Mild electrocatalytic hydrogenation of lactic acid to lactaldehyde and propylene glycol

Article abstract of DOI:10.1016/j.jcat.2006.11.009

Reduction of fermentation-derived lactic acid (LA) offers a renewables-based pathway to propylene glycol (PG), a large-scale commodity chemical, currently manufactured by the oxidation of petroleum-derived propene. Complementing our previously described catalytic hydrogenation of LA to PG, we now report electrocatalytic hydrogenation (ECH) of LA in an aqueous electrolyte using constant current electrolysis. A reticulated vitreous carbon (RVC) electrode serves to agglomerate, support, and supply current to a 5% Ru/C powder catalyst, the same catalyst used in the classical hydrogenations. The ECH conditions are mild (ambient pressure, 70 °C vs 1500 psi H₂, 150 °C) relative to the chemical hydrogenation. More surprisingly, the major electrohydrogenation product is lactaldehyde (LAL), with small quantities of PG also formed. Variable current studies in the range of 10-100 mA show an increase in product yields and a shift in selectivity toward PG with increasing current. Experiments carried out with different acids as electrolytes reveal a distinct effect of the anion on the yields of the two products. In situ ATR-FTIR studies of the ECH of LA point to a chelating bidentate carboxylate adsorption mode for lactate on the Ru surface and offer insight into the effects of electrolyte anions on surface adsorption and reactivity.

Full text of DOI:10.1016/j.jcat.2006.11.009

Journal of Catalysis 246 (2007) 15–28
www.elsevier.com/locate/jcat
Mild electrocatalytic hydrogenation of lactic acid to lactaldehyde and
propylene glycol
a
a,∗
a
b
Tulika S. Dalavoy , James E. Jackson , Greg M. Swain , Dennis J. Miller ,
c
c
Jie Li , Jacek Lipkowski
a
Department of Chemistry, Michigan State University, East Lansing, MI 48824, USA
b
Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA
c
Department of Chemistry and Biochemistry, University of Guelph, Guelph, ON N1G2W1, Canada
Received 13 August 2006; revised 31 October 2006; accepted 1 November 2006
Available online 12 December 2006
Abstract
Reduction of fermentation-derived lactic acid (LA) offers a renewables-based pathway to propylene glycol (PG), a large-scale commodity chem-
ical, currently manufactured by the oxidation of petroleum-derived propene. Complementing our previously described catalytic hydrogenation of
LA to PG, we now report electrocatalytic hydrogenation (ECH) of LA in an aqueous electrolyte using constant current electrolysis. A reticulated
vitreous carbon (RVC) electrode serves to agglomerate, support, and supply current to a 5% Ru/C powder catalyst, the same catalyst used in the
◦
◦
classical hydrogenations. The ECH conditions are mild (ambient pressure, 70 C vs 1500 psi H , 150 C) relative to the chemical hydrogenation.
2
More surprisingly, the major electrohydrogenation product is lactaldehyde (LAL), with small quantities of PG also formed. Variable current stud-
ies in the range of 10–100 mA show an increase in product yields and a shift in selectivity toward PG with increasing current. Experiments carried
out with different acids as electrolytes reveal a distinct effect of the anion on the yields of the two products. In situ ATR-FTIR studies of the ECH
of LA point to a chelating bidentate carboxylate adsorption mode for lactate on the Ru surface and offer insight into the effects of electrolyte
anions on surface adsorption and reactivity.
©
2006 Elsevier Inc. All rights reserved.
Keywords: Electrocatalytic hydrogenation; Lactic acid; Lactaldehyde; Propylene glycol; Reticulated vitreous carbon; Organic electrochemistry; Electrosynthesis;
Green chemistry; ATR; SEIRAS; FTIR
1
. Introduction
in water, this conversion requires high temperature and pressure
conditions (150 C, 1200 psi H2) [1]. Interestingly, lactic and
◦
Prompted by the long-term imperative to replace fossil hy-
α-amino acids have turned out to be especially favorable sub-
strates, undergoing reduction substantially faster than unsubsti-
tuted propanoic acid [1,2]. Indeed, conditions severe enough to
reduce simple alkanoic acids lead to degraded selectivities, as
C–O and C–C bond hydrogenolyses begin to compete with car-
boxylate reduction. Because the overall heats of hydrogenation
of the substituted and the parent carboxylic acids are almost
drocarbons with renewable resources as the basis for the chem-
ical and energy industries, the search is on for alternative paths
to chemical building blocks from biomass-derived intermedi-
ates such as glucose, glycerol, and organic acids. In recent
years, we described aqueous phase heterogeneous catalytic hy-
drogenation of lactic acid (LA; IUPAC name, 2-hydroxypro-
panoic acid) to propylene glycol (PG; IUPAC name, 1,2-pro-
+
equal, it appears that the α-OH or –NH groups exert a kinetic
3
9
panediol), an important (>10 kg/year in the United States)
influence independent of the underlying thermochemistry.
Electrocatalytic hydrogenation (ECH) offers a mild alterna-
tive to traditional chemical catalytic methods. Here the unsatu-
rated organic compound (Y=Z) is reduced by hydrogen formed
from electrolytic acid decomposition at the surface of an elec-
trode made of catalytic metal, as shown in Scheme 1. Because
commodity chemical used widely across industrial sectors. Al-
though its yield is >95% in the presence of a 5% Ru/C catalyst
*
Corresponding author. Fax: +1 517 353 1793.
E-mail address: jackson@chemistry.msu.edu (J.E. Jackson).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.11.009

16
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
Scheme 1. Electrocatalytic hydrogenation and hydrogen evolution.
the hydrogen is produced in situ on the catalyst surface, this
strategy eliminates three difficulties: (1) the need for high pres-
sures to achieve adequate concentrations of (poorly soluble)
hydrogen gas, (2) its mass transport, and (3) its splitting into
hydrogen atoms on the catalyst surface. Compared with chem-
ical catalytic hydrogenation, ECH should enable milder, and
hence more selective, reaction. Safety would be improved as
well; with no external hydrogen source required, the need to
store and handle potentially hazardous high-pressure hydrogen
gas would be obviated.
Although a number of authors have explored ECH of unsatu-
rated substrates such as phenol [3,4], diketones [5], enones [6],
and aldehydes [7], on electrodes such as Pt, Raney Ni, Raney
Co, and Pd/carbon felt, to the best of our knowledge, no such
studies have been reported for free carboxylic acids. The use
of aqueous electrolyte mirrors our chemical catalytic studies, in
which water was chosen as the native (and “greenest”) medium
for biomass-derived organic acids. We have since found that
the aqueous environment is uniquely favorable; organic sol-
vents generally degrade reaction rates and selectivities relative
to those in water [8].
drogenation, and the mechanism by which the α-position func-
tional groups can facilitate hydrogenation of carboxylic acids.
This paper describes a series of ECH studies of LA using
a RVC electrode suffused with a powder catalyst. The same
5% Ru/C powder catalyst was used in our studies of the chem-
ical catalytic hydrogenation [2], so the physical form of the
catalyst, and presumably the essential components of the cat-
alytic process, were retained, allowing maximum comparability
between the new electrochemical and the chemical processes.
The RVC’s three-dimensional network of large pores provides
electrically conductive host sites to support the powder catalyst
particles. With a high overpotential for hydrogen evolution, it is
an electrochemically stable support material and not itself ex-
pected to effect electrohydrogenation or water reduction, as our
control experiments confirm. Previous work using this type of
electrode was reported by Menard and coworkers [3,9–12] in
the context of ECH of phenol and other organic compounds.
This work’s unexpected finding is that lactaldehyde (2-hy-
droxypropanal) (LAL) is the major product instead of the ex-
pected PG, which only occasionally appears as a minor part
of the product mixture. This thermodynamically surprising re-
sult apparently reflects the relative rates of reduction and sur-
face desorption/adsorption, together with the aqueous hydration
properties of the starting materials and the two products.
This project’s long-term goal is a general process for large-
1
scale, aqueous-phase stereoretentive catalytic hydrogenation
of renewables-derived carboxylic acids under temperature and
pressure conditions milder than those of current state-of-the-
art chemical conversions (batch or continuous). Electrochem-
istry is also amenable to in situ observation via attenuated total
reflection (ATR) infrared and Raman spectroscopies, offering
insights into the mechanistic details of metal-catalyzed hy-
drogenations. We have used such spectroelectrochemical tech-
niques to probe the mode of binding of the substrate on the elec-
trode surface, the nature of the intermediates formed during hy-
2. Experimental
S-LA and PG were purchased from Aldrich and were used
without further purification. 5% Ru/C was purchased from
PMC Catalysts.
2
.1. Electrochemical cell setup
Electrohydrogenation studies of LA in several aqueous elec-
1
In the heterogeneous catalytic hydrogenation of chiral lactic acid [1], the
trolyte solutions were carried out in the glass electrochemical
cell depicted below (Fig. 1). It is a two-compartment cell with
the anode and the cathode compartments separated by a glass
frit to prevent the diffusion of gaseous products. The reference
high temperature and pressure causes some racemization at the stereocenter of
lactic acid, lowering the enantiomeric excess to ∼90%. The rate of racemiza-
tion decreases with decrease in temperature. The milder conditions of alanine
hydrogenation [2] allow for completely stereoretentive reduction.

T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
17
Fig. 1. Schematic diagram of the electrochemical cell.
0
electrode was Ag/AgCl (E = −0.35 V vs SCE). The counter
electrode is Pt wire. The RVC working electrode (dimensions
4
.0 × 3.0 × 0.5 cm; pore size, 100 pores per inch [ppi]) was
connected to the external circuit via a copper wire, positioned
to minimize surface area exposed to solution; electrochemistry
on the copper surface can thus be assumed to be negligible.
A voltmeter was connected between the reference and work-
ing electrode to measure the potential of the working electrode.
Controlled current was applied using a model 273 potentio-
stat/galvanostat from EG&G Princeton Applied Research. The
temperature of the system was controlled using a heating tape
wrapped around the cell and connected to a temperature con-
troller from Omega.
(a)
2
.2. Electrocatalytic hydrogenation
Before each experiment, 1.0 g (dry weight) of 5% Ru/C was
added to the electrolyte solution and the solution was stirred
for 2 h under cathodic polarization (I = 20–40 mA) to ensure
proper agglomeration of the catalyst in the pores of the RVC.
The potential of the working electrode during this process was
in the range of −100 to −400 mV (vs Ag/AgCl) depending on
the electrolyte and the temperature of the cell. The stirring rate
was kept low to prevent the agglomerated catalyst from being
washed off the pores. The RVC surfaces before and after ag-
glomeration were examined by scanning electron microscopy
(b)
Fig. 2. SEM of RVC working electrode (100 ppi): (a) bare RVC electrode,
b) RVC electrode after entrapment of Ru/C via agglomeration for 2 h at 20 mA
(
under cathodic polarization.
(SEM) as shown in Fig. 2. Elemental composition as measured
via energy-dispersive spectral (EDS) analysis showed the aver-
age surface Ru content to be about 5.5 ± 0.1%. After the ag-
glomeration, 1 mL of LA solution in the respective electrolyte
was added so that the final concentration of LA in solution was
HPLC with refractive index detection, using a Biorad HPX87H
ion exchange column at 65 C. The mobile phase was 5 mmol
H2SO4, run isocratically at a flow rate of 0.6 mL/min.
◦
0
7
.1 wt% or 11.1 mmol, and the total volume of the solution was
5 mL. Samples were withdrawn for analysis immediately after
2.4. In situ ATR-FTIR
addition of LA and at intervals during the electrohydrogenation.
The sample cell consisted of a hemispherical Si window
2
.3. Analysis of products
(25 mm diameter) and a Teflon holder, screwed onto a stainless
steel base plate. The glass ATR cell (about 20 mL in volume)
was mounted on the base plate and sealed with an o-ring to pre-
vent leaks. The surface area of the internal reflection element
Reaction mixtures were filtered using Millipore syringe fil-
ters (0.25 µm) to remove the catalyst and then analyzed by

18
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
2
(
IRE) exposed to the solution was 3.14 cm . The cell consisted
Table 1
Yields of products of ECH of lactic acid at various temperatures in 0.01 M
of a water jacket through which water was circulated to main-
tain a constant temperature. The counter electrode was a Pt
wire, which was attached to the cell, and the Ag/AgCl refer-
ence electrode was connected through the salt bridge. The Si
window, coated with a thin film of Ru on Au, was used as the
working electrode. Electrical contact with the window was at-
tained through a gold wire sealed to a Teflon holder, which was
mounted in a glass tube. The gold wire was sealed to the glass
tube at the top and the entire assembly was screwed onto the
cell. A constant potential was applied using a PAR 173 poten-
tiostat. Before the ECH experiments, an Au film was chemically
deposited on the Si IRE using the method described by Osawa
and co-workers [13]. Once the Au film was formed, the window
was rinsed with milli-Q water and then immersed in a solution
H SO after 9 h; current = 40 mA
2
4
Temperature
◦
LA remaining
LAL
(%)
PG
(%)
Material balance
(%)
(
C)
(%)
2
5
7
9
5
71.8
63.0
66.3
67.0
1.2
3.3
4.2
5.7
1.6
3.5
5.1
9.8
74.6
69.8
75.6
82.5
0
0
0
2
solution was 3.14 cm . Each spectrum was acquired by inte-
−1
grating 1000 interferograms with a resolution of 4 cm , to
achieve the required sensitivity. The acquisition time was 8 min
per spectrum. All spectra are shown in absorbance units, de-
fined as A = −log(I/I ), where I and I are the intensities of
0
0
the reflected radiation at the sample and reference solutions, re-
spectively.
3+
of 1 mmol RuCl3·xH2O to deposit an adlayer of Ru on the
Au surface. The electroreduction of Ru3 to Ru was carried
+
0
out in the spectroelectrochemical cell by applying a potential of
2
.5. Ab initio calculations
0
.0 mV (vs Ag/AgCl).
After the cell was assembled, the water circulator was turned
Heats of formation were calculated from atomization en-
◦
on. Once the temperature was stabilized at 30 C, about 19 mL
of water was added to the cell, and an IR spectrum was
recorded. Then the required amount of electrolyte was added to
the water so that the final electrolyte concentration was about
.01 M. Argon gas was bubbled through the solution for about
0 min to remove oxygen and other dissolved gases. A con-
thalpies using the G3 composite ab initio method [14] as im-
plemented in the Gaussian 98 code [15]. A 32-processor SGI
Origin 3400 supercomputer was used for all of the calculations.
The energetic consequences of solvation in water were approx-
imated using the polarized continuum model (CPCM), calcu-
lated at the B3LYP/6-311 + G(d,p)//MP2/6-31G* level [16].
0
3
stant potential of 0.0 V (vs Ag/AgCl) was applied for 10 min
to reduce the Ru3 adlayer to metallic Ru. The working elec-
trode was then polarized at −400 mV, and an IR spectrum was
recorded. Then the required amount of LA solution was added
so that the final concentration of LA in the system was 0.01 M
and this point marked the beginning of ECH. The working elec-
trode was maintained at a constant potential of −400 mV, and
IR spectra were recorded every 8 min for 20 h.
+
3. Results and discussion
As summarized in Tables 1 and 2 and in Scheme 2, ECH of
LA yields LAL and/or PG under all conditions studied. The re-
action’s electrical efficiency was found to be highly dependent
on the nature of the cell and the stability and dynamics of the
system. The reductive catalyst agglomeration procedure before
substrate addition was used to ensure that all Ru in contact with
the RVC was in the metallic state and saturated with hydrogen.
This step is equivalent to catalyst reduction in traditional batch
hydrogenation.
The present ECH studies were carried out with 5% Ru/C/
RVC to allow us to compare the results with the previously
reported chemical hydrogenation of LA [1], because this cat-
alyst was found to give the maximum rate of hydrogenation in
FTIR spectra were taken with a Nicolet 20SX/C FTIR spec-
trometer, equipped with a liquid nitrogen-cooled linearized
narrow band MCT-B detector. The sample compartment was
purged using CO2 and H2O free air from a Puregas heat-
less dryer. The source (Everglow) was focused at the elec-
trode/electrolyte interface by passing it through the Si window.
◦
The incident angle was 30 from the surface normal, the beam
diameter was 8 mm, and the area of the IRE exposed to the
Scheme 2. Lactic acid hydrogenation reaction along with the results of the ab initio calculations of heats of formation (kcal/mol). Total heats of formation listed
include implicit H (ꢀH = 0) and H O (ꢀH (gas) = −57.8 kcal/mol; ꢀH (liq) = −68.3 kcal/mol) reactants and byproducts along with the species explicitly
2
f
2
f
f
shown.

T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
19
Table 2
1H NMR. Almost complete conversion of glycolic acid to gly-
colic aldehyde was observed. Similarly, glyceric acid and ala-
nine were converted to glyceraldehyde and 2-amino propanal,
respectively. All of these products are well known, and their
identification and quantification is straightforward.
Final yields (%) of products after lactic acid (11.1 mM) electrohydrogenation
in various electrolytes; current = 100 mA, temperature = 70 C
◦
Electrolyte
Lactic acid
Lactal- 1,2-pro- Material Current
dehyde panediol balance efficiency
(% remaining)
Time = 48 h
As an additional control experiment, LAL, synthesized
chemically using the procedure described by Nambiar et al.
[17], was subjected to electrohydrogenation in 0.01 M HClO4
−
−
−
7
6
6
0
0
0
.1 M H SO
.1 M HCl
.1 M HClO
91.8
2.6
64.0
25.4
2.7
6.6
0.7
97.1
79.5
80.6
7.48 × 10
7.17 × 10
2.48 × 10
2
4
8.9
54.5
4
◦
at 70 C and 100 mA current. There was negligible conver-
Time = 21 h
sion of LAL to PG. LAL produced in situ from pyruvaldehyde
dimethyl acetal and from pyruvaldehyde was also not signifi-
cantly hydrogenated to PG.
−
−
−
6
5
5
0
0
0
.01 M H SO 37.4
.01 M HCl
.01 M HClO
29.1
80.3
58.6
4.5
2.9
1.1
71.0
92.6
59.9
8.09 × 10
1.83 × 10
1.29 × 10
2
4
9.4
0.2
4
3
.2. Effect of temperature
those studies. The same factors that lead to the high rate in the
chemical hydrogenation—namely the stabilization of surface
bound intermediates, electronic properties and particle size of
the metal, structure and properties of the support and extent of
adsorption of various species on the metal and support—would
also be expected to favor the ECH process.
Electrohydrogenation experiments with 11.1 mM LA in
◦
0
.01 M H SO were carried out at 25, 50, 70, and 90 C for 9 h
2
4
using a constant current of 40 mA. The temperatures used span
the practical temperature window for an aqueous electrolyte at
atmospheric pressure, and they approach the lowest tempera-
ture (90 C) at which conversion was observed in the chemical
hydrogenation of LA.
This choice drew on our early attempts at ECH of LA using
◦
Ru/carbon felt and 10% Ru/Pt mesh electrode in 0.01 M H SO4
2
under similar conditions. These efforts produced only LAL and
in much lower yields compared with the current work; forma-
tion of PG was seen only when the electrohydrogenation was
carried out using 5% Ru/C catalyst agglomerated in the RVC
working electrode. Evidently, the hydrogenation of LA to PG
depends on specific characteristics of the Ru/C catalyst, such as
the nature and distribution of pores and of the Ru nanoparticles
in these pores. This notion was confirmed by similar experi-
With rising temperature, yields of LAL and PG increased
(see Table 1) indicating increasing current efficiency. At higher
temperatures, both ECH and hydrogen desorption by the Tafel
reaction accelerate; the ECH relative rate increase, however,
suggests that ECH competes more effectively for surface H
atoms at higher temperatures. Relative rates of hydrogenation
and hydrogen evolution are thought to depend on the intrin-
sic thermochemistry of the reduction, adsorption modes and
energies of the reactants and products, and the activity of
chemisorbed hydrogen [4a]. Similar increases in the conver-
sion and current efficiency with increasing temperature have
been previously noted in the ECH of (a) lignin models such
as guaiacol [9] and 4-phenoxy phenol [10] to phenol and cy-
clohexanol over Raney Ni powder and 5% Pd/C in an RVC
electrode using 1 mol NaOH as the electrolyte, where Menard
et al. found a slight increase in conversions with increase in the
ments with 5% Ru/SiO and 5% Ru/BaSO particles suffused
2
4
into the RVC. Here the catalyst supports were of lower poros-
ity compared with carbon, and the Ru particles were larger and
distributed mainly on the outer surface. In both cases, LAL was
the only ECH product.
To maximize current efficiency, ECH should be carried out
under conditions that minimize the competing hydrogen evo-
lution by the Tafel or Heyrovski processes [3,4]. Beyond the
control experiments below, reaction optimization studies were
carried out to explore the effects of temperature, current, and
electrolyte identity and concentration.
◦
temperature from 25 to 50 C; (b) 4-amino-5-nitrosodimethyl
uracil on a foamed nickel cathode in (NH4)2SO4 solution, stud-
ied by Hu et al. [18]; (c) phenol to cyclohexanol on dispersed
Pt electrode in 0.05 mol H2SO4, studied by Amouzegar and
Savadogo [19]. Interestingly, in the ECH of vat dyes such as
indigo in 1 mol NaOH, Roessler et al. [20] observed an in-
3
.1. Control experiments
◦
A control experiment carried out at 90 C using the RVC
◦
with no catalyst in 0.01 M H2SO4 showed no LA conversion.
Thus, the RVC alone is incapable of hydrogenation. Another
control with Ru/C catalyst and gaseous hydrogen (1 atm) run
crease in conversion up to 60–80 C, but current efficiency
and conversion then dropped off above this temperature; these
findings were interpreted in terms of the competition between
hydrogen incorporation into the substrate and decreasing sur-
face concentration due to H2 desorption. Bryan and Grimshaw
[21] found that efficiency increased with temperature without
changing the C=C versus C=O product selectivity in the ECH
of 4-phenylbutene-2-one to 4-phenylbutan-2-one on a Ni cath-
ode in ethanol containing H2SO4. In all of these cases, the
reported trends were believed to be due to greater increase
in the rate of hydrogenation compared with hydrogen evolu-
tion.
◦
at 90 C in the absence of the RVC similarly showed no reac-
tion, confirming that Ru needs to be in electrical contact for
hydrogenation of LA to occur at the mild conditions of this
study.
To confirm the unexpected formation of LAL via ECH of
LA, glycolic (2-hydroxyacetic) acid, the C analogue of lac-
2
tic, was subjected to electrohydrogenation in 0.01 mol HClO4
◦
at 70 C and 100 mA current for 21 h. The products of elec-
trohydrogenation were analyzed by HPLC-RI, ESI-MS, and

20
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
◦
Fig. 3. Yield of lactic acid reduction products after electrohydrogenation in 0.01 M H SO for 9 h at 90 C. Only lactaldehyde and propylene glycol were detected;
2
4
“total conversion” refers to the sum of these products.
3
.3. Effect of current
The case of LA hydrogenation appears similar to the afore-
mentioned findings of Mahdavi et al., but more complex due
to the intermediate dehydration step involved before LAL can
be further hydrogenated to PG (Scheme 2). With increasing cur-
rents, yields of LAL and PG decrease and increase, respectively,
while the total conversion remains almost constant. As shown
in Scheme 2, the simplest interpretation is that LA conversion
to PG occurs via two sequential hydrogen additions with LAL
hydrate and LAL as discrete intermediates. At low current, dis-
placement of the relatively weakly bound LAL hydrate (1,1,2-
propanetriol) by LA is faster than its dehydration and capture
by surface hydrogens. At higher currents LAL (also relatively
weakly bound) would be more quickly converted to PG due to
the availability of more surface hydrogen atoms. But the nearly
complete absence of PG at low currents suggests another factor:
base-promoted acceleration of LAL formation at the surface via
dehydration of LAL hydrate in the diffusion layer. Like other
carbonyl hydrates, LAL hydrate’s equilibration with the free
carbonyl form is accelerated at pH values both above and be-
low neutral [23]. At higher currents, solution protons are more
quickly adsorbed and reduced, so the pH at the cathode surface
is also higher. Not only are more hydrogen atoms available for
reaction, but also the concentration of free LAL in proximity to
the surface is enhanced, increasing PG formation.
Electrohydrogenation of 11.1 mM LA in 0.01 M H2SO4
◦
for 9 h at 90 C at different currents indicated that with in-
creasing current, the yield of PG increases and that of LAL
decreases slightly, as shown in Fig. 3. Like Polcaro et al. [7]
in their studies of benzaldehyde and acetophenone reduction,
we assume that the density of surface H atoms increases with
increasing current, as does reaction of LA, but the compet-
ing Tafel reaction (second order in H atoms) accelerates more.
Viewed this way, the fact that the current efficiency of LA
reduction is maximal at 10 mA suggests only a need for iso-
lated hydrogen atoms. On the other hand, the conversion of
LA to PG at low current (10 mA) is close to zero and rises
with increasing current, in spite of the attendant increased rate
of hydrogen evolution. Assuming the sequential mechanism of
Scheme 2, the aldehyde reduction step appears to require ad-
sorbed molecular hydrogen (or at least closely spaced atoms)
on the surface. Although one might hope that the kinetic stud-
ies of classical hydrogenation would allow a distinction to be
made between surface-bound H molecules and H atoms, mod-
2
els invoking both situations have been successful at modeling
the kinetic behaviors of LA and the closely analogous alanine
reductions [1,2].
In the ECH of cyclohexenone using a nickel boride elec-
trode, Mahdavi et al. [6] observed a decrease in the conversion
efficiency with increasing current density due to a relatively
greater increase in the rate of hydrogen evolution compared
with hydrogenation. However, there was an increase in the fur-
ther hydrogenation of cyclohexanone to cyclohexanol. This was
explained by the idea that the more negative potential at higher
current densities would activate the chemisorbed hydrogen for
faster hydrogenation of the cyclohexanone molecules generated
at the electrode surface by cyclohexenone hydrogenation. Cy-
clohexanone is expected to be more weakly adsorbed on the
electrode surface compared with the conjugated enone, cyclo-
hexenone. At lower current densities, cyclohexanone hydro-
genation would be too slow to compete with desorption and dif-
fusion to the bulk solution. On the other hand, Senda et al. [22]
saw an increase in ECH of 4-t-butylcyclohexanone to alcohol
with decreasing current.
The increase in the rate of formation of PG at higher cur-
rent may also be favored by the higher electrode surface en-
ergy at more negative potential, resulting in higher mobility
of H atoms, similar to the phenomenon observed by Mahdavi
et al. [6] in the ECH of cyclohexenone. The trend observed
here also points to the involvement of overpotential deposited
hydrogen (OPD-H), rather than underpotential deposited hy-
drogen (UPD-H) in the ECH of LA. UPD-H atoms are strongly
bound to the electrode surface and are believed to be formed
on Ru at around 0.0 V, whereas the weakly bound OPD-H
atoms are formed at more negative potentials and may coex-
ist with UPD-H. These weakly bound H atoms were found to
be involved in the hydrogen evolution reaction [44] and in the
ECH of the olefinic double bond in substrates such as maleic
acid [24].
An alternative to the sequential interpretation of Scheme 2
involves independent paths: parallel reactions might form LAL

T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
21
and PG simultaneously from LA adsorbed on two different
types of catalytic sites. In the ECH of benzaldehyde and ace-
tophenone by Polcaro et al. [7], the aldehydes were converted to
two different products, the corresponding alcohol and hydrocar-
bon; the formation of the two types of products were assumed
to involve parallel reactions of the aldehyde adsorbed on differ-
ent catalytic sites. Adsorption of aldehyde on low coordination
sites such as steps and kinks was assumed to lead to the forma-
tion of hydrocarbon whereas the alcohol was thought to form
by adsorption on high co-ordination sites. Consistent with this
key role for catalyst structural features (e.g., terraces, step edges
and kinks), the ratio of the two types of products were found
to depend on catalyst particle size, morphology and method of
preparation. Lacking information on the reactivity of multiple
catalyst morphologies, we cannot rule out this two-site interpre-
tation of the LAL and PG production. However, the sequential
scheme is simple, consistent with all observations, and serves
as the working model for this discussion.
(a)
3
.4. Effect of electrolyte
Electrohydrogenation of 11.1 mmol LA was carried out in
◦
various electrolytes at 70 C and 100 mA current. Electrolytes
studied were H2SO4, HClO4, and HCl at concentrations of 0.01
◦
and 0.1 M. Although the ECH rate was maximal at 90 C, the
7
◦
0 C temperature was chosen to prevent excessive water evap-
oration from the system during the long run times. The experi-
ments with 0.01 M electrolyte could run only for 21 h because
of the system’s increasing resistance as protons were depleted.
The results, summarized in Figs. 4a and 4b, show consider-
able variation in the rates of LAL and PG formation in different
electrolytes due to differences in the anions’ reactivity and cat-
alyst surface adsorption. The final yields of the products and
current efficiencies are listed in Table 2. The low material bal-
ances may be due to sequestration of products in the catalyst’s
activated carbon support. Alternatively, losses to volatiles are
possible; formation of volatile products such as methane have
been seen in the more severe chemical hydrogenations of car-
boxylic acids with 5% Ru/C.
(
b)
Fig. 4. Time course of product formation from electrohydrogenation of lac-
tic acid in various electrolytes (0.1 M) at 70 C. (a) ( ) Lactaldehyde
◦
(
0.1 M HClO ); ( ) 1,2-propanediol (0.1 M HClO ); ( ) lactaldehyde
4
4
(0.1 M H SO ); ( ) 1,2-propanediol (0.1 M H SO ); ( ) lactaldehyde (0.1 M
HCl); ( ) 1,2-propanediol (0.1 M HCl). (b) ( ) Lactaldehyde (0.01 M
HClO ); ( ) 1,2-propanediol (0.01 M HClO ); ( ) lactaldehyde (0.01 M
H SO ); ( ) 1,2-propanediol (0.01 M H SO ); ( ) lactaldehyde (0.01 M HCl);
) 1,2-propanediol (0.01 M HCl); ( ) lactaldehyde (no electrolyte).
2
4
2
4
4
4
2
4
2
4
(
tion’s resistance was high, however, limiting the practical ECH
current to half the usual value (i.e., 50 mA). Nonetheless, at this
current, the quantity of H atoms produced should be sufficient
to completely hydrogenate all of the LA present in solution.
As expected, no PG formation was observed in this case; pre-
sumably, the low solution concentration of protons could not
provide the high surface coverage of H needed for PG forma-
tion. Importantly, there was no induction period, a finding that
supports the Cl adsorption explanation for the initial low ECH
in HCl electrolyte. However, the rate of ECH at later stages in
the reaction was not as high as that seen with HCl; presumably,
once Cl has desorbed, the relative ECH rates reflect the differ-
ences in current (50 vs 100 mA) between experiments with LA
alone and in the HCl electrolyte. The anomalous behavior ob-
served with HCl may also reflect an electronic enhancement or
surface restructuring effect of Cl on the Ru electrode surface,
but the present results cannot make that determination.
In general, in both 0.1 and 0.01 M electrolyte concentrations,
the rate of formation of LAL was found to decrease in the order
HCl > HClO4 > H2SO4. Formation of PG was generally slow,
maximal in 0.1 M HCl and almost negligible in 0.1 M HClO4
−
due to an inhibitory effect of the ClO4 anion. In the case of
ECH of LA in 0.01 and 0.1 mol HCl, the rate of conversion of
LA to LAL and PG was low initially and increased over time.
Apparently, as the electrohydrogenation proceeds, Cl2 evolu-
−
tion at the anode decreases the Cl concentration in solution,
−
leading to the desorption of Cl from the cathode surface and
leaving more sites for the adsorption of LA, hydrogen, or other
active species. It is also possible that partial surface coverage
by adsorbed Cl is beneficial for the ECH reaction, because the
electronegative Cl atoms modify the electronic state of the Ru
atoms and enhance adsorption of reactants. To test this hypothe-
sis, ECH of LA was performed in the absence of an electrolyte.
With a pKa of 3.86, LA at 11.1 mmol is expected to be largely
ionized and thus able to function as the electrolyte. The solu-
The adsorption of anions on electrode surfaces and their
effects on hydrogen adsorption on metal surfaces have been
extensively studied. The relative strengths of adsorption of the

22
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
−
anions were found by Horanyi to decrease in the order Cl >
−
−
HSO > ClO on Pt electrode [25]. In studies of chemical cat-
4
4
−
alytic hydrogenation of LA, Cl was found to significantly in-
hibit hydrogenation, whereas the inhibitory effect was observed
only in the early stages in ECH. This discrepancy reflects the
fundamental difference in the initial step of the two processes.
In catalytic hydrogenation, the first step in the process is the dis-
sociation of the H2 molecule into atoms, which is equivalent to
the electrocatalytic oxidation of H2. In the case of ECH, the ini-
tial step involves the cathodic reduction of protons in solution.
Adsorption of anions was found to inhibit the hydrogen oxida-
tion process on Pt, but not the electroreduction step [24]. There-
fore, the effect of anions depends on the source of H atoms (i.e.,
molecular H2 or protons). This difference was attributed to the
involvement of overpotential deposited (OPD) H atoms in ECH
and not underpotential deposited (UPD) H atoms, the adsorp-
tion of which is influenced by anion adsorption. OPD hydrogen
atoms are weakly bound on the cathode surface and have a cer-
tain degree of mobility, thus enabling them to reach the active
sites. However, in the case of ECH of LA on Ru, the adsorp-
tion of anions seem to be strong and accompanied by some
amount of charge transfer because they are not easily desorbed
even in the hydrogen evolution region. The increase in ECH
activity at more negative potentials also may be attributed to
the increased mobility of OPD H atoms and a change in the
electronic properties of the metal surface due to cathodic polar-
(a)
−
ization [24]. Several studies found that the adsorption of Cl
on the electrode surface shifted the metal oxide formation peak
in the cyclic voltammogram to more positive potentials and
the hydrogen adsorption peak to more negative potentials [26].
(b)
Fig. 5. Cyclic voltammogram of Ru polycrystalline electrode in various elec-
−
Therefore, Cl adsorption was found to inhibit the formation
2
trolytes (0.01 M) (A = 0.2 cm , υ = 10 mV/s): (a) without lactic acid and (b) in
of metal oxides such as RuOH, resulting in increased hydrogen
the presence of lactic acid (0.01 M). (—) HClO , (- - -) H SO and (-·-·-) HCl.
4
2
4
−
adsorption. Some Cl adsorption was assumed to occur even
at more negative potentials of hydrogen evolution. Therefore,
the enhancing effect of HCl toward ECH of LA observed in our
studies may be attributed to the inhibition of Ru oxide forma-
Comparison of rates of product formation between the various
electrolytes reveals that SO4 has a significant detrimental ef-
fect toward LAL formation, whereas perchlorate significantly
2−
−
tion. The adsorption of Cl does inhibit LA conversion initially,
inhibits PG formation at only high concentration. These results
but the effect is overcome at longer times by the desorption of
−
indicate a general adsorption of Cl on all types of active sites,
−
−
Cl resulting from anodic oxidation of Cl and reduced solu-
tion concentration.
whereas perchlorate and sulfate show preferential adsorption
on specific active sites. On the basis of in situ FTIR studies,
Marinkovic et al. [31] reported that due to a near-ideal geomet-
rical fit to the Ru lattice’s dimensions and hexagonal symmetry,
bisulfate ions adsorb strongly on Ru(0001) surface at potentials
about 300 mV more negative than for adsorption on Pt, whereas
their adsorption on polycrystalline Ru is inhibited by surface
oxide formation. The potential at which anion adsorption on a
metal surface occurs is determined by its work function [32].
−
−
FTIR spectroscopy has shown that HSO and ClO adsorb
4
4
on metals through three of the O atoms in a C3v configura-
−
tion [27–29]. Even though ClO is considered weakly adsorbed
4
on electrode surfaces in many cases, Colom and Gonzalez-
−
−
Tejera [30] observed electroreduction of ClO4 to Cl on Ru
electrodes. The extent of reduction was found to depend on
−
the concentration of ClO , hydrogen adsorption, and temper-
4
ature. Perchlorate reduction process requires the presence of
adsorbed hydrogen and is enhanced at higher temperatures.
Therefore, despite the similar modes of adsorption of the an-
ions, the enhancement of LAL formation with perchlorate may
3
.5. Voltammetric studies
Figs. 5a and 5b show cyclic voltammograms of polycrys-
−
2
be attributed to some extent to its decomposition to Cl on
talline Ru electrode (0.2 cm ) in the three electrolytes at 0.01 M
concentration in the absence and presence of LA (11.1 mM)
recorded at a scan rate of 10 mV/s. In all cases, the cathodic
hydrogen adsorption and anodic hydrogen oxidation/desorption
peaks can be observed. In Fig. 5a, in HClO and HCl, the
H adsorption peaks overlap with the hydrogen evolution cur-
the surface, followed by the desorption and anodic oxidation of
−
Cl . Cl2 gas evolution at the anode was observed when HClO4
was used as the electrolyte. However, it is interesting to note
that there is significant concentration-dependent inhibition of
LAL and PG formation with 0.01 and 0.1 M H2SO4 and HClO4.
4

T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
23
rent. In addition, oxygen evolution occurs at much lower po-
tentials in the presence of LA. However, in case of H2SO4,
there is significant hysteresis in both the absence and presence
of LA, which can be attributed to the difference in the proper-
ties of the surface caused by anion adsorption on Ru [33]. The
peak currents are also much lower in H2SO4. Unlike HClO4
and HCl, there is a delay in the onset of hydrogen desorp-
tion in H2SO4, indicating the formation of a stable adlayer
of H which is resistant to oxidation. All of these results are
consistent with adsorption of sulfate on Ru at the H adsorp-
tion/desorption potential. There is no cathodic peak after the
hydrogen desorption peak during the positive sweep in 0.01 M
HClO4, indicating that perchlorate reduction to chloride, which
is known to be a slow and irreversible process at room temper-
ature [27], does not occur at this scan rate and at this concen-
tration.
In the presence of LA (Fig. 5b), the hydrogen adsorption
peak currents decrease in the case of H2SO4 and HClO4 due
to partial blocking of surface sites by LA adsorption, but re-
main constant in the case of HCl. This indicates that in HCl,
in spite of LA adsorption, the surface coverage of adsorbed hy-
drogen remains constant due to an enhancement of hydrogen
adsorption by adsorbed Cl. Using radiotracer analysis, Horanyi
and Rizmayer [34] reported strong adsorption of Cl on Ru in
the hydrogen adsorption/evolution region at negative potentials
even with high surface coverage of H and stated the possibility
(a)
−
of Cl ions influencing the adsorption of H and O on Ru. They
−
also showed much stronger adsorption of Cl ions compared
−
2−
with ClO4 and SO4 . The cyclic voltammograms yield some
information about the extent of anion adsorption in the hydro-
gen adsorption region, but not in the more negative hydrogen
evolution region, where ECH actually occurs. ATR-SEIRAS
studies of the ECH of LA, described in the subsequent section
gives more detailed information about the effect of anion ad-
sorption on ECH.
3
.6. In situ ATR-FTIR studies of ECH of LA
◦
The IR spectra of LA recorded at various times at 30 C and
400 mV in the three electrolytes H2SO4, HCl, and HClO4,
−
(b)
using the spectra recorded after 1200 min as the reference, are
shown in Figs. 6a, 7a and 8a. The O–H stretching and H–O–H
bending regions are shown separately in Figs. 6b, 7b and 8b.
These spectra were referenced against the spectra of the re-
spective electrolytes alone taken under identical conditions, to
clearly see the changes occurring during ECH of LA without
significant interference from solution LA. The spectra below
Fig. 6. Real time ATR-IR spectra of the ECH of lactic acid in 0.01 M H SO
2
4
◦
(
T = 30 C, E = −400 mV) (a) using the spectrum taken after 1200 min as
the reference, and (b) using the spectrum of 0.01 M H SO taken under identi-
2
4
cal conditions as the reference, showing the O–H stretch and H–O–H bending
bands.
in bonding to the metal catalyst. The weak or negative bands ob-
−
1
−
1
served around 1710 cm may be attributed to the C=O stretch
of solution phase LA. The monodentate and the dissociative ad-
sorption modes may be ruled out as these modes would result
in much stronger C=O stretching bands.
1
200 cm are not shown in the figures because the cutoff fre-
−
1
quency for the Si window is 1200 cm . The LA concentration
used in all cases was 0.01 M, which was the concentration cal-
culated to give minimum interference from solution phase LA.
The peak positions and assignments of the initial spectrum in
each case, taken immediately after addition of LA to the elec-
trolyte are shown in Table 3. The spectra clearly show strong
bands due to the COO symmetric stretch and the alcoholic C–O
stretch, indicating that LA is adsorbed on Ru in the chelating
bidentate mode. Therefore, the α-hydroxy group is not involved
The spectra show strong symmetric stretching (1435 cm 1)
and weak asymmetric stretching bands. These observations can
be explained by the surface selection rule [35] and strongly sup-
port the chelating bidentate mode of adsorption and the fact
that spectra show predominantly adsorbed LA species, because
SEIRAS is sensitive to surface species and less sensitive to
−

24
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
(a)
(a)
(b)
(b)
Fig. 7. Real time ATR-IR spectra of the ECH of lactic acid in 0.01 M HCl
Fig. 8. Real time ATR-IR spectra of the ECH of lactic acid in 0.01 M HClO
4
◦
◦
(
T = 30 C, E = −400 mV) (a) using the spectrum taken after 1200 min as
(T = 30 C, E = −400 mV) (a) using the spectrum taken after 1200 min as
the reference, and (b) using the spectrum of 0.01 M HCl taken under identi-
cal conditions as the reference, showing the O–H stretch and H–O–H bending
bands.
the reference, and (b) using the spectrum of 0.01 M HClO taken under identi-
4
cal conditions as the reference, showing the O–H stretch and H–O–H bending
bands.
In 0.01 M H2SO4 (Fig. 6a), as the reaction progresses, sev-
eral changes are observed in the spectra. The peak at 3168 cm
solution species. The transition dipole of the symmetric COO
vibration is parallel to the diagonal of the carboxylate group and
when LA is coordinated to the metal surface through the car-
boxylate group has a strong component in the direction normal
to the surface, which explains the high intensity of this band.
In contrast, the direction of transition dipole of the COO asym-
metric stretch is along the line joining the two oxygen atoms,
which in the adsorbed molecule is parallel to the surface and
hence this vibration is IR inactive. The O–H deformation mode
−1
due to the O–H stretching vibration of strongly hydrogen
−1
bonded water reappears and the peaks at 1435 and 1283 cm
from adsorbed LA decrease in intensity. This indicates that as
the reaction progresses, the Ru surface is poisoned and this re-
sults in a decrease in the adsorption of LA. The poison may be
associated with reaction intermediates, sulfate or S-containing
species derived from the sulfate ions of the electrolyte, because
the decrease in LA adsorption is associated with the appearance
of strongly hydrogen bonded water near the electrode surface,
which are known to be associated with the hydrophilic sulfate
anions, from previous studies of interfacial water molecules on
−
1
also gives a weak band at 1357 cm due to this bond being
parallel to the surface when LA is adsorbed in the chelating
bidentate orientation.

T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
25
Table 3
with O–H stretching and H–O–H bending vibrations of water.
The appearance of these bands corresponds approximately to
the increase in the rate of ECH of LA after a slower initial
phase, observed while using HCl as the electrolyte with 5%
Ru/C agglomerated in RVC. This suggests that the partial de-
sorption of chloride, as a result of decrease in the solution con-
centration of chloride as Cl2 gas is evolved at the anode, frees
up sites, allowing the adsorption of water molecules, which are
required for ECH. From the frequencies of the water bands,
they appear to be relatively isolated or poorly hydrogen bonded,
which may be a result of association with the surface and the
hydrophobic nature of the co-adsorbed chloride, resulting in
disruption of hydrogen bonding [40–42]. These observations
are consistent with the conclusion that adsorbed water plays
an important role in LA hydrogenation, derived from the ex-
periments indicating a decrease in the rate of hydrogenation
of LA in organic solvent/water mixtures [8]. From ab initio
studies of water layer on Pt(111) and Rh(111) surfaces, Vas-
silev et al. [43] discovered that water molecules adsorbed on
a metal surface have a slight metal-like conductivity result-
ing from mixing of local states of the oxygen atom with the
electron bands of the metal substrate. The extent of mixing
of the atomic states from oxygen and metal was found to de-
pend on the orientation of the water molecules, the distance
from the surface, and the polarization of the surface. Analy-
sis of the electronic states indicated an electron charge transfer
from the water molecules toward the surface, resulting in a
reduction of the work function of the system. This phenom-
enon may be responsible for facilitating the electron transfer
IR peak assignments for the initial spectrum taken at 8 min during ECH of lactic
acid
−
1
Wave number (cm
)
Vibrational mode
0
1
1
1
1
3
.01 M H SO
2
4
283
357
435
630
610
υC–O (α-hydroxy)
O–H deformation
υCOO symmetric
δH–O–H
υO–H (water and lactic acid)
0
1
1
1
1
3
.01 M HCl
283
357
435
595
120
υC–O (α-hydroxy)
O–H deformation
υCOO symmetric
δH–O–H
υO–H (water and lactic acid)
0
1
1
1
1
3
.01 M HClO
4
275
340
435
625
400
υC–O (α-hydroxy)
O–H deformation
υCOO symmetric
δH–O–H
υO–H (water and lactic acid)
Au electrode by Osawa and co-workers [36,37]. This poisoning
effect explains the lower rate of ECH of LA in H2SO4 com-
pared with HCl and HClO4.
In 0.01 mol H2SO4 (Fig. 6b), the negative band at 3168
cm 1 indicates the displacement of strongly hydrogen-bonded
−
water molecules from the surface by LA. However, the bands at
−1
1
630 and 3610 cm are due to the H–O–H bending and O–H
stretching vibrations of non-hydrogen-bonded water molecules.
Apparently, adsorption of LA on the electrode surface isolates
bound water molecules, and such non-hydrogen-bonded water
+
to protons or H3O to form H atoms, which are required for
ECH. In the case of H2SO4 and HCl electrolytes, a weak fea-
+
ture due to H3O was observed in the IR spectra. However,
−
1
molecules show O–H stretching frequencies around 3600 cm
38]. The shift in the O–H stretch frequency from this value
the peak corresponding to the M–H bond, known to appear
[
−1
around 2100 cm [44,45] was not observed in any of the spec-
gives an estimate of the extent of hydrogen bonding. The H–O–
H bending frequency of water adsorbed on Ru(001) surface in
tra, indicating that either the surface concentration of adsorbed
H is too low to be detected by IR or it is a very short-lived
species, with lifetime much shorter than the time scale of the
scans.
−
1
UHV has been reported to be in the range of 1560–1520 cm
36,39]. A downward shift of the O–H stretch frequency and
[
an upward shift of the δH–O–H mode are indicative of strongly
hydrogen bonded water [36]. The O–H stretching vibrations of
LA cannot be distinguished from those of water, and the peak
The cyclic voltammetry results described earlier indicate
that adsorbed Cl on the electrode surface when HCl is used
as the electrolyte favors H adsorption. Therefore, the enhanced
ECH rate in HCl is probably a result of the electronic effects of
both adsorbed chloride and water. The specific sites occupied
by the water molecules and the exact configuration required
for involvement in the hydrogenation rate enhancement remains
unclear.
−
1
is believed to overlap with the 3610-cm peak from water.
The close-packed monolayer of LA adsorbed on Ru through
the carboxylate groups are also expected to have –O–H groups,
which are isolated from one another and from water molecules
in the vicinity, forming very few hydrogen bonds, thus giving
−
1
the high-frequency band around 3610 cm
.
In 0.01 mol HClO (Fig. 8), there is a slight decrease in the
4
In 0.01 M HCl (Fig. 7b), the O–H stretching bands from
water and LA and H–O–H bending bands from water appear
intensity of the adsorbed LA bands with time at 1000 min, indi-
cating poisoning of the surface by perchlorate or chloride, but
the decrease in intensity is not as pronounced as in the case
of H SO and occurs at much longer time compared with the
−
1
at 3120 and 1595 cm , respectively. Therefore, the LA may
be associated with strongly hydrogen-bonded water. However,
in this case, it is clear from Fig. 7a that there is no significant
decrease in intensity with time of bands due to adsorbed LA.
Apparently, the Ru surface is not poisoned by electrolyte anions
in this case.
2
4
H SO case. The negative C=O stretching band in all the three
2
4
electrolytes may be explained by the fact that some LA is still
present in solution at the end of ECH (i.e., at 1200 min), and in
the absence of significant amounts of adsorbed LA (due to elec-
trode surface poisoning), show stronger IR bands in comparison
to the spectra taken during the earlier stages of ECH.
It is also interesting to note the appearance of new bands at
−1
3
400 and 1647 cm at around 200 min in Fig. 7b, associated

26
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
Why does electrohydrogenation of LA stop at LAL, rather
sure. Acetaldehyde formation by the vapor-phase hydrogena-
tion of acetic acid over silica-supported copper catalyst was
also described by Cortright and Dumesic in an earlier report
[49]. In these cases, aldehyde formation was attributed to the
hydrogenation of intermediate surface acyl species. However,
in the present work, dissociative adsorption of LA and conse-
quent surface acyl formation can be ruled out from the FTIR
results. Falorni et al. [50] reported the Pd/C catalyzed hydro-
genation of substituted amino acids to aldehydes at atmospheric
pressure in ethanol. The conversion involved the preparation of
an ester with 2-chloro-4,6-dimethoxyl[1,3,5]triazine, followed
by hydrogenation. Reduction of the aldehyde to alcohol was
reported only at higher pressures and longer reaction times.
However, as with our mandelide studies, alcohol was formed as
the major product when aromatic carboxylic acids were used.
These studies reflect a dependence of the relative rates of prod-
uct formation on the surface hydrogen concentration, reaction
conditions, and mode of adsorption of the substrate.
than progressing on to 1,2-propanediol as in the chemical re-
duction? The most likely reason is that the LA reduction leads
directly to the hydrated form of LAL, as shown in Scheme 2.
Like other α-hydroxy aldehydes, LAL strongly favors the hy-
drated form at equilibrium. Under the relatively low tempera-
ture and acidic conditions of ECH, the dehydration required to
enable further hydrogenation to the diol would be slower than
its desorption, thus leading to the formation of small quantities
of PG. The formation of PG is also favored by higher surface
concentration of H. This notion is also supported by the fact that
when mandelic acid, an aromatic analog of LA, was subjected
to ECH under similar conditions, the completely hydrogenated
product cyclohexyl ethylene glycol was formed to an approxi-
mate 50% extent, much higher than the LA-to-PG conversion.
The intermediate cyclohexyl glycolic aldehyde also present in
the product mixture is expected to be less hydrated than LAL
due to the greater electron-donating effect of the cyclohexyl
substituent.
A similar selectivity to the corresponding aldehyde was ob-
served by Yokoyama and Yamagata [46] in the vapor-phase
hydrogenation of benzoic acid using Cr/ZrO2. In this case, the
carboxylic acid interacts strongly with the catalyst surface, dis-
placing the weakly bound aldehyde as soon as it is formed,
preventing successive hydrogenation to the alcohol and result-
ing in high aldehyde selectivity (96 at 98% conversion in the
Yokoyama and Yamagata [46] case). Studies of LA, PG, and
glycerol binding on Ru surfaces have shown stronger LA bind-
ing by at least 2 orders of magnitude. In terms of chemical
functionality, the hydrated LAL (1,1,2-propanetriol) is much
more like PG and glycerol than LA, and its expected affin-
ity for the surface should be similarly weak [47]. Thus, like
Yokoyama and Yamagata [46], we interpret the LAL prefer-
ence as evidence for the desorption rate exceeding the dehy-
dration/hydrogenation rate at these low temperatures. The re-
sults of the ab initio calculations of heats of formation shown
in Scheme 2 indicate that the dehydration of hydrated LAL in
the liquid phase is endothermic by 6.8 kcal/mol. However, the
other two steps—hydrogenation of LA to hydrated LAL and the
hydrogenation of LAL to PG—are exothermic, for an overall
LA-to-PG exothermicity of 18.5 kcal/mol. Although gas-phase
heats of formation calculated this way are typically within 1–
4. Conclusion
Electrocatalytic hydrogenation of LA has been accom-
plished using 5% Ru/C agglomerated in RVC under mild condi-
tions of temperature and pressure compared with chemical cat-
alytic hydrogenation. The major product of electrocatalytic hy-
drogenation was LAL, with small quantities of 1,2-propanediol.
This unexpected process can provide a clean, convenient, and
efficient method for preparing LAL, a compound that is not
commercially available and is difficult to obtain in pure form.
To the best of our knowledge, this is the first ECH study con-
ducted on a carboxylic acid. The low current efficiency is due
to the slow kinetics of the ECH reaction. In the 5% Ru/C/RVC
system, the amount of active electrocatalyst at any given time
is very low, because the technique relies on efficient agglom-
eration of the catalyst, which is affected by various factors,
including particle size, fluid dynamics, and others. The kinet-
ics can be improved by developing better electrode materials
based on Ru and C, which will consist of well-dispersed Ru
nanoparticles on solid-phase carbon of similar porosity.
In preparative and mechanistic studies [1,51] of the Ru/C-
catalyzed hydrogenation of LA to 1,2-propanediol, LAL was
proposed as an intermediate but was never isolated. Using elec-
trohydrogenation, LAL is the primary LA conversion prod-
uct, supporting its proposed intermediacy en route to 1,2-
propanediol. Although this project was begun as a way to carry
out LA-to-PG reduction under mild conditions, the discovery of
a pathway that directly reduces aqueous carboxylic acid to alde-
hyde and then stops is substantially more interesting in terms of
both fundamental chemistry and practical processes. The fact
that the product ratio of ECH depends on the electrolyte used
is also intriguing and can be exploited in future applications to
control the selectivity of the reaction toward specific products
by using specific electrolytes or a combination of electrolytes.
The ECH of LA on Ru electrode surface was studied in real
time using ATR-SEIRAS spectroscopy. The conditions used for
the SEIRAS studies were close to those used for bulk ECH
studies, so that the spectra can be correlated with changes oc-
2
kcal/mol of experimental values, the simulation of aqueous
solution via continuum methods ignores specific hydrogen-
bonding contributions; thus, these “solution” energy changes
are based on the approximation that errors in representing such
interactions are of similar magnitude for all species.
It is also interesting to note that although no LAL formation
was observed in the Ru/C-catalyzed aqueous-phase chemical
hydrogenation of LA, Dumesic and co-workers [48] did ob-
serve LAL formation in the vapor-phase hydrogenation of LA
using 10% Cu/SiO2 catalyst. The formation of the aldehyde was
found to be dependent on the equilibrium between the aldehyde
and diol at a given hydrogen partial pressure. The highest yield
of LAL (22.2%) was obtained at a hydrogen partial pressure of
0
.10 MPa at 473 K, and the yield was found to decrease with
decreasing temperature and increasing hydrogen partial pres-

T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
27
curring on or near the electrode surface during the course of
the experiment. The differences in rate of ECH observed in
the three electrolytes can be correlated with the extent of sur-
face poisoning by the electrolyte anions or species derived from
them. Because from the ATR-FTIR studies, LA was found to
adsorb on the Ru surface in the chelating bidentate mode with-
out involving the α-hydroxy groups, it can be concluded that
the α-hydroxy group does not facilitate the hydrogenation of
LA (and other α-hydroxy acids) by forming a more stable ad-
sorbed intermediate or by influencing the initial adsorption of
LA, but it may indirectly increase the hydrogenation rate rel-
ative to simple alkanoic acids by facilitating the adsorption of
water on the metal surface through the formation of hydrogen
bonds. This might be expected from the fact that in the chelat-
ing bidentate mode, the LA is adsorbed with the –O–H groups
exposed to the solution. It was also determined that chloride
does not inhibit the adsorption of LA or hydrogen, and that the
enhancement of the ECH rate in HCl may be due to partial des-
orption of chloride from the electrode surface at longer times
and subsequent adsorption of water molecules, which play an
important role in ECH. In future work, a more detailed pic-
ture of the effect of anion adsorption on LA conversion may
be obtainable using in situ Raman spectroscopic studies with
Ru single-crystal electrodes to determine the exact mode of ad-
sorption and the nature of surface sites occupied by each type
of anion.
[6] B. Mahdavi, P. Chambrion, J. Binette, E. Martel, J. Lessard, Can. J. Chem.
3 (1995) 846.
7] A.M. Polcaro, S. Palmas, S. Dermini, Ind. Eng. Chem. Res. 32 (1993)
315.
7
[
1
[8] J. Farrugia, unpublished work.
[9] A. Cyr, F. Chiltz, P. Jeanson, A. Martel, L. Brossard, J. Lessard, H.
Menard, Can. J. Chem. 78 (2000) 307.
[
[
[
10] P. Dabo, A. Cyr, J. Lessard, L. Brossard, H. Menard, Can. J. Chem. 77
1999) 1225.
11] L. Ruest, H. Menard, V. Moreau, F. Laplante, Can. J. Chem. 80 (2002)
662.
(
1
12] P. Dube, F. Kerdouss, P. Proulx, L. Brossard, H. Menard, J. Appl. Elec-
trochem. 33 (2003) 541.
13] H. Miyake, S. Ye, M. Osawa, Electrochem. Commun. 4 (2002) 973.
14] L.A. Curtiss, K. Raghavachari, P.C. Redfern, V. Rassolov, J.A. Pople,
J. Chem. Phys. 109 (1998) 7764.
15] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R.
Cheeseman, V.G. Zakrzewski, J.A. Montgomery Jr., R.E. Stratmann, J.C.
Burant, S. Dapprich, J.M. Millam, A.D. Daniels, K.N. Kudin, M.C. Strain,
O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C.
Pomelli, C. Adamo, S. Clifford, J. Ochterski, G.A. Petersson, P.Y. Ayala,
Q. Cui, K. Morokuma, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B.
Foresman, J. Cioslowski, J.V. Ortiz, A.G. Baboul, B.B. Stefanov, G. Liu,
A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R.L. Martin, D.J.
Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, C. Gonzalez,
M. Challacombe, P.M.W. Gill, B.G. Johnson, W. Chen, M.W. Wong, J.L.
Andres, M. Head-Gordon, E.S. Replogle, J.A. Pople, Gaussian 98 (Revi-
sion A.7), Gaussian, Inc., Pittsburgh, PA, 1998.
[
[
[
[
16] (a) S. Miertus, E. Scrocco, J. Tomasi, J. Chem. Phys. 55 (1981) 117;
S. Miertus, J. Tomasi, Chem. Phys. 65 (1982) 239;
(b) M. Cossi, V. Barone, R. Cammi, J. Tomasi, Chem. Phys. Lett. 255
(1996) 327;
(
c) V. Barone, M. Cossi, J. Tomasi, J. Comput. Chem. 19 (1998) 404.
Acknowledgments
[
17] K.P. Nambiar, D.M. Stauffer, P.A. Kolodziej, S.A. Benner, J. Am. Chem.
Soc. 105 (1983) 5886.
18] X.E. Hu, H.W. Yang, X.J. Wang, R.S. Bai, J. Appl. Electrochem. 32 (2002)
321.
[19] K. Amouzegar, O. Savadogo, Electrochim. Acta 39 (4) (1994) 557.
20] A. Roessler, O. Dossenbach, P. Rys, J. Electrochem. Soc. 150 (1) (2003)
The authors gratefully acknowledge the Biochemistry Mass
Spectrometry Laboratory at Michigan State University for help
in the characterization of the products of electrocatalysis by
ESI-MS. They are also grateful to Professor Hughes Menard
from the University of Sherbrooke, Sherbrooke, QC, who de-
veloped the electrocatalytic hydrogenation method using RVC
electrodes, for his generous help and advice. Financial sup-
port was provided by National Science Foundation grant CTS-
[
[
D1.
[
[
21] A. Bryan, J. Grimshaw, Electrochim. Acta 42 (13–14) (1997) 2101.
22] Y. Senda, M. Tateoka, H. Itoh, J. Ishiyama, Bull. Chem. Soc. Jpn. 64
(1991) 3302.
[23] H. Nielsen, P.E. Sorensen, Acta Chem. Scand. A 31 (1977) 739.
24] E. Lamy-Pitara, S. El Mouahid, J. Barbier, Electrochim. Acta 43 (2000)
[
0
328112.
4299.
[
[
25] G. Horanyi, Electrochim. Acta 36 (1991) 1453.
26] S. Hadzi-Jordanov, H. Angerstein-Koziowska, M. Yukovic, B.E. Conway,
J. Phys. Chem. 81 (24) (1977) 2271.
Supporting information
The online vision on this article contains additional supple-
mentary material.
[27] Y. Sawatari, J. Inukai, M. Ito, J. Electron Spectrosc. Relat. Phenom. 64/65
1993) 515.
28] N. Hoshi, M. Kuroda, T. Ogawa, O. Koga, Y. Hori, Langmuir 20 (2004)
066.
29] Y. Shingaya, M. Ito, Electrochim. Acta 37 (1992) 365.
(
[
[
Please visit doi: 10.1016/j.jcat.2006.11.009.
5
References
[30] F. Colom, M.J. Gonzalez-Tejera, J. Electroanal. Chem. 190 (1985) 243.
[
31] N.S. Marinkovic, J.X. Wang, H. Zajonz, R.R. Adzic, J. Electroanal.
Chem. 500 (1–2) (2001) 388.
[
1] (a) Z. Zhang, J.E. Jackson, D.J. Miller, Appl. Catal. A 219 (2001) 89;
b) Z. Zhang, J.E. Jackson, D.J. Miller, Ind. Eng. Chem. Res. 41 (2002)
91.
2] (a) F.T. Jere, D.J. Miller, J.E. Jackson, Org. Lett. 5 (2003) 527;
b) F.T. Jere, J.E. Jackson, D.J. Miller, Ind. Eng. Chem. Res. 43 (2004)
297.
(
6
[32] Y. Shingaya, M. Ito, J. Electroanal. Chem. 467 (1999) 299.
[33] A. Ahmadi, R.W. Evans, G. Attard, J. Electroanal. Chem. 350 (1993) 279.
[34] G. Horanyi, E.M. Rizmayer, J. Electroanal. Chem. 181 (1984) 199.
[35] M. Osawa, M. Ikeda, J. Phys. Chem. 95 (1991) 9914.
[36] K. Ataka, T. Yotsuyanigi, M. Osawa, J. Phys. Chem. 100 (1996) 10664.
[37] K. Ataka, M. Osawa, Langmuir 14 (1998) 951.
[38] E. Pretsch, J. Seibl, D. Clerc, W. Simon, Tables of Spectral Data for Struc-
ture Determination of Organic Compounds, second ed., Springer-Verlag,
1985, p. I85.
[
(
3
[
[
3] F. Laplante, L. Brossard, H. Menard, Can. J. Chem. 81 (3) (2003) 258.
4] (a) H. Ilikti, N. Rekik, M.J. Thomalla, J. Appl. Electrochem. 32 (2002)
6
(
1
03;
b) H. Ilikti, N. Rekik, M.J. Thomalla, J. Appl. Electrochem. 34 (2004)
27.
5] A.K. Cheong, Y. Bolduc, J. Lessard, Can. J. Chem. 71 (1993) 1850.
[39] (a) P.A. Thiel, E.M. Hoffman, W.H. Weinberg, J. Chem. Phys. 75 (1981)
5556;
[

28
T.S. Dalavoy et al. / Journal of Catalysis 246 (2007) 15–28
(b) P.A. Thiel, R.A. DePaola, W.H. Weinberg, J. Chem. Phys. 80 (1984)
[45] K. Kunimatsu, T. Senzaki, M. Tsushima, M. Osawa, Chem. Phys. Lett. 401
5326.
(2005) 451.
[
[
40] G. Brink, M. Falk, Can. J. Chem. 48 (1970) 2096.
41] (a) G.E. Walrafen, J. Chem. Phys. 36 (1962) 1035;
[46] T. Yokoyama, N. Yamagata, Appl. Catal. A Gen. 221 (2001) 227.
[47] L. Peereboom, unpublished results.
(
b) G.E. Walrafen, J. Chem. Phys. 55 (1971) 768.
42] F. Kitamura, N. Nanbu, T. Ohsaka, K. Tokuda, J. Electroanal. Chem. 452
1998) 241.
43] P. Vassilev, R.A. Van Santen, M.T.M. Koper, J. Chem. Phys. 122 (2005)
54701.
44] K. Kunimatsu, H. Uchida, M. Osawa, M. Watanabe, J. Electroanal. Chem.
87 (2006) 299.
[48] R.D. Cortright, M. Sanchez-Castillo, J.A. Dumesic, Appl. Catal. B 39
[
[
[
(2002) 353.
(
[49] R.D. Cortright, J.A. Dumesic, Adv. Catal. 46 (2001) 161.
[50] M. Falorni, G. Giacomelli, A. Porcheddu, M. Taddei, J. Org. Chem. 64
0
(1999) 8962.
5
[51] D.G. Kovacs, in preparation.