Cu-Catalyzed Decarboxylative Borylation.

A simple method for the conversion of carboxylic acids to boronic esters via redox-active esters (RAEs) is reported using copper catalysis. The scope of this transformation is broad, and compared with the known protocols available, it represents the most inexpensive, rapid, and operationally simple option. In addition to a full exploration of the scope, a kinetic study was performed to elucidate substrate and reagent concentration dependences.

The metamorphosis of alkyl carboxylic acids into boronic acids is a powerful transformation that provides a rapid gateway into useful chemical space.[1] In 2017, our lab reported a straightforward method to achieve this through the intermediacy of n class="Chemical">redox-active esters (RAEs), an inexpensive Ni-catalyst, and an in situ derived boron ate complex (from B2pin2).[2] Shortly thereafter, two separate reports emerged from the Aggarwal[3] and Li[4] groups on photochemical and photoinduced electron transfer (PET)-based approaches to the same transformation.[5] Both methods employed RAEs and either a photoactive boron-donor (B2cat2) in the case of the former method, or an Ir-based photocatalyst and B2pin2 as the boron donor in the latter. These three complementary approaches[6] are clearly of use on a preparative scale, but for process applications, they all suffer certain drawbacks, such as the use of excess/expensive boron donors, low concentrations, or oxygen sensitivity. In addition, certain regions of the world prefer to avoid the use of homogeneous Ni catalysis on scale,[7] or expensive photocatalysts and continuous flow apparatus which add additional cost and engineering considerations. These pragmatic constraints were brought to our attention in the context of the recently approved β-lactamase inhibitor vaborbactam (1, Figure A, one of two active ingredients in Vabomere).[8] The chiral ester 2 previously served as a key intermediate en route to 1 and was assembled using standard polar transformations in a four-step sequence and required the use of toxic acrolein, as well as kinetic resolution step, resulting in a 50% loss of material.[9] In contrast, a radical disconnection enabled the use of carboxylic acid 4 (prepared in three steps[10]) directly to access 2 in 48% isolated yield. In considering options for the commercial scale-up of this step, the drawbacks of the known decarboxylative borylation (as mentioned above) were magnified. Here we report a new protocol to achieve this transformation featuring a simple procedure, short reaction times (0.1 h), and a catalyst system based on inexpensive Cu without the need for external ligands. The cost associated with the scale up of this process is roughly an order of magnitude less than its Ni-catalyzed counterpart. The kinetics of this intriguing transformation have also been studied in depth.

Figure 1

(A) Decarboxylative borylation to access vaborbactam. (B) Discovery of a Cu-catalyzed decarboxylative borylation and optimization of reaction conditions. B2pin2 = bis(pinacolato)diboron, TMEDA = N,N,tetramethylethylenediamine.

The development of Cu-catalyzed decarboxylative borylation was guided by industrial n class="Chemical">feedback regarding the minimization of B2pin2 employed, as this was not only a cost driver but also difficult to remove from the product. Thus, this reagent was limited to 1.5 equiv for all explorations on the conversion of model RAE 5 to boronic ester 6 (Figure B, see the Supporting Information for an exhaustive list of conditions evaluated). As shown in entry 1, standard Ni-catalyzed conditions with this reduced B2pin2 loading resulted in 43% yield of 6. A promising hit was obtained by employing conditions reported by the Cook lab for Mn-catalyzed borylation of alkyl halides (6%, entry 2).[11] Oestreich’s decarboxylative silylation using Cu catalysis inspired the conditions attempted in entry 3 (11%).[12] Systematic screening of Cu sources, solvents, and ligands led to a gradual increase in yield (entry 4).[13] The use of an old bottle of tBuOLi led to the realization that LiOH was actually the active base for our reaction and the first promising yields were observed when used in excess (entry 5). Subsequent ligand screening unveiled that no external ligand was actually needed (entry 6) and additional evaluation of both the Mg source and Cu catalyst led to optimized conditions shown in entry 7 (69% yield). If the loading of B2pin2 and Cu were increased, the yield could be further increased to 86% (entry 8). The experimental simplicity of this new protocol is worthy of additional comment. All components are simply added to a flask, the air is purged and replaced with inert gas, followed by addition of solvent—the reaction is complete within minutes at ambient temperature.[14]

With an optimized protocol in hand, the scope of this Cu-catalyzed borylation methodology was explored (Scheme ). To test the generality of this protocol, 3.0 equiv of n class="Chemical">B2pin2 and 30 mol % of Cu(acac)2 was used for the small-scale preparation of substrates in Scheme . Under standard conditions, a variety of primary and secondary RAE substrates were converted to their Bpin derivatives smoothly (Scheme A,B). Simple cyclic (21–25, 27, and 29) and acyclic (6, 11–20, 26, 28, and 30) boronic esters are easily accessed in higher or comparable yield to known decarboxylative borylation methods (comparative yields are shown for Ni2, hν3 and PET4-based conditions where available). Scalability of this reaction was demonstrated through the preparation of 6 on 3.5 mmol scale with 1.5 equiv of B2pin2 and 20 mol % of Cu(acac)2. An in situ protocol from the parent alkyl carboxylic acids was also developed, providing access to several boronic esters (11, 18, 24, and 27) in similar yields. RAEs with β-nitrogen substitution were tolerated and afforded the β-amino boronates (18, 25 and 26), which are difficult to access otherwise. Furthermore, heterocyclic moieties were tolerated under the mild reaction conditions, even in the absence of an external ligand on copper; the 3-pyridyl Bpin ester product (17) was formed in moderate yield. Despite the use of LiOH·H2O, a variety of functionalities such as esters (6), Boc groups (14, 16, 18, 22, 25, and 26), epoxides (15), ketones (19), and ethers (20 and 28) were intact under the mild conditions, presumably due to the limited solubility of LiOH·H2O in organic solvents. Cyclopropyl-based tertiary RAEs were also found to be suitable substrates for this transformation (31–33). Although active low-valent Cu species are likely generated during this reaction, aryl chloride (32) was left unscathed, while aryl iodide containing substrates (33) could be obtained in synthetically useful yield with judicious choice of solvent.

Scheme 1

Scope of the Cu-Catalyzed Decarboxylative Borylation with NHPI Redox-Active Esters

Reaction conditions: (a) RAE (1.0 equiv), Cu(acac)2 (30 mol %), LiOH·H2O (15.0 equiv), MgCl2 (1.5 equiv), B2pin2 (3.0 equiv), dioxane/DMF, rt, 10 min. (b) Cu(acac)2 (20 mol %) and B2pin2 (1.5 equiv) were used. (c) 3.5 mmol scale. (d) In situ reaction with NHPI (1.0 equiv) and DIC (1.0 equiv). (e) [Ni] yield referred to reported yield in ref (2). (f) [] yield referred to reported yield in ref (3). (g) [PET] yield referred to reported yield in ref (4). (h) solvent MTBE/DMF was used. Ts = tosyl, Boc = tert-butyloxycarbonyl, Fmoc = fluorenylmethyloxycarbonyl. NHPI = N-hydroxyphthalimide, DIC = N,N’-diisopropylcarbodiimide.

Reaction conditions: (a) RAE (1.0 equiv), n class="Chemical">Cu(acac)2 (30 mol %), LiOH·H2O (15.0 equiv), MgCl2 (1.5 equiv), B2pin2 (3.0 equiv), dioxane/DMF, rt, 10 min. (b) Cu(acac)2 (20 mol %) and B2pin2 (1.5 equiv) were used. (c) 3.5 mmol scale. (d) In situ reaction with NHPI (1.0 equiv) and DIC (1.0 equiv). (e) [Ni] yield referred to reported yield in ref (2). (f) [] yield referred to reported yield in ref (3). (g) [PET] yield referred to reported yield in ref (4). (h) solvent MTBE/DMF was used. Ts = tosyl, Boc = tert-butyloxycarbonyl, Fmoc = fluorenylmethyloxycarbonyl. NHPI = N-hydroxyphthalimide, DIC = N,N’-diisopropylcarbodiimide.

As a testament to the chemoselectivity of the method, complex natural products and drug molecule acid derivatives (34–42) were further evaluated in this study (Scheme C). Thus, diverse scaffolds containing displacement-prone n class="Chemical">chlorides (34), ketones (35 and 38), lactones (37), acetonides (36), base-labile Fmoc groups (41) and alkenes (37 and 40) were left unperturbed under the reaction conditions. Six of the nine substrates investigated were obtained with an enhanced yield and reduced reaction time when compared to established methods.

Although this protocol was proved to be successful on a range of substrates, there are some limitations. Diminished yields were observed for acids containing aryl (43) or n class="Chemical">alkyl (44) bromides due to competitive protodebromination or borylation of bromide. Tertiary (45), benzylic (46), and α-amino (47) substrates are also nonproductive or low-yielding.

Following the substrate scope exploration, a preliminary mechanistic inquiry was conducted (Scheme E). Radical clock experiments, including cyclopropyl ring opening (48) and 5-exon class="Chemical">-trig radical cyclization (49), suggested the presence of alkyl radicals during this protocol, which is consistent with reported RAE decarboxylative coupling studies.[15]

After achieving the process requirements and demonstrating the generality of this Cu-catalyzed decarboxylative borylation class="Chemical">n, this protocol was ultimately tested on the genuine substrate 50. Gratifyingly, the desired product 2 was obtained in 64% yield with 1.5 equiv of B2pin2 and 20 mol % of Cu(acac)2 (Scheme ). As a further test of its scalability, boronate 2 was isolated in similar yield during 2.5 mmol scale synthesis, and the reaction was complete in less than 10 min. Not only was a higher yield achieved compared with our previous Ni-catalyzed borylation conditions, but the decreased stoichiometry of B2pin2, removal of strong base, and absence of exogenous ligands led to a dramatically less expensive decarboxylative borylation protocol ($224 per mole vs Ni $1565 per mole). This procedure also compares favorably from a cost perspective to the other known decarboxylative borylation methods.[16]

Scheme 2

Cu-Catalyzed Decarboxylative Borylation to Synthesize Target Compound 2 and Cost Comparisons with Other Methods

(a) 0.1 mmol scale. (b) 2.5 mmol scale. (c) The cost of raw materials for one mole reaction setup. See the Supporting Information for details. (d) B2cat2 prepared from B2(NMe2)4 and catechol. e) Ir[(ppy)2dtbpy]PF6 prepared from IrCl3 × H2O in 2 steps.

Kinetic studies of the borylation reaction were carried out on RAE derived from n class="Chemical">5-phenylvaleric acid to understand the concentration dependences and the role of the reaction components. Following the time course of the reaction under our initial standard conditions (Scheme A, entry 8) revealed that the reaction occurs extremely rapidly, with full consumption of the RAE within 3 min of initiating the reaction and 5-phenylvaleric acid forming as the major byproduct (see the Supporting Information for details). The concentration dependences were probed by varying initial concentrations separately for each component and comparing reaction rate to that under standard conditions. Parameters for increased rate include increased [B2pin2], [Cu(acac)2], and equivalents of LiOH·H2O, whereas reaction rate was not significantly influenced by changes in equivalents of MgCl2 or [RAE] (Scheme A). Interestingly, the fraction yield increased substantially in reactions run at lower [RAE] from 60% to 85%, indicating that under these conditions, a larger percentage of the RAE is directed toward the productive reaction pathway. This suggests that while the productive reaction exhibits zero order kinetics in [RAE], unproductive side reactions have a positive dependence on RAE concentration.

Scheme 3

(A) Kinetic Studies of Cu-Catalyzed Decarboxylative Borylation and Concentration Dependences for This Transformation; (B) Reactions Orders in [Cu] for Formation of Product and Major Byproduct; (C) Relationship between the Relative Rates of Productive and Side Reactions to mol % Cu

Selectivity to desired product is enhanced by high [Cu] and low [RAE]. Yields were determined by LC-UV.

We examined the role of each component in turn by removing them one at a time and probing the effect on both product yield and consumption of n class="Chemical">RAE (see the Supporting Information for details). It was found that a Cu(I) source does not significantly reduce product yield from that observed with the Cu(II) source used under standard conditions. Reactions carried out with either source of Cu removed from the reaction mixture afforded no product, but did consume RAE.

Removal of MgCl2 also resulted in no product formation class="Chemical">n, although RAE was consumed, albeit more slowly than under standard reaction conditions. Interestingly, removal of either LiOH·H2O or B2pin2 completely suppressed the consumption of RAE, suggesting a role for both species in activating the catalyst for product turnover as well as for side reactions of RAE.

Further examination of the role of [Cu(acac)2] in both productive and unproductive reactions is shown in Scheme B, where the concentration of product 11 and the major byproduct 5-phenylvaleric acid is plotted as a function of (time·[n class="Chemical">Cu]) according to the methodology of Burés.[17] Overlay for reactions at different [Cu] in Scheme B (top) indicates first order kinetics in [Cu] for the borylation reaction, while the lack of overlay for acid formation in the plot of Scheme B (bottom) suggests that the consumption of RAE in unproductive side reactions are not a simple function of the concentration of Cu. These results suggest the rate laws shown in equations and 2 that describe the formation of the borylation product and the overall consumption of RAE, respectively. The productive Cu-catalyzed reaction may be maximized at the expense of the Cu-free byproduct formation by maximizing the ratio given in eq . Employing dilute conditions of [RAE] and/or high catalyst loadings thus offers a practical means of increasing reaction efficiency. The selectivity dependence on the [RAE]/[Cu] ratio is illustrated in the model in Scheme C,[18] where the ratio of k′/k = 0.25 is derived from the data in Scheme B.

These results suggest the mechanistic picture shown in Scheme . The observed positive order kinetics in [LiOH·n class="Chemical">H2O] and [B2pin2] may be rationalized by rate-determining formation of a complex with Cu. This complex may then react with the RAE forming a radical complex that fragments to release CO2 and phthalimide.[19] Finally, radical recombination will result in the borylated alkyl product and thus close the productive catalytic cycle. The formation of side products from the RAE is noncatalyzed and occur in the presence of B2pin2 and LiOH·H2O.

Scheme 4

Proposed Cu-Catalyzed Borylation Mechanism

This work constitutes another example of the use of Cu-catalysis for inducing decarboxylation of n class="Chemical">redox-active esters.[12,20] In the present case, inexpensive and rapid access to boronic esters combined with a clear analysis of concentration dependences may facilitate large-scale adoption. The simple experimental protocol described herein might inspire other radical-based cross couplings and functional group interconversions using Cu.