Transformations of polyols to organic acids and hydrogen in aqueous alkaline media

Article abstract of DOI:10.1039/c4cy00249k

In this paper we show that carbon nanofiber supported copper and nickel nanoparticles can selectively transform ethylene glycol and glycerol into value added oxygenates (organic acids) under anaerobic aqueous conditions. During aqueous phase oxidation Cu based catalysts showed a nearly quantitative yield (96% selectivity at 82% conversion) of glycolic acid from ethylene glycol. The reaction was carried out under alkaline conditions at relatively mild temperatures (150-180 °C) and produced H₂ as co-product. The high selectivity towards glycolic acid was independent of the temperature. For glycerol oxidation a high selectivity (67% at full conversion) towards lactic acid was observed using Cu with competitive formation of glyceric acid, 1,2-propanediol tartronic acid and formation of H₂ as co-product. The activity of Ni was comparable to that of Cu but it was less selective for the formation of desired oxygenates, glycolic acid (31%) and lactic acid (24%), due to the formation of formic acid.

Full text of DOI:10.1039/c4cy00249k

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Transformations of polyols to organic acids and
hydrogen in aqueous alkaline media
Cite this: Catal. Sci. Technol., 2014,
4, 2353
a
a
a
ab
T. van Haasterecht, T. W. van Deelen, K. P. de Jong and J. H. Bitter*
In this paper we show that carbon nanofiber supported copper and nickel nanoparticles can selectively
transform ethylene glycol and glycerol into value added oxygenates (organic acids) under anaerobic
aqueous conditions. During aqueous phase oxidation Cu based catalysts showed a nearly quantitative
yield (96% selectivity at 82% conversion) of glycolic acid from ethylene glycol. The reaction was carried
out under alkaline conditions at relatively mild temperatures (150–180 °C) and produced H
2
as
Received 25th February 2014,
Accepted 22nd May 2014
co-product. The high selectivity towards glycolic acid was independent of the temperature. For glycerol
oxidation a high selectivity (67% at full conversion) towards lactic acid was observed using Cu with
2
competitive formation of glyceric acid, 1,2-propanediol tartronic acid and formation of H as co-product.
DOI: 10.1039/c4cy00249k
The activity of Ni was comparable to that of Cu but it was less selective for the formation of desired
oxygenates, glycolic acid (31%) and lactic acid (24%), due to the formation of formic acid.
www.rsc.org/catalysis
catalysts have also been explored. The liquid phase catalytic
oxidation of glycerol with heterogeneous Au and Pd catalysts
Introduction
Plant derived biomass can serve as a primary source for the
synthesis of renewable chemicals and fuels. Cellulose is of
special interest in this regard as it is one of the major compo-
nents in biomass and does not compete directly with food
production. It has been shown that cellulose can be directly
converted into smaller polyols like glycerol (1) and ethylene
yields dihydroxy acetone (6) plus a variety of organic acids,
14,15
most notably glyceric acid (4).
acid (5) have also been reported with an Au–Pt/TiO
High yields (85%) of lactic
16
2
catalyst.
A chemical route is also known for the selective production
of glycolic acid (3) by catalytic oxidation of ethylene glycol
1
7
using Cu or precious metal (Au Pd, and Au–Pd bimetallic)
1
–3
18
13,19
glycol (2) which can serve as renewable building blocks for
the fine-chemical industry. In addition glycerol is widely
available as a by-product from biodiesel production (see
Scheme 1). Known strategies for the valorisation of these
polyols are outlined in Scheme 1 and include oxidation,
catalysts and with Au and Pd supported on carbon.
These
processes are conducted at low temperatures (40–90 °C) using
molecular oxygen and typically require alkaline conditions to
in order to achieve appreciable activity.
Next to oxidation the hydrogenolysis of polyols is another
route for the production of value added chemicals (Scheme 3).
This process requires the addition of hydrogen at high pres-
sures (20–50 bar) at 180–220 °C. For instance the production
of 1,2-propanediol (7) from glycerol has been investigated
4
,5
hydrogenolysis of a C–O bond
reforming.
and aqueous phase
6
The aerobic oxidation (Scheme 2) of small polyols can
be accomplished via chemical and biochemical routes.
For instance, glycolic acid (2-hydroxyethanoic acid) (3) can
be produced by aerobic fermentation of ethylene glycol with
3
,4,20–24
extensively with noble and base metal catalyst.
How-
ever d'Hondt et al. were, to the best or our knowledge, the
first to report that hydrogenolysis also proceeds without
7
,8
high selectivity. Value added glycerol derivatives like glyceric
9
10,11
acid (2,3-dihydroxypropanoic acid) (4) and lactic acid (5)
external H addition in inert atmosphere. In that case aque-
2
1
2
6
can be produced via a variety of microbial processes. These
ous phase reforming of a part of the glycerol feed to H
2
yields the required hydrogen in situ, which has since
then been corroborated by others.
has been studied under acidic and alkaline conditions
and, besides the type of catalysts, also pH and gas phase
composition were shown to affect selectivity.
2
and
2
5
chemicals are valuable products for the pharmaceutical,
CO
1
3
26–28
cosmetic and polymer industry.
Glycerol hydrogenolyis
2
9
21
Since these biochemical processes suffer from low space-
time-yields alternative chemical processes using heterogeneous
3
0,31
Higher
a
Utrecht University, Debye Institute for Nano Materials, Department of Inorganic
1
,2-propanediol (7) and ethylene glycol selectivities were
Chemistry and Catalysis, Universiteitsweg 99, Utrecht, Netherlands.
E-mail: harry.bitter@wur.nl
obtained under hydrogen atmosphere while more gaseous
hydrocarbons were formed under inert atmosphere.
b
Wageningen University, Netherlands
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2353

View Article Online
Paper
Catalysis Science & Technology
Scheme 1 Roadmap summarizing different strategies for the valorization of glycerol and ethylene glycol obtained from cellulose or biodiesel
production via aqueous phase processes; (an)aerobic oxidation, hydrogenolysis and aqueous phase reforming.
Scheme 2 Possible products obtained by the catalytic transformation of glycerol and ethylene glycol via chemical or biochemical aerobic
oxidation.
Scheme 3 Possible products obtained via the hydrogenolysis of glycerol.
Interestingly also lactic acid (5) was observed as a by-product in
e.g. when lactic acid is converted further into propionic acid
2
1
the hydrogenolysis of glycerol without external hydrogen
or acrylic acid. In most cases noble metals (Pt and Ru or Ir
2
6
30,31
addition. Several studies showed that the presence of base
significantly enhanced the reaction rate but shifted the selec-
tivity from 1,2-propanediol (7) to lactic acid (5) as major
and Rh
were investigated. Obviously the use of non-noble
3
3
metal catalyst would be advantageous. Recently Roy et al.
showed that copper catalysts are active and selective for the
production of lactic acid (5) from glycerol in alkaline aqueous
solutions and reported high selectivities at relatively low
temperatures (200–240 °C). Unfortunately only bulk copper
catalysts were stable and supported Cu/SiO and CuO/Al O
2 2 3
deactivated due to support degradation in the alkaline
reaction medium.
2
1
product, both with a hydrogen atmosphere or under inert
3
1,32
atmosphere (i.e. no external hydrogen addition).
The synthesis of lactic acid without external oxygen or
hydrogen using metal catalysts is an interesting alternative to
the aerobic chemical routes and to the biochemical route for
application where enantio pure lactic acid (5) is not required,
2354 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
In this paper we investigate both ethylene glycol and
glycerol transformations (Scheme 4) using copper and nickel
nanoparticles supported on carbon nanofibers (CNF) as cata-
lysts. The main advantage of the CNF support is its inertness
Crystallite sizes were estimated from the Scherrer equation
using a shape factor of 0.89.
Hydrogen chemisorption was performed on a Micro-
meretics ASAP 2020C instrument. The catalysts were reduced
3
4–36
−1
under the basic conditions applied.
We show for the
in H (60 ml min ) at 300 °C for 2 h after which the system
2
first time that ethylene glycol can be nearly quantitatively
converted to glycolic acid (3) in alkaline aqueous solutions
under anaerobic conditions at temperature between 150–180 °C
over Cu while over Ni also formic acid (8) was observed as
product. Pathways for the conversion of ethylene glycol and
glycerol over Ni and Cu catalysts will be discussed.
was outgassed (<1 Pa) at that temperature for 2 h. Subse-
quently the system was cooled to 150 °C and outgassed for
another 0.5 h before the measurement was started. The
2
monolayer H uptake was found by extrapolating the linear
part of the isotherm (150–600 mbar) to zero pressure. The
monolayer uptake was calculated assuming a stoichiometry
Ni : H of 1.
Transmission Electron Microscopy (TEM) analysis was
carried out on an FEI Technai20F operated at 200 KeV. The
samples were ground to a fine powder in a mortar and a
small amount of powder was supported onto a thin carbon
film on a copper grid.
Experimental procedures
Catalyst preparation
Carbon nanofibers were prepared by a procedure explained
3
5,37
elsewhere.
impregnation of surface oxidized carbon nanofibers (CNF)
with aqueous solutions of Ni(NO )·6H O (Acros 99%) or
Cu(NO )·3H O (Acros 99.5%) to obtain metal loadings of
0 wt%. The impregnated samples were dried at 120 °C for
6 hours in static air. The dried samples were then reduced
The catalysts were prepared by incipient wetness
Catalyst testing
3
2
3
2
Catalytic performance was evaluated using 10 ml batch auto-
clave reactors. The reactors were loaded in glove box with
reduced catalyst (10–100 mg), base additive (0–0.65 M KOH)
and substrate (0.1–10 wt%), typically a KOH/substrate molar
ratio of 2 and a total liquid volume of 5 ml was used. The
reactors were placed in a preheated oil bath and the reaction
mixture was stirred using a magnetic spinner (1000 rpm). The
reaction was continued at the desired temperature (150–180 °C)
for 0.5–15 h after which the reactors were quenched and
depressurized. The reaction mixture was diluted 10× with
water and adjusted to pH 2–3 by addition of diluted H SO .
1
1
−
1
in a 150 ml min flow of 30% v/v H
2 2
/N at 300 °C during
2
hours. Part of the reduced sample (Cu/CNF, Ni/CNF) was
removed from the reactor in a glove box under nitrogen
atmosphere and stored before further use. Prior to characteri-
zation part of the sample was slowly passivated by diffusion
of air into the reactor through a 26 cm long tube with a
3
mm inner diameter (Cu/CNFp, Ni/CNFp). Part of the passiv-
ated sample was then calcined in static air at 250 °C for
hours to obtain the CuO/CNF and NiO/CNF catalyst.
2
4
2
The composition of the liquid phase was analysed by
HPLC using a refractive index detector and an Animex Column
−
1
(
0.55 ml min of 0.005 M H SO at 60 °C and 35 bar). The
2 4
Catalyst characterization
substrate conversion (Xsubstrate) was calculated as moles of
substrate reacted/moles of substrate fed × 100%. The product
yields (Yproduct) were calculated on a carbon basis as moles of
carbon in the formed product/n times the moles of substrate
fed to the reactor × 100%, with n = 2 for ethylene glycol and
n = 3 for glycerol. The selectivity (Sproduct) was calculated as
X-ray powder diffraction (XRD) was carried out using a
Bruker-AXS D2 Phaser powder X-ray diffractometer with Co
Kα radiation (λ = 1.79026 Å), between 2theta 20–80°. The
experimental data was fitted with a pseudo-Voight function
using a linear combination of the support (CNF) and metal
contributions and a correction for instrumental broadening.
S
product= Yproduct/Xsubstrate × 100%.
Scheme 4 Possible products obtained via hydrothermal anaerobic oxidation of glycerol and ethylene glycol.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2355

View Article Online
Paper
Catalysis Science & Technology
The gas phase composition was investigated by
For the pre-reduced Cu/CNF catalyst different base to
ethylene glycol ratios were explored (Table 1, entries 5–8).
The activity increased with increasing base concentration up
to a ratio of 2 mole KOH per mole ethylene glycol. The maxi-
mum achievable conversion under the applied conditions
was 20%. Glycolic acid and trace amounts of glycolaldehyde
(2-hydroxyacetaldehyde) and oxalic acid (ethanedioic acid) were
the only products observed. The type of base (KOH, NaOH or
preforming the reaction in an Autoclave Engineers 100 ml
batch EZE seal batch reactor. The reactor vessel was filled
with 0.5 g catalyst (Ni/CNF or Cu/CNF) and 50 ml 1 wt%
aqueous solution of ethylene glycol or glycerol. The reactor
−
1
2
was pressurized to 5 bars with N and heated (10 °C min )
to 150 °C and the reaction was continued for 4 hours before
cooling down to room temperature. An increase in pressure
was observed. To analyze the gas phase the reactor was
depressurized via a back pressure regulator (BPR) and the
gas composition was measured using an online GC (dual
channel Interscience CompactGC with Poraplot Q and
Molsieve 5A columns, equipped with TCD detectors using He
2
Ca(OH) also affected the activity (Table 1, entries 7, 9–10). We
found that KOH was comparable to NaOH and superior to
Ca(OH) . This is in agreement with effectiveness of base addi-
2
tives in the non-catalytic high temperature hydrothermal
3
8
process and probably mainly the result of the difference in
pH as a result of the solubility of the base.
and N as carrier gas).
2
Turn-over-frequencies (TOFs) were calculated based on the
moles of glycerol converted per hour and normalized by the
number of metal surface sites of the fresh catalyst as mea-
sured by H chemisorption (Ni) or calculated based on the
2
average crystallites size observed from XRD (Cu).
With Ni/CNF, at a KOH : ethylene glycol ratio of 2, a high
conversion (49%) was realized in 2 hours and besides glycolic
acid also formic acid was observed (entry 11). These results
clearly show that the formation of glycolic acid from ethylene
glycol requires both the metal catalyst and the base at
this temperature. Therefore all further experiments were
performed at a KOH : substrate ratio of 2.
Results and discussion
Effect of base promoters
Reaction profiles: Ni versus Cu
The effect of the concentration and nature of the base on the
conversion of ethylene glycol and on the product formation
was investigated first. Table 1 shows the conversion of ethyl-
ene glycol and the corresponding product yields after 2
hours. Without base addition (entries 1–2), at neutral pH,
almost no ethylene glycol was converted (<1%) over the
Ni/CNF or Cu/CNF catalyst. We also performed blank experi-
ments with base but without metal catalyst (entries 3–4). At a
KOH : ethylene glycol ratio of 2 no activity was observed (con-
version <1%). This is not surprising since the non-catalytic
hydrothermal process (i.e. without metal catalyst) is known to
The activity of pre-reduced Ni/CNF and Cu/CNF was studied
over time in order to gain insight into the evolution of the
reaction products (Fig. 1 and 2). Fig. 1 shows the conversion
of ethylene glycol and product yield between 1 and 15 hours
over (a) Ni/CNF and (b) Cu/CNF. For Ni/CNF we observed the
formation of glycolic acid (as glycolate) and formic acid (as
formate). Glycolic acid is the major product and reached an
optimum yield of 37% after 4 hours and then decreased
again to a final yield of 31% after 15 hours. Glycolic acid is
apparently prone to secondary reactions. The decrease in
glycolic acid yield between 4 and 16 hours was accompanied
by an increase in formic acid yield. The formic acid yield con-
tinued to increase over the whole course of the reaction and
a yield of 24% was reached after 15 hours. This suggests that
glycolic acid is the intermediate product and degrades to
formic acid (Scheme 5). With Cu/CNF (Fig. 1b) glycolic acid
3
8,39
occur only at significantly higher temperatures (e.g. 300 °C).
The CNF support did show some activity (conversion 5%) but
no products were detected in significant quantity. Therefore
we concluded that the observed conversion is due to the for-
mation of undetectable condensation products or the result of
adsorption of reactant on the support.
Table 1 Effect of the substrate base ratio on the product yield and conversion of ethylene glycola
Entry
Catalyst
Base
Base/EG (mol/mol)
Conversion (%)
GOA yield (%C)
1
2
3
4
5
6
7
8
9
Cu/CNF
Ni/CNF
None
None
None
KOH
KOH
KOH
KOH
KOH
KOH
NaOH
Ca(OH)
KOH
0
0
2
2
1.1
1.5
2
3
2
<1
2
<1
5
14
18
20
16
17
8
<1
<1
<1
<1
12
17
20
14
16
7
CNF
Cu/CNF
Cu/CNF
Cu/CNF
Cu/CNF
Cu/CNF
Cu/CNF
Ni/CNF
10
1
2
2
2
1
49
29
a
Reaction conditions: 1 wt% EG, 150 °C, 100 mg pre-reduced catalyst (Cu/CNF or Ni/CNF), reaction time 2 hours; EG = ethylene glycol,
GOA = glycolic acid.
2356 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Fig. 1 Conversion profiles of (●) ethylene glycol and (carbon based)
product yields of (■) glycolic acid and (▲) formic acid. Reaction
conditions: KOH : EG = 2, 1 wt% EG in water, 150 °C, 100 mg catalyst.
Fig. 2 Conversion profiles of glycerol (○) and product yields (carbon
based) of (△) lactic acid, (▲) formic acid, (*) 1,2-propanediol, (□)
glyceric acid (●) ethylene glycol, and (■) glycolic acid. a) Ni/CNF and
b) Cu/CNF. Conditions: KOH : glycerol = 2, 1–15 hours, 1 wt% glycerol,
150 °C, 100 mg catalyst.
was also formed but, unlike with Ni/CNF, did not react
further to formic acid and glycolic acid was quantitatively
obtained (Scheme 5).
The product selectivities and carbon balances after 15 hour
using the Cu and Ni catalyst and ethylene glycol and glycerol
as feedstock are reported in Table 2. With ethylene glycol as
reactant and Cu/CNF as catalyst the selectivity towards
glycolic acid was 98% while with Ni/CNF the selectivity was
only 31%. In addition the carbon balance of the liquid phase
shows a significant carbon deficit with the Ni/CNF catalyst.
C–C cleavage is an important pathway over the nickel cata-
lyst. With the exception of glyceric acid these products have
been reported previously in the transformation of glycerol to
3
0,31
31,32
21
lactic acid using Rh/C,
Ir/C,
Pt and Ru on carbon
3
3
and Cu catalysts. Lactic acid and 1,2PDO yields increased
to 19% for both products in the first 2 hours. At this point
almost complete conversion of glycerol was realized, however
the lactic acid yield increased further to 24% after 15 hours
whereas the 1,2PDO yield decreased to 9% during this
period. The yields of ethylene glycol and glyceric acid also
decreased after full glycerol conversion was realized while
formic acid and glycolic acid yields increased continuously
over the whole course of the reaction.
2
This might be explained by the formation of CO which is
captured by the base as will be discussed below. Table 2 also
lists the initial activity expressed as turn-over-frequency (TOF)
at 20% conversion. These rates are normalized by the number
−
1
−1
of surface Ni and Cu atoms (288 μmol g and 58 μmol g
respectively) estimated from H -chemisorption (Ni) and the
With Cu/CNF complete glycerol conversion was achieved
within 6 hours (Fig. 2b). The major products formed were
lactic acid and glyceric acid and also significant amounts of
1,2PDO and glycolic acid were formed. Also here the forma-
tion of lactic acid increased even after complete glycerol con-
version was realized and a final yield of 52% was obtained.
Glyceric acid and 1,2PDO yield decreased after glycerol was
fully converted.
2
XRD (Cu). The TOFs were comparable for the Ni and Cu
−
1
catalyst, 11 and 14 h respectively.
Fig. 2 shows the conversion profiles of glycerol and
product yields between 1 and 15 hours over Ni/CNF (a) and
Cu/CNF (b). With Ni/CNF complete conversion of glycerol
was obtained after 2 hours and the product spectrum is dom-
3
inated by the C products lactic acid (as lactate) and 1,2-
propanediol (1,2PDO) and glyceric acid (as glycerate). Also
When the product distributions over Cu and Ni are com-
pared (Scheme 6) it appears that 1,2PDO can react to lactic
acid and glyceric acid can degrade to form glycolic acid over
significant quantities of C (glycolic acid and ethylene glycol)
2
1
and C (formic acid) products were formed. This shows that
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2357

View Article Online
Paper
Catalysis Science & Technology
Scheme 5 Product formation in the anaerobic oxidation of ethylene glycol with Cu/CNF and Ni/CNF catalysts in aqueous alkaline media.
Table 2 Product selectivities in the processing of ethylene glycol and glycerol over Ni/CNF and Cu/CNF catalysts
a
Product selectivity (%)
b
X
(%)
TOF
(h
C
bal
(%)
−
1
#
Reactant
Catalyst
)
GOA
FA
LA
GEA
1,2PDO
EG
1
2
3
4
EG
EG
GLY
GLY
Ni/CNF
Cu/CNF
Ni/CNF
Cu/CNF
89
82
100
100
11
14
n.a
30
31
96
5
27
0
15
0
0
0
24
52
0
0
5
0
0
9
0
0
2
0
65
98
59
99
5
33
10
a
b
Activity expressed as turn-over-frequency (TOF) at ± 20% conversion. Percentage of carbon accounted for in the liquid phase after 15 hours
reaction. Reaction conditions: 150 °C, 1 wt% solutions, 15 hours, 100 mg catalyst. EG = ethylene glycol, GLY = glycerol, GOA= glycolic acid,
FA = formic acid, LA = lactic acid, GEA = glyceric acid, EG = ethylene glycol, n.a = not available.
Scheme 6 Product formation in the anaerobic oxidation of glycerol with Cu/CNF and Ni/CNF catalysts in aqueous alkaline media.
both catalysts. Formation of additional C–C cleavage prod-
ucts, formic acid and ethylene glycol, occurs only with the Ni
catalyst. Ethylene glycol might be formed directly from glyc-
erol and formic acid could be formed from ethylene glycol
via glycolic acid. The selectivity after 15 hours toward desir-
able lactic acid and glyceric acid was thus also higher with
the Cu catalyst than with the Ni catalyst, 52% versus 24%
molecular oxygen and should thus be accompanied by hydro-
gen production. Overall the reaction can be written as (1) for
the formation of glycolic acid from ethylene glycol and (2) for
the formation of lactic acid from glycerol. We also observed
the formation of 1,2PDO from glycerol which is accompanied
by consumption of hydrogen (3).
(
Table 2). Again the carbon balance was closed for the Cu
C H O (ethylene glycol) + H O → C H O (glycolic acid) + 2H (1)
2
6
2
2
2
4
3
2
catalyst (99%) while a significant carbon deficit was observed
with Ni (carbon balance 59%), likely due to formation of CO2
C H O (glycerol) → C H O (lactic acid) + H
(2)
(3)
3
8
3
3
6
3
2
and subsequent capture of CO by KOH as (bi)carbonate.
2
C
3
H
8
O
3
(glycerol) + H
2
→ C
3
H
8 2 2
O (12PDO) + H O
Hydrogen formation
During aqueous phase hydrogenolysis of glycerol without
external H in alkaline conditions Roy et al. detected small
2
3
3
Organic acids are the main products from both glycerol and
ethylene glycol under the prevailing conditions. Apparently
the oxidation of these polyols proceeds in the absence of
amounts of hydrocarbons and CO but did not report the
2
3
1
production of H . Auneau et al. on the other hand reported
2
2358 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
the formation of H during glycerol hydrogenolysis both in
not observed in the gas phase under alkaline conditions due
to the formation of carbonates.
2
41
inert and hydrogen containing atmosphere. Since we also
observed a pressure increase in the reactor during the experi-
ments the composition of the gas phase was investigated for
the reactions of ethylene glycol and glycerol with both Cu
Effect of temperature and concentration
and Ni catalyst. H
amounts. Fig. 3 shows the H selectivity expressed as mole of
2
was the only gas formed in significant
The influence of temperature on the activity and selectivity
was investigated for Cu/CNF, the most selective catalyst
towards the desired oxygenated products (i.e. glycolic acid
and lactic acid). Table 3 summarizes the activity and product
distribution as function of temperature. It was already shown
in Fig. 1 that at 150 °C ethylene glycol was converted selec-
2
H produced per mole of converted substrate. For the copper
2
2
catalyst the observed production of 1.9 mole of H per mole
of ethylene glycol is in good agreement with the stoichiometry
of the formation of glycolic acid from ethylene glycol (reac-
−
1
tion 1). For the reactions with glycerol lower H
2
selectivities
tively to glycolic acid at a rate of 14 h (Table 2, entry 2). By
were obtained as compared to starting with ethylene glycol.
This is because the formation of lactic acid only yields one
mole of H and also the production of 1,2PDO from glycerol
2
increasing the temperature to 180 °C (Table 3, entry 1) the
−
1
TOF increased to 39 h while still maintaining high selec-
tivity (97%). Thus the temperature increased the activity
while maintaining the high selectivity.
consumes 1 mole of H per mole converted glycerol. With the
2
nickel catalyst the H selectivity was always higher. This is in
agreement with the observation that significant amounts of
formic acid are formed which results in additional produc-
The effect of the reaction temperature on the reactivity of
glycerol in the range 150–180 °C is also shown in Table 3
2
−
1
(entries 2–4). The glycerol TOF increased from 33 h at
−
1
tion of H
2
.
150 °C to 124 h at 180 °C. With Cu/CNF complete conver-
sion was obtained at 160 °C after 4 hours. From 150–180 °C
the lactic acid selectivity increased steadily from 42% to 64%
while the selectivity to glyceric acid and 1,2PDO decreased.
The increase in lactic acid selectivity can be partially attrib-
uted to the higher conversions at higher temperature and the
conversion of 1,2PDO to lactic acid (Fig. 1). Since at higher
temperature the increase in selectivity towards lactic acid is
larger than the corresponding decrease in 1,2PDO selectivity
it must be concluded that lactic acid formation is also
favoured over the formation glyceric acid. In addition, above
Since only H
2
was detected in the gas phase the observed
carbon deficiency with the Ni/CNF catalyst cannot be attrib-
uted to the formation of hydrocarbons. It is therefore most
likely that with the Ni catalyst also partial reforming to H
and CO occurs, possibly via decarboxylation or decarbonylation
2
2
4
0
2
of the acid or aldehyde intermediates. The formed CO is
1
50 °C the formation of tartronic acid (2-hydroxypropanedioic
acid) becomes significant. This also contributes to the
decrease in glyceric acid selectivity.
Since the use of dilute (1 wt%) feed solution is not
economical (e.g. glycerol from trans-etherification typically
contains only 20 wt% water) we also investigated the use of
higher feedstock concentrations. The effect of the initial
glycol concentration on the reaction rate and lactic acid yield
is shown in Table 4. When the glycerol concentration was
increased from 5 to 10 wt% (entries 1–2) the glycerol TOF at
−
1
2
0% conversion almost doubled from 106 to 203 h . The use
2
Fig. 3 Hydrogen selectivity expressed as mole H per mole converted
of even higher glycerol concentrations would also require
higher base concentrations and therefore risk severe corro-
sion of the reactor. We note that such limitation can easily
substrate for ethylene glycol and glycerol using Cu/CNF and Ni/CNF
catalysts at 150 °C. Reaction conditions: 150 °C, 1 wt% solutions,
4
hours, KOH : substrate ratio 2.
Table 3 Influence of temperature on the glycerol conversion and product selectivities over Cu/CNF
Selectivity (%)
a
−1
Entry
Reactant
T (°C)
X (%)
TOF (h
)
LA
GEA
1,2PDO
TA
GOA
1
2
3
4
EG
180
150
160
180
76
68
100
100
39
33
50
0
42
51
64
0
39
24
7
0
16
9
0
2
11
14
97
1
5
GLY
GLY
GLY
124
2
13
a
Surface normalized reaction rates expressed as turn-over-frequency (TOF) at ±20% conversion. Reaction condition: 1 wt% solutions, 100 mg
Cu/CNF, 4 hours, 120–180 °C. LA = lactic acid, GEA = glyceric acid, 1,2PDO = 1,2-propanediol, TA = tartronic acid and GOA = glycolic acid.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2359

View Article Online
Paper
Catalysis Science & Technology
Table 4 Effect of initial glycerol concentration on conversion and lactic
acid selectivity with Cu/CNF
reference (rxn 3 and 7). The aerobic oxidation of both ethyl-
18
14
ene glycol and glycerol also requires the addition of base
and thus forms the corresponding carboxylic salts and are
therefore more accurately written as reactions 4 and 8.
a
Time
(h)
C
(wt%)
i,GLY
T
(°C)
X
(%)
GLY
TOF
(h
S
LA
(%C)
−
1
Entry
)
The Gibbs free energies are lower for the reaction of both
substrates if KOH addition (i.e. neutralization) is taken into
account. The Gibbs free energy for the formation of glycolic
acid from ethylene glycol (160 °C) has a small positive value
1
2
3
2
2
4
5
10
10
150
150
180
22
21
100
106
203
765
49
50
67
a
Surface normalized reaction rates expressed as turn-over-frequency
−
1
(TOF) at ±20% conversion. Reaction conditions: 150–180 °C, 2 hours,
(8 kJ mol ) and thus the equilibrium lies on the side of eth-
ylene glycol and the reaction does not proceed (with any sig-
nificant conversion). This explains our experimental finding
that without the addition of base the reaction does not pro-
ceed (Table 1). From glycerol the formation of lactic acid is
5
0 mg Cu/CNF. XGLY = glycerol conversion. SLA = selectivity toward
lactic acid.
be overcome for an industrial process in which the pH can
be controlled for instance by operating in semi-batch mode.
The use of more concentrated glycerol solutions did not
affect the lactic acid yield which was ~50%. By increasing the
reaction temperature to 180 °C almost full conversions was
achieved after 4 hours with a lactic acid selectivity of 67%.
Thus by increasing the glycerol concentration and tempera-
ture lactic acid could be produced at high selectivity and with
−
1
favourable even without KOH addition (−89 kJ mol ). Experi-
mentally however no glycerol conversion was observed over
Ni or Cu without KOH (Table 1). The transformation of glyc-
erol is thus kinetically limited over Ni and Cu catalysts in the
absence of the base as co-catalyst. The same is true for the
aerobic oxidation reactions (rxn 4 and 8) which have very neg-
ative Gibbs free energies values and are thus very favourable.
The role of the base in lowering the activation energy for
−
1
high reaction rate (765 h ). We note that these reaction rates
are comparable to highly active noble metals, e.g. 600 h for
−
1
14
these reactions has been discussed in literature. When we
1
8
Au–Pd bimetallic catalyst, during aerobic oxidation reaction
at 60 °C. Further enhancement of these rates is expected
at higher concentrations and temperatures because, unlike
the aerobic oxidation, the reaction is not limited by oxygen
diffusion.
compare the enthalpy (Δ H°) of these reactions we see that
r
the aerobic oxidation reactions are also very exothermic
(large negative Δ H° value). This implies that such a process
r
requires a large cooling duty and, because of the low reaction
temperature used to prevent over oxidation (e.g. 60 °C), thus
produces substantial amounts of low temperature waste heat.
The anaerobic process uses water as the oxidant and does
not require molecular oxygen. As a consequence the heat
produced in the reaction of glycerol to lactic acid is signifi-
Reaction thermodynamics
To further illustrate the benefit of the oxidation of polyols in
water we calculated the Gibbs free energy and enthalpy for
the transformations of ethylene glycol and glycerol. The most
important reaction products detected in this study were
glycolic acid from ethylene glycol and lactic acid from
−
1
−1
cantly lower (−27 kJ mol versus −338 kJ mol for the aero-
bic reaction). In case of the aerobic glycolic acid production
from ethylene glycol the reaction becomes endothermic by
−
1
−1
68 kJ mol (versus −520 kJ mol for the aerobic reaction). In
fact the difference in reaction enthalpy between the aerobic
and the hydrothermal process equals the heat of combustion
r
glycerol. Table 5 lists the standard Gibbs free energy (Δ G°)
and enthalpy (Δ H°) of these reactions for a neutral reaction
r
media (rxn 1 and 5) and for the reaction including the
neutralization in alkaline media (rxn 2 and 6). The aerobic
oxidation reactions (with molecular oxygen) are given as a
2
of H . Since both process require stoichiometric amounts of
base the anaerobic hydrothermal process can be viewed as
advantageous compared to the aerobic process because,
Table 5 Standard reaction Gibbs free energies and enthalpies of reaction for transformations of ethylene glycol and glycerol
Rxn
No.
Δ
r
G°
r
Δ H°
−1
Reaction
(kJ mol substrate)
Ethylene glycol to glycolic acid:
a
1
2
3
4
C
C
C
C
2
2
2
2
H
H
H
H
6
6
6
6
O
O
O
O
2
2
2
2
(l) + H
(l) + KOH(ia) = KC
2
O(l) = C
2
H
4
O
3
(l) + 2H
2
(g)
8
101
68
a
b
b
2
H
3 3
O (a) + 2H
2
(g)
−63
−410
−498
(l) + O
(l) + O
2
2
(g) = C H
4
O
3
(a) + H
2
O(l)
−466
2
(g) + KOH(ia) = K(C
2 3 3 2
H O )(a) + 2H O(l)
−520
Glycerol to lactic acid:
a
5
C
C
3
H
H
8
O
O
3
(l) = C
3
H
6
3
O (a) + H
2
(g)
−89
−4
a
6
3
8
3
(l) + KOH(ia) = KC
3
H
O
5 3
(a) + H
(a) + 2H O(l)
(g) + 2KOH(ia) = 2KC
2
(g) + H
2
O(l)
−155
−301
−360
−27
−294
−338
b
7
2C
2C
3
H
H
8
O
O
3
(l) + O
(l) + O
2
(g) = 2C
2
3
H
6
O
3
2
b
8
3
8
3
H
3 5
3
O (a) + 4H
2
O(l)
4
2 a
b
Calculated using HSC Chemistry for Windows.
ia) aqueous dissociated, (l) liquid, (g) gaseous.
Reaction temperature: 160 °C. Reaction temperature: 60 °C. Aggregation states (a) aqueous
(
2360 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
instead of low temperature heat (which cannot be used
economically), this process produces hydrogen as a valuable
co-product. For the production of glycolic acid from ethylene
glycol part of the hydrogen can be combusted in order to
drive the reaction while the remainder can be used elsewhere
in a bio-refinery system.
in trace amounts. Acetol was not observed but both glyceral-
dehyde and acetol have been proposed as intermediate in the
2
1,39
transformation of glycol to lactic acid.
We investigated
the reactivity of the possible intermediates and reaction prod-
ucts in aqueous solutions, in alkaline aqueous solutions and
in alkaline aqueous solution in the presence of Cu/CNF. The
results are summarized in Table 7. Only glyceraldehyde was
converted under solely hydrothermal conditions (i.e. without
metal catalyst and base) or with only the base. For all other
intermediates activity was only observed in the presence of
both the base and the Cu catalysts. Under solely hydrother-
mal conditions the observed products from glyceraldehyde
were lactic acid (89%) and 1,2PDO (11%) however only 40%
of the carbon could be accounted for probably due to
condensation/polymerization reactions (entry 1), indeed the
formation of solids was observed for these reactions. With
the addition of base (KOH) and without metal catalyst only
lactic acid formation was observed from glyceraldehyde but
again most of the starting carbon could not be accounted for
Effect of metal oxidation state
Copper was previously reported to be active both as metallic
0
(
Cu ) and metal oxide (Cu (I)O and Cu(II)O) catalyst for glycerol
2
3
3
transformation to lactic acid. The effect of the oxidation state
of our Cu catalyst was investigated by comparing the activity
of the catalyst in a reduced, passivated and oxidized state
(
(
Table 6). XRD was used to confirm the presence of the metallic
Cu ) and Cu(II)O crystalline phases. When a reduced catalyst
0
was exposed to air at room temperature, formation of a CuO
phase was observed in addition to the metallic Cu phase
(
2
Cu/CNFp). When this catalyst was calcined in static air at
50 °C the metallic phase disappeared completely and CuO
(
entry 2) and again solid formation was observed. The fact
was the only phase observed (CuO/CNF). The highest conver-
sion at 150 °C was observed for the reduced catalyst both
with Ni/CNF (97%) and with Cu/CNF (83%). The passivated
Ni catalyst (Ni/CNFp) was slightly less active (90% conver-
sion) than the reduced catalyst, whereas the activity of the
passivated Cu catalyst (Cu/CNFp) was significantly lower
the no 1,2PDO was formed with base only suggests that the
formation of lactic acid is enhanced much more significantly
by the presence of base then the formation of 1,2PDO.
Next we discuss the stability of the substrates when tested
with both the copper and the base catalyst (entries 3–7).
Under these conditions glycerol was one of the products
obtained from glyceraldehyde (entry 4). This shows that the
glycerol dehydrogenation is reversible. However, next to glyc-
erol, mostly oxidation products, glyceric acid, glycolic acid
(
29% conversion). The CuO and the NiO catalysts were
almost inactive. A possible explanation for the lower activity
of the passivated catalysts is that the CuO layer of the passiv-
ated metal catalyst first needs to be reduced in situ by the
substrate subsequently yielding the active metal catalyst. It is
expected that the in situ reduction of CuO/CNF and NiO/CNF
is more difficult which explains that these catalyst were inac-
tive. We therefore conclude that the metallic Cu is most likely
the active phase. The activity of the aforementioned Cu(II)O
catalyst might be due to the higher temperature (200–240 °C)
compared to this study (150 °C).
and tartronic acid were observed. This implies that also H
2
must be formed which is then likely the source of the hydro-
gen required to hydrogenate the glyceraldehyde to glycerol. It
also shows that the metal catalyst is essential to catalyse the
formation of glyceric acid. Starting with glyceric acid as sub-
strate tartronic and glycolic acid were formed, indicating that
the formation of glyceric acid from glyceraldehyde is not
reversible since no lactic acid and 1,2PDO were observed.
This shows that formation of glycolic acid might occur through
C–C cleavage of glyceric acid, although it cannot be excluded
that additionally it forms directly from glyceraldehyde.
Intermediate and stability study
In order to gain more insight into the possible reaction route
for the conversion of glycerol and to attempt to differentiate
between reactions taking place on the catalyst surface and in
the solution we investigated the reactivity of the observed
products (lactic acid, 1,2PDO and glyceric acid) and possible
intermediates; glyceraldehyde (2,3-dihydroxypropanal) and
The reactivity of acetol was investigated since it is the
product that would be formed if glycol undergoes a base
catalysed dehydration reaction. Both 1,2PDO and lactic acid
were formed. Thus both acetol and glyceraldehyde are possi-
ble intermediate in the formation of lactic acid and 1,2PDO
from glycerol. Finally, the reaction with 1,2PDO as substrate
(entry 6) showed that it was partially converted to lactic acid
whereas the lactic acid (entry 7) was found to be a stable
product under reaction conditions. This is in agreement with
−
1
acetol (1-hydroxyacetone) at low concentrations (1 g L ).
Glycolaldehyde was observed in some experiments but only
Table 6 Effect of metal oxidation state on catalytic activitya
31
previous reports and it can be concluded that the formation
of lactic acid from glyceraldehyde is irreversible.
Catalyst
Ni/CNF Ni/CNFp NiO/CNF Cu/CNF Cu/CNFp CuO/CNF
Reaction route
XGLY (%) 97
90
0
68
29
6
a
Based on the experimental results and literature a reaction
route is proposed for the formation of glycolic acid from
Reaction conditions: 4 hours, 150 °C, 1 wt% glycerol solution and
KOH : glycerol ratio 2.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2361

View Article Online
Paper
Catalysis Science & Technology
Table 7 Reactivity of intermediates and products over a Cu/CNF catalyst under alkaline hydrothermal reaction conditions
a
Product distribution (%)
C
bal
GLY
GLYAL
GEA
TA
GOA
1,2PDO
LA
Entry
Reactant
GLYAL
(%)
b
1
0
0
17
0
0
0
0
0
63
0
0
4
0
0
6
11
0
1
89
100
9
40
24
85
c
2
d
3
d
4
GEA
0
0
50
34
16
0
0
94
d
5
Acetol
1,2PDO
LA
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
40
0
95
60
94
98
95
d
6
d
7
100
a
b
−1
Percentage of carbon accounted for in the liquid phase after 15 hours reaction. Reactions with 1 g L of substrate at 150 °C for 4 hours in
c
−1
d
−1
water. Reactions with 1 g L of substrate at 150 °C for 4 hours in aqueous solution of KOH (KOH : substrate ratio 2). Reactions with 1 g L
of substrate at 150 °C for 4 hours with 50 mg Cu/CNF and KOH : substrate ratio 2. GLYAL = glyceraldehyde, GLY = glycerol, GEA = glyceric acid,
TA = tartronic acid, GOA = glycolic acid, 1,2PDO = 1,2-propanediol and LA = lactic acid.
ethylene glycol (Scheme 7) and for the reactions of glycerol
Scheme 8). From ethylene glycol we found that glycolic acid
was formed exclusively with Cu and both glycolic acid and
formic acid were formed using Ni. Based on the identifica-
tion of trace amount of glycolaldehyde in some of the reac-
observed the formation of glycolaldehyde as a by-product in
3
5
(
the aqueous phase reforming of ethylene glycol. The fact
that the dehydrogenation only occurs in alkaline media can
1
4
be explained based on the work of Davis et al. on the oxida-
tive dehydrogenation of ethanol. These authors show that
O–H bond activation for ethanol in aqueous media over Pt/C
catalysts is unlikely but the dissociative adsorption is facili-
tated in the presence of surface hydroxides. The resulting
alkoxide intermediate can also be formed by deprotonation
tion mixtures and the formation of H gas it is reasonable to
2
assume that the formation of glycolic acid proceeds via
3
3
the dehydrogenation of ethylene glycol to glycolaldehyde,
catalysed by both Ni and Cu. In a previous study we also
Scheme 7 Proposed reaction route for the formation of glycolic acid from ethylene glycol. For clarity products are shown as free acids.
2362 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Scheme 8 Proposed reaction routes for glycerol. For clarity products are shown as free acids. M denotes the metal catalyst.
of ethylene glycol in strong alkaline solution (the pK
a
of
temperatures it becomes possible to remove the adsorbed
hydrogen atoms from the metal as molecular hydrogen.
The difference between the Cu and Ni catalyst is the addi-
tional formation of formic acid with the latter which was
shown to proceed via glycolic acid. Because Ni is effective for
C–C cleavage (opposed to Cu) the formation of formic acid
ethylene glycol is ~14). Subsequent activation of the C–H of
the alkoxide intermediate then yields the glycolaldehyde. The
activation barrier for this step is also lowered due to the pre-
1
4
sents of adsorbed OH groups. The formation of glycolic
acid might proceed via a base catalysed Cannizzaro reaction
of the glyceraldehyde by disproportionation of two aldehydes
to form an alcohol and an acid. However the base alone cannot
catalyse this reaction and it is therefore more likely that the
reaction takes place on the metal surface where oxygen inser-
tion could occur via addition of hydroxide to the aldehyde,
4
5,46
might proceed via oxalic acid (as oxalate, not shown)
which was detected in trace amounts. Decomposition of
4
7
formic acid then yields CO (as (bi)carbonate) and H .
2
2
Scheme 8 shows the proposed mechanism starting from
2
1
glycerol based on that of Davis et al. and adjusted to incor-
porate the products observed in this study. Glycolic acid was
not reported in previous studies on the transformation of
1
4
after which a hydrogen abstraction will yield the acid. Since
aldehydes are partially hydrated to geminal diols in aqueous
solution it was suggested that the reaction proceeds through
2
1,30
glycerol in alkaline media with external hydrogen addition
or in inert atmosphere
4
3,44
30,33
a diol intermediate (Scheme 7 route I).
Another explana-
but was observed in this study with
tion is that the reaction proceeds via the metal catalysed
dehydrogenation of glyceraldehyde to glyoxal (ethanedial)
which is then followed by a rapid and irreversible internal
both Ni and Cu catalysts. The formation of glyceric acid from
glycol can proceed via a mechanism equivalent to the forma-
tion of glycolic acid from ethylene glycol (Scheme 7). Glycerol
is first dehydrogenated to glyceraldehyde over the metal
(both Ni and Cu) catalyst which reacts further to glyceric
acid. Glyceric acid can follow the same route again with
oxidation of the primary alcohol to obtain tartronic acid
although higher temperatures were required for this reaction.
The formation of glycolic acid was suggested to proceed through
a retro-aldol reaction of the glyceraldehyde since Cu is not
effective for C–C bond cleavage. This will yield glycolaldehyde
which can reacts to glycolic acid as shown in Scheme 7. Our
experiments confirm that glycolic acid is one of the products
when starting from glyceraldehyde but it is also formed from
glyceric acid and only in the presence of the Cu catalyst.
1
9
disproportionation (Scheme 7 route II). The difference
between the oxidative dehydrogenation and the anaerobic
hydrothermal process lies in the removal of the adsorbed
hydrogen. During the aerobic oxidation of alcohols oxygen is
not directly involved in the mechanism but is required to
4
4
remove the hydrogen for the catalyst surface. The adsorbed
hydrogen atoms are removed from the surface by reaction
with dissociatively adsorbed oxygen and close the catalytic
2
1
1
4,18
cycle by regenerating surface-bound hydroxide ions.
It
has been established that in the absence of molecular oxygen
no reaction occurs at reaction temperatures typical for cata-
lytic oxidation of alcohols (60–90 °C). We show that at higher
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2363

View Article Online
Paper
Catalysis Science & Technology
Since glyceric acid can react to form glyceraldehyde and this
reaction is irreversible, it appears that glycolic acid formation
actually proceeds via glyceric acid.
The dehydration of glycerol will yield acetol (in equilibrium
with 2-hydroxy allyl alcohol via keto–enol tautomerization)
which was never observed in this study. However their study
was performed without metal catalyst in basic solutions at
much higher temperatures (300 °C). Because glycerol dehy-
dration is endothermic it is possible that at higher tempera-
ture dehydration becomes the dominant pathway.
The reaction path for hydrogenolysis of a C–O bond in
glycerol to form 1,2PDO has also been discussed in literature.
Montassier proposed a mechanism for the hydrogenolysis of
polyols with Cu catalysts. They found that metallic copper is
responsible for the (de)hydrogenation and copper surface
hydroxyl groups are responsible for dehydroxylation via dehy-
The difference between the Cu and Ni catalyst is the for-
mation of additional C–C cleavage products with Ni; ethylene
glycol and formic acid (not shown in Scheme 8). Ethylene gly-
2
3
dration followed by hydrogenation. Whether this reaction
proceeds first via a dehydration (Scheme 7, route I) or a dehy-
drogenation (Scheme 7, route II) has also been investigated.
As noted by Davis et al. the dehydration of glycerol is unlikely
in dilute aqueous glycerol solutions due to the unfavourable
2
1,22
col can be formed by direct hydrogenolysis of glycerol
with Ni/CNF and can subsequently react to formic acid as
shown in Scheme 7. However more research is required to
identify the additional pathway availably over the Ni/CNF
catalyst in more detail.
In conclusion we propose that the formation of glycolic
and glyceric acid from polyols proceeds via a mechanism
similar to the oxidative dehydrogenation but hydrogen is
desorbed from the surface as molecular hydrogen. The mech-
anism for C–O scission leading to 1,2PDO and lactic acid is
under debate but the metal catalyst is required for both the
2
1
equilibrium. Auneau et al. showed that the reaction rate
for glycerol hydrogenolyis in alkaline media was actually
increased when the gas phase is switched from hydrogen to
inert, which point towards a dehydrogenation as the first
step. They evaluated reaction mechanisms over Ir and Rh
catalyst by calculation of the energy of the intermediates and
the transition states using DFT. They concluded that the
dehydrogenation pathway is kinetically favoured and the
formation of glyceraldehyde intermediate was favoured over
3
0,31
dihydroxyacetone
although the role of the base was not
taken into account. The formation of glyceraldehyde as the
first step is also in agreement with our results since it is also
required to explain the formation of glyceric acid. In order
to form 1,2PDO a competing reaction occurs where instead
of glyceric acid formation the aldehyde intermediate can
then undergo a base catalysed dehydration reaction to form
pyruvaldehyde (2-oxopropanal), which is in equilibrium with
the enol, 2 hydroxyacroline via keto–enol tautomerization.
Hydrogenation of these intermediates then gives 1,2PDO. For
this reaction the metal catalyst is required which explains
why 1,2PDO was not formed from glyceraldehyde with only
the base.
Route I can also explain the formation of lactic acid
which, starting from glyceraldehyde can be formed without
metal catalyst. The isomerization of glyceraldehyde to lactic
acid may occurs through dehydration of the glyceraldehyde
with subsequent disproportionation via intermolecular
3
0,31
Cannizzaro rearrangement of pyruvaldehyde,
also referred
3
3
to as benzilic rearrangement. This step is the equivalent of
the formation of glycolic acid from glyoxal in the mechanism
for ethylene glycol (Scheme 7). The base catalysed elimina-
tion of water from glyceraldehyde is expected to be easier
than the direct dehydration of glycerol because of the more
4
8
acidic α-hydrogens of glyceraldehyde. However when this
reaction was performed with only the base it also leads to the
formation of solid by-products, which is not the case in the
presence of the metal catalyst. The metal catalyst thus also
facilitates the formation of lactic acid, possibly by providing
a different dehydration pathway.
3
9
Base on isotope studies Zhang et al. came to the oppo-
site conclusion and suggest that lactic acid formation pro-
ceeds via dehydration as the first step (Scheme 2, route II).
Fig. 4 TEM micrographs of passivated Cu/CNF catalysts; fresh catalyst
(top) and spent catalyst after 15 hours reaction (bottom).
2364 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
dehydration and hydro-dehydrogenation steps, suggesting
that the mechanism takes place entirely on the metal surface
although the assistance of base is required.
Conclusion
Anaerobic oxidation of polyols into mainly glycolic acid and
lactic acid proceeds with high selectivity in alkaline aqueous
media at low temperatures, 150–180 °C, using CNF supported
Ni and Cu catalyst. The combination of both supported metal
catalyst and homogeneous base is essential. From ethylene
glycol we can produce glycolic acid with near 100% selectivity
at high conversion levels (82%) with Cu supported on CNF.
Compared to the aerobic process for the production of
glycolic acid (oxygen atmosphere, 60 °C) the use of anaerobic
conditions at somewhat higher temperatures has the benefit
of producing hydrogen as valuable co-product, however stoi-
chiometric amounts of base are required in both processes.
Starting from glycerol mainly lactic acid is formed with glyceric
acid and 1,2-propanediol as by-products. Higher temperatures
and longer reaction times favour the production of lactic acid
for which a maximum selectivity 67% was be obtained at full
glycerol conversion using Cu/CNF. With the Ni/CNF catalyst
C–C cleavage products, formic acid and ethylene glycol, were
also formed resulting in high hydrogen selectivity but lower
selectivity to valuable glycolic and lactic acid.
Catalyst stability
Since Cu is the most promising catalyst for the selective pro-
duction of glycolic acid from ethylene glycol and lactic acid
from glycerol we focussed our stability studies on the Cu/CNF
catalyst. The catalyst was characterized with Transmission
Electon Microscopy (TEM) as shown in Fig. 4. For both the
fresh and the spent catalyst Cu particles with sizes of 10–100 nm
were observed evenly distributed over the CNF support. The
Cu particles appeared to be somewhat larger for the spent
catalyst which was also apparent from XRD.
Fig. 5 shows the XRD diffractograms of the fresh catalysts
and of the spent catalyst after 15 hours reaction. We observed
mainly diffraction lines at 51° and 59° 2θ attributed to the
metallic Cu phase. In addition broad contributions due to
CuO at 41° and 45° 2θ were present. This is likely the result
of the passivation treatment before the measurement and
indicates a layer of CuO exist on the metallic nano-particles.
However we cannot exclude that a fraction of the smallest
particles is completely oxidized by the passivation treatment.
By applying the Scherrer equation an average value of ~30 nm Acknowledgements
was obtained for the metal crystallites of the fresh catalyst.
This research has been performed within the framework of
An increase to 44 nm was found was found after 15 hours
reaction. The loss of dispersion was also evident from the
decrease in catalytic activity when the catalyst was reused. A
decrease in conversion was observed from 40% to 13% after
using the catalyst in four consecutive runs (see Table 8).
the CatchBio program. The authors gratefully acknowledge
the support of the Smart Mix Program of the Netherlands
Ministry of Economic Affairs, Agriculture and Innovation and
the Netherlands Ministry of Education, Culture and Science.
References
1
Z. H. Xiao, S. H. Jin, M. Pang and C. H. Liang, Green Chem.,
013, 15, 891.
2
2
N. Ji, T. Zhang, M. Y. Zheng, A. Q. Wang, H. Wang,
X. D. Wang, Y. Y. Shu, A. L. Stottlemyer and J. G. G. Chen,
Catal. Today, 2009, 147, 77.
3
4
J. Y. Sun and H. C. Liu, Green Chem., 2011, 13, 135.
C. Montassier, D. Giraud and J. Barbier, Polyol Conversion
by Liquid Phase Heterogeneous Catalysis Over Metals, in
Studies in Surface Science and Catalysis, ed. M. Guisnet,
J. Barrault, C. Bouchoule, D. Duprez, C. Montassier and
G. Pérot, Elsevier, 1988, p. 165.
5
6
7
8
C. H. Zhou, J. N. Beltramini, Y. X. Fan and G. Q. Lu, Chem.
Soc. Rev., 2008, 37, 527.
R. D. Cortright, R. R. Davda and J. A. Dumesic, Nature, 2002,
Fig. 5 XRD diffractograms of the fresh catalysts and of the spent
catalyst after 15 hours reaction.
4
18, 964.
W. Dischert, C. Colomb and G. Bestel-Corre, US Pat.,
0130210097 A1, 2013.
Table 8 Recycle experiment for Cu/CNFa
2
Run#
Conversion (%)
M. Kataoka, M. Sasaki, A. R. G. D. Hidalgo, M. Nakano and
S. Shimizu, Biosci., Biotechnol., Biochem., 2001, 65, 2265.
1
2
3
4
40 ± 2
27 ± 2
15 ± 1
13 ± 1
9 H. Habe, Y. Shimada, T. Yakushi, H. Hattori, Y. Ano,
T. Fukuoka, D. Kitamoto, M. Itagaki, K. Watanabe,
H. Yanagishita, K. Matsushita and K. Sakaki, Appl. Environ.
Microbiol., 2009, 75, 7760.
a
Reaction conditions: T = 150 °C, 1 wt% glycerol solution, run time
4
hours.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 2353–2366 | 2365

View Article Online
Paper
Catalysis Science & Technology
1
1
1
1
1
1
1
1
1
0 A. A. Hong, K. K. Cheng, F. Peng, S. Zhou, Y. Sun, C. M. Liu
and D. H. Liu, J. Chem. Technol. Biotechnol., 2009, 84, 1576.
1 S. Mazumdar, J. M. Clomburg and R. Gonzalez, Appl.
Environ. Microbiol., 2010, 76, 4327.
2 J. R. M. Almeida, L. C. L. Favaro and B. F. Quirino,
Biotechnol. Biofuels, 2012, 5.
29 T. Miyazawa, Y. Kusunoki, K. Kunimori and K. Tomishige,
J. Catal., 2006, 240, 213.
30 F. Auneau, C. Michel, F. Delbecq, C. Pinel and P. Sautet,
Chem. - Eur. J., 2011, 17, 14288.
31 F. Auneau, L. S. Arani, M. Besson, L. Djakovitch, C. Michel,
F. Delbecq, P. Sautet and C. Pinel, Top. Catal., 2012, 55, 474.
32 F. Auneau, S. Noel, G. Aubert, M. Besson, L. Djakovitch and
C. Pinel, Catal. Commun., 2011, 16, 144.
33 D. Roy, B. Subramaniam and R. V. Chaudhari, ACS Catal.,
2011, 1, 548.
34 J. H. Bitter, J. Mater. Chem., 2010, 20, 7312.
35 C. C. I. L. T. van Haasterecht, K. P. de Jong and J. H. Bitter,
J. Energy Chem., 2013, 22, 257.
36 K. P. De Jong and J. W. Geus, Catal. Rev. Sci. Eng., 2000,
42, 481.
37 M. L. Toebes, J. H. Bitter, A. J. van Dillen and K. P. de Jong,
Catal. Today, 2002, 76, 33.
38 Z. Shen, F. M. Jin, Y. L. Zhang, B. Wu, A. Kishita and
K. Tohjiand H. Kishida, Ind. Eng. Chem. Res., 2009, 48(19), 8920.
39 Y. L. Zhang, Z. Shen, X. F. Zhou, M. Zhang and F. M. Jin,
Green Chem., 2012, 14, 3285.
40 A. Wawrzetz, B. Peng, A. Hrabar, A. Jentys, A. A. Lemonidou
and J. A. Lercher, J. Catal., 2010, 269, 411.
41 J. Liu, X. W. Chu, L. J. Zhu, J. Y. Hu, R. Dai, S. H. Xie, Y. Pei,
S. R. Yan, M. H. Qiao and K. N. Fan, ChemSusChem, 2010,
3, 803.
42 A. Roine, in HSC Chemistry, Outotec, 2007.
43 T. Mallat and A. Baiker, Chem. Rev., 2004, 104, 3037.
44 M. Besson and P. Gallezot, Catal. Today, 2000, 57, 127.
45 J. A. Onwudili and P. T. Williams, Green Chem., 2010,
12, 2214.
3 C. Bianchi, F. Porta, L. Prati and M. Rossi, Top. Catal., 2000,
1
3, 231.
4 B. N. Zope, D. D. Hibbitts, M. Neurock and R. J. Davis,
Science, 2010, 330, 74.
5 S. Carrettin, P. McMorn, P. Johnston, K. Griffin and
G. J. Hutchings, Chem. Commun., 2002, 696.
6 Y. H. Shen, S. H. Zhang, H. J. Li, Y. Ren and H. C. Liu,
Chem. - Eur. J., 2010, 16, 7368.
7 A. M. Sakharov, P. A. Sakharov and G. E. Zaikov, Mol. Cryst.
Liq. Cryst., 2012, 555, 168.
8 M. B. Griffin, A. A. Rodriguez, M. M. Montemore,
J. R. Monnier, C. T. Williams and J. W. Medlin, J. Catal.,
2
013, 307, 111.
1
2
9 L. Prati and M. Rossi, J. Catal., 1998, 176, 552.
0 A. Torres, D. Roy, B. Subramaniam and R. V. Chaudhari,
Ind. Eng. Chem. Res., 2010, 49, 10826.
2
2
1 E. P. Maris and R. J. Davis, J. Catal., 2007, 249, 328.
2 E. P. Maris, W. C. Ketchie, M. Murayama and R. J. Davis,
J. Catal., 2007, 251, 281.
2
2
2
2
2
2
3 C. Montassier, D. Giraud, J. Barbier and J. P. Boitiaux, Bull.
Soc. Chim. Fr., 1989, 148.
4 C. Montassier, J. C. Menezo, L. C. Hoang, C. Renaud and
J. Barbier, J. Mol. Catal., 1991, 70, 99.
5 E. D'Hondt, S. V. de Vyver, B. F. Sels and P. A. Jacobs, Chem.
Commun., 2008, 6011.
6 D. Roy, B. Subramaniam and R. V. Chaudhari, Catal. Today,
46 A. Costine, J. S. C. Loh, F. Busetti, C. A. Joll and A. Heitz,
Ind. Eng. Chem. Res., 2013, 52, 5572.
47 J. A. Onwudili and P. T. Williams, Appl. Catal., B, 2013,
132–133, 70.
2
010, 156, 31.
7 C. Pendem, P. Gupta, N. Chaudhary, S. Singh, J. Kumar,
T. Sasaki, A. Datta and R. Bal, Green Chem., 2012, 14, 3107.
8 R. B. Mane and C. V. Rode, Green Chem., 2012, 14, 2780.
48 J. ten Dam and U. Hanefeld, ChemSusChem, 2011, 4, 1017.
2366 | Catal. Sci. Technol., 2014, 4, 2353–2366
This journal is © The Royal Society of Chemistry 2014