Hydrodeoxygenation of 2,5-hexanedione and 2,5-dimethylfuran by water-, air-, and acid-stable homogeneous ruthenium and iridium catalysts

Article abstract of DOI:10.1021/cs501202t

The complexes [(4′-Ph-terpy)Ru(H₂O)₃](OTf)₂ and [(4′-Ph-terpy)Ir(OTf)₃] have been evaluated as catalysts for the conversion of 2,5-hexanedione and 2,5-dimethylfuran to hydrodeoxygenated products in aqueous acidic medium at elevated temperature (150-225 °C) under hydrogen gas (5.5 MPa). These two substrates form part of a value chain leading from C₆ sugars to 2,5-hexanediol, 2,5-dimethyltetrahydrofuran, and hexane, which can be generated by the homogeneously acting ruthenium catalyst in up to 69%, 80%, and 10% yield, respectively, while at T > 175 °C the iridium system decomposes to a highly active but heterogeneously acting coating in the reactor defeating the premise of a homogeneous catalyst system. The deactivation and decomposition pathway of both catalysts leads to the formation of a series of isostructural complexes [M(4′-Ph-terpy)₂]ⁿ⁺ (M = Fe, Ni, Ru, Ir; n = 2, 3) characterized by ESI-MS and single crystal X-ray crystallography, in which the source of the Fe and Ni is the 316SS reactor body.

Full text of DOI:10.1021/cs501202t

Research Article
pubs.acs.org/acscatalysis
Hydrodeoxygenation of 2,5-Hexanedione and 2,5-Dimethylfuran by
Water‑, Air‑, and Acid-Stable Homogeneous Ruthenium and Iridium
Catalysts
Ryan J. Sullivan, Elnaz Latifi, Benjamin K.-M. Chung, Dmitriy V. Soldatov, and Marcel Schlaf*
Department of Chemistry, University of Guelph, The Guelph-Waterloo Centre for Graduate Work in Chemistry (GWC)2, 50 Stone
Road East, Guelph, Ontario, Canada, N1G 2W1
S
* Supporting Information
ABSTRACT: The complexes [(4′-Ph-terpy)Ru(H2O)3](OTf)2
and [(4′-Ph-terpy)Ir(OTf)3] have been evaluated as catalysts for
the conversion of 2,5-hexanedione and 2,5-dimethylfuran to
hydrodeoxygenated products in aqueous acidic medium at
elevated temperature (150−225 °C) under hydrogen gas (5.5
MPa). These two substrates form part of a value chain leading
from C6 sugars to 2,5-hexanediol, 2,5-dimethyltetrahydrofuran,
and hexane, which can be generated by the homogeneously
acting ruthenium catalyst in up to 69%, 80%, and 10% yield,
respectively, while at T > 175 °C the iridium system decomposes
to a highly active but heterogeneously acting coating in the
reactor defeating the premise of a homogeneous catalyst system.
The deactivation and decomposition pathway of both catalysts leads to the formation of a series of isostructural complexes
[M(4′-Ph-terpy)2]n+ (M = Fe, Ni, Ru, Ir; n = 2, 3) characterized by ESI-MS and single crystal X-ray crystallography, in which the
source of the Fe and Ni is the 316SS reactor body.
KEYWORDS: homogeneous catalysis, aqueous media, hydrogenation, hydrodeoxygenation, biomass conversion, catalyst decomposition
chloro-2,2′-bipyridine) as an effective catalyst for the selective
hydrogenation of furfural to tetrahydrofurfurylalcohol in
ethanol solvent.18 The same catalyst (as the corresponding
trifluoromethanesulfonic acid salt) had previously been
introduced by Lau as a water-soluble alkene hydrogenation
catalyst and used by us for the partial and total deoxygenation
of terminal diols to primary alcohols, alkenes, and alkanes in
aqueous acidic medium.19−21 A further interesting alternative
approach is the methyltrioxorhenium mediated deoxygenation
of polyols.22−27
While the use of homogeneous systems for the hydro-
deoxygenation of sugars and sugar derived substrates poses
substantial challenges in catalyst recovery and reuse, it may
offer advantages with respect to long-term catalyst stability,
activity, and selectivity as well as mass transport limitations.
Specifically, the operation of heterogeneous hydrogenation
catalysts based on a combination of a hydrogenating metal (e.g.,
Ru, Rh, Ni, Pd, Pt) with a typical support (e.g., Al2O3, SiO2) in
the necessarily aqueous acidic medium of hydrodeoxygenation
reactions can lead to rapid catalyst coking and fouling due to
polymerization and decomposition of the typically highly polar
sugar(-derived) substrates.28 In contrast, single metal-site
INTRODUCTION
■
In the last two decades enormous progress has been made in
the development of new catalysts and processes for the
conversion of biomass into chemicals and fuels by hydro-
deoxygenation, i.e., the iterative combination of acid-catalyzed
loss of water and metal catalyzed hydrogenations, which has
emerged as one of the main strategies toward reducing the
oxygen content of overfunctionalized biomass-derived sub-
strates. Starting from the seminal work by Descotes,1 advances
in these processes have mainly focused on the use of
heterogeneous catalyst systems,2−7 while attempts to use and
integrate homogeneous hydrodeoxygenation catalysts into
value chains originating from biomass-derived C5 and C6
molecules identified as platform molecules remain compara-
tively few.8−10 Examples are the hydrogenation of levulinic acid
(LA) to γ-valerolactone (GVL) using ruthenium or iridium
phosphine complexes,11,12 the conversion of LA to 1,4-
pentanediol and 2-methyltetrahydrofuran by an in situ
generated ruthenium phosphine complex combined with an
acidic ionic liquid,13,14 the principal demonstration of a reaction
cascade leading from sucrose to GVL, 1,4-pentanediol, 2-
methyl-THF, and alkanes,15 and the direct conversion of
fructose to 2,5-dimethyltetrahydrofuran by RhI3/HI with
successful catalyst recycling but at very high acid/sugar
substrate ratios.16,17 Ladipo et al. have reported the use of cis-
[Ru(6,6′-Cl2bpy)2(OH2)2](BArF)2 (6,6′-Cl2bpy = 6,6′-di-
Received: August 15, 2014
Revised: October 3, 2014
Published: October 6, 2014
© 2014 American Chemical Society
4116
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Chart 1. Structures of the Two Isoelectronic Ruthenium and Iridium Complexes Evaluated As HDO Catalysts
Scheme 1. Conceivable Modes of Hydrogen Activation by Catalysts 1 and 2
homogeneous catalysts are not susceptible to this type of
deactivation process but may instead suffer from substrate,
(side-) product or−for cationic complexes−counterion specific
irreversible coordinative inhibition.29
Cl2bpy)2(OH2)2](OTf)2, which in our hands decomposed to
Ru(0) by loss of ligand at temperatures >125 °C.21 In contrast
the tris-chelate complex 1 is stable in aqueous acidic medium
up to 250 °C, which we interpreted to be a direct consequence
of the enhanced chelate effect. This emphasizes that the key
role of the ligand in these systems operating at such elevated
temperatures is not the induction of reaction selectivity or
specificity but rather the stabilization of the metal center against
reduction to bulk metal which would then act as a
heterogeneous rather than homogeneous catalyst.
By definition homogeneous catalysts to be used in
hydrodeoxygenation reactions must be water- and acid-stable
and−for ease of use, handling−should also be air-stable, at least
in their pro-catalyst form (i.e., the species actually added to the
reaction mixture). Empirically, the acid-catalyzed loss of water
from polyalcohols and other biomass derived substrates and the
ring-opening of biomass-derived furans to diketones also
requires temperatures of ≥150 °C, which then constitutes the
lower limit of temperature stability for the catalysts used.30−32
Together these criteria constitute a remarkable challenge to the
design and synthesis of suitable catalysts that we chose to
address through the combination of ruthenium(II) centers
stabilized by air-stable nonphosphine containing chelate ligands
based on pyridine donors and one or several easily dissociated
labile ligands to allow for the activation of hydrogen gas and/or
the substrate.33 Postulating that cationic complexes containing
aquo ligands would both be stable and soluble in aqueous
medium while at the same time allowing heterolytic activation
of hydrogen gas by facile displacement of one or several aquo
ligands at a d6 transition metal center,34 this type of ruthenium
complexes was specifically targeted.21
Here we report the application of 1 and its iridium analogue
[(4′-Ph-terpy)Ir(OTf)3] (2) to the hydrodeoxygenation of 2,5-
hexanedione (2,5-HD) and 2,5-dimethylfuran (2,5-DMF). The
structures of the two (pro)-catalysts are shown in Chart 1. In
these complexes trifluoromethanesulfonate (OTf−; triflate) and
its corresponding acid (HOTf; triflic acid) were chosen as the
counterion and acid cocatalysts (see below), as HOTf is the
strongest known nonoxidizing acid and consequently its anionic
form one of the weakest anionic ligands resulting in minimal
coordinative inhibition of the catalysts by competition with also
only weakly coordinating H2(g) and substrates.36,37 On this
basis we postulated that the tris-triflate iridium complex 2
would in hot aqueous conditions be converted into the
corresponding aquo-complex 2′, i.e., generate a species
completely analogous to 1 as also shown in Chart 1.
Previously we reported that the complex [(4′-Ph-terpy)Ru-
(H2O)3](OTf)2 (terpy = 2,2′:6′,2″-terpyridine) (1) is an active
and very robust catalyst for the total deoxygenation of glycerol
to propene/propane in acidic sulfolane solution.35 Catalyst 1 is
a logical evolution of the bis-chelate complex cis-[Ru(6,6′-
The previously described complex 2 was targeted as a catalyst
on the basis of its isoelectronic relationship (RuII/d6 → IrIII/d6)
to complex 1 and hypothesizing that in oxidation state +3
slightly stronger metal−ligand bonds and therefore an even
more stable catalyst would result.38 For both catalysts an ionic
4117
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Scheme 2. Value Chain from Cellulose to Deoxygenated Value-Added Products via the Two Target Substrates 2,5-
Dimethylfuran and 2,5-Hexanedione
hydrogenation modus operandi is postulated,39 in which the base
taking up the proton generated by a heterolytic splitting of
dihydrogen would be the counterion, substrate, product or in
the solvent-leveled aqueous reaction mixture most likely water
resulting in the formation of H3O+ as the strongest possible
corresponding acid.35 For catalyst 2 we also considered the
possibility of an in situ reduction to IrI as shown in Scheme 1
with the generation of two equivalents of free acid. In this
scenario the activation of hydrogen would then occur via
oxidative addition of H2 to form the corresponding iridium
trihydride as an active species.
The choice of the targeted substrates is based on the value
chain cellulose → HMF → 2,5-dimethylfuran →2,5-hexanedione
→2,5-hexanediol →2,5-dimethyltetrahydrofuran and/or hexane
shown in Scheme 2. 2,5-DMF has been suggested as a possible
fuel, and routes to 2,5-DMF from glucose/fructose (mainly by
use of heterogeneous catalysts and ultimately to be derived
from cellulose) have been proposed by several authors.40−44
Further value could be added to this product by conversion
into 2,5-hexanediol (2,5-HDO) − a potential polyester or
polyurethane component or cross-linker in paint and resin
formulations−or 2,5-methyltetrahydrofuran (2,5-DMTHF) and
hexane (HEX) that could serve as biomass derived solvents or
high energy density fuels.16,17,44−46
deactivated polyethylene glycol) column. Quantification was
carried out using internal standard calibration against 100 mmol
L−1 dimethyl sulfone (DMS) in a three level calibration. GC-
MS analyses were performed on a Varian Saturn 2000 GC/MS
using a 30 m Rtx-1701 (14% cyanopropylphenyl/86% dimethyl
polysiloxane) column or a 30 m Stabilwax-da (acid-deactivated
polyethylene glycol) column running in CI mode. Reaction
products were identified by comparison to the retention times
and MS fragmentation patterns of authentic samples. Head-
space gas analyses were carried out on a SRI 8610 micro-GC
with a TCD detector against authentic gas samples (1000 ppm
of C1−C6 alkanes and C2−C6 alkenes in helium, GRACE
Davison Discovery Sciences). For those reactions where phase
separation of an organic product layer occurred the organic
layer was weighed, diluted with MeOH, and analyzed by GC.
No partitioning of the internal standard from the polar phase
into these phase separated layers was observed. Assuming equal
FID responses for all components of the organic layers (mainly
2-hexanone, 2,5-dimethyltetrahydrofuran, and hexane) they
were quantified by relative peak areas on the basis of their equal
carbon numbers (C6).47 Normalization to the molecular
weights of each component allowed calculation of the mass
of each component in the phase separated layer. These
amounts were then added to those quantified in the aqueous
phase, and the sum was reported as a percentage (by mole) of
the starting material.
All hydrogenation experiments employed industrial grade H2
gas (99.995%) and were carried out in an Autoclave Engineers
(AE) Mini-reactor with a 50 mL 316 stainless steel (316SS)
reactor vessel and impeller. At a total reaction solution volume
of 25 mL the reactor has a gas-phase headspace of 50 mL
(unused reactor body space and enclosed and pressurized
magnet-drive stirring assembly). Unless otherwise specified (cf.
control experiments) reactor vessel and impeller were
thoroughly cleaned and polished after every run by lathe at
600 rpm with 3 M Abrasive pads and sand-blasting,
respectively. Regular control reactions without addition of
catalyst showed only marginal conversion (<5%) of substrate to
hydrogenated products (see main text) caused by the
background activity of the reactor walls.48
With the goal of developing a homogeneous catalyst system
capable of the hydrodeoxygenation of the components of this
value chain to 2,5-DMTHF or ultimately hexane, we here
present a study of the application of the catalysts 1 and 2 to 2,5-
hexanedione and 2,5-dimethylfuran.
EXPERIMENTAL SECTION
■
General. All reagents and solvents were purchased from
readily available commercial sources and used as received
unless otherwise specified. 2,5-Hexanedione, 2,5-dimethylfuran,
and 1,4-dioxane were passed through a short plug of neutral
Al2O3 (Brockmann Activity I) immediately before use to
remove any peroxides or stabilizers present. All water used was
HPLC grade. Trifluoromethanesulfonic (triflic) acid was stored
under argon atmosphere in a Rotaflo Schlenk tube sealed with a
Teflon stopcock and dispensed into reaction mixture using a
microliter syringe. Catalyst syntheses were performed under
argon atmosphere using standard Schlenk-line techniques. All
NMR spectra were obtained on 300, 400, or 600 MHz
spectrometers and calibrated to the residual protonated solvent
signal. GC analyses were performed on a Varian 3800 with FID
detectors using a 30 m Rtx-1701 (14% cyanopropylphenyl/86%
dimethyl polysiloxane) column or a 30 m Stabilwax-da (acid-
4′-Phenyl-2,2′:6′,2″-terpyridine was prepared following the
solvent-less procedure developed by Cave and Raston.49 In our
hands this protocol consistently generated a product containing
an orange contaminant of unknown structure and without a
discernible NMR signature. Passing a CHCl3 solution of this
crude orange product through a short plug of basic Al2O3
(Brockman Activity I) efficiently removes this impurity
rendering a snow-white product in 67% overall yield after
4118
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Table 1. Hydrogenation of 2,5-Hexanedione by Catalysts 1 and 2 in Water
b
b
b
b
b
b
b
cat. decomp.
[Y/N]
2,5-HD
[%]
2,5-HDO
[%]
2,5-DMTHF
[%]
HMA
[%]
HXO
[%]
HEX
[%]
MBD
[%]
phase
sep.
a
entry cat. T [°C]
1
2
3
4
5
6
1
1
1
1
2
c
150
175
200
225
200
225
N
8
0
45
69
14
0
17
25
80
78
80
0
6
1
3
2
0
2
0
0
0
2
0
0
0
0
0
3
0
0
24
4
N
N
N
Y
N
N
2
1
N
6
9
N
0
13
1
7
N
N
n/a
60
37
a
Reaction conditions: 2,5-hexanedione [1000 mmol/L] in water, 5.5 MPa (800 psi) H2(g), dimethylsulfone (ISTD) [100 mmol/L], catalyst load [1
b
mmol/L= 0.1 mol % w.r.t substrate], reaction time = 16 h. By quant. GC; 1%; 2,5-HD = 2,5 hexanedione; 2,5-HDO = 2,5-hexanediol; 2,5-
DMTHF = 2,5-dimethyltetrahydrofuran (sum of cis and trans isomers); HMA = hemiacetal = 2-hydroxy-2,5-dimethyl-tetrahydrofuran; HXO = 2-
hexanone; HEX = hexane; MBD = mass balance deficiency: gas phase products and substrate decomposition to solids not quantifiable by GC.
c
Control reaction without catalyst added.
Scheme 3. Overall Reaction Cascades and Observed Reaction Products for the Hydrodeoxygenation of 2,5-Hexanedione and
2,5-Dimethylfuran in Aqueous Acidic Medium
removal of solvent in vacuo. [(4′-Ph-terpy)Ru(H2O)3](OTf)2
(terpy =2,2′:6′,2″-terpyridine) (1) was prepared as previously
described (the final purple solid is air-stable for extended
periods, i.e., hours or days, but should be stored under inert
atmosphere for the longer term).35 [(4′-Ph-terpy)Ir(OTf)3]
(2) was prepared according to the method developed by Lo et
al.38
Single Crystal X-ray Diffraction Analyses. Crystals were
taken directly from the samples recovered from the reactor and
studied at 150 K. Ruthenium complex (CCDC deposition
number: 1018797): dark-red truncated prism; [RuL2]-
(triflate)2*(H2O)*0.5(trans-2,5-diMeTHF), where L = 4′-
phenyl-2,2′:6,2″-terpyridine; FW = 1086.02; triclinic, P-1; a =
12.4822(2), b = 12.5899(2), c = 16.3218(2) Å, α =
85.1949(13)o, β = 73.0090(14)o, γ = 67.1062(17) o, V =
2258.47(7) Å3; Z = 2; R1 = 0.038; GOF = 1.114. Mixed FeNiIr
crystal I (CCDC deposition number: 1018795): dark-red
prism; [ML2](triflate)2*(H2O)*0.5(trans-2,5-diMeTHF),
where M = Fe:Ni:Ir with 61.29(8):26.27(4):12.4(1) %,
respectively; FW = 1058.52; triclinic, P-1; a = 12.5227(3), b
= 12.5493(4), c = 16.3000(4) Å, α = 85.591(2)o, β =
o
73.366(2)o, γ = 67.274(3) , V = 2262.17(12) Å3; Z = 2; R1 =
0.052; GOF = 1.053. Mixed FeNiIr crystal II (CCDC
deposition number: 1018796): dark-red prism; [ML2]-
(triflate)2*(H2O)*0.5(trans-2,5-diMeTHF), where M = Fe:-
Ni:Ir with 65.46(8):28.06(3):6.5(1) %, respectively; FW =
1050.45; triclinic, P-1; a = 12.5186(5), b = 12.5124(4), c =
16.2959(5) Å, α = 85.669(2)o, β = 73.376(3)o, γ = 67.447(3) o,
V = 2257.13(15) Å3; Z = 2; R1 = 0.055; GOF = 1.121.
Representative Procedure for a Catalytic Hydro-
deoxygenation Reaction. In a 25 mL volumetric flask
were combined catalyst 1 (0.0197 g, 0.025 mmol) and ∼10 mL
of HPLC water to form a dark purple solution. Methylsulfone
(0.2333 g, 2.5 mmol, GC internal standard), 2,5-hexanedione
4119
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Table 2. Hydrogenation of 2,5-Hexanedione by 1 and 2 in the Presence of Added Acid
2,5-
HD
[%]
2,5-
HDO
[%]
2,5-
DMTHF
[%]
b
b
b
b
b
b
b
b
T
acid [equiv
w.r.t. Ru]
SA
HMA
[%]
HXO
[%]
HEX
[%]
MBD
[%]
phase
sep.
a
entry cat. [°C]
acid
cat. decomp.[Y/N]
[%]
1
1
1
1
1
1
2
2
2
2
2
125
150
175
200
225
200
200
200
200
200
225
HOTf
HOTf
HOTf
HOTf
HOTf
HOTf
HOTf
HOTf
H3PO4
H3PO4
HOTf
10
10
10
10
10
10
5
N
N
N
N
N
82
2
4
1
0
0
0
0
1
0
1
0
10
37
41
33
7
3
7
1
0
0
0
0
0
0
0
0
3
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
35
46
39
53
52
16
43
6
N
N
Y
Y
Y
Y
Y
Y
Y
Y
N
2
15
8
3
1
1
4
7
9
12
3
c
5
24
12
0
d
6
Y
0
4
29
65
56
91
55
1
19
15
0
d
7
Y
0
d
8
1
Y
0
0
d
9
5
Y
0
0
3
d
10
11
1
Y
0
0
3
41
53
e
-
10
n/a
46
0
0
a
Reaction conditions: 2,5-hexanedione [1000 mmol/L] in water, 5.5 MPa (800 psi) H2(g), dimethylsulfone (ISTD) [100 mmol/L], catalyst load [1
b
mmol/L = 0.1 mol % w.r.t substrate], HOTf 10 mmol/L = 10 equiv w.r.t. Ru metal, reaction time = 16 h. By quant. GC; 1%; 2,5-HD = 2,5
hexanedione; 2,5-HDO = 2,5-hexanediol; 2,5-DMTHF = 2,5-dimethyltetrahydrofuran (sum of cis and trans isomers); SA = self-aldol products; HMA
= hemiacetal = 2-hydroxy-2,5-dimethyltetrahydrofuran; HXO = 2-hexanone; HEX = hexane; MBD = mass balance deficiency: gas phase products
c
d
and substrate decomposition to polymers and solids not quantifiable by GC. <1% of 2,5-DMF also formed. Black coating formed on reactor wall.
e
Control reaction without catalyst added.
(2.853 g, 25 mmol), or 2,5-dimethylfuran (2.4105 g, 25 mmol)
and (where applicable) triflic acid were added, and the volume
made up to 25.00 mL with water. The solution was thoroughly
mixed, then placed in an Autoclave Engineers MiniReactor, and
purged three times with hydrogen. The reactor was pressurized
to 5.5 MPa (800 psi) H2, sealed, and then heated to the set
temperature (125−238 °C), which in all cases was reached in
≤30 min. The reaction was stirred at temperature for 16 h and
then cooled to ambient. A sample of the headspace gas was
taken for micro-GC-TCD analysis before the reactor was
vented. The reaction products were analyzed by GC-FID.
Complex 1 is freely soluble in water forming initially purple
solutions that turn red after the reaction. Complex 2 is not
soluble in water at ambient temperature but forms pale yellow
solutions in 5:1 1,4-dioxane/water. In either water or 5:1 1,4-
dioxane purple-red solutions result after the reaction with 2.
With 2,5-dimethylfuran as the substrate initially biphasic
reaction mixtures result. The appearance of the final reaction
mixtures is discussed in the main text and indicated in the
tables listing the observed conversion and yields.
added acid cocatalyst a complex reaction network of iterative
dehydration, cyclizations, and hydrogenations evolves in which
2,5-dimethyltetrafuran (2,5-DMTHF) and hexane (HEX)
appear to be the terminal stable products. Small amounts of
the partial hydrodeoxygenation intermediates 2-hydroxy-2,5-
dimethyl-tetrahydrofuran (hemiacetal = HMA) and 2-hexanone
(HXO) as well as the self-aldol products (SA, left-most branch
in Scheme 3, ≪1%) were also observed in some of the
reactions.
The best result for catalyst 1 is achieved at 200 °C (Entry 3,
Table 1) realizing 94% of hydrogenated products 2,5-HDO
(14%) and 2,5-DMTHF (80%) with no catalyst decomposition
and an almost full mass balance (MBD ≤ 1%). Trace amounts
of hexane (HEX) were identified in a gas phase sample from
the head space of the reactor (by micro-GC). At the higher
temperature 225 °C (Entry 4, Table 1) phase separation of the
reaction mixture occurs into an aqueous polar and a small
amount of an organic phase containing 2,5-DMTHF, 2-
hexanone (HXO), and HEX as the total deoxygenation
product. The mass balance deficiency (MBD) for this reaction
(9%) and all others listed in the following is caused by partial
condensation into and decomposition of the substrate and
products into the reactor headspace and/or decomposition into
nonvolatile oligo- or polymeric products in solution and−where
applicable−partition of some of the hexane generated into the
gas phase.50 At the lower temperature of 175 °C a higher
selectivity to 2,5-HDO is observed at comparable conversion,
while at 150 °C substantial decomposition/polymerization of
the substrate competes with hydrogenation indicating very low
catalyst activity at this temperature. For comparison, a control
reaction without added catalyst at 225 °C (Entry 6, Table 1)
results in 37% decomposition to brown solids and 60%
recovery of the substrate with only marginal conversion to
hydrogenated products caused by the background activity of
the 316SS reactor body.48 Control reactions at lower
temperatures equally gave no conversion to hydrogenated
substrates but higher recoveries of unconverted substrate. In all
reactions the initially purple solution of 1 turns red, i.e., a
different color than the catalytically active yellow dimer [((4′-
phenyl-2,2′:6′,2″-terpy)Ru(CO))2(μ-OCH3)2](OTf)2 ob-
served when 1 was employed in the hydrodeoxygenation of
RESULTS
■
Catalytic Hydrogenation Reactions. Hydrodeoxygena-
tion of 2,5-Hexanedione by 1 and 2. Complexes 1 and 2 were
evaluated as catalysts for the hydrogenation of 2,5-hexanedione
(2,5-HD) exploring the influence of temperature and type and
amount of acid cocatalyst added on catalyst activity and product
distribution. Since 2,5-HD is freely soluble in water at the
substrate concentration selected (1000 mmol/L), all reactions
with this substrate were carried out in aqueous solution. Table
1 summarizes the results of experiments as a function of
temperature and catalyst employed but without the addition of
an acid cocatalyst. However, note that even in the absence of
added acid the reaction mixtures are still acidic due to the
generation of free HOTf by the heterolytic activation of
hydrogen (cf. Scheme 1). Scheme 3 shows the overall reaction
cascade observed with all pathways, inferred and actually
observed intermediates and products.
Under the aqueous conditions no acid-catalyzed ring closure
of the substrate to 2,5-DMF was observed in any of the
reactions starting from 2,5-HD, but even in the absence of
4120
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Table 3. Hydrogenation of 2,5-Dimethylfuran by Catalyst 1 in Water
2,5-
DMTHF
[%]
b
b
b
b
b
b
b
b
b
T
[°C]
HOTf [equiv
w.r.t. Ru]
cat. decomp. 2,5-DMF
2,5-HD
[%]
2,5-HDO
[%]
SA
HMA
[%]
HXO
[%]
HEX
[%]
MBD
[%]
phase
sep.
a
entry
(Y/N)
[%]
[%]
1
2
3
4
150
175
200
225
225
200
225
225
225
225
n/a
n/a
n/a
n/a
n/a
1
N
N
N
N
n/a
N
Y
47
0
0
22
3
7
19
12
0
4
14
55
48
0
0
2
0
0
0
0
0
0
0
0
0
8
4
3
0
5
3
0
2
0
0
0
0
3
0
0
8
6
0
0
0
0
0
2
0
0
4
0
0
0
41
34
27
32
48
40
28
49
43
9
Y
N
N
Y
0
0
12
50
47
16
40
49
83
c
5
2
0
N
N
Y
6
0
0
8
d
7
1
1
0
40
3
d
8
e
5
Y
1
0
Y
9
1
Y
0
0
6
N
N
c
10
1
n/a
2
0
6
a
Reaction conditions: 2,5-dimethylfuran [1000 mmol/L] in water, 5.5 MPa (800 psi) H2(g), dimethylsulfone (ISTD) [100 mmol/L], catalyst load [1
b
mmol/L = 0.1 mol % w.r.t. substrate], reaction time = 16 h. By quant. GC; 1%; 2,5-HD = 2,5 hexanedione; 2,5-HDO = 2,5-hexanediol; 2,5-
DMTHF = 2,5-dimethyltetrahydrofuran (sum of cis and trans isomers); SA = self-aldol products; HMA = hemiacetal = 2-hydroxy-2,5-
dimethyltetrahydrofuran; HXO = 2-hexanone; HEX = hexane; MBD = mass balance deficiency: gas phase products and substrate decomposition to
polymers and solids not quantifiable by GC. Control reactions without catalyst added. Catalyst decomposition to blue metallic coating in reactor
body observed. Control reaction with blue coating from reaction of Entry 7.
c
d
e
Table 4. Hydrogenation of 2,5-Dimethylfuran by Catalyst 2 in Water
b
b
b
b
b
b
T
acid [equiv w.r.t.
Ru]
cat. decomp.
(Y/N)
2,5-DMF
[%]
2,5-HD
[%]
2,5-HDO
[%]
2,5-DMTHF
[%]
HEX
[%]
MBD
[%]
phase
sep.
a
entry
[°C]
acid
n/a
1
2
3
4
5
175
200
225
175
225
225
225
225
225
225
n/a
n/a
n/a
1
N
Y
Y
Y
Y
Y
Y
Y
Y
Y
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
4
0
0
23
1
46
74
38
87
69
71
17
9
0
1
31
24
55
12
18
23
54
62
26
23
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
n/a
n/a
0
7
HOTf
HOTf
HOTf
HOTf
HOTf
H3PO4
H3PO4
0
2
1
0
13
7
c
6
1
0
7
5
0
29
25
13
31
8
10
1
0
9
0
61
47
10
5
0
a
Reaction conditions: 2,5-dimethylfuran [1000 mmol/L] in water, 5.5 MPa (800 psi) H2(g), dimethylsulfone (ISTD) [100 mmol/L], catalyst load [1
b
mmol/L = 0.1 mol % w.r.t. substrate], reaction time = 16 h. By quant. GC; 1%; 2,5-HD = 2,5 hexanedione; 2,5-HDO = 2,5-hexanediol; 2,5-
DMTHF = 2,5-dimethyltetrahydrofuran (sum of cis and trans isomers); HEX = hexane; MBD = mass balance deficiency: gas phase products and
c
substrate decomposition to polymers and solids not quantifiable by GC. Control reaction with black coating generated from the reaction of Entry 5.
glycerol.35 At 200 °C the iridium catalyst 2 achieves very similar
results (Entry 5, Table 1) resulting in a clear purple-red
solution and was not further evaluated under these conditions.
With reference to Scheme 3 and as observed for Entry 4,
Table 1, an iterative hydrodeoxygenation sequence can in
principle result in the total deoxygenation of the substrate to
hexane, which requires further acid-catalyzed dehydration of the
initial products of 2,5-HD hydrogenation. We therefore
evaluated both catalysts in the presence of added HOTf.
Table 2 summarizes the results of the 2,5-HD reactions with
added HOTf. For 1 with 10 equiv of acid added no catalyst
decomposition (as evidenced by the formation of Ru0 or a
metallic coating in the reactor − vide infra) was observed, and
depending on the temperature used the original purple (T ≤
150 °C) to orange-red (T ≥ 150 °C) homogeneous solutions
were recovered with some organic substrate decomposition to
brown solids (humins) observed at the higher reaction
temperatures. For T ≥ 175 °C phase separation of an organic
nonpolar layer from the aqueous polar reaction mixture was
observed. In the presence of acid overall lower conversions to
hydrodeoxygenated products result, and higher amounts of
acid-catalyzed decomposition to nonvolatile products, reaching
53% for T = 225 °C, are observed, which is the same amount of
decomposition as in the control reaction (Entry 11, Table 2) .
The presence of added acid also causes the stable 2,5-DMTHF,
most likely formed by acid-catalyzed dehydration ring-closure
of 2,5-HDO, to dominate the product distribution. At the
higher temperature traces of 2,5-DMF were also identified in
the reaction mixture, i.e., even in aqueous solvent the acid
catalyzed ring-closure of 2,5-HD to its cyclic bis-enol (i.e., the
furan ring) is possible. The best result for 1 is achieved at 200
°C (Entry 4, Table 2) generating 12% of hexane.51 Catalyst 2
was then evaluated at this temperature exploring the effect of
different amounts of acid (Entries 6−8, Table 2). A high acid
load (10 equiv as for 1) not only gives the highest yield of the
total deoxygenation product hexane but also results in excessive
substrate decomposition. A better balance of total deoxygena-
tion and overall conversion to hydrodeoyxgenated products is
achieved at 5 equiv of HOTf added (Entry 7). In order to
explore the possible effect of a weaker, noncoordinating and
nonoxidizing acid two reactions with H3PO4 were also carried
out giving a very high conversion and selectivity to 2,5-
DMTHF (91%, Entry 9, Table 2) and low substrate
−
decomposition. The buffering capacity of the H2PO4
⇔
2−
3−
HPO4 ⇔ PO4 system likely plays an important role here
and suggests that the reaction could be further optimized for
the highly selective generation of 2,5-DMTHF. However, in all
cases the reactions of 2 in the presence of acid generated a
4121
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Table 5. Hydrogenation of 2,5-Dimethylfuran by Catalyst 1 in 5/1 1,4-Dioxane/Water
2,5-
DMF
[%]
2,5-
HD
[%]
2,5-
HDO
[%]
2,5-
DMTHF
[%]
b
b
b
b
b
b
b
b
b
T
[°C]
time cat. load [% w.r.t. to acid/base [eq
SA
HMA
[%]
HXO
[%]
HEX
[%]
MBD
[%]
a
entry
[h]
substrate]
w.r.t. Ru]
[%]
1
2
175
200
225
225
225
225
225
225
16
16
16
16
16
16
16
40
16
16
16
0.1
0.1
n/a
15
6
35
44
46
41
19
2
6
0
0
0
0
0
1
0
0
0
0
14
17
11
9
7
5
3
1
0
0
0
1
2
1
1
0
0
0
0
0
0
21
27
30
38
35
30
38
19
24
41
13
n/a
3
0.1
n/a
9
3
0
0
4
0.1
5/HOTf
1/HOTf
5/NaOH
1/NaOH
n/a
8
1
3
1
5
0.1
3
26
2
2
10
0
5
6
0.1
64
5
1
0
7
0.1
31
0
12
78
71
44
0
10
0
1
0
8
0.1
0
0
3
c
9
238
0.1
n/a
0
1
0
0
4
10
225
225
0.25
n/a
n/a
0
1
0
3
11
0
d
11
n/a
83
3
0
0
a
Reaction conditions: 2,5-dimethylfuran [1000 mmol/L] in 5:1 1,4-dioxane:water, 5.5 MPa (800 psi) H2(g), dimethylsulfone (ISTD) [100 mmol/L],
b
catalyst load [1 mmol/L = 0.1 mol % w.r.t. substrate], reaction time = 16 h. By quant. GC; 1%; 2,5-HD = 2,5 hexanedione; 2,5-HDO = 2,5-
hexanediol; 2,5-DMTHF = 2,5-dimethyltetrahydrofuran (sum of cis and trans isomers); SA = self-aldol products; HMA = hemiacetal = 2-hydroxy-
2,5-dimethyltetrahydrofuran; HXO = 2-hexanone; HEX = hexane; MBD = mass balance deficiency: gas phase products and substrate decomposition
c
d
to polymers and solids not quantifiable by GC. Maximum sustained internal reactor temperature attainable by the equipment used. Control
reaction without catalyst.
black coating on the reactor walls along with clear purple
solutions. The nature and activity of this coating and of the
purple species in solution was further investigated in detail with
2,5-DMF as the substrate (vide infra) ultimately suggesting that
all reactions at T ≥ 200 °C and in the presence of acid with 2 as
the starting catalyst are in fact heterogeneously catalyzed by this
coating likely composed of Ir0.
inhibition and very high levels of substrate decomposition the
amount of acid added was lowered to 5 and ultimately 1 equiv
w.r.t. ruthenium content, which at 200 and 225 °C (Entries 6
and 7, Table 3) gave 8 and 40% of hydrodeoxygenated
products, i.e., again lower conversion to desired products and/
or higher decomposition of substrate than without acid added
(Entries 3 and 4, Table 3). For the reactions at 225 °C in the
presence of acid a blue coating with a metallic sheen was
observed along with the clear red reaction solutions obtained at
the end of the reaction, indicating at least partial catalyst
decomposition. However, in a control reaction without cleaning
the reactor this coating generated from 1 proved to be
essentially inactive giving similar amounts of hydrodeoxygen-
ated products as the control reaction without catalyst in a clean
reactor (Entry 10, Table 3). The control reaction (Entry 11,
Table 3) also clearly established that 2,5-DMF hydrolyzes to
2,5-HD under the reaction conditions as we also observed in a
previous study employing HMF derived substrates with longer
alkyl side-chains.31 Control reactions at 225 °C with 2,5-
DMTHF as the substrate with or without acid added showed
no conversion to ring-opened products or further deoxygena-
tion, i.e., once formed 2,5-DMTHF is stable under even the
harshest reaction conditions tested here.
Table 4 lists the results for the reaction carried out with
catalyst 2. With the exception of the first reaction at 175 °C
without added acid, which realized at total of 69% of
hydrodeoxygenated products, catalyst decomposition to a
black coating, very similar in appearance to that seen in the
reactions with 2,5-HD, was observed in all cases. In contrast to
the coating generated from 1, the coating from the iridium
system 2 is very active as a catalyst, achieving essentially the
same conversion and product distribution as the homogeneous
catalyst, i.e., the observed activities are likely heterogeneous in
nature based on the formation of an Ir0 species on the reactor
wall surfaces (Entries 5 and 6, Table 4). The coating maintains
∼50% of its activity even after twice charging the reactor with
pure water only and heating to 225 °C for 16 h but is easily
removed by mechanical polishing returning the reactor to its
baseline activity (cf. control reaction, Entry 10, Table 3). In the
presence of acids relatively high conversions to the hexane (up
Hydrogenation of 2,5-Dimethylfuran. Taking “one step
back” in the value chain illustrated in Scheme 2 the more
challenging hydrodeoxygenation of 2,5-DMF was attempted as
a function of temperature, solvent, and concentration of acid
cocatalyst added. While 2,5-DMF is only marginally soluble in
water at room temperature, both the relative permittivity (=
dielectric constant ε0) and surface tension σ [mN/m] of water
150
drop substantially with rising temperature (ε020 = 80.36 → ε0
= 32.32 and σ20 = 72.74 → σ150 = 48.75, respectively),52 which
led us to postulate that the solubility of this substrate would be
much higher at the temperatures required for catalyst activity
and ring-opening of the substrate (≥150 °C) and that a
homogeneous single-phase reaction solution be present at
elevated temperatures inside the reactor.31 It is also important
to realize that at temperatures 150−250 °C the pKw of water
drops from 14.00 at 25 °C to 11.64−11.20, i.e., the relative
concentration of hydronium H3O+ (and hydroxide OH−) is
approximately 3 orders of magnitude higher than at room
temperature providing enhanced intrinsic acid/base-catalysis
for the interconversion of 2,5-DMF and 2,5-HD (cf. Scheme 3)
even without the addition of an acid cocatalyst above and
beyond any acidic species generated by the heterolytic
activation of H2(g) at the metal center (Scheme 1). A first
series of experiments with 2,5-DMF as the substrate was
therefore carried out in water as the reaction medium. Tables 3
and 4 list the results for catalysts 1 and 2, respectively, with and
without the addition of HOTf.
The results listed in Table 3 show that the hydro-
deoxygenation of 2,5-DMF in water is in fact possible. The
best results for 1 without added acid were achieved at 200 and
225 °C (Entries 3 and 4, Table 3), showing a small amount of
hexane as the phase separated total deoxygenation product
without catalyst decomposition occurring. As reactions with 10
equiv of added HOTf (cf. Table 2) led to complete catalyst
4122
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Table 6. Hydrogenation of 2,5-Dimethylfuran by Catalyst 2 in 5/1 Azeotrope 1,4-Dioxane/Water or 9/1 Sulfolane/Water
2,5-
DMTHF
[%]
b
b
b
b
b
b
b
b
b
T
[°C]
cat. decomp. 2,5-DMF
2,5-HD
[%]
2,5-HDO
[%]
SA
HMA
[%]
HXO
[%]
HEX
[%]
MBD
[%]
phase
sep.
a
entry
1
solvent
(Y/N)
[%]
[%]
175 1,4-diox/
H2O
N
49
22
44
32
59
37
43
1
2
0
1
1
1
2
1
1
2
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
24
25
36
18
48
35
N
N
N
N
N
N
2
3
4
5
6
200 1,4-diox/
H2O
Y
Y
12
7
7
9
8
0
0
0
225 1,4-diox/
H2O
16
16
9
175 sulfolane/
H2O
N
Y
6
200 sulfolane/
H2O
5
225 sulfolane/
H2O
Y
14
6
a
Reaction conditions: 2,5-dimethylfuran [1000 mmol/L] in the solvent mixture indicated. 5.5 MPa (800 psi) H2(g), dimethylsulfone (ISTD) [100
b
mmol/L], catalyst load [1 mmol/L = 0.1 mol % w.r.t. substrate], reaction time = 16 h. By quant. GC; 1%; 2,5-HD = 2,5 hexanedione; 2,5-HDO
= 2,5-hexanediol; 2,5-DMTHF = 2,5-dimethyltetrahydrofuran (sum of cis and trans isomers); SA = self-aldol products; HMA = hemiacetal =2-
hydroxy-2,5-dimethyltetrahydrofuran; HXO = 2-hexanone; HEX = hexane; MBD = mass balance deficiency: gas phase products and substrate
decomposition to polymers and solids not quantifiable by GC.
to 29%, Entry 7, Table 4) were observed, however, also
accompanied by high levels of substrate decomposition.
decomposition products not quantifiable by GC being formed
reversibly and converted to volatile hydrodeoxygenation
products over time. The large amounts of 2,5-HD and the
presence of the self-aldol products in the reactions of Entries
1−4 suggest that reversibly formed aldol condensates (dimers
etc.) of 2,5-HD constitute a substrate reservoir. At a higher
catalyst load (Entry 10, Table 5) not only a moderate yield of
2,5-DMTHF but also a high amount of decomposition is
observed, which we attribute to the higher amount of acid
released by the catalyst in this case. The corresponding results
for catalyst 2 are summarized in Table 6, which also lists the
results of a temperature series in a 9:1 sulfolane/water mixture.
Again catalyst 2 decomposed at temperatures >175 °C, and in
both solvent systems only marginal conversions to hydro-
deoxygenated products were observed indicating coordinative
inhibition by either solvent.
Attempts at Catalyst Recycling and Catalyst Decom-
position Pathways. On the basis of the results discussed that
showed that catalyst 2 is not stable at T > 175 °C no attempts
were made to recycle this catalyst. For catalyst 1 the overall best
result was obtained by the reaction in water at 200 °C without
addition of acid cocatalyst (Table 3, entry 3), which gave a total
yield of 12 + 55 = 67% of the desired hydrodeoxygenation
products 2,5-HDO and 2,5-DMTHF, respectively. Attempting
to reuse the catalyst, fresh 2,5-DMF was added to the clear
deep red single-phase solution obtained from repeats of this
reaction. However, resubjecting this solution to the same
reaction conditions gave only 3% conversion to 2,5-DMTHF
(≈ reactor baseline activity) with 76% recovery of ring-opened
substrate (2,5-HD) along with some substrate decomposition
(18%), i.e., the recovered catalyst/product solution is
completely inactive. Postulating that the red solution represents
an irreversibly inhibited catalyst resting state, generated, e.g., by
coordination of one of the intermediate or side-products listed
in Scheme 3, we hypothesized that offering the catalyst system
an alternative stable but reactivatable resting state may change
this behavior. We therefore repeated the experiment in the
presence of propanal (5 mol %) as an additional substrate and
source of CO ligands, essentially recreating a reaction mixture
that resembles more closely that with glycerol as the
substrate,35 where the dimer [((4′-phenyl-2,2′:6′,2″-terpy)Ru-
(CO))2(μ-OCH3)2](OTf)2 was identified as a reversibly
Recognizing that both the catalyst and 2,5-DMF substrate
decomposition (up 50%) in the aqueous reaction mixtures may
be rooted in the fact that the system starts out as a biphasic
aqueous/organic mixture, a series of reactions at 175−225 °C
with and without added HOTf (1 and 5 equiv) was also
attempted in a 5:1 mixture of 1,4-dioxane:water representing
the azeotropic mixture of these two solvents. This mixture
completely dissolves 2,5-DMF at the substrate concentration
employed and ambient conditions. 1,4-Dioxane was selected as
it was anticipated to be stable under the reaction conditions and
chemically resembles the stable product 2,5-DMTHF (cf.
control reaction with this substrate discussed above) with the
azeotrope having a sufficiently different boiling point from 2,5-
HDO, 2,5-DMTHF, and HEX to potentially allow separation
by distillation. Table 5 gives the results of the reactions for
catalyst 1 in the azeotrope solvent mixture as a function of
temperature, reaction time, catalyst load, and cocatalyst,
which−postulating that base catalysis could also lead to ring
opening and hydrogenation−in this series was expanded to
include addition of NaOH at 1 and 5 equiv w.r.t to catalyst. For
1 none of the reactions listed in Table 5 showed catalyst
decomposition to solids or phase separation giving deep red
homogeneous solutions in all cases.
Under otherwise identical reaction conditions overall lower
conversions to hydrodeoxygenated products than in water were
observed, while the amount of decomposition−if somewhat
lower than in water−was still appreciable. E.g., at 200 °C
without added acid the reaction in the dioxane mixture
generated only 17% of 2,5-DMTHF with 27% decomposition
(Entry 2, Table 5) compared to 55% and 34%, respectively in
water (Entry 3, Table 3). At a longer reaction time of 40 h at
225 °C (Entry 8, Table 5) 78% of 2,5-DMTHF with 19%
decomposition are realized compared to only 11% and 30%,
respectively after 16 h (Entry 3, Table 5). Similar results were
obtained by running the reaction at 238 °C (Entry 9, Table 5),
the highest temperature reachable by the equipment currently
available to us. These results can be rationalized by 1,4-dioxane
causing coordinative inhibition of the catalyst, possibly by
acting as a reversibly binding weak chelate from a boat
conformation of the solvent and/or by some of the nonvolatile
4123
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Figure 1. ESI-MS of [Ru(4′-Ph-terpy)2]2+ (MeOH solution, +-ve ionization mode).
Figure 2. ORTEP drawing of the complex cation in the crystal structure of [Ru(4′-phenyl-2,2′:6,2″-terpyridine)2](OTf)2·(H2O)·0.5(trans-2,5-
DMTHF). Thermal ellipsoids are drawn at the 50% probability level.
formed resting state of the catalyst formed at T > 200 °C that
can serve as a pro-catalyst equivalent to 1. However, these
reactions again resulted in deep red solutions with no catalytic
activity.
identity of the material as the bis-chelate complex [Ru(4′-Ph-
terpy)2](OTF)2. Figure 1 shows the ESI-MS and Figure 2 the
ORTEP of this bis-chelate complex, which−with different
counterions and solvent inclusions−has previously been
described by Constable and Housecroft as well as McMurtrie
and Dance.53,54
The analogous analysis of a purple solution of redissolved
crystals (in MeOH or acetone) obtained from a reaction
solution of 2 by ESI-MS gave a dominant peak at m/z = 337.08.
This species does on the basis of both its mass charge ratio and
the absence of an iridium isotope pattern not contain iridium.
Instead, we successfully modeled this complex as a 2.3:1
mixture of iron and nickel by overlaying the known isotope
patterns of these two metals as shown in Figure 3.55 Again the
Insight into the catalyst decomposition pathway for both 1
and 2 was gained by the observation that red or red-purple
crystals, respectively, deposited from recovered aqueous
reaction solutions of both catalyst systems after being stored
under ambient conditions for several weeks. The ESI-MS
spectrum of a crystal (redisolved in MeOH) deposited from a
reaction solution of 1 showed a dominant peak with a distinct
ruthenium isotope pattern at m/z = 360.08. The same crop of
crystals also allowed a structure determination by single-crystal
X-ray diffraction. Together these unambiguously established the
4124
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Figure 3. ESI-MS of [(Fe)0.7(Ni)0.3(4′-Ph-terpy)2]2+ (MeOH solution, +-ve ionization mode).
same crop of crystals also allowed a structure determination by
single-crystal X-ray diffraction from two different crystals,
which−in agreement with the ESI-MS spectrum−showed these
crystals to be a blend of the complexes [Fe(4′-Ph-terpy)2]-
(OTf)2 and [Ni(4′-Ph-terpy)2](OTf)2, isostructural to those of
1 and to each other. However, the presence of relatively large
amounts of unaccounted electron density required fixing the
Fe:Ni ratio at 2.3:1 on the basis of the MS results (Figure 3)
and incorporation of smaller proportions of Ir for optimal
refinement, giving final crystal compositions of Fe, 61.29(8)%/
Ni, 26.27(4)%/Ir, 12.4(1)% and Fe, 65.46(8)%/Ni,
28.06(3)%/Ir, 6.5(1)%, respectively.56
For both catalysts this established the ligand loss and
redistribution deactivation/decomposition pathways shown in
Scheme 4, which for ruthenium forms an inactive and for
iridium a catalytically active coating in the reactor body. With 1
the bis-chelate complex [Ru(4′-Ph-terpy)2]2+ is formed by a
ligand redistribution reaction generating the blue inactive
coating in the reactor from ∼50% of the original metal content.
With 2 as the starting complex the black catalytically very active
coating, presumably composed of Ir0 is generated releasing free
ligand into solution, which then captures FeII/NiII (and possibly
small amout of MnII) ions leached into the reaction solution
under the reducing aqueous-acidic conditions or directly
scavenges metals from the 316SS of the reactor body, which
in either case is the only source of iron and nickel in the
reaction mixture.
DISCUSSION
■
Hydrodeoxygenation Reaction Cascade. The complex-
ity of the overall hydrodeoyxgenation reaction cascade of the
2,5-DMF/2,5-HD system (Scheme 3) is reflected in the
complex product distributions observed, which are further
complicated by acid-catalyzed substrate decomposition reac-
tions that compete with the desired hydrodeoxygenation
pathways. However, the product distribution data and the
results of the control reactions without catalysts and for 2,5-
DMTHF as the substrate allow some interpretation of the
actually operating pathways as a function of substrate and
reaction conditions. For 2,5-HD as the substrate a pathway via
ring-closure to 2,5-DMF followed by hydrogenation of the
furan ring to 2,5-DMTHF appears unlikely as only trace
amounts of 2,5-DMF were observed in these reactions. The
observation of ring-opening of 2,5-DMF to 2,5-HD in the
presence of only acid or both acid and metal catalyst (with the
catalyst also generating some acid by heterolytic H2 activation−
cf. Scheme 2) supports the notion that the actual initial
hydrogenation substrate for 2,5-DMF as the starting point is
likely also 2,5-HD. However, a direct furan ring hydrogenation
cannot be excluded in this case. Other homogeneously
catalyzed hydrogenations of substituted furans are well
established and have been shown to proceed at much lower
temperatures (albeit at typically high catalyst loads asking for
<100 turnovers) under basic conditions and can be highly
enantioselective, precluding a ring-opening as the first step.57,58
On the basis of our earlier study on the hydrodeoxygenation
of HMF-acetone condensate 2-hydroxymethyl-5-butan-3′-one-
furan to nonane by the heterogeneous Pd0/C system in glacial
HOAc,31 the results of the control reaction with 2,5-DMTHF
For all three metals the bis-chelate complexes [M(4′-Ph-
terpy)2]2+; M = Ru, Fe, Ni, once formed are stable and inactive
as a catalyst on the basis of the failed recycling experiments.
4125
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
Scheme 4. Catalyst Decomposition Pathways via Ligand Loss and Redistribution
Scheme 5. Possible Pathway for the Decomposition of Biomass Derived Furans to Humin via Aldol Condensations and Diels-
Alder Reactions
that gave no substrate conversion at all and the fact that 2-
hexanone (HXO) is only observed if hexane (HEX) is also
formed suggest that for the open chain products 2,5-HD must
be the initial hydrogen acceptor, as the formation of the very
stable THF ring−while in principle hydrolyzable to the diol
(2,5-HDO)59 − is a “dead end” under the reaction conditions
employed. As a trend, higher acid concentrations result in a
higher degree of substrate loss due to formation of nonvolatile
dimers, oligomers, and ultimately polymers (humins) as an acid
catalyzed competition reaction to hydrodeoxygenation. As the
hydrogenation of CC and CO bonds in the substrates and
any of the intermediates lowers the overall reactivity of any
intermediate and products toward humin formation, achieving
better mass balances and higher conversion to the stable
products 2,5-DMTHF and/or hexane will require a more active
hydrogenation catalyst competing more effectively with acid
catalyzed humin formation, which is believed to mainly occur
through aldol condensation reactions.60 In addition to the aldol
pathway, we propose that−in particular under acidic conditions
and high temperature−a Diels−Alder pathway starting by the
dimerization of furans followed by ring-opening and aromatiza-
tion to phenolic species as outlined in Scheme 5 may also play a
4126
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
role in humin formation. Such a pathway would also be
applicable to furfural, HMF, and other biomass derived furans.
Acknowledging the relatively low solubility of hydrogen gas
in water and postulating that either hydrogen activation or
hydride transfer from the metal center to substrate are part of
the rate-determining step in the hydrogenation reactions, a
better hydrodeoxygenation:decomposition ratio may also be
achieved by carrying out the reactions at (much) higher
hydrogen pressures,61 which does however pose substantial
technical and economic challenges.62
Table 4), while 1 at best achieves at best 78% (at extended
reaction times: Entry 8, Table 5) and 11% (at increased catalyst
load: Entry 10, Table 5), respectively.
The formation of even small amounts of hexane by the
homogeneous system 1 is however intriguing, as it leads to
phase separation of the nonpolar product from an aqueous or
other highly polar reaction medium and thus offers in principle
a pathway to recycling a homogeneous catalyst system based on
water-soluble cationic complexes applicable to biomass hydro-
deoxygenation. At the temperatures required for the
dehydration and ring-opening reactions in a value chain leading
from C6 biomass-derived substrates to the desired deoxy-
genated solvent and fuel products, the present system is
however still not sufficiently stable against decomposition to
catalytically inactive species and−at the hydrogen pressures
tested−also lacks sufficient overall activity to prevent
substantial substrate decomposition by the competing humin
forming side reactions. Catalysts that are both more stable
against deactivation and overall more active may be accessible
by a logical extension of the design sequence bipy →
terpyridine to pyridine-based tetradentate ligands, possibly
including coordinated or pendant amines that could act as
proton shuttles in a metal−ligand bifunctional heterolytic
activation of hydrogen gas under aqueous acidic conditions,63
with the ultimate goal to be able to directly convert sugars and/
or sugar alcohols to alkanes in a single reactor with recycling
and reuse of the homogeneous catalyst enabled by phase
separation of the reaction solution from the nonpolar products.
Catalysts and Catalyst Deactivation. The two isostruc-
tural and isoelectronic catalysts tested combine a very open
structure with three potential coordination sites for the
activation of hydrogen and/or substrate with a strongly
chelating meridional binding pocket, which does however still
not sufficiently stabilize the catalysts. For 2 decomposition may
also occur via a planar complex formed in situ by reduction to
IrI and shown at the bottom right of Scheme 2. While the open
structure is in principle desirable for an effective nonspecific
and nonselective catalysis as targeted here, it also makes the
metal center more susceptible to coordinative inhibition, either
by the counterion of excess acid added, by coordination of two
ligands forming the inert bis-chelate complexes discussed
above, or by irreversible coordination of substrate or product
condensates−in particular as chelating dimers, oligomer, or
polymers, i.e., the humins or humin precursors or components
referred to above (Scheme 5). In a reverse argument, the
stability and inactivity of the bis-chelate complexes [M(4′-Ph-
terpy)2]2+ (M = Ru, Fe, Ni) is then a direct consequence of
their lack of a free coordination site required for any kind of
catalytic activity. The fact that the addition of excess acid
appears to make both catalysts more susceptible to
decomposition (Scheme 4) also suggests that the formation
of the bis-chelate complexes may be acid-mediated, possibly
starting with decoordination and protonation of one of the
terminal pyridine-arms of the 4′-Ph-terpy ligand.
ASSOCIATED CONTENT
■
S
* Supporting Information
Details of the single crystal X-ray structure determination. This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*Phone: 519-824-4120 x 53002. E-mail: mschlaf@uoguelph.ca.
CONCLUSION AND OUTLOOK
■
The complex [(4′-Ph-terpy)Ru(H2O)3](OTf)2 (1) is a
homogeneous catalyst for the hydrodeoxygenation of 2,5-
hexanedione and 2,5-dimethylfuran in aqueous acidic medium
at temperatures up to 225 °C but eventually decomposes to the
inactive bis-chelate complex [(4′-Ph-terpy)2Ru]2+ and a
metallic but catalytically inactive ruthenium coating, making
the system nonrecyclable. The analogous iridium system [(4′-
Ph-terpy)Ir(H2O)3](OTf)3 (2′), postulated to form in situ
from the tris-triflate precursor [(4′-Ph-terpy)Ir(OTf)3] (2),
decomposes at T > 175 °C, but in contrast to 1 forms−in
particular in the presence of added acid−a catalytically highly
active coating, i.e., in this case the complex merely serves as a
precursor to a heterogeneous catalyst effectively defeating the
purpose and premise of a homogeneous precursor. The 4′-Ph-
terpy ligand liberated by the decomposition to Ir0 reacts with
iron and nickel from the 316SS reactor body to form a mixture
of the stable and catalytically inactive bis-chelate complexes
[M(4′-Ph-terpy)2]n+; M = Fe, Ni. The homogeneous system is
inhibited by both the addition of acid or the use of a
coordinating solvent (1,4-dioxane, sulfolane) and overall the
activity of the coating formed by 2 is higher than that of 1 in
homogeneous phase. Starting from 2,5-dimethylfuran the
heterogeneous catalyst formed from 2 realizes−depending on
reaction conditions−up to 87% conversion to 2,5-dimethylte-
trahydrofuran (Entry 4, Table 4) or 29% of hexane (Entry 7,
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank the Natural Science and Engineering
Council (NSERC) Canada, the Bioeconomy Program of the
Ontario Ministry for Agriculture, Food and Rural Affairs
(OMAFRA), and the Los Alamos National Laboratory (LANL)
Laboratory Directed Research and Development (LDRD)
program for supporting this research. D.V.S. thanks the
Canadian Foundation for Innovation (CFI), the Ontario
Ministry for Research and Innovation (MRI), and the
University of Guelph for funding for the X-ray diffraction
facility.
REFERENCES
■
(1) Schiavo, V.; Descotes, G.; Mentech, J. Bull. Soc. Chim. Fr. 1991,
128, 704−711.
(2) Huber, G. W.; Iborra, S.; Corma, A. Chem. Rev. 2006, 106, 4044−
4098.
(3) Chheda, J. N.; Huber, G. W.; Dumesic, J. A. Angew. Chem., Int.
Ed. 2007, 46, 7164−7183.
(4) Corma, A.; Iborra, S.; Velty, A. Chem. Rev. 2007, 107, 2411−
2502.
4127
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128

ACS Catalysis
Research Article
(5) Alonso, D. M.; Bond, J. Q.; Dumesic, J. A. Green Chem. 2010, 12,
(37) Howells, R. D.; Crown, J. D. M. Chem. Rev. 1977, 77, 69−92.
(38) Lo, K. K.-W.; Chung, C.-K.; Ng, D. C.-M.; Zhu, N. New J. Chem.
2002, 26, 81−88.
1493−1513.
(6) Besson, M.; Gallezot, P.; Pinel, C. Chem. Rev. 2014, 114, 1827−
1870.
(39) Bullock, R. M. Chem.Eur. J. 2004, 10, 2366−2374.
(40) Roman-Leshkov, Y.; Barrett, C. J.; Liu, Z. Y.; Dumesic, J. A.
Nature 2007, 447, 982−985.
(7) Climent, M. J.; Corma, A.; Iborra, S. Green Chem. 2014, 16, 516−
547.
(8) White Paper: Top Value Added Chemicals from Biomass; Werpy,
T., Peterson, G., PNNL/NREL, Eds.; U.S. Department of Energy:
2004.
(41) Jae, J.; Zheng, W. Q.; Lobo, R. F.; Vlachos, D. G. ChemSusChem
2013, 6, 1158−1162.
(42) Huang, Y. B.; Chen, M. Y.; Yan, L.; Guo, Q. X.; Fu, Y.
ChemSusChem 2014, 7, 1068−1072.
(9) Deuss, P. J.; Barta, K.; de Vries, J. G. Catal. Sci. Technol. 2014, 4,
(43) Chidambaram, M.; Bell, A. T. Green Chem. 2010, 12, 1253−
1262.
1174−1196.
(10) Schlaf, M. J. Chem. Soc., Dalton Trans. 2006, 4645−4653.
(11) Delhomme, C.; Schaper, L. A.; Zhang-Presse, M.; Raudaschl-
(44) Thananatthanachon, T.; Rauchfuss, T. B. Angew. Chem., Int. Ed.
2010, 49, 6616−6618.
Sieber, G.; Weuster-Botz, D.; Kuhn, F. E. J. Organomet. Chem. 2013,
̈
(45) Pittman, C. U., Jr.; Liang, Y. F. J. Org. Chem. 1980, 45, 5048−
724, 297−299.
5052.
(12) Li, W.; Xie, J.-H.; Lin, H.; Zhou, Q.-L. Green Chem. 2012, 14,
(46) Kotkar, D.; Mahajan, S. W.; Mandal, A. K.; Ghosh, P. K. J. Chem.
Soc., Perkin Trans. 1 1988, 1749−1752.
(47) Scanlon, J. T.; Willis, D. E. J. Chromatogr. Sci. 1985, 23, 333−
340.
2388−2390.
(13) Geilen, F. M. A.; Engendahl, B.; Harwardt, A.; Marquardt, W.;
Klankermayer, J.; Leitner, W. Angew. Chem., Int. Ed. 2010, 49, 5510−
5514.
(48) Di Mondo, D.; Ashok, D.; Waldie, F.; Schrier, N.; Morrison, M.;
(14) Geilen, F. M. A.; Engendahl, B.; Holscher, M.; Klankermayer, J.
̈
Schlaf, M. ACS Catal. 2011, 355−364.
R.; Leitner, W. J. Am. Chem. Soc. 2011, 133, 14349−14358.
(49) Cave, G. W. V.; Raston, C. L. J. Chem. Soc., Perkin Trans. 1 2001,
3258−3264.
́ ́
(15) Mehdi, H.; Fabos, V.; Tuba, R.; Bodor, A.; Mika, L. T.; Horvath,
I. T. Top. Catal. 2008, 48, 49−54.
(50) This is an unavoidable artifact of the benchtop reactor design in
which the unheated top of the reactor acts as a cold sink accumulating
some of the substrate and/or products.
(16) Grochowski, M. R.; Yang, W. R.; Sen, A. Chem.Eur. J. 2012,
18, 12363−12371.
(17) Yang, W. R.; Sen, A. ChemSusChem 2010, 3, 597−603.
(18) Gowda, A. S.; Parkin, S.; Ladipo, F. T. Appl. Organomet. Chem.
2012, 26, 86−93.
(51) For both catalysts very small amounts of a black solid were
observed at the phase boundary between the aqueous polar and
nonpolar organic phase. Using these solids as the catalyst in a control
reaction gave no conversion, which together with the low density of
the material suggest that it is not bulk Ru or Ir metal but carbonized
organics originating from either starting material or product
decomposition. (This also applies to the reactions with 2,5-DMF as
the substrate.)
(19) Lau, C. P.; Cheng, L. Inorg. Chim. Acta 1992, 195, 133−134.
(20) Lau, C.-P.; Cheng, L. J. Mol. Catal. 1993, 84, 39−50.
(21) Xie, Z.; Schlaf, M. J. Mol. Catal. A: Chem. 2005, 229, 151−158.
(22) Wegenhart, B. L.; Abu-Omar, M. M. Inorg. Chem. 2010, 49,
4741−4743.
(23) Yi, J.; Liu, S.; Abu-Omar, M. M. ChemSusChem 2012, 5, 1401−
1404.
(24) Shiramizu, M.; Toste, F. D. Angew. Chem., Int. Ed. 2012, 51,
(52) See http://www.iapws.org/release.html for the up-to-date
collection of physical data on water.
(53) McMurtrie, J.; Dance, I. CrystEngComm 2009, 11, 1141−1149.
(54) Constable, E. C.; Housecroft, C. E.; Medlycott, E. A.;
Neuburger, M.; Reinders, F.; Reymann, S.; Schaffner, S. Inorg. Chem.
Commun. 2008, 11, 805−808.
8082−8086.
(25) Shiramizu, M.; Toste, F. D. Angew. Chem., Int. Ed. 2013, 52,
12905−12909.
(26) Raju, S.; Jastrzebski, J.; Lutz, M.; Gebbink, R. ChemSusChem
2013, 6, 1673−1680.
(55) The smaller peak at 381.28 is tentatively assigned to the species
(4′-Ph-terpy)Mn(OH)]+.
(27) Qu, S. L.; Dang, Y. F.; Wen, M. W.; Wang, Z. X. Chem.Eur. J.
(56) Further details on the single crystal X-ray analyses (parameters,
tables of bond distances and angles, etc.) are given in the Supporting
Information.
2013, 19, 3827−3832.
(28) Generally referred to as ″humin formation″ this process is
effectively an uncontolled caramelization by loss of water followed by
aldol, Diels−Alder and other cross-condensation reactions on the
surface of the catalyst.
(57) Wang, D.-S.; Chen, Q.-A.; Lu, S.-M.; Zhou, Y.-G. Chem. Rev.
2011, 112, 2557−2590.
(58) Wysocki, J.; Ortega, N.; Glorius, F. Angew. Chem., Int. Ed. 2014,
53, 8751−8755.
(29) Cornils, B.; Herrmann, W. A. Applied Homogeneous Catalysis
with Organometallic Compounds; VCH: Weinheim, 1996; Vol. 1, pp 3−
27.
(59) Copes, J. P.; McKinley, C. General Aniline & Film Corporation,
1954, USA, US 2,686,817.
(30) Dykeman, R. R.; Luska, K. L.; Thibault, M. E.; Jones, M. D.;
Schlaf, M.; Khanfar, M.; Taylor, N. J.; Britten, J. F.; Harrington, L. J.
Mol. Catal. A: Chem. 2007, 277, 233−251.
(60) van Zandvoort, I.; Wang, Y.; Rasrendra, C. B.; van Eck, E. R. H.;
Bruijnincx, P. C. A.; Heeres, H. J.; Weckhuysen, B. M. ChemSusChem
2013, 6, 1745−1758.
(61) Pray, H. A.; Schweickert, C. E.; Minnich, B. H. Ind. Eng. Chem.
1952, 44, 1146−1151.
(62) We recently acquired equipment that will allow reactions at up
4500 psi hydrogen pressure and may revisit these catalysts and
reactions in the future.
(31) Sutton, A. D.; Waldie, F. D.; Wu, R.; Schlaf, M.; Silks, L. A. ‘P.’;
Gordon, J. C. Nat. Chem. 2013, 5, 428−432.
(32) Oltmanns, J. U.; Palkovits, S.; Palkovits, R. Appl. Catal., A 2013,
456, 168−173.
(33) While there is a large and very successful body of work on
thermally highly stable pincer-type complexes based on phosphines,
we made the conscious decision to eschew these ligands in favor of air-
stable nitrogen-donor only based ligands.
(63) DiMondo, D.; Thibault, M. E.; Britten, J.; Schlaf, M.
Organometallics 2013, 32, 6541−6554.
(34) Kubas, G. Metal Dihydrogen and σ-Bond Complexes. Structure,
Theory and Reactivity; Kluwer Academic/Plenum Publishers: New
York, 2001.
(35) Taher, D.; Thibault, M. E.; Mondo, D. D.; Jennings, M.; Schlaf,
M. Chem.Eur. J. 2009, 10132−10143.
(36) Marziano, N. C.; Ronchin, L.; Tortato, C.; Zingales, A.; Sheikh-
Osman, A. A. J. Mol. Catal. A: Chem. 2001, 174, 265−277.
4128
dx.doi.org/10.1021/cs501202t | ACS Catal. 2014, 4, 4116−4128