D. E. De Vos et al.
explained below. Much better results were obtained when the
heterogeneous catalysts were replaced by a homogeneous cat-
alyst precursor, that is, HRuCl(CO)(PPh3)3 (Table 1, entry 2). Such
ruthenium hydride compounds have been reported in litera-
ture as hydrogenolysis catalysts for fructose and glucose,[17,18]
for example in N-methyl-2-pyrrolidinone. In the presence of
HRuCl(CO)(PPh3)3 and KOH, and in BMImCl as the solvent, cel-
lobiose conversion was almost complete, with sorbitol as a
major product. Minor amounts of C4-polyols and C5-polyols
were also recovered, originating from the hydrogenolysis of
glucose, as reported by Andrews and Klaeren.[17] The role of hy-
droxide ions in these reactions is probably manifold: basicity
could promote the retro-aldol cleavage that results in fragmen-
tation of the sugar molecules; but the base may also promote
coordination of the sugar molecules to ruthenium as alcohol-
ates, and bases are known to facilitate the heterolytic forma-
tion of hydrides from dihydrogen on ruthenium complexes.
The latter effect has not only been reported for hydrogenation
in conventional solvents;[19] it is known that addition of bases
promotes hydrogenations in ionic liquids as well.[20]
Results and Discussion
Initially, reductive splitting of ionic-liquid-dissolved cellulose
was approached as a hydrogenolysis of an acetal, for which
several methods have been reported.[14–16] In order to investi-
gate the effectiveness of reported catalysts for acetal or ketal
hydrogenolysis, 1,1-diethoxycyclohexane was selected as the
reference compound. While this compound is not soluble in
BMImCl, it is a liquid at room temperature and thus can even
be hydrogenolyzed in solvent-free conditions. As a mixture of
Pt and Rh catalysts has been reported as being most effective
for ketal hydrogenolysis,[16b] these catalysts were applied to the
hydrogenolysis of 1,1-diethoxycyclohexane. Attempts to hydro-
genolyze this substrate in the absence of Lewis acids gave no
result. In the presence of a Lewis-acidic promoter hydrogenoly-
sis was conducted successfully. BF3·Et2O and AlCl3 were tested,
and the former was found to be more effective. Neat 1,1-dieth-
oxycyclohexane (0.5 mL) was hydrogenolyzed successfully by
using a combination of the heterogeneous catalysts Rh/C
(0.003 g; 5 wt%) and Pt/C (0.007 g; 0.5 wt%) in the presence
of the Lewis acid BF3·Et2O (0.5 mL) at 208C under 2 MPa of hy-
drogen gas during 7 h. The only product was ethoxycyclohex-
ane, which was obtained in 100% yield, together with ethanol.
The next compound chosen for investigation was cellobiose.
Cellobiose consists of two glucose molecules linked in a b(1!
4) bond, and thus represents a dimeric model for the cellulose
polymer. Cellobiose is well-soluble in BMImCl; cellobiose and
its reaction products were easily and unequivocally quantified
by GC and GC–MS after derivatization with a silylating agent.
Our initial approach to apply the same conditions in BMImCl
as for the ketal hydrogenolysis in conventional media was not
fruitful, even when a range of conditions were tested. A repre-
sentative result is shown in Table 1 (entry 1): while some glu-
cose was obtained by hydrolysis of the cellobiose, the glucose
did not react further to sorbitol, despite the presence of metal
catalysts and a hydrogen pressure of 3.5 MPa in the vessel.
Some of the glucose was transformed to levoglucosan and 5-
hydroxymethylfurfural, which are typical glucose degradation
products. This highlights that in certain conditions, the hydro-
genation in the ionic liquid medium is problematic, as will be
Eventually, cellulose was applied as the reagent in the form
of the microcrystalline commercial product Avicel. As reported,
BMImCl proved very effective in dissolving cellulose.[4] In a first
attempt, we used the mixture of heterogeneous catalysts that
was effective for ketal hydrogenolysis, that is, Rh/C and Pt/C,
together with the Lewis acid BF3 (Table 2, entry 1). However, no
monomeric products with 6 or less carbon atoms or dimers
were detected at all, confirming that selective depolymeriza-
tion of cellulose is a nontrivial task. Attention therefore turned
to the homogeneous catalyst HRuCl(CO)(PPh3)3, in combination
with the base KOH, in the same ratio as applied for cellobiose.
After extended reaction times, a modest conversion (20%) was
obtained, with surprisingly glucose, rather than hydrogenolysis
or hydrogenation products, as the principal reaction product
(Table 2, entry 2). This indicates that hydrolysis of the b-1,4
chain is the initial reaction, even if water was not intentionally
added to the mixture. Minor contamination of BMImCl by
water should always be taken into account; Karl–Fischer titra-
tion showed that the level of contamination was ca. 0.1 wt%.
The cellulose that is used as a reagent may contain some addi-
tional water (typically 4–7 wt%
based on cellulose).[21] As the hy-
drolysis reaction is expected to
Table 1. Hydrogenation of cellobiose in BMImCl.
be counteracted by base, the
ratio base/Ru complex was low-
Entry Catalyst
1[a]
Amount of product in final mixture Total conversion Yield of hydro-
ered, which resulted in a strong
increase of the conversion to
72% (Table 2, entry 3). Addition
of a minute amount of water
(0.25 wt%) to provide a stoichio-
metric amount smoothly raised
the conversion to ca. 100%,
without compromising the dis-
solution of the cellulose.[21] How-
ever, in these conditions only a
minor fraction of the hexoses
was hydrogenated, and glucose
genated product
Rh/C (0.002 g; 5 wt%),
dimers: 52%
48%
90%
0%
Pt/C (0.0006 g; 0.5 wt% ) glucose: 10%
levoglucosan: 21%
hydroxymethylfurfural 17%
HRuCl(CO)(PPh3)3 (0.01 g) dimers: 10%
2[b]
76%
glucose: 6%
sorbitol: 43%
C5 alcohols: 24%
C4 alcohols: 17%
[a] Reaction conditions: cellobiose (0.0416 g), solvent (0.7 g BMImCl), 1508C, hydrogen (3.5 MPa, measured at
258C), 15 h. [b] Reaction conditions: cellobiose (0.01 g), solvent (1 g BMImCl), KOH (0.0072 g), 1508C, hydrogen
(3.5 MPa, measured at 258C), 24 h.
92
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2010, 3, 91 – 96