Decarboxylative allylation of amino alkanoic acids and esters via dual catalysis.

A combination of photoredox and palladium catalysis has been employed to facilitate the room temperature decarboxylative allylation of recalcitrant α-amino and phenylacetic allyl esters. This operationally simple process produces CO2 as the only byproduct and provides direct access to allylated alkanes. After photochemical oxidation, the carboxylate undergoes radical decarboxylation to site-specifically generate radical intermediates which undergo allylation. A radical dual catalysis mechanism is proposed. Free phenylacetic acids were also allylated utilizing similar reactions conditions.

The site-specific generation of reactive intermediates from easily accessible carboxylic acids while producing CO2 as the only stochiometric byproduct is an attractive hallmark of decarboxylative coupling reactions.[1] One limiting factor of these transformations is that the rate of decarboxylation qualitatively tracks with the pKa of the resulting anion.[1a] Thus, decarboxylations to generate alkyl anion equivalents with pKa’s > 25 require high temperatures or do not occur at all. We sought to overcome this limitation by taking advantage of the facile radical decarboxylation of alkanoates as opposed to the recalcitrance of more common anionic decarboxylation. Ultimately, we hypothesized that photoredox decarboxylation could be combined with Pd-catalysis to effect the allylation of alkyl species that do not undergo anionic decarboxylation.[2]

The radical decarboxylation of various aryl acetic and α-amino acids following photo-[3] and electrochemical[4] oxidation has been studied for many decades. We were interested in beginning our investigation with (aminoaryl)acetic acids given that the nitrogen lone pair is more prone to oxidation than carboxylates and the resulting radical has been trapped post-decarboxylation.[3h,5] For example, photoredox catalysis[6] has been utilized for the decarboxylative radical addition of aminoarylacetic acids to electron-deficient alkenes by Nishibayashi[7] and for the arylation of α-amino acid derivatives by MacMillan.[8]

Palladium-catalyzed benzylic allylation is especially challenging because benzyl nucleophiles have pKa’s > 40, which limits their accessibility under mild conditions.[9] In 2011 the Walsh group disclosed the use of Cr(CO)3 to increase the acidity of the benzylic C–H bonds, which allowed for their deprotonation and use in Pd-catalyzed allylic substitution reactions (Scheme 1).[10] While this was a significant advancement in the allylation of benzylic nucleophiles, the approach requires the use of a stoichiometric base and chromium and suffers from the need to preform the organometallic complex and remove chromium after the reaction. We hypothesized that a combination of photoredox and palladium catalysis may allow for the direct allylation of benzyl species without the need for a stoichiometric base or activating agents.

Scheme 1

Benzylic Allylation

We began by screening conditions for the decarboxylative coupling of allyl ester 1a by combining Pd(PPh3)4 with various visible-light-mediated photoredox catalysts and monitoring the conversion to 2a/3a by GC/MS (Table 1). A solvent screen revealed that a polar aprotic solvent was required for reactivity and that acetonitrile gave the best conversion. After 9 h good conversion of starting material to desired product 2a and homocoupled benzylic radical product 3a was observed when 1 mol % of Ru(bpy)3[BF4]2 was used as the photocatalyst (entry 1). Employing more oxidizing[11] ruthenium based photocatalysts led to extremely low conversions, even after extended reactions times (entries 5, 6). Decreased conversion was also observed when strongly reducing[12] tris-cyclometalated iridium photocatalysts were used (entries 2, 4). In contrast, complete conversion was observed when Ir(ppy)2(bpy)[BF4], which possesses intermediate photoredox properties,[13,14] was employed (entry 3). Additional optimization revealed that the reaction reached full conversion in just 1 h (entry 7) and that both the palladium and photocatalyst loading could be reduced at the expense of a slightly longer reaction time (entry 8). No conversion was detected when the reaction was run in the absence of light, photocatalyst, Pd(PPh3)4, or in the presence of 2 equiv of the radical scavenger TEMPO (entries 9, 10, and 11).

Table 1

Screening of Photocatalystsa

0.25 mmol of 1a in 2.5 mL of MeCN was stirred under Ar with Pd(PPh3)4 and photocatalyst for the indicated time in front of a white LED. The crude reaction mixture was analyzed by GC/MS.

GC/MS ratios

32% isolated yield of 2a.

Isolated 44% 2a; 0.052 mmol, 41% 3a.

In the absence of light.

Next, different reaction concentrations, temperatures, and palladium sources/ligands were examined in an attempt to control the ratio of 2a and 3a. Despite an extensive screen (see Supporting Information), the ratio of desired allylation product to benzylic homodimerization adduct did not appreciably change from 2:1 in reactions that reached complete conversion. This observation led us to hypothesize that both benzyl and allyl radicals were being produced during the reaction leading to a statistical mixture of products.[15] Importantly, monitoring the reaction via 1H NMR spectroscopy revealed the formation of an allyl–allyl coupling product (1,5-hexadiene) and bibenzyl (3a) in a 1:1 ratio. While the nature of the radical coupling limits the maximum yield of the coupling, this disadvantage is far outweighed by the operational simplicity of accessing allylated alkane products without any stoichiometric additives.

With these conditions in hand, we began evaluating decarboxylative couplings of para-amino substituted phenylacetic acid allyl esters (Scheme 2, 2a–2h). Both cyclic and acyclic tertiary amines underwent radical decarboxylation to provide the desired product. Steric bulk about the benzyl radical did not dramatically improve the yield by decreasing the amount of homodimerization (2c). Substrates with an additional heteroatom that could undergo oxidation without participating in direct resonance with the carboxylate (2f–2h) were also successfully allylated. Importantly, no allylation was observed at the remote benzylic positions [−N(CH2Ph)2] of substrate 1d, which suggests that decarboxylation allows the site-specific generation and reaction of radicals.

Scheme 2

Scope of Amino Alkanoic Esters,

Reactions performed on a 0.25 mmol scale. Isolated yields. ∼95% pure see SI for more information.

Next, α-amino substituted phenyl acetic acid allyl esters were evaluated (2i–2m). Tertiary amines with benzyl (2l, 2m), aryl (2i, 2k), and alkyl (2j) substituents were all compatible with the standard reaction conditions. In addition, the successful coupling of a phenylalanine derivative to provide 2l shows that radical formation is not strictly limited to benzylic positions.

To further test the site-specificity of radical coupling, substrate 1m was synthesized and subjected to decarboxylative allylation conditions (Scheme 3). Indeed, only allylation at the site of the carboxylate was observed. Thus, photochemical decarboxylation allows the generation of the thermodynamically less favored radical, and that radical does not isomerize to the more stabilized radical under the reaction conditions.

Scheme 3

Regiospecific Allylation

Lastly, avoiding the preformation of allyl esters by generating the same reactive species in situ from phenylacetic acids and an allyl source such as allyl methyl carbonate was pursued. We hypothesized that oxidative addition to Pd(0) would generate a Pd-π-allyl species along with methyl carbonate which provides a base capable of generating the requisite carboxylate anion. Indeed, employing the previously optimized conditions to a 1:1 ratio of para-amino substituted phenyl acetic acid with allyl methyl carbonate resulted in the generation of the desired allylation products in comparable yields to those from the intramolecular process (Scheme 4).

Scheme 4

Allylation of Amino Alkanoic Acids,

Reactions performed on a 0.25 mmol scale. Isolated yields. The ratio of 2a/3a was measured by GC/MS to be 2:1.

On the basis of the observed experimental results, we tentatively propose a dual catalysis mechanism (Scheme 5). First, oxidative addition of the allyl ester to Pd(0) forms a Pd-π-allyl species and the carboxylate counterion. This carboxylate can be oxidized by the photoactivated iridium catalyst which triggers radical decarboxylation. Next, the IrII species may reduce the Pd(allyl) species by one electron,[16] resulting in homolysis to provide an allyl radical and regenerate the active Pd(0) and IrIII catalysts. Alternatively, it is possible that the benzyl radical adds to the Pd(allyl) complex to generate a Pd(III)-allyl species that undergoes reduction by the photocatalyst and reductive elimination.[2a,2g,17] In either case, it is noteworthy that the photocatalyst performs both the oxidation of the carboxylate and reduction of the Pd(allyl) complex. This electron shuttling mechanism avoids the need for stoichiometric sacrificial oxidants or reductants that are commonly required in photocatalysis. Additionally, the need to perform both oxidation and reduction is consistent with our observation that a photocatalyst with intermediate redox properties performs better in the decarboxylative allylation reaction.

Scheme 5

A Plausible Mechanism

In conclusion, we have developed an additive-free method for the allylation of alkanes that cannot undergo standard anionic decarboxylative allylation. The dual catalytic coupling proceeds by the site-specific generation of radicals and has the potential to greatly expand the scope of decarboxylative allylation chemistry.