Deprotection of benzyl ethers via solvent anion-mediated EDA complex

Abstract

Benzyl ethers are among the most robust and indispensable protecting groups in the synthesis of complex natural products and pharmaceuticals, providing effective neutralization of the reactive hydroxy group in phenols and alcohols. However, their cleavage necessitates the use of expensive catalysts, hazardous conditions, or excessive organic additives. Herein, we report a catalyst-free, photoinduced debenzylation protocol of phenolic and aliphatic ethers. By leveraging in situ generated solvent-anions as the reductant, the protocol achieves chemoselective debenzylation in the presence of alkenes, amides, carbamates, alcohols, and boronic esters in high yields under mild conditions. This strategy circumvents the need for costly hydrogenation catalysts or toxic reagents, providing a sustainable and economically viable alternative for debenzylation.

Benzyl ether deprotection is achieved under mild contions using only base, solvent, and light. The reaction proceeds via an EDA complex between the solvent anion and the benzyl group, providing good functional group compatibility.

Green foundation

1. This work introduces a catalyst-free, atom-economical reductive deprotection of benzyl ethers via electron donor–acceptor (EDA) complexes. By utilizing solvent anions as donors, the method eliminates the need for expensive photocatalysts, sacrificial additives, and hazardous pressurized hydrogen gas required by traditional Pd/C hydrogenation. The only reagents needed for this deprotection are solvent, base, and light.

2. We achieved excellent yields for phenols and aliphatic alcohols with superior functional group tolerance compared to traditional DDQ or Pd/C methods with orthogonality for N-boc protected amines and alkenes. The method significantly reduces the cost of debenzylation by 85%–99% when compared to photocatalytic procedures and is kinetically fast. When the reaction is scaled up to 5.8 mmol, we were able to obtain an 83% yield.

3. Future research will focus on the implementation of flow procedures for large scale deprotection. We aim to eliminate hazardous chromatography solvents like hexane/ethyl acetate and further optimize efficiency for aliphatic substrates to enhance broad synthetic utility.

Introduction

Phenols and alcohols are fundamental structural components in a wide range of natural products. Found in over 33% of pharmaceuticals, these functional groups are often essential for the antioxidant and antimicrobial properties of bioactive molecules.^1–3 Additionally, the hydroxy group (–OH) is among the most versatile synthetic handles for molecular diversification.^4,5 However, the inherent nucleophilicity and acidity of these moieties prove troublesome during multistep syntheses, requiring the use of protecting groups (PG) to adequately prevent unwanted side reactions.

While silyl groups are the most prevalent protecting groups for hydroxyl motifs in industry due to their straightforward installation and removal, their utility is limited by instability in acidic or fluoride-rich environments.⁶ Consequently, the benzyl (Bn) motif remains an indispensable alternative in synthetic strategies, offering easy installation and exceptional stability across a broad range of reaction conditions.^7–9

Traditionally, the cleavage of benzylic C–O bonds has been dominated by two established manifolds: metal-catalyzed hydrogenolysis and oxidative fragmentation (Fig. 1A). While hydrogenation is a classic and widely used approach, the inherent ability for spontaneous ignition of Pd/C and the requirement for pressurized hydrogen gas represent significant safety risks when it comes to large-scale synthetic applications.^10–14 Additional downsides to hydrogenolysis include poor chemoselectivity in presence of alkenes, amides, and nitro groups, as well as catalyst poisoning/deactivation in presence of substrates with amines and thiol moieties.^10,15 On the other hand, oxidative strategies utilizing stoichiometric reagents such as ceric ammonium nitrate (CAN) or DDQ suffer from poor atom economy, generation of substantial toxic waste that cannot be reused (including hazardous heavy metals or difficult to remove organic byproducts), and potential over oxidation of the desired product.^16–21 These protocols also require the use of the costly and difficult to procure para-methoxybenzyl (PMB) protecting group, which is easier to oxidize.²² While photo-excited DDQ strategies have been developed by Seeberger, Pieber, and Toshima to enable the deprotection of unactivated benzyl ethers, these methods still suffer from the aforementioned toxicity and waste limitations.^23,24

Fig. 1. (A) Traditional methods for benzyl deprotection of phenols and alcohols. (B) Photocatalytic protocols for debenzylation. (C) Our previous work in EDA chemistry with solvent anions. (D) Reaction cost analysis of photocatalytic systems (photocatalysts, additives, and protection precursor Bn vs. PMB) for a 0.2 mmol reaction. (E) This work.

As the structural complexity of modern targets increases, so has the demand for cost effective, sustainable, orthogonal, and robust deprotection methodologies that can streamline advanced synthetic routes.²⁵ As such, the development of visible light photoredox catalysis has provided a platform for the mild deprotection of benzyl ethers, enabling complementary and chemoselective strategies compared to the harsh conditions previously mentioned. Early efforts focused primarily on oxidative pathways; seminal work by the Stephenson group demonstrated that iridium photocatalysts could cleave benzyl ethers via single-electron transfer (SET) oxidation. In spite of the elegance of this strategy, the method is limited to aliphatic alcohols protected by the electron rich PMB group. Additionally, the method requires the use of toxic BrCCl₃ as a terminal oxidant.^26,27 Subsequent work by the Cheng, Woo, and Jiang groups further expanded this manifold utilizing Eosin Y, acridinium, and uranyl-based photocatalysts, respectively (Fig. 1B).^28–30 Despite these advancements, the protocols are hindered by the generation of persistent organic byproducts and poor atom economy, typically requiring multiple equivalents of sacrificial additives or oxidants such as hydrogen peroxide or the persistent pollutant trifluoro acetic acid (TFA).³¹

In contrast to oxidative approaches, reductive photocatalytic strategies for C–O bond cleavage have remained underexplored despite their ability to deprotect commonly used and relatively cheaper electron-neutral benzyl groups. The Matsubara group was the first to disclose a reductive cleavage method utilizing a carbazole photocatalyst. While the system enabled efficient phenolic benzyl ether cleavages, the reaction had poor functional group tolerance and only one example of aliphatic alcohol was presented in low yield.³² More recently, the Xia group expanded this reductive manifold using 2,6-di-tert-butyl-4-phenylphenol (DBPP), demonstrating excellent functional group tolerance (Fig. 1B).³³ However, the protocol is constrained by the high cost of the catalyst and the requirement for multiple additives involving cesium salts and n-butanethiol, a volatile and malodorous HAT agent. More recently, the Ota and Yamaguchi group demonstrated a reductive benzylic deoxygenation with zirconocene and an iridium photocatalyst, however, the reliance on a complex dual catalytic system undermines the atom economical requirements for multistep synthesis.³⁴

Our lab has explored the versatility of electron donor–acceptor (EDA) complexes, as a means to eliminate costly and potentially toxic external photocatalyst demonstrating a wide variety of photoinduced transformations: from aryl halide functionalizations,^35–37 to C(sp³)–C(sp³) cross-couplings.^38–40 Specifically, we demonstrated that solvent anions form EDA complexes capable of reducing electron-rich aryl halides, thereby enabling the chalcogenation, phosphonation, and borylation of these aromatic building blocks (Fig. 1C). See reviews for further work involving EDA systems.^41–47 From these precedents, we surmised that EDA complexes between solvent anion and benzyl ethers could lead, upon reduction, to the mesolytic cleavage of the etherial C–O bond (Fig. 1E), thereby offering an atom-economical and sustainable deprotection approach that eliminates the requirement for costly photocatalysts, toxic reagents, and hazardous set-ups. This method would significantly reduce the cost of debenzylation by 85%–99% when compared to photocatalytic strategies (Fig. 1D).

Experimental

Initial deprotection investigations used 1a (4-(benzyloxy)-1,1′-biphenyl) as our benzyl protected model substrate, potassium tert-butoxide (KO^tBu) as a base, and DMF under 390 nm irradiation, which afforded quantitative deprotection to the phenol 1 after 16 h (Table 1). Solvent screening revealed that MeCN (entry 2) resulted in a significant decrease in yield to 57%. This result further corroborates previous observations from our lab showing that the acetonitrile anion could also behave as a donor in EDA complexes, given that acetonitrile radicals have been trapped in some reactions.³⁶ DMA and DMSO (entries 3 and 4) showed full conversion to the desired phenol. Interestingly, both DMF and DMSO were found to be optimal solvents for phenolic substrates, while DMF was preferred for aliphatic alcohols.

Table 1. All reactions performed at 0.2 mmol, ¹HNMR yields obtained using 1,2-dibromomethane as internal standard.

Entry	Variation of standard conditions	Yield (%)
1	None	100
2	MeCN instead of DMF	57
3	DMSO instead of DMF	99
4	DMA instead of DMF	95
5	NaOtBu instead of KOtBu	91
6	Cs2CO3 instead of KOtBu	70
7	NaH instead of KOtBu	80
8	427 nm instead of 390 nm	100
9	370 nm or 440 nm	86, 34
10	30 min reaction time	98
11	No light	0
12	No base	0

Evaluation of the base indicated that the reaction is relatively insensitive to the counterion used, showing only a slight decrease in yield to 91% when using NaO^tBu (entry 5), presumably due to lower solubility (Table 1). When switching to a weaker carbonate base (entry 6) the yield dropped to 70%, in light of recent carbonate-mediated EDA chemistry from our group, this entry suggests that a less efficient base-assisted photochemical pathway may operate under non-optimal conditions.³⁷ On the other hand, the strong base NaH (entry 7) had a yield of 80%. Investigation into the light source confirmed that lower energy 427 nm irradiation was sufficient for conversion (entry 8), while 390 nm was selected for the standard. However, higher energy UV irradiation (370 nm) and lower energy 440 nm lights (entry 9) resulted in diminished yields of 86% and 34%, respectively. Notably, the reaction kinetics are relatively high; quantitative conversion was achieved within 30 minutes (entry 10) with this model substrate, and times as short as 10 minutes afforded 87% yield. Reaction times varied significantly across substrates, therefore all subsequent reactions were performed overnight for 16 hours. Control reactions without light (entry 11) or without base (entry 12) showed no product formation, confirming the photochemical nature of the transformation and the need for a base. A complete table of optimization is presented in the SI, page SI–S10.

Results and discussion

With the optimized conditions established, we evaluated the substrate scope of our protocol, starting with a series of substituted phenolic benzyl ethers (Scheme 1). The presence of a strongly electron-withdrawing p-cyano group (2) resulted in a noteworthy decrease in yield (61%), possibly due to the competitive reduction of the nitrile-substituted ring by the solvent anion, which may interfere with the intended C–O bond fragmentation. Electron-rich and electron-neutral substrates, such as para-methoxy phenol 3 and para-methyl phenol 4, afforded the corresponding products in excellent yields of 84% and 88%, respectively. The synthetic handle boronic ester showed a decreased yield to 55% (5), possibly due to competitive binding of the alkoxide to the boron, with most of the remainder mass balance being recoverable starting material. The method proved exceptionally effective for polycyclic aromatics, with 1- and 2-benzyloxynaphthalenes (6 and 7) delivering quantitative or near quantitative deprotections based on ¹H-NMR yields. This efficiency extended to highly electron-rich and sterically demanding systems, as shown by the ortho-dimethyl phenol (8, 84%). 3,4,5-Trimethoxy phenol derivative 9 showed a diminished yield (50%), but the remainder mass balance corresponded to unreacted starting material. Finally, highly sterically demanding ortho-tert-butyl phenol 10 afforded the desired deprotected product in 94% yield.

Scheme 1. Substrate scope. Reactions were performed on a 0.2 mmol scale. Yields in parentheses represent ¹HNMR yields obtained using 1,2-dibromomethane as internal standard.

To further demonstrate the synthetic utility and broad applicability of this protocol, we extended our investigation to more structurally complex substrates and natural product derivatives. We first examined an α-O-4 lignin model dimer (11), a motif prevalent in biomass degradation studies;^48,49 however, a diminished yield of 25% was observed. This result may be attributed to the electronic-rich nature of the dimethoxy-substituted benzyl goup in this lignin motif. In contrast, the deproctection of benzyl-protected β-estradiol 12 proceeded with high efficiency (83%) despite the presence of a free secondary alcohol.

The method also showed excellent chemoselectivity toward sensitive nitrogen containing and unsaturated substrates. N-Boc protected dopamine derivative 13, afforded the desired double debenzylation in 59% yield, while also leaving the Boc protecting group intact. Furthermore, benzyl-protected capsaicin 14 was successfully deprotected in 59% yield, leaving the alkene and secondary amide unreacted, thereby demonstrating that this debenzylation approach is complementary to hydrogenation strategies, and orthogonal to alkene functionalizations. Finally, the debenzylation of tocopherol 15 (Vitamin E) was achieved, albeit in a modest 25% yield. These results highlight the ability of this solvent-anion system to operate in the presence of various sensitive moieties (alkene, amide, N-boc, alcohol, boron) in complex molecular architectures, including steroids, neurotransmitters, and antioxidants.

Next, we investigated the impact of substitutions on the benzyl protecting group. Deprotection of para-tert-butyl benzyl (16) and para-fluoro benzyl (17) derivatives afforded the [1,1-biphenyl]-4-ol in 99% and 74% yields, respectively. Despite its electron-richness, we were pleased to see that PMB-protected phenol 18 was efficiently deprotected in 96% yield. Similarly, the meta-methyl benzyl derivative 19 afforded the desired phenol in 84% yield. When the protecting group was extended to a 2-naphthylmethyl motif (20), a slight decrease in yield (70%) was observed. We further explored the effect of substitution at the benzylic position. The use of a secondary benzylic ether (21, 1-phenylethyl) resulted in a diminished yield of 76%. We also tested the highly hindered trityl protecting group (22), which was successfully cleaved in 80% yield, thereby providing a complementary basic method for trityl deprotection compared to traditional acidic procedures.^50,51 Finally, we tried the cleavage of an allyl ether, which delivered the phenol in 48% yield. These results demonstrate that while the unsubstituted benzyl group remains optimal, our system is capable of cleaving a variety of substituted benzylic and related protecting groups.

Next, we applied our deprotection strategy to the more challenging benzyl protected aliphatic alcohols. Deprotection of a secondary benzyl protected alcohol 24 afforded the corresponding product in 38% yield. It is worth highlighting that for all aliphatic alcohol substrates, the remainder mass balance was unreacted recoverable starting material. To investigate the effect of increased steric bulk around the alcohol, we subjected benzyl-protected 1-adamantol (25) to the reaction conditions, which afforded the desired product in 51% yield. Similarly, the debenzylation of a protected menthol derivative (26) proceeded with 56% conversion to the alcohol. Medicinally important scaffolds, such as N-Boc-nortropine (27), also underwent deprotection in 44% yield. Additionally, the protocol proved effective for primary benzylic ethers, as demonstrated by the cleavage of 28 in 41% yield. Finally, a double cleavage was successfully performed on substrate 29, despite the presence of both phenolic and aliphatic benzyl ethers. This dual C–O bond cleavage afforded the desired product in 41% yield.

To gain insight into the reaction mechanism, a series of radical trapping experiments were conducted (Scheme 2A). The addition of radical scavengers such as TEMPO, 1,1-diphenylethylene (1,1-DPE), and 2,6-di-tert-butyl-4-methylphenol (BHT) resulted in complete inhibition of the deprotection, with no detectable formation of phenol 1. Further analysis via GC-MS identified the trapped benzyl radical species, and HRMS analysis supported the formation of a trapped DMF radical. These results strongly suggest a radical-mediated pathway and support the hypothesis that the solvent anion acts as the electron-donor in the system.

Scheme 2. (A) Radical trapping experiments performed on a 0.2 mmol scale with each radical scavenger; trapped species detected using GC-MS and HRMS. (B) Scaled up deprotection experiment using 1a. (C) Competition experiment between the benzyl and PMB ether showing the yield of collected starting material. Yield in parenthesis is crude yield by ¹HNMR. (D) UV-Vis experiment.

To demonstrate the practical scalability of this protocol, a large-scale deprotection was performed using 1.5 g (5.8 mmol) of starting material (1a) in a standard round-bottom flask. Under irradiation from two Kessil lamps, the reaction afforded the desired phenol (1) in an 83% (0.81 g) isolated yield. Notably, the benzyl homocoupling byproduct, resulting from the dimerization of the generated benzyl radicals, was isolated and characterized, providing further evidence for the proposed mesolytic fragmentation mechanism (Scheme 2B). Successful scale-up underscores the operational simplicity of the method and its potential for broader synthetic and industrial application.

We surmised that we could achieve chemoselective deprotection of alcohols based on the electronic nature of the benzyl protecting groups; with electron-rich PMB-protected alcohols being slower to remove than electron-neutral or deficient benzyl derivatives. To demonstrate this hypothesis a competition experiment between benzyl-protected phenol 1a and PMB-protected 18a was performed (Scheme 2C). We were pleased to observe good selectivity where substrate 1a was completely deprotected under the reaction conditions while 18a remained majorly unreacted. Such deprotection selectivity could be used in the synthesis of complex scaffolds containing multiple alcohols or phenols.

Our proposal of an EDA complex was further corroborated by UV-Vis spectroscopic studies; a significant increase in absorbance was observed upon the combination of 1a, KO^tBu, and DMF, indicative of a complexation between the carbamoyl anion and the phenol (Scheme 2D).

To gain insight into our mechanism, density functional theory (DFT) calculations were performed. First, we probed the nature of the EDA complex using a series of pre-associated geometries (see SI, page SI–S83). All optimized structures correspond to weakly bound, non-covalent assemblies, consistent with EDA complexes in polar media. Two geometries, that were initiated from distinct orientations, converged to minima within <1 kcal mol⁻¹ of each other, indicating a shallow association landscape with multiple accessible binding modes. Among these, geometries featuring interaction between the DMF-derived donor and the aryl acceptor fragment were identified as the most stable and were selected for further analysis. Notably, the lowest-energy conformer exhibits a large ground-state dipole moment (μ = 17.06 D), consistent with significant donor–acceptor polarization. Formation of the EDA complex is slightly endergonic (ΔG ≈ 4.4 kcal mol⁻¹) which supports weak, reversible association in polar media.

Time-dependent DFT (TD-DFT) calculations of this conformer reveal a manifold of low-lying singlet excited states in the visible region (λ = 458.7, 436.6, and 426.3 nm). Among these, the S₃ state (f = 0.0166) (see Table S4, page SI–S79) was identified as the most relevant electronic excitation based on its appreciable oscillator strength and clear donor-to-acceptor charge-transfer character.^52–54 Natural transition orbital (NTO) analysis shows a near-unity hole-particle eigenvalue (λ = 0.99994), consistent with a well-defined single excitation, with the hole localized on the DMF-derived donor fragment (Scheme 3A) and the electron density distributed over the benzyl ether π-system (Scheme 3B) (see SI Fig. S1, page SI–S82).^55,56 This assignment is further supported by frontier orbital analysis, indicating a HOMO (donor, Scheme 3C) → LUMO (acceptor, Scheme 3D) transition. The computed excitation energy (∼2.70 eV, ∼62 kcal mol⁻¹) places this state within the experimental conditions.

Scheme 3. The natural transition orbital visualization of the (A) hole and (B) particle. The frontier molecular orbital visualization for the (C) HOMO and (D) LUMO of the lowest energy complex. (E) Potential energy profile for the EDA-mediated C–O bond cleavage pathway. Formation of the EDA complex is slightly endergonic (4.4 kcal mol⁻¹), followed by photoexcitation to a charge-transfer state (S₃, 66.9 kcal mol⁻¹). Electron transfer generates a radical ion and DMF radical (−4.1 kcal mol⁻¹) and the reaction is overall exergonic. Energies (kcal mol⁻¹) are referenced to separated reactants and obtained at the APFD/DGTZVP level with CPCM (DMSO). (F) Proposed mechanism.

To further evaluate the feasibility of the proposed mechanism, a free energy profile was constructed based on DFT calculations (Scheme 3E). The combination of the model substrate 1a (A) and carbamoyl anion (B) was arbitrarily set at 0 kcal mol⁻¹. The formation of the ground state EDA complex (C) in a polar media required 4.4 kcal mol⁻¹ and photoexcitation to the excited state (C*) required 62.5 kcal mol⁻¹, giving a 66.9 kcal mol⁻¹ difference from the reference point. The LED used at 390 nm provides up to 73.3 kcal mol⁻¹, which is sufficient to achieve this photo excitation. Subsequent electron transfer generates a radical ion pair intermediate (D) and carbamoyl radical (E) that is significantly stabilized relative to the excited state and lies at −4.1 kcal mol⁻¹ from the ground-state. From this intermediate, C–O bond cleavage occurs to generate the phenolate anion (F), benzyl radical (G), and carbamoyl radical (E). Repeated attempts to locate a discrete transition state for this C–O bond cleavage led to either failure to converge or relaxation back to the ground-state D; this is consistent with a shallow and dissociative potential energy surface of similar mesolytic cleavages.⁵⁷ Full thermodynamic analysis of the phenolate anion and benzyl radical indicates that C–O bond cleavage is possibly exergonic. We also considered a possible product-assisted EDA complex in which the phenoxide generated after initial C–O bond cleavage acts as an electron-rich donor toward another equivalent of benzyl ether which may contribute as a secondary photoactive donor after initial conversion has begun, consistent with the precedent that phenolate-type photocatalysts can promote reductive cleavage of benzyl-derived protecting groups.³³ However, this secondary pathway is expected to be most relevant for electron-rich phenolic products that resembles the DBPP-type phenol catalyst used by Xia and coworkers mentioned in the introduction, and is less likely to be general for less electron-rich phenols or aliphatic alcohol products³³ (see SI, page SI–S77 for full computational details).

Based on our experimental observations, spectroscopic data, and computational analysis, we propose a mechanism initiated by the deprotonation of the solvent (DMF or DMSO) by potassium tert-butoxide which generates the corresponding dimsyl or carbamoyl anions (Scheme 3F).⁵⁸ An EDA complex III is formed between starting benzyl protected phenol I and solvent anion II. Upon irradiation of the complex at 390 nm a SET occurs and generates charge-separated complex IV, which contains a solvent radical and a benzyl radical anion species. Dissociation of charge-separated complex IV leads to the formation of solvent radical V and mesolytic cleavage of the benzylic C–O bond, thereby generating the desired phenoxide VI and a stabilized benzyl radical VII. Protonation of the phenoxide during workup gives the desired phenol product.

Conclusion

In conclusion, we have developed a mild and metal-free photochemical protocol for the deprotection of phenols and aliphatic alcohols. This strategy leverages the unique reducing capabilities of solvent anions to achieve efficient C–O bond cleavage. The method exhibits broad functional group tolerance and orthogonality with the N-Boc protecting group as well as alkenes, while also being scalable. Importantly, we have also demonstrated that this method enables selective deprotection of electronically differentiated benzyl groups, thereby providing potential synthetic utility in complex scaffolds containing multiple benzyl-protected alcohols and phenols. Lastly, by eliminating the need for hazardous reagents and expensive transition metals, this approach offers a more sustainable and atom economical manifold for the debenzylation of complex organic molecules.

Conflicts of interest

The authors declare no conflicts of interest.

Data availability

All data supporting the findings of this study are available within the article and its supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6gc02664h.