Effects of hydrogen and water on the activity and selectivity of acetic acid hydrogenation on ruthenium

Article abstract of DOI:10.1039/c4gc00011k

Kinetic flow reactor experiments have been carried out to study acetic acid hydrogenation on a Ru/C catalyst in both three-phase (catalyst, aqueous, and gaseous) and two-phase (catalyst and gaseous) regimes. In addition, density functional theory calculations have been performed and combined with microkinetic modeling to better understand the activity and selectivity observed in the experiments. Our experiments show that ethanol selectivity varies strongly from <10% to a maximum of ～70% with increasing hydrogen partial pressure (pH₂) at 185 °C in the three-phase reactor. Co-fed water also enhances ethanol selectivity, from ～60% to ～70% in the two-phase reactor and ～40% to ～65% in the three-phase reactor, at 185 °C, but only up to a certain concentration. The aqueous phase is not necessary for high ethanol selectivity. The first-principles microkinetic analysis is able to reasonably capture the apparent activation energy, ethanol selectivity, and reaction orders of acetic acid and ethanol with respect to pH₂, providing a theoretical explanation for the crucial role that hydrogen plays in the selectivity of this reaction. Our findings provide insights into why high activity and selectivity for acetic acid hydrogenation to ethanol can be achieved on Ru, which may have general relevance to the catalytic hydrogenation of organic oxygenates on Ru and other metals.

Full text of DOI:10.1039/c4gc00011k

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Effects of hydrogen and water on the activity and
selectivity of acetic acid hydrogenation on
ruthenium
Cite this: Green Chem., 2014, 16, 911
Hakan Olcay,†a Ye Xu*‡b and George W. Huber*§a
Kinetic flow reactor experiments have been carried out to study acetic acid hydrogenation on a Ru/C
catalyst in both three-phase (catalyst, aqueous, and gaseous) and two-phase (catalyst and gaseous)
regimes. In addition, density functional theory calculations have been performed and combined with
microkinetic modeling to better understand the activity and selectivity observed in the experiments. Our
experiments show that ethanol selectivity varies strongly from <10% to a maximum of ∼70% with increas-
ing hydrogen partial pressure (pH ) at 185 °C in the three-phase reactor. Co-fed water also enhances
2
ethanol selectivity, from ∼60% to ∼70% in the two-phase reactor and ∼40% to ∼65% in the three-phase
reactor, at 185 °C, but only up to a certain concentration. The aqueous phase is not necessary for high
ethanol selectivity. The first-principles microkinetic analysis is able to reasonably capture the apparent
activation energy, ethanol selectivity, and reaction orders of acetic acid and ethanol with respect to pH
,
2
Received 5th January 2014,
Accepted 15th January 2014
providing a theoretical explanation for the crucial role that hydrogen plays in the selectivity of this reac-
tion. Our findings provide insights into why high activity and selectivity for acetic acid hydrogenation to
ethanol can be achieved on Ru, which may have general relevance to the catalytic hydrogenation of
organic oxygenates on Ru and other metals.
DOI: 10.1039/c4gc00011k
www.rsc.org/greenchem
reducing undesirable decomposition. These processes include
the hydrodeoxygenation of bio-oils,2 hydrogenation of fermen-
Introduction
Increases in the cost of petroleum combined with growing con- tation broths,3,4 and alkane production by aqueous phase pro-
cerns about the environmental impact of fossil fuels are cessing.5,6 One of the slowest steps in the hydrotreating of bio-
prompting researchers worldwide to investigate biomass as a oils is the hydrogenation of organic acids.7
feedstock for the production of liquid fuels and commodity
Ru/C has been shown to be an excellent aqueous-phase
chemicals.1 Biomass feedstock differs from petroleum feed- hydrogenation (APH) catalyst for organic acids. Miller and co-
stock in being highly rich in oxygen-containing functional workers converted lactic3,8,9 and propionic3 acids to the corres-
groups. The selective reduction of oxygen in the biomass feed- ponding aldehydes and alcohols on Ru/C catalysts by APH at
stock will be an integral part of future bio-refineries. Many pro- 70–150 °C and 30–100 bar of hydrogen pressure (pH ). They
2
cesses currently being developed for the conversion of reported high activity and selectivity (up to 95%) toward propy-
biomass into biofuels involve hydrogenation in the aqueous lene glycol from lactic acid.8 We have shown that Ru/C has the
phase. The aqueous phase is an ideal carrier for large hydro- highest activity and ethanol selectivity in the APH of acetic
philic molecules and permits mild-temperature processes acid among several monometallic transition metal catalysts
(Ru, Rh, Pt, Pd, Ir, Ni, and Cu) at mild temperature and up to
∼50 bar of pH .4 Supported Ru catalysts can also hydrogenate
2
many other organic oxygenates with high activity and
aDepartment of Chemical Engineering, University of Massachusetts, Amherst,
selectivity.10–14 All this stands in contrast to the reactivity of Pt
MA 01003, USA. E-mail: huber@engr.wisc.edu
bCenter for Nanophase Materials Sciences, Oak Ridge National Laboratory,
group metals at UHV and low-pressure conditions, which leads
primarily to C–C bond scission in acetic acid,15–21 although
alloying18 and strong interaction with oxides19,22,23 can mark-
edly alter the selectivity in favor of ethanol. These studies show
that the main challenge with conversion of oxygenates over Ru
catalysts is selectivity, and that the reaction environment plays
a crucial role.
Oak Ridge, TN 37831, USA. E-mail: yexu@lsu.edu
†Current address: Department of Aeronautics and Astronautics, Massachusetts
Institute of Technology, Cambridge, MA 02139, USA.
‡Current address: Department of Chemical Engineering, Louisiana State Univer-
sity, Baton Rouge, LA 70803, USA.
§Current address: Department of Chemical and Biological Engineering, Univer-
sity of Wisconsin-Madison, Madison, WI 53706, USA.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 911

View Article Online
Paper
Green Chemistry
Previously, we have demonstrated that the high activity of a
Ru/C catalyst relative to other metals at given conditions can
be explained by the ability of Ru to activate the initial C–O
Methods
Experimental
bond scission.4 The origin of the high ethanol selectivity, All experiments are carried out in a continuous up-flow
which stands in contrast to the UHV and ambient packed-bed reactor over a 5% Ru/C catalyst (Evonik H1, Strem
pressure experiments, remains unclear and unaddressed in Chemicals). The catalyst, obtained as 50% wetted, is first dried
the existing literature. We have, therefore, carried out a series overnight at 100 °C unless indicated otherwise. It is then
of kinetic experiments to explore what is responsible for the reduced in situ under a hydrogen flow of 155 mL min−1 at
selectivity for ethanol in the hydrogenation of acetic acid on 290 °C for two hours following an 8-hour temperature
Ru/C. Selectivity to ethanol (eqn (1)) is entirely a kinetic ramping. Based on the electrochemical behavior of Ru26,27 and
phenomenon since CH4 and CO2 are the thermodynamically previous experiments employing Ru in aqueous solutions,10,28
more favored products via decarbonylation (eqn (2)) and de- we conclude that the Ru catalysts in our study are stable
carboxylation (eqn (3)):
against oxidation or dissolution in the O2-free acetic acid solu-
tion with an initial concentration of 10 wt% (or 3.2 mol%;
pH = 2.3) under high pH .4 The metal dispersion of the fresh
2
CH3COOHðgÞ þ 2H2ðgÞ ! CH3CH2OH þ H2O
ðgÞ
ðgÞ
catalyst is measured, via hydrogen adsorption with Quanta-
chrome Autosorb 1C, to be 6.7%.
ð1Þ
ð2Þ
ΔG°rxn ¼ ꢀ13:6 kJ molꢀ1
Acetic acid (glacial, Fisher) feed (with or without water) is
pumped by an HPLC pump (Jasco PU980) into the reactor
system. The aqueous feed then mixes with hydrogen (and
helium as an inert gas, when dilution is needed), and enters a
1/4′′ OD tubular stainless steel reactor. The reactor runs
through a cylindrical aluminum insert that is used to provide
better heat conduction and is placed inside a tubular split
furnace (Lindberg) that is heated to the desired reaction temp-
erature. The liquid feed is heated up or evaporated, depending
on the reaction conditions, on glass beads packed within the
CH3COOHðgÞ þ H2ðgÞ ! CH4ðgÞ þ CO þ H2O
ðgÞ
ðgÞ
ΔG°rxn ¼ ꢀ33:4 kJ molꢀ1
CH3COOH ! CH4ðgÞ þ CO2ðgÞ ΔG°rxn ¼ ꢀ62:0 kJ molꢀ1
ðgÞ
ð3Þ
Our experiments indicate that higher pH strongly favors reactor between two end-plugs of glass wool (Fisher). Reaction
2
ethanol formation vs. non-selective C–C bond cleavage in takes place over the catalyst packing fixed between the two sets
acetic acid. Furthermore, co-feeding water is found to promote of glass wool. After the reaction, the effluent cools down at the
the ethanol selectivity in both the three-phase and the two- exit of the reactor. Another set of glass beads is used after the
phase regimes, and the three-phase conditions are not necess- catalyst packing to prevent the condensed phase at the exit of
ary to obtain high ethanol selectivity. The strong effects of pH
the reactor from coming in contact with the reacting phase(s)
2
are further corroborated by density functional theory (DFT) cal- on the catalyst surface. The cooled-down products separate
culations and microkinetic analysis based on a mechanism into different phases in a 150 mL sample cylinder (Swagelok).
involving the dissociation of acetic acid and the hydrogenation The gaseous products are analyzed on-line. The liquid pro-
of the monoxy (CxHyO) intermediates. The effects of H2 and ducts are drained from the phase separator periodically and
water on acetic acid conversion and ethanol formation have analyzed off-line. A back-pressure regulator (Swagelok) is used
been noted in previous batch reactor experiments,24,25 but to to keep the reactor pressure constant at desired values.
our knowledge detailed kinetic experiments and first-prin-
ciples-based theoretical analysis have not been performed Hydrogen and carbon dioxide are separated in Hayesep DB
before for acetic acid hydrogenation on Ru. packed column and detected using thermal conductivity detec-
In the following, we first recap the results of a set of experi- tor (HP 5890II GC). Gaseous and liquid hydrocarbons are sepa-
ments that we previously carried out in three-phase rated in AT-Q capillary column and analyzed by flame
Three gas chromatographs are employed in the analyses:
a
flow reactor varying the temperature as the main variable, ionization detector (HP 5890II GC). Liquid products are also
which entails a corresponding change in other variables analyzed by a Shimadzu GC-2010 Plus system equipped with
such as pH . We then describe the results of our new FID and Restek RTX-VMS capillary column.
2
experiments, in which the partial pressures/concentrations
The reaction parameters for the experiments performed in
of H2 and water have been varied while holding other this study are described in various tables. The set values indi-
variables constant, in order to demonstrate the effects of H2 cate the parameters physically set on the experimental setup at
and water on the activity and selectivity of this reaction on normal temperature and pressure (NTP), i.e. 20 °C and 1 atm.
Ru/C. In the remaining portion of this article we present a DFT The volumetric flow rates of the gas-phase acetic acid-water
study of the reaction pathways and the results of microkinetic mixture (for the two-phase reactions), and of hydrogen and
analysis, which demonstrates that our theoretical understand- helium at the reaction temperature and pressure (RTP) are calcu-
ing can explain both the activity and ethanol selectivity of this lated in the following way: The compressibility factors (Z) cal-
reaction on Ru.
culated using the Peng–Robinson equation of state29 are used
912 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
in the non-ideal gas law, pV = ZnRT, to calculate the molar den- layer of the four-layer slab differs by 0.01 eV from that of a five-
sities (V/n) of hydrogen and helium at NTP, which are used to layer slab. Thus, we consider the energies to be converged with
˙
obtain the inlet molar flow rates (n) using the volumetric flow respect to the given parameters.
˙
rates (V) at NTP. From the inlet molar flow rates, the volu-
Adsorption is allowed on one side of the slab only, with
metric flow rates at RTP can then be obtained. The volumetric dipole decoupling.37 All adsorbates and the top two Ru layers
and molar flow rates of the liquid-phase acetic acid–water are relaxed, and the bottom two Ru layers are held fixed at
mixture (for the three-phase reactions) at the reaction tempera- bulk positions. Geometry optimization is converged to below
ture are calculated through the density equations given by Sun 0.03 eV Å−1 in each degree of freedom for all relaxed atoms.
et al.30 Whether the acetic acid–water mixture remains in the The binding energy (ΔE) of an adsorbate is calculated as ΔE =
liquid phase or is vaporized at the RTP is determined through Etotal − Eclean − Egas, where Etotal, Eclean, and Egas are the total
dew point and bubble point calculations. If the calculated energies of the slab with the adsorbate on it, the clean metal
bubble point pressure is below the total pressure then the slab without any adsorbate, and the adsorbate atomic, radical,
mixture is assumed to remain in the liquid phase. If the calcu- or molecular species in the neutral state in the gas phase,
lated dew point pressure is above the total pressure then the respectively. The (3 × 3) surface unit cell corresponds to a cov-
mixture is assumed to completely evaporate into gas phase. erage (θ) of 1/9 ML for one adsorbate per unit cell.
The concentration for a species A is calculated by taking the
The free energy of a gas-phase species, A, is calcu-
ratio of the inlet molar flow rate of A (nA) to the sum of individ- lated as:38
˙
˙
ual volumetric flow rates of all species (Vi) in all phases (ϕ)
at RTP:
ꢀ
ꢁ
pA
p0
GA ¼ EA þ EAZPE þ ΔhAðT; p0Þ ꢀ TsAðT; p0Þ þ kBT ln
ð5Þ
X
˙
nA
˙
Vi
P
cA ¼
ð4Þ
ϕ
i
The pressure dependence of GA enters through pA, with the
reference pressure p0 being 1 bar. ΔhA is h(T, p0) − h(0 K, p0)
and the values of the enthalpies and entropies of the gas-
phase species tabulated by NIST are used.39 EA and EAZPE are
the DFT total energy and zero-point energy (ZPE) of A in gas
phase. kB is the Boltzmann constant. The gas-phase confor-
mation used in this study for acetic acid is the more stable
cis-(CvO, C–OH).
The total pressure, along with the other parameters to be
set at NTP on the experimental system, is varied in a trial-and-
error procedure controlled by a computer program to achieve
the concentrations at the reaction conditions that are to be
tested. The liquid hourly space velocities and gas hourly space
velocities (LHSV and GHSV, respectively) are calculated from
the volumetric flow rates at RTP and the volume of the catalyst
packing unless indicated otherwise.
For an adsorbed species A*, its free energy is calcu-
lated as:40
Theoretical
ZPE
ꢁ
ꢁ
ꢁ
ꢁ
GA ¼ EA þ EAꢁ þ ΔuA ðTÞ ꢀ TsA ðTÞ
ð6Þ
Spin-polarized periodic DFT calculations are performed in the
generalized gradient approximation (GGA-PBE)31 using the
Vienna Ab initio Simulation Package (VASP).32 The core elec- where ΔuA* replaces Δh for gas-phase species and is the
trons are described by the PAW method,33 and the Kohn–Sham change in internal energy from 0 K to T. EA* is the DFT
valence states are expanded in a plane wave basis set up to a total energy of the species A when adsorbed on the surface
kinetic energy of 400 eV. A first-order Methfessel–Paxton and is equal to Etotal – Eclean. EZPE, ΔuA*, and sA* are calculated
scheme is used for the smearing of electronic states with a from the vibrational frequencies associated with the normal
smearing factor of 0.075. All total energies are extrapolated modes of the adsorbed species. The normal modes are
to 0 K.
calculated in the harmonic approximation by diagonalizing
The equilibrium bulk lattice constants for Ru are calculated the mass-weighted Hessian matrix, which was built from
to be 2.726/1.578 Å, in good agreement with the experimental calculated two-sided differences of the first derivatives of
values of 2.70/1.58 Å.34 The surfaces of bulk Ru are represented the total energy estimated from small geometrical pertur-
by the thermodynamically most stable (0001) surface, which is bations (0.01 Å) in each degree of freedom of the optimized
in line with the low dispersion of our Ru/C catalyst. Ru step geometry of A*.
edges are expected to be poisoned by low molecular weight
The minimum-energy reaction path and transition state
carbon-containing species from decomposition at moderate (TS) for each elementary step are determined using a combi-
temperatures.4,35,36 The (0001) surface is modeled by a (3 × 3) nation of the climbing-image nudged elastic band method41,42
surface unit cell consisting four layers of metal atoms and is and the dimer method43 as appropriate. Each TS is verified to
separated from neighboring slabs in the z direction by seven possess only one imaginary mode in the direction of the
layers equivalent of vacuum. The surface Brillouin zone is corresponding bond breaking/forming process. The free
sampled with a 5 × 5 × 1 Monkhorst–Pack k-point mesh. The energy of the TS complex is calculated as described above for
total energy of the slab changes by 0.05 eV of sampling by a an adsorbed species, except with the imaginary mode
7 × 7 × 1 Monkhorst–Pack k-point mesh, and the energy per excluded.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 913

View Article Online
Paper
Green Chemistry
Results and discussion
Effects of temperature
Previously, we reported the effect of temperature on the activity
and selectivity of the APH of acetic acid over a 5% Ru/C catalyst,
the results of which are summarized here.4 The experiments
were carried out at a constant total pressure of 750 psi and
three different temperatures (110, 160, and 185 °C). The reac-
tion parameters are given in Table 1. At 750 psi and below
246 °C, the content of the reactor remains as a three-phase
mixture (gas phase, aqueous phase, and catalyst). We follow pre-
vious examples and assume our reactor to behave as a differen-
tial reactor because the conversion was less than 20%.44,45 Our
previous work also included experiments at 245 °C,4 but those
results are excluded here because the conversion was greater
than 20%. The apparent activation energy (Eaapp),
Fig. 1 Carbon turnover frequencies of acetic acid hydrogenation on
Ru/C at 750 psi total pressure and 110–185 °C in the 3-phase flow
ꢀ
ꢁ
@ðlnðrÞÞ
@ð1=TÞ
Eaapp ¼ ꢀkB
ð7Þ
●
reactor ( ; the temperature experiments). See Table 1 for experimental
pi
parameters. The turnover frequencies of acetic acid conversion pre-
dicted by the microkinetic models at corresponding experimental con-
for acetic acid conversion was calculated to be 33 kJ mol−1 (or
0.35 eV; Fig. 1), somewhat lower than the 53.1 kJ mol−1
reported by Shih et al. for the hydrogenation of acetic acid on
Ru in a batch reactor.24 The ethanol selectivity was 84% at
110 °C and 74% at 185 °C (Fig. 2). The main side products
included ethyl acetate, acetaldehyde, and alkanes (mainly
methane), while CO and CO2 were detected only at ppm levels.
□
△
ditions are plotted for comparison. : two-site model; : one-site
model.
Effects of hydrogen pressure
sat
H2O
Increasing temperature causes p
to increase and concur-
rently pH to decrease at a fixed total pressure. Thus, the
2
Table 1 Parameters in the temperature experiments on acetic acid (AA)
hydrogenationa
T (°C)
110
160
185
Set values
% Liquid concentration
AA
Liquid
H2
10.00
0.100
155
0
Volumetric flow rate (mL min−1
)
Fig. 2 Ethanol selectivity in acetic acid hydrogenation on Ru/C at
750 psi and 110–185 °C in the three-phase flow reactor (black bars). See
Table 1 for experimental parameters. Striped and grey bars are ethanol
selectivity predicted at corresponding experimental conditions, using
the two-site and one-site models respectively.
He
LHSV (h−1
)
4.1
Calculated values
Molar flow rate (mol min−1
)
Total
AA
H2O
H2
0.0117
1.69 × 10−4
5.07 × 10−3
6.46 × 10−3
0
temperature experiments (Tables 1 and 2) actually varied temp-
erature and pH together. To isolate the effect of pH , we have
2
2
He
performed two additional sets of experiments, both with the
concentration of acetic acid held constant and the hydrogen
concentration allowed to vary. Here hydrogen concentration is
Volumetric flow rate (mL min−1
)
AA-H2O 0.106 0.111
H2
He
AA
H2O
H2
0.114
4.81
0
0.0342
1.03
1.31
0
4.03
0
4.55
0
0.0407 0.0361
Concentration (mol L−1
)
taken to be directly proportional to pH at constant T because
2
1.22
1.56
0
1.08
1.38
0
of the ideal gas behavior and low solubility in water of hydro-
gen. In the first set shown in Table 3, the hydrogen concen-
tration was varied at a fixed reactor temperature of 160 °C by
allowing the total pressure to increase from 250 to 750 psi.
Both the turnover frequency (TOF) of acetic acid conversion
and the ethanol selectivity increased with increasing total
pressure (Fig. 3). Since the water partial pressure was constant
He
H2/AA 38.3
38.3
38.3
a Total pressure is 750 psi. Catalyst is Ru/C (0.85 g, 5%). Reaction is in
3-phase regime. Set values are parameters physically set at 20 °C and
1 atm. LHSV is weight-based. Calculated values are inlet values based
on the reaction temperature and pressure. Concentration is based on
the molar and volumetric flow rates at RTP.
914 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Table 2 Measured activity and selectivity for acetic acid hydrogenation
in the temperature experiments
T (°C)
110
160
185
Carbon conversion (%)
2.936
20.823
12.269
86.862
16.611
120.702
Carbon TOF (h−1
)
% Carbon selectivity
Acetaldehyde
Ethanol
Ethyl acetate
Methane
Ethane
0.970
83.663
12.638
1.867
0.863
0
0.185
80.207
11.254
7.137
1.218
0
0.128
73.952
10.322
13.971
1.627
0
Propane
CO
CO2
0
0
0
0
0
0
Fig. 3 Ethanol selectivity and carbon turnover frequency for acetic acid
hydrogenation at 160 °C and three different pressures on Ru/C in the
three-phase flow reactor. Black: ethanol + ethyl acetate selectivity;
checked: methane + ethane selectivity; points: turnover frequency. The
percentage due to ethyl acetate in the liquid phase is indicated for
750 psi (striped area). See Table 3 for experimental parameters.
Table 3 Parameters in the hydrogen pressure experiments on acetic
acid (AA) hydrogenation by varying total pressurea
P (psi)
250
500
750
Set values
% Liquid concentration
AA
Liquid
H2
10.00
0.100
155
0
Volumetric flow rate (mL min−1
)
pressure had the same effect as lowering the partial pressure of
H2 by decreasing the overall pressure. At 185 °C, diluting H2
significantly reduced the selectivity for ethanol while promot-
ing undesired C–C and 2nd C–O dissociation products, primar-
ily methane and ethane. CO and CO2 were detected only at
ppm levels over the range of H2 concentration tested and are
therefore ignored.
He
LHSV (h−1
)
6.0
Calculated values
Molar flow rate (mol min−1
)
Total
AA
H2O
H2
He
AA-H2O
H2
He
AA
H2O
H2
He
0.0117
1.68 × 10−4
5.05 × 10−3
6.44 × 10−3
0
0.111
6.79
If the reaction rates can be written as a power law with
respect to H2 concentration, then on a log–log scale the slope
of the TOF versus H2 concentration gives the apparent reaction
orders with respect to pH (aH ); i.e.,
Volumetric flow rate (mL min−1
)
13.5
0
4.55
0
0.0361
1.08
1.38
0
0
Concentration (mol L−1
)
0.0124 0.0244
0.370 0.731
0.473 0.933
2
2
ꢀ
ꢁ
@ðlnðrÞÞ
ap
¼
ð8Þ
H
2
0
0
38.3
@ðlnðpH ÞÞ
T
2
H2/AA 38.3
38.3
The TOF of acetic acid conversion and ethanol, methane,
and ethane formation are plotted as a function of H2 concen-
a Temperature is 160 °C. Catalyst is Ru/C (1.0 g, 5%). Reaction is three-
phase. Set values are parameters physically set at 20 °C and 1 atm.
LHSV is weight-based. Calculated values are inlet values based on the tration on a log–log scale in Fig. 4b. The TOF of methane and
reaction temperature and pressure. Concentration is based on the
ethane formation clearly deviate from straight lines, indicating
molar and volumetric flow rates at RTP.
that the rate equations for the two species cannot be expressed
as a simple power law. The TOF of acetic acid also deviates
from a straight line, but less so than the TOF of methane and
ethane. Fitting a straight line to the rate of acetic acid conver-
sion yields an apparent reaction order with respect to hydro-
gen concentration of 0.20. A reaction order of 0.4–0.6 with
at constant temperature, the change in total pressure was
entirely due to an increase in the inlet pH
.
2
Another way to change pH was to add an inert gas (helium)
2
to the gas stream while maintaining a constant total gas flow
rate as shown in Table 4. The temperature, total pressure, and
inlet liquid concentrations were also held constant. The total
molar flow rate, the volumetric flow rate of acetic acid plus
water, and the concentration of acetic acid were identical to
those used at 185 °C in Table 1 when no helium was added.
Fig. 4 shows the effect of changing the hydrogen concentration
on the TOF and product selectivity. The ethanol selectivity
respect to pH was reported previously for acetic acid hydrogen-
2
ation on Pt/TiO2.19 The TOF for ethanol formation is much
closer to being linear, and its apparent reaction order with
respect to hydrogen concentration is 1.38.
Effects of water concentration
Besides hydrogen pressure, another factor affecting the
increased and the ethane and methane selectivity decreased ethanol selectivity is the water concentration. Fig. 5 shows the
(Fig. 4a), while the overall TOF of acetic acid increased slightly, effect of water concentration based on the sum of the liquid-
with increasing hydrogen concentration (Fig. 4b). As expected,
phase (if any) and gas-phase flow rates on the TOF and selecti-
adding an inert gas such as He at constant temperature and vity for acetic acid and the major products. The concentration
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 915

View Article Online
Paper
Green Chemistry
Table 4 Parameters in the hydrogen pressure experiments on acetic acid (AA) hydrogenation by adding heliuma
Set values
% Liquid concentration
AA
Liquid
H2
10.00
0.100
155
0
9.90
0.099
102
52
10.00
0.095
50
10.00
0.100
20
Volumetric flow rate (mL min−1
)
He
98
136
Calculated values
Molar flow rate (mol min−1
)
Total
AA
H2O
H2
He
AA-H2O
H2
0.0117
1.69 × 10−4
5.07 × 10−3
6.46 × 10−3
0
0.114
4.81
0.0116
1.65 × 10−4
5.02 × 10−3
4.25 × 10−3
2.17 × 10−3
0.112
0.0111
1.60 × 10−4
4.79 × 10−3
2.08 × 10−3
4.07 × 10−3
0.108
0.0117
1.68 × 10−4
5.05 × 10−3
8.31 × 10−4
5.65 × 10−3
0.114
Volumetric flow rate (mL min−1
)
3.17
1.55
0.621
He
0
4.96
209
0.0342
1.03
1.60
4.87
207
0.0338
1.03
3.02
4.70
199
0.0342
1.02
4.19
4.96
209
0.0342
1.03
Space velocity (h−1
Concentration (mol L−1
)
LHSV
GHSV
AA
H2O
H2
)
1.31
0.868
0.444
0.169
He
H2/AA
0
38.3
0.443
25.7
0.870
13.0
1.15
4.94
a Temperature is 185 °C, and total pressure is 750 psi. Catalyst is Ru/C (0.50 g, 5%). Reaction is three-phase. Set values are parameters physically
set at 20 °C and 1 atm. LHSV is weight-based. Calculated values are inlet values based on the reaction temperature and pressure. Concentration
is based on the molar and volumetric flow rates at RTP.
of acetic acid and the ratio of the acetic acid : H2 concen- expect increased concentration of water in the reactant stream
trations were both kept constant in these experiments (see to decrease the rate of the reaction according to Le Châtelier’s
Table 5 for experimental parameters). The temperature was principle if the system is in equilibrium. As the microkinetic
fixed at 185 °C, while the total pressure was varied so that the models suggest (see following sections), the reaction is far
reaction took place in the two-phase regime as well as the from equilibrium under given conditions, so water does not
three-phase regime for comparison. These experiments were affect the reaction thermodynamically. Possible explanations
carried out at lower hydrogen concentrations compared to the for the enhancement effects of water include: (1) water stabil-
preceding experiments to accentuate the effect of water.
izes the desired product, ethanol. Although acetic acid and
For the rates of ethanol, methane and ethane formation, ethanol both have appreciable vapor pressures under standard
the calculated apparent reaction orders are 0.74, 0.51 and 0.33, conditions and although the reaction does not have to be
respectively, in the two-phase regime, and 0.60, −0.26 and carried out in aqueous phase unlike larger organic molecules,
0.17, respectively, in the three-phase regime. In the two-phase the high affinity of ethanol for water may help stabilize
regime, increasing the concentration of co-fed water mildly ethanol relative to CO2 and alkanes, which are the thermo-
enhanced the ethanol selectivity by ca. 10% (mainly at the dynamically more stable products than ethanol in gas phase
expense of methane; Fig. 5a) and consistently increased the as mentioned in the Introduction. (2) Water enhances bond
TOF of acetic acid conversion (Fig. 5b). On the other hand, in scission/formation steps through hydrogen bonding or proton
the three-phase regime an over 20% increase in the ethanol shuttling.46 However, to the extent that the formation of both
selectivity was obtained by increasing the concentration of co- ethanol and decomposition products involve C–H bond scis-
fed water, while the rate of acetic acid conversion remained sion/formation, and that the stabilizing effect of water on the
nearly constant. In both the two-phase and the three-phase less polar C–H bond is not as significant as on the C–O
regimes, increasing water concentration was beneficial to bond,47 the significance of the kinetic effect of water on the
ethanol selectivity up to a certain point, beyond which it had selectivity of this reaction is unclear. (3) Water dissociation on
little additional effect (Fig. 5a). The highest rate of acetic acid Ru provides a source of oxidant to remove surface poisons, e.g.
conversion and the highest ethanol selectivity were obtained removing CO through water-gas shift that would produce
in the two-phase regime. At the given conditions, the TOF of additional hydrogen in situ. This could explain why water
acetic acid conversion was somewhat lower in the three-phase benefits both the activity and selectivity of this reaction.
regime than in the two-phase regime. However, aqueous phase However, the production of CO and CO2 was at the ppm levels,
reactions may be beneficial from an economic point of view, whereas methane was consistently the main side product, indi-
because the process excludes an expensive step such as cating that residual carbon is primarily removed reductively,
volatilization.
not oxidatively. Overall, the beneficial role of water in acetic
The exact mechanism for the beneficial effects of water is acid hydrogenation requires further study to be fully
unclear. Since water is a product of the reaction, one might understood.
916 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Fig. 5 Effect of water concentration on (a) product selectivity and (b)
turnover frequency in acetic acid hydrogenation at 185 °C on Ru/C in
■□
: ethanol;
◆◇
the two- and three-phase flow reactors.
: acetic acid;
: ethane; hollow symbols: two-phase; filled symbols:
●○
▲△
: methane;
three-phase. See Table 5 for experimental parameters.
give rise to a series of monoxy intermediates, including acet-
aldehyde (CH3CHO), 1-hydroxyethylidene (CH3COH), 1-hydro-
xyethyl (CH3CHOH), ethoxy (CH3CH2O), comprising three
different pathways to ethanol. Fig. 6 shows the snapshots of
Fig. 4 Effect of hydrogen concentration on the (a) selectivity and (b)
turnover frequencies of acetic acid and the major products in acetic acid
hydrogenation at 750 psi and 185 °C on Ru/C in the three-phase flow
reactor. No ethyl acetate is produced. See Table 4 for experimental the minimum-energy configurations of these intermediates
parameters. (c) turnover frequencies of acetic acid conversion and
ethanol formation predicted using the microkinetic models and selecti-
(see Table 6 for the corresponding adsorption energies). Fig. 6
also shows the reaction network for the formation of ethanol
■
○
◆
△
vity descriptor. : acetic acid; : ethanol; : methane; : ethane. In (c),
black symbols are results of the one-site model and the grey symbols
are the results of the two-site model.
from acetic acid hydrogenation based on these intermediates,
along with the activation and reaction energies of the elemen-
tary steps in the network. All bond scission/formation steps
are assumed to be surface-mediated, as has been done in pre-
vious works on aqueous-phase catalytic conversion of organic
acids.3,4,19,48
Reaction mechanism for ethanol formation
In a previous study,4 we concluded that the rate of acetic acid
In Table 7, the TS energies of the various monoxy hydrogen-
conversion on different metal catalysts can be understood as ation steps are listed. It can be seen that the acetyl → acet-
aldehyde → ethoxy pathway has the lowest overall activation
energy. We designate this the primary hydrogenation pathway
controlled by the rate of the first C–O bond scission step for
acetic acid, and that the first C–O bond scission occurs directly
in acetic acid or acetate on Ru(0001). Here we investigate the (see Fig. 7 for snapshots of the TS’s on this pathway). Esterifi-
cation forming ethyl acetate is not included here for simplicity
because a variety of processes, including metal-free, acid-cata-
reaction pathway from acetic acid to ethanol on Ru(0001) in
more detail. Following C–O bond scission in acetic acid or
acetate, the successive hydrogenation of acetyl (CH3CO) may lyzed ones, may produce esters from acetic acid.23,49
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 917

View Article Online
Paper
Green Chemistry
Table 5 Parameters in the water/water vapor concentration experiments on acetic acid (AA) hydrogenationa
P (psi)
100
120
140
220
240
260
Set values
Liquid concentration (%)
AA
Liquid
H2
28.54
0.028
126.4
0
15.40
0.050
120.1
0
10.40
0.074
120.3
0
28.54
0.014
63.2
15.44
0.024
57.8
10.46
0.037
60.1
Volumetric flow rate (mL min−1
)
He
108
113
132.1
Calculated values
Reaction
2-Phase
6.54 × 10−3
1.37 × 10−4
1.14 × 10−3
5.25 × 10−3
0
2-Phase
7.51 × 10−3
1.31 × 10−4
2.39 × 10−3
4.99 × 10−3
0
2-Phase
8.85 × 10−3
1.30 × 10−4
3.72 × 10−3
5.00 × 10−3
0
3-Phase
7.76 × 10−3
6.86 × 10−5
5.72 × 10−4
2.63 × 10−3
4.49 × 10−3
0.0160
3-Phase
3-Phase
Molar flow rate (mol min−1
)
Total
AA
H2O
H2
He
AA-H2O
H2
8.31 × 10−3
6.27 × 10−5
1.15 × 10−3
2.40 × 10−3
4.70 × 10−3
0.0273
9.91 × 10−3
6.52 × 10−5
1.86 × 10−3
2.50 × 10−3
5.49 × 10−3
0.0420
Volumetric flow rate (mL min−1
)
6.79
29.1
11.1
23.0
14.4
19.8
6.62
5.55
5.33
He
0
0
0
0
0
0
11.3
2.83
10.8
4.83
11.7
7.43
Space velocity (h−1
Concentration (mol L−1
)
LHSV
GHSV
AA
H2O
H2
6352
6035
6053
3172
2894
3014
)
3.82 × 10−3
0.0319
0.146
3.82 × 10−3
0.0700
0.146
3.79 × 10−3
0.109
3.82 × 10−3
0.0319
3.82 × 10−3
0.0698
3.82 × 10−3
0.109
0.146
0.146
0.146
0.146
He
H2:AA
0
38.3
0
38.3
0
38.6
0.250
38.3
0.286
38.3
0.322
38.3
a Temperature is 185 °C. Catalyst is Ru/C (0.12 g, 5%). Set values are parameters physically set at 20 °C and 1 atm. Calculated values are inlet
values based on the reaction temperature and pressure. The liquid space velocity under normal conditions was kept within the ranges of
2.48–6.55 h−1 and 4.96–13.1 h−1 for three-phase and two-phase reactions, respectively. Concentration is based on the molar and volumetric flow
rates at RTP.
Descriptor for ethanol selectivity
and 1-hydroxyethyl, decompose via C–C, 2nd C–O, and methyl
C–H bond scission, which we designate as the primary
The non-selective decomposition of acetic acid to alkanes may decomposition steps. The decomposition of further dehydro-
be initiated anywhere along the selective pathway. For this genated species (CHx<3R) is not investigated because previous
reason, we have investigated how the monoxy surface inter- studies have shown that they tend to undergo facile C–C bond
mediates, including acetyl, acetaldehyde, 1-hydroxyethylidene, scission.18,50 C–C bond scission in acetic acid and acetate is
Fig. 6 Reaction network for the formation of ethanol from acetic acid hydrogenation. Only the selective pathways are shown. The surface inter-
mediates on Ru(0001) are: (a) acetic acid; (b) acetate; (c) acetyl; (d) acetaldehyde; (e) 1-hydroxyethylidene; (f) ethoxy; (g) 1-hydroxyethyl; (h) ethanol;
(i) O; ( j) OH; (k) H2O. Snapshots for the minimum-energy configuration for each intermediate are included (in each panel, left: side view; right: top
view). Large grey, medium grey and black, and small white spheres represent Ru, O, C, and H atoms, respectively. The DFT-calculated total energy
forward activation energy (Ea) and reaction energy (ΔE) for each elementary step are indicated on each arrow.
918 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Table 6 Minimum-energy adsorption configuration and energies
(ΔE, in eV) of surface intermediates on Ru(0001)a
Table 7 Free (G°TS) and total (ET° S) energies of the transition states of the
primary hydrogenation and decomposition steps for the monoxy
intermediatesa
Configuration
ΔE
ΔZPE
T (°C)
Acetic acid
Acetate
Acetyl
Acetaldehyde
1-Hydroxyethylidene
Ethoxy
1-Hydroxyethyl
Ethanol
O
μ-O,H
−0.49
−3.13
−2.41
−0.72
−2.74
−2.74
−1.69
−0.40
−5.90
−3.32
−0.40
−2.87
−0.02
+0.08
+0.08
+0.04
+0.04
+0.16
+0.08
+0.03
+0.08
+0.12
+0.07
+0.16
μ-O,O
Step
110 160 185
C off top
C off top
C off top
O hcp
C atop
O atop
hcp
O fcc
Flat, atop
fcc
G°TS
E°TS
Acetyl + H → acetaldehyde
Acetaldehyde + H → ethoxy
Ethoxy + H → ethanol
0.41 0.45 0.47 0.50
0.61 0.67 0.70 0.42
0.71 0.79 0.83 0.31
Acetyl + H → 1-hydroxyethylidene
Acetaldehyde + H → 1-hydroxyethyl
1-Hydroxyethylidene + H → 1-hydroxyethyl
1-Hydroxyethyl + H → ethanol
0.68 0.70 0.71 0.98
1.00 1.07 1.10 0.89
1.02 1.08 1.11 0.54
0.65 0.75 0.80 0.32
OH
H2O
H
a All energies are in eV. ΔE is based on DFT total energies. Adding ΔZPE
to ΔE yields ZPE-corrected adsorption energy. Coverage is 1/9 ML.
Acetyl
C–C
C–O
C–H
C–C
C–O
C–H
C–C
C–O
C–H
C–C
C–O
C–H
C–C
C–O
C–H
0.73 0.73 0.72 1.26
0.74 0.74 0.75 1.20
0.67 0.67 0.67 1.24
1.28 1.32 1.34 1.39
1.12 1.16 1.17 1.22
0.31 0.34 0.36 0.49
1.12 1.12 1.12 1.39
0.71 0.74 0.76 0.83
0.61 0.65 0.66 0.76
1.62 1.68 1.71 1.44
1.00 1.06 1.08 0.81
0.81 0.88 0.91 0.61
2.04 2.10 2.12 1.92
1.00 1.06 1.09 0.80
0.75 0.82 0.85 0.54
Acetaldehyde
1-Hydroxyethylidene
1-Hydroxyethyl
Ethoxy
not considered further because of the significant activation
energies for these steps (DFT total energy Ea = 1.33 and
1.70 eV, respectively).
We propose that the ethanol selectivity can be estimated by
comparing the rate of the primary decomposition step with
the lowest-energy TS, to that of the primary hydrogenation step
with the highest-energy TS. A comparison of the primary
decomposition steps shows that the TS of methyl C–H scission
in acetaldehyde (acetaldehyde → ketenal) is the lowest in
energy, whereas ethoxy → ethanol is the primary hydrogen-
ation step with the highest-energy TS (Table 7). By assuming
that the steps leading up to acetaldehyde and ethoxy formation
are quasi-equilibrated compared to methyl C–H scission in
acetaldehyde and ethoxy hydrogenation to ethanol, respect-
ively, the rates of the decomposition and hydrogenation of
monoxy intermediates are seen as controlled by the following
two lumped steps, where ‡ indicates a transition state:50
a All energies are in eV. All TS’s are calculated at 1/9 ML coverage. Free
energies are calculated at the conditions of the temperature experiments
(Table 1). E°TS is based on DFT total energies and included for reference.
G°TS and E°TS in this table are relative to the free and total energies,
respectively, of gas-phase ethanol plus a balance of gas-phase H2.
S and θ* are solved for together iteratively through the
microkinetic models as described below. As formulated, the
selectivity descriptor, S, is a function of gas-phase H2 through
both the pH and the GWH terms. This is consistent with the
CH3COOHðgÞ þ 1=2H2ðgÞ þ 3* ! ðCH2CHOÞ* ꢂ ꢂ ꢂ H*‡ þ OH*
ð9Þ
2
2
experimental insights, and we shall see below that it also leads
to satisfactory theoretical predictions for ethanol selectivity
based on the DFT calculations.
CH3COOHðgÞ þ 3=2H2ðgÞ þ 2* ! ðCH3CH2OHÞ*‡ þ OH*
ð10Þ
The rates of these two steps are given by the transition state
theory as:
Microkinetic models of acetic acid hydrogenation to ethanol
=°
decomp
°
°
1
2
°
1
G
þG
ꢀG
T
ꢀ
G
3
The purpose of our microkinetic models is not to simulate a
chemical reactor but to predict the TOF and selectivity based
on the proposed reaction mechanism and given reaction con-
ditions, using the thermodynamic and kinetic parameters that
are entirely derived from DFT calculations (except for the
enthalpies and entropies of the gas-phase species, for which
tabulated values are used; see Methods). The predictions are
to be compared with the observed activity and selectivity, thus
providing a measure of the validity of the proposed reaction
mechanism.
OH
AA
H
2
2
kBT
h
pAApH θ*
2
k
B
rdecomp
¼
eꢀ
ð11Þ
ð12Þ
θOH
=°
hydrog
°
°
3
°
3
2
G
þG
ꢀG
ꢀ
G
2
2
OH
AA
H
2
kBT
h
pAApH θ*
eꢀ
2
k
T
B
rhydrog
¼
θOH
here h is the Planck constant. The ratio of the two rates is:
=°
=°
decomp
°
H
G
ꢀG
ꢀG
hydrog
2
rhydrog
p
θ*
H2
k
T
B
S ¼
¼ e
ð13Þ
rdecomp
and the ethanol selectivity is therefore S/(S + 1). The standard
In the present work we propose a simplified reaction mech-
free energies for the two TS’s (TShydrog and TSdecomp) are listed anism consisting of the following nine of the steps shown in
in Table 8.
Fig. 6 that include the primary hydrogenation pathway, i.e.,
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 919

View Article Online
Paper
Green Chemistry
Table 8 Temperature (T) dependent standard free energy (G°) for gas-
phase species and surface intermediates, including transition states, on
Ru(0001) used to parameterize the microkinetic model for acetic acid
hydrogenation to ethanola
Species
G°
Acetic acid(g)
Ethanol(g)
Water(g)
H2(g)
−0.00318T − 44.795
−0.00314T − 44.455
−0.00203T − 13.540
−0.00144T − 6.370
Acetyl*
−0.00109T − 35.004
Acetaldehyde*
Ethoxy*
OH*
−0.00103T − 38.084
−0.00117T − 41.788
−0.000430T − 10.602
(0.00451T + 140.729)(2.219θATA*3 − 1.135θATA*
0.161θATA* − 0.308)
2
Acetate*
+
O*
H*
(0.000126T + 4.425)(0.0281θO*2 + 0.395θO* − 1.752)
(0.000470T + 30.032)(0.00469θH*2 + 0.00148θH*
0.127)
−
TSI
TSII
TSIII
TSIV = TShydrog
TSV
TSVI
TSVIII
TSIX
−0.00126T − 44.855
−0.000997T − 37.942
−0.00113T − 41.156
−0.00154T − 44.358
−0.000327T − 9.938
−0.000547T − 13.257
−0.00144T − 45.123
−0.00107T − 41.607
−0.000946T − 38.068
TSdecomp
a All energies are in eV. G° for each species is calculated at the
conditions of the temperature experiments and fitted to a linear
function in T. For acetate, O, and H, G° is the coverage-dependent
differential standard free energy and θ° is equal to θ. For all other
species θ° is 1/9 ML. H2 dissociation on Ru(0001) is calculated to be
barrier-less even with 7/9 ML of pre-adsorbed atomic H, so the
standard free energy of activation of Step 7 is assigned a small
arbitrary value of 0.10 eV independent of temperature, which does not
affect the outcome of the microkinetic model. Multiple reactants or
products in a step are calculated in separate unit cells.
Fig. 7 Snapshots of the transition states of the steps included in the
microkinetic models: (a) acetic acid → acetyl + OH; (b) acetate → acetyl
+ O; (c) acetyl + H → acetaldehyde; (d) acetaldehyde + H → ethoxy; (e)
ethoxy + H → ethanol; (f) O + H → OH; (vii) OH + H → H2O; (g) acetic
acid → acetate + H. In each panel, side view is on left and top view is on
right. Large grey, medium grey and black, and small white spheres rep-
resent Ru, O, C, and H atoms, respectively.
simplifies the microkinetic model by neglecting the large
numbers of non-selective steps.
I. acetic acid(g) + 2* ↔ acetyl* + OH*
As will be shown below, the coverages of all surface species
are lower than 0.01 ML under the experimental conditions
except for acetate, O, and H. Given the negligible interactions
between the monoxy species (CxHyOz) suggested by the low
coverages, and given the unknown nature of how free sites,
hydrogen, and water are arranged around the organic species
to affect the adsorption and reaction of the organic species, we
use the DFT results at 1/9 ML without co-adsorbed species to
calculate the standard free energies for the monoxy species
and the TS’s, and parameterize them with respect to tempera-
ture (G°(T); Table 8). The forward standard activation free
energy (G°a) for a step i is calculated as the difference between
the G° of the TS and the sum of the G° of the reactants:
II. acetyl* + H* ↔ acetaldehyde* + *
III. acetaldehyde* + H* ↔ ethoxy* + *
IV. ethoxy* + H* ↔ ethanol(g) + 2*
V. O* + H* ↔ OH* + *
VI. OH* + H* ↔ H2O(g) + 2*
VII. H2(g) + 2* ↔ 2H*
VIII. acetic acid(g) + 2* ↔ acetate* + H*
IX. acetate* + * ↔ acetyl* + O*
For simplicity, the weakly adsorbing species, including
acetic acid, water, and ethanol (cf. Table 6), are ignored so that
these species react from/form into the gas phase directly. For
acetate, the η2 configuration has been seen in our calculations
at high coverage, in which both of the carboxylic O atoms coor-
dinate to a single Ru atom.
X
G°a; i ¼ G°TS; i
ꢀ
G°reactants; i
ð14Þ
The microkinetic model is solved self-consistently for the
coverages of the surface intermediates and the ethanol selecti-
vity as described above, for each of the three sets of conditions
used in the temperature experiments (Table 1), by assuming
dθi
The forward and backward reaction rate constants for the
step are then calculated as:
°
a; i
G
all surface species to be at steady state, i.e.,
¼ 0, and by
kBT
h
kf; i
Ki
eꢀ T ; and kr; i
¼
ð15Þ
dt
k
B
kf; i
¼
including the selectivity descriptor as described above, which
920 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
where Ki is the equilibrium constant for step i. Possible co-
adsorbate interactions, e.g. between the organic species, water,
O/OH, and hydrogen, and the determination of the interaction
energetics will be the subject of future studies. For acetate,
atomic O, and atomic H, which are the more abundant surface
species, we express their differential standard free energies as
functions of both coverage and temperature (Table 8).
Table 9 Results of the microkinetic models for acetic acid hydrogen-
ation to ethanol calculated at the conditions of the temperature experi-
ments: rate of acetic acid conversion (−rI), coverage of adsorbed species
(θ), and reversibility for each step (β)a
T (°C)
110
160
185
Two-site model
−rI (1 s−1
)
0.28
72
2.39
71
19
52
The effective partial pressures of the gas-phase species,
including acetic acid, hydrogen, water, and ethanol, are deter-
mined based on the experimental conditions of the tempera-
ture experiments (Table 1) and the predicted selectivity. In our
experiments the content of the reactor can either be three-
phase or two-phase. In the three-phase regime, the gas phase
contains water vapor, taken to be at its saturation pressure
(pH O ¼ psHatOðTÞ); hydrogen; and small amounts of acetic acid
Ethanol selectivity (%)
Coverage (ML)
*
H*
1.4 × 10−4
0.84
1.8 × 10−4
0.62
4.1 × 10−4
0.61
O*
Acetate*
0.11
0.05
0.09
0.29
0.10
0.29
Reversibility
βI
0.04
0.22
0.39
βII
1.00
1.00
1.00
βIII
0.77
0.81
0.83
2
2
βIV
0.01
0.09
0.15
and ethanol (with decomposition products ignored). The
partial pressures of acetic acid and ethanol (pAA and pEtOH) are
calculated as the saturation pressure corresponding to the
average concentration between the inlet and outlet of each
species (zero at inlet for ethanol) in the aqueous phase accord-
βV
0.48
0.61
0.68
βVI
0.02
0.05
0.08
βVII
1.00
1.00
1.00
βVIII
βIX
1.00
1.00
1.00
1.1 × 10−5
6.5 × 10−5
2.3 × 10−4
Partial pressure (bar)
Acetic acid
Ethanol
H2O
H2
3.3 × 10−2
1.3 × 10−2
1.43
0.21
0.38
6.14
0.45
0.86
11.18
39.21
6.2 × 10−7
ing to Henry’s Law (for acetic acid: k°H = 5400 mol (kg bar)−1
d(ln(kH))/d(1/T) = 6300 K; for ethanol: k°H = 190 mol (kg bar)−1
;
;
d(ln(kH))/d(1/T) = 6600 K39). Since the solubility of H2 in water
is very low (0.00078 mol (kg bar)−1; d(ln(kH))/d(1/T) = 500 K39),
H2 remains predominantly in the gas phase and the balance of
50.23
44.98
Reaction quotient
3.2 × 10−11
3.2 × 10−8
One-site model
−rI (1 s−1
)
33
85
171
80
819
63
the total pressure is assigned to pH (in the absence of helium).
2
Ethanol selectivity (%)
Since the decomposition products are ignored in this pro-
Coverage (ML)
*
H*
cedure, pH is slightly overestimated particularly at low ethanol
6.6 × 10−5
0.75
1.1 × 10−4
0.53
2.6 × 10−4
0.52
2
selectivity. The calculated effective partial pressures are listed
in Table 9.
O*
Acetate*
0.20
0.05
0.18
0.29
0.18
0.29
The predicted TOF of acetic acid conversion (−rI, the net
rate of Step I, which is equal to the rate of ethanol formation,
the net rate of Step IV, by the steady-state assumption), the cov-
erage for each surface species (θ), and the reversibility (β) of
each step predicted by the model are listed in Table 9. The low
surface coverage of O species corroborates that the metallic Ru
surface is not oxidized under the reaction conditions. The first
set of results are based on acetic acid and acetate each disso-
ciating over a pair of sites (the two-site model). The predicted
TOF and ethanol selectivity are plotted in Fig. 1 and 2 for com-
parison with the experimental TOF and selectivity. The pre-
dicted TOF is too low by two orders of magnitude at 110 °C
but comes within a factor of ∼5 from the measured TOF at
185 °C. The ethanol selectivity decreases continuously with the
increase in temperature, but the decrease is too rapid com-
pared to the experiments. The Eaapp is measured experimentally
to be 0.33 eV, whereas the Eaapp based on the calculated TOF is
0.78 eV. Incidentally, the alternate acetyl → 1-hydroxyethy-
lidene → 1-hydroxyethyl → ethanol and acetyl → acetaldehyde
→ 1-hydroxyethyl → ethanol pathways lead to TOF that are ∼10
times smaller (results not shown). These alternate pathways
are therefore omitted from subsequent analysis.
Reversibility
βI
βII
βIII
βIV
βV
βVI
βVII
0.71
1.00
0.83
1.00
0.85
1.00
0.77
0.79
0.81
2.4 × 10−5
0.76
5.9 × 10−4
0.87
2.1 × 10−3
0.88
3.7 × 10−5
1.00
2.5 × 10−4
1.00
8.9 × 10−4
1.00
βVIII
βIX
0.94
0.99
0.96
1.00
0.97
1.00
Partial pressure (bar)
Acetic acid
Ethanol
H2O
H2
Reaction quotient
Equilibrium constant
3.3 × 10−2
1.5 × 10−2
1.43
0.21
0.43
6.14
44.93
8.3 × 10−8
6.87
0.45
1.04
11.18
39.03
1.1 × 10−6
3.51
50.23
3.5 × 10−10
34.29
a Equilibrium constant is based on gas-phase acetic acid
9
Q
hydrogenation to ethanol. Reaction quotient is equal to
βi.
i¼1
experimental and predicted TOF, we have also tested an ad hoc
one-site ansatz. It involves changing the free site requirement
for the C–O dissociation steps from 2 to 1, i.e.,
I. acetic acid(g) + * ↔ acetyl* + OH*
IX. acetate* ↔ acetyl* + O*
and re-solving the model with all other parameters kept the
same. The results of this modified model (the one-site model)
The vanishing coverage of free sites coupled with the two-
site requirement for C–O bond dissociation appears to be the
primary reason for the under-prediction of the TOF at the
lower temperatures. To address this inconsistency between the
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 921

View Article Online
Paper
Green Chemistry
are also listed in Table 9 and plotted in Fig. 1 and 2 for com- apparent reaction orders with respect to pH (aH ) to be 0.003
2
2
parison. The TOF of the one-site model agrees more closely for acetic acid conversion and 1.09 for ethanol formation. The
with the experiments, and the selectivity is overall in better one-site model predicts aH to be 0.30 and 1.29, respectively, in
2
agreement with the experiments. The predicted Eaapp is now better agreement with aH exhibited in the experiments (0.20
2
0.59 eV, also in closer agreement with the experiments than and 1.38, respectively; cf. Fig. 4b). The predicted positive aH
2
the two-site model is. Whereas in the two-site model the acetic for ethanol formation demonstrates that high pH is a requisite
2
acid channel is the dominant C–O bond scission mechanism for high ethanol selectivity. Furthermore, the positive aH are
2
(βI ≫ βIX), the acetate channel is somewhat more dominant consistent with the calculated reversibility that suggests the
(βIX > βI) in the one-site model.
hydrogenation steps to be rate-controlling, and corroborates
The one-site ansatz may be rationalized in the following the validity of the selectivity descriptor S. The descriptor does
way: The scission of the carboxylate C–O bond requires a tri- not account for the distribution of different decomposition
angular ensemble of three metal atoms on close-packed metal products, including CH4, C2H6, and CO2 (see Table 2). More
surfaces (see Fig. 7).51,52 This precludes the adsorption of detailed microkinetic analysis will be needed to fully account
other species in the threefold site in the center of the ensem- for the formation of the different decomposition products.
ble, and so the TS effectively occupies one threefold site just
like H, O, and OH. It should be noted that, whereas most of
the organic species bond to Ru(0001) through C or O to the
top of Ru atoms, the atomic species as well as OH and ethoxy
Conclusions
occupy the threefold sites, so a single site-type in any Lang- We have performed a series of kinetic flow reactor experiments
muir–Hinshelwood reaction mechanism to model the hydro- to better understand the high selectivity for ethanol that can
genation of acetic acid or other organic oxygenates may be be achieved in acetic acid hydrogenation on Ru. The reaction
inherently deficient. Two different site-types may need to be has a measured Eaapp of 33 kJ mol−1 for acetic acid conversion
invoked,3 together with an improved description of how inter- and well over 70% selectivity for ethanol at and below 185 °C
mediates are co-adsorbed on the surface, in order to more in the three-phase (gaseous, aqueous, and catalyst) regime at a
accurately describe this reaction with microkinetic or kinetic fixed total pressure of 750 psi. Microkinetic models based on
Monte Carlo modeling. Interestingly, the two-site and one-site the dissociative adsorption of acetic acid and the successive
TOF values bracket the experimental measurements, with the hydrogenation of acetyl to acetaldehyde and ethoxy, using
former better matching the experimental values at the higher DFT-calculated thermodynamic and kinetic parameters except
temperatures and the latter better matching the experimental for the enthalpies and entropies of the gas-phase species, are
values at the lower temperatures. Whether this is a reflection able to predict the rate of acetic acid conversion to within an
of an actual transition from a one-site to a two-site C–O bond order of magnitude of the experimental rates, with an calcu-
scission mechanism remains to be determined.
Both the two-site and the one-site models predict a highly bond scission is taken to occur over two sites or one site.
covered surface at all three of the temperatures, with the free Our experiments clearly show that increasing the hydrogen
lated Eaapp of 0.78 eV (0.59 eV) depending on whether C–O
site coverage being on the order of ∼10−4, and atomic H being pressure increases both the ethanol selectivity and the rate of
the most abundant surface species followed by up to 20% of O ethanol formation. This conclusion is closely corroborated by
and 30% of acetate. According to both models, all the other including a simple theoretical selectivity descriptor in the
surface species have negligibly small coverages (omitted from microkinetic model that is based on the rates of the most
Table 9). Including direct interactions between major surface difficult of the hydrogenation steps and the most facile of the
species may lead to a significantly higher coverage of free decomposition steps, for the monoxy intermediates (CxHyOz).
sites.53 It remains to be seen what effect such an increase in The predicted ethanol selectivity is 72% (two-site model) and
free site coverage would have on the ethanol selectivity. The 85% (one-site model) at 110 °C and decreases with increasing
reaction quotient indicates that the overall reaction is far from temperature, in close agreement with the experimental results
equilibrium to the side of the reactants in both models.
(84% at 110 °C). Moreover, at 185 °C and a total pressure of
Both models predict the ethanol selectivity to decrease with 750 psi (three-phase regime), the measured TOF of acetic acid
increasing temperature in agreement with the temperature conversion and ethanol formation have apparent reaction
experiments, which suggests that the ethanol selectivity is orders of 0.20 and 1.38 in hydrogen concentration, respect-
affected by pH . To further probe the effects of hydrogen on the ively, with which the predicted apparent reaction orders also
2
reaction, we have calculated the TOF of acetic acid conversion come in good agreement (0.003 and 1.09 (two-site model), and
and ethanol formation at several different values of pH
0.30 and 1.29 (one-site model)). Co-feeding water is identified
2
(including 5, 10, and 20 bar and the pH corresponding to as another way to enhance the ethanol selectivity in both the
2
185 °C in the temperature experiment), assuming the same three-phase and two-phase regimes, up to a certain concen-
temperature and acetic acid conversion as for the 185 °C temp- tration. We offer some thoughts on the origin of the water
erature experiment. The results show that the TOF of both effect, but its exact mechanism remains to be clarified. Our
acetic acid conversion and ethanol formation indeed increases findings provide insights into why high activity and selectivity
with increasing pH (Fig. 4c). The two-site model predicts the for acetic acid hydrogenation can be achieved on Ru, which
2
922 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
may have general relevance and applicability for the hydrogen- 17 R. D. Haley, M. S. Tikhov and R. M. Lambert, Catal. Lett.,
ation of organic oxygenates on Ru and other metals, for both 2001, 76, 125–130.
two- and three-phase thermal catalytic hydrogenation 18 R. Alcalá, J. W. Shabaker, G. W. Huber, M. A. Sanchez-
reactions.
Castillo and J. A. Dumesic, J. Phys. Chem. B, 2005, 109,
2074–2085.
19 W. Rachmady and M. A. Vannice, J. Catal., 2000, 192, 322–
334.
20 G. Hoogers, D. C. Papageorgopoulos, Q. Ge and D. A. King,
Surf. Sci., 1995, 340, 23–35.
Acknowledgements
Experimental work was performed at University of Massachu-
setts and supported with a grant from the ACS Petroleum 21 M. Bowker, C. Morgan and J. Couves, Surf. Sci., 2004, 555,
Research Fund. Theoretical work was performed at the Center 145–156.
for Nanophase Materials Sciences, which is sponsored at Oak 22 R. Pestman, R. M. Koster, J. A. Z. Pieterse and V. Ponec,
Ridge National Laboratory (ORNL) by the Scientific User Facili- J. Catal., 1997, 168, 255–264.
ties Division, Office of Basic Energy Sciences, U.S. Department 23 G. Onyestyák, Catal. Commun., 2013, 38, 50–53.
of Energy (US-DOE), and used computing resources of ORNL 24 Y. S. Shih and C. K. Lee, J. Chin. Chem. Soc., 1985, 32,
and the National Energy Research Scientific Computing
Center, which is supported by Office of Science, US-DOE, 25 H. Wan, R. V. Chaudhari and B. Subramaniam, Energy
under Contract DE-AC02-05CH11231. G. W. Huber was sup- Fuels, 2012, 27, 487–493.
ported as part of the Institute for Atom-efficient Chemical 26 CRC Handbook of Chemistry and Physics, CRC Press, Boca
Transformations (IACT), an Energy Frontier Research Center Raton, FL, 2009.
funded by the US-DOE, Office of Science, Office of Basic 27 T. Loucka, J. Appl. Electrochem., 1989, 20, 522–523.
29–34.
Energy Sciences.
28 W. C. Ketchie, E. P. Maris and R. J. Davis, Chem. Mater.,
2007, 19, 3406–3411.
29 S. I. Sandler, Chemical and Engineering Thermodynamics,
Wiley, 1999.
30 T. Sun, D. Ly and A. S. Teja, Ind. Eng. Chem. Res., 1995, 34,
1327–1331.
31 J. J. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,
1996, 77, 3865.
32 G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter,
1996, 54, 11169.
33 G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter,
1999, 59, 1758.
34 N. W. Ashcroft and N. D. Mermin, Solid State Physics, Saun-
ders College, Orlando, FL, 1976.
35 P. Ferrin, D. Simonetti, S. Kandoi, E. Kunkes,
J. A. Dumesic, J. K. Nørskov and M. Mavrikakis, J. Am.
Chem. Soc., 2009, 131, 5809–5815.
36 S. B. Vendelbo, M. Johansson, D. J. Mowbray, M. P. Andersson,
F. Abild-Pedersen, J. H. Nielsen, J. K. Nørskov and
I. Chorkendorff, Top. Catal., 2010, 53, 357–364.
37 J. Neugebauer and M. Scheffler, Phys. Rev. B: Condens.
Matter, 1992, 46, 16967.
References
1 G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006,
106, 4044–4098.
2 T. P. Vispute, H. Zhang, A. Sanna, R. Xiao and G. W. Huber,
Science, 2010, 330, 1222–1227.
3 Y. Q. Chen, D. J. Miller and J. E. Jackson, Ind. Eng. Chem.
Res., 2007, 46, 3334–3340.
4 H. Olcay, L. Xu, Y. Xu and G. W. Huber, ChemCatChem,
2010, 2, 1420–1424.
5 G. W. Huber, R. D. Cortright and J. A. Dumesic, Angew.
Chem., Int. Ed., 2004, 43, 1549–1551.
6 R. Xing, A. V. Subrahmanyam, H. Olcay, W. Qi, G. P. van
Walsum, H. Pendse and G. W. Huber, Green Chem., 2010,
12, 1933–1946.
7 D. C. Elliott, Energy Fuels, 2007, 21, 1792–1815.
8 Z. G. Zhang, J. E. Jackson and D. J. Miller, Ind. Eng. Chem.
Res., 2002, 41, 691–696.
9 T. S. Dalavoy, J. E. Jackson, G. M. Swain, D. J. Miller, J. Li
and J. Lipkowski, J. Catal., 2007, 246, 15–28.
10 P. Gallezot, N. Nicolaus, G. Fleche, P. Fuertes and
A. Perrard, J. Catal., 1998, 180, 51–55.
38 M. V. Bollinger, K. W. Jacobsen and J. K. Nørskov, Phys.
Rev. B: Condens. Matter, 2003, 67, 085410.
39 R. Sander, in NIST Chemistry WebBook, NIST Standard Refer-
ence Database Number 69, ed. P. J. Linstrom and
W. G. Mallard, National Institute of Standards and Tech-
nology, Gaithersburg, MD, 2012.
11 E. P. Maris and R. J. Davis, J. Catal., 2007, 249, 328–337.
12 K. P. Pimparkar, D. J. Miller and J. E. Jackson, Ind. Eng.
Chem. Res., 2008, 47, 7648–7653.
13 L. E. Manzer, Appl. Catal., A, 2004, 272, 249–256.
14 J. Lee, Y. Xu and G. W. Huber, Appl. Catal., B, 2013,
140–141, 98–107.
15 C. J. Houtman, N. F. Brown and M. A. Barteau, J. Catal.,
1994, 145, 37–53.
40 M. P. Andersson, E. Abild-Pedersen, I. N. Remediakis,
T. Bligaard, G. Jones, J. Engbwk, O. Lytken, S. Horch,
J. H. Nielsen, J. Sehested, J. R. Rostrup-Nielsen, J. K. Nørskov
and I. Chorkendorff, J. Catal., 2008, 255, 6–19.
41 G. Henkelman and H. Jónsson, J. Chem. Phys., 2000, 113, 9978.
16 A. R. Garcia, J. L. da Silva and L. M. Ilharco, Surf. Sci., 1998, 42 G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem.
415, 183–193.
Phys., 2000, 113, 9901.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 911–924 | 923

View Article Online
Paper
Green Chemistry
43 G. Henkelman and H. Jónsson, J. Chem. Phys., 1999, 111, 48 B. N. Zope, D. D. Hibbitts, M. Neurock and R. J. Davis,
7010–7022. Science, 2010, 330, 74–78.
44 A. D. Cowan, R. Dumpelmann and N. W. Cant, J. Catal., 49 Y. Liu, E. Lotero and J. G. Goodwin Jr., J. Catal., 2006, 242,
1995, 151, 356–363. 278–286.
45 M. J. Vincent and R. D. Gonzalez, Appl. Catal., A, 2001, 217, 50 R. Alcalá, M. Mavrikakis and J. A. Dumesic, J. Catal., 2003,
143–156. 218, 178–190.
46 C. D. Taylor and M. Neurock, Curr. Opin. Solid State Mater., 51 V. Pallassana and M. Neurock, J. Catal., 2002, 209, 289–305.
2005, 9, 49–65. 52 L. Xu and Y. Xu, Catal. Today, 2011, 165, 96–105.
47 X. Nie, M. R. Esopi, M. J. Janik and A. Asthagiri, Angew. 53 J. Lu, S. Behtash, M. Faheem and A. Heyden, J. Catal.,
Chem., Int. Ed., 2013, 52, 2459–2462. 2013, 305, 56–66.
924 | Green Chem., 2014, 16, 911–924
This journal is © The Royal Society of Chemistry 2014