Analysis of kinetics and reaction pathways in the aqueous-phase hydrogenation of levulinic acid to form γ-Valerolactone over Ru/C

Article abstract of DOI:10.1021/cs401177p

Reaction pathways and kinetics governing the Ru-catalyzed hydrogenation of levulinic acid (LA) in the aqueous phase to form γ-valerolactone (GVL) were considered in a packed bed reactor. GVL can be produced by two distinct hydrogenation pathways; however, over Ru/C at temperatures below 423 K, it forms exclusively via intramolecular esterification of 4-hydroxypentanoic acid (HPA). Over Ru/C, reasonable hydrogenation rates of LA to HPA were observed at near-ambient temperatures (e.g., 0.08 s^-1 at 323 K), but GVL selectivities are poor (<5%) under these conditions. Apparent barriers for LA hydrogenation and HPA esterification are 48 and 70 kJ mol^-1, respectively, and GVL selectivity improves at higher temperatures alongside increasing mass transfer limitations in 45-90 μm catalyst particles. Reactivity and selectivity trends in LA hydrogenation below 343 K are well-described by an empirical kinetic model capturing sequential hydrogenation and esterification. Coupling stacked beds of Ru/C and Amberlyst-15 delivers high GVL yields (～80%) at near ambient temperatures (323 K) and practical residence times.

Full text of DOI:10.1021/cs401177p

Research Article
pubs.acs.org/acscatalysis
Analysis of Kinetics and Reaction Pathways in the Aqueous-Phase
Hydrogenation of Levulinic Acid To Form γ‑Valerolactone over Ru/C
Omar Ali Abdelrahman,† Andreas Heyden,‡ and Jesse Q. Bond†,*
†Department of Biomedical and Chemical Engineering, Syracuse University, Syracuse, New York 13244, United States
‡Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina 29208, United States
S
* Supporting Information
ABSTRACT: Reaction pathways and kinetics governing the Ru-
catalyzed hydrogenation of levulinic acid (LA) in the aqueous phase
to form γ-valerolactone (GVL) were considered in a packed bed
reactor. GVL can be produced by two distinct hydrogenation
pathways; however, over Ru/C at temperatures below 423 K, it
forms exclusively via intramolecular esterification of 4-hydroxypen-
tanoic acid (HPA). Over Ru/C, reasonable hydrogenation rates of
LA to HPA were observed at near-ambient temperatures (e.g., 0.08
s−1 at 323 K), but GVL selectivities are poor (<5%) under these
conditions. Apparent barriers for LA hydrogenation and HPA
esterification are 48 and 70 kJ mol−1, respectively, and GVL
selectivity improves at higher temperatures alongside increasing
mass transfer limitations in 45−90 μm catalyst particles. Reactivity and selectivity trends in LA hydrogenation below 343 K are
well-described by an empirical kinetic model capturing sequential hydrogenation and esterification. Coupling stacked beds of Ru/
C and Amberlyst-15 delivers high GVL yields (∼80%) at near ambient temperatures (323 K) and practical residence times.
KEYWORDS: levulinic acid, ruthenium, γ-valerolactone, hydroxypentanoic acid, angelicalactone, aqueous phase kinetics, hydrogenation
which offer good hydrogenation rates, high GVL selectivity, and
(particularly in the case of bimetallics) good stability.7,41,42,45
Thus, detailed consideration of the kinetics of LA hydro-
genation over Ru is warranted.
Between temperatures of 298 and 473 K and over supported
Ru catalysts, LA is converted selectively to GVL via reduction
with molecular H2 (10−35 bar).7,46 Producing GVL requires
both hydrogenation and dehydration of LA, and we may
envision two different pathways for the transformation,
depending on the order in which dehydration and hydro-
genation occur (Scheme 1).
Pathway 1 illustrates the sequence initiated by hydrogenation
of the ketone group in LA to form 4-hydroxypentanoic acid
(HPA). Subsequently, HPA undergoes acid-catalyzed, intra-
molecular esterification (ring closure) to form the thermody-
namically preferred lactone, GVL. Alternatively, angelicalac-
tones (AL) can form via endothermic dehydration of LA, and
they become increasingly prevalent in acidic media and at
elevated temperatures (pathway 2). In the presence of Ru/C
and under H2 atmospheres, angelicalactones are anticipated to
rapidly hydrogenate, forming GVL. To date, the relative
contributions of each pathway have not been delineated, and
it is unclear which reactions in the above network are kinetically
significant. Further, depending on operating conditions,20,47
1. INTRODUCTION
γ-Valerolactone (GVL) is of interest as a lignocellulosic
platform chemical that offers tremendous flexibility in down-
stream applications and upgrading. For example, GVL can serve
directly as a gasoline blender1 or be subsequently processed to
yield relatively energy-dense fuel additives, such as methylte-
trahydrofuran,2 valeric biodiesel,3 or liquid alkanes.4−10 GVL is
also remarkably versatile as a biorefining solvent,11 particularly
for expediting the production of sugars, furans, levulinic acid
(LA), and their numerous derivatives from lignocellulose.12−17
Finally, GVL can be converted to chemical intermediates such
as 1,4 pentanediol,2 alkyl pentenoates,18 and α-methylene-γ-
valerolactone,19 any of which may find application in the
production of biobased polymers. The primary route
envisioned for the production of lignocellulosic GVL is
hydrogenation of levulinic acid,20 which may be prepared
from both 5- and 6-carbon sugars present in cellulose and
hemicellulose through hydrolysis of either 5-hydroxymethylfur-
fural or furfuryl alcohol.21−30 There are multiple options for
selective hydrogenation of LA, and both homogeneous2 and
heterogeneous systems20 have been employed. Despite
burgeoning interest in transfer hydrogenation for GVL
production,31−33 the majority of LA hydrogenation processes
have used molecular H2 as a reducing agent and supported
metals, such as Cu,34,35 Ru,19,36−39 Ir,40 Au,33 or bimetallics,
such as RuRe41,42 or RuSn43,44 as catalysts. Further, recent
studies focusing on the scale-up and economic feasibility of a
GVL-centered biorefinery have favored Ru or Ru-bimetallics,
Received: December 10, 2013
Revised: February 14, 2014
Published: February 28, 2014
© 2014 American Chemical Society
1171
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
Levulinic acid (LA, 98%, Sigma Aldrich), γ-valerolactone
(GVL, 98%, Sigma Aldrich), sulfuric acid (95−98 wt %, Sigma
Aldrich), and acetonitrile (HPLC grade 99.9% purity, Fisher
Scientific) were employed as supplied by the manufacturer. A 4-
hydroxypentanoic acid (HPA) calibration standard was
synthesized by alkaline hydrolysis of GVL in sodium hydroxide
(1 M) to produce 4-hydroxypentanoate, which forms HPA
upon protonation. HPA used in esterification kinetic studies
was synthesized via partial hydrogenation of aqueous LA (0.5
M) over Ru/C in a packed bed reactor (298 K, 23 bar H2).
After synthesis, esterification feeds were refrigerated (275 K) to
minimize ring closure during storage (<24 h), but some
amount of GVL formation was inevitable. Zero-time concen-
trations of HPA, LA, and GVL were thus rigorously determined
by HPLC analysis (see online Supporting Information) prior to
starting each experiment. Concentrations of HPA (0.04−0.06
M), LA (0.15−0.45 M), and GVL (5 × 10−4 to 1.5 × 10−3 M)
varied depending upon the intended experiment. Water used in
preparing reactor feeds, calibration standards, and HPLC
mobile phases was purified in house by sequential reverse
osmosis, UV oxidation, and double ion exchange. H2 (99.999%,
Airgas), N2 (99.999%, Airgas), CO (99.99% Praxair) employed
in flow systems and chemisorption experiments were used
without further purification.
Scheme 1. Pathways Leading to γ-Valerolactone during
Hydrogenation of Levulinic Acid
choice of solvent,46,48−50 presence of metal promoters (e.g., Re
or Sn),42,44 and presence of residual impurities in levulinic acid
feeds (e.g., H2SO4 or acid-soluble lignin),42 Ru-based catalysts
can display pronounced differences in both hydrogenation
activity20,36 and on-stream stability.42
Here, we present an investigation of the reaction pathways
and kinetics of aqueous-phase LA hydrogenation over
supported Ru, which is representative of LA feedstocks
obtained via acid hydrolysis of lignocellulose. Kinetic studies
have been carried out only for monometallic Ru/C having a
single metal loading (5 wt %). Our motivation in doing so is to
establish governing phenomena on a practically employed
catalyst in the absence of confounding effects, thus providing a
foundation for subsequent studies examining the influence of
Ru promoters and feed impurities in greater detail. In
consideration of this model system, we have decoupled HPA-
and angelicalactone-mediated hydrogenation pathways to
illustrate that GVL formation occurs primarily through ketone
hydrogenation, followed by intramolecular esterification (path-
way 1, Scheme 1). Upon identifying acid catalyzed intra-
molecular esterification of HPA as the kinetic bottleneck in
GVL formation at low temperatures, we demonstrate that high
yields of GVL can be achieved at 323 K using stacked beds of
Ru/C followed by Amberlyst 15 (A15).
2.1. Kinetic Studies. Kinetic data were collected for three
separate reactions illustrated in Scheme 1. First, the rate of LA
hydrogenation over Ru/C was determined in a packed bed
reactor under a H2 atmosphere. Second, LA dehydration rates
were quantified under typical LA hydrogenation conditions
over Ru/C in a packed bed reactor under a nonreducing
atmosphere. Finally, HPA esterification to form GVL was
carried out in a batch reactor. Details for each experiment are
provided in the following subsections.
2.1.1. Levulinic Acid Hydrogenation. Prior to use in kinetic
studies, 5 wt % Ru/C was graded through a series of standard
mesh sieves. Unless otherwise noted, particles in the range of
45−90 μm were used. Before loading into reactors, the stock
catalyst was diluted to 1 g 5 wt % Ru/C per 99 g 45−90 μm
quartz particles, which were obtained by milling and grading
fused quartz granules (4−20 mesh SiO2, Sigma Aldrich). One
to five milligrams of undiluted 5 wt % Ru/C was generally
required for typical reactor operation; as such, we employed a
high dilution ratio to improve precision in catalyst mass
loading. Reference experiments confirmed that hydrogenation
rates were independent of the dilution scheme, suggesting that
this protocol does not induce bypassing of the catalyst particles.
Hydrogenation of LA was carried out using the above-
described Ru/C dilution in a concurrently fed, stainless steel
packed bed reactor (1/4 in. × 12 in.). The reactor was operated
in an upflow configuration over a range of temperatures (323−
423 K) and H2 pressures (4.1−41.5 bar). Relatively small
amounts of catalyst were required for all of the studies reported
here, and bed lengths were on the order of 0.5−1.0 in. Catalyst
beds were positioned at the center of the heated section of the
reactor and held in place by two quartz wool plugs. The tube
upstream of the catalyst bed was packed with coarse (850−
1200 μm) quartz granules to minimize dead volume, and the
section immediately downstream of the catalyst bed was packed
with smaller quartz granules (<45 μm) to minimize entrain-
ment of carbon fines. Quartz packing was fixed in place with
two quartz wool end plugs, and the reactor was placed in line
using compression fittings.
2. MATERIALS AND METHODS
Ru/C (5 wt %) was purchased from Strem Chemicals, and its
surface area and average pore diameter were determined to be
756 m2/g and 5.04 nm via N2 adsorption at 77 K. Ru dispersion
was calculated to be 40.4% from irreversible CO uptake at 308
K, assuming 1:1 adsorption stoichiometry. For 5 wt % Ru/C,
this corresponds to a Ru site density of 200 μmol g−1. Both N2
and CO adsorption were performed using volumetric dosing in
a commercial system (Micromeritics ASAP 2020). Before N2
dosing experiments, samples were heated under vacuum (6 h,
623 K) to remove physisorbed impurities. Samples were
prepared for CO chemisorption by reducing in flowing H2 (3 h,
623 K, 3 K min−1), evacuating at 623 K (1 h) to remove
chemisorbed hydrogen, and cooling to 308 K under vacuum.
Analysis was performed by collecting a CO adsorption isotherm
at 308 K, evacuating the sample at 308 K (1 h) to remove
physisorbed CO, and collecting a second isotherm at 308 K.
Irreversible uptake was determined as the difference in CO
uptake between the two isotherms.
1172
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
Aqueous LA feeds (0.025−1.5 M) were introduced to the
system using an HPLC pump (Lab Alliance Series 1). H2 feeds
were regulated by mass flow controller (Brooks 5850S) and
mixed with the aqueous feed prior to introduction into the
reactor. To ensure thermal equilibration of the reactor feed,
both the H2 and the aqueous LA feeds were circulated through
a preheater section that was maintained at reactor temperature.
Feed temperature was monitored by an in-line type K
thermocouple and controlled using a PID controller (Love
16A-3010). The combined feed was then introduced to the
packed bed, which was positioned in the center of a 2 in.
aluminum rod held within a ceramic furnace (Applied Test
Systems). Reactor temperature was monitored at the external
wall of the catalyst bed using a type K thermocouple and
controlled with a PID controller. System pressure was
controlled using a back-pressure regulator (Tescom model
26-1766-24) and monitored both upstream and downstream of
the packed bed using analog pressure gauges. The effluent of
the system flowed directly into a vapor−liquid separator.
Because no volatile products are formed during this reaction,
the gas phase was vented continuously without analysis. The
aqueous product was collected at regular intervals (15−30 min)
and analyzed immediately using HPLC (Agilent 1100 Series).
Full details of the analytical protocols are given in the
Supporting Information. Prior to use, catalyst beds were
reduced in situ (673 K, 4 h hold, 1 K min−1) under flowing H2
(100 SCCM, 1 bar).
Since LA hydrogenation occurs in a three-phase system, we
anticipate that interphase, interparticle, and intraparticle
transport limitations may govern concentrations of LA,
dissolved H2, and temperature at catalytic centers and thus
control the rate of hydrogenation. A thorough consideration of
transport and kinetically controlled operating regimes, along
with the calculation of dissolved H2 concentrations, is
presented in the Supporting Information. For the experiments
described here, we observed that interphase H2 transport is
rapid such that dissolved H2 concentrations are well-
approximated by their equilibrium value, as determined by
gas-phase H2 pressures and Henry’s law. Further, linear fluid
velocities at or beyond volumetric aqueous feed rates of 0.4 mL
min−1 are sufficiently high to eliminate any external
concentration or temperature gradients. Finally, for 5 wt %
Ru/C, intraparticle transport limitations were negligible at and
below 343 K for catalyst particles smaller than 125 μm.
Where higher conversions are not otherwise indicated,
reactors operated differentially (<3% LA conversion), and the
only reaction products observed were HPA and GVL. Because
conversion ranges for differential operation are within the
precision expected of mass balance closure and HPLC analysis,
LA conversion and product selectivity were determined on the
basis of product formation as defined in eqs 1 and 2, where
nHPA and nGVL are the total molar quantities of HPA and GVL
recovered in a given reactor sample, and nLA0 is the total molar
quantity of LA fed into the system. Yield is calculated as the
product of LA conversion, XLA, and selectivity, Si.
Site time yields (STY) are reported for both HPA and GVL
over a range of conversions. They were calculated according to
eq 3.
F
i
STY =
i
mcat·φRu
(3)
where Fi is the effluent molar flow rate of either HPA or GVL,
m
cat is the mass of catalyst in the bed, and φRu is the loading of
Ru surface sites per gram of catalyst as determined by CO
chemisorption. In all data summarized here, site time yields
have units of moles product per mole of Ru surface sites per
second, which is reported as s−1 for convenience.
2.1.2. LA Dehydration. To probe the extent to which
angelicalactones contribute to GVL formation, the baseline rate
of LA dehydration was determined in an inert atmosphere over
Ru/C. This experiment was motivated by the fact that CC
bond hydrogenation is anticipated to be rapid over Ru/C
relative to LA dehydration, making it difficult to distinguish
between GVL formed via LA dehydration and LA hydro-
genation in a reducing environment. To provide an accurate
accounting of GVL fractions formed by pathways 1 and 2
(Scheme 1), it is thus necessary to quantify the rate of
dehydration independent of hydrogenation. To this end, a
reference experiment was carried out identically to the protocol
described in section 2.1.1 (LA Hydrogenation), with the
exception that the H2 feed was replaced with N2 to suppress
both LA and angelicalactone hydrogenation. Liquid phase
products were analyzed by HPLC and gas chromatography, as
described in the Supporting Information.
2.1.3. Intramolecular Esterification of HPA. The intra-
molecular esterification (ring closure) of 4-hydroxypentanoic
acid (HPA) to form GVL was studied in batch reactors in dilute
aqueous solutions of LA. A feed HPA solution was prepared by
partial hydrogenation of LA, stored at 275 K to inhibit ring
closure, divided into aliquots, and loaded into magnetically
stirred autoclave reactors (10 mL) that were subsequently
placed in a temperature-controlled oil bath. Batch experiments
were conducted at ambient pressure and temperatures ranging
from 300 to 339 K. To determine whether ring closure is
heterogeneously catalyzed in this system, identical experiments
were carried out with the addition of both Ru/C and quartz
granules (0.05 g/g feed). Batch vessels were stirred at 700 rpm
to eliminate extraparticle gradients, and solid particles of 45−90
um were employed to minimize internal diffusion limitations
where relevant. LA, HPA, and GVL concentrations were
monitored as a function of time by withdrawing 200 μL
aliquots from the reactor using a syringe. Each sample was
analyzed by HPLC as described in the Supporting Information.
Intramolecular esterification of HPA was the only reaction
observed in these experiments, and LA concentrations
remained constant throughout. Solution pH was measured at
ambient temperature at the beginning and end of every
experiment, and we observed a maximum change of 2.3% in
dissolved H+ concentration over the course of a single
experiment, indicating that HPA consumption did not have a
significant effect on H+ concentration. On the basis of
previously compiled data regarding the effect of temperature
on the dissociation of LA51 and related carboxylic acids,51−53
dissolved H+ concentrations were determined to vary minimally
(<5%) over the experimental temperature range (300−339 K).
This observation is consistent with prior results,54 and we
n4HPA + nGVL
XLA
=
n
LA0
(1)
(2)
ni
nHPA + nGVL
Si =
1173
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
Table 1. Experimentally Observed LA Conversions and Product Selectivities during Hydrogenation of LA over Ru/C under
Inert and Reducing Environments and at Temperatures Ranging from 323 to 423 K
STY (s−1
GVL
)
selectivity (%)
GVL
run
gas
T (K)
WHSV (hr−1
)
XLA (%)
HPA
AL
HPA
AL
1
2
H2
H2
H2
N2
N2
H2
H2
H2
H2
H2
323
323
323
323
423
343
363
383
403
423
550
37
1
0.09
0.09
0.06
0.004
0.002
0.003
96
98
95
0
4
2
0
20
99
trace
trace
2
0
0
3
5.0
5
4
550
550
730
720
920
790
1500
5 × 10−7
1 × 10−6
0
100
100
0
5
0
0
6
0.17
0.18
0.23
0.25
0.12
0.02
0.04
0.13
0.17
0.48
89
81
63
59
19
11
19
36
40
81
7
3
0
8
3
0
9
5
0
10
3
0
therefore consider the measured starting pH to be constant
throughout each experiment.
dehydration (10−6 s−1) remain well below hydrogenation rates
summarized in entries 1−3. Since LA dehydration is essentially
not observed over Ru/C relative to the scale of HPA and GVL
production over the range of temperatures considered here
(303−423 K), we conclude that GVL formation over Ru/C
occurs exclusively through the HPA-mediated pathway (path-
way 1, Scheme 1) in this temperature range.
2.2. Stacked Bed Experiments. Ru/C and Amberlyst-15
were coupled in stacked beds to facilitate both LA hydro-
genation and HPA ring closure in a single reactor at low
temperatures. To accommodate the relatively low thermal
stability of A15, Ru/C used in stacked beds was reduced ex situ
in H2 at 673 K and subsequently passivated at 298 K in 1% O2/
N2. Passivated Ru/C and unmodified A15 were sequentially
loaded into a single 1/4 in. stainless steel tube to create two
stacked beds separated by a quartz wool plug. The entire bed
was then reduced at 373 K (3 K min−1) in 100 sccm of H2 for 2
h to remove the oxygen monolayer from Ru surface sites.
Control experiments confirmed that Ru/C reduced and
passivated ex situ (used in stacked beds) performed identically
to Ru/C reduced in situ (used in LA hydrogenation kinetic
studies). As such, this experiment reveals only the effect of
adding secondary acid functionality downstream of the
hydrogenation system.
Entries 6−10 compare HPA and GVL selectivity and site
time yields at reaction temperatures from 343 K to 423 K. As
the reaction temperature increases, LA hydrogenation rates
increase, and GVL selectivity improves relative to HPA. This is
attributed to an increase in the rate of ring closure relative to
the rate of hydrogenation. Intramolecular esterification of HPA
is reversible such that HPA/GVL distributions in hydro-
genation products will ultimately be determined by chemical
equilibrium. A concentration-based equilibrium constant for
esterification of HPA in aqueous solution at 298 K was
experimentally determined here to be 14.5, which is in good
agreement with the mean value of those previously reported for
this reaction (13.8).55 HPA ring closure is estimated to be only
slightly exothermic (−3 kJ/mol), on the basis of prior accounts
of γ-hydroxybutyric acid esterification;56 thus, equilibrium GVL
selectivities are not expected to change drastically with
increasing temperature and should exceed 90% at each
experimental condition described here. This was confirmed
by experimental observation, where nearly complete GVL
selectivity was observed at chemical equilibrium over a range of
temperatures from 298 to 423 K. We therefore conclude that
HPA ring closure is not equilibrated in any of the experiments
described in Table 1 and attribute the improved GVL selectivity
to an increased rate of HPA esterification relative to the rate of
LA hydrogenation. This suggests that HPA ring closure
proceeds with a higher activation barrier than LA hydro-
genation and may also reflect increasing mass transfer
limitations in LA hydrogenation at high temperatures (see
the Supporting Information).
3. RESULTS AND DISCUSSION
3.1. Primary Analysis of LA Hydrogenation Products
Formed over Ru/C. Table 1 summarizes experimental
observations during the aqueous-phase hydrogenation of LA
over Ru/C.
From data in entries 1−3, we observe that hydrogenation of
LA over Ru/C at near ambient temperatures (323 K) yields
only two products, HPA and GVL, over LA conversions
ranging from 1 to 100%. Over the entire conversion range, we
note that GVL selectivity is poor (<5%) at 323 K, and the
major LA hydrogenation product is HPA (>95% selectivity).
This suggests that the HPA-mediated pathway illustrated in
Scheme 1 likely dominates GVL production and may indicate
that intramolecular esterification of HPA, rather than hydro-
genation of LA, controls the rate of GVL formation over Ru/C.
To further probe the extent to which angelicalactone
formation contributes to GVL production, we consider entry
4, which is identical to entry 1 except that the reducing gas
(H2) has been replaced with an inert (N2). Under these
conditions, we observe trace LA conversion with complete
selectivity to angelicalactones, indicating that in the absence of
H2, LA is consumed only by dehydration. However, the site
time yield to angelicalactones over Ru/C (5 × 10−7 s−1) is
several orders of magnitude lower than site time yields
observed for HPA (0.085 s−1) and GVL (0.005 s−1) in the
presence of hydrogen (entry 1). Even upon increasing the
system temperature to 423 K (entry 5), site time yields for
3.2. Kinetics of LA Hydrogenation. Because LA
dehydration is demonstrated to have a negligible contribution
to GVL production below 423 K (Section 3.1), the sum of
production rates for GVL and HPA can be taken as the total
rate at which the ketone group in LA is hydrogenated. Because
differential conditions were maintained in kinetic studies
reported hereafter, turnover frequencies for LA (ketone)
hydrogenation are well-approximated as the sum of site time
yields for HPA and GVL.
TOF = STYHPA + STYGVL
(4)
LA
1174
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
◊
□
○
Figure 1. (a) Rates of LA hydrogenation over Ru/C at 323 K ( ), 343 K ( ), and 363 K ( ) as a function of time on-stream. (b) Linearized rate
data illustrating second-order deactivation and method employed for estimation of rates at zero time on-stream. For all experiments summarized
here, the aqueous phase concentrations of LA and H2 were 0.50 and 0.016 M.
Figure 1 plots turnover frequencies of LA hydrogenation
over Ru/C as a function of time on-stream at several
representative reaction conditions. It is evident that LA
hydrogenation rates decay rapidly in this aqueous-phase system,
even at mild temperatures (323 K). Ultimately, we were unable
to identify an operating regime in which deactivation does not
occur; as such, initial turnover frequencies at a given
experimental condition were estimated by extrapolation of
rate data to zero time on-stream. Independent of the reaction
conditions, the time decay of hydrogenation rates at short times
on-stream (below 8 h) was well-described by a second-order
model such that declining reaction rates can be linearized by
plotting inverse turnover frequencies against time on-stream, as
described by Bartholomew.57 Regression of linearized data sets
permits quantification of initial hydrogenation turnover
frequencies via estimation of a y-intercept, and all initial
turnover frequencies reported here for LA hydrogenation were
approximated using this method.
Figure 2. Observed LA hydrogenation rates as a function of time on-
Detailed consideration of Ru/C deactivation is outside the
focus of this article; however, some discussion is warranted
because of the ubiquity of noble metal-on-carbon catalysts in
emerging aqueous phase hydrogenation processes targeting
biorenewables. Figure 2 summarizes on-stream performance of
Ru/C during LA hydrogenation at 323 K. Here, we observe
that TOFs for LA hydrogenation ultimately stabilize at 20−30%
of their original values after 48−72 h on stream. Regeneration
attempts via in situ reduction reveal that this loss of activity
occurs through a combination of reversible and irreversible
phenomena. After reaching steady state, catalysts can be
restored to only 50−60% of their initial activity, and their
renewed activity decays quickly (within 5 h) to the steady state
value after being placed on-stream again. To date, the mode of
reversible deactivation has not been conclusively identified, but
it may be attributed to either strongly bound hydrocarbon
intermediates or surface oxidation of Ru nanoparticles.
With respect to irreversible deactivation, we have not
observed Ru leaching or attrition of the carbon support in
this system; neither does the carbon support undergo
significant physical changes as determined by N2 adsorption.
As such, irreversible deactivation observed here is most likely
attributed to particle sintering. Ru dispersion, determined by
CO chemisorption, decreases from 40% to 21% in pre- and
postreaction samples. Because leaching, attrition, and structural
○
stream at T = 323 K, CH = 0.016 M, CLA = 0.1 M. ( ) Fresh catalyst,
2
◊
□
(Δ) first regeneration, ( ) second regeneration, ( ) third
regeneration. Regeneration was performed in situ by interrupting
aqueous feeds and reducing the catalyst under flowing H2 (100 sccm, 4
h, 673 K, 1 K min−1).
degradation have not been observed, this loss of dispersion
suggests an increase in Ru cluster size from roughly 3.6 nm in
fresh samples to 6.8 nm in samples recovered after 65 h on-
stream. Sintering is typically considered a high temperature
phenomenon; however, water can facilitate particle agglomer-
ation close to room temperature.58 These observations are
supported by the works of Davis, who reported metal cluster
growth during aqueous phase hydrogenation of glucose over
Ru/SiO2 at 373 K,59 and Marin, who observed particle growth
from 2.2 to 3.2 nm during the aqueous phase oxidation methyl-
α-D-glucoside over Pt/C.60 Sintering is thus a realistic
consideration in this system, and future efforts geared toward
rational design of stable hydrogenation catalysts for biomass
processing should consider strategies for maintaining high Ru
dispersions in the aqueous phase. In the interest of brevity and
maintaining a focus on the kinetics of LA hydrogenation, we
will defer a more comprehensive characterization of spent
catalyst samples to subsequent communications. In the
1175
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
remainder of this section, we discuss observed trends in initial
LA hydrogenation turnover frequency as a function of LA
concentration, dissolved H2 concentration, and reaction
temperature.
3.2.1. Reaction Orders. Levulinic Acid. The concentration of
LA in the aqueous phase was found to have a minimal and
slightly inhibitory effect on the initial rate of hydrogenation.
This trend is illustrated in Figure 3, and regression of the data
Figure 4. Correlation between dissolved aqueous phase hydrogen
concentration and hydrogenation rate. Experiments illustrated here
were carried out at 323 K, 0.5 M LA, and varying WHSV to maintain
LA conversions below 3%. Turnover frequencies are reported in s−1,
and H2 concentrations are reported in mol L−1.
were observed by Vannice during benzene and acetic acid
hydrogenation over supported Pt,63,67 Dumesic during ethylene
hydrogenation over Pt,65 and Mahajani in the hydrogenation of
n-valeraldehyde68 and iso-valeraldehyde69 over supported Ru.
Reconciliation between this observation and a proposed
reaction pathway will be discussed in subsequent analysis
(section 3.2.3). At this stage, our consideration of apparent
reaction orders suggests that LA hydrogenation is well-
represented under the conditions reported here by the
empirical rate law given in eq 5.
Figure 3. Correlation between levulinic acid concentration and
hydrogenation rate. Experiments illustrated here were carried out at
323 K, 0.016 M H2 concentration, and varying WHSV to maintain LA
conversions below 3%. Turnover frequencies are reported in s−1, and
LA concentrations are reported in mol L−1.
there reveals an apparent reaction order of −0.04 0.04 over a
range of LA concentrations (0.025−1.5 M), indicating that the
rate of LA hydrogenation is nearly independent of bulk LA
concentrations over practical values. This observation is
consistent with prior studies of both gas and liquid phase
reactions in which apparent zero-order kinetics with respect to
the hydrocarbon are generally observed for both CO and
CC hydrogenations over group VIII metals. For example,
both hydrogenation of D-glucose61 and arabinonic acid62 are
zero-order in the oxygenate over Ru/C above concentrations of
0.3 and 1 M, respectively. Other supported, noble metals
exhibit similar behavior. For example, Vannice has reported that
benzene63 and citral64 hydrogenation over supported Pt and Pd
catalysts, respectively, are zero-order in organic concentration.
Similarly, for the gas phase hydrogenation of ethylene over Pt,
Dumesic observed zero-order dependence on ethylene at low
temperatures and high partial pressures of ethylene.65 Apparent
zero-order dependencies on organic species during hydro-
genation is typically attributed to the presence of strongly
bound hydrocarbon intermediates that saturate available metal
surface sites during hydrogenation.66 A more detailed
interpretation will be provided in section 3.2.3.
1/2
H2
rLA = k′·c
(5)
3.2.2. Temperature Effects. Assuming the empirical rate law
derived in the preceding section (eq 5) is valid over the range
of experimental conditions tested, we can estimate apparent
rate constants for the hydrogenation of LA by normalizing
measured turnover frequencies by the square root of the
concentration of dissolved hydrogen. Subsequently, we may
examine their temperature dependence to extract apparent
activation energies and pre-exponential factors for this reaction.
Apparent rate constants are plotted on a logarithmic scale
against inverse temperature in Figure 5, which reveals two
distinct regimes. At relatively high temperatures (363−423 K),
the data suggest an apparent activation barrier of 20
6 kJ
mol−1, which is lower than that anticipated of kinetic rate
control and likely indicates that internal pore diffusion
dominates at higher temperatures. This observation is
consistent with our preliminary analysis of the Weisz−Prater
criteria (see Supporting Information), which suggests that pore
diffusion becomes significant at temperatures around 363 K. At
lower temperatures (303−343 K), regression of linearized rate
data indicate an apparent activation energy of 48 5 kJ mol−1
and an apparent pre-exponential factor of roughly 3.1 × 107 L0.5
mol−0.5 s−1 on a Ru site basis or 6.2 × 103 (L·mol)0.5 (g·s)−1 on
a catalyst mass basis. Importantly, data in the low temperature
region were demonstrated to be free of internal diffusion
limitations through the observation of invariant hydrogenation
turnover frequencies as a function of mean particle size at 343
K (see Supporting Information). In line with our observations,
Hydrogen. Figure 4 illustrates the dependence of hydro-
genation turnover frequency on the concentration of dissolved
hydrogen in the aqueous phase. Here, a pronounced effect is
observed, and regression of the rate data reveals a fractional
order of 0.6 0.2. This outcome suggests that H2 adsorption is
rapid, such that surface reactions involving the addition of
atomic hydrogen to adsorbed organic species control the rate of
LA hydrogenation. Similar half-order hydrogen dependencies
1176
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
of hydrogenation would exhibit apparent reaction orders of −1
in LA and 0 in H2, which are significantly different from our
observed apparent orders of −0.04 and 0.6 for LA and H2,
respectively. Alternatively, our preliminary density functional
theory results suggest that on flat terrace sites of Ru, adsorbed
H-atoms are the dominant surface species. In this case,
competitive adsorption of H2 and hydrocarbon species would
predict apparent reaction orders of +1 in LA and 0 in H2, which
is again significantly different from our observations. Data
collected in the regimes summarized here are thus more easily
reconciled with the assumption that surface-bound hydro-
carbons and hydrogen atoms adsorb noncompetitively and can
be considered in separate site balances. In the case of
noncompetitive adsorption, we predict via Langmuir−Hinshel-
wood analysis that the overall hydrogenation rate should take
the form given by eq 11.
k4K1K2K3CLAC
H2
r =
(1 + K1CLA + K1K21/2K3CLAcH 1/2)(1 + K21/2cH
)
1/2
2
Figure 5. Correlation between reaction temperature and levulinic acid
hydrogenation apparent rate constant. Experiments illustrated here
were carried out at 0.5 M LA, 0.016 M H2, and varying WHSV to
maintain LA conversions below 3%. Apparent rate constants here are
reported in L0.5 mol0.5 g−1 s−1.
2
(11)
Considering that the half-hydrogenated LAH intermediate is
likely bound strongly and difficult to hydrogenate, we may
apply the simplifying assumptions that the coverage of the
hydrocarbon intermediate approaches saturation on sites
available for hydrocarbon adsorption, and the coverage of
levulinic acid approaches zero. The assumption of a small LA
coverage is reasonable because close structural analogs, such as
2-butanone, are reported to bind weakly relative to hydro-
genation intermediates at Ru sites.73 Because we observe a
distinct, positive reaction order with respect to hydrogen, we
additionally assume that sites accessible to atomic hydrogen are
far from saturation. Applying these limiting assumptions, the
overall hydrogenation rate manifests as in eq 12, which
reconciles well with our observed reaction orders of −0.04 and
0.6 in LA and H2.
prior studies have reported activation barriers ranging from 34
to 64 kJ mol−1 over supported Ru for cinnamaldehyde,70 D-
lactose,71 D-glucose,61 and arabinonic acid.62 Similar activation
energies of hydrogenation were also reported over supported Pt
for benzene63 and ethylene.65 We therefore consider that an
apparent barrier on the order of 48 kJ mol−1 is reasonable for
levulinic acid hydrogenation over Ru/C.
3.2.3. Analysis of Kinetic Data. The observed reaction
orders and apparent activation energy align well with previously
documented studies considering both CO and CC
hydrogenation. All observations can be reconciled with a
conventional Horiuti−Polanyi interpretation,72 which is illus-
trated in a simplified scheme of LA hydrogenation in eqs 6−9.
Briefly, according to this mechanism, LA adsorbs molecularly
and H2 adsorbs dissociatively at Ru surface sites. Surface-bound
LA is subsequently reduced in two steps by sequential addition
of hydrogen atoms, ultimately forming the hydroxyacid (HPA),
which we assume desorbs irreversibly from the surface.
1/2
H2
r = k4·K21/2c
(12)
3.3. Kinetics of the Intramolecular Esterification of
HPA. As demonstrated in section 3.1, intramolecular
esterification of HPA is the final step in the low-temperature
production of GVL. Examination of selectivity trends suggests
that HPA esterification is kinetically significant in the
production of GVL over Ru/C, and we expect that this step
is acid catalyzed. In the system considered here, the primary
source of acidity is likely solvated protons dissociated from LA
and HPA, both of which are weak organic acids having pKa
values of 4.59 and 5.69,55 respectively. However, because
oxidized carbon and Ru may also exhibit some acidity, we
cannot conclude a priori that ring closure is exclusively a
homogeneous reaction in this system. Prior to investigating
reaction kinetics, control experiments were carried out to
determine whether Ru/C and quartz influence the rate of HPA
ring closure. At 303 K, we observed no difference in
esterification rates observed with and without the addition of
Ru/C and quartz, indicating that heterogeneous reactions do
not contribute to HPA ring closure during hydrogenation.
Subsequent sections describe experiments designed to capture
reaction orders and temperature dependencies in the intra-
molecular esterification of HPA.
(6)
1. LA + * ↔ LA*
2. H2 + 2* ↔ 2H*
(7)
(8)
3. LA* + H* ↔ LAH* + *
(9)
4. LAH* + H* → HPA + 2*
On the basis of prior computational studies,73 the half-
hydrogenated intermediate (LAH, formed in step 3) is
expected to be the dominant hydrocarbon species bound to
Ru surface sites, and the second addition of atomic hydrogen
(step 4) is generally considered to be rate-determining. This
assumption will yield the overall rate expression for LA
hydrogenation given by eq 10.
r = k4·θLAHθH
(10)
Half-hydrogenated species are expected to adsorb favorably
on Ru surface sites under most conditions.66 If we make the
assumption that H2 adsorption and hydrocarbon adsorption are
fully competitive and that the two species occupy identical
surface sites, then we would expect (in the limit of a surface
dominated by bound hydrocarbon intermediates) that the rate
3.3.1. Reaction Orders. On the basis of the intramolecular
esterification mechanism, we expect that the rate of ring closure
1177
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
+
+
▽
Figure 6. (a) Batch kinetic data for intramolecular esterification of HPA at (Δ) 300 K, CH = 1.65 mM, CHPA = 47.7 mM, ( ) 318 K, CH = 1.65
0
+
+
□
○
mM, CHPA = 47.7 mM, ( ) 339 K, CH = 1.65 mM, CHPA = 47.7 mM, ( ) 339 K, CH = 2.69 mM, CHPA = 41.2 mM. (b) Apparent rate constants
0
0
0
−1
○
for intramolecular esterification of HPA determined at (Δ) 339 and ( ) 318 K at various pH values. k′ is reported here in units of min .
will exhibit some dependence on the concentrations of HPA
and dissolved protons, as summarized in eqs 13 and 14.
α
r = k′·CHPA
(13)
+ β
k′ = k·CH
(14)
For a set of batch kinetic studies having varied HPA
concentrations, LA concentrations, and reaction temperatures,
Figure 6a illustrates HPA concentration normalized by its initial
value as a function of time on a logarithmic scale. The
demonstration of linearity in each data set indicates that HPA
ring closure is first-order in HPA concentration, and regression
of each data set yields an apparent rate constant into which the
concentration of dissociated protons has been lumped (eq 14).
Figure 6b plots these apparent rate constants calculated against
the measured concentration of solvated protons on a
logarithmic scale, and examination of the correlation between
+
k′ and CH again reveals a first-order dependence. First-order
Figure 7. Arrhenius plot illustrating the temperature dependence of
dependencies on HPA and proton concentrations are
rate constants for intramolecular esterification of HPA. Apparent rate
consistent with prior descriptions of the intramolecular
constants are reported here in units of L mol−1 min−1.
esterification of HPA in homogeneous systems.74,75
3.3.2. Temperature Effects. Having established reaction
orders for both HPA and H+, the rate constant, k, can be
derived from experimental observations summarized to this
explicitly determined, and batch kinetic studies were carried out
at various temperatures to quantify Arrhenius parameters. Rate
constants determined at each temperature are plotted on a
logarithmic scale against inverse temperature in Figure 7.
Regression of this data indicates that the pre-exponential factor
and activation barrier for the intramolecular esterification of
HPA are, respectively, 2.0 × 1010 L mol−1 s−1 and 70 0.4 kJ
mol−1. We note that this barrier is slightly higher than that
observed for LA hydrogenation (48 kJ mol−1, Section 3.2.2).
This result is consistent with our observations that ring closure
appears to control GVL production rates at low temperatures
and that GVL selectivities improve relative to HPA with
increasing reaction temperature.
3.4. Kinetic Model for GVL Production via LA
Hydrogenation. With reaction orders and temperature
dependencies for both heterogeneously catalyzed LA hydro-
genation and homogeneously catalyzed HPA ring closure
established, the two reactions can be modeled independently to
predict both LA hydrogenation rates and GVL selectivities over
Ru/C. The reaction pathway and empirical kinetic parameters
point are compiled in Scheme 2.
Using these parameters and rate equations, packed bed
reactors employed for LA hydrogenation were modeled to
reconcile predicted hydrogenation rates and product selectiv-
ities with those observed experimentally. The system can be
described using the material balances summarized in eqs
15−17.
dF
dV
LA
1/2
2
= −k1·ρb·cH
= k1·ρb·cH
(15)
(16)
(17)
dFHPA
dV
1/2
+
− k2·cHPA·cH
2
dFGVL
dV
+
= k2·cHPA·cH
The catalyst mass in the system is calculated as a product of
the system volume and ρb, the bulk density of the catalyst bed,
and the concentration of solvated protons is calculated on the
basis of acid dissociation constants and concentrations of LA
1178
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
Scheme 2. Summary of the Empirical Kinetic Model Employed Here To Predict LA Hydrogenation Rates and GVL Selectivities
in a Differential Packed Bed Reactor Operating between T = 303 and 343 K, CLA = 0.025−1.5 M, and CH = 0.0028−0.028 M
2
Figure 8. Comparison between model predicted trends and experimental observations at various LA and H2 concentrations below 363 K. (a) Trends
observed in GVL production rates, which captures selectivity during LA hydrogenation. (b)Trends in TOF of LA hydrogenation.
and HPA. As detailed in the Materials and Methods section,
carboxylic acid dissociation is expected to vary minimally over
the range of temperatures considered here, and its effect was
neglected in our calculations. Modeling this system is
complicated slightly by the fact that LA hydrogenation is
heterogeneously catalyzed on Ru sites, and HPA ring closure is
homogeneous. Thus, LA hydrogenation occurs only in the
relatively small Ru/C bed, whereas HPA ring closure occurs
homogeneously throughout both the Ru/C bed and in reactor
sections filled with inert packing. To address this issue, reactors
were simulated as being composed of separate hydrogenation
and inert sections, and the effluent from the hydrogenation
section was used as the input to the inert section. As illustrated
in Figure 8, we were able to reconcile, without further
adjustment of parameters, predicted LA hydrogenation turn-
over frequencies and GVL production rates in each of the
differential kinetic studies summarized in sections 3.2.1 and
3.2.2, indicating that the empirical model developed here is
sufficient for predicting both LA conversion and GVL
selectivity in kinetically controlled regimes. Since the catalysts
employed here are deactivating, the steady state balances given
by eqs 15−17 apply only for prediction of initial LA
hydrogenation and GVL production rates, and the data
summarized in Figure 8 indicate values estimated at zero
time on-stream. On the basis of replicates of rate data collected
at a well-defined reference condition (CLA = 0.5 M, CH = 0.016
2
M, T = 323 K), rates of LA hydrogenation and GVL production
measured in this system deviate by roughly 5% from the mean,
and the error bars in Figure 8 reflect this uncertainty.
3.5. Stacked Bed Reactors. Results from the preceding
sections indicate that LA hydrogenation turnover frequencies
are reasonable at near ambient temperatures (e.g., ∼0.08 s−1 at
323 K, 0.5 M LA, 0.016 M H2) and that transport limitations
become significant as temperatures increase substantially
beyond this point. At low temperatures, we additionally
observe that HPA ring closure appears to control the rate of
GVL production such that GVL selectivity is <5%, independent
of LA conversion. Increasing reaction temperature improves
GVL selectivity because ring closure has a slightly larger
activation barrier (70 kJ mol−1) than LA hydrogenation (48 kJ
1179
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
mol−1), and we observe nearly 80% selectivity to GVL during
differential LA hydrogenation at 423 K. However, high
temperature operation is perhaps an inefficient choice for
increasing productivity over a 5 wt % Ru/C catalyst. Pore
diffusion quickly becomes rate controlling such that dramatic
increases in reaction temperature yield only marginal enhance-
ment in hydrogenation rates (e.g., 0.08 s−1 at 323 K compared
with 0.6 s−1 at 423 K); thus, the additional investment in energy
input is not fully recovered as substantially decreased residence
times for LA hydrogenation. Results summarized thus far
demonstrate that the rate of HPA esterification scales with
proton concentration. Therefore, an alternate approach to
improving GVL selectivity is to introduce a second, acidic
catalyst to expedite ring closure. Although many homogeneous
and heterogeneous combinations of Ru and acidity are likely
possible, we have employed a simple, stacked bed of Ru/C
followed by A15 to facilitate sequential hydrogenation of LA
and intramolecular esterification of HPA. Although the same
outcome could be more elegantly accomplished using a
bifunctional catalyst, introducing acidity in parallel to hydro-
genation may encourage angelicalactone formation and, thus,
alter the kinetics of GVL formation according to Scheme 1. By
separating metal and acid functionalities, we were able to
directly probe the hypothesis that expediting HPA ring closure
improves GVL production rates without altering LA hydro-
genation rates. Observed stacked-bed hydrogenation rates and
product selectivities are summarized in Table 2.
CONCLUSION
■
We have shown that LA can be hydrogenated at near ambient
temperatures, proceeding primarily through a HPA-mediated
pathway in which hydrogenation occurs first and is followed by
acid-catalyzed dehydration. At low temperatures, intramolecular
esterification of HPA appears to control the rate of GVL
formation, whereas at high temperatures, mass transfer limits
the rate of hydrogenation. By recognizing this and developing
an understanding of the kinetics of the sequential hydro-
genation and dehydration steps, we have demonstrated that it is
possible to leverage a strongly acidic catalyst in conjunction
with a hydrogenation metal to significantly improve GVL
production rates at low temperatures. With respect to rational
design, it appears that LA hydrogenation is best-suited to
bifunctional catalysts exhibiting hydrogenation functionality
alongside acidity. We have additionally observed pronounced
deactivation of monometallic Ru/C in this model system, even
under mild conditions. Although underlying causes have not
been conclusively established, sintering appears to cause an
irreversible loss of activity, suggesting that it is critical to
identify strategies for retaining metal dispersion during aqueous
phase hydrogenation.
ASSOCIATED CONTENT
* Supporting Information
■
S
Consideration of interphase transport of H2 to allow calculation
of dissolved H2 concentrations; consideration of interparticle
heat and mass transfer on LA hydrogenation rates; consid-
eration of intraparticle heat and mass transfer on LA
hydrogenation rates; operational limits for kinetic control in
LA hydrogenation; analytical procedures for quantification of
LA, GVL, and HPA in reactor samples. This information is
available free of charge via the Internet at http://pubs.acs.org/.
Table 2. Experimentally Observed LA Conversions and
Product Selectivities during Hydrogenation of LA over Ru/C
a
and A15
selectivity
(%)
STY (s−1
)
b
c
WHSV
(h−1
WHSV
XLA
(%)
TOF
(s−1
(h−1
)
)
HPA
GVL HPA GVL
AUTHOR INFORMATION
Corresponding Author
*Address: Department of Biomedical and Chemical Engineer-
ing, 329 Link Hall, Syracuse University, Syracuse, NY 13244.
Phone: 315-443-2550. E-mail: jqbond@syr.edu.
run
)
■
1
2
3
4
550
560
5
0
16.8
0
2
2
0.107
0.093
0.104
0.074
0.056
0.007
0.003
0.019
0.003
0.048
97
80
95
12
3
20
5
99
92
5
0.15
88
a
T = 323 K, CLA = 0.5 M (Entries 1 and 2), 0.1 M (Entries 3 and 4),
Notes
b
CH = 0.016 M. LA weight hourly space velocity based on Ru/C
The authors declare no competing financial interest.
2
c
loadings. LA weight hourly space velocity based on Amberlyst-15
ACKNOWLEDGMENTS
loadings.
■
The authors acknowledge financial support for this work from
the National Science Foundation (CBET, Award No.
1159739). We also thank William P. Dossert and Richard W.
Chave (Syracuse University College of Engineering) for their
invaluable support in the design and manufacture of laboratory
equipment.
Control experiments carried out under differential conditions
(entries 1 and 2) illustrate that, despite employing different Ru/
C pretreatment protocols (i.e., in situ vs ex situ reduction),
hydrogenation rates over Ru/C are not altered in the stacked
bed reactor. Entry 2 shows that even a small quantity of A15
resin drastically shifts product selectivity toward GVL; however,
conversions (<2%) and selectivities (<20%) remain below what
would be desirable in a practical implementation. Entry 3
demonstrates that complete conversion of LA via hydro-
genation over Ru/C at 323 K is achieved at a weight hourly
space velocity of 5 h−1; however, GVL selectivity remains poor
over the metal catalyst. By adding a second bed of A15 (entry
4), product selectivity can be shifted almost entirely to GVL at
323 K, facilitating nearly 80% yield of GVL at practical
residence times in a single reactor.
REFERENCES
■
(1) Horvath, I. T.; Mehdi, H.; Fabos, V.; Boda, L.; Mika, L. T. Green
Chem. 2008, 10, 238−242.
(2) Mehdi, H.; Fabos, V.; Tuba, R.; Bodor, A.; Mika, L. T.; Horvath,
I. T. Top. Catal. 2008, 48, 49−54.
(3) Lange, J. P.; Price, R.; Ayoub, P.; Louis, J.; Petrus, L.; Clarke, L.;
Gosselink, H. Angew. Chem. Int. Ed. 2010, 49, 4479−4483.
(4) Bond, J. Q.; Martin Alonso, D.; Wang, D.; West, R. M.; Dumesic,
J. A. Science 2010, 327, 1110−1114.
(5) Bond, J. Q.; Martin Alonso, D.; West, R. M.; Dumesic, J. A.
Langmuir 2010, 26, 16291−16298.
(6) Bond, J. Q.; Wang, D.; Martin Alonso, D.; Dumesic, J. A. J. Catal.
2011, 281, 290−299.
1180
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181

ACS Catalysis
Research Article
(7) Serrano-Ruiz, J. C.; Braden, D. J.; West, R. M.; Dumesic, J. A.
Appl. Catal., B 2010, 100, 184−189.
(42) Braden, D. J.; Henao, C. A.; Heltzel, J.; Maravelias, C. C.;
Dumesic, J. A. Green Chem. 2011, 13, 1755−1765.
(43) Martin Alonso, D.; Wettstein, S. G.; Bond, J. Q.; Root, T. W.;
Dumesic, J. A. ChemSusChem 2011, 4, 1078−1081.
(44) Wettstein, S. G.; Bond, J. Q.; Martin Alonso, D.; Pham, H. N.;
Datye, A. K.; Dumesic, J. A. Appl. Catal., B 2012, 117−118, 321−329.
(45) Sen, S. M.; Gurbuz, E. I.; Wettstein, S. G.; Martin Alonso, D.;
Dumesic, J. A.; Maravelias, C. T. Green Chem. 2012, 14, 3289−3294.
(46) Al-Shaal, M. G.; Wright, W. R. H.; Palkovits, R. Green Chem.
2012, 14, 1260−1263.
(47) Upare, P. P.; Lee, J. M.; Hwang, D. W.; Halligudi, S. B.; Hwang,
Y. K.; Chang, J. S. J. Ind. Eng. Chem. 2011, 17, 287−292.
(48) Akpa, B. S.; D’Agostino, C.; Gladden, L. F.; Hindle, K.; Manyar,
H.; McGregor, J.; Li, R.; Neurock, M.; Sinha, N.; Stitt, E. H.; Weber,
D.; Zeitler, J. A.; Rooney, D. W. J. Catal. 2012, 289, 30−41.
(49) Wan, H.; Raghunath, A. V.; Chaudhari, V.; Subramaniam, B. J.
Catal. 2014, 309, 174−184.
(8) Martin Alonso, D.; Bond, J. Q.; Dumesic, J. A. Green Chem. 2010,
12, 1493−1513.
(9) Martin Alonso, D.; Bond, J. Q.; Serrano-Ruiz, J. C.; Dumesic, J. A.
Green Chem. 2010, 12, 992−999.
(10) Martin Alonso, D.; Bond, J. Q.; Wang, D.; Dumesic, J. A. Top.
Catal. 2011, 54, 447−457.
(11) Horvath, I. T. Green Chem. 2008, 10, 1024−1028.
(12) Wettstein, S. G.; Martin Alonso, D.; Chong, Y. X.; Dumesic, J.
A. Energy Environ. Sci. 2012, 5, 8199−8203.
(13) Martin Alonso, D.; Gallo, J. M. R.; Mellmer, M. A.; Wettstein, S.
G.; Dumesic, J. A. Catal. Sci. Technol. 2013, 3, 927−931.
(14) Gurbuz, E. I.; Gallo, J. M. R.; Martin Alonso, D.; Wettstein, S.
G.; Lim, W. Y.; Dumesic, J. A. Angew. Chem., Int. Ed. 2013, 52, 1270−
1274.
(15) Martin Alonso, D.; Wettstein, S. G.; Dumesic, J. A. Green Chem.
2013, 15, 584−595.
(16) Martin Alonso, D.; Wettstein, S. G.; Mellmer, M. A.; Gurbuz, E.
I.; Dumesic, J. A. Energy Environ. Sci. 2013, 6, 76−80.
(17) Luterbacher, J. S.; Rand, J. M.; Martin Alonso, D.; Han, J.;
Youngquist, J. T.; Maravelias, C. T.; Pfleger, B. F.; Dumesic, J. A.
Science 2014, 343, 277−280.
(50) Luo, W.; Deka, U.; Beale, A. M.; van Eck, E. R. H.; Bruijnincx, P.
C. A.; Weckhuysen, B. M. J. Catal. 2013, 301, 175−186.
(51) Walde, A. W. J. Phys. Chem. 1939, 43, 431−438.
(52) Martin, A. W.; Tartar, H. V. J. Am. Chem. Soc. 1937, 59, 2672−
2675.
(53) Harned, H. S.; Ehlers, R. W. J. Am. Chem. Soc. 1933, 55, 2379−
2383.
(18) Lange, J. P.; Vestering, J. Z.; Haan, R. J. Chem. Commun. 2007,
3488−3490.
(19) Manzer, L. E. Appl. Catal., A 2004, 272, 249−256.
(20) Wright, W. R. H.; Palkovits, R. ChemSusChem 2012, 5, 1657−
1667.
(54) Reijenga, J. C.; Gagliardi, L. G.; Kenndler, E. J. Chromatogr., A
2007, 1155, 142−145.
(55) Washburn, E. W. In International Critical Tables of Numerical
Data, Physics, Chemistry and Technology; McGraw-Hill: New York,
1930; Vol. 7, p 140.
(21) Lange, J. P.; van de Graaf, W. D.; Haan, R. J. ChemSusChem
́
(56) Perez-Prior, M. T.; Manso, J. A.; del Pilar García-Santos, M.;
2009, 2, 437−441.
Calle, E.; Casado, J. J. Org. Chem. 2004, 70, 420−426.
(57) Bartholomew, C. H. Appl. Catal., A 2001, 212, 17−60.
(58) Besson, M.; Gallezot, P. Catal. Today 2003, 81, 547−559.
(59) Maris, E. P.; Ketchie, W. C.; Oleshko, V.; Davis, R. J. J. Phys.
Chem. B 2006, 110, 7869−7876.
(22) Gurbuz, E. I.; Wettstein, S. G.; Dumesic, J. A. ChemSusChem
2012, 5, 383−387.
(23) Maldonado, G. M. G.; Assary, R. S.; Dumesic, J. A.; Curtiss, L.
A. Energy Environ. Sci. 2012, 5, 8990−8997.
(24) Gallezot, P. Chem. Soc. Rev. 2012, 41, 1538−1558.
(25) Van de Vyver, S.; Thomas, J.; Geboers, J.; Keyzer, S.; Smet, M.;
Dehaen, W.; Jacobs, P. A.; Sels, B. F. Energy Environ. Sci. 2011, 4,
3601−3610.
(60) Schuurman, Y.; Kuster, B. F. M.; van der Wiele, K.; Marin, G. B.
Appl. Catal., A 1992, 89, 47−68.
(61) Crezee, E.; Hoffer, B. W.; Berger, R. J.; Makkee, M.; Kapteijn,
F.; Moulijn, J. A. Appl. Catal., A 2003, 251, 1−17.
(26) Fitzpatrick, S. W. World Intellectual Property Organization,
Patent No. 96/40609 1996.
(27) Fitzpatrick, S. W.; United States Patent Office, Patent No.
(62) Fabre, L.; Gallezot, P.; Perrard, A. J. Catal. 2002, 208, 247−254.
(63) Singh, U. K.; Vannice, M. A. AIChE J. 1999, 45, 1059−1071.
(64) Singh, U. K.; Vannice, M. A. J. Catal. 2000, 191, 165−180.
(65) Cortright, R. D.; Goddard, S. A.; Rekoske, J. E.; Dumesic, J. A. J.
Catal. 1991, 127, 342−353.
5608105 1997.
(28) Bozell, J. J.; Moens, L.; Elliott, D. C.; Wang, Y.; Neuenscwander,
G. G.; Fitzpatrick, S. W.; Bilski, R. J.; Jarnefeld, J. L. Resour., Conserv.
Recycl. 2000, 28, 227−239.
(29) Girisuta, B.; Janssen, L. P. B. M.; Heeres, H. J. Ind. Eng. Chem.
Res. 2007, 46, 1696−1708.
(30) Mascal, M.; Nikitin, E. B. Green Chem. 2009, 12, 370−373.
(31) Heeres, H.; Handana, R.; Chunai, D.; Rasrendra, C. B.; Girisuta,
B.; Heeres, H. J. Green Chem. 2009, 11, 1247−1255.
(32) Deng, L.; Li, J.; Lai, D. M.; Fu, Y.; Guo, Q. X. Angew. Chem., Int.
Ed. 2009, 48, 6529−6532.
(33) Chia, M.; Dumesic, J. A. Chem. Commun. 2011, 47, 12233−
12235.
(34) Yan, K.; Liao, J. Y.; Wu, X.; Xie, X. M. RSC Adv. 2013, 3, 3853−
3856.
(35) Hengne, A. M.; Rode, C. V. Green Chem. 2012, 14, 1064−1072.
(36) Luo, W. H.; Deka, U.; Beale, A. M.; van Eck, E. R. H.;
Bruijnincx, P. C. A.; Weckhuysen, B. M. J. Catal. 2013, 301, 175−186.
(37) Wu, Z. J.; Ge, S. H.; Ren, C. X.; Zhang, M. H.; Yip, A.; Xu, C.
M. Green Chem. 2012, 14, 3336−3343.
(38) Deng, L.; Zhao, Y.; Li, J. A.; Fu, Y.; Liao, B.; Guo, Q. X.
ChemSusChem 2010, 3, 1172−1175.
́
(66) Boudart, M.; Djega-Mariadassou, G. Kinetics of Heterogeneous
Catalytic Reactions; Princeton University Press: Princeton, NJ, 1984.
(67) Rachmady, W.; Vannice, M. A. J. Catal. 2000, 192, 322−334.
(68) Vaidya, P. D.; Mahajani, V. V. Chem. Eng. Sci. 2005, 60, 1881−
1887.
(69) Tike, M. A.; Mahajani, V. V. Can. J. Chem. Eng. 2006, 84, 452−
458.
(70) Neri, G.; Bonaccorsi, L.; Galvagno, S. Ind. Eng. Chem. Res. 1997,
36, 3554−3562.
(71) Jyrki, K.; Mikkola, J. P.; Sparv, M.; Warna, J.; Karhu, H.; Salmi,
T. Chem. Eng. J 2008, 139, 69−77.
(72) Horiuti, I.; Polanyi, M. Trans. Faraday Soc. 1934, 30, 1164−
1172.
(73) Sinha, N. K.; Neurock, M. J. Catal. 2012, 295, 31−44.
(74) Taylor, H. S.; Close, H. W. J. Phys. Chem. 1924, 29, 1085−1098.
(75) Fishbein, W. N.; Bessman, S. P. J. Biol. Chem. 1966, 241, 4842−
4847.
(39) Luque, R.; Clark, J. H. Catal. Commun. 2010, 11, 928−931.
(40) Du, X. L.; Liu, Y. M.; Wang, J. Q.; Cao, Y.; Fan, K. N. Chin. J.
Catal. 2013, 34, 993−1001.
(41) Sen, S. M.; Henao, C. A.; Braden, D. J.; Dumesic, J. A.;
Maravelias, C. T. Chem. Eng. Sci. 2012, 67, 57−67.
1181
dx.doi.org/10.1021/cs401177p | ACS Catal. 2014, 4, 1171−1181