Visible light photoredox-catalyzed deoxygenation of alcohols

Article abstract of DOI:10.3762/bjoc.10.223

Carbon-oxygen single bonds are ubiquitous in natural products whereas efficient methods for their reductive defunctionalization are rare. In this work an environmentally benign protocol for the activation of carbon-oxygen single bonds of alcohols towards a reductive bond cleavage under visible light photocatalysis was developed. Alcohols were activated as 3,5-bis(trifluoromethyl)-substituted benzoates and irradiation with blue light in the presence of [Ir(ppy)₂( dtb-bpy)](PF₆ ) as visible light photocatalyst and Hünig's base as sacrificial electron donor in an acetonitrile/water mixture generally gave good to excellent yields of the desired defunctionalized compounds. Functional group tolerance is high but the protocol developed is limited to benzylic, α-carbonyl, and α-cyanoalcohols; with other alcohols a slow partial C-F bond reduction in the 3,5-bis(trifluoromethyl)benzoate moiety occurs.

Full text of DOI:10.3762/bjoc.10.223

Visible light photoredox-catalyzed deoxygenation of alcohols
Daniel Rackl, Viktor Kais, Peter Kreitmeier and Oliver Reiser*
Full Research Paper
Open Access
Address:
Beilstein J. Org. Chem. 2014, 10, 2157–2165.
Institute of Organic Chemistry, University of Regensburg, 93053
Regensburg, Germany
doi:10.3762/bjoc.10.223
Received: 04 June 2014
Email:
Accepted: 27 August 2014
Published: 10 September 2014
Oliver Reiser* - Oliver.Reiser@chemie.uni-regensburg.de
* Corresponding author
This article is part of the Thematic Series "Organic synthesis using
photoredox catalysis".
Keywords:
C–O bond activation; deoxygenation; photochemistry; photoredox
catalysis; visible light
Guest Editor: A. G. Griesbeck
© 2014 Rackl et al; licensee Beilstein-Institut.
License and terms: see end of document.
Abstract
Carbon–oxygen single bonds are ubiquitous in natural products whereas efficient methods for their reductive defunctionalization
are rare. In this work an environmentally benign protocol for the activation of carbon–oxygen single bonds of alcohols towards a
reductive bond cleavage under visible light photocatalysis was developed. Alcohols were activated as 3,5-bis(trifluoromethyl)-
substituted benzoates and irradiation with blue light in the presence of [Ir(ppy)2(dtb-bpy)](PF6) as visible light photocatalyst and
Hünig’s base as sacrificial electron donor in an acetonitrile/water mixture generally gave good to excellent yields of the desired
defunctionalized compounds. Functional group tolerance is high but the protocol developed is limited to benzylic, α-carbonyl, and
α-cyanoalcohols; with other alcohols a slow partial C–F bond reduction in the 3,5-bis(trifluoromethyl)benzoate moiety occurs.
Introduction
The dwindling supply of hydrocarbons from fossil resources A classical radical deoxygenation reaction using over-stoichio-
calls for the usage of renewable resources for the synthesis of metric amounts of highly noxious chemicals [2] is the
fine chemicals in the future [1]. This strategy suffers from Barton–McCombie reaction, although nowadays several impro-
the relative high degree of functionalization of feedstock ved protocols are available [3]. Radical deoxygenations can also
materials, which is often not desired in fine chemicals and be carried out electrochemically [4] or photochemically [5-9].
further leads to compatibility issues in chemical transforma- In all these cases an activation of the hydroxy group is neces-
tions. Carbon–oxygen single bonds are common elements in sary, either via conversion to the corresponding halide or forma-
natural materials and their reduction to non-functionalized tion of an ester derivative, which consequently generates over-
carbon–hydrogen bonds decreases complexity and increases stoichiometric amounts of byproducts. Related to this work,
compatibility of those materials in further chemical manipula- Stephenson et al. elegantly succeeded in the direct deoxygena-
tions in accordance with established oil-based protocols devel- tion of alcohols by their in situ conversion to iodides using
oped in the chemical industry during the last century.
triphenylphospine and iodine followed by visible light-medi-
2157

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
ated reduction with amines as stoichiometric sacrificial electron Results and Discussion
donor and fac-Ir(ppy)3 (ppy = 2-phenylpyridine) as photoredox Following the lead of photochemical deoxygenations under UV
catalyst (Scheme 1) [10]. This protocol is applicable to a broad light irradiation (vide supra) we envisioned carboxylic ester
range of alcohols. It also greatly advanced the quest to develop derivatives as substrates for initial test reactions. Benzoate
sustainable methods for the deoxygenation of alcohols. Yet, esters were chosen due to the potentially very facile recovery of
several redox steps, i.e., the stoichiometric transformation of the benzoic acids used for activation. Through variation of the
triphenylphosphine to triphenylphospine oxide and iodine to substitution pattern of benzoates we intended to shift the elec-
iodide, are required, which appears to be problematic for estab- trochemical reduction potentials of the substances into a region
lishing a sustainable protocol that allows the recycling and that could be accessed by common visible light photocatalysts.
reuse of the reagents involved. Herein we report a redox The substituents should be as inert as possible in order not to
economic deoxygenation method of alcohols in which forma- interfere with the photochemical reaction itself. Therefore
tion of radicals is achieved under visible light photocatalysis different trifluoromethyl-substituted benzoates were prepared
and the auxiliary activation group can be readily recovered and and subjected to cyclovoltametric measurements. 3,5-Bis(triflu-
reused.
oromethyl)-substituted benzoate 3 showed the most promising
reduction potential of the compounds investigated (Scheme 2)
This method, ultimately requiring only energy and a tertiary and was therefore expected to be most susceptible for an initial
amine as stoichiometric reductant, gives high yields after short photoredox electron transfer that we considered in analogy to
illumination times under mild conditions for the deoxygenation the Barton–McCombie technology to be crucial to trigger
of benzylic alcohols, α-hydroxycarbonyl, and α-cyanohydrin deoxygenations.
compounds. Moreover, the selective catalytic monoacylation of
diols is possible, thus allowing efficient monodeoxygenations as Initial deoxygenation experiments were carried out with either
exemplified in the conversion of (+)-diethyl tartrate to unnat- Ru(bpy)3Cl2·6H2O [bpy = 2,2'-bipyridine] or [Ir(ppy)2(dtb-
ural (+)-diethyl malate.
bpy)](PF6) [ppy = 2-phenylpyridine; dtb-bpy = 4,4′-di-tert-
Scheme 1: Strategies for the visible light-catalysed deoxygenation of alcohols (reagents needed in (over-)stoichiometric quantities are depicted in
blue).
2158

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
Scheme 2: Reduction potentials of investigated derivatives 1–3 in DMF.
butyl-2,2′-bipyridine] as photocatalysts, Hantzsch ester (diethyl
1,4-dihydro-2,6-dimethyl-3,5-pyridinedicarboxylate) as
hydrogen donor, and iPr2NEt as sacrificial electron donor in
DMF (Scheme 3). Light generated from a high power LED was
channeled into the reaction solution in a Schlenk tube through a
glass rod from above while magnetic stirring and heating was
applied from below. This reaction setup is superior compared to
conventional ones that are carried out in a vessel with external
irradiation: a high light intensity can be achieved for irradiation
and reaction temperatures can be manipulated very conve-
niently by heating in a conventional oil bath while carrying out
the photoreaction. Also light pollution is kept to a minimum,
which renders the apparatus an optimal device for the fast setup
of test reactions.
In agreement with the reduction potentials, 3,5-bis(trifluo-
romethyl)-substituted benzoate 3 gave the best results in the
deoxygenation reaction while esters 1 and 2 led to incomplete
reactions after 16 h of irradiation (Table 1). The performance of
Scheme 3: Initial reaction conditions for deoxygenation candidates
1–3.
[Ir(ppy)2(dtb-bpy)](PF6) was in all cases superior as compared the conversion of 3 (Table 2). Gratifyingly, toxic DMF could be
to Ru(bpy)3Cl2·6H2O, possibly due to its increased reduction replaced with more benign acetonitrile without appreciable
potential (E0 = −1.51 V vs E0 = −1.31 V) [11].
decreasing the yield (Table 2, entry 2). The reaction also
proceeded in less polar solvents (Table 2, entries 3 and 4), albeit
Having identified a promising activation group for deoxygena- yields were significantly lower. When the reaction was
tion in combination with an iridium-based photocatalyst, performed at ambient temperature 4 was only formed in 41%
different solvents and reaction temperatures were examined for yield after 16 h of irradiation (Table 2, entry 5). Control experi-
2159

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
omitted significantly lower yields were obtained. Leaving out
Hantzsch ester (Table 2, entry 8) apparently does not impede
the deoxygenation while carrying out the reaction without
Hünig’s base lowers the yield (Table 2, entry 9), nevertheless, 4
was still formed to a significant extent. These results suggest
that Hantzsch ester is not necessary as the hydrogen source but
likewise that reductive quenching of the photocatalyst is not
exclusively accomplished by Hünig’s base.
Table 1: Comparison of different esters and photocatalysts in deoxy-
genation reaction.
Photocatalyst
Compound yield [%]a
1
2
3
We assumed that the mechanism of the deoxygenation reaction
involves an electron uptake of the ester moiety from the reduc-
tively quenched photocatalyst followed by mesolysis and subse-
quent hydrogen abstraction (Scheme 4). Quantum mechanical
calculations (B3LYP/6-31G*) for benzhydryl 3,5-bis(trifluo-
romethyl)benzoate (3) revealed that the electron density of the
presumed transient radical anion is mainly located at the phenyl
moiety of the benzoate – and not in the desired anti-bonding
σ*(C–O) (Scheme 5). Protonation of the radical anion would
lead to a neutral radical species, which in the calculations
reflects in a shift of electron density towards the C–O bond to
be cleaved.
Ru(bpy)3Cl2·6H2O
5
8
10
85
[Ir(ppy)2(dtb-bpy)](PF6)
18
20
aAll yields determined by GC–FID with dodecane as internal standard.
Table 2: Solvent/temperature dependence and control experiments of
deoxygenation reaction with 3,5-bis(trifluoromethyl)benzoate 3a.
Entry
Solvent, modification
Yield (%)b
1
2
3
4
5
6
7
8
9
DMF
85
80
20
22
41
7
MeCN
DCM
THF
Consequently, a mixture of acetonitrile/water (14:1) was
explored as reaction medium, resulting to our delight in dramat-
ically reduced reaction times to achieve full conversion.
Starting from bis(trifluoromethyl)benzoates, quantitative forma-
tion of the deoxygenated products could be already observed
after 20 minutes of irradiation.
MeCN, rt
MeCN, w/o photocatalyst
MeCN, w/o light source
MeCN, w/o Hantzsch ester
MeCN, w/o iPr2NEt
15
91
53
aConditions see Table 1. bAll yields determined by GC–FID with dode-
cane as internal standard.
A simple iridium-catalyzed hydrogenation mechanism as an
alternative for a photochemical pathway of the reaction could
ments suggest that the deoxygenation reaction of 3,5-bis(trifluo- be ruled out; in the presence of 5 atm H2 without irradiation
romethyl)benzoate 3 is indeed a photochemically mediated under otherwise unchanged reaction conditions no deoxygena-
process (Table 2, entries 6 and 7): when either the photocata- tion of the benzoates could be observed even after prolonged
lyst (Table 2, entry 6) or the light source (Table 2, entry 7) was reaction times.
Scheme 4: Proposed reaction mechanism with and without additional water.
2160

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
groups with different electronic properties, i.e., an electron-
donating p-methoxy substituent (Table 3, entry 3), electron-
withdrawing p-nitro substituent (Table 3, entry 4), ester group
containing systems (Table 3, entry 5), chlorinated derivatives
(Table 3, entry 6) and electron-deficient heteroaromatic systems
(Table 3, entry 7) were well tolerated, giving the corresponding
deoxygenated products in analytical pure form in high yields
after filtration through a short plug of silica gel. Noteworthy, no
reduction of reducible groups such as nitro (Table 3, entry 4) or
chloro (Table 3, entry 5) was observed. Moving to monobenzyl
alcohols, e.g., replacement of one aromatic group with an alkyl
chain resulted in prolonged reaction times but nevertheless
acceptable yields of the deoxygenated products (Table 3, entries
8 and 9). With α-carbonyl-substituted benzylic alcohol deriva-
tives irradiation times could be reduced again and defunctional-
ized materials were isolated in moderate to good yields
(Table 3, entries 10 and 11). Bis(trifluoromethyl)benzoic acid 5
could easily be recovered (>90%) in an acid–base extraction
Scheme 5: Calculated spin densities of the radical anion and its proto-
nated species.
With the newly optimized reaction conditions at hand different step.
benzylic alcohol derivatives were investigated (Table 3).
Uniformly very good isolated yields after short reaction times Also bis(trifluoromethyl)benzoates of non-benzylic α-cyanohy-
were achieved in case of dibenzylic alcohol derivatives. Steric drin 6a and α-hydroxycarbonyl compounds 6b–e turned out to
bulk (Table 3, entry 2), as well as a broad range of functional be amenable for the desoxygenation process (Table 4).
Table 3: Preparative deoxygenation reactions.
Entry
1
Substrate
Product
Yield (%)a
95
3a
4a
2
3
86
87
3b
3c
4b
4c
4
91
3d
4d
2161

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
Table 3: Preparative deoxygenation reactions. (continued)
5
93
3e
4e
4f
6
92
86
3f
7
3g
4g
4h
8
66b,c
79c
3h
9
3i
4i
4j
10
83
67
3j
11
3k
4k
aIsolated yields of reactions conducted at a 0.2–1.0 mmol scale. bDetermined by GC with dodecane as internal standard. c16 h reaction time.
Table 4: Preparative deoxygenation reactions of non-benzylic benzoates. a
Entry
Substrate
Product
Yield (%)
86
1
6a
6b
7a
7b
2
3
79
14b
6c
7c
2162

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
Table 4: Preparative deoxygenation reactions of non-benzylic benzoates. a (continued)
4
69c,d
99
6d
7d
7e
5
6e
aConditions see Table 3. bParent compound was prone to hydrolysis under reaction conditions. c16 h reaction time. d1H NMR yield.
Especially interesting from a preparative point of view, the presumably due to the energetically unfavorable primary radical
monodeoxygenation of diethyl tartrate to maleate could be intermediate that would result via the desired cleavage of the
achieved from 6e in excellent yields (Table 4, entry 5). Notably, C–O bond. Instead, carbon–fluorine cleavage, leading to a
after screening various Lewis acids (Supporting Information benzylic radical, is the preferred pathway.
File 1) it was found that 6e could be selectively prepared by
copper(II) chloride-catalyzed benzoylation of diol 8 with anhy- To unambiguously rule out that the methyl group in 12 origi-
dride 9, which in turn can be generated from its acid 5 by treat- nates from a substitution process with acetonitrile as the methyl
ment with acetic anhydride (Scheme 6).
source, the reaction was carried out in deuterated acetonitrile.
No deuterium incorporation was observed which proved that
Since acetic anhydride is technically being produced by thermal acetonitrile is not responsible for the presence of the methyl
dehydration of acetic acid [12], the overall sequence to the group in 12. Performing the reaction in the presence of TEMPO
benzoylated starting material 6e does not require any type of ac- (2,2,6,6-tetramethylpiperidine-1-oxyl) gave, beside reduction
tivation reagents such as thionyl chloride or DCC that are often product 12, adduct 13 which suggests that the methyl group
used for ester formation, but ultimately only requires energy in originates from a sequential reduction of the C–F bonds through
form of heat. After the photochemical deoxygenation, 3,5- a radical pathway. In addition a test for fluoride with
bis(trifluoromethyl)benzoic acid (5) is formed, which can be [Fe(SCN)(H2O)5]2+ in an evaporated aliquot of the irradiated
easily recovered in high yield and from which anhydride 9 can reaction mixture was positive. Increasing the amount of Hünig’s
be regenerated as described above.
base acting as sacrificial electron donor in the initial reduction
step of 10 led to full conversion of the starting material and
Attempts to deoxygenate simple alkyl-substituted alcohols (pri- gave 12 as the only reaction product in 77% isolated yield.
mary, secondary and tertiary) were not successful; for example
under typical conditions 3,5-bis(trifluoromethyl)benzoates such For larger scale applications it would be desirable to install the
as 10 gave 12 where one trifluoromethyl group was completely activating benzoate group in situ rather than in a foregoing reac-
reduced to a methyl group in 52% yield (Scheme 7). Appar- tion step. Also, considering its high price the employment of
ently, electron transfer to the benzoate group is still possible only small amounts of iridium-based catalyst and lower priced
however, the subsequent C–O bond cleavage does not occur, triethylamine instead of costly diisopropylethylamine would be
Scheme 6: Synthesis of monobenzoate 6e.
2163

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
Scheme 7: Reduction of benzoate moiety in case of non-benzylic alcohols.
desirable. Taking 14 as model compound it could be shown that deoxygenation protocol described here could become also
the overall deoxygenation process can be optimized and simpli- attractive for large-scale applications.
fied in this regard by the in situ formation of the 3,5-bis(trifluo-
romethyl)benzoate 3a which in turn could then be converted to Experimental
deoxygenated compound 4a in 91% yield (Scheme 8) in a flow General procedure for photochemical deoxygenations: A
process using a microreactor (see Supporting Information Schlenk tube was charged with photocatalyst (20 µmol,
File 1).
2.0 mol %), benzoate (1.00 mmol, 1.00 equiv), sealed with a
screw cap and subsequently evacuated and backfilled with N2
(3×). Solvent (20 mL), Hünig’s base (0.35 mL, 2.0 mmol,
Conclusion
In summary a protocol for the deoxygenation of benzylic alco- 2.0 equiv), and degassed water (1.8 mL, 100 mmol, 100 equiv)
hols, α-hydroxycarbonyl and α-cyanohydrin compounds under were added and the reaction mixture was magnetically stirred
visible light photocatalysis was developed using 3,5-bis(trifluo- until a homogeneous solution was obtained. The reaction mix-
romethyl)benzoic anhydride for alcohol activation. Since 3,5- ture was degassed by freeze-pump-thaw (5×) and the screw cap
bis(trifluoromethyl)benzoic acid can be recycled and reacti- was replaced with a Teflon-sealed inlet for a glass rod, through
vated under redox neutral conditions, and moreover, the in situ which irradiation with a 455 nm high power LED took place
activation of alcohols with this auxiliary is possible we envi- from above while the reaction was magnetically stirred and
sion that an overall continuous process could be developed for heated in an aluminum block from below. After the reaction the
the deoxygenation of alcohols by this protocol. That ultimately mixture was diluted with 100 mL Et2O, washed with 50 mL
only requires heat, triethylamine as a sacrificial electron donor 10% Na2CO3, 50 mL H2O, 50 mL brine, and dried over
and visible light, forming water as the only byproduct. There- Na2SO4. After evaporation the crude mixture was purified by
fore, we believe that despite the relatively expensive activation filtration through a short plug of flash silica gel with a mixture
of alcohols as 3,5-bis(trifluoromethyl)benzoic acid esters, the of petrol ether and ethyl acetate.
Scheme 8: Optimized conditions for larger scale applications.
2164

Beilstein J. Org. Chem. 2014, 10, 2157–2165.
Supporting Information
License and Terms
Supporting Information File 1
This is an Open Access article under the terms of the
Creative Commons Attribution License
Experimental details, characterization data and spectra.
[http://www.beilstein-journals.org/bjoc/content/
supplementary/1860-5397-10-223-S1.pdf]
(http://creativecommons.org/licenses/by/2.0), which
permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Acknowledgements
The license is subject to the Beilstein Journal of Organic
Chemistry terms and conditions:
This work was supported by the GRK 1626 (“Chemical Photo-
catalysis”) of the DFG.
(http://www.beilstein-journals.org/bjoc)
References
The definitive version of this article is the electronic one
which can be found at:
1. Dodds, D. R.; Gross, R. A. Science 2007, 318, 1250–1251.
doi:10.1126/science.1146356
doi:10.3762/bjoc.10.223
2. Barton, D. H. R.; McCombie, S. W. J. Chem. Soc., Perkin Trans. 1
1975, 1574–1585. doi:10.1039/p19750001574
3. Studer, A.; Amrein, S. Angew. Chem., Int. Ed. 2000, 39, 3080–3082.
doi:10.1002/1521-3773(20000901)39:17<3080::AID-ANIE3080>3.0.CO
;2-E
4. Lam, K.; Markó, I. E. Org. Lett. 2008, 10, 2773–2776.
doi:10.1021/ol800944p
5. Saito, I.; Ikehira, H.; Kasatani, R.; Watanabe, M.; Matsuura, T.
J. Am. Chem. Soc. 1986, 108, 3115–3117. doi:10.1021/ja00271a057
6. Prudhomme, D. R.; Wang, Z.; Rizzo, C. J. J. Org. Chem. 1997, 62,
8257–8260. doi:10.1021/jo971332y
7. Shen, B.; Jamison, T. F. Aust. J. Chem. 2013, 66, 157–164.
doi:10.1071/CH12426
8. Bordoni, A.; de Lederkremer, R. M.; Marino, C. Carbohydr. Res. 2006,
341, 1788–1795. doi:10.1016/j.carres.2006.04.012
9. Bordoni, A.; de Lederkremer, R. M.; Marino, C. Tetrahedron 2008, 64,
1703–1710. doi:10.1016/j.tet.2007.12.005
10.Nguyen, J. D.; Reiß, B.; Dai, C.; Stephenson, C. R. J. Chem. Commun.
2013, 49, 4352–4354. doi:10.1039/c2cc37206a
11.Nguyen, J. D.; D’Amato, E. M.; Narayanam, J. M. R.;
Stephenson, C. R. J. Nat. Chem. 2012, 4, 854–859.
doi:10.1038/nchem.1452
12.Arpe, H.-J. Industrielle Organische Chemie, 6th ed.; Wiley-VCH:
Weinheim, Germany, 2007; p 536.
2165