Intramolecular Hydroamination of Unbiased and Functionalized Primary Aminoalkenes Catalyzed by a Rhodium Aminophosphine Complex

Abstract

We report a rhodium catalyst that exhibits high reactivity for the hydroamination of primary aminoalkenes that are unbiased toward cyclization and that possess functional groups that would not be tolerated in hydroaminations catalyzed by more electrophilic systems. This catalyst contains an unusual diaminophosphine ligand that binds to rhodium in a κ³-P,O,P mode. The reactions catalyzed by this complex typically proceed at mild temperatures (room temperature to 70 °C), occur with primary aminoalkenes lacking substituents on the alkyl chain that bias the system toward cyclization, occur with primary aminoalkenes containing chloride, ester, ether, enolizable ketone, nitrile, and unprotected alcohol functionality, and occur with primary aminoalkenes containing internal olefins. Mechanistic data imply that these reactions occur with a turnover-limiting step that is different from that of reactions catalyzed by late transition metal complexes of Pd, Pt, and Ir. This change in the turnover-limiting step and resulting high activity of the catalyst stem from favorable relative rates for protonolysis of the M-C bond to release the hydroamination product vs reversion of the aminoalkyl intermediate to regenerate the acyclic precursor. Probes for the origin of the reactivity of the rhodium complex of L1 imply that the aminophosphine groups lead to these favorable rates by effects beyond steric demands and simple electron donation to the metal center.

Introduction

The hydroamination of unactivated alkenes is a conceptually simple method for the preparation of alkylamines.^1-5 The intramolecular version of this process is an attractive route to nitrogen-containing heterocycles, a common structural motif in biologically active molecules,⁶ from readily available aminoalkenes. Due to the utility of this reaction in academic, pharmaceutical, and industrial settings, much effort has been spent on developing catalysts to effect this transformation.

Specifically, complexes based on rare-earth,^7-10 alkali-earth,^11,12 and Group IV^13-19 metals are known to catalyze the intramolecular hydroamination of primary and secondary aminoalkenes; however, these catalysts are often highly air and moisture sensitive and incompatible with many functional groups. For example, aminoalkenes containing acidic protons, such as unprotected alcohols, and carbonyl functionality, such as ketones and esters, have not been reported to undergo hydroamination with the highly basic rare-earth, and alkali-earth metal catalysts.^7,13,20 In a similar vein, group IV metal catalysts have not catalyzed hydroaminations of aminoalkenes containing carbonyl or ether functionalities. The group IV catalyst that reacts with the broadest scope was reported recently by Schafer and is based on a tethered ureate ligand.¹⁹ Even this catalyst, however, has been shown to tolerate a limited range of functionality (an aromatic dioxolane, an N-alkyl pyrrole, and a tertiary aromatic amine), and substrates containing functional groups are all secondary amines with a bias toward cyclization. All primary aminoalkenes that react with this catalyst have lacked additional functionality, and all reported reactions of primary aminoalkenes lacking gem-diphenyl substituents with this catalyst were conducted at 145 °C.

Catalysts derived from zinc complexes^21,22 and Bronsted acids^23-26 have also been shown to catalyze intramolecular hydroamination with a limited range of reactivities. For example, the reaction of a secondary aminoalkene lacking gem-substitution in the presence of a zinc catalyst derived from diethylzinc and a protic additive required 21 days at 180 °C. Primary aminoalkenes containing gem-disubstitution of the alkyl linker required 120 °C, and a single example of a substrate lacking gem-disubstitution on the alkyl chain (2-amino-hex-5-ene) gave the product in only 19% conversion after 36 h.²¹ Hydroaminations catalyzed by Brønsted acids required high temperatures or protection of the nitrogen with an electron-withdrawing group.^24-26

Complexes of late-transition metals are more stable to air and moisture and are more tolerant of polar functional groups than complexes containing early transition metals and lanthanides.²⁷ However, intramolecular hydroaminations catalyzed by late transition metal complexes often require an N-H donor in which the nitrogen is electron-deficient and contained in an amide, carbamate or sulfonamide functionality.^26,28-31 In the last five years, complexes of platinum,^32,33 copper,³⁴ rhodium,^35-37 and iridium^38-40 have been shown to catalyze the intermolecular hydroamination of aminoalkenes containing more basic, secondary alkylamines. In 2008, the author's group reported a cationic rhodium complex that catalyzed the cyclization of both secondary and primary aminoalkenes.³⁵

Although data with this original catalyst³⁵ and that with catalysts based on copper,³⁴ rhodium,³⁷ and iridium^40,46 reported since this time demonstrated that primary aminoalkenes can undergo cyclization, reactions of such such amines catalyzed by these systems have significant limitations. First, the cyclization of primary aminoalkenes catalyzed by late-metal systems has been limited to cases that are biased toward cyclization by a Thorpe-Ingold effect; second, these reactions were limited to aminoalkenes containing terminal olefins; third, these reactions required high temperatures (>100 °C); and fourth, reactions of primary aminolkenes containing a second functional group have not been reported with any catalyst. Thus, the identification of a complex containing a late transition metal that catalyzes the cyclization of unactivated, unbiased aminoalkenes, as well as an understanding of the factors controlling the reactions of primary amines catalyzed by late transition metal complexes, is needed.^26,41,42

We previously investigated the cyclization of a secondary aminoalkene catalyzed by a series of rhodium complexes containing bisphosphine ligands.³⁵ One example of the cyclization of a secondary aminoalkene catalyzed by the combination of bis(diethylamino)xantphos ligand L1’⁴³ (Table 1) and a cationic rhodium precursor was included in this study. Complexes of aminophosphines have been investigated rarely as components of catalysts, let alone for hydroamination.⁴⁴ Thus, we investigated the structure of this catalyst and its reactivity toward a broader range of aminoalkenes.

Table 1.

Rh-Catalyzed Hydroamination of Primary Aminoalkenes: Initial Results

entry	R	mol % [Rh]	ligand	solvent	time (h)	conversion (%)a	ratioa2:3:4:5	Isolated yield
1	Ph	1	L1	tBuOH	2	100	18:1:1:0.7	80%
2	Ph	1	L1’	tBuOH	2	100	11:1:1:0.2
3	Ph	1	L1	THF	2	100	14:1:1:0.4	71%
4	H	3	L1	tBuOH	15	100	9:1:1:0.5	66%c
5	H	3	L1’	tBuOH	15	100	12:1:1:0.5
6	H	3	L1	THF	15	77	6:1:0.6:1.6

^aDetermined by ¹H NMR spectroscopy.

^b A trace amount of internal alkene from isomerization observed.

^cA 1.3 : 1 ratio of diastereomers was obtained.

Here we report a full account of our findings that reveal the high reactivity of this unusual metal-ligand system for the hydroamination of primary amines, along with detailed mechanistic studies that reveal the origins of the high reactivity of this system. These new studies show that the active catalyst posseses a rhodium POP-pincer structure and is highly active for the hydroamination of primary amines. The reactions catalyzed by this complex typically proceed at mild temperatures (room temperature to 70 °C), occur with primary aminoalkenes lacking any substituents on the alkyl chain that would bias the system toward cyclization, and occur with a high degree of tolerance for polar functional groups, including aryl chlorides, esters, ethers, enolizable ketones, nitriles, and unprotected alcohols. Initial results imply that the aminophosphine groups on the ligand are involved in creating the fast rates and high selectivity, and mechanistic data show that the reactions of primary amines catalyzed by these complexes occur with a turnover-limiting step that is different from that of secondary aminoalkenes catalyzed by complexes of Pd,⁴⁵ Pt,³² and Ir.⁴⁶

Results and Discussion

Our prior observation that bis(diethylamino)xantphos L1' and Rh(COD)₂BF₄ catalyzed the cyclization of a secondary aminoalkene, the novelty of a diaminophosphine as a component for catalysis, the potential flexibility and ease of synthesis of this class of ligand, and the possibility that this catalyst combination could lead to more hydroaminations than just the cyclization of secondary amines with gem-disubstitution led us to explore the reactivity of this catalyst for aminoalkenes that have resisted cyclization with previous late transition metal catalysts. To do so, we explored the activity of the catalyst generated from L1 and cationic rhodium catalyst precursors for reactions of primary aminoalkenes possessing and lacking gem-disubstitution on the alkyl linker. We found that just 1 mol% of the complex generated from [Rh(CH₃CN)₂COD]BF₄ and L1 catalyzed the hydroamination of primary aminoalkene 1a containing 3,3-diphenyl substituents in THF at 70 °C to form the pyrrolidine product 2a with high selectivity (14:1:1:0) for formation of the cyclic secondary amine product over the combination of imine product 3a and hydrogenation product 4a (Table 1, entry 1) Pure amine was isolated in 71% yield. The pairwise formation of the imine and hydrogenation products would occur by oxidative amination and transfer of the hydrogen to the olefin unit of the starting aminoalkene. The internal alkene 5a, resulting from isomerization of the terminal olefin in the starting material, was observed in only trace amounts. Based on the potential that proton transfer is the turnover-limiting step of the catalytic cycle (vide infra), we also studied reactions in alcohol solvents. The same reaction of 1a occurred with higher selectivity and higher isolated yield in tert-butanol (Table 1, entry 3).

Encouraged by these initial results, we investigated reactions of primary aminoalkenes that have not undergone cyclization with previous late transition metal catalysts. Initial studies conducted on the hydroamination of aminoalkene 1b lacking gem-disubstitution on the alkyl linker in the presence of catalytic amounts of [Rh(CH₃CN)₂COD]BF₄ and L1 showed that the rate and selectivity of reactions in ethereal solvents were significantly lower than those of the gem-disubstituted substrate 1a (Table 1, compare entries 3 and 6). However, the same reaction in tBuOH led to the first cyclization of an unbiased primary amine with a late metal catalyst with acceptable rates and high selectivity (9:1 amine:imine) for the desired amine product (entry 4). The pyrrolidine product 2b was isolated from this reaction in 66% yield as a 1.3:1 mixture of diastereomers (see also Table 2, entry 2).

Table 2.

Scope of Rh-Catalyzed Hydroamination of Primary Amines^a

entry	mol % [Rh]		aminoalkene	product	time (h)	yieldb (dr)c	ratio (amine:imine)c
1	3				15	66% (dr = 1.3:1)	9:1
2d	4				7	a; n=1; 76%e	>95:5
3d	4				8	b; n=2; 77%	>95:5
4f,g	18				15	40%	>95:5
5f,h	3				14	64%i	7:1
6j	3		a; R=OTBDPS		18	71% (dr = 1.6:1)	>95:5
7	3		b; R=4-(CO2Me)C6H4		21	64% (dr = 1.6:1)	10:1
8	5		c; R=4-(CH2CN)C6H4		12	79% (dr = 1.5:1)	>95:5
9	5		d; R=4-(CH3(O)C)C6H4		16	61% (dr = 1.5:1)	10:1
10	5		e; R=4-(CH2OH)C6H4		8	56%k (dr = 1.1:1)	7:1
11	1		a; R = H	a; R = H	5	87%	>95:5
12	2		b; R = Cl	b; R = Cl	6	84%	>95:5
13	1				7	74% dr = (1.1:1)	>95:5

^aReaction conditions: 0.5 mmol aminoalkene, [Rh(CH₃CN)₂COD]BF₄ (1-18 mol %), and L1 (1.2-19 mol %) in 1 mL of tBuOH at 70 °C for unless otherwise specified.

^bIsolated yield.

^cRatio determned by ¹H NMR spectroscopy.

^dProduct isolated as Boc carbamate.

^eNMR yield; the product was isolated in 67% yield as an 10:1:1 mixture of 7a-Boc:BocNEt₂:N-Boc-1-amino-3-pentene

^fReaction run at 100 °C in dioxane.

^g3.3:1 mixutre of E:Z isomers.

^h10:1 mixture of E:Z isomers.

ⁱIsolated as a 1.8:1 mixture of olefin isomers (internal:terminal).

^jReaction run on a 0.25 mmol scale.

^kIsolated as a 97:3 mixture of alcohol:aldehyde products.

Control experiments showed that the tBu groups on the xanthene backbone had little effect on the reactivity or selectivity of the catalyst (Table 1, compare entries 1 and 2 and entries 4 and 5). Thus, L1 (R = tBu) or L1’ (R = H) can be used interchangeably in catalytic hydroaminations. During the latter course of this work, L1’ became available commercially, but both ligands are prepared by a similar one-pot metalation of 4,5-dibromo-2,7-di-t-butyl-9,9-dimethylxanthene or 9,9-dimethylxanthene and quenching with bis(diethylamino)chlorophosphine.⁴⁷

1. Scope of the Intramolecular Hydroamination

1.1. Cyclization of unbiased aminoalkenes and internal alkenes

The scope of the hydroamination of primary amino alkenes that lack gem-disubstitution on the alkyl chain or that contain internal olefins or dienes catalyzed by the combination of [Rh(CH₃CN)₂COD]BF₄ and L1 is illustrated in Table 2. As noted in the previous section, the reaction of substrate 1b containing just one substituent on the carbon beta to the amino group and alkene unit (Table 2 entry 1) gave a 9:1 ratio of amine to imine and led to isolation of the amine in 66% yield. The high activity of this catalyst system for reactions of primary aminoalkenes in tBuOH solvent was particularly exemplified by reactions with aminoalkenes lacking any substituents on the linker between the amino group and the alkene. Reactions of these aminoalkenes occurred to generate both 5- and 6-membered ring products. For example, 1-aminopentene and 1-aminohexene underwent cyclization in the presence of 4 mol % [Rh(CH₃CN)₂COD]BF₄ and L1 to afford 2-methylpyrrolidine 7a and 2-methylpiperidine 7b (Table 2, entries 2 and 3). Internal olefins also underwent cyclization. In these cases, the solvent had little effect on the conversion or selectivity of hydroamination, and intramolecular additions to monoene 8 and the diene 10 (entries 4 and 5) were accomplished in dioxane at 100 °C. A higher catalyst loading (18 mol %) was required to achieve high conversion of the unactivated internal alkene 8 due to competitive catalyst decomposition, but these reactions constitute the first examples of cyclization of a primary amine with an internal alkene catalyzed by a complex of a late transition metal.

Free diethylamine was detected in the crude reaction mixtures by GC and ¹H NMR spectroscopy, and this observation suggests that one pathway for catalyst decomposition occurs by displacement of the diethylamino groups on the phosphorus by the primary amine substrate. For substrates with slower rates of hydroamination (i.e unactivated internal olefins), the rate of catalyst decomposition becomes competitive with the rate of cyclization.

1.2. Tolerance for auxiliary functional groups

The objective of developing a late transition metal catalyst for hydroamination stems from the expectation that hydroamination with this type of catalyst will occur with a high tolerance for auxiliary functional groups. To determine if this assertion was valid, we studied cyclizations of substrates that are both unbiased for cyclization and possessing potentially reactive functional groups. Indeed, such cyclizations catalyzed by the combination of [Rh(CH₃CN)₂COD]BF₄ and L1 occurred with aminoalkenes containing a siloxy group (entry 6), a methoxycarbonyl group (entry 7), a cyano group (entry 8), an acetyl group (entry 9) and even an unprotected hydroxyl group (entry 10). Cyclization of the aminoalkene containing the unprotected alcohol (12e) occurred with a small amount of competitive formation of an aldehyde byproduct at high conversions, but the hydroxyl-substituted product 13e was isolated in 56% yield. Cyclizations of substrates containing esters, ketones, and alcohols have not been reported with catalysts based on lanthanides or group IV metals.

Tetrahydroisoquinolines are a common structural motif in numerous biologically active molecules.^48,49 3-Methyl-tetrahydroisoquinolines were readily synthesized by hydroamination of 2-allylbenzylamines in excellent yield and selectivity (entries 11-13) with low catalyst loadings (1-2 mol %). The hydroamination of the parent 2-allylbenzylamine 14a even occurred at room temperature (10 mol % Rh, THF, 24 h) to afford the product in 77% isolated yield. Although reactions of a broad range of functionalized 2-allylbenzylamines were not explored, the aryl chloride function was tolerated, as demonstrated by the high-yielding cyclization of substrate 14b (entry 13). In addition, reactions of aminoalkenes that are branched at the α position have not occurred previously with catalysts of the late transition metals,⁵⁰ but the α-branched (entry 14) aminoalkene 16 readily formed the tetrahydroisoquinoline product in the presence of catalytic amounts of [Rh(CH₃CN)₂COD]BF₄ and L1.

1.3. Reactions of secondary aminoalkenes

The reactivity of this catalyst was not limited to primary amines. Secondary amines also underwent hydroamination in good to excellent yields with low catalyst loadings (1-3 mol %) to furnish N-alkyl pyrrolidines and N-alkyl tetrahydroisoquinolines (Table 3). The commonly studied aminoalkene 18 containing gem-diphenyl substituents cyclized in high yield with a high ratio of amine to imine. However, secondary aminoalkene 20 containing no substituents in the linking unit also underwent cyclization with a high ratio of amine to imine to give 65% yield of isolated pyrrolidine product. Even the sterically encumbered N-iPr-substituted aminoalkene 22a readily underwent cyclization (entry 3). Finally, hydroamination with the secondary aminoalkene 22b occurred at room temperature (10 mol % Rh, THF, 24 h) to give the corresponding product and 23b in good yield (74%), further exemplifying the high reactivity of the catalyst derived from [Rh(CH₃CN)₂COD]BF₄ and L1.

Table 3.

Rh-Catalyzed Hydroamination of Secondary Amines.^a

entry	mol% [Rh]	aminoalkene	product	time (h)	yieldb	ratio (amine:imine)c
1	1			5	83%	>95:5
2	3			5	65%d	>95:5
3	1	a; R = iPr	a; R = iPr	8	78%	>95:5
4	1	b; R = Bn	b; R = Bn	5	86%	>95:5

^aReaction conditions: 0.5 mmol aminoalkene, [Rh(CH₃CN)₂COD]BF (1-3 mol %), L1 (1.2-3.5 mol %) in 1 mL of tBuOH at 70 °C unless otherwise specified.

^bIsolated yield.

^cRatio determined by ¹H NMR spectroscopy.

^dNMR yield.

1.4. Current limitations in the scope of the cyclizations of aminoalkenes

The results in the previous three sections demonstrate that complex derived from [Rh(CH₃CN)₂COD]BF₄ and L1 is the most active late-transition metal catalyst for intramolecular hydroaminations of unactivated primary aminoalkenes reported. However, we also identified limitations of this catalyst system. For example, substrates containing internal olefins, such as substituted styrenes and cyclohexenes (i.e. to generate bicyclic ring systems), did not undergo cyclization. Moreover, a basic amine was required; aminoalkenes that lacked a basic amine, such as N-aryl- and N-Cbz-protected aminoalkenes did not react with this catalyst under these conditions. Finally, the reaction of 2,2-diphenylhept-6-en-1-amine with catalytic amounts of [Rh(CH₃CN)₂COD]BF₄ and L1 afforded only small amounts of the seven-membered ring product. Isomerization of the starting aminoalkene was the major reaction of the heptenylamine substrate.

2. Identification of the Structure of the Active Catalyst

To understand the origins of the high reactivity of this catalyst system for cyclization of primary aminoalkenes, we identified the structure of the active catalyst and the factors that control the reaction rates and selectivity for formation of hydroamination vs oxidative amination products. The reaction of [Rh(CH₃CN)₂COD]BF₄ with L1 or L1’ in THF at room temperature formed [(L1)Rh(NCMe)]BF₄ and [(L1’)Rh(NCMe)]BF₄ in 89% and 67% yield, respectively (Eq. 1). Complexes of L1 and L1’ were characterized by ¹H, ¹³C, and ³¹P NMR spectroscopy and elemental analysis, and the structure of [(L1’)Rh(NCMe)]BF₄ was determined by X-ray diffraction. An ORTEP diagram of this complex is shown in Figure 1. This structure shows that L1 is bound in a κ³-fashion to the rhodium atom to generate a square-planar POP-pincer structure. The Rh-O bond distance (Rh-O = 2.129 Å) is similar to that in known cationic Rh-POP complexes reported previously.^51,52 These complexes catalyzed the hydroamination of aminoalkenes with reactivity and selectivity that are similar to those of reactions catalyzed by the combination of the rhodium precursor [Rh(CH₃CN)₂COD]BF₄ and ligand L1 or L1’. For example, cyclization of substrate 1b with 5 mol% [(L1)Rh(NCMe)]BF₄ in tBuOH at 70 °C resulted in full conversion of starting material after 7 h to afford a 10:1 ratio of amine:imine products 2b:3b as determined by ¹H spectroscopic analysis of the crude reaction mixture. Under otherwise identical reaction conditions, the combination of 5 mol% [Rh(CH₃CN)₂COD]BF₄ and 6 mol % ligand L1 gave complete conversion of aminoalkene to a 9:1 ratio of 2b:3b. The tridentate, “pincer” coordination mode of the xantphos derivative is likely integral to the selectivity for hydroamination over oxidative amination by inhibiting the β-hydride elimination step that leads to the formation of the imine byproduct (Eq. 2).⁵³ Additional discussion on the influence of the pincer structure on selectivity is presented in the mechanistic conclusions section.

(Eq 1)

(Eq 2)

Figure 1.

ORTEP drawing of catalyst [(L1’)Rh(NCMe)]BF₄ (35% ellipsoids, hydrogens and BF₄ counterion omitted). Selected bond distances (Å) and angles (deg): Rh1-O1 2.129; Rh1-P1 2.250; Rh1-P2 2.242; Rh1-N5 1.943; P1-Rh1-P2 168.0; N5-Rh1-O1 178.8; O1-Rh1-P1 84.1; O1-Rh1-P2 83.9; N5-Rh1-P1 96.9; N5-Rh1-P2 95.0.

3. Effect of Phosphine Ligand on Hydroamination Reactivity

To probe the influence of the phosphine substituents on reactivity and selectivity, we investigated the reactions of aminoalkene 1a catalyzed by complexes of xantphos ligands containing diaryl- and dialkylphosphino groups. Specifically, we evaluated the parent xantphos ligands L2 and L2’, containing diphenylphosphino groups (Table 4, entries 3 and 4),⁵⁴ and Cy-xantphos L3,⁵⁵ containing dicyclohexylphosphino groups (entry 5). Reactions catalyzed by complexes of these ligands occurred with lower conversions and selectivity than those catalyzed by the complex of bis(diethylamino) xantphos L1 or L1’ (entries 1 and 2). Specifically, the reaction catalyzed by the combination of ligand L2 or L2’ and [Rh(CH₃CN)₂COD]BF₄ afforded mainly the isomerized product 5a, and the reaction catalyzed by the combination of ligand L3 and the same rhodium species occurred with low conversions and poor selectivity for hydroamination versus oxidative amination (2:1 ratio of 2a:3a).

Table 4.

Evaluation of Xantphos Ligands in the Hydroamination of 1a.^a

entry	% Rh	ligand	time (h)	%convb	ratiob2a:3a:4a:5a
1	1	L1	2	100	18:1:1:0.7
2	1	L1’	2	100	11:1:1:0.2
3	3	L2	2	83	1:0.1:0:4.4
4	3	L2’	2	62	1:0.2:0:9.1
5	3	L3	2	47	2:1:1:0c
6	3	L4	2	100	5.7:1:1:0.4

^aReaction conditions: 0.2-0.3 mmol 1a, 1-3 mol % [Rh(CH₃CN)₂COD]BF₄, and 1.2-3.5 mol % ligand in 0.4-0.6 mL of tBuOH at 70 °C.

^bDetermined by ¹H NMR spectroscopy.

^cTrace amount of isomerization observed.

We hypothesized that the amino groups on the phosphorus ligand could be responsible for the enhanced rate and selectivity observed in the catalytic hydroamination reaction. To distinguish between steric and electronic effects of the amino groups on this reactivity and selectivity, we prepared L4,⁵⁶ an aminophosphine analogue of Cy-xantphos L3. Ligand L4 contains piperidinyl groups that are nearly isosteric with cyclohexyl groups. The cyclization of 1a catalyzed by the combination of L4 and [Rh(CH₃CN)₂COD]BF₄ generated the amine product with higher conversion and selectivity for hydroamination (100% conversion, 6:1 ratio of 2a:3a, entry 6) than the reaction catalyzed by the complex of L3 (47% conversion, 2:1 ratio of 2a:3a, entry 4) and close to the selectivity of the reactions catalyzed by rhodium and L1 or L1’. These results suggest that amino groups on the phosphorus ligand are responsible for the enhanced rate and selectivity.

The electron-donating property of the aminophosphine was assessed by preparing the carbonyl adducts of LRh⁺ fragments (Eq. 3). The v_CO value in the infrared spectrum of the aminophosphine complex [(L1’)Rh(CO)]⁺ and the Cy-xantphos complex [(L3)Rh(CO)]⁺ were nearly identical (1993.3 cm^-1 vs 1992.7 cm^-1). Thus, the aminoxantphos and Cy-xantphos ligands have similar electron-donating properties. These results, in combination with the above results comparing the difference in the reactivity of the sterically similar Cy-xantphos and piperidinyl xantphos complexes, demonstrates that the effect of the substituents on phosphorus likely stem from properties beyond steric demands and degree of electron-donation to the metal center.

(Eq 3)

4. Mechanistic Studies

4.1. Identification of Catalyst Resting States in Hydroamination Reactions of Primary Aminoalkenes

Two mechanistic pathways are typically proposed for the hydroamination of aminoalkenes catalyzed by late transition metals, one that occurs by insertion of the alkene into the metal-amide bond (path A) and one that occurs by nucleophilic addition of the amine onto a coordinated olefin (path B, Eq 4).^3,57 With few exceptions, electron-deficient late transition metal catalysts, such as those based on platinum,^32,33 palladium,⁴⁵ and iridium,⁴⁶ have been shown to react by nucleophilic attack of an amine onto a coordinated olefin. To distinguish between these two mechanisms and to determine which step of the deduced catalytic cycle is influenced by the aminophosphine ligand we determined the rhodium species present in the catalytic reaction and measured the kinetic isotope effect.

(Eq 4)

To identify catalyst resting states, we monitored reactions of aminoalkene 1b with catalytic and stoichiometric quantities of [(L1)Rh(NCMe)]BF₄ by ¹H and ³¹P NMR spectroscopy and compared the spectra to those of rhodium complexes prepared independently. Likely intermediates were readily prepared (vide infra) and were distinguishable by the magnitude of the Rh-³¹P coupling constant. Reactions were conducted in both THF and tBuOH to gain insight into the origin of the accelerating effect of the tBuOH solvent.

Model rhodium complexes containing alkene ([(L1’)Rh(1-hexene)]BF₄), primary amine ([(L1’)Rh(4a)]BF₄), and imine ligands ([(L1’)Rh(3a)]BF₄) ligands were synthesized from the acetonitrile complex [(L1’)Rh(NCMe)]BF₄ (Scheme 1). The imine ligand 3a was found to displace the acetonitrile ligand at room temperature in THF after 1 h to afford the corresponding imine complex [(L1’)Rh(3a)]BF₄ in 75% isolated yield. The primary amine complex [(L1’)Rh(4a)]BF₄ was prepared in 89% yield by heating a solution of the acetonitrile complex [(L1’)Rh(NCMe)]BF₄ and 2.3 equiv. of primary amine 4a at 50 °C in THF for 14 h. Displacement of the coordinated acetonitrile with 1-hexene required a large excess of olefin (>50 equiv) to drive the equilibrium. Even after the mixture fully equilibrated, ~10% of the acetonitrile complex remained when analyzed by ³¹P NMR spectroscopy after 21 h in THF at room temperature. After removal of the free acetonitrile under vacuum and resubjection of the crude mixture containing this ~10:1 ratio of rhodium complexes to the reaction conditions, complete conversion of [(L1’)Rh(NCMe)]BF₄ to the olefin complex [(L1’)Rh(1-hexene)]BF₄ was achieved. [(L1’)Rh(1-hexene)]BF₄ was obtained in 76% isolated yield. All three model rhodium complexes were characterized by elemental analysis, ¹H, ¹³C, and ³¹P NMR spectroscopy.

Scheme 1.

Synthesis of model rhodium complexes for the determination of catalyst resting states.

The phosphorous-rhodium coupling constants for the acetonitrile, olefin, and amine/imine complexes are significantly different, thus enabling the identification of intermediates containing coordinated alkenes, primary amines, imines, or acetonitrile by ³¹P NMR spectroscopy (Scheme 1). The ³¹P chemical shifts and phosphorous-rhodium coupling constants for the amine and imine complexes [(L1’)Rh(4a)]BF₄ and [(L1’)Rh(3a)]BF₄ were found to be similar (δ 106 ppm, J_Rh-P = 166 and 168 Hz in tBuOH, respectively); however, a diagnostic downfield signal near ~4.5 ppm in the ¹H NMR spectrum of the imine complex, assigned to the methylene protons alpha to the imine nitrogen, allowed us to distinguish between these two species.

Spectral data for these model complexes matched closely with the spectral data for species observed in the reaction of 1b catalyzed or mediated by [(L1)Rh(NCMe)]BF₄. Specifically, the reaction of 1b in THF or tBuOH in the presence of stoichiometric amounts of [(L1)Rh(NCMe)]BF₄ was monitored by ¹H and ³¹P NMR spectroscopy. In THF, substrate 1b displaced the acetonitrile ligand, and a 10:1 ratio of amine complex:olefin complex (I:II) was observed at room temperature (Scheme 2). Further heating to 40 °C resulted in formation of the cyclized product. After heating at 60 °C for 25 min, aminoalkene 1b was consumed, and the catalyst reverted to the acetonitrile complex [(L1)Rh(NCMe)]BF₄. Similar results were obtained for the reaction of 1b and [(L1)Rh(NCMe)]BF₄ in tBuOH. However, only ~60% conversion of [(L1)Rh(NCMe)]BF₄ to a 1:1 ratio of I:II was observed upon addition of aminoalkene 1b to the rhodium complex. These results demonstrate that the coordination ability of the primary amine to the rhodium center is attenuated in tBuOH and that the equilibrium ratio of amine complex to olefin complex lies more toward the olefin complex in tBuOH than it does in THF.⁵⁸

Scheme 2.

Reaction of aminoalkene 1b with stoichiometric [(L1)Rh(NCMe)]BF₄.

The reaction of aminoalkene 1b catalyzed by 5 mol % [(L1)Rh(NCMe)]BF₄ in tBuOH at 62 °C was also monitored by ¹H and ³¹P NMR spectroscopy. In this experiment, amine and olefin complexes I and II were identified as the catalyst resting states throughout the course of the reaction (~3:1 to 6:1 ratio I:II) (Figure 2, (a-d)). However, the imine complex was observed (Figure 2, (b-e)) as the reaction progressed and the concentration of the imine byproduct 3b increased. The imine complex was identified as the final resting state of the catalyst upon consumption of 1b (Figure 2, (e)). Although the imine product 3b is formed in minor amounts, it binds tightly to the cationic rhodium center and out competes binding of the primary amine reactant, the pyrrolidine product, and acetonitrile.

Figure 2.

Stacked plot of ³¹P NMR spectra obtained at various conversions of aminoalkene 1b catalyzed by 5 mol % [(L1)Rh(NCMe)]BF₄ in tBuOH at 62 °C.

The presence of the coordinated olefin intermediate II and the absence of neutral rhodium-amido and rhodium-hydride species in reactions mediated or catalyzed by [(L1)Rh(NCMe)]BF₄, is consistent with a mechanism involving nucleophilic attack of the primary amine onto a coordinated olefin (Eq 4, Path B). The observation of the amine and olefin complexes I and II as the resting state suggests two possible reaction sequences. Nucleophilic attack of the primary amine onto the coordinated olefin could be irreversible and turnover-limiting. Alternatively, nucleophilic attack of the primary amine onto the coordinated olefin could be reversible, and followed by irreversible protonolysis of the Rh-C bond. In the latter case, the combination of reversible nucleophilic attack and irreversible protonolysis together would control the rates of these reactions (see Scheme 4, steps ii and iii).

Scheme 4.

Proposed catalytic cycle

4.2. Measurment of Kinetic Isotope Effects

A kinetic isotope effect (KIE) would distinguish between a pathway in which nucleophilic attack was the first irreversible (“turnover-limiting”) step and one in which proton transfer is the first irreversible step. Thus, we measured the KIE for reaction of the aminoalkene 1b and the N-deuterated substrate 1b-D₂ (Eq 5). A small KIE would be expected for the pathway involving turnover-limiting nucleophilic attack because N-H(D) bond-cleavage is not involved in this step; however, a large primary KIE would be expected for the pathway involving turnover-limiting Rh-C bond cleavage by proton transfer from an amine or ammonium nitrogen. At 61 °C with 5 mol % [Rh(CH₃CN)₂COD]BF₄ and 6 mol % L1 in tert-butanol-d₉ or -d₁₀, the value of kH/kD of aminoalkenes 1b/1b-D₂ deduced from initial rates was small (1.16±0.10).⁴⁷ This small value is consistent with turnover-limiting nucleophilic attack of the amine onto the alkene. This result differs from the large primary isotope effect of 3.4 measured for cyclization of an aminoalkene catalyzed by Ir and the observation of aminoalkyl-metal complexes as the resting states in reactions catalyzed by complexes of Pd and Pt. The small KIE implies that the relative rates for protonolysis of the metal-carbon bond vs reversion to the acyclic reactant (reverse of step ii) is more favorable for the rhodium aminophosphine complex than it is for other systems containing late metals.

(Eq 5)

4.3. Investigation of Relative Rates

The relative rates of reaction of the primary and secondary amines were different in separate systems and in the same system. When measured independently, the rate of hydroamination of the secondary amine 18 to form 19 was similar to the rate of hydroamination of primary amine 1a to form 2a (k_1a/k₁₈ deduced from initial rates was 1.3) However, the primary amine reacted exclusively before the secondary amine when present in the same system. This difference reflects a difference in the resting state of the catalyst in separate systems and with the two substrates together.

The resting state of reactions of secondary amines consists of a combination of the olefin-bound secondary aminoalkene 18 complex and the acetonitrile complex [(L1)Rh(NCMe)]BF₄. The resting state of the catalyst in reactions of primary aminoalkenes in THF is the primary amine complex. The order of affinities of a primary amine, secondary amine, and alkene follow the trend 1° amine > alkene > 2° amine. Treatment of alkene complex [(L1’)Rh(1-hexene)]BF₄ with 1.5 equiv of benzylamine in THF led to a 26:1 ratio of the primary amine complex to the alkene complex complex after 16 h, whereas treatment of this alkene complex with the secondary amine N-methylbenzylamine formed a 2.5:1 ratio of the olefin complex to the secondary amine complex. Thus, the relative energies for the resting states and transition states can be represented by those in Figure 3, and reaction of the substrates together yields the product from the primary amine over the secondary amine.

Figure 3.

Relative ground and transition state energies for reactions of primary and secondary aminoalkenes deduced from competition binding and catalytic reations.

5. Mechanistic Conclusions

Based on these experiments we propose the mechanism in Scheme 4 involving nucleophilic attack of amine onto a coordinated olefin. In this mechanism, substitution of the tethered olefin for the primary amine in complex I occurs to deliver olefin complex II. The equilibrium ratio of the amine and olefin complexes was significantly influenced by the solvent (i.e. THF vs tBuOH). In the protic solvent tBuOH the ratio of olefin to amine complex was larger than it was in the aprotic THF solvent. Increased concentrations of the olefin complex should increase the rate of nucleophilic addition, and this increased concentration is one factor that explains the faster rates of catalytic reactions performed in tBuOH than those of reactions conducted in THF. The coordinated olefin subsequently undergoes rate-limiting nucleophilic attack (step ii) to generate an ammonium alkylrhodium adduct III that could either undergo protonolysis to generate the hydroamination product or β-hydride elimination to afford the oxidative amination product.

The pincer coordination mode of the xantphos derivative likely affects the selectivity for hydroamination over oxidative amination. Previously, a dicationic PNP-palladium complex was shown to catalyze the hydroamination of amide- and carbamate-protected aminoalkenes.²⁸ It was proposed that the tridentate PNP ligand inhibits β-hydride elimination of an alkylmetal intermediate because it occupies the remaining three coordination sites in the square-plane of the metal. β-Hydride elimination is typically faster when a coordination site is available in the square-plane (Eq. 2).⁵³ The POP-pincer structure of the rhodium complex likely favors hydroamination over oxidative amination for related reasons. The POP pincer structure, although less tightly bound to monocationic rhodium in a κ³ structure than the PNP ligand is bound to dicationic palladium, could still discourage β-hydrogen elimination and favor formation of hydroamination over oxidative amination products. The use of pincer-type ligands in late metal-catalyzed hydroamination reactions, therefore, appears to be a general strategy for inhibiting an oxidative amination pathway and favoring a hydroamination pathway.

The complementarity of hydroaminations with basic amines catalyzed by rhodium and hydroamination with less basic amides and carbamates catalyzed by palladium²⁸ likely results from differences in the charge of the two complexes. The monocationic rhodium complex competitively binds the alkene and the amine, but it bears a less electrophilic coordinated alkene. Thus, a more nucleophilic amine is needed for cyclization to occur by nucleophilic attack on the coordinated alkene. On the other hand, the dicationic palladium complex bears a more electrophilic coordinated alkene, and less nucleophilic amides will attack this coordinated alkene. Hydroamination with more basic nitrogen nucleophiles, such as alkylamines, do not occur with this palladium system, perhaps because these nucleophiles bind tightly to the highly electrophilic Pd center and prevent the alkene binding that precedes C-N bond formation.

Our measured kinetic isotope effect and the observed catalyst resting states imply that the series of proton transfer steps (step iii), including cleavage of the rhodium-carbon bond to generate the cyclized product, occurs faster than reversion to the reactant by C-N bond cleavage. Cleavage of the rhodium-carbon bond could occur by either direct protonation of the sigma bond or by a mechanism involving initial formation of a Rh^III-H intermediate, followed by reductive elimination to release the product.⁵⁹ We propose that the aminophosphine ligand affects these relative rates by promoting proton transfer, perhaps by assistance by the electron pair. We are currently investigating this hypothesis further. As the reaction progresses, and the concentration of the imine increases, the imine complex IVb becomes the resting state. In contrast to platinum-, palladium-, and iridium-catalyzed hydroamination of secondary amines, protonation of the metal-carbon bond is not the turnover-limiting step for the hydroamination of primary amines catalyzed by [(L1)Rh(NCMe)]BF₄.

Summary

We have developed a highly active catalyst for the intramolecular hydroamination of unprotected, primary aminoalkenes based on an unusual diaminophosphine ligand. For the first time with a late metal catalyst, primary aminoalkenes that do not contain gem-disubstitution on the alkyl chain undergo cyclization in high yields at mild temperatures. The substrate scope includes aminoalkenes containing various reactive functional groups that would not otherwise be tolerated with previously reported catalysts for hydroamination of primary amines. We have also demonstrated initial reactivity of this catalyst with internal alkenes.

The diaminophosphine ligand appears to be important for achieving high rates and selectivity in the intramolecular hydroamination of primary amines. A detailed mechanistic study of this catalytic reaction has been conducted. Our data support a mechanism occurring by turnover-limiting nucleophilic attack of amine onto a coordinated olefin, followed by cleavage of the metal-carbon bond. We suggest that the aminophosphine groups on the ligand create favorable relative rates for the proton transfer steps.

We have also identified the primary pathway for catalyst decomposition that will allow us to design future catalysts with improved stability. We are currently seeking to use these conclusions to develop even more active catalysts for intermolecular hydroaminations of unactivated olefins and to develop asymmetric hydroaminations with catalysts bound by ligands containing chiral diamino groups.³⁷

Scheme 3.

Competition experiment between primary and secondary aminoalkenes.