Small Phosphine Ligands Enable Selective Oxidative Addition of Ar—O over Ar—Cl Bonds at Nickel(0)

Abstract

Current methods for Suzuki-Miyaura couplings of non-triflate phenol derivatives are limited by their intolerance of halides including aryl chlorides. This is because Ni(0) and Pd(0) often undergo oxidative addition of organohalides at a similar or faster rate than most Ar—O bonds. DFT and stoichiometric oxidative addition studies demonstrate that small phosphines, in particular PMe₃, are unique in promoting preferential reaction of Ni(0) with aryl tosylates and other C—O bonds in the presence of aryl chlorides. This selectivity was exploited in the first Ni-catalyzed C—O-selective Suzuki-Miyaura coupling of chlorinated phenol derivatives where the oxygen-containing leaving group is not a fluorinated sulfonate such as triflate. Computational studies suggest that the origin of divergent selectivity between PMe₃ and other phosphines differs from prior examples of ligand-controlled chemodivergent cross-couplings. PMe₃ effects selective reaction at tosylate due to both electronic and steric factors. A close interaction between nickel and a sulfonyl oxygen of tosylate during oxidative addition is critical to the observed selectivity.

Graphical Abstract

INTRODUCTION

The Suzuki-Miyaura cross-coupling reaction is among the most widely used strategies for C—C bond formation in organic synthesis. This transformation is traditionally catalyzed by palladium and employs aryl halides or arenes substituted by highly labile fluorinated sulfonates (especially triflates) as electrophilic coupling partners.¹ However, recent advances in nickel-catalyzed cross-couplings have facilitated the use of more robust phenol derivatives such as aryl sulfamates, tosylates, mesylates, carbamates, and esters.² The ability to exploit less labile phenol derivatives as alternatives to aryl halides has several advantages. First, phenols can be cheaper or more accessible than organohalides. Second, many phenol derivatives can be carried through multiple synthetic steps, including those involving Pd-catalysis (e.g., Scheme 1A).³ Third, some phenol derivatives can direct other synthetic steps prior to their use in cross-coupling, such as ortho metalation, catalytic C—H functionalization, and electrophilic aromatic substitution.^4,5

Scheme 1.

Ni- and Pd-Catalyzed Suzuki-Miyaura Couplings of Non-Triflate Chlorophenol Derivatives.

Despite numerous reports of Ni-catalyzed cross-coupling of robust phenol derivatives, a critical limitation of current methods is that aryl halides such as chloride are not tolerated. This is because organohalides tend to react with Ni(0) at a similar or faster rate than Ar—O bonds.^6,7 Efforts to achieve Ni-catalyzed Suzuki-Miyaura coupling of chlorophenol derivatives have rarely been reported, but the few existing examples demonstrate preferential coupling of the chloride (e.g., Scheme 1B), poor yields, and/or low selectivity to form a mixture of products (e.g., Scheme 1C).^8,9,10 This functional group incompatibility severely limits the utility of phenol-derived electrophiles, as myriad biologically-relevant synthetic targets contain halogens. For example, about 40% of drugs currently on the market or in clinical trials and about 50% of molecules in high-throughput screening are halogenated.¹¹ Among these, a large percentage (~38%) are chlorinated.¹² As such, development of methods for selective cross-coupling of chlorinated phenol derivatives via C—O cleavage (Scheme 1D) could streamline access to pharmacologically relevant targets.

Herein we demonstrate that small phosphine ligands for Ni(0), especially PMe₃, facilitate selective oxidative addition of Ar—OTs bonds in the presence of Ar—Cl bonds. This selectivity is accurately predicted by DFT calculations, and a series of stoichiometric studies confirm that methylphosphines are unique in enabling this C—O-selective oxidative addition. By using a Ni/PMe₃ catalytic system, we provide proof-of-principle for a highly selective Suzuki-Miyaura cross-coupling of chloroaryl tosylates through C—O cleavage.^10,13,14 Our DFT studies suggest that trimethylphosphine’s unique effect on selectivity can be attributed to a combination of electronic and steric factors. In particular, trimethylphosphine’s small size enables a close stabilizing interaction between a sulfonyl S=O and Ni during oxidative addition at C—OTs.

RESULTS AND DISCUSSION

DFT Calculations With PMe₃.

We conducted density functional theory (DFT) calculations on oxidative addition at Ni(0) to gather insight into the effect of ligands on this step.¹⁵ Our initial DFT studies examined the reaction of Ni(PMe₃)₂ with 4-chlorophenyl tosylate (1). We chose to focus on aryl tosylates because they are generally reactive toward Ni(0), but are often inert toward other reaction conditions including Pd-catalysis (e.g., Scheme 1A). PMe₃ was chosen as a model phosphine ligand due to its computational simplicity. Previous DFT studies suggested that bis-phosphine ligated nickel is likely favored over mono-ligated analogues during oxidative addition when dispersion is considered,¹⁶ so two PMe₃ ligands were included in these calculations. Geometry optimizations were conducted using the functional MN15L,¹⁷ and energies were further refined with MN15L using a larger basis set and 1,4-dioxane as an implicit solvent (see SI for details).

Surprisingly, our initial calculations predict that oxidative addition of C—OTs at Ni(PMe₃)₂ should be significantly faster than oxidative addition of C—Cl (Figure 1). Transition structure 4a-TS is ~3.0 kcal mol⁻¹ lower in energy than 3-TS, suggesting that selective cross-coupling at tosylate might occur with Ni/PMe₃. We initially assumed that this prediction reflected error in the DFT energies, since it contradicts literature results with PCy₃,^8a 1,2-bis(diphenylphosphino)ethane,^8h and triarylphosphines^8f—ligands that are more conventional for catalysis. To our knowledge, PMe₃ had not been experimentally evaluated in Ni-catalyzed Suzuki-Miyaura cross-couplings.^18–20

Figure 1.

Free energy profile for oxidative addition of the C—O and C—Cl bonds of 1 at Ni(PMe₃)₂.

DFT Calculations Using PCy₃ and PPh₃.

We next calculated oxidative addition transition structures with the more experimentally relevant phosphine ligands PCy₃ and PPh₃ (Figure 2). In contrast to the calculations with PMe₃, oxidative addition at Ni(PPh₃)₂ is predicted to be faster at chloride than tosylate (ΔΔG^‡ = 2.0 kcal mol⁻¹). Ni(PCy₃)₂ favors reaction at tosylate, but only by 1.2 kcal mol⁻¹ (compared to 3.0 kcal mol⁻¹ with PMe₃). These predictions are consistent with experimental reports using PCy₃^8a or triarylphosphines in catalysis,^8f and contrast with our DFT results using PMe₃.

Figure 2.

Calculated free energies of activation (in kcal mol⁻¹) for oxidative addition with PMe₃, PPh₃, and PCy₃, measured from the preceding π-complex.

Several additional DFT methods were evaluated with each of the ligands PMe₃, PCy₃, and PPh₃. Most methods predict the same trend in preference for oxidative addition of tosylate (PMe₃ > PCy₃ > PPh₃; see SI for details). With the majority of methods tested, Ni(PMe₃)₂ is predicted to favor reaction at tosylate, while Ni(PPh₃)₂ is usually predicted to react at chloride. As such, DFT calculations suggest that the small phosphine PMe₃ is more likely to facilitate the desired chemoselectivity than traditional ligands like PCy₃ and PPh₃.

Stoichiometric Oxidative Addition Studies.

Stoichiometric oxidative addition studies with PCy₃, PPh₃, and PMe₃ were undertaken and the results were compared to the DFT predictions. When Ni(cod)₂ and PCy₃ (1:2) are combined in 1,4-dioxane, a new signal appears in the ³¹P NMR spectrum assigned to a Ni(0)/PCy₃ adduct (46.0 ppm, Figure 3A). Free PCy₃ (9.9 ppm) is also still detectable after 2 h. Upon addition of 1-chloronaphthalene (7) a new signal grows in within 2 h (11.9 ppm) corresponding to the oxidative addition adduct 9 (Figure 3B). A smaller signal at 12.9 ppm was identified as the Ni(II)-2-naphthyl adduct resulting from oxidative addition of 2-chloronaphthalene, which is a contaminant in commercial 7.²¹ Alternatively, addition of aryl tosylate 8 leads to a new signal at 10.9 ppm corresponding to 10 (Figure 3C). Finally, when a Ni(cod)₂/PCy₃ solution is combined with a mixture of 7 and 8 (1.0 equiv each), signals for both 9 and 10 appear in a 1.5 : 1 ratio (Figure 3D).²² These results show that Ni(0)/PCy₃ has a slight preference for oxidative addition of Ar—Cl over Ar—OTs in an intermolecular competition.

Figure 3.

³¹P{¹H} NMR studies on oxidative addition at Ni/PCy₃ (reaction conditions = 2 h at room temperature in 1,4-dioxane).

The analogous experiments using PPh₃ demonstrate much higher selectivity for oxidative addition of C—Cl. In a competition reaction between 7 and 8 with Ni(cod)₂/PPh₃ (1:2), the only oxidative addition adduct detected results from reaction at chloride (see SI for details). These results are consistent with the trends predicted by DFT, as well as Zou’s catalytic studies using a triarylphosphine (Scheme 1B).^8f

In stark contrast to PCy₃ and PPh₃, the combination of a mixture of 7 and 8 with a 1:2 solution of Ni(cod)₂/PMe₃ gives preferential oxidative addition at C—OTs (Figure 4). The putative oxidative addition adducts 11 and 12 are formed in about a 1 : 6.3 ratio based on ³¹P NMR integrations. Taken together, the results of the intermolecular competition studies with the three ligands are consistent with the DFT-predicted trend in preference for oxidative addition at C—OTs (PMe₃ > PCy₃ > PPh₃).

Figure 4.

³¹P{¹H} NMR studies on oxidative addition at Ni/PMe₃ (reaction conditions = 2 h at room temperature in 1,4-dioxane).

A variety of additional phosphine ligands were evaluated for their effect on oxidative addition selectivity (Table 1). The results show a trend that triarylphosphines strongly favor reaction at chloride (entries 1–3), while most alkyl phosphines give a mixture of products (entries 4–12). The best selectivity for reaction at tosylate is obtained with PMe₃ (entry 12), but PPhMe₂ gives similar selectivity (entry 4). Interestingly, the analogous ethyl phosphines PPhEt₂ and PEt₃ slightly favor reaction at chloride instead of tosylate (entries 5 and 11), suggesting that the small size of a methyl group on phosphine is important.

Table 1.

Ligand Effect on Selectivity of Oxidative Addition.

entry	ligand	A : B	entry	ligand	A : B
1	PPh3	>99 : 1	7	PCy3	1.5 : 1
2		>99 : 1	8	P(n-Bu)3	1.1 : 1
			9	P(i-Bu)3	2.9 : 1
3		>99 : 1	10a		>99 : 1
4	PPhMe2	1 : 5.7
5	PPhEt2	2.2 : 1	11	PEt3	1.8 : 1
6	PPh2Me	1 : 1	12	PMe3	1 : 6.3

^aWith 1 equiv of bisphosphine relative to nickel.

We next evaluated the selectivity when competing 1-chloronaphthalene 7 against a variety of other 1-naphthol derivatives for oxidative addition at Ni(0)/PMe₃ (Table 2). Excitingly, reaction of an aryl triflate (entry 1), as well as the more robust phenol derivatives mesylate (entry 3) and sulfamate (entry 4) are preferred over reaction of naphthyl chloride when using PMe₃. The use of PCy₃ or PPh₃ results in a slight preference for reaction of triflate over chloride, but oxidative addition of chloride is preferred over all other types of C—O bonds depicted in Table 2 with these ligands.

Table 2.

Selectivity of Ni/PR₃ for Reaction of Various Phenol Derivatives in Competition with 1-Chloronaphthalene.

entry	–OR	PMe3 A : B	PCy3 A : B	PPh3 A : B
1	–OTf	1 : 88.6	1 : 2.3	1 : 2.1
2	–OTs	1 : 6.3	1.5 : 1	>99 : 1
3	–OMs	1 : 10.0	2.7 : 1	>99 : 1
4	–OSO2NMe2	1 : 4.6	1.7 : 1	>99 : 1
5	–OCONEt2	>99 : 1	>99 : 1	>99 : 1
6	–OPiv	>99 : 1	>99 : 1	>99 : 1

^aThe reported amounts of A include the Ni(Cl)(2-naphthyl) product resulting from the 2-Chloronaphthalene contaminant in 7, when detected.

The chemoselectivity of oxidative addition with PMe₃ was further evaluated in an intramolecular competition. A 1:2 solution of Ni(cod)₂/PMe₃ was combined with chloronaphthyl tosylate 13 (Scheme 2). Tentative assignment of product signals was made by analogy to the reaction of Ni(cod)₂/PMe₃ with 7 or 8. The product resulting from reaction at tosylate is favored over the putative nickel chloride adduct by about 90:1, suggesting that tosylate selectivity can be even stronger in an intramolecular versus intermolecular competition.²³ The major oxidative addition adduct 14 was isolated in 80% yield and its identity was confirmed by X-ray crystallography.

Scheme 2.

Chemoselective Oxidative Addition with PMe₃.

Taken together, these stoichiometric studies indicate that PMe₃ and PPhMe₂ are unique in providing good selectivity for oxidative addition of aryl tosylates over aryl chlorides at Ni(0).

Catalytic Studies on Chemoselective Cross-coupling.

Trimethylphosphine was investigated as a ligand for a chemoselective Ni-catalyzed Suzuki-Miyaura coupling of chloroaryl tosylates. Initial conditions were modeled after previously reported cross-couplings of aryl tosylates.^8a,24 Substrate 1, 4-methoxyphenylboronic acid (15), and K₃PO₄ were combined with Ni(cod)₂ and PMe₃ in 1,4-dioxane (Table 3). Excitingly, heating to 80 °C led to formation of products 17 and 18 in about a 1:13 ratio favoring C—OTs cleavage (entry 1). However, the yield of 18 is fair and mass balance is poor under these conditions. Moreover, PMe₃ and Ni(cod)₂ are both highly air-sensitive, and PMe₃ is a volatile liquid that is difficult to measure accurately in small volumes. This latter feature of PMe₃ proved to be particularly problematic for reproducibility, as the reaction outcome is sensitive to the PMe₃/Ni(cod)₂ ratio. The best ratio is PMe₃:Ni ≈ 2:1 (9–10 mol % of PMe₃, entries 1 and 3), but a higher or lower ratio gives worse yield of 18 and/or worse mass balance (entries 2, 4, and 5).

Table 3.

Optimization of a Chemoselective Suzuki-Miyaura Cross-Coupling of 4-Chlorophenyl Tosylate.^a

entry	[B]	[Ni]	PMe3 (mol%)	1 (%)	17 (%)	18 (%)
1	15	Ni(cod)2	10%	9	4	54
2	15	Ni(cod)2	6%	27	7	39
3	15	Ni(cod)2	9%	12	6	62
4	15	Ni(cod)2	12%	3	4	55
5	15	Ni(cod)2	15%	14	10	38
6	16	19	--	37	3	60
7b	16	19	--	5	2	88 (72)c
8b	16	20	--	1	2	83
9b,d	16	21	--	21	3	61 (57)c
10b,e	16	22	--	9	5	73

^aGC yields calibrated against undecane as the internal standard. Average of two runs. Small quantities of diarylation observed in all cases (~3–10%). PMP = para-methoxyphenyl.

^bWith H₂O (50 mol %).

^cIsolated yield in parentheses.

^dSet up on benchtop.

^eWith 4 mol % of the mono-THF adduct of precatalyst 22.

For this reason, we synthesized the Ni(II) precatalyst 19. This precatalyst ensures a phosphine:nickel ratio of 2:1. Use of 19 together with boronic ester 16 led to much better mass balance and better reproducibility (entry 6). Finally, addition of a small quantity of water (0.5 equiv) enabled high yield of 18 with excellent mass balance (entry 7). Use of the chloride analogue of the precatalyst (20) gave similar results (entry 8).²⁵ The reaction could be set up on the benchtop by using the semi-air-stable precatalyst 21 and sparging the reaction mixture briefly with N₂ prior to heating, although the yield of 18 was somewhat eroded (entry 9).²⁶ For comparison, a variety of other phosphine ligands were also evaluated in this catalytic system. Consistent with the stoichiometric studies, PPhMe₂ gave comparable results to PMe₃, albeit with slightly lower selectivity and yield (entry 10). All other phosphines evaluated in this catalytic reaction favored reaction at choride or gave poor selectivity and/or yield (see SI).

The generality of the Ni/PMe₃ catalytic system was further explored with a modest scope of boronic ester coupling partners and chlorophenol derivatives (Table 4). Unsurprisingly, the high selectivity for reaction at tosylate was unaffected by the electronics of the arylboronic ester (entries 1–6), although in some cases larger amounts of diarylated products were detected (entries 1 and 6).²⁷ High selectivity for reaction at tosylate was retained using chloroaryl tosylates with different substitution patterns (33 and 35). However, ortho-chlorophenyl tosylate 31 gives diarylation as the major product (32).²⁸ 4-Chlorophenyl dimethylsulfamate and triflate also provide 18 as the major product, resulting from oxidative addition of the C—O bond (entries 8–9). A considerable amount of unproductive substrate decomposition occurred when using the aryl triflate 30.²⁹ An aryl tosylate was successfully cross-coupled with a chloro-substituted boronic ester (eq. 1). However, the attempted cross-coupling of heteroaryl substrates or boronic esters was unproductive and led to recovery of starting material or a complex mixture of products (see SI).³⁰ We anticipate that the generality of this chemoselective cross-coupling can be improved in the future by better understanding the mechanisms of catalyst deactivation or conversion to less-selective catalytic species.³¹

(1)

Table 4.

Suzuki-Miyaura Cross-coupling of Chlorophenol Derivatives through Selective Cleavage of C—O Bonds.

entry	substrate	major product	% GC yield (selectivity)b	isolated yield (%)	entry	substrate	major product	% GC yield (selectivity)b	isolated yield (%)
1c			80 (>50:1)	72	7d			88 (44:1)	72
2	13		80 (>50:1)	61	8d		18	55 (8:1)	39
3	13		95 (>50:1)	78	9d		18	49 (>50:1)	48
4	13		86 (>50:1)	75	10			48	35
5	13		86 (>50:1)	62	11			73 (>50:1)	35
6c	13		87 (>50:1)	64	12			89 (30:1)	67

^aPrecatalyst for entries 1–6:19; entries 7–11:12; entry 12: 20. PMP = para-methoxyphenyl. Diarylated product observed in ≤10% estimated yield by GC unless noted.

^bCalibrated GC yield relative to undecane standard. Approximate selectivities of major product to minor monoarylated product based on calibrated GC yield of major product and uncalibrated GC yield of minor product, where the minor product results from cleavage of the C—Cl bond instead of C—OTs.

^cDiarylated product observed in ≤20% estimated yield by GC.

^dSelectivity based on calibrated GC yields of both the major and minor products.

Computational Analysis of Selectivity Origin.

Previous examples of ligand-controlled chemoselective cross-coupling have a mechanistic origin in the metal’s ligation state during oxidative addition. Perhaps the best-studied example is the case of Pd/phosphine-catalyzed Suzuki-Miyaura coupling of chloroaryl triflates.³² In the coupling of 30, electron-rich bisligated Pd(PCy₃)₂ prefers to react at C—OTf due to a stronger attractive interaction energy in the transition state (Figure 5A). In contrast, the less electron-rich monoligated Pd(P^tBu₃) reacts at C—Cl due to lower distortion energy.^32b Ligation state is likely also the determining factor in other examples of Pd-catalyzed chemodivergent cross-couplings of chloro-³³ or bromoaryl triflates.³⁴ Very bulky monodentate ligands which promote a monoligated active catalyst can favor reaction at the halide, and bidentate or smaller monodentate ligands that promote a bisligated catalyst favor reaction at triflate. Similarly, bulky monodentate and bidentate phosphine ligands can effect divergent selectivity between activation of C(aryl)—O and C(acyl)—O bonds of aryl esters by Ni(0) (Figure 5B).³⁵ DFT studies suggest that this selectivity relates to nickel’s coordination number by phosphine.^35b,36

Figure 5.

Chemodivergent oxidative addition with selectivity controlled by the metal’s ligation state.

In contrast, nickel’s ligation state is unlikely to be relevant to the selectivity observed in the current chemodivergent system.^37,38 PMe₃ and PPh₃ give opposite preferences for reaction at tosylate and chloride, respectively, yet both are monodentate and neither ligand is particularly bulky. DFT calculations suggest that, for PMe₃, PPh₃, and PCy₃, the lowest-energy transition state for oxidative addition at C—Cl or C—OTs involves bisphosphine-ligated nickel¹⁶ (moreover, a 2:1 ratio of phosphine to nickel was used in our experimental studies). As such, it appears that selectivity in this system has a different origin than prior examples of ligand-controlled chemodivergent cross-coupling. We undertook further DFT calculations to understand the unique selectivity preference of Ni/PMe₃.

One hypothesis to explain the divergent behavior between PMe₃ and PPh₃ relates to the greater electron-donating character of PMe₃.⁴⁰ Nickel is calculated to be more negatively charged when ligated by PMe₃ than by PPh₃ (Figure 6A). Furthermore, Ni(PMe₃)₂ has a higher-energy HOMO than Ni(PPh₃)₂. Because C—O bonds are more polarized than C—Cl bonds, the carbon of C—OTs is more electrophilic (compare the charges at carbon in Figure 6B). A more electron-rich Ni(PMe₃)₂ might be expected to have a stronger attractive interaction at the more electrophilic site (C—OTs). A similar argument has been used to explain the divergent selectivity of mono- vs. bis-ligated palladium in the cross-coupling of chloroaryl triflates (Figure 5A).^32b However, this electronic explanation is unsatisfying in the current system when considering PCy₃, which is usually thought of as more electron-donating than either PPh₃ or PMe₃.⁴⁰ Ni(PCy₃)₂ has a higher-energy HOMO than PMe₃ and a more negatively-charged metal center, yet PCy₃ effects only poor selectivity between tosylate and chloride. As such, ligand electronics are not sufficient to explain the unique selectivity observed with PMe₃, especially when compared to other electron-rich alkyl phosphines like PCy₃.

Figure 6.

(A) Calculated NBO charges and HOMO energies for optimized Ni(PR₃)₂ structures, (B) NBO charges at carbon in chlorophenyl tosylate.³⁹

To understand the difference between the behavior of PMe₃ and PCy₃, we more carefully examined the calculated oxidative addition transition structures using these ligands (Figure 7). With both of these ligands, insertion of Ni(0) into C—Cl proceeds through a concerted 3-centered transition structure in which nickel interacts with the departing chloride (3-TS and 38-TS). In contrast, 3-centered concerted transition structures for tosylate activation could not be located. Instead, C—O cleavage occurs through a nucleophilic displacement mechanism: nickel does not interact significantly with the departing oxygen. During reaction at tosylate, nickel has a closer association with the ortho carbon of the substrate: the Ni••••C_ortho distances range from 2.10–2.31 Å in the transition structures for oxidative addition at C—OTs (4a-TS, 4b-TS, 37a-TS, and 37b-TS) compared to 2.33–2.37 Å for reaction at C—Cl (3-TS and 38-TS, Ni••••C_ortho distances not labeled). Oxidative addition of tosylate can occur through a conformation in which the S=O bonds point away from nickel, referred to as a “dissociation” mechanism (4b-TS and 37b-TS), or through a conformation in which one of the sulfonyl oxygens interacts with nickel, referred to as a “5-centered” mechanism⁴¹ (4a-TS and 37a-TS). With PMe₃, this Ni••••O=S interaction results in a lower energy structure compared to the dissociation mechanism. However, with PCy₃, the two mechanisms for reaction at tosylate are energetically similar.

Figure 7.

Oxidative addition transition structures using Ni(PMe)₃ and Ni(PCy₃)₂. Most of the carbons and hydrogens of tosylate, as well as the hydrogens on PCy₃, are hidden for clarity (see SI for complete structures).

The geometries and energetics of these transition structures are consistent with a steric argument for the different selectivity seen with PMe₃ compared to PCy₃. If reaction at tosylate is desired, then PCy₃ has the advantage of being more electron-donating than PMe₃ (see discussion above). However, PMe₃ is also electron-donating—albeit perhaps not as strongly as PCy₃—and it has the added advantage of minimal sterics. The reason why small ligand size promotes reaction at tosylate is because of the possibility for a close stabilizing interaction between a sulfonyl S=O and Ni during the 5-centered transition state (4a-TS). This interaction is also possible with PCy₃ (37a-TS), but it is less stabilizing due to a longer Ni••••O=S distance (2.52 vs. 2.33 Å). Apparently, PCy₃ sterically shields nickel from having a close interaction with oxygen. As a result, the 5-centered mechanism with PCy₃ is not more facile than the dissociation mechanism. In contrast, the 5-centered mechanism with PMe₃ is about 3 kcal mol⁻¹ lower in energy than both the corresponding dissociation mechanism and the mechanism for reaction at chloride.

We conducted a distortion-interaction analysis on the relevant PMe₃- and PCy₃-containing transition structures to gather further evidence for the role of ligand sterics on selectivity. This type of analysis, also called an activation-strain analysis, involves dissecting the activation energy (ΔE^‡) into a distortion (ΔE_dist) and an interaction term (ΔE_int, Figure 8A).⁴² The distortion term is typically a positive value (unfavorable) and represents the energy penalty for distorting the reactants into the transition state geometry. The interaction term is usually attractive (a negative value), and represents the favorable interaction between substrate and catalyst due to factors like orbital mixing, electron correlation, and coulombic pairing of opposite charges. The relationship between distortion and interaction is defined as ΔE^‡ = ΔE_dist + ΔE_int. Distortion and interaction energies of the transition structures were measured from the lowest-energy preceding π complexes.

Figure 8.

(A) Definition of distortion and interaction energy. (B) Selectivity is determined by the difference in distortion energy (ΔΔE_dist) and interaction energy (ΔΔE_int) when comparing oxidative addition at tosylate vs. chloride.

With both PCy₃ and PMe₃, reaction at tosylate involves greater distortion energy than reaction at chloride (compare total ΔE_dist values in Table 5). This is due to greater distortion of both the substrate and the catalyst. The substrate is more distorted during reaction at tosylate because of nickel’s more extensive interaction with the arene π system (both the ipso and the ortho carbons) as well as distortion of several bond angles in the departing tosylate group. The larger catalyst distortion energies during reaction at tosylate reflect the more constrained geometry of NiL₂. During reaction at chloride, the ligands have more room to spread out because Ni is further away from the substrate aryl ring. In fact, the catalyst distortion energy is actually slightly negative (favorable) during insertion of Ni(PCy₃)₂ into C—Cl (38-TS), reflecting a relaxation of the P–Ni–P bond angle in the TS compared to the preceding π complex (122° in 38-TS vs. 115° in the corresponding π-complex). Although distortion energy favors reaction at chloride for both ligands, interaction energy strongly favors (is more negative) reaction at tosylate. This can be rationalized by (1) the stronger coulombic attraction between Ni and the electron-deficient carbon of C—OTs and (2) the ability for another oxygen of S=O to interact with Ni during a 5-centered mechanism.

Table 5.

Distortion-Interaction Analysis of Relevant PMe₃- and PCy₃-Containing Transition Structures.

Selectivity of oxidative addition depends on the differences between the distortion and interaction energies for reaction at the two possible sites. These differences are defined as ΔΔE_dist = ΔE_dist(OTs) – ΔE_dist(Cl) and ΔΔE_int = ΔE_int(OTs) – ΔE_int(Cl) (Figure 8B). ΔΔE_dist is always a positive value, and a larger ΔΔE_dist signifies a more powerful influence of distortion energy in favor of reaction at chloride. Conversely, ΔΔE_int is always negative. A more negative value of ΔΔE_int indicates a more powerful influence of interaction energy in favor of reaction at tosylate.

If reaction at tosylate is desired, the dissociation mechanism with Ni(PCy₃)₂ (37b-TS) has a larger distortion disadvantage than either 5-centered mechanism 37a-TS or 4a-TS (compare the values of ΔΔE_dist in Table 5). This may be due to the larger bending of tosylate out-of-plane of the arene in 37b-TS. However, this bending also leads to a large interaction energy advantage for the dissociation mechanism (ΔΔE_int), likely due to better orbital overlap between nickel’s HOMO and the C—OTs σ* (the Ni–C–O bond angle is 117° in 37b-TS compared to 107° in 37a-TS and 101° in 4a-TS). The effects of ΔΔE_dist and ΔΔE_int nearly cancel each other out for 37b-TS, and this dissociation mechanism for reaction at tosylate using PCy₃ is only slightly lower-energy than reaction at chloride. The 5-centered mechanism with PCy₃ (37a-TS) has a much smaller distortion disadvantage than either 37b-TS or 4a-TS due to the earliness of the transition structure 37a-TS.⁴³ However, 37a-TS also has less of an interaction energy advantage than the corresponding transition structure using PMe₃ (4a-TS). This is because the stabilizing Ni••••O=S interaction in 37a-TS is less significant, as evidenced by the longer distance. In fact, when the intrinsic reaction coordinate for 37a-TS is followed, the Ni••••O distance remains longer than that of 4a-TS until well after the transition state (when the breaking C••••OTs bond is elongated to 2.55 Å). As a result, the effects of ΔΔE_dist and ΔΔE_int again nearly cancel out and 37a-TS has a similar energy barrier as that for reaction at chloride (38-TS). In contrast, the 5-centered mechanism with Ni(PMe₃)₂ has a stronger interaction between Ni and O=S, resulting in a value of ΔΔE_int for 4a-TS that outweighs the influence of distortion energy (ΔΔE_dist) and makes reaction at tosylate more facile than reaction at chloride.

CONCLUSION

Prior Ni-catalyzed Suzuki-Miyaura cross-couplings of non-triflate phenol derivatives do not tolerate aryl chlorides, a feature that limits their synthetic utility. We have shown that DFT calculations accurately predict the following ligand trend for reaction of tosylate over chloride at Ni(0): PMe₃ > PCy₃ > PPh₃. In stoichiometric studies evaluating a wider range of phosphine ligands, we demonstrate that methylphosphines are uniquely capable of promoting chemoselective oxidative addition of aryl tosylates in the presence of aryl chlorides. Good to excellent selectivity for reaction of aryl triflates, mesylates, and sulfamates over aryl chlorides was also observed. This selectivity was exploited to demonstrate proof-of-principle for a C—O-selective Suzuki-Miyaura coupling of chlorinated non-triflate phenol derivatives. We anticipate that a detailed understanding of catalyst decomposition mechanisms will lead to improvements to the scope of the catalytic process.

It appears that the unique behavior of methyl phosphines compared to other phosphines does not relate to the catalyst’s ligation state by ancillary ligands, and thus differs from prior examples of ligand-controlled chemodivergent cross-couplings. Computational analysis suggests that the unusual selectivity of Ni(PMe₃)₂ for insertion into C—OTs is due to both electronic and steric factors. In contrast to arylphosphines, the use of an alkyl phosphine results in an electron-rich active catalyst that can more effectively donate into the more electrophilic carbon of C—OTs. Furthermore, the small size of PMe₃ allows for a close interaction between Ni and a sulfonyl oxygen of tosylate during a 5-centered oxidative addition mechanism. This interaction helps to stabilize the buildup of positive charge at nickel during its oxidation from Ni(0) to Ni(II), and is less significant with bulkier alkyl phosphines. Notably, the differences observed between PMe₃ and PCy₃ caution against relying on PMe₃ as a computational model for more complex alkylphosphines.