Synthesis of Heterodiarylmethanes via Metallaphotoredox Decarboxylative Arylation

Abstract

A metallaphotoredox-catalyzed synthesis of heterodiarylmethanes using mild reaction conditions, commercially available substrates, and bench-stable catalysts is demonstrated. Moderate yields are obtained, and further derivatization of the newly formed benzylic position is shown.

Key words

metallaphotoredox - decarboxylative arylation - heterodiarylmethanes - photochemistry

Photochemistry is an exciting field of organic chemistry which has allowed for access to synthetic transformations that are not feasible under thermal conditions. When combined with transition-metal catalysis, the door has been opened towards the formation of bonds that were previously considered challenging, such as Csp²–Csp³ bonds.[1] The ability to form this type of bond provides access to various privileged scaffolds such as heterodiarylmethanes. This motif provides the backbone structure of many bioactive molecules (Figure [1])[2`] [b] [c] [d] and is therefore important for the development of novel pharmaceutically active compounds. As such, it would be advantageous to be able to synthesize these motifs under mild reaction conditions and from readily available starting materials.

While there is an abundance of synthetic routes towards the synthesis of diarylmethanes, fewer methods have been developed towards the synthesis of heterodiarylmethanes, most of which proceed via decarboxylative arylation.[3] Decarboxylation is an attractive method as it is more environmentally friendly, atom-economical, and synthetically more straightforward as a wide variety of carboxylic acids are commercially available.

Figure 1 Bioactive molecules with heterodiarylmethane core

As decarboxylative couplings typically require high temperatures, the use of metal catalysis is needed for these reactions to occur at room temperature. To date, only a handful of decarboxylative arylation methodologies have been developed towards the synthesis of heterodiarylmethanes using transition-metal catalysis,[4] electrophotochemistry,[5] and metallaphotoredox reaction.[6] The metallaphotoredox approach is appealing as it uses mild reaction conditions and commercially available catalysts. One drawback of many metallaphotoredox methodologies is the need for pre-functionalization of the alcohol or acid nucleophile into redox-active esters (e.g., xanthate[7a] or NHP esters[7b] [c]), thereby adding an additional synthetic step. Moreover, a collaborative effort between the Doyle and the MacMillan groups has demonstrated the use of metallaphotoredox decarboxylative arylation to couple α-heteroatom carboxylic acids with (hetero)aryl halides, without the need for pre-functionalization (Scheme [1a]). [8] Despite amino acids being the main focus of the scope of this methodology, there are two examples where aryl acetic acids are used to form the corresponding diarylmethanes. Inspired by these efforts, we were interested in applying a metallaphotoredox approach towards the synthesis of heterodiarylmethanes (see Scheme [1b]).

Scheme 1 Metallaphotoredox decarboxylative arylations towards the synthesis of diarylmethanes

At the onset of our reaction development, our goal was to synthesize a broad scope of heterodiarylmethanes using readily available and inexpensive commercial starting materials with no pre-functionalization of the substrates required. Additionally, we envisioned using a commercially available and bench-stable nickel catalyst and iridium photocatalyst to maximize the ease and adaptability of this approach. A Penn PhD Photoreactor M2 equipped with 450 nm blue LEDs was used in order to allow for consistency and ease of the reaction setup.[9] For optimization, we chose 2-(naphthalen-2-yl)acetic acid and 3-bromopyridine as our model substrates and by using MacMillan’s reaction conditions[8] as a starting point, we altered various reaction parameters from there (Table [1]).

Table 1 Reaction Optimization

Entry	Changes to above reaction conditionsa	NMR yield of 3a (%)b	NMR yield of 3a′ (%)b
1	none	trace	35
2	DMF (0.2 M)	trace	10
3	DMF (0.05 M)	23	42
4	DMA or DMSO or MeCN or dioxanes (0.05 M)	trace	2–20
5	DMF (0.05 M), Cs2CO3	23	35
6	DMF (0.05 M), K2CO3	51	8
7	DMF (0.05 M), K2CO3 c	73 (55)d	trace
8	DMF (0.05 M), K2CO3, 2a (1.5 equiv)c	52	12
9	DMF (0.05 M), K2CO3, 1a (1.5 equiv)c	44	20
10	DMF (0.05 M), K2CO3, phthalimide (1 equiv)c	75	trace

^a Reaction mixture was sparged with nitrogen for 5 min.

^{b 1}H NMR yields determined by using 1,3,5-trimethoxybenzene as an internal standard.

^c The headspace of the reaction mixture was purged with nitrogen for 10 s.

^d Isolated yields in parenthesis.

The use of MacMillan’s previously developed conditions[8] with our model substrates provided trace amounts of the desired product and 35% of the major byproduct, 2-methylnaphthalene (3a′), that results from the quenched radical after decarboxylation (entry 1). We first looked into changing the concentration of the reaction mixture, as shown in entry 2, doubling the solvent concentration was detrimental to product formation. As metallaphotoredox reactions often proceed better under more dilute conditions, we found that lowering the concentration from 0.1 M to 0.05 M (entry 1 vs. 3) increased product formation to 23%. A quick solvent screen (entry 4) showed that other polar aprotic solvents resulted in minimal product formation, confirming that DMF was the best solvent for this transformation. After executing a base screen, we found inorganic bases to be superior to organic bases. Using Cs₂CO₃, did not exhibit an improvement in the yield of 3a but did slightly decrease the amount of 3a′ formed (entry 3 vs. 5). After exploring other inorganic bases, we found that K₂CO₃ increased the yield of the desired product to 51% while suppressing the formation of 3a′ (entry 6).

Amid our optimization trials, we found a publication which detailed the impact of oxygen specifically on metallaphotoredox decarboxylative arylations.[10] This paper concluded that the complete exclusion of molecular oxygen can result in a decrease in product formation. Therefore, rather than sparging the reaction prior to irradiation, we purged only the headspace of the vial for 10 s. As a result, we observed an increase in product formation from 51% to 73% (entry 6 vs. 7), with the formation of 3a′ being completely suppressed.

Additionally, we explored whether having an excess of either substrate would be beneficial to the reaction. Using 1.5 equiv of 2a resulted in a similar yield as when 1.1 equiv was used (entry 7 vs. 8) but led to an increase in the formation of 3a′. Alternatively, using 1.5 equiv of the acid substrate resulted in an increase in byproduct formation and a decrease in the amount of product formed (entry 9).

Also during the optimization, the group of MacMillan published a complimentary paper highlighting how the use of phthalimide as an additive to metallaphotoredox decarboxylative arylations could dramatically increase the yields of more challenging substrates.[11] Albeit, under our newly identified optimal reaction conditions, the addition of 1.0 equiv of phthalimide (entry 10) was not beneficial Moreover, our control experiments confirmed the necessity of each component of the reaction, as conducting the reaction without either catalyst or without a light source resulted in no product formation (see the Supporting Information for details).

After uncovering that the rigorous exclusion of oxygen from the reaction mixture was not necessary, we were interested in investigating the impact of other external factors on the reaction. Using the optimized reaction conditions as the control (Table [1], entry 7), a sensitivity screen was conducted following the method detailed by Glorius.[12] Figure [2] illustrates the main parameters that were evaluated, such as the sensitivity of the reaction to water, oxygen, and changes in solvent concentrations (see the Supporting Information for details on the parameters). The yields presented in the figure are ¹H NMR yields, determined using 1,3,5-trimethoxybenzene as the internal standard and each reaction was performed on a 0.13 mmol scale (with the exception of the larger-scale entry).

As observed during optimization, increasing the solvent concentration to 0.2 M was detrimental. While the reaction benefits from more dilute reaction conditions, reducing the concentration further to 0.025 M gave an NMR yield similar to that of our optimized conditions. The screen also revealed the sensitivity of the reaction to water as adding 1% water reduced the yield, whereas using 4 Å molecular sieves did not dramatically change the yield compared to that of the control. Additionally, the reaction did not proceed effectively when open to air but as mentioned previously, we found that only purging the headspace of the vial with nitrogen was required, as sparging the reaction mixture for 5 min did not greatly increase the amount of product formed. Unsurprisingly, the reaction did not proceed when lowering the intensity of the LED lamp from 90% to 10%, as expected for a photochemical transformation. One of our final tests was to use 3-chloropyridine as the halide substrate which afforded only trace amounts of product. On the other hand, the amount of product formed when using 3-iodopyridine was comparable to that of the control. We also tested the scalability of the reaction where we found that despite increasing the scale from 0.13 mmol to 1.3 mmol, minimal impact on product formation was observed.

Figure 2 Sensitivity screen

With the optimized reaction conditions in hand, we began exploring various (hetero)aryl bromides and aryl acetic acid derivatives to give the corresponding heterodiarylmethanes in modest yields (Scheme [2]). To start, a nitrogen walk around the pyridine core was completed, where the use of 4-bromopyridine (3b) and 2-bromopyridine (3c) resulted in 23% and 17% isolated yields, respectively. Furthermore, in altering the heteroaryl bromide, we were able to successfully incorporate both a pyrimidine and a pyrazine, 3d and 3e in 42% and 23% yield, respectively. Using 3,5-dibromopyridine, we initially believed that if we doubled the amount of acid and iridium catalyst used, we would obtain the double-addition product. Instead, we found that the single-addition product was the major product to give 3f in 21% yield, and only trace amounts of the double addition product or the monodebrominated product were observed by ¹H NMR spectroscopy (<5% of each).

Scheme 2 Substrate scope of heterodiarylmethanes

Next, using differentially substituted 3-bromopyridines, we explored how the electronics impacted the reaction. Incorporating either an EWG or an EDG demonstrated no improvement in yield when comparing 3a to 3g or 3h. Employing the same exercise on the aryl acetic acid substrate, we found that EWGs shut down the reaction entirely (see the Supporting Information for a list of unsuccessful substrates). On the other hand, we thought that adding an EDG could aid in stabilizing the benzylic radical and thereby help to further suppress the formation of the major byproduct to increase conversion into the desired product. Unfortunately, the use of 4-methoxyphenylacetic acid did not show a great improvement in yield (3i). Alternatively, the combination of 4-methoxyphenylacetic acid and 5-trifluoromethyl-3-bromopyridine gave a boost in yield as 3j was isolated in 48% yield. At this point, we realized that having an EDG on the (hetero)aryl acetic acid and an EWG on the (hetero)aryl bromide was the ideal balance of electronics. This allowed us to flip the coupling partners to use a pyridyl acetic acid substrate with 1-bromo-4-(trifluoromethyl)benzene to provide the corresponding heterodiarylmethane product (3k) in 33% yield. Following suit, we used the balanced electronics to isolate our first diheteroarylmethane (3l) in a 41% yield.

Scheme 3 Benzylic functionalization

The newly formed benzylic position has advantages and drawbacks. A potential disadvantage of this scaffold, from a drug development perspective is that the reactivity of the benzylic position creates a metabolic hotspot. This reactive site can be blocked by functionalization, ideally using small substituents that do not drastically alter the physiochemical properties of the molecule. Fortunately, the reactivity of the benzylic position allows for late-stage mono- or difunctionalization to provide further diversification of medicinally relevant compounds.[13] Thereby, to demonstrate the synthetic utility of the heterodiarylmethane scaffold, we sought to derivatize the benzylic position of 3a (Scheme [3]) using classical transformations. Of note, these transformations were performed only to explore the potential reactivity of these scaffolds, thus the reaction conditions were not optimized.

To start, oxidizing the benzylic position would give a useful intermediate for several subsequent transformations. Using a copper-catalyzed radical oxidation afforded the ketone intermediate 4a in 52% yield.[14] From there, we performed a Grignard addition using methyl magnesium bromide to get the methyl alcohol derivative 5a.[15] Furthermore, as we aimed to install small substituents, we sought to difluorinate the benzylic position which was achieved by simply stirring 4a in neat DAST to give 6a in a 23% yield.[16] Another small substituent that we were interested in incorporating onto the heterodiarylmethane motif was a cyclopropane ring. From 4a, a Wittig reaction was done to synthesize 7a (56% yield) which underwent a Simmons–Smith cyclopropanation to afford 8a in a 19% yield.[17]

In summary, we have synthesized a variety heterodiarylmethanes via decarboxylative metallaphotoredox using commercial starting materials and bench-stable catalysts.[18] We have also identified that conducting the reaction with the complete exclusion of air was unnecessary and led to a decrease in product formation. Furthermore, late-stage functionalization of these scaffolds was explored by further functionalizing the newly formed benzylic position to expand the synthetic utility of this motif.

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgment

We thank Paraza Pharma Inc. and BioTalent Canada (Co-Op internship for A.Y.). We also thank Serge Bourg for the HRMS analysis, Augustin Péneau for the NMR assistance and Dilan Polat for the fruitful discussions.

• References and Notes

• 1a Chan AY, Perry IB, Bissonnette NB, Buksh BF, Edwards GA, Frye LI, Garry OL, Lavagnino MN, Li BX, Liang Y, Mao E, Millet A, Oakley JV, Reed NL, Sakai HA, Seath CP, MacMillan DW. C. Chem. Rev. 2021; 122: 1485
• 1b Levin MD, Kim S, Toste FD. ACS Cent. Sci. 2016; 2: 293
• 2a Papaverine: https://pubchem.ncbi.nlm.nih.gov/compound/Papaverine
• 2b Pheniramine: https://pubchem.ncbi.nlm.nih.gov/compound/4761
• 2c Trimethoprim: https://pubchem.ncbi.nlm.nih.gov/compound/Trimethoprim
• 2d Furegrelate: https://pubchem.ncbi.nlm.nih.gov/compound/Furegrelate
• 3 Mondal S, Panda G. RSC Adv. 2014; 4: 28317
• 4 Moon PJ, Lundgren RJ. ACS. Catal. 2020; 10: 1742
• 5 Yang K, Lu J, Li L, Luo S, Fu N. Chem. Eur. J. 2022; 28: e202202370
• 6a Beil SB, Chen TQ, Intermaggio NE, MacMillan DW. C. Acc. Chem. Res. 2022; 55: 3481
• 6b Gesmundo NJ, Tu NP, Sarris KA, Wang Y. ACS Med. Chem. Lett. 2023; 14: 521
• 7a Vara BA, Patel NR, Molander GA. ACS Catal. 2017; 7: 3955
• 7b Behnke NE, Sales ZS, Li M, Herrmann AT. J. Org. Chem. 2021; 86: 12945
• 7c Tao M, Zeng L.-Y, Li W, Pu G, Jia J, Yao Q, Li X, He C.-Y. Adv. Synth. Catal. 2023; 365: 854
• 8 Zuo Z, Ahneman DT, Chu L, Terrett JA, Doyle AG, MacMillan DW. C. Science 2014; 345: 437
• 9 Le C, Wismer MK, Shi Z.-C, Zhang R, Conway DV, Li G, Vachal P, Davies IW, MacMillan DW. C. ACS Cent. Sci. 2017; 3: 647
• 10 Oderinde MS, Varela-Alvarez A, Aquila B, Robbins DW, Johannes JW. J. Org. Chem. 2015; 80: 7642
• 11 Prieto Kullmer CN, Kautzky JA, Krska SW, Nowak T, Dreher SD, MacMillan DW. C. Science 2022; 376: 532
• 12 Pitzer L, Schäfers F, Glorius F. Angew. Chem. Int. Ed. 2019; 58: 8572
• 13 Gulati U, Gandhi R, Laha JK. Chem. Asian. J. 2020; 15: 3135
• 14 Nawratil S, Grypioti M, Menendez C, Mallet-Ladeira S, Lherbet C, Baltas M. Eur. J. Org. Chem. 2014; 654
• 15 Khadka DB, Le QM, Yang SH, Van H TM, Le TN, Cho SH, Kwon Y, Lee K.-T, Lee E.-S, Cho W.-J. Bioorg. Med. Chem. 2011; 19: 1924
• 16 Shook BC, Rassnick S, Wallace N, Crooke J, Ault M, Chakravarty D, Barbay JK, Wang A, Powell MT, Leonard K, Alford V, Scannevin RH, Carroll K, Lampron L, Westover L, Lim H.-K, Russell R, Branum S, Wells KM, Damon S, Youells S, Li X, Beauchamp DA, Rhodes K, Jackson PF. J. Med. Chem. 2012; 55: 1402
• 17 Wang D, Xue X.-S, Houk KN, Shi Z. Angew. Chem. Int. Ed. 2018; 57: 16861
• 18 Representative Experimental Procedure: 3-(Naphthalen-2-ylmethyl)pyridine (3a) To a vial charged with 2-(naphthalen-2-yl)acetic acid (100 mg, 532 μmol), 3-bromopyridine (57.5 μL, 585 μmol), dtbbpy (14.6 mg, 53.2 μmol), NiCl₂(glyme) (11.9 mg, 53.2 μmol), K₂CO₃ (225 mg, 1.59 mmol), and {Ir[dF(CF₃)ppy]₂(dtbpy)}PF₆ (5.96 mg, 5.32 μmol) was added anhydrous DMF (10.6 mL, 0.05 M). The headspace of the vial was purged with nitrogen for 10 s, placed in a Penn PhD Photoreactor M2 equipped with a 450 nm blue LED light and left to stir for 16 h under irradiation. To the reaction was added saturated NaHCO₃ and was extracted with EtOAc (3×), the combined organic layers were washed with water (4×) then brine, dried over anhydrous Na₂SO₄, filtered, and evaporated. The residue was purified on a 10 g silica gel column using a Biotage Isolera automated purification system with a gradient elution of EtOAc/heptanes (0–100%) and a flow rate of 36 mL/min over 20 min to afford the title product 3a as a tan solid. Yield 64 mg (55%). ¹H NMR (400 MHz, CDCl₃): δ = 8.58 (d, J = 1.7 Hz, 1 H), 8.49 (dd, J = 4.9, 1.4 Hz, 1 H), 7.84–7.76 (m, 3 H), 7.63 (s, 1 H), 7.57 (d, J = 7.9 Hz, 1 H), 7.50–7.42 (m, 2 H), 7.31–7.27 (m, 2 H), 4.16 (s, 2 H). ¹³C NMR (101 MHz, CDCl₃): δ = 150.3, 147.8, 137.4, 136.7, 136.6, 133.7, 132.3, 128.6, 127.8, 127.7, 127.4, 127.3, 126.4, 125.8, 123.7, 39.3. LC–MS (ESI): m/z [M + H]⁺ = 220.1, t _R = 1.17 min; 5–100% MeCN/H₂O (0.1% of 10 mM ammonium formate buffer) over 3 min.

Publication History

Received: 04 December 2023

Accepted after revision: 28 December 2023

Article published online:
26 January 2024