CHEMCATCHEM
FULL PAPERS
Results and Discussion
tention both in homogeneous and heterogeneous catal-
ysis.[4a,7–8] Levulinic acid is now considered amongst the top 12
chemical building blocks from biomass because of the versatile
applications and synthetic utility of GVL as depicted in
Scheme 1.[3a,b] So far, the hydrogenation of levulinic acid to
GVL was accomplished by heterogeneous catalysts (e.g., ruthe-
nium or platinum catalysts supported on carbon, aluminum
oxide, or titanium dioxide).[7] Although this approach offers the
advantage of easy catalyst recycling, high pressure and/or tem-
peratures are necessary to obtain high conversion. Moreover,
such processes are usually associated with the formation of
over-hydrogenated 2-methyltetrahydrofuran, which has safety
issues as the latter product can readily be converted into the
corresponding hazardous peroxide.[8f] These problems prompt-
ed researchers to look into analogous homogeneous catalytic
pathways.[4a,8] Already in 1982, Yoshikawa et al. reported the
catalytic hydrogenation of levulinic acid using RuCl2(PPh3)2 as
the catalyst at 1808C.[8a] Almost 25 years later in 2008, Horvꢁth
et al. observed that Ru(acac)3 (acac=acetylacetonate) ligated
with PBu3, tris(3-sulfonatophenylphosphine) (TPPTS) provides
good to moderate yields of GVL at 80 bar hydrogen pressure
and high temperature (2008C).[8b] More recently, Guo and Fu
et al. developed a RuCl3/PPh3-based catalytic system for levu-
linic acid hydrogenation using formic acid as hydrogen source
(no external hydrogen gas applied). Again, high temperature
(2008C) and 10 mol% base were necessary to obtain sufficient
conversion to achieve high yield of GVL.[8c] In 2010, Leitner and
co-workers improved these earlier methods by using Ru(acac)3
as the catalyst, PnOct3 as the ligand and NH4PF6 as an additive
at 100 bar hydrogen pressure and 1608C.[8d] The following year,
a biphasic RuCl3–TPPTS system was developed by Heeres et al.
in which the catalytic system can be recycled but the product
yield dropped by approximately 30% after the first catalytic
run.[8e] Although these catalysts (see below) allowed for high
yields of the desired product, still the turnover numbers (TON,
moles of products/moles of catalyst) need to be improved. The
maximum TON does not exceed 1560 if using the reported ho-
mogeneous ruthenium-based catalyst (Supporting Information,
Scheme S1).[8a–e] In this aspect, the work of Zhou et al. is note-
worthy. By using sophisticated iridium pincer complexes, excel-
lent catalyst TONs (71000) were achieved at 1008C. However,
addition of 120 mol% of base eventually limits its practical po-
tential.[8f] In 2013, a high TON of 78000 was achieved by using
half-sandwich iridium complex at 1208C; reported again by
Guo and Fu et al. (Scheme S1).[8g]
The initial optimizations were performed in the presence of
Ru(acac)3 as the catalyst and p-toluenesulfonic acid (PTSA) as
an additive by using methyl levulinate as the model substrate.
Unlike most of the relevant reports under homogeneous con-
ditions (Scheme S1),[4a,8] here only a catalytic amount of the ad-
ditive (1.75 equivalents with respect to the catalyst) was used.
In the absence of both acid and ligand, only 19% yield of GVL
was obtained. The addition of PTSA increased the yield merely
to 21% (Figure 1, Table S1).
Figure 1. Hydrogenation reaction of methyl levulinate under variation of li-
gands (reaction conditions: 0.08 mol% Ru(acac)3, 0.08 mol% ligand,
0.14 mol% additive, substrate, 24 mmol, 50 bar H2, solvent: THF, 1408C,
22 h). For details, see Table S1.
Next, various mono-, bi-, and tridentate phosphine-based li-
gands were added to the model reaction (Scheme 2, Figure 1).
The product yield was slightly improved (up to 27%) in the
presence of monodentate PPh3 (L1) but only traces of activity
was noted in the case of bidentate Xantphos (L2) as the
ligand. To our delight, the commercially available triphos,
bis(2-diphenylphosphinoethyl)phenylphosphine (L3) led to
82% yield under identical reaction conditions. Notably, another
commercially available tridentate ligand (tris[2-(diphenylphos-
phino)ethyl]phosphine, L4) gave rise to 35% of GVL. Next, ac-
cording to the literature we prepared tris((diphenylphosphino)-
methyl)amine (L5); however, subsequent catalytic tests re-
vealed low activity (24% of GVL). These initial results prompted
us to explore the structural aspects of the triphos backbone of
L3 in more detail (Figure 1, Scheme 2). Hence, three newly de-
signed triphos analogue ligands {[(phenylphosphinediyl)bis-
(2,1-phenylene)]bis(methylene)}bis-(diphenylphosphine) (L6),
{[(phenylphosphinediyl)bis(2,1-phenylene)]bis(methylene)}bis-
(di-tert-butylphosphine) (L7) and {[(phenylphosphinediyl)bis-
Herein, we report three ruthenium–triphos-based efficient
catalytic systems, which yielded GVL in excellent yield (up to
95%) with very high TON (75855) under relatively mild reac-
tion conditions (1408C) on a preparative scale (substrate load-
ing as high as 185 mmol) and under solvent-free (neat) condi-
tions. In essence, we believe these catalyst systems constitute
an important step towards a practical GVL production in a sus-
tainable manner under homogeneous conditions (Scheme S1).
(methylene)]bis(2,1-phenylene)}bis-(diphenylphosphine)
(L8)
were prepared and fully characterized. The synthetic routes of
these ligands (L6–L8) are depicted in Scheme 3 and the de-
tailed procedures are described in the Supporting Information.
Interestingly, L6 and L7 revealed significantly different cata-
lytic performances. In the presence of L6, 81% of GVL was ob-
tained, but L7 gave only 7% of the desired product under
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 7
&
2
&
ÞÞ
These are not the final page numbers!