LiCl-Accelerated Multimetallic Cross-Coupling of Aryl Chlorides with Aryl Triflates.

While the synthesis of biaryls has advanced rapidly in the past decades, cross-Ullman couplings of aryl chlorides, the most abundant aryl electrophiles, have remained elusive. Reported here is the first general cross-Ullman coupling of aryl chlorides with aryl triflates. The selectivity challenge associated with coupling an inert electrophile with a reactive one is overcome using a multimetallic strategy with the appropriate choice of additive. Studies demonstrate that LiCl is essential for effective cross-coupling by accelerating the reduction of Ni(II) to Ni(0) and counteracting autoinhibition of reduction at Zn(0) by Zn(II) salts. The modified conditions tolerate a variety of functional groups on either coupling partner (42 examples), and examples include a three-step synthesis of flurbiprofen.

The synthesis of biaryls has become one of the most commonly used reactions in pharmaceutical, agrochemical, and materials science industries,[1] yet access to arylmetal reagents remains limiting. The low commercial availability of arylmetal reagents has inspired a number of active areas of research (Scheme A), including improved methods for arylmetal synthesis,[2] C–H arylation,[3] and decarboxylative cross-coupling.[4]

Scheme 1

Cross-Ullman Reaction in Biaryl Synthesis

The relative abundance of aryl electrophiles (Scheme B[5]) would make the cross-Ullman reaction[6,7] an attractive approach, but our recently reported catalytic nickel and palladium method was not broadly effective with the most abundant and versatile aryl electrophiles, aryl chlorides.[8] In addition to opening up more chemical space, aryl chlorides are often lower in cost, and their lower reactivity would allow for sequential coupling in fragment-based drug discovery[9] or late-stage coupling on complex molecules.[10]

Although significant advances in the use of aryl chlorides in cross-coupling have been made recently,[7c,11−14]there are no general methods for the direct cross-coupling of electron-neutral or electron-rich aryl chlorides with other aryl electrophiles.[15,16] In our prior report, we established that in order to promote a successful cross-Ullman reaction, the electrophiles employed had to be orthogonally paired in reactivity: the Ni catalyst activated aryl bromides at a higher rate than aryl triflates; the Pd catalyst activated aryl triflates at a higher rate than aryl bromides. When sufficiently electron-deficient aryl chlorides were substituted for aryl bromides, they were still activated enough to maintain catalyst selectivity. However, when less activated aryl chlorides were used, poor results were obtained. Preliminary studies attempting to couple more electron-rich chlorides with aryl triflates led to production of the triflate-derived dimer and incomplete conversion of both the aryl chloride and the aryl triflate. Herein we report a general multimetallic solution that achieves the selective coupling of a variety of aryl chlorides with aryl triflates (Scheme C).

On the basis of the mechanism proposed in our earlier studies with nickel and palladium multimetallic catalysis (Table A),[8] the slow consumption of the aryl chloride and aryl triflate suggested that arylnickel (II) formation was being inhibited (Table A). Arylpalladium (IV) will not consume aryl triflate without arylnickel (II) present. The inhibition could arise from slow oxidative addition (I to II),[17] slow reduction (III to I), or an off-cycle loss of nickel catalyst.

Table 1

Mechanistic Study and Optimization of the Ar–Cl Cross-Ullman Reaction

entry	change from the optimized conditionsc	3a (%)d
1	none	84
2	NaCl instead of LiCl	62
3	LiBr instead of LiCl	59
4	Bu4NCl instead of LiCl	53
5	TMSCl instead of LiCl	16
6	no LiCl	<10
7	Mn instead of Zn	62
8	Mn instead of Zn, LiBr instead of LiCl	77
9	without PdCl2 and dppb	44
10	without NiCl2(dme) and dtbbpy	<5
11	reaction set up on the benchtope	80
12	1.2 equiv of 2a	90 (89f)

In DMF. See the Supporting Information for details on electrochemical studies.

Reduction of III was conducted in DMF at a concentration of 0.025 M with Zn powder (40 equiv). Cyclooctadiene (0.125 M) was added to stabilize the product. Salts (1–40 equiv) were added in some cases. See the Supporting Information for additional results and experimental details.

Reactions were run on a 0.5 mmol scale in 2 mL of solvent. NMP = N-methyl-2-pyrrolidinone.

GC yield vs dodecane as an internal standard.

The reaction was set up under air with dry solvent.

Isolated yield.

Reduction of the (dtbbpy)NiIIX2 complexes III-Cl and III-OTf was studied by both electrochemical and chemical methods (Table B). Cyclic voltammetry (CV) studies, which are commonly used to assess the ease of reduction of metal complexes,[18] showed no difference between III-Cl and III-OTf (Table B and Figures S7 and S8 in the Supporting Information). While CV provides information on the thermodynamic driving force for a reduction, it does not account for the complex kinetic picture of reduction at a metal surface.[19,20] Indeed, the reduction of complexes III-OTf and III-Cl over zinc flakes in the presence or absence of additives showed that III-OTf is not reduced unless chloride salts are present (Table B,C, Figures S8 and S9, and Table S2). There is also a cation effect: while LiCl enhances the rate of reduction of both nickel complexes III-OTf and III-Cl, ZnCl2 does not. In fact, zinc chloride and zinc triflate, salts formed during the reaction, inhibit reduction of (dtbbpy)NiIICl2 (37% yield with no salt, 2–5% yield with 1 equiv of ZnCl2 or Zn(OTf)2). Lithium chloride[21] can overcome zinc inhibition and was generally the most useful additive studied (Table , entries 1–6, and Table S2).[22,23] While we found that reduction of octyl bromide to octylzinc bromide was also inhibited by zinc salts,[21d,24] reduction of palladium(II) phosphine complexes to palladium(0) was fast with or without added LiCl or Zn (Figures S11–S16).[25,26]

These studies show that the low reactivity observed for the coupling of aryl chlorides with aryl triflates (Scheme B) is due to autoinhibition: the zinc salts (ZnCl2 and Zn(OTf)2) formed in the reduction of III to I inhibit subsequent reductions of III. While it had previously been noted that halide anions accelerate reduction of NiX2 intermediates at zinc surfaces,[27,28] the inhibitory effect of less-coordinating anions (OTf–, BF4–, PF6–)[29] and zinc salts has not been previously reported. This result has broad implications for cross-electrophile coupling reactions that rely upon heterogeneous metallic reductants.

The catalytic reaction behaved as expected from the stoichiometric studies: the addition of LiCl enabled turnover (Table C, entries 1–6, and Figures S1 and S2).[30] Consistent with previous reports,[8] these reactions were still promoted by the cooperativity between the two metal catalysts: reactions without palladium were poorly selective, and reactions without nickel did not consume starting materials (Table , entries 9 and 10). Similar to other cross-electrophile coupling reactions,[28a] the reaction was tolerant of adventitious O2, allowing reactions to be set up on the benchtop (Table , entry 11), although O2 in the reaction headspace resulted in an induction period (Figure S6).[31] Both Zn and Mn could be utilized as reductants. As in our previous report, LiBr was superior to LiCl with Mn (Table , entries 7 and 8, and Figure S5).[20a] Finally, while dtbbpy and dppb were generally the best pair of ligands for this coupling, PCy3 was also effective (Figures S2 and S3). While 6,6′-dibromo-2,2′-bipyridine was not an effective ligand for the model reaction, it was effective for couplings of electron-poor aryl chlorides (Scheme ).

Scheme 2

Reaction Scope and Applications

Reactions were run on a 0.5 mmol scale in 2 mL of solvent for 2–24 h.

Using 5 mol % 6,6′-dibromo-2,2′-bipyridine instead of dtbbpy.

The aryl bromide was used instead of the aryl chloride.

With these modified reaction conditions and an effective way to promote aryl chloride reactivity, we examined the couplings of a variety of aryl chlorides and triflates containing an array of functional groups and steric environments (Scheme ). Electron-poor fluorine-containing substrates as well as electron-neutral and electron-rich substrates were well-tolerated, including sensitive functionalities such as a Boc-protected amine (3c), an aldehyde (3i), an alkyl Bpin ester (3ab), and a phosphonate ester (3ac). More reactive aryl halides, such as aryl chlorides bearing strongly electron-withdrawing groups, heteroaryl halides, or aryl bromides, could be selectively coupled with an aryl triflate by the use of the hindered, electron-poor ligand 6,6′-dibromo-2,2′-bipyridine (3g, 3i, 3j, 3o, 3t, and 3u). Under these reaction conditions, ortho-substituted (3q–s) and 2,6-disubstituted aryl bromides and chlorides (3t–v) were also coupled efficiently. In contrast, steric hindrance was not as well tolerated in our previous report.[32,33] The ability to couple unactivated aryl chlorides can be beneficial in synthesis when the corresponding aryl bromide is either more expensive or not commercially available (3w–ac). The most challenging combination was electron-rich aryl chlorides with electron-poor aryl triflates (3l), which suffered from lower selectivity (about 2.5:1 biaryl to product).

The scope of the aryl triflate was also examined (Scheme ), demonstrating good yields with both electron-donating and electron-withdrawing substituents (3ad–am). The lower yields observed for the coupling of electron-poor aryl triflates with electron-poor aryl chlorides (3ah and 3aj) were due to competing homodimer formation. In these cases, the use of 6,6′-dibromo-2,2′-bipyridine as the ligand did not improve the yield. The couplings with 2-cyano-1-chlorobenzene form biaryls that could be useful for the synthesis of angiotensin II receptor antagonists (3ad–ah).[34]

Besides their improved availability and lower cost, an additional benefit of using aryl chlorides is that their lower reactivity facilitates multistep synthesis (Scheme ). For example, cross-electrophile coupling with an alkyl bromide (5), C–H arylation (7), and reductive α-arylation (9) can all be conducted while preserving the C–Cl bond.[35] As an example of how this can be applied in synthesis, a concise three-step synthesis of flurbiprofen (9) was demonstrated that would be amenable to rapid analogue synthesis.[36]

This report shows how the nickel and palladium system can be rationally modulated to couple less reactive substrates: an unselective multimetallic reaction was made selective with the use of an additive, LiCl, that facilitates the reduction of the nickel catalyst at the zinc surface. Combined with our previous reports, these results suggest that the Ni/Pd system is general and that multimetallic catalysis may have broad generality. Finally, this work demonstrates how reactivity in cross-electrophile coupling reactions can be influenced by the reductant choice as much as the ligand choice: salts formed in the reaction may be autoinhibitory, and new reductant combinations can unlock new reactivity.