrations under aerobic conditions (Table , entries 21 and 22).

Scheme 2

Desired Catalyzed Reaction vs the Uncatalyzed Side Reaction (Urea Formation)

The copper-catalyzed three-component reaction produces only CO2 as a byproduct. In view of the atom efficiency of the process, we tested the efficacy of this reaction with a lower ratio of phenylacetylene and diisopropylamine relative to the dioxazolone substrate (see the SI).

The use of fewer equivalents of acetylene has a larger effect on the yield than lowering the amount of n class="Chemical">diisopropylamine (56% and 84% yield, respectively). When one equivalent of all reactants was used, we still obtained 70% of the N-acyl amidine. It should be noted that a lower yield is obtained when using just one equivalent of phenylacetylene in combination with a higher concentration of diisopropylamine. This indicates that fast formation of a copper acetylide species is important and that this process must outcompete the diisopropylamine-assisted decomposition of the dioxazolone to obtain high yields of the desired N-acyl amidine.

Scope of the (Xantphos)Cu-Catalyzed Reaction

The catalytic reaction is not limited to the use of phenylethyl dioxazolone 1a (Table ) and actually has a relatively large substrate scope. Longer dioxazolone substrates can be used without a decrease in yield (4baa–4daa), and double bonds are tolerated as well (4eaa). The yield dropped significantly for n class="Chemical">benzyl dioxazolone 1f and cyclohexyl dioxazolone 1g. Presumably, slower activation of the bulkier substrates on the copper acetylide species leads to faster aerobic decomposition of the catalyst. The fact that under an atmosphere of N2 products 4faa and 4gaa are obtained in good yields supports that hypothesis. An oxygen atom at the β-position of the dioxazolone is deleterious for the yield, as 62% of urea 5ia was obtained when 1i was used. We suspect that this is the result of chelation of the substrate or product to copper, inhibiting the formation of the acetylide on the metal. A first attempt to apply para-tolyl acetylene as a terminal alkyne in the copper-catalyzed three-component reaction yielded only 35% of N-acyl amidine 4aba. However, again under a protective atmosphere of nitrogen we observed high yields of the desired product (see Table ). Para-, meta-, and ortho-tolyl acetylene all gave a similar high yield, suggesting a negligible influence of sterics at the acetylene (4aba–4ada). The electron-rich 4-ethynyl anisole gave slightly lower yields (4aea), which could be related to the lower acidity of the acetylene C–H bond, thus leading to less efficient formation of the copper acetylide complex. When using an alkyl acetylene (2j), which is even less acidic, product 4aja was not observed. Electron-withdrawing phenyl acetylenes, on the other hand, were converted with high efficiency (4afa–4aga). Even the use of the heteroaromatic acetylenes 3-ethynylpyridine and 3-ethynylthiophene resulted in efficient formation of the desired products 4aha and 4aja (79% and 98%, respectively).

Table 2

Scope of the [Cu(OAc)(Xantphos)]-Catalyzed Three-Component Reactiona

Reaction of 1x (0.5 mmol), 2y (2 equiv), 3a (2 equiv), and [Cu(OAc)(Xantphos)] (5 mol %) in CHCl3 (1 mL), stirred for 10 min in a closed 4 mL vial at room temperature. Isolated yields are given. *Performed under an N2 atmosphere. #Quantitative formation of product according to 1H NMR with 1,3,5-trimethoxybenzene as internal standard. $The urea was formed as the main product.

Reaction of 1x (0.5 mmol), 2y (2 equiv), 3a (2 equiv), and [Cu(OAc)(Xantphos)] (5 mol %) in CHCl3 (1 mL), stn class="Chemical">irred for 10 min in a closed 4 mL vial at room temperature. Isolated yields are given. *Performed under an N2 atmosphere. #Quantitative formation of product according to 1H NMR with 1,3,5-trimethoxybenzene as internal standard. $The urea was formed as the main product.

Next, we explored the scope of using different amines as coupling partners in reactions catalyzed by n class="Chemical">[Cu(OAc)(Xantphos)] (Table ). We initially were disappointed to see that benzylamine, isopropylamine, and diethylamine did not yield any N-acyl amidine (3b–3d). Dicyclohexylamine (3h), however, did show a very high yield for the formation of the desired product. We hypothesized that the less bulky amines might coordinate too strongly to the copper atom and compete with coordination of the dioxazolone and thereby inhibit the reaction. To verify our assumption, we further explored the effect of changing the steric bulk of the amine. Indeed, the less bulky amines N-ethylisopropylamine 3e and N-methyl-tert-butylamine 3f do not result in any desired product formation, while the bulkier amine N-ethyl-tert-butylamine 3g produced the desired product 4aag in high yield (96%). The change from 3f to 3g might seem abrupt, but note that in amine 3g the ethyl group is pushed away from the tert-butyl group when coordinated to Cu(I), thus making it much bulkier and hence a weaker donor. The combined results suggest that smaller amines poison the catalyst, while the bulkier amines fit poorly in the “Xantphos pocket”, resulting in negligible catalyst inhibition and thus a high product yield (see Figure ). Upon increasing the steric bulk of the amine further, i.e., employing 2,2,6,6-tetramethylpiperidine (3i) as nucleophile, we observed formation of the desired product as the enamine tautomer (4aai, see the SI).

Table 3

Scope of Amines Used for the [Cu(OAc)(Xantphos)]-Catalyzed Three-Component Reactiona

Reaction of 1a (0.5 mmol), 2a (2 equiv), 3z (2 equiv), and [Cu(OAc)(Xantphos)] (5 mol %) in CHCl3 (1 mL), stirred for 10 min in a closed 4 mL vial at rt. Isolated yields for 4aaz are given. *A tautomer (enamine) was isolated.

Figure 3

Proposed catalyst poisoning by small amine substrates.

Mechanistic Investigations

To gain more information about the reaction mechanism, we studied the formation of copper species with 31P NMR spectroscopy, investigated the kinetics of the reaction, and performed supporting density functional theory (DFT) studies.

We used 31P NMR spectroscopy to study the formation of [Cu(CCPh)(Xantphos)] under optimized reaction conditions, both in an class="Chemical">ir and under a nitrogen atmosphere. [Cu(OAc)(Xantphos)] in CDCl3 gave a 31P NMR signal at δ = −17 ppm (Figure A), which remained unchanged upon addition of phenylacetylene or diisopropylamine (Figure B/C). When both reactants were added to the solution of [Cu(OAc)(Xantphos)] in CDCl3 in air, a new signal was observed at δ = +29 ppm, which corresponds to Xantphos-P-dioxide, and a small broad peak at δ = −13 ppm emerged as well (Figure D). Under an N2 atmosphere, we did not observe ligand oxidation and detected a new singlet at δ = −15 ppm, along with a signal at δ = −19 ppm corresponding to free Xantphos (Figure E). Coordination of Xantphos to copper acetylide or salt metathesis of [CuCl(Xantphos)] with lithium phenylacetylide (Figure F) led to the same 31P NMR signal at δ = −15 ppm, supporting the formation of [Cu(CCPh)(Xantphos)]. Rapid formation of copper acetylide complex [Cu(CCPh)(Xantphos)] upon addition of a base to a reaction mixture containing both [Cu(OAc)(Xantphos)] and phenylacetylene is further associated with a characteristic color change from colorless to bright yellow.[87] The influence of air on the yield of the catalytic reactions in the case of lower catalyst loadings or when sterically more demanding dioxazolones are used implies that [Cu(CCPh)(Xantphos)] reacts with water or dioxygen from air.

Figure 4

31P NMR studies on the formation of [Cu(CCPh)(Xant-phos)]. 31P NMR spectra measured under an N2 atmosphere in CDCl3. (A) [Cu(OAc)(Xantphos)]; (B) [Cu(OAc)(Xantphos)] with 40 equiv of phenylacetylene; (C) [Cu(OAc)(Xantphos)] with 40 equiv of iPr2NH; [Cu(OAc)(Xantphos)] with 40 equiv of both phenylacetylene and iPr2NH under (D) air or (E) N2; (F) [Cu(CCPh)(Xantphos)] formed from salt metathesis from [CuCl(Xantphos)] and lithium phenylacetylide.

Together with the formation of the acetylide complex, we could always observe a broad peak at δ = −13 ppm and free Xantphos, which might result from clustering of copper species. Dimers could be observed in mass spectrometry, and indication for the formation of [Cu3(CCPh)3(Xantphos)2] species was obtained by single-crystal X-ray diffraction (low-quality data set; see the SI for a connectivity plot).

To get more insight into the reaction mechanism, we investigated the kinetics of the reaction, following the pressure buildup caused by CO2 release during the reaction in a dedicated closed system as a measure of the substrate conversion over time.[88,89] The observed rates are plotted as a function of the concentrations of the catalyst and each of the substrates 1a, 2a, and 3a (Figure A–D).

Figure 5

Kinetic experiments and X-ray structure of the catalyst interacting with phenyl acetylene. We varied the concentration of the catalyst and all three substrates and calculated the slope of the steepest part in the graph as the reaction rate. These rates have been plotted over the related concentration to give graphs A, B, C, and D. Graph E shows the pressure during the reaction toward 4aaa for [Cu(OAc)(Xantphos)] (red) and [Cu(NCMe)2(Xantphos)]BF4 (blue). F shows the crystal structure obtained by crystallization in the presence of acetylene, showing a H-bonding interaction with the acetate.

Kinetic experiments and X-ray structure of the catalyst interacting with phenyl acetylene. We varied the concentration of the catalyst and all three substrates and caln class="Chemical">culated the slope of the steepest part in the graph as the reaction rate. These rates have been plotted over the related concentration to give graphs A, B, C, and D. Graph E shows the pressure during the reaction toward 4aaa for [Cu(OAc)(Xantphos)] (red) and [Cu(NCMe)2(Xantphos)]BF4 (blue). F shows the crystal structure obtained by crystallization in the presence of acetylene, showing a H-bonding interaction with the acetate.

The reaction is clearly first order in both [catalyst] and [dioxazolone] and (nearly) zero order in [iPr2NH]. The rate dependence of [n class="Chemical">PhCCH] (Figure C) is more complex and suggestive of saturation kinetics. This behavior is most likely caused by a (slightly endergonic) pre-equilibrium between [Cu(OAc)(Xantphos)] and [Cu(CCPh)(Xantphos)]. The existence of such an equilibrium could be confirmed by H/D exchange studies. Burk and co-workers showed with H/D scrambling that copper acetates and copper acetylides are in fast equilibrium.[90] In a similar experiment, using [Cu(OAc)(Xantphos)] as the catalyst, we also observed H/D scrambling of phenylacetylene-D and 4-CF3-phenylacetylene (see the SI for details). Combined with the 31P NMR studies shown in Figure and the small effect of [iPr2NH] on the rate, the data suggest that this pre-equilibrium shifts toward the acetylide complex in the presence of the amine.

From the catalyst screening studies presented in Table , it is clear that the “unsaturated” [Cu(NCMe)2(Xantphos)]BF4 complex is less selective than n class="Chemical">[Cu(OAc)(Xantphos)] for the formation of the N-acyl amidine, indeed confirming that the acetate plays a prominent role in the catalytic cycle. When comparing the conversion over time between the reaction catalyzed by [Cu(OAc)(Xantphos)] and the [Cu(NCMe)2(Xantphos)]BF4 (Figure E), we observed a very large difference in rate, confirming that the acetate is important for rapid formation of the acetylide complex.[91] In fact, the X-ray structure of the [Cu(OAc)(Xantphos)]-HCCPh adduct, obtained by crystallization of [Cu(OAc)(Xantphos)] in the presence of an excess of phenylacetylene, reveals a direct hydrogen-bonding (H-bonding) interaction between the copper-bound κ2-acetate fragment and the acidic proton of the acetylene (Figure F).

Further mechanistic information was obtained using DFT calculations (Figure , see the SI for details). While rapid formation of n class="Chemical">acetylide complex B was observed experimentally during the NMR studies, DFT calculations show that the acetylene activation pre-equilibrium step is endergonic by about +3 kcal/mol. This small discrepancy is likely a medium effect, a result of the reaction being performed under nonstandard concentration conditions or a combination thereof.[92]N-Coordination and subsequent activation of the dioxazolone substrate on the Cu center, leading to CO2 dissociation, over TS1 produces acyl nitrene acetylide intermediate D. Interestingly, two possible transition states need to be considered for the C–N bond formation step, involving acyl nitrene insertion into the Cu–C bond of the acetylide fragment. In TS2-1, the acyl nitrene is κ1-N coordinated and C–N coupling leads to a η1-N coordination of the amide moiety. In TS2-2, the acyl nitrene coordinates in a κ2-N,O fashion and C–N bond formation leads to a chelating coordination of the carbonyl and the triple bond.

Figure 6

DFT (B3LYP-D3-TZ2P)-computed energies of the acyl nitrene activation and insertion.

The transition state barriers of these processes (from D) are very similar (∼7.5 kcal mol–1). The resulting intermediates, E for TS2-1 and F for TS2-2, are different by only 3.5 kcal/mol and connected by a low energy barrier transition state TS3 and thus can both be present during the reaction. The undesired n class="Chemical">Curtius rearrangement from intermediate D was calculated to be ∼3 kcal higher in energy than TS2-1 and TS2-2, which is in line with the experimental observations (see the SI for further details). For the protodemetalation of the vinylideneamide by an incoming equivalent of acetylene, acetic acid, or even diisopropylamine, many pathways can be considered from either E or F. The protodemetalation was therefore not included in the DFT studies.

The combined information gathered from the catalytic studies and the mechanistic investigations led us to the proposed catalytic cycle shown in Scheme . The reaction sequence involves activation of acetylene 2a by complex A to form n class="Chemical">acetylide complex B, most likely assisted by the amine. Subsequently, B activates the dioxazolone to form acyl nitrene intermediate D, which leads to either E or F, followed by protodemetalation to regenerate A and to form vinylideneamide G. Finally, reaction of the amine with the electrophilic intermediate G produces the N-acylamidine product 4.

Scheme 3

Proposed Catalytic Cycle

Expanding the Scope to Benzoyl Amidines Using CuI

As shown in Table , phenyl dioxazolone 1h is not converted by n class="Chemical">[Cu(OAc)(Xantphos)], thus preventing the one-pot three-component synthesis of benzoyl amidines when using this catalyst. However, we argued that it might well be possible to expand the substrate scope of the newly discovered one-pot Cu-catalyzed three-component reaction from alkyl dioxazoles to aryl dioxazolones if we would use a less bulky Cu catalyst. Such reactions are desirable, as (just like for N-acyl amidines) there are currently no practical or high-yielding reactions available for the efficient one-pot (multicomponent) synthesis of benzoyl amidines. Sharpless and Chang published a method to synthesize the corresponding amidine from a benzoyl azide, but the desired compound was obtained in only very low yield (9% based on 1H NMR spectroscopy).[44] Hence, we considered that it remained important to develop new catalytic methods to convert aryl dioxazolones to benzoyl amidines. Therefore, we decided to investigate the reactivity of less bulky, phosphine-free copper salts in reactions using these substrates. Gratifyingly, CuI proved to be a suitable catalyst to extend the protocol to the synthesis of N-benzoyl amidine derivatives. Using 10 mol % of CuI as the catalyst and phenyl dioxazolone as the substrate yielded the desired product in 68% isolated yield (see Table ). This method works for electron-rich as well as electron-poor aryl dioxazolones (4jaa, 4kaa). The reaction still seems to be affected by sterics, but o-phenyl dioxazolone still produced 4laa in 34% isolated yield.

Table 4

Synthesis of N-Benzoyl Amidines from Aryl-1,4,2-dioxazol-5-ones Catalyzed by CuIa

Reaction of 1a (0.5 mmol), 2a (2 equiv), 3z (2 equiv), and CuI (10 mol %) in CHCl3 (1 mL), stirred for 30 min in a closed 4 mL vial at room temperature. Isolated yields are shown.

Conclusion

In this work we have presented a new, fast, efficient, and practical three-component reaction of an acyl nitrene pren class="Chemical">cursor catalyzed by copper to produce N-acylamidines and benzoyl amidines. To the best of our knowledge, this is the first reported catalytic procedure to synthesize these types of products. This atom-efficient reaction uses dioxazolone substrates that are easily accessible and only generates CO2 as byproduct under mild conditions, making this an attractive methodology for both small- and large-scale organic synthesis.