Metal-free deoxygenation of carbohydrates

Article abstract of DOI:10.1002/anie.201306864

The conversion of readily available cellulosic biomass to valuable feedstocks and fuels is an attrative goal but a challenging transformation that requires the cleavage of multiple nonactivated C-O bonds. Herein, the Lewis acid trispentafluorophenylborane (B(C₆F₅)₃) is shown to catalyze the metal-free hydrosilylative reduction of monosaccharides and polysaccharides to give hydrocarbons with reduced oxygen content. The choice of the silane reductant influences the degree of deoxygenation, with diethylsilane effecting the complete reduction to produce hexanes while tertiary silanes give partially deoxygenated tetraol and triol products. A spoonful of sugar: The Lewis acid B(C₆F₅)₃ catalyzes the complete deoxygenation of carbohydrates to give a mixture of hexane and hexene isomers, with diethylsilane (Et₂SiH₂) providing the hydride equivalent. A variety of carbohydrates including methyl cellulose can be deoxygenated by this metal-free method, and the system can be tuned for selective deoxygenation at certain sites. Copyright

Full text of DOI:10.1002/anie.201306864

.
Angewandte
Communications
DOI: 10.1002/anie.201306864
Biomass Deoxygenation
Hot Paper
Metal-Free Deoxygenation of Carbohydrates**
Laura L. Adduci, Matthew P. McLaughlin, Trandon A. Bender, Jennifer J. Becker, and
Michel R. Gagnꢀ*
Abstract: The conversion of readily available cellulosic
biomass to valuable feedstocks and fuels is an attrative goal
but a challenging transformation that requires the cleavage of
to silyl ethers (with concomitant H2 evolution). For this
reason experiments to contrast the deoxygenation activity of
catalysts 1 and 2 used sugars that were preprotected
(Figure 1), while experiments to gauge the breadth of the
substrate scope of 2 used unprotected sugars for convenience.
13C NMR analysis, which proved the most useful method for
À
multiple nonactivated C O bonds. Herein, the Lewis acid
trispentafluorophenylborane (B(C6F5)3) is shown to catalyze
the metal-free hydrosilylative reduction of monosaccharides
and polysaccharides to give hydrocarbons with reduced
oxygen content. The choice of the silane reductant influences
the degree of deoxygenation, with diethylsilane effecting the
complete reduction to produce hexanes while tertiary silanes
give partially deoxygenated tetraol and triol products.
À
monitoring C O bond cleavage, indicated that the metal-free
borane-catalyzed reaction proceeds faster at the 5 mol%
catalyst level than the iridium-catalyzed version at the same
loading. For example, after one hour at room temperature, the
B(C6F5)3-catalyzed deoxygenation of per-TMS-protected glu-
cose (TMS-3) displayed exclusively deoxygenated products
by 13C NMR spectroscopy, while oxygenates persisted in the
iridium-catalyzed reaction (Figure 1).
T
he ubiquity of petroleum-based materials in everyday
products has driven a growing interest in developing renew-
able sources for important feedstocks and fuels.[1] As many
feedstocks and fuels are composed of hydrocarbons or
partially oxygenated hydrocarbons, one natural source for
their production might be cellulosic biomass, which is readily
available but requires deoxygenation for further use,[2] a need
with significant inherent challenges.[3]
In both cases, however, the reactions eventually pro-
À
ceeded to a point where all observable C O bonds had been
cleaved. Like the (POCOP)IrH+-catalyzed deoxygenation,
the B(C6F5)3-catalyzed deoxygenation yields a product mix-
ture that is dominated by hexane isomers, principally n-
We recently reported that the hydrosilylative reduction of
glucose to a mixture of hexane isomers can be catalyzed by
a
(POCOP)IrH+ species ([(POCOP)IrH(acetone)]+[B-
(C6F5)4]À, 1; POCOP = 2,6-[PO(tBu)2]2C6H3), with Et2SiH2
serving as the terminal hydride source.[4] Reported herein
are investigations showing that the commercially available
Lewis acid trispentafluorophenylborane (B(C6F5)3, 2),
a known catalyst for the hydrosilylative reduction of primary
and secondary alcohols,[5] also catalyzes the hydrosilylative
reduction of carbohydrates (Scheme 1),[6] and moreover that
the reaction can be tuned to selectively deoxygenate glucose
to value-added products.
Scheme 1. Example of B(C6F5)3-catalyzed hydrosilylative carbohydrate
deoxygenation. 2-MP=2-methylpentane, 3-MP=3-methylpentane.
When unprotected carbohydrates are utilized, the deox-
ygenation is preceded by an in situ conversion of the alcohols
[*] L. L. Adduci,[+] Dr. M. P. McLaughlin,[+] T. A. Bender,
Prof. Dr. M. R. Gagnꢀ
Department of Chemistry
University of North Carolina at Chapel Hill
Chapel Hill, NC 27599-3290 (USA)
E-mail: mgagne@unc.edu
Dr. J. J. Becker
U.S. Army Research Office
P.O. Box 1221, Research Triangle Park, NC 27709 (USA)
[+] These authors contributed equally to this work.
[**] M.P.M. thanks the National Research Council for a postdoctoral
fellowship, L.L.A. thanks the Progress Energy Foundation/UNC
Institute for the Environment, M.R.G. thanks the Department of
Energy (DE-FG02-05ER15630), and J.J.B. thanks the Army Research
Office for funding.
Figure 1. 150 MHz 13C{1H} NMR spectra of a) TMS-3 starting material
with cyclooctane standard, b) reaction mixture from the hydrosilylative
reduction of TMS-3 catalyzed by 5 mol% 1 with 20 equiv of Et2SiH2
after 1 hour, c) reaction mixture from the hydrosilylative deoxygenation
of TMS-3 catalyzed by 5 mol% 2 with 20 equiv Et2SiH2 after 1 hour.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201306864.
1646
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1646 –1649

Angewandte
Chemie
and smallest for the ring-opened glucitol (6)
(Figure 2).[4] Intriguingly, the relative proportions
of 2-MP and 3-MP were also substrate dependent,
with 2-deoxyglucose providing a disproportion-
ately high amount of 3-MP. In all cases, other C6
hydrocarbons including methylcyclopentane, cy-
clohexane, 2,2-dimethylbutane, and 2,3-dimethyl-
butane were either not observed or present in
quantities too small to be reliably quantified by
Scheme 2. Iridium- versus borane-catalyzed deoxygenation of TMS-4.
hexane, 2-methylpentane (2-MP), and 3-methylpentane (3-
MP). The observation of 2-MP and 3-MP points to the
likelihood of alkyl shifts and carbenium ions in the deoxyge-
nation sequence.[7]
NMR spectroscopy.
As noted above, the iridium catalyst 1 reacted to give little
detectable elimination, whereas the borane catalyst 2 gave
significant elimination (up to 20% overall yield). Spiking
product mixtures with authentic samples revealed that
alkenes observed in the 13C NMR spectra of the B(C6F5)3-
catalyzed reactions included trans-2-hexene > trans-3-
hexene > cis-2-hexene, with cis-3-hexene not being observed
in any appreciable quantity. Because the absolute yield of
alkenes was small, not exceeding a sum of 25%, a correlation
between the quantity of hexenes and experimental parame-
ters was challenging.
A full accounting for the mass balance of the completely
deoxygenated products was hampered both by the volatility
of the products and the complexity of the mixture. Looking at
only the aforementioned hexane and hexene isomers, a sum
of the products consistently gave an estimated total yield
between 60 and 90% for various sugar substrates, although
the absolute yield of each hydrocarbon typically varied 5–
10% between trials. This variability also characterized the
product ratios in most of the runs.[9] 13C NMR spectra
revealed the wide variety of species contributing to the total
mass balance, and although several products including C2–C4
hydrocarbons were identified by GC–MS analysis, many
remain unidentified. Additionally, while 19F NMR spectros-
copy revealed minor catalyst decomposition during the
experiment, the addition of a second portion of carbohydrate
and silane after the first portion had been fully consumed
showed the second portion of carbohydrate to be reduced in
good yield (80%). Catalysis also continued in the presence of
While catalysts 1 and 2 were able to effect complete
deoxygenation, the hydrocarbon product distribution was
different in several cases. For example, the B(C6F5)3-catalyzed
deoxygenation of TMS-3 affords a significant quantity of
hexene isomers in addition to the alkanes observed in the
iridium-catalyzed hydrosilylation. Additionally, for TMS-
protected methyl glucoside (TMS-4), the product distribution
arising from the iridium-catalyzed reaction favored the alkyl-
shifted 2-MP and 3-MP isomers,[4] while the major deoxyge-
nation product of TMS-4 under B(C6F5)3 catalysis was n-
hexane (Scheme 2 and Figure S7 in the Supporting Informa-
tion). The build-up of hexene isomers for 2 but not for 1 is also
a point of differentiation (Figure 1). As recently discussed,[6]
studies by Brookhart et al. for 1[8] and by the Piers, McRae,
and Gevorgyan groups for 2[5] indicate that there are
numerous similarities and some differences between these
two catalysts. The commercial availability of 2 and ease of
handling were considered major benefits, prompting an
exploration of its reactivity with a range of carbohydrates.
Employment of B(C6F5)3 in combination with diethylsi-
lane enabled the deoxygenation of a wide range of glycosidic
substrates. While complete deoxygenation was observed by
13C NMR spectroscopy in all cases, the relative yields of
alkane and alkene products were dependent on the substrate,
catalyst, and protecting group (Figure 2). The proportion of
alkyl-shifted C6 products was greatest for 1-deoxyglucose (5)
Figure 2. Hydrosilylative defunctionalization of representative carbohydrates. The proportions of each of the hexane and hexene products are
reported as a percentage of the total. Alkane and alkene yields were determined by semiquantitative 13C NMR spectroscopy, corroborated in select
cases by GC analysis (see the Supporting Information for details and absolute yields). General reaction conditions: 5 mol% catalyst 2, 24 equiv of
Et2SiH2, RT, 18 h. Data are the average of two trials, except where * denotes the result of one trial.
Angew. Chem. Int. Ed. 2014, 53, 1646 –1649
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1647

.
Angewandte
Communications
small amounts of added water (6 equiv relative to B(C6F5)3),
which was quickly hydrosilylated to provide dry reaction
conditions, indicating that rigorous drying of catalyst, sub-
strate, or solvent is not necessary. Polymethylhydrosiloxane,
(-(OSiMe(H))n-), an inexpensive waste product of the silicone
industry, also reduces carbohydrates under B(C6F5)3 catalysis
to give hexane and hexene isomers in 80% yield.[10]
Reactions in chlorobenzene and methylene chloride both
resulted in facile reduction of glucose and a good yield of the
fully deoxygenated products in similar proportions. Although
the reaction mixture appeared homogeneous when pentane
was used as the solvent, the rate and yield of the reaction was
dramatically reduced, perhaps speaking to the importance of
generating reactive ion pairs during silane activation and/or
oxygen abstraction. The rate and yield of the reaction were
also reduced with the tertiary silane Me2EtSiH, indicating
a significant steric effect. This difference is evident in the
reduction of Me2EtSi-protected glucitol (Me2EtSi-6) with
Me2EtSi-H, which proceeds with only partial consumption of
Scheme 3. Et3Si-3 is reduced first to glucitol (6) and then to 1,6-deoxy
À
C O bonds after 1 week [Eq. (1)]. Under these conditions it
glucitol.
À
is evident that reduction of primary C O bonds is kinetically
favored, though other minor products are also observed. In
deoxyglucitol (Scheme 3).[12] The preference for ring opening
is likely the result of a comparatively increased Lewis basicity
and steric accessibility of the pyranose oxygen.
Finally, the direct deoxygenation of cellu-
lose has historically been limited by its low
solubility and steric congestion, therefore often
requiring pretreatment to separate the chains
or cleave the glycosidic linkages before defunc-
tionalization.[13] Not surprisingly then, cellulose
itself was completely insoluble and unreactive
contrast, complete reduction of Me2EtSi-6 occurs with
Et2SiH2 after only 2 h.
under our conditions, even at elevated temperatures. Com-
mercially available 30% methylated cellulose (7), however,
was significantly more soluble. While displaying no apparent
solubility in CH2Cl2, this material is rapidly solubilized by the
addition of B(C6F5)3 and Et2SiH2, which presumably silylates
at least some of the unprotected hydroxy groups and allows
for compete dissolution under reaction conditions. Although
the degree of silyl versus methyl protection at the time of
reduction is unknown, fully deoxygenated products are
observable within 20 min (Scheme 4 and Figure 2), with
yields reaching 80% after our 18 h assay. This suggests both
that the method is exceptionally active and can consequently
activate congested, minimally soluble substrates, and that the
cellulose problem may be exclusively one of solubility.
In summary, we have demonstrated a new mild method
for the metal-free deoxygenation of carbohydrates to give
Under silane-limited conditions (1 equiv), we observed
that Me2EtSi-H converted Me2EtSi-6 to 1,6-deoxyglucitol as
the major product. On a preparative scale (0.8 mmol),
treating unprotected 6 with 9 equiv of Me2EtSi-H in the
presence of B(C6F5)3 gave 1,6-deoxy glucitol in 70% yield (by
1
in situ H NMR spectroscopy). The tetraol was isolated in
62% yield, with 2,3,4-hexanetriol (6:1 d.r., the product of one
additional reduction) also contributing to the mass balance
( ꢀ 7%). With 2 equiv of Et2SiH2, the product mixture was
considerably more complex, but 1,6-deoxyglucitol was also
identified as the major observed product by 13C NMR
spectroscopy (> 50%, see Figure S11 in the Supporting
Information). These data suggest that the reaction with
Et2SiH2 may be inherently less selective or that the O-
SiEt2(H) groups generated by silyl exchange may participate
in further, presumably intramolecular, reductions. The selec-
À
tive reduction of primary over secondary C O bonds parallels
that observed for simple alcohols.[5a,b]
Slowing the reaction further by using Et3Si-protected
glucose (Et3Si-3) with Me2EtSiH allowed us to observe the
initial steps of the reaction for a ring-closed sugar. Curiously,
the first step causes ring opening over the course of minutes to
give spectroscopically identified glucitol (6) as the major
product,[11] and not reduction of the primary C6 or anomeric
C1 positions. Ring opening was followed by a considerably
slower (hours) reduction of the primary sites to give 1,6-
Scheme 4. B(C6F5)3-catalyzed hydrosilylative deoxygenation of 30%
methylcellulose (7). Major products observed by 13C NMR spectros-
copy.
1648
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1646 –1649

Angewandte
Chemie
Organometallics 2010, 29, 1045 – 1048; k) J. M. Nichols, L. M.
mixtures of alkane and alkene hydrocarbons. The product
distribution and rate of reduction are sensitive to substrate,
solvent, silane, protecting group, and catalyst. This feature
makes it possible to develop conditions for the selective
deoxygenation of glucose to more valuable products like 1,6-
deoxyglucitol. The mechanism of deoxygenation is currently
under investigation.
Bishop, R. G. Bergman, J. A. Ellman, J. Am. Chem. Soc. 2010,
132, 12554 – 12555; l) R. Palkovits, K. Tajvidi, A. M. Ruppert, J.
Procelewska, Chem. Commun. 2011, 47, 576 – 578; m) A. G.
Sergeev, J. F. Hartwig, Science 2011, 332, 439 – 443; n) M. E.
Thibault, D. V. DiMondo, M. Jennings, P. V. Abdelnur, M. N.
Eberlin, M. Schlaf, Green Chem. 2011, 13, 357 – 366; o) T. J.
Ahmed Foskey, D. M. Heinekey, K. I. Goldberg, ACS Catal.
2012, 2, 1285 – 1289; p) V. Choudhary, S. I. Sandler, D. G.
Vlachos, ACS Catal. 2012, 2, 2022 – 2028; q) M. R. Grochowski,
W. Yang, A. Sen, Chem. Eur. J. 2012, 18, 12363 – 12371; r) J. Julis,
W. Leitner, Angew. Chem. 2012, 124, 8521 – 8521; Angew. Chem.
Int. Ed. 2012, 51, 8395 – 8395; s) R. S. Assary, A. C. Atesin, Z. Li,
L. A. Curtiss, T. J. Marks, ACS Catal. 2013, 3, 1908 – 1914; t) T.
Prasomsri, T. Nimmanwudipong, Y. Roman-Leshkov, Energy
Environ. Sci. 2013, 6, 1732 – 1738; u) A. D. Sutton, F. D. Waldie,
R. Wu, M. Schlaf, L. A. Silks, J. C. Gordon, Nat. Chem. 2013, 5,
428 – 432; v) D. B. Lao, A. C. E. Owens, D. M. Heinekey, K. I.
Goldberg, ACS Catal. 2013, 3, 2391 – 2396.
Received: August 5, 2013
Published online: January 13, 2014
Keywords: biomass · carbohydrates · deoxygenation ·
.
hydrosilylation · renewable feedstocks
[1] a) A. M. Ruppert, K. Weinberg, R. Palkovits, Angew. Chem.
2012, 124, 2614 – 2654; Angew. Chem. Int. Ed. 2012, 51, 2564 –
2601; b) F. M. A. Geilen, B. Engendahl, A. Harwardt, W.
Marquardt, J. Klankermayer, W. Leitner, Angew. Chem. 2010,
122, 5642 – 5646; Angew. Chem. Int. Ed. 2010, 49, 5510 – 5514;
c) M. Besson, P. Gallezot, C. Pinel, Chem. Rev. 2013, DOI:
10.1021/cr4002269.
[4] M. P. McLaughlin, L. L. Adduci, J. J. Becker, M. R. Gagnꢂ, J.
Am. Chem. Soc. 2013, 135, 1225 – 1227.
[5] a) V. Gevorgyan, J.-X. Liu, M. Rubin, S. Benson, Y. Yamamoto,
Tetrahedron Lett. 1999, 40, 8919 – 8922; b) V. Gevorgyan, M.
Rubin, S. Benson, J.-X. Liu, Y. Yamamoto, J. Org. Chem. 2000,
65, 6179 – 6186; c) D. J. Parks, J. M. Blackwell, W. E. Piers, J.
Org. Chem. 2000, 65, 3090 – 3098; d) R. D. Nimmagadda, C.
McRae, Tetrahedron Lett. 2006, 47, 5755 – 5758.
[6] T. Robert, M. Oestreich, Angew. Chem. 2013, 125, 5324 – 5326;
Angew. Chem. Int. Ed. 2013, 52, 5216 – 5218.
[7] Similar carbon skeleton rearrangements have been observed
during hydrodefluorinations. See: C. Douvris, O. V. Ozerov,
Science 2008, 321, 1188 – 1190.
[2] a) T. Werpy, G. Petersen, Top Value Added Chemicals from
Biomass, Volume I, Results of Screening for Potential Candi-
dates from Sugar and Synthesis Gas, U.S. Department of Energy
DOE/GO-102004-1992, August 2004; b) A. J. Ragauskas, C. K.
Williams, B. H. Davison, G. Britovsek, J. Cairney, C. A. Eckert,
W. J. Frederick, J. P. Hallett, D. J. Leak, C. L. Liotta, J. R.
Mielenz, R. Murphy, R. Templer, T. Tschaplinski, Science 2006,
311, 484 – 489.
[3] For reviews in this area, see: a) A. Corma, S. Iborra, A. Velty,
Chem. Rev. 2007, 107, 2411 – 2502; b) P. Gallezot, Chem. Soc.
Rev. 2012, 41, 1538 – 1558; c) M. Schlaf, E. Thibault Michelle, D.
DiMondo, D. Taher, E. Karimi, D. Ashok, Int. J. Chem. React.
Eng. 2009, 7, 1542 – 6580. For other examples, see: d) G. W.
Huber, J. N. Chheda, C. J. Barrett, J. A. Dumesic, Science 2005,
308, 1446 – 1450; e) Y. Romꢀn-Leshkov, J. N. Chheda, J. A.
Dumesic, Science 2006, 312, 1933 – 1937; f) H. Zhao, J. E.
Holladay, H. Brown, Z. C. Zhang, Science 2007, 316, 1597 –
1600; g) E. L. Kunkes, D. A. Simonetti, R. M. West, J. C.
Serrano-Ruiz, C. A. Gꢁrtner, J. A. Dumesic, Science 2008, 322,
417 – 421; h) E. Arceo, P. Marsden, R. G. Bergman, J. A. Ellman,
Chem. Commun. 2009, 3357 – 3359; i) M. Schlaf, P. Ghosh, P. J.
Fagan, E. Hauptman, R. M. Bullock, Adv. Synth. Catal. 2009,
351, 789 – 800; j) P. J. Fagan, M. H. Voges, R. M. Bullock,
[8] J. Yang, P. S. White, M. Brookhart, J. Am. Chem. Soc. 2008, 130,
17509 – 17518.
[9] In reactions with presilylated sugars the yields varied by as much
as 15–17%.
[10] Under conditions optimized for scale-up using polymethylhy-
drosiloxane (PMHS) and ortho-xylene, hexanes formed in 50%
yield (NMR analysis), of which 60% could be isolated in 97%
purity.
[11] The product was identified by comparison to authentic glucitol.
[12] See the Supporting Information for characterization data.
[13] a) K. Petzold, A. Koschella, D. Klemm, B. Heublein, Cellulose
2003, 10, 251 – 269; b) J. Fan, M. De bruyn, V. L. Budarin, M. J.
Gronnow, P. S. Shuttleworth, S. Breeden, D. J. Macquarrie, J. H.
Clark, J. Am. Chem. Soc. 2013, 135, 11728 – 11731.
Angew. Chem. Int. Ed. 2014, 53, 1646 –1649
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 1649