Scandium‐Catalyzed Highly Selective Deoxygenation of Alcohols by Using Hydrosilanes as Reductants

Abstract

Herein, we present a straightforward and efficient scandium‐based catalytic system that enables highly selective reduction of σ bonds (C─O) in the presence of more reactive π (C═O) or O─H bonds. Employing 5 mol% of Sc(OTf)₃ as the catalyst, a diverse range of aryl and alkyl hydrosilanes proved to be highly effective in reducing secondary and tertiary alcohols to the corresponding alkanes. Furthermore, this protocol was extended to the deoxygenation of ketones and the hydrodehalogenation of alkyl halides.

A new protocol for selective C─O bond cleavage using Sc(OTf)₃ as the catalyst in the presence of hydrosilanes as reductants is reported. Alcohols, ketones, and alkyl halides were efficiently reduced to the corresponding alkanes, with excellent regio‐ and chemoselectivity.

1. Introduction

Given the depletion of oil reserves and the urgency of combating climate change, enhancing energy efficiency and resource utilization in chemical processes is critical for sustainable development. Among these efforts, the targeted conversion of abundant functional groups found in nature is pivotal in improving organic synthesis efficiency.^{[ 1} , ² , ^{3 ]} A crucial aspect lies in the deoxygenation reaction of alcohols, which facilitates the conversion of biomass‐derived alcohols into environmentally friendly fuels by cleaving C─O bonds.^{[ 4} , ^{5 ]} Alcohol deoxygenation methods generally follow two main strategies. The first is a two‐step approach, exemplified by the Barton‐McCombie reaction, which involves converting the hydroxyl group into reactive xanthate intermediates, followed by a reduction step using a stoichiometric amount of a highly toxic tin hydride reagent. While effective, the toxicity and practicality issues of this method limit its broader application.^{[ 6} , ⁷ , ⁸ , ⁹ , ¹⁰ , ^{11 ]} The second approach is direct deoxygenation, which faces significant challenges due to the high bond dissociation energy of C─O bonds. To address these challenges, various transition metals have been investigated, including Ti^{[ 12} , ¹³ , ^{14 ]} Pd,^{[ 15} , ^{16 ]} Ir,^{[ 17} , ¹⁸ , ^{19 ]} Ru,^{[ 20} , ^{21 ]} Mn,^{[ 22 ]} Co,^{[ 23 ]} and Ni,^{[ 24 ]} yet the need for relatively elevated temperatures, excessive bases, additives, and the high costs of noble metal catalysts has significantly limited the widespread application of these methods.

Since the mid‐1900s, Lewis acids have been recognized as effective agents for activating C─O bonds in deoxygenation reactions when used in conjunction with hydride sources.^{[ 25} , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ^{32 ]} Tris(pentafluorophenyl)borane (BCF), stands out as one of the most extensively utilized Lewis acids in such transformations, particularly when paired with hydrosilanes as reductants.^{[ 33 ]} This combination has proven effective in reducing a variety of carbonyl derivatives including ketones, aldehydes, and carboxylic acid derivatives.^{[ 32} , ³⁴ , ³⁵ , ³⁶ , ^{37 ]} Gevorgyan et al.^{[ 32} , ^{35 ]} demonstrated the successful reduction of alcohols to their corresponding alkanes using BCF as a catalyst with excess Et₃SiH (Scheme 1a). Although this system efficiently deoxygenates primary alcohols, it is largely ineffective for secondary and tertiary alcohols unless they contain substituents capable of strongly stabilizing carbocations. Baba et al. explored the deoxygenation of ketones^{[ 38 ]} using InCl₃ with Me₂SiClH and subsequently expanded this method to alcohols,^{[ 29} , ^{30 ]} successfully reducing secondary and tertiary alcohols to alkanes. However, the practical utility of this system was limited by the requirement for moisture‐sensitive chlorohydrosilanes (Scheme 1b).

Scheme 1.

Lewis acid‐catalyzed deoxygenation of alcohols.

In this work, we introduce a simple and efficient protocol for the deoxygenation of secondary and tertiary alcohols to their corresponding hydrocarbons, using Sc(OTf)₃ as a catalyst and hydrosilanes as reductants (Scheme 1c). Additionally, we extended the system's reactivity to effectively deoxygenate ketones and promote the hydrodehalogenation of alkyl halides. Importantly, this method addresses the limitations of previous systems, providing a highly selective strategy for the reduction of σ bonds (C─O) in the presence of π (C═O) or O─H bonds.

2. Results and Discussion

We initiated our investigations on the catalytic activity of Sc(OTf)₃ in the direct deoxygenation of alcohols, using 1‐phenylethanol as a model substrate (Table 1). Reaction of 1‐phenylethanol 1a with 2.0 equiv. of Et₃SiH in the presence of 10 mol% of Sc(OTf)₃ in CD₂Cl₂ at 60 °C for 4 hours resulted in complete conversion (>98%) to the desired ethylbenzene 2a (Table 1, entry 1). Lowering the amount of Et₃SiH to 1.1 equiv. maintained the reaction's efficiency, achieving full conversion with 98% yield of 2a and only a minor formation (2%) of the ether product 3a, (Table 1, entry 2). However, reducing the reaction time to 1 hour under identical conditions significantly lowered the yield of ethylbenzene 2a to 6%, while increasing the formation of the ether product 3a to 65% (Table 1, entry 3). Further, increasing the reaction temperature to 80 °C restored full conversion and produced 98% yield of 2a within 1 hour (Table 1, entry 4). Reducing the reaction time further to 30 minutes under these conditions led to a moderate decrease in yield, with 2a obtained in 75% yield and 3a in 17% (Table 1, entry 5). Notably, lowering the catalyst loading from 10 mol% to 5 mol% at 80 °C for 1 hour maintained excellent conversion and product selectivity (Table 1, entry 6). However, a further reduction to 1 mol% resulted in a marked decrease in selectivity, yielding 2a in 39% and 3a in 52% (Table 1, entry 7). A control experiment without the catalyst confirmed the essential role of Sc(OTf)₃, as no product formation was observed (Table 1, entry 8). Furthermore, running the reaction under air had no adverse effect on its efficacy, with 2a obtained in 96% yield (Table 1, entry 9).

Table 1.

Optimization of reaction conditions for the deoxygenation of 1‐phenylethanol 1a^{[ a ]}

Entry	Sc(OTf)3 (mol%)	Hydrosilane (equiv.)	Temp. (°C)	Time (h)	Conversion %[ b ]	2a %[ b ]	3a %[ b ]
1	Sc(OTf)3 (10)	Et3SiH (2)	60	4	>98	>98	0
2	Sc(OTf)3 (10)	Et3SiH (1.1)	60	4	>98	98	2
3	Sc(OTf)3 (10)	Et3SiH (1.1)	60	1	71	6	65
4	Sc(OTf)3 (10)	Et3SiH (1.1)	80	1	>98	98	2
5	Sc(OTf)3 (10)	Et3SiH (1.1)	80	0.5	94	75	17
6	Sc(OTf)3 (5)	Et3SiH (1.1)	80	1	>98	98	2
7	Sc(OTf)3 (1)	Et3SiH (1.1)	80	1	95	39	52
8	None	Et3SiH (1.1)	80	1	0	0	0
9[ c ]	Sc(OTf)3 (5)	Et3SiH (1.1)	80	1	>98	96	4
10	Sc(OTf)3 (5)	Et2SiH2 (1.1)	80	1	>98	>98	0
11	Sc(OTf)3 (5)	PhSiH3 (1.1)	80	1	>98	81	19
12	Sc(OTf)3 (5)	PhMe2SiH (1.1)	80	1	>98	95	6
13	Sc(OTf)3 (5)	Ph2SiH2 (1.1)	80	1	>98	89	11
14	Sc(OTf)3 (5)	PMHS (1.1)	80	1	>98	93	3
15	Sc(OTf)3 (5)	TMDS (1.1)	80	1	>98	>98	0
16	Sc(OTf)3 (5)	TMDS (1.1)	80	30	>98	>98	0
17	Sc(OTf)3 (5)	TMDS (0.5)	80	1	>98	59	40

^[a] General conditions: 1a (0.1 mmol, 12 µL), Hydrosilane (x mmol, x µL), Sc(OTf)₃ (x mol%), and CD₂Cl₂ (0.5 mL),

^[b] Conversions and yields are determined by ¹H NMR with mesitylene as the internal standard,

^[c] Reaction was performed under air atmosphere.

Employing the optimized reaction conditions (Table 1, entry 6), we explored alternative catalysts, solvents, and hydrosilanes to identify the optimal system for deoxygenation. Screening a range of metal ion catalysts, (Table S1, S5, SI), revealed Sc(OTf)₃ as the most effective, achieving 98% yield of the target product 2a. Among the lanthanide triflate catalysts tested, Ce(OTf)₃ exhibited moderate activity, delivering 2a in 63% yield, alongside 36% of the ether product 3a. In contrast, La(OTf)₃ showed no catalytic activity, while other catalysts such as Sm(OTf)₃, Eu(OTf)₃, and Nd(OTf)₃ predominantly produced the ether product 3a, with only trace amounts of the ethylbenzene 2a and the dehydrated product 4a. Additionally, Yb(OTf)₃ resulted in a mixture comprising 58% of the ether intermediate 3a and 34% of styrene 4a. Interestingly, Fe(OTf)₃ which possesses higher Lewis acidity than Sc(OTf)₃, furnished a complex mixture of three different products (2a, 33%; 3a, 38%; and 4a, 29%), showing reduced selectivity relative to Sc(OTf)₃. Furthermore, when FeCl₃ was tested, almost no catalytic activity was observed, producing only 6% of the ether product 3a. The formation of the dehydrated product 4a aligns with a previous study by Repo et al.,^{[ 39 ]} which demonstrated that alcohols can be converted into their corresponding alkenes using highly oxophilic and Lewis acidic metal triflates. Catalysts such as Hf(OTf)₄ and Fe(OTf)₃ were identified as highly efficient in such reaction, whereas Sc(OTf)₃, with its comparatively lower acidity, was deemed inactive in that context. These results suggest that neither the strongest nor the weakest Lewis acids optimize the system's efficiency; rather, an appropriate balance of Lewis acidity is essential for identifying the ideal catalyst.

With the optimized catalyst in hand, we evaluated various solvents, finding that solvent selection was crucial for the reaction's success. Dichloromethane was uniquely effective, while solvents such as toluene, acetonitrile, and chloroform primarily formed the ether product 3a, and tetrahydrofuran was entirely inactive (Table S2, S6, Supporting Information).

The effect of hydrosilane was also investigated. Interestingly, different aryl and alkylsilanes such as Et₃SiH, Et₂SiH₂, tetramethyldisiloxane (TMDS), Ph₂SiH₂, Ph₃SiH, and polymethylhydrosiloxane (PMHS), were all exhibited high efficiency, providing full conversions with excellent yields (>81%) (Table 1, entries 10–15). Moreover, the silylation of the alcohols was never observed, indicating a preferential activation of the C─O bond compared to the O─H bond in the C─O─H moiety.

We selected TMDS for further studies as it is a readily available and cheap hydrosilane.^{[ 40 ]} Notably, the reaction time could be reduced to just 30 minutes without compromising efficiency (Table 1, entry 16). However, using only 0.5 equiv. of TMDS led to a decreased yield of the desired product 2a (59%), accompanied by the formation of 40% of the ether product 3a (Table 1, entry 17). Under the optimized conditions (TMDS 1.0 equiv., Sc(OTf)₃ 5 mol%, 80 °C, in CD₂Cl₂), the substrate scope was investigated (Scheme 2). Secondary and tertiary benzylic alcohols (1a–1d) underwent smooth deoxygenation, yielding their corresponding alkanes in excellent yields (>94%). Substrate bearing electron‐donating group 1‐(4‐methoxyphenyl)ethan‐1‐ol 1e, showed enhanced reactivity, delivering 1‐ethyl‐4‐methoxybenzene 2e quantitatively within 10 minutes. In contrast, electron‐withdrawing groups slowed the reaction, as observed with 1‐(4‐(trifluoromethyl)phenyl)ethanol 1f, that required 5 hours to achieve 59% yield of 2f. Aliphatic alcohols 1g and 1i were efficiently converted to their respective products 2g and 2i in yields exceeding 98%. However, 2‐methylcyclohexanol 1h required longer reaction time (22 hours) to achieve 94% of 2h. Notably, primary alcohol 1j, remained unreactive even after extending the reaction time to 16 hours. Furthermore, we showcased the synthetic utility of this approach by scaling the reaction to a 1 mmol scale. Diphenylmethane 2b was isolated in 96% yield with Et₃SiH and 89% yield when using TMDS (S7, SI).

Scheme 2.

Substrate scope for the deoxygenation of alcohols. Reaction conditions : 1 (0.1 mmol), TMDS (0.1 mmol, 1.0 equiv.), Sc(OTf)₃ (5 mol%), and CD₂Cl₂ (0.5 mL), Conversions and yields are determined by ¹H NMR with mesitylene as the internal standard.

2.1. Catalytic Deoxygenation of Ketones

Catalytic reduction of ketones using hydrosilanes as hydride donors can yield diverse products. The predominant pathway, hydrosilylation reduction, leads to the formation of silyl ethers,^{[ 41} , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ^{46 ]} whereas a less common deoxygenation process generates the corresponding alkane.^{[ 23} , ⁴⁷ , ^{48 ]} The Lewis acid‐catalyzed deoxygenation of alcohols approach has often been adapted for the deoxygenation of ketones. For instance, both the BCF/hydrosilane^{[ 34} , ⁴⁹ , ^{50 ]} and InCl₃/hydrochlorosilane^{[ 38 ]} systems have been employed in the defunctionalization of ketones to their corresponding methylene derivatives. Consequently, we extended this methodology to the deoxygenation of ketones (Scheme 3). Having optimized the reaction conditions (Table S3, S8, SI), Ketones (1k–1z) were treated with 5 mol% of Sc(OTf)₃ and 2.0 equiv. of TMDS in CD₂Cl₂ at 80 °C. Under these conditions, acetophenone 1k, benzophenone 1l, and butyrophenone 1m were quantitatively converted to their corresponding alkanes (2k–2m). Halo‐substituted aryl ketones (1o–1r) were successfully reduced to methylene compounds (2o–2r) without affecting the aryl halide functionalities. Similarly, the alkyl halide group in 1s remained intact, affording (2‐bromoethyl)benzene 2s in 98% yield. Interestingly, the electronic effects observed with ketones differed from those with alcohols. The electron‐rich para‐methoxyacetophenone 1n required 20 hours to yield the desired product 2n in 96% yield, whereas the electron‐deficient para‐(trifluoromethyl)acetophenone 1t was deoxygenated in just 1 hour to afford 2t in 87% yield. This inverse trend suggests distinct mechanistic pathways or rate‐determining steps between ketone and alcohol deoxygenation. In the case of ketones bearing potentially coordinating groups (1u–1x), the formation of the silyl ethers was dominated. Para‐nitroacetophenone 1u was fully converted instantly at room temperature to the corresponding silyl ether 2u' and yielded 39% of the alkane product 2u after heating at 80 °C for 22 hours. In the same manner, the para‐cyano acetophenone 1v and 4‐acetyl pyridine 1w were also quantitatively converted to the corresponding silyl ethers (1v' and 1w') and no formation of the alkane products was observed even by continuing to heat the reaction mixture at 80 °C for 22 hours. 2‐acetyl furan 1x yielded a mixture of the silyl ether 2x' 79% and 21% of the deoxygenated product 2x in 4 hours. Additionally, 2,6‐dimethylcyclohexan‐1‐one 1y was smoothly converted into 1,3‐dimethylcyclohexane 2y in excellent yield >98%. In contrast, the aliphatic ketone 3‐pentenone (1z) underwent a reductive dehydration process, yielding a mixture of alkene isomers (2z, trans 60%, and 2z' cis 27%) rather than the expected alkane ((Scheme 4)).

Scheme 3.

Substrate scope for the deoxygenation of ketones. Reaction conditions: 1 (0.1 mmol), TMDS (0.2 mmol, 2.0 equiv.), Sc(OTf)₃ 5 mol%, and CD₂Cl₂ (0.5 mL), Conversions and yields are determined by ¹H NMR with mesitylene as the internal standard.

Scheme 4.

Substrate scope for the hydrodehalogenation of alkyl halides. Reaction conditions: 1 (0.1 mmol), TMDS (0.1 mmol, 1.0 equiv.), Sc(OTf)₃ (5 mol%), and CD₂Cl₂ (0.5 mL), Conversions and yields are determined by ¹H NMR with mesitylene as the internal standard. [a] 1.0 equiv. of Et₃SiH was used instead of TMDS.

2.2. Hydrodehalogenation of Alkyl Halides

Developing a catalytic system that is highly efficient for activating C─O bonds has prompted us to assess its reactivity with C–X derivatives, where X represents a halogen. Hydrodehalogenation, which replaces a halogen atom with a hydrogen, is a highly effective strategy for neutralizing hazardous organohalides. Methods employing boron Lewis acids in combination with hydrosilanes have been utilized for cleaving carbon–halogen bonds.^{[ 51} , ⁵² , ⁵³ , ^{54 ]} Thus the potential of our system for hydrodehalogenation of alkyl halides was evaluated. Using 5 mol% of Sc(OTf)₃, and 1.0 equiv. of TMDS, secondary and tertiary organo(chlorides, fluorides, and bromides) (1aa–1ad) displayed excellent reactivity, achieving complete conversion with yields exceeding 98% (Scheme 4). Additionally, 1‐phenylbromoethane 1ae and 1‐bromoadamantane 1af were hydrodehalogenated in high yields (98% and 95%, respectively) under longer reaction times and elevated temperatures. However, cinnamyl bromide 1ag was completely consumed but did not yield the expected product, instead forming unidentified byproducts. Meanwhile, benzyl chloride 1ah and aliphatic 1‐chlorohexane 1ai showed no conversion to the corresponding hydrocarbons, even under harsher reaction conditions.

2.3. Chemoselectivity

Having in hand a catalytic system capable of reducing various chemical bonds, including C═O, C─O, and C─X, we wished to explore its potential for other selective reductions, specifically investigating whether it could preferentially cleave σ bonds (C─O) in the presence of more reactive (C═O) double bonds.

Our initial investigations focused on assessing the ability of the Sc(OTf)₃/Et₃SiH system to selectively reduce alcohols in the presence of ketones. Under optimized conditions and using only one equivalent of Et₃SiH, mixing 1‐phenylethanol 1a with 1‐(2‐bromophenyl) ethan‐1‐one 1r resulted in the full conversion of alcohol 1a into the corresponding alkane 2a in an excellent yield of 96 %, while ketone 1r was quantitatively recovered. Similarly, 1‐(4‐methoxyphenyl)ethan‐1‐ol 1e and diphenylmethanol 1b were selectively reduced to their respective alkanes, 2e (98 %) and 2b (98 %), in the presence of acetophenone 1k, which remained fully untouched. (Scheme 5).

Scheme 5.

Chemoselective reduction of alcohols in the presence of ketones catalyzed by Sc(OTf)_3.

We then conducted a comparative study against the established BCF/Et₃SiH and InCl₃/Ph₂ClSiH protocols (Table 2). Reactions involving 1‐phenylethanol 1a in the presence of 1‐(2‐bromophenyl)ethan‐1‐one 1r and 10 mol% of BCF, using either 1 or 2 equiv. of Et₃SiH, yielded a complex mixture of silyl ether products 2a′ and 2r′ along with ether products 3a and 3a′. However, only 5% yield of the desired alkane 2a was obtained when 2 equiv. of Et₃SiH were used. A similar lack of selectivity was observed with the InCl₃/ Ph₂ClSiH system. With 1 equiv. of Ph₂ClSiH, 1‐phenylethanol 1a was converted to the silyl ether product 2a″ in 75% yield, with only 10% of the desired alkane 2a, while 1‐(2‐bromophenyl)ethan‐1‐one 1r remained unaffected. Increasing to 2 equiv. of Ph₂ClSiH led to complete conversion of 1a to ethylbenzene (2a, 69%) accompanied by unidentified byproducts, while ketone 1r was fully converted into its silyl ether 2r″ (89%), with a minor formation of alkane 2r (11%).

Table 2.

Chemoselectivity studies (Alcohols versus Ketones), with Sc(OTf)₃/Et₃SiH, BCF/Et₃SiH, and InCl₃/Ph₂ClSiH systems.

Sc(OTf)3 (5 mol%) Et3SiH (1.0 equiv.)[ a ]	‐	>98%	2a (96%)	‐	‐	‐
BCF (10 mol%) Et3SiH (1.0 equiv.)	‐	56%	‐	2a′ (64%)	2r′ (36%)	3a (36%) 3a′ (8%)
BCF (10 mol%) Et3SiH (2.0 equiv.)	‐	20%	2a (5%)	2a′ (25%)	2r′ (50%)	3a (69%) 3a′ (30%)
InCl3 (10 mol%) Ph2ClSiH (1.0 equiv.)	‐	>98%	2a (10%)	2a″ (75%)	‐	3a (15%)
InCl3 (10 mol%) Ph2ClSiH (2.0 equiv.)	‐	‐	2a (69%) 2r (11%)	‐	2r″ (89%)	‐

^[a] The reaction mixture was heated for 1 hour at 80 °C.

These findings highlight the high chemoselectivity of our system, which prioritizes σ bonds C─O bond cleavage over π(C═O) or O─H bonds, with the use of only one equivalent of hydrosilane. This is in contrast to existing catalytic systems such as BCF/hydrosilane or InCl₃/hydrochlorosilane, who necessitate the formation of silyl ether intermediates, with BCF generating H₂ ^{[ 32} , ^{55 ]} and InCl₃ producing HCl^{[ 30 ]} as by‐products, thereby requiring an excess of hydrosilane to achieve comparable selectivities. These efficiency and selectivity make it a valuable tool for targeted reductions in synthetic chemistry.

Leveraging the inactivity of our system toward primary alcohols, we sought to explore the possibility of selectively reducing secondary alcohols in the presence of primary alcohols. Such regioselective deoxygenation represents a critical transformation in the valorization of alcohol‐rich feedstocks for high‐value chemical synthesis.^{[ 56} , ⁵⁷ , ^{58 ]} Morandi and co‐workers reported a catalytic strategy for the regioselective deoxygenation of terminal 1,2‐diols at the primary position, enabling the synthesis of 2‐alkanols, using BCF as a catalyst and a combination of Ph₂SiH₂ and Et₃SiH as reductants.^{[ 59 ]}

Gratifyingly, reacting 1 equiv. of either TMDS or Et₃SiH with 1‐phenyl‐1,2‐ethanediol (1aj) resulted in highly regioselective monodeoxygenation at the secondary alcohol position, and affording the desired product 2aj in an excellent 98% yield after heating the reaction mixture at 100 °C for 24 hours (Scheme 6a). Additionally, ketoprofen 1ak underwent selective and efficient reduction using 2 equiv. of TMDS under mild conditions (80 °C for 30 minutes), leaving the carboxylic acid functionality unaltered (Scheme 6b). This selective methodology highlights its broad utility and potential for application in complex molecular transformations.

Scheme 6.

Regioselective a) and chemoselective b) deoxygenation catalyzed by Sc(OTf)_3.

2.4. Mechanism of the Reaction

To gain insights into the mechanism of the deoxygenation of alcohols, a series of control experiments were conducted, as illustrated in Scheme 7. In the absence of an hydrosilane, the reaction of 1‐phenylethanol 1a with 5 mol% of Sc(OTf)₃ resulted in complete conversion to the ether product 3a, accompanied by the generation of water. Upon addition of 1.1 equiv. of Et₃SiH to the same reaction mixture, 3a was fully converted into the alkylated product 2a. Such a dehydrative etherification of alcohols has been reported in the presence of various Lewis acids.^{[ 60} , ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ , ^{67 ]} Interestingly, upon addition of 1 equiv. of water under optimal reaction conditions, 85% of ethylbenzene 2a was obtained after 22 hours at 80 °C, indicating the scandium stability in aqueous environments, although the reaction was slowed. This behavior contrasts with that of BCF, which tends to form hydroxide adducts in the presence of water, thereby destroying its Lewis acidity and halting catalytic activity.^{[ 68 ]} Additionally, when we started the reaction with the ether intermediate 3a in the absence of water, it was fully converted to the desired product 2a within just 10 minutes at 80 °C. To investigate the hypothesized involvement of a silylated alcohol intermediate C (Scheme 8), the reaction was initiated with the silyl ether 2a′ which was fully converted to the desired product 2a within 10 minutes at room temperature.

Scheme 7.

Control experiments for alcohol deoxygenation. a) sequential reaction, b) ether deoxygenation and cleavage, c) deoxygenation in the presence of water, d) deoxygenation of the silylated alcohol.

Scheme 8.

Proposed mechanism for Sc(OTf)₃ catalyzed deoxygenation of alcohol.

Based on these experimental findings, we propose a mechanism for the scandium‐catalyzed deoxygenation of alcohols as illustrated in Scheme 8. The observed lack of reduction in primary alcohols and the enhanced reaction efficiency with electron‐rich substrates suggest that the reaction proceeds through an intermediate that behaves like a carbocation. The proposed mechanism initiates with either 1‐phenylethanol 1a or the ether intermediate 3a, which, in the presence of Sc(OTf)₃, could form an oxonium complex (A or B). These intermediates subsequently dissociate to generate the corresponding carbocation. In pathway 1, the carbocation undergoes reduction by hydrosilane, yielding the hydrocarbon product 2a, silanol, and regenerating the scandium catalyst. Similarly, in pathway 2, the addition of Et₃SiH induces hydride transfer to the carbocation, producing the alkane product 2a alongside an oxonium complex C. The oxonium species C can dissociate again into a carbocation, which reacts with another molecule of hydrosilane to afford the desired product 2a and hexaethyldisiloxane, then latter was isolated and confirmed via GC/MS analysis, while restoring the scandium catalyst to complete the catalytic cycle.

In the case of hydrodehalogenation, we propose that the mechanism proceeds via a pathway analogous to that of alcohol deoxygenation. This involves the generation of a carbocation intermediate through halide abstraction, followed by hydride transfer from the hydrosilane to the carbocation, ultimately yielding the desired product along with the corresponding halosilane byproduct (S22‐S26, Supporting Information). In contrast, the deoxygenation of ketones is well established to proceed via the formation of a silyl ether intermediate, which undergoes further reduction in the presence of an additional equivalent of hydrosilane.^{[ 42} , ^{50 ]}

3. Conclusion

We developed a highly efficient and simple catalytic system that exhibit excellent chemoselectivity, preferentially reducing σ bonds (C─O) over more reactive π (C═O) or O─H bonds, utilizing only one equivalent of hydrosilane. This approach enabled us to reduce alcohols in the presence of ketones, to monodeoxygenate secondary alcohols in a regioselective manner and in the presence of primary alcohol, and to selectively reduce a ketone in the presence of acid‐functional group. Additionally, secondary and tertiary alcohols were effectively transformed into their corresponding hydrocarbons, with primary alcohols remaining inert, unlike BCF system, which in general selectively deoxygenates primary alcohols. Furthermore, all hydrosilanes tested in this study performed effectively in the deoxygenation process, setting this method apart from other catalytic systems that often require specific types of hydrosilanes. Beyond alcohol deoxygenation, the system also showed high efficiency in the deoxygenation of ketones and the hydrodehalogenation of alkyl halides, highlighting its versatility and broad applicability.

4. Experimental Section

General Procedure for the deoxygenation of alcohols

A 2.5 mL J. Young NMR tube in a glovebox was charged with alcohol (0.1 mmol), TMDS (1.0 mmol, 1.0 equiv.), Sc(OTf)₃ (5 mol%), CD₂Cl₂ (0.5 mL), and mesitylene (5 µL) as internal standard. The tube was sealed and brought out of the glovebox, and the mixture was then stirred at 80 °C for the required time. Yields were determined by ¹H NMR integration versus mesitylene.

General procedure for the deoxygenation of ketones

In a 2.5 mL J. Young NMR tube in a glovebox, ketone 1 (0.1 mmol, 1.0 equiv.) and TMDS (0.2 mmol, 2.0 equiv.), and mesitylene (5 µL) were added to a solution of Sc(OTf)₃ (5 mol%) in deuterated dichloromethane (0.5 mL). The tube was sealed and brought out of the glove box, and the solution was then heated at 80 °C for the required time. The reaction progress was monitored by ¹H NMR spectroscopy. Yields were determined by ¹H NMR integration versus mesitylene as an internal standard.

Supporting Information

The authors have cited additional references within the Supporting Information.^{[ 62} , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² , ⁸³ , ⁸⁴ , ⁸⁵ , ⁸⁶ , ^{87 ]}

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.