Hydrogenolysis of glycerol over carbon-supported Ru and Pt catalysts

Article abstract of DOI:10.1016/j.jcat.2007.05.008

Commercial carbon-supported Ru and Pt catalysts were evaluated in the batchwise hydrogenolysis of glycerol in aqueous solution at 473 K and 40 bar H₂, with and without added base. At neutral pH, Ru was more active than Pt at converting glycerol to glycols. However, Ru favored the production of ethylene glycol over propylene glycol and also catalyzed methane formation. Although less active, Pt catalyzed propylene glycol formation with high selectivity. Addition of base enhanced the reactivity of Pt to a greater extent than Ru, but lactate formation was significant at high pH in the presence of either Pt or Ru. The cleavage of C{single bond}C bonds leading to the formation of ethylene glycol from glycerol is proposed to occur primarily through a metal-catalyzed reaction on Ru, whereas this cleavage is thought to occur through a base-catalyzed reaction in the presence of Pt. An overall reaction network for glycerol hydrogenolysis is presented.

Full text of DOI:10.1016/j.jcat.2007.05.008

Journal of Catalysis 249 (2007) 328–337
www.elsevier.com/locate/jcat
Hydrogenolysis of glycerol over carbon-supported Ru and Pt catalysts
Erin P. Maris, Robert J. Davis ∗
Department of Chemical Engineering, University of Virginia, 102 Engineers Way, Charlottesville, VA 22904-4741, USA
Received 21 March 2007; revised 9 May 2007; accepted 10 May 2007
Available online 19 June 2007
Abstract
Commercial carbon-supported Ru and Pt catalysts were evaluated in the batchwise hydrogenolysis of glycerol in aqueous solution at 473 K and
40 bar H , with and without added base. At neutral pH, Ru was more active than Pt at converting glycerol to glycols. However, Ru favored the
2
production of ethylene glycol over propylene glycol and also catalyzed methane formation. Although less active, Pt catalyzed propylene glycol
formation with high selectivity. Addition of base enhanced the reactivity of Pt to a greater extent than Ru, but lactate formation was significant
at high pH in the presence of either Pt or Ru. The cleavage of C–C bonds leading to the formation of ethylene glycol from glycerol is proposed
to occur primarily through a metal-catalyzed reaction on Ru, whereas this cleavage is thought to occur through a base-catalyzed reaction in the
presence of Pt. An overall reaction network for glycerol hydrogenolysis is presented.
© 2007 Published by Elsevier Inc.
Keywords: Glycerol hydrogenolysis; Ruthenium; Platinum; Retro-aldol; Hydrogenation; Ethylene glycol; Propylene glycol; Lactic acid
1. Introduction
industry; however, the greatest opportunity for the utilization
of lactic acid is the production of polylactic acid (PLA). With
global demands of over 1, 3, and 5 billion pounds for lactic
acid [8], propylene glycol [9], and ethylene glycol [10], respec-
tively, the production of these chemicals from biorenewable
glycerol can be both environmentally and economically attrac-
tive. In fact, the industrial scale utilization of glycerol to pro-
duce these smaller chemicals will occur in the near future. Both
Archer Daniels Midland (ADM) and Cargill have announced
plans to commercialize a glycerol-to-glycols process [11,12].
Previous studies have demonstrated the effectiveness of het-
erogeneous catalysts in the hydrogenolysis of polyols to lower
molecular weight glycols and acids [13–15]. Furthermore, the
addition of base has been reported to enhance the conversion
of polyols [16–18]. For example, Montassier et al. [14] investi-
gated the use of silica-supported Co, Pt, Ru, Rh, and Ir as cat-
alysts for the hydrogenolysis of glycerol. However, due to the
high oxygen functionality and low thermal stability of polyol
feedstocks, reaction typically occurs in the aqueous phase at
moderate temperatures and pressures. Previous research in our
laboratory has shown that silica supports are not suitable for
aqueous-phase processing because the silica is hydrolyzed un-
der reaction conditions leading to coalescence and sintering of
metal particles on the support [19]. In addition, silica com-
It is generally accepted that the primary source of renew-
able organic fuels, chemicals, and materials will be from plant-
derived biomass [1–4]. In 2004, only 3% of all chemicals pro-
duced in the U.S. were derived from biomass [5]. The U.S. De-
partment of Energy has set the goal to increase this percentage
to 10% by the year 2020 and 50% by the year 2050 [5]. Glycerol
has been identified by the DOE as one of the top-12 building
block chemicals that can be derived from sugar and converted
to high-value bio-based chemicals or materials [6]. Addition-
ally, glycerol is the major by-product of biodiesel production
by transesterification of vegetable oil [7]. Therefore, using the
growing supply of glycerol is a logical step in moving toward
a more sustainable economy. One route for the conversion of
glycerol to oxygenated chemicals involves hydrogenolysis to
ethylene glycol, propylene glycol, and lactic acid (Scheme 1).
Ethylene and propylene glycol are industrially important
chemicals used in the production of polymers, resins, functional
fluids (antifreeze, deicing), foods, and cosmetics. Lactic acid
and lactate salts are also important for the food and beverage
*
Corresponding author. Fax: +1 434 982 2658.
E-mail address: rjd4f@virginia.edu (R.J. Davis).
0021-9517/$ – see front matter © 2007 Published by Elsevier Inc.
doi:10.1016/j.jcat.2007.05.008

E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
329
Scheme 1. Hydrogenolysis of glycerol.
Scheme 2. Mechanism suggested by Montassier et al. for cleavage of C–C and C–O bonds during polyol hydrogenolysis [14,15,22].
pletely dissolves in high pH solutions that may be of interest in
the aqueous-phase hydrogenolysis of polyol feedstocks. More
recent studies on the hydrogenolysis of glycerol have focused
on the use of Raney metals [14,20,21] and carbon-supported
catalysts [17,21–27].
Ruthenium is well known to be active in the hydrogenoly-
sis of glycerol [14,17,22,23,25]. Unfortunately, Ru often pro-
motes excessive C–C cleavage resulting in a high selectivity
to hydrocarbons (mainly methane) [14,17,22]. Platinum has
also been recognized as a catalyst for glycerol hydrogenolysis
[14,21,26,27]. Whereas Pt is typically found to be less active
than Ru, previous studies of Pt-based catalysts were only con-
ducted under neutral conditions.
Because both a transition metal catalyst and the solution pH
influence the rate of glycerol hydrogenolysis, it is quite likely
that the reaction network is composed of steps occurring on
the catalyst surface as well as in the solution. Whereas sev-
eral mechanisms have been suggested for the hydrogenolysis
of polyols, a widely accepted one was proposed by Montassier
et al. [14,15,22], shown in Scheme 2. In this mechanism, the
cleavage of C–C bonds is proposed to occur through a base-
catalyzed retro-aldol reaction, whereas C–O cleavage occurs
through a base-catalyzed dehydration reaction. The initial de-
hydrogenation step is thought to occur on the transition metal
catalyst.
work aims to compare the differences in activity and selectiv-
ity of Ru/C and Pt/C catalysts during glycerol hydrogenolysis,
with the intent of elucidating a relationship between metal type
and the mechanism of hydrogenolysis. The importance of solu-
tion pH will also be investigated through the addition of basic
promoters such as NaOH and CaO.
2. Experimental
Activated carbon-supported Ru and Pt catalysts were ob-
tained from Acros Organics and Sigma-Aldrich, respectively.
The Ru/C (5 wt% Ru, 50% w/w water) catalyst was dried in
air at 493 K overnight before use; Pt/C (3 wt% Pt) was used
as received. The metal dispersion of the Ru/C and Pt/C cata-
lysts was determined by H2 adsorption using a Micromeritics
ASAP 2020 chemisorption analyzer at 303 K in the pressure
range of 10–450 Torr. Before chemisorption, the catalysts were
heated from ambient conditions to 523 K at 4 K min−1 in flow-
ing H2 (GT&S 99.999%). The catalysts were reduced in H2 at
523 K for 90 min followed by evacuation at 523 K for 4 h. Af-
ter cooling to 308 K, the catalysts were evacuated again for 2 h
followed by analysis at 308 K. Surface Ru and Pt were evalu-
ated by the total amount of H2 adsorbed at 303 K extrapolated
to zero pressure, assuming a stoichiometry (H/Msurf, M = Ru,
Pt) equal to unity.
Whereas the mechanism proposed by Montassier et al. pro-
vides a good foundation to begin a study on glycerol hy-
drogenolysis, the role of the transition metal component relative
to the solution phase still needs to be elucidated. Therefore, this
Glycerol hydrogenolysis reactions were conducted semi-
batchwise in a 300 mL stainless steel reactor (Parr Instruments)
equipped with an electronic temperature controller, a mechani-
cal stirrer, a catalyst addition device, and a dip tube for periodic

330
E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
sampling of the liquid phase. In a typical reaction, glycerol
(Acros Organics, 99%) was diluted with distilled, deionized wa-
ter to form a 1 wt% solution. One hundred fifty milliliters of this
solution was loaded into the reactor along with the appropriate
amount of NaOH (Mallinckrodt) or CaO (Fisher) when nec-
essary. The appropriate amount of Ru/C and Pt/C was loaded
into the catalyst addition device to maintain a substrate/surface
metal (S/Msurf) ratio of approximately 700. The reactor was
sealed and flushed with flowing N2 (GT&S, 99.998%) at 1 bar
for 10 min to remove air in the headspace. To remove the N2,
the reactor was subsequently flushed with flowing H2 (GT&S,
99.995%) at 2 bar for 20 min. The reactor was then pressur-
ized to 5 bar with H2 and heated under moderate agitation
(100 rpm) to a final reaction temperature of 473 K. Once this
temperature was reached, an initial liquid sample was removed
to mark the start of the reaction. Catalyst was subsequently in-
troduced into the reaction medium through the catalyst addition
device, the pressure was increased to 40 bar with H2, and the
rate of agitation was increased to 475 rpm. The reaction was al-
lowed to proceed under these conditions for 5 h whereas liquid
samples were periodically removed. The liquid samples were
allowed to cool to room temperature before the pH was mea-
sured. The reactor was back filled with H2 after each sample to
maintain constant pressure. Liquid samples were analyzed by
high performance liquid chromatography (HPLC) using a Ther-
moSeparations Products (TSP) AS1000 autosampler equipped
with a TSP P2000 pump, an Aminex HPX-87H (Bio Rad) col-
umn, a Waters R-401 refractive index detector, and Millenium
data acquisition software. The HPLC column was maintained
at 333 K, with a mobile phase of 5 mM H2SO4 flowing at
0.7 mL min−1. The major liquid-phase products observed were
ethylene glycol, propylene glycol, lactic acid, and formic acid;
other by-products detected in trace amounts by HPLC include
glyceraldehyde, methanol, and ethanol.
At the conclusion of the 5 h reaction, the reactor was allowed
to cool to room temperature. Gas samples in the headspace were
removed by gas-tight syringe and analyzed for CO2 and CH4
by gas chromatography. The HP 5890 GC was equipped with
an Alltech CTR1 column and TCD detector.
To aid our understanding of the mechanism of glycerol hy-
drogenolysis, the batchwise reactions of lactic acid and propy-
lene glycol with dihydrogen over the commercial Ru/C (Acros,
5 wt%) and Pt/C (Aldrich, 3 wt%) catalysts were also per-
formed. The hydrogenation of a 10 wt% aqueous solution
of lactic acid (ICN Biomedicals Inc., 85–90%) to propylene
glycol was conducted at 473 K and 40 bar H2 using a sub-
strate:catalyst ratio of 345:1 (mol lactic acid:mol total metal).
The hydrogenolysis of a 1 wt% aqueous solution of propylene
glycol was conducted at 473 K and 40 bar H2 using a sub-
strate:catalyst ratio of 700:1 (mol propylene glycol:mol surface
metal). Products were analyzed by HPLC as described above.
Table 1
Results from H chemisorption
2
a
Catalyst
Metal loading
(wt%)
Metal dispersion
Metal particle
size (nm)
b
(H/M
)
surf
c
Ru/C
Pt/C
a
5.0
0.43
0.43
2.3
2.3
d
3.0
Determined by H chemisorption.
Estimated as the inverse of metal dispersion.
Determined by elemental analysis performed by Galbraith Laboratories
2
b
c
(Knoxville, TN).
d
Provided by manufacturer.
catalysts have a similar metal dispersion (43%) and presumably
a similar metal particle size (2.3 nm).
Fig. 1 illustrates the effects of transition metal and base ad-
dition on the glycerol hydrogenolysis reaction. Under neutral
conditions, Ru/C was more active than Pt/C. The addition of
base (both NaOH and CaO) enhanced the rate of reaction over
both catalysts. These results are consistent with the previous
studies mentioned earlier.
Table 2 summarizes the reaction studies of Ru/C and Pt/C
catalysts for glycerol hydrogenolysis in the presence and ab-
sence of base. The overall turnover frequency (TOFoverall) is
reported as the rate of glycerol reacted at 20% conversion per
surface metal atom counted by H2 chemisorption. The turnover
frequency for glycol production (TOFglycol) is based on the eth-
ylene glycol and propylene glycol formation rates at 20% con-
version of glycerol, normalized by the number of surface metal
atoms. The carbon balance is reported as the percentage of car-
bon accounted for in the system (both liquid and gas phase) at
the end of 5 h. The amount of missing carbon in the system
can be explained by loss of gas phase products during sam-
pling of the reaction mixture. Because the reactor system was
not equipped to separate and analyze the gas and liquid phase
products simultaneously, a small amount of gas phase products
were inevitably lost during liquid sampling. Thus, larger errors
in the material balance are expected for runs with larger gas
phase product inventories. The product selectivities of the reac-
tions are summarized in Table 3. The selectivities are reported
both at 20% conversion of glycerol and after the reaction was
stopped at 5 h. The selectivity of each product is determined as
a carbon selectivity, where
moles of carbon in specific product
selectivity =
.
moles of carbon in all observed products
As shown by run #1 in Table 2, the presence of catalyst was
required for hydrogenolysis to proceed. Inspection of the prod-
uct distributions reported in Table 3 when Ru/C was used as the
catalyst reveals significant methane formation. Methane forma-
tion could be the result of methanation of the carbon support as
suggested by Rodriguez-Reinoso [28]. However, a blank reac-
tion run in the absence of glycerol (run #2 of Tables 2 and 3)
failed to produce methane, indicating all methane formed dur-
ing the reactions can be attributed to the hydrogenolysis of
glycerol and not methanation of the carbon support.
3. Results and discussion
Runs #3 and 4 of Table 2 revealed that in the absence of base
(pH ∼6), Ru/C was more active than Pt/C for the hydrogenol-
ysis of glycerol. This result is consistent with other studies that
The results of the H2 adsorption studies are reported in Ta-
ble 1. Characterization of the Ru/C and Pt/C revealed that both

E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
331
(a)
(b)
Fig. 1. Conversion of glycerol over (a) Ru/C and (b) Pt/C in the absence and presence of base. Reaction conditions: 1 wt% glycerol, S/M
(when present), 473 K, 40 bar H . Error bars represent 95% confidence limits.
2
= 700, 0.8 M base
surf
Table 2
a
Summary of catalyst activity for glycerol hydrogenolysis
Run #
Catalyst
Base
Base conc.
(M)
Conversion
(%)
Carbon
balance
TOF
(s
TOF
overall
−1 d
)
glycol
)
b
c
−1 e
(s
1
2
3
4
5
6
7
8
9
10
11
12
13
None
Ru/C
None
None
None
None
NaOH
NaOH
NaOH
NaOH
NaOH
CaO
–
–
–
–
0.8
0.01
0.01
0.8
0.8
0.01
0.01
0.8
0.8
<0.7
0
40
13
4
48
25
100
92
50
40
85
100
100
100
84
97
97
93
95
85
83
87
93
75
80
–
–
–
–
f
Ru/C
Pt/C
None
Ru/C
Pt/C
Ru/C
Pt/C
Ru/C
Pt/C
0.02 0.003
0.006 0.002
–
0.04 0.01
0.01 0.006
0.2 0.03
0.3 0.05
0.07 0.03
0.04 0.04
0.1 0.03
0.02 0.003
0.004 0.0006
–
0.01 0.002
0.005 0.001
0.07 0.003
0.09 0.01
0.02 0.01
0.02 0.01
0.04 0.006
g
g
CaO
CaO
CaO
Ru/C
Pt/C
h
h
0.4 0.07
0.2 0.03
a
Reaction conditions: 1 wt% glycerol solution, S/M
Conversion determined after 5 h of reaction.
Percentage of carbon accounted for after 5 h of reaction.
Determined at 20% conversion of glycerol. Error represents 95% confidence limits.
Calculated using the sum of ethylene and propylene glycol formation determined at 20% conversion of glycerol. Error represents 95% confidence limits.
Reaction run without glycerol to check for possible methanation of carbon support.
Determined after 5 h of reaction as glycerol conversion never reaches 20%.
= 700, 5 wt% Ru/C or 3 wt% Pt/C, T = 473 K, P = 40 bar.
H
2
surf
b
c
d
e
f
g
h
Determined at first time sample of reaction (30% conversion).
compare the activity of Ru- and Pt-based catalysts for the hy-
drogenolysis of glycerol [14,21,26,27]. Although no lactate was
observed in the absence of base, regardless of metal, close ex-
amination of the product selectivities given in Table 3 shows
an interesting dependence on the nature of the metal regard-
ing glycol formation. At 20% conversion of glycerol and in the
absence of base, Ru/C promoted 68% selectivity to ethylene
glycol with only 32% selectivity to propylene glycol. Even after
5 h of reaction, the selectivity to ethylene glycol (47%) was still
greater than that to propylene glycol or methane (26% each).
Platinum, on the other hand, favored formation of propylene
glycol (79% selectivity) over ethylene glycol (17% selectivity),
even after 5 h of reaction. These results suggest that a favorable
C–C cleavage pathway exists on Ru that is not available on Pt.
The proposed mechanism of Montassier et al. shown in
Scheme 2 suggests that the first step in the reaction sequence
is the dehydrogenation of glycerol over the transition metal
catalyst to form a glyceraldehyde intermediate. Although C–C

332
E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
Table 3
a
Summary of product selectivities during glycerol hydrogenolysis
b
Run #
Catalyst
Base
Base conc.
(M)
Conversion
(%)
Carbon selectivity
c
d
d
EG
PG
LA
CH
CO
4
2
e
2
3
Ru/C
None
None
–
–
0
20
40
–
13
–
0
0
0
0
0
–
0
–
0
–
0.26
–
0
–
0
–
0.31
–
0
–
0.01
–
0
–
0.06
0
–
0.01
–
0
–
0
–
Ru/C
0.68
0.47
–
0.17
–
0.32
0.26
–
0.79
–
4
5
Pt/C
None
NaOH
NaOH
NaOH
NaOH
NaOH
CaO
–
None
Ru/C
Pt/C
0.8
0.01
0.01
0.8
4
0
0
Trace
0.65
0.21
0.43
0.39
0.47
0.34
0.62
0.48
0.54
0.32
0.37
0.19
0.54
0.48
0.58
0.58
6
20
48
20
25
20
0.21
0.19
0.04
0.05
0.12
0.01
0.02
0.02
0.14
0.16
0.08
0.09
0.09
0.11
0.05
0.04
0.14
0.27
0.51
0.55
0.37
0.19
0.30
0.46
0.32
0.46
0.55
0.71
0.35
0.36
0.36
0.36
Trace
7
–
0
–
0.01
–
0
–
0
–
8
Ru/C
Pt/C
f
g
100
9
0.8
0.01
0.01
0.8
20
92
20
50
20
40
20
85
10
11
12
13
Ru/C
Pt/C
CaO
–
0
–
0.06
–
0.007
Ru/C
Pt/C
CaO
–
0
–
0
h
CaO
0.8
30
100
0
a
Reaction conditions: 1 wt% glycerol solution, substrate/surface metal = 700, 5 wt% Ru/C or 3 wt% Pt/C, T = 473 K, P = 40 bar. EG: ethylene glycol;
H
2
PG: propylene glycol; LA: lactic acid (in the form of lactate).
b
Run numbers are the same as those listed in Table 2.
c
Carbon selectivities were determined both at 20% conversion of glycerol and at the final conversion attained after 5 h of reaction.
The gas phase was only sampled after the completion of the reaction. Therefore gas phase compositions at 20% conversion of glycerol are not available.
Reaction run in the absence of glycerol to investigate possible methanation of carbon support.
Balance in the carbon selectivity is to formic acid.
In the form of carbonate.
Selectivities determined at first timed-sample during reaction.
d
e
f
g
h
cleavage is speculated to occur through a base-catalyzed retro-
over a factor of 50 in the presence of base. Evidently the ob-
served activity on Ru and Pt depends strongly on the solution
pH. A study performed by Ukisu and Miyadera explored the
enhancement in activity of 2-propanol dehydrogenation over
Ru/C and Pt/C in the presence of basic compounds such as
NaOH and Ca(OH)2 [30]. They found that the addition of
NaOH had relatively little effect on the rate of dehydrogenation
over Ru/C resulting in an enhancement factor of only 1.06. In-
terestingly, the dehydrogenation activity of Pt/C was enhanced
by a factor of 18.1 in the presence of NaOH. Thus, they sug-
gested the rate-determining step for alcohol dehydrogenation
is different over Ru than over Pt, i.e., the dissociation of the
methine C–H bond may be limiting over Pt, whereas forma-
tion of molecular hydrogen may be limiting over Ru. Ukisu and
Miyadera concluded that the addition of alkaline compounds
assisted in the dissociation of the methine hydrogen on Pt,
thereby increasing the rate of dehydrogenation. These findings
are consistent with our results showing that the addition of base
has a greater effect on the rate of glycerol hydrogenolysis on
Pt/C than on Ru/C. Ukisu and Miyadera also explored the pro-
motional effect of Ca(OH)2. However, due to the limited solu-
bility of Ca(OH)2 in 2-propanol, an enhancement factor of only
1.47 for dehydrogenation was obtained over Pt/C. As CaO read-
ily forms Ca(OH)2 in an aqueous environment and a pH of ∼12
aldol reaction, whereas C–O cleavage is thought to proceed
through a base-catalyzed dehydration, the authors suggest that
both reactions occur in the adsorbed state [14,22]. The resulting
intermediates are then re-hydrogenated on the metal catalyst to
form ethylene glycol, propylene glycol, and methanol, with the
latter being easily converted to methane [22]. Wang et al. [29]
supported this mechanism through studying the hydrogenolysis
of 1,3-diol model compounds.
Our results seem to be in agreement with the idea that a
metal catalyst is necessary for the initial dehydrogenation of
glycerol to glyceraldehyde, because essentially no reaction was
observed in the absence of catalyst (runs #1 and 5 of Table 2).
However, the mechanism described in Scheme 2 suggests that
addition of base should increase the conversion of glycerol to
both ethylene glycol and propylene glycol because the retro-
aldol and dehydration reactions are both thought to be catalyzed
by adsorbed hydroxyls. As reported in Table 2 and illustrated in
Fig. 1, the addition of 0.8 M base to maintain a constant pH
(pHNaOH ∼14; pHCaO ∼12) increased the overall rate of glyc-
erol hydrogenolysis over both catalysts. Comparison of runs #3,
8, and 12 in Table 2 show that the addition of base increased the
TOFoverall on Ru/C by an order of magnitude. Runs #4, 9, and
13, however, show that the TOFoverall on Pt/C is enhanced by

E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
333
The product selectivities reported in Table 3 reveal that the
large increase in overall rate of hydrogenolysis in the presence
of base can be attributed to the formation of lactate. Although
this observation is important, it will be addressed later in the
discussion. To better understand the routes to C–C and C–O
cleavage over Ru and Pt, the product selectivities for ethylene
and propylene glycol as reported in Table 3 are discussed next.
Inspection of runs #3, 8, and 12 in Table 3 shows that the
addition of base to a reaction over Ru/C switched the glycol
selectivity to favor propylene glycol over ethylene glycol. The
actual rate of ethylene glycol formation did not change signif-
icantly with the addition of base. Similar comparisons of runs
#4, 9, and 13, however, show that over Pt/C, the addition of
base enhanced the initial rate of ethylene glycol formation by
an order of magnitude or more. In accordance with Scheme 2,
Montassier et al. [14] proposed that C–C cleavage of glycerol
to ethylene glycol occurred by a retro-aldol step in the adsorbed
state catalyzed by surface hydroxyl groups. This speculation
implies that addition of base should increase the rate of eth-
ylene glycol formation on both catalysts compared with that
without base. Instead, we observed that on Ru/C, the rate of eth-
ylene glycol formation changed by less than a factor of 2 with
the addition of base, but on Pt/C, the change was more than
an order of magnitude. Therefore, we propose that over Ru, the
C–C cleavage involved in the conversion of glycerol to ethylene
glycol is primarily metal-catalyzed instead of base-catalyzed.
Studies on the ability of supported Ru and Pt to catalyze the
hydrogenolysis of ethane to methane indicate Ru is more ef-
fective for the cleavage of C–C hydrocarbon bonds than Pt.
Under similar reaction conditions, a Pt/SiO2 catalyst was found
to be approximately three orders of magnitude less active than
Ru/SiO2 for ethane hydrogenolysis [34,35]. Because Pt is less
effective for the cleavage of C–C bonds, the yield and selec-
tivity to ethylene glycol are low in neutral solution. However,
increasing the pH increased the rate of ethylene glycol forma-
tion over Pt/C, indicating a base-catalyzed retro-aldol reaction
may be an important route to form ethylene glycol from glyc-
erol over Pt. Additional experiments in our laboratory have
confirmed that at 473 K and 40 bar H2, the hydrogenolysis of
a 1 wt% propylene glycol solution yielded no ethylene glycol
over Ru/C or Pt/C in the presence or absence of base. There-
fore, the formation of ethylene glycol does not result from the
subsequent hydrogenolysis of propylene glycol.
Fig. 2. Reaction profile for the hydrogenolysis of glycerol on Ru/C in the pres-
ence of 0.8 M NaOH. Reaction conditions: 1 wt% glycerol, S/M
= 700,
surf
473 K, 40 bar H . (ꢀ) Glycerol; (B) formic acid; (!) lactic acid; (P) propy-
2
lene glycol; (1) ethylene glycol; (+) methanol; (×) carbon dioxide trapped as
carbonate.
is maintained during reaction, our system does not appear to be
limited by solubility to the same extent as the 2-propanol sys-
tem. In fact, Cameron and Patten have shown that the solubility
of Ca(OH)2 is actually greater in aqueous solutions of glycerol
than in water itself [31]. Therefore, it appears that Pt/C is actu-
ally more active than Ru/C for the hydrogenolysis of glycerol
in the presence of base, and this increased activity may be at-
tributed to the base promoting the initial dehydrogenation of
glycerol to glyceraldehyde.
Inspection of Table 3 indicates that the activity of Pt/C was
enhanced to a greater extent by the presence of CaO than
NaOH. Because a constant pH was maintained in the runs
performed with 0.8 M base, this behavior is not attributed to
buffering of the solution by CaO. Instead, previous authors
have proposed that the cation may have a role in the complexa-
tion of aldehydes leading to saccharinic acid formation [32,33],
however, our work provides no evidence to further support or
dispute these claims.
It is also interesting to note that, in the presence of 0.8 M
NaOH, Ru/C promoted significant formation of formic acid
(run #8 in Table 3), which did not appear under other reaction
conditions. Shown in Fig. 2 is the reaction profile for glycerol
hydrogenolysis on Ru/C in the presence of 0.8 M NaOH. It
appears that as the conversion of glycerol neared completion,
degradation of propylene glycol and ethylene glycol began.
The degradation product was primarily formic acid, with small
amounts of methanol and CO2 produced as well. In the basic
environment, CO2 was trapped in the form of carbonate. These
findings are consistent with those of Montassier et al. [14] who
reported that degradation of glycols commenced on complete
conversion of glycerol.
Addition of base to the glycerol hydrogenolysis reaction in-
creased the formation rate of propylene glycol over both Ru/C
and Pt/C. Comparison of the selectivities reported at 20% con-
version of glycerol with the TOFglycol’s reported in Table 2
shows that the rate of propylene glycol formation was en-
hanced by a factor of 5–10 over Ru/C and 27–67 over Pt/C
in the presence of 0.8 M base. There are two possible routes
of C–O cleavage leading to the formation of propylene glycol
from glycerol: (1) metal-catalyzed and (2) base-catalyzed. The
metal-catalyzed reaction is analogous to the route proposed by
Suppes et al. shown in Scheme 3 [21,36]. In this path, the glyc-
erol is directly dehydrated over the metal catalyst to form an
acetol intermediate, which can then be hydrogenated over the
metal to form propylene glycol. However, in the work of Sup-

334
E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
Scheme 3. Direct dehydration of glycerol.
Scheme 4. Alkali induced formation of lactic acid from glyceraldehyde [32,37,38].
can be attributed mainly to the formation of lactate. Moreover,
it is worth noting that lactate was only detected in the presence
of base. In runs #6, 7, 10, and 11, only 0.01 M base was used
so that the transient behavior of the product yields and solu-
tion pH could be monitored throughout the reaction. For each
of these runs, the lactate yield remained constant once the pH
dropped to neutral. A representation of this behavior is shown
in Fig. 3 for the case of Pt/C in the presence of 0.01 M CaO.
The formation of lactic acid is known to occur in solution by
the base-catalyzed isomerization of glyceraldehyde [32,37,38].
A schematic of this process is shown in Scheme 4. Therefore,
if the metal catalyst were necessary for the initial dehydrogena-
tion of glycerol to glyceraldehyde, as indicated by Scheme 2
and runs #1 and 5 of Table 2, then the conversion of glyceralde-
hyde to lactic acid (in the form of lactate) may be catalyzed
by hydroxyl groups. With the intent of better understanding
the role of lactate formation in the mechanism of glycerol hy-
drogenolysis, we have conducted studies on the hydrogenation
of lactic acid under the standard hydrogenolysis conditions of
473 K and 40 bar H2.
The Miller group reported that lactic acid can be easily hy-
drogenated to propylene glycol over a Ru/C catalyst [39,40].
As shown in Fig. 4a, Ru/C was more active than Pt/C for the
hydrogenation of aqueous lactic acid, which is consistent with
the well-recognized fact that Ru is superior to other Pt metal
catalysts for the hydrogenation of carboxylic acids [41]. How-
ever, the reaction profiles shown in Fig. 3 indicate that lactic
acid is not a reactive intermediate to propylene glycol, but in-
stead remains bound as a lactate salt in the presence of base.
The Miller group has also studied the hydrogenation of potas-
sium and calcium lactate salts, and found that these salts are
unable to undergo direct hydrogenation to propylene glycol un-
til addition of sulfuric acid converted the lactate salt to free acid
[39]. To confirm that the lactate salts formed during glycerol
hydrogenolysis in the presence of NaOH and CaO are not be-
ing further reacted, we have also investigated the hydrogenation
of sodium and calcium lactate salts over Ru/C under typical hy-
drogenolysis conditions of 473 K and 40 bar H2. In agreement
with the work from the Miller group, sodium and calcium lac-
Fig. 3. Product formation during glycerol hydrogenolysis over Pt/C with 0.01 M
CaO. The pH of the reaction mixture was monitored as a function of the reac-
tion time. PG: propylene glycol; LA: lactic acid (in the form of lactate); EG:
ethylene glycol; GA: glyceraldehyde.
pes et al., glycerol hydrogenolysis was carried out using an 80%
glycerol solution. In our studies, a 1% glycerol solution was
used, and therefore the equilibrium for direct dehydration is less
favorable in our system. In addition, the acetol intermediate that
was detected by Suppes et al. was never detected in our studies.
Furthermore, this direct dehydration route does not account for
the enhanced formation of propylene glycol with the addition
of base. Therefore, it is likely in our studies that propylene gly-
col is formed by a mechanism other than metal-catalyzed C–O
cleavage.
The second proposed route accounts for the enhanced rate
of propylene glycol formation at high pH. Although the overall
rate of glycerol hydrogenolysis (TOFoverall) increased with the
addition of base, the rate of glycol formation (TOFglycol) did not
increase as significantly. Inspection of the product selectivities
reported in Table 3 reveals that the increase in the TOFoverall

E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
335
(a)
(b)
Fig. 4. (a) Hydrogenation of lactic acid to propylene glycol over Ru/C and Pt/C. (b) Hydrogenation of calcium and sodium lactate over Ru/C. Reactions were carried
out at 473 K and 40 bar H . Error bars represent 95% confidence limits.
2
tate salts do not readily hydrogenate to propylene glycol over
Ru/C (Fig. 4b). Furthermore, hydrogenation of free lactic acid
to propylene glycol over Ru/C was also inhibited by NaCl, sug-
gesting that merely the presence of the alkali cation is enough
to bind lactic acid into a nonreactive salt.
It is possible that, in the absence of base, free lactic acid
is formed during glycerol hydrogenolysis and is subsequently
hydrogenated to propylene glycol. In the presence of an alkali
metal or alkaline earth cation, however, the lactic acid might
be bound as lactate salt and therefore does not react further
to propylene glycol. To investigate this hypothesis, the hy-
drogenolysis of glycerol was conducted over Ru/C in the pres-
ence of 0.8 M NaCl to determine if the Na+ cation would bind
intermediate lactic acid. The solution pH remained constant
(pH ∼6) throughout the course of the reaction. The resulting
TOFoverall on Ru/C was 0.01 0.008 s−1, whereas the TOFglycol
was 0.008 0.001 s−1 at 20% conversion of glycerol, which
were significantly lower than the rates on Ru/C in neutral so-
lution (run #3, Table 2). No formation of lactate was observed.
As an interesting aside, glycerol hydrogenolysis over Ru/C in
the presence of NaCl resulted in greater selectivity to propylene
glycol than ethylene glycol (69% PG vs 31% EG at 20% con-
version of glycerol). This contrasted what was observed when
glycerol hydrogenolysis was conducted over Ru/C in the ab-
sence of base (run #3 in Table 3). Apparently, the presence of
Cl inhibited the C–C bond cleavage that lead to ethylene gly-
col formation. Whereas the exact Cl content of the fresh Ru/C
was not evaluated, washing Ru/C with distilled deionized water
failed to produce any Cl in the filtrate as measured by a 0.1 M
AgNO3 test. Because Cl is a known catalyst poison for transi-
tion metals, the lower productivity of ethylene glycol in NaCl
solution further supports the idea that ethylene glycol is formed
through a metal-catalyzed C–C cleavage route on Ru. The fact
Fig. 5. Turnover frequency for the production of propylene glycol on Ru/C and
Pt/C at 20% conversion of glycerol under various conditions. Reaction con-
ditions: 1 wt% glycerol, S/M
40 bar H . Error bars represent 95% confidence limits.
2
= 700, 0.8 M base (when present), 473 K,
surf
that lactate was not observed during the hydrogenolysis of glyc-
erol in the presence of NaCl indicates that the presence of base
is required for the formation of lactic acid. Therefore, under
neutral conditions, formation of propylene glycol may be at-
tributed to the hydrogenolysis of glyceraldehyde (as shown in
Scheme 2) and not the hydrogenation of intermediate lactic
acid. Further support for this idea is obtained by comparing
the rates of propylene glycol formation over Ru/C and Pt/C.
As shown in Fig. 5, the turnover frequency for the formation
of propylene glycol at 20% conversion of glycerol is actually
higher on Pt/C than Ru/C in the presence of base. Because Pt/C
is ineffective for the hydrogenation of lactic acid to propylene

336
E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
Scheme 5. Adjusted scheme for glycerol hydrogenolysis on Ru- and Pt-based catalysts. M denotes either Ru or Pt.
glycol, and in the presence of base, lactate salts are nonreactive,
propylene glycol formation likely occurs through C–O cleavage
without passing through a lactate intermediate.
For verification, the hydrogenation of a 4 wt% aqueous solution
of pyruvaldehyde on Ru/C (S/Msurf = 350) under standard hy-
drogenolysis conditions of 473 K and 40 bar H2 yielded propy-
lene glycol. However, a large percentage (∼50%) of the carbon
was unaccounted for during this experiment due to the ther-
mal degradation of pyruvaldehyde under the hydrogenolysis
reaction conditions. Nevertheless, the formation of propylene
glycol is consistent with the mechanism proposed in Scheme 5.
Propylene glycol formation does not appear to occur via hydro-
genation of lactate. The formation of CO2 reported in Table 3
can be attributed to a Cannizzaro reaction of the intermediate
aldehydes, as proposed by Wang et al. [29].
Scheme 5 shows a proposed mechanism for glycerol hy-
drogenolysis that is based loosely on that of Montassier et al.,
but has been adjusted to account for the results of our study. In
this mechanism, the first step is the dehydrogenation of glycerol
to glyceraldehyde on the transition metal catalyst as suggested
by others. Although this step may be enhanced by the presence
of a base, the base promotion is greater over Pt than over Ru.
Formation of ethylene glycol occurs by two routes. Because
Ru is an effective C–C bond cleavage catalyst, glyceraldehyde
can be directly converted and hydrogenated to ethylene gly-
col and methanol on Ru. Because Pt is not effective for the
cleavage of C–C bonds, formation of ethylene glycol likely pro-
ceeds through a base-catalyzed retro-aldol route. The interme-
diate formaldehyde and glycol aldehyde is then hydrogenated
over the metal to form ethylene glycol and methanol. Although
methanol is sometimes detected as a product, it also can be
converted to methane, as proposed by Montassier et al. [22].
The cleavage of C–O bonds has not been altered from the
mechanism in Scheme 2 where glyceraldehyde undergoes base-
catalyzed dehydration. This route has only been adjusted to
account for the formation of lactate in the presence of base ac-
cording to Scheme 4. The base-catalyzed dehydration of glycer-
aldehyde yields pyruvaldehyde, which in the presence of base
forms lactate. The subsequent hydrogenation of the dehydra-
tion intermediates on the metal catalyst gives propylene glycol.
4. Conclusion
In this work, the activity and selectivity of Ru/C and
Pt/C catalysts were compared for the hydrogenolysis of glyc-
erol. The effects of NaOH and CaO addition on the reaction
rates were used to help elucidate metal-catalyzed versus base-
catalyzed routes in the mechanism of glycerol hydrogenolysis.
The presence of both 0.8 M NaOH and CaO enhanced the
rate of glycerol hydrogenolysis over both catalysts, however
the extent of enhancement is greater over Pt/C than Ru/C. This
trend is consistent with the observation by others that the rate
of alcohol dehydrogenation over Pt is enhanced by base to a
greater extent than over Ru. Because the dehydrogenation of
glycerol to glyceraldehyde is proposed to be the first step in the
mechanism of glycerol hydrogenolysis, the activity of Pt/C is

E.P. Maris, R.J. Davis / Journal of Catalysis 249 (2007) 328–337
337
therefore enhanced more than Ru/C with the addition of NaOH
and CaO.
[11] ADM plans polyols unit using renewable feedstock (Archer Daniels Mid-
land Co. to build renewable plant) (Brief Article), Chem. Week 23 (1)
(2005) 167.
In the absence of base, Ru/C was more active than Pt/C for
the hydrogenolysis of glycerol. Under neutral conditions, Ru
favors the formation of ethylene glycol over propylene glycol.
Because the rate of ethylene glycol formation was not enhanced
significantly in the presence of base, C–C cleavage is thought
to occur over Ru primarily via a metal-catalyzed reaction. The
addition of Cl− to the system resulted in decreased selectivity
to ethylene glycol, further supporting the idea that ethylene gly-
col is formed via a metal-catalyzed route. Because Pt is a less
effective C–C cleavage catalyst, ethylene glycol is proposed to
form through base-catalyzed retro-aldol reaction.
[12] E. Chynoweth, ICIS Chem. Business (Weekly) 39 (1) (2006) 1.
[13] I.T. Clark, Ind. Eng. Chem. 50 (1958) 1125.
[14] C. Montassier, D. Giraud, J. Barbier, Polyol Conversion by Liquid Phase
Heterogeneous Catalysis over Metals, Elsevier Science Publishers, Ams-
terdam, 1988, p. 165.
[15] D.K. Sohounloue, C. Montassier, J. Barbier, React. Kinet. Catal. Lett. 22
(1983) 391.
[16] A.K. Sirkar, US Patent 4 338 472 (1982).
[17] B. Casale, A.M. Gomez, US Patent 5 276 181 (1994).
[18] L. Schuster, M. Eggersdorfer, US Patent 5 616 817 (1997).
[19] E.P. Maris, W.C. Ketchie, V. Oleshko, R.J. Davis, J. Phys. Chem. B 110
(2006) 7869.
[20] A. Perosa, P. Tundo, Ind. Eng. Chem. Res. 44 (2005) 8535.
[21] M.A. Dasari, P.-P. Kiatsimkul, W.R. Sutterlin, G.J. Suppes, Appl. Catal.
A 281 (2005) 225.
Acknowledgments
[22] C. Montassier, J.C. Menezo, L.C. Hoang, C. Renaud, J. Barbier, J. Mol.
Catal. 70 (1991) 99.
[23] D.G. Lahr, B.H. Shanks, Ind. Eng. Chem. Res. 42 (2003) 5467.
[24] J. Chaminand, L. Djakovitch, P. Gallezot, P. Marion, C. Pinel, C. Rosier,
Green Chem. 6 (2004) 359.
[25] D.G. Lahr, B.H. Shanks, J. Catal. 232 (2005) 386.
[26] Y. Kusunoki, T. Miyazawa, K. Kunimori, K. Tomishige, Catal. Commun. 6
(2005) 645.
[27] T. Miyazawa, Y. Kusunoki, K. Kunimori, K. Tomishige, J. Catal. 240
(2006) 213.
[28] F. Rodriguez-Reinoso, Carbon 36 (1998) 159.
[29] K.Y. Wang, M.C. Hawley, T.D. Furney, Ind. Eng. Chem. Res. 34 (1995)
3766.
[30] Y. Ukisu, T. Miyadera, React. Kinet. Catal. Lett. 81 (2004) 305.
[31] F.K. Cameron, H.E. Patten, J. Phys. Chem. 15 (1911) 67.
[32] J.C. Sowden, E.K. Pohlen, J. Am. Chem. Soc. 80 (1958) 242.
[33] J. Kenner, G.N. Richards, J. Chem. Soc. (1957) 3019.
[34] R.D. Cortright, R.M. Watwe, J.A. Dumesic, J. Mol. Catal. A Chem. 163
(2000) 91.
[35] Y. Doi, H. Miyake, K. Soga, J. Mol. Catal. 48 (1988) 123.
[36] G.J. Suppes, W.R. Sutterlin, M.A. Dasari, WO 2005095536 (2005).
[37] P.A. Shaffer, T.E. Friedemann, J. Biol. Chem. 86 (1930) 345.
[38] G. Machell, G.N. Richards, J. Chem. Soc. (1960) 1924.
[39] Z. Zhang, J.E. Jackson, D.J. Miller, Appl. Catal. A 219 (2001) 89.
[40] Z.G. Zhang, J.E. Jackson, D.J. Miller, Ind. Eng. Chem. Res. 41 (2002)
691.
This work was funded by the National Science Foundation
(grants CTS-0313484 and CTS-0624608) with partial support
by the U.S. Department of Energy, Office of Basic Energy Sci-
ences (grant DE-FG02-95ER14549).
References
[1] L.R. Lynd, C.E. Wyman, T.U. Gerngross, Biotechnol. Progr. 15 (1999)
777.
[2] B.E. Dale, J. Chem. Technol. Biotechnol. 78 (2003) 1093.
[3] B. Kamm, M. Kamm, Appl. Microbiol. Biotechnol. 64 (2004) 137.
[4] A.J. Ragauskas, C.K. Williams, B.H. Davison, G. Britovsek, J. Cairney,
C.A. Eckert, W.J. Frederick, J.P. Hallett, D.J. Leak, C.L. Liotta, J.R. Mie-
lenz, R. Murphy, R. Templer, T. Tschaplinski, Science 311 (2006) 484.
[5] D.J. Miller, NSF Workshop Report on Catalysis for Biorenewables Con-
version, http://www.nsf.gov/eng/cbet/workshops/catalysis_for_biofuels_
workshop_2004.pdf, 2004.
[6] T. Werpy, G. Petersen, Top Value Added Chemicals from Biomass, vol. 1,
Results of Screening for Potential Candidates from Sugars and Synthesis
Gas, US DOE Report, 2004.
[7] C. Chiu, Ind. Eng. Chem. Res. 45 (2006) 791.
[8] M. Altaf, M. Venkateshwar, M. Srijana, G. Reddy, J. Appl. Microbiol.
(2007), early online article.
[9] P.H. Sim, Chem. Week 24 (1) (2005) 167.
[10] M. Kirschner, Chem. Market Reporter 35 (1) (2004) 266.
[41] R.L. Augustine, Heterogeneous Catalysis for the Synthetic Chemist, Mar-
cel Dekker, New York, 1996.