Platinum-catalyzed oxidations of organic compounds by ferric sulfate: Use of a redox fuel cell to mediate complete oxidation of ethylene glycol by dioxygen at 80°C

Article abstract of DOI:10.1006/jcat.1996.0009

A number of alcohols, aldehydes, and carboxylic acids were oxidized under mild conditions (60-80°C, 3 M H₂SO₄ in water) by Fe(III) using a catalytic amount of Pt black to generate Fe(II) and protons. The extent of oxidation depended on the structures of the organic reductant and on reaction conditions. A redox fuel cell that catalyzed the complete oxidation of ethyl-ene glycol by dioxygen at 80°C was assembled. Fe(II) generated by oxidation of ethylene glycol was reoxidized to Fe(III) at the anode of the cell. A V(V)/V(IV)/HNO₃/O₂ redox system was used to catalyze the electrochemical reduction of dioxygen. Both the anode and cathode consisted of WDF graphite felt, the surface of which had been oxidized by boiling concentrated nitric acid for 15 min. A maximum power density of 9.9 mW/ cm³ of graphite felt was obtained at a cell voltage of 197 mV.

Full text of DOI:10.1006/jcat.1996.0009

JOURNAL OF CATALYSIS 158, 92–96 (1996)
ARTICLE NO. 0009
Platinum-Catalyzed Oxidations of Organic Compounds by Ferric
Sulfate: Use of a Redox Fuel Cell to Mediate Complete Oxidation of
Ethylene Glycol by Dioxygen at 80ЊC
Christopher B. Gorman, Steven H. Bergens, and George M. Whitesides
Department of Chemistry, Harvard University, Cambridge, Massachusetts 02138
Received January 17, 1995; revised July 28, 1995
design a catalytically active electrode that has a substantial
A number of alcohols, aldehydes, and carboxylic acids were volume. However, with this scheme, it becomes difficult
2 4
oxidized under mild conditions (60–80؇C, 3 M H SO in water)
to compare performance with conventional fuel cells that
employ planar electrodes (7, 8).
We previously reported such a redox fuel cell that oper-
ated with methane as fuel (9). In this paper we demonstrate
both catalytic oxidation of a number of substrates with
the Fe(III)/Pt system and complete oxidation of ethylene
by Fe(III) using a catalytic amount of Pt black to generate
Fe(II) and protons. The extent of oxidation depended on the
structures of the organic reductant and on reaction conditions.
A redox fuel cell that catalyzed the complete oxidation of ethyl-
ene glycol by dioxygen at 80؇C was assembled. Fe(II) generated
by oxidation of ethylene glycol was reoxidized to Fe(III) at the
3 2
anode of the cell. A V(V)/V(IV)/HNO /O
redox system was glycol to carbon dioxide under mild conditions using dioxy-
used to catalyze the electrochemical reduction of dioxygen. gen as stoichiometric oxidant in a redox fuel cell operating
Both the anode and cathode consisted of WDF graphite felt,
the surface of which had been oxidized by boiling concentrated
nitric acid for 15 min. A maximum power density of 9.9 mW/
at 80ЊC. This is the first report of a redox fuel cell operating
with ethylene glycol as fuel (5, 10–12).
3
cm of graphite felt was obtained at a cell voltage of 197
METHODS
mV.
 1996 Academic Press, Inc.
Materials. Methanol, ethanol, and ethylene glycol were
distilled before use. Acetic acid (glacial) was stirred with
INTRODUCTION
acetic anhydride and CrO (2 g/100 ml) to remove water
3
and trace aldehyde and distilled before use. Glycolic acid
The catalytic oxidation of organic compounds using di-
oxygen as the ultimate oxidant is widely used in synthesis
and is a prevailing strategy for removal of contaminants
from waste streams and flue gases (1–3). We have found
an inorganic redox system (Fe(III)/Fe(II)) that oxidized
various alcohols, aldehydes, and carboxylic acids in the
presence of a catalytic amount of platinum black. The
system was made catalytic in Fe(III), and dioxygen was
used as stoichiometric oxidant, by electrochemically reoxi-
dizing Fe(II) to Fe(III) at the anode of a redox fuel cell.
Figure 1 shows a schematic of the fuel cell. Fe(II) was
electrochemically oxidized to Fe(III) at the anode, and O2
(
Aldrich), glyoxyilic acid (50% solution, Aldrich), po-
tassium oxalate (Mallinckrodt), glyoxal (40% solution,
Aldrich), glycerol (Fisher), sorbitol (Aldrich), and glucose
(
Aldrich) were used as received. Platinum black (Aesar,
fuel cell grade) was heated to 100ЊC under vacuum for 30
min before use. The activity of the platinum catalyst was
Ϫ
4
determined to be 3.3 ϫ 10 mol/g Pt (6.44% dispersion)
by alternating titration with hydrogen and oxygen as de-
scribed previously by other researchers (13). Ferric sulfate
was purchased from Aesar (puratronic grade) and used
directly or from Aldrich (as the pentahydrate) and recrys-
tallized from 6 M H SO before use. Solutions approxi-
2
4
was reduced to H O at the cathode using a V(IV)/V(V)/
2
mately 0.5 M in ferric sulfate and 0.8 M in sulfuric acid
HNO redox system developed by Kummer and Oei (4–6).
3
3
ϩ
(
Aesar, puratronic grade) were prepared and [Fe ] was
A Nafion membrane separated the anode and the cathode
compartments of the cell and allowed transfer of protons
from the anolyte to the catholyte. Both the anode and the
cathode consisted of WDF graphite felt (Union Carbide)
that was surface oxidized with boiling, concentrated nitric
determined by titrating aliquots of the solution with EDTA
at 40ЊC using N-(4-methoxyphenyl)-1,4-phenylenediamine
hydrochloride (Aldrich) as the indicator.
Open-circuit oxidation of methanol by Fe(III) on a plati-
acid for 15 min. This scheme may have advantages in the nized platinum gauze electrode. A two-compartment elec-
design of fuel cells. Particularly, it is not clear how to trochemical cell similar to one described previously (14)
9
2
0
021-9517/96 $12.00
Copyright  1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

PLATINUM-CATALYZED OXIDATIONS OF ORGANIC COMPOUNDS
93
allowing the reaction system to come to steady-state tem-
perature (80ЊC) and pressure (1 atm), the stopcock to a
gas buret filled with mineral oil was opened. The organic
compound to be oxidized was added as a neat liquid or as
an aqueous solution via syringe. The amount added was
selected so that the ratio of Fe(III) to reducing equivalents
of fuel was approximately 2 : 1. Reaction generally oc-
curred after a short (1–2 min) initiation period and was
evidenced by gas production (an increase in the volume
of gas in the buret). During the reaction, the volume of
the system was allowed to expand by maintaining the pres-
sure of the system at atmospheric by means of a sidearm
reservoir on the gas buret. Reaction was carried out until
gas production had slowed to virtually zero or after approx-
imately 3–4 h. Afterward, the reaction solution was filtered
under argon atmosphere to remove the platinum black and
FIG. 1. Schematic of the ethylene glycol/oxygen fuel cell. Details of
the cell components and operation are provided in the Methods section.
4
ϩ
titrated with a standard Ce solution to determine the
amount of iron reduced. In a control experiment, it was
verified that small amounts of residual organics in the acidic
solution were not directly oxidized by the ceric ion at a
rate that would interfere with the redox titration of the
ferrous ion. Oxidation of methanol was carried out in a
similar procedure except that 2 mol% of Pt based on the
reducing equivalents of fuel was used, and the reaction
was carried out at a temperature of 60ЊC.
was used to study the oxidation. In this case, however, the
cell compartments were separated by a Nafion membrane,
a saturated calomel electrode was used as the reference
electrode, and stirring was accomplished with a Teflon-
coated magnetic stirbar. The catalyst/electrode (roughness
2
factor of 684, geometric area of 6.25 cm ) was prepared as
previously described (15). The platinized catalyst/elec-
trode was first carefully rinsed with a solution of 30% H O
2
2
Ethylene glycol fuel cell. The assembly and operation
of the cell was similar to that described for a methane
redox fuel cell (9). The oxidation of ethylene glycol at
dispersed platinum black by Fe(III) and the reduction of
dioxygen by V(IV) were carried out in separate reformer
flasks. The anolyte and catholyte were circulated between
reformer flasks and cell compartments using peristaltic
pumps. The solutions in the reformer flasks were filtered
through glass frits before being admitted to the fuel cell.
in concentrated H SO (ȁ10% H O solution in H SO by
2
4
2
2
2
4
volume) and then held at alternating potentials of ϩ1.4
and Ϫ0.1 V in 1 M solutions of H SO in water at 25ЊC
2
4
until constant cyclic voltammograms (swept from 0 to 1.2
V at 50 mV/s) were obtained. It was finally held at ϩ1.4
V for 20 s before being rinsed with water and transferred
to the reaction vessel. The system was set up under an
atmosphere of argon with equal volumes of Fe(III) solution
(
0.6 M, Fe(III)/Ptsurface ϭ 978 : 1) and of 3 M H SO solution
2 4
in the catalyst/electrode and reference electrode compart- The reduction of Fe(III) to Fe(II) due to oxidation of
ments, respectively. Control experiments showed that only ethylene glycol was carried out for 2 h before the solution
slight variation in measured potentials of the catalyst/elec- was admitted to the anode compartment and the cell was
trode occurred when the reference electrode was moved operated. Currents at various cell potentials were mea-
between cell compartments. The system was allowed to sured under steady-state conditions by varying the load
equilibrate in a 60ЊC bath before addition of methanol. across the cell. The cell compartments were 3 cm in diame-
The potential of the working catalyst/electrode varied little ter and 5.5 cm deep and were separated by a Nafion mem-
with stir rate unless stirring was stopped, which would brane. A saturated calomel electrode was placed in each
typically result in an increase of potential by ȁ50 mV.
compartment as a reference electrode. Both the solutions
in the reformer flasks and the solutions in the fuel cell
chambers were stirred with Teflon-coated magnetic
stirbars.
Oxidation reactions. Except for the oxidation of meth-
anol (see below), all reactions were performed as follows.
All reactions were performed under argon using argon-
purged solutions. After heating the platinum black [3 mol%
The reformer flask for the anode consisted of a flask
based on the number of reducing equivalents of fuel (x in containing 75 ml of a 0.6 M Fe (SO ) solution in 30%
2
4 3
Eq. [1] and Table 1)], the flask was cooled to the reac- H SO solution, 1 ml of ethylene glycol [a four-fold excess
2
4
tion temperature, and ferric sulfate solution (15 ml, relative to starting Fe(III)], and 79.8 mg of Pt black [0.9
Ϫ
4
[
Fe(III)] ϭ 0.586 M) was added. The stirring rate was mol% relative to starting Fe(III); 3.3 ϫ 10 mol active
established at 2400 rpm (1-in. Teflon-coated stirbar in a sites/g Pt as determined by H /O titrimetry (13) (6.4%
2
2
100-ml, three-necked flask) using a stroboscope. After dispersion)]. The oxygen-reforming reaction is discussed

94
GORMAN, BERGENS, AND WHITESIDES
in detail elsewhere by Kummer and Oei (4–6). Cyclic volt-
ammograms of the catholyte and anolyte were identical to
those of authentic solutions of V(V) and Fe(III), respec-
tively.
Both the anode and the cathode of the fuel cell consisted
of rectangles of WDF graphite felt (Union Carbide, 2 ϫ
TABLE 1
Reaction Conditions and Resulting Stoichiometries for the
a
Oxidation of Various Functionalized Fuels
x (mol Fe(II)/mol
2
z (mol CO /mol
fuel oxidized)b
fuel oxidized)b
3
ϫ 0.5 cm). The felt was surface oxidized with boiling,
R
Expected Foundc
Expected Foundd
concentrated nitric acid for 15 min. The density of the felt
before this treatment was 8.49 ϫ 10 g/cm . The surface CH OH
area and density of the felt after treatment with boiling HOCH
Ϫ
2
3
3
6
10
6
6
4
6.4
9.6
6.4
6.6
4.6
2.1
1
2
2
2
2
2
0.94
1.7
1.6
1.8
1.8
2.0
2
CH
2
OH
HOCH
2
COOH
nitric acid were not measured. The current collectors were
OCHCHO
OCHCOOH
KOOCCOOK
Pt wires (0.5 mm in diameter and ȁ4 cm long). Since
this redox fuel cell scheme permits the use of porous felt
electrodes (i.e., the electrode reaction can take place in
the volume of a porous electrode) and since the cost of CH
the graphite felt is relatively unimportant, but the volume
2
3
CH
2
OH
COOH
12
8
5.8
Ȃ0.02
2
2
0
0
CH
3
of the felt is important, current densities are reported per
unit volume of carbon felt used (15).
C
C
C
3
6
6
H
H
H
8
O
3
(glycerol)
(glucose)
(sorbital)
14
24
26
15.0
3.8
7.2
3
6
6
1.8
0
0
12
O
6
14
O
6
RESULTS
a
b
c
2 4
Conditions 80ЊC, 3 M H SO .
From Eq. [1].
Oxidation of methanol. Methanol is rapidly and quan-
2
ϩ
Determined by potentiometric titration of Fe produced with a stan-
titatively oxidized to CO by excess Fe(III) in the presence
4
ϩ
2
ϩ
2
dard Ce solution and expressed as the ratio of [Fe ] at the end of the
of a catalytic amount of platinum black (1 mol% Pt versus reaction to [R] at the beginning of the reaction. Error in this ratio is Ϯ0.1.
d
Fe(III); [Fe(III)] ϭ 0.7 M; 60ЊC; 30% w : w H SO in water;
Determined by measuring the volume of gas produced at constant
2
4
temperature and pressure using a gas buret. Error in zfound is Ϯ0.1.
under argon). The rate of CO evolution was effectively
2
constant for the majority of reaction. The turnover fre-
quency during this period was 0.39 turnover/atom Ptsurface
/
s (one turnover ϭ production of 1 CO or of 6 Fe(III),
Ptsurface is an active site on Pt black). Substantial evolution
2
Fe(III) Х 16 : 1) caused a rapid increase of catalyst/elec-
trode potential to 0.61 V and evolution of CO to resume.
2
of CO was observed even when methanol was in excess
2
We interpret this behavior as formation of the active cata-
lyst during a short induction period by reduction of surface
oxides by methanol (14), followed by catalytic oxidation
by Fe(III). The potential of the catalyst/electrode during
catalytic oxidation of methanol remained in the region
of Fe(III) (e.g., MeOH : Fe(III) ȁ 2 : 1; 12 reducing equiva-
lents per Fe(III)), demonstrating that products of partial
oxidation were more reactive than methanol.
We obtained information about the surface of an op-
erating catalyst by measuring the open-circuit potential
(
0.4–0.7 V) (15) where coverage of Pt by hydrides and by
(
the catalyst potential with no applied current or voltage)
oxides is low. We believe, therefore, that two possible roles
of Fe(III)/Fe(II) during this oxidation are to act as a sink
for electrons and to maximize the number of active sites
available on Pt for oxidation of methanol by keeping the
surface of Pt free of oxides and hydrides.
of a Pt catalyst/electrode while it catalyzed oxidation of
methanol by Fe(III). The catalyst/electrode was a plati-
nized (15) square of 52-mesh Pt gauze with a roughness
factor of 684 (15). The potential of the catalyst/electrode
was 1.15 V versus the standard hydrogen electrode (SHE;
all potentials in this paper are reported versus SHE) before
addition of methanol. This potential is consistent with that
of a Pt surface covered with oxides (14). The oxides re-
sulted from the method we used to prepare the catalyst/
electrode surface (see the Methods section). Addition of
excess methanol (MeOH : Fe(III) ȁ 3.3 : 1) resulted in im-
mediate reduction of the catalyst/electrode potential to
Rates/stoichiometries for the Pt/Fe (SO ) oxidations of
2
4 3
functionalized hydrocarbons. Oxidation of the substrates
was carried out using the Fe(III)/Pt redox system:
Pt(0)
3
ϩ
2
ϩ
ϩ
R ϩ yH O ϩ xFe
Ǟ zCO ϩ xFe ϩ xH .
[1]
2
2
0
.65 V and evolution of CO from the catalyst/electrode Table 1 summarizes the results of this reaction for various
2
surface. The potential gradually decreased as Fe(III) was reductants (R). Our first interest was to establish the stoi-
consumed, reaching a minimum (0.48 V versus SHE) at chiometries of these reactions. Reaction stoichiometry was
the end of reaction, when evolution of CO had ceased. quantified in two complementary ways: by measuring the
2
Addition of more Fe(III) (remaining MeOH : added amount of CO produced using a gas buret and by measur-
2

PLATINUM-CATALYZED OXIDATIONS OF ORGANIC COMPOUNDS
95
oxidations using different samples of substrate in each.
Experimentally, there is ambiguity about the mechanisms
of oxidation at a platinum surface (18). Empirically, multi-
electron oxidation is possible under these conditions from
functionalized carbon compounds with no detectable
slow steps.
Operation of a fuel cell using ethylene glycol as the fuel.
Ethylene glycol was oxidized completely to carbon dioxide
in a prototype fuel cell similar to one using methane (9)
that was reported previously. Oxidation of ethylene glycol
to CO was carried out in a separate reformer flask, and
2
the reduced iron solution was pumped into the anode
chamber of the fuel cell to deliver the electron equivalents
produced in the oxidation reaction. Since the redox couple
FIG. 2. Rates of carbon dioxide production measured by gas buret. was the same as in the methane cell (9), the same electro-
Conditions are the same as those reported in Table 1.
chemical behavior is expected if ethylene glycol does not
oxidize much directly at the electrode. Indeed, similar be-
havior was observed and is illustrated in Fig. 3.
ing the amount of ferric ion reduced to ferrous ion by
redox titration.
The first general class of compounds listed in Table 1 is
DISCUSSION
C fuels in which each carbon atom is functionalized. These
2
In this article, it has been shown that several alcohols,
aldehydes, and carboxylic acids were oxidized under mild
conditions (60–80ЊC, 3 M H SO in water) by Fe(III) using
compounds were completely oxidized to CO at 80ЊC. The
2
extent of this reaction after 3–4 h was, at a minimum,
2
4
7
9% of completion as evidenced by the amount of CO2
a catalytic amount of Pt black to generate Fe(II) and pro-
tons. Under these conditions, complete oxidization of sev-
eral reductants was effected. On the basis of these results,
a redox fuel cell that catalyzed the complete oxidation of
ethylene glycol by dioxygen at 80ЊC was assembled. A
produced. By titration, it was found that at least enough
ferric ion was reduced to correspond to the expected
amount of fuel oxidized (16).
Ethanol and acetic acid both contain one unfunction-
alized (methyl) carbon. Under the experimental conditions
employed (80ЊC, 3–4 h reaction time), these reductants
3
maximum power density of 9.9 mW/cm of graphite felt
was obtained at a cell voltage of 197 mV. These results
indicate that the oxidation of the ethylene glycol in the
reformer section (Fig. 2) is not a limiting factor in the
electrical performance of the cell. The power density in
this porous felt electrode is not straightforwardly compared
with those in planar and gas diffusion electrodes. However,
as the porous felt employed has substantial surface area,
it is clearly low compared to that in gas diffusion electrodes
were not completely oxidized to CO . No gas evolution
2
was detected. The reduction of ferric ion suggests, how-
ever, that partial oxidation of the ethanol occurred, proba-
bly to at least the oxidation level of acetic acid (expected
x ϭ 4; see Eq [1]). We suspect that complete oxidation of
these species can be effected, albeit under conditions of
higher temperature and pressure (9).
Functionalized fuels with greater numbers of carbons
gave results on oxidation that depended on the structure.
Glycerin (three carbons) was oxidized completely under
these reaction conditions. When six carbon substrates (glu-
cose, sorbital) were employed as fuels, no gas evolved
and only a small fraction of the expected ferrous ion was
produced. Dehydrations of sugars in acidic media to form
functionalized furans are documented in the literature (17).
Given the results presented in this work, it is likely that
such fuels are difficult to oxidize under these conditions
and that cyclization/aromatization may be occurring.
The rates of oxidation of several fuels to carbon dioxide
were monitored with a gas buret; CO was identified by
2
GC/MS. Figure 2 summarizes rates. Interestingly, a long
FIG. 3. Current/voltage characteristics of the ethylene glycol/oxygen
fuel cell. Anode and cathode voltages were measured using separate
(
20–30 min) initiation period was observed in the oxidation
of glycolic acid. This observation was made in triplicate reference electrodes in each compartment.

96
GORMAN, BERGENS, AND WHITESIDES
4
5
6
7
. Oei, D.-G., J. Appl. Electrochem. 12, 41 (1982).
(
7, 8). However, by decoupling the catalytic oxidation of
. Kummer, J. T., and Oei, D.-G., J. Appl. Electrochem. 12, 87 (1982).
. Kummer, J. T., and Oei, D.-G., J. Appl. Electrochem. 15, 619 (1985).
. Prentice, G., Chemtech 14, 684 (1984).
the fuel from the electrochemical transfer of electrons to
the circuit, redox fuel cells provide several advantages over
conventional, direct fuel cells. The electrode material is
constructed at substantially lower cost and with greater
ease than the conventional gas diffusion electrode that has
been employed in a methanol fuel cell (19). As we have
shown previously, a high pressure and temperature re-
8. Lindstr o¨ m, O., Chemtech 18, 490 (1988).
9. Bergens, S. H., Gorman, C. B., Palmore, G. T. R., and Whitesides,
G. M., Science 265, 1418 (1994).
1
1
0. Carson, W. M., and Feldman, M. L., Proc. Annu. Power Sources
Conf. 13, 111 (1959).
1. Posner, A. M., Fuel 24, 330 (1955).
former that catalyzes the complete oxidation of methane 12. Matsen, J. M., Adv. Chem. Ser. 64, 277 (1967).
1
3. Carballo, L., Serrano, C., Wolf, E. E., and Carberry, J. J., J. Catal.
2, 507 (1978).
4. DiCosimo, R., and Whitesides, G. M., J. Phys. Chem. 93, 768 (1989).
5. Woods, R., in ‘‘Electroanalytical Chemistry: A Series of Advances’’
(A. J. Bard, Ed.), p. 1. Dekker, New York, 1976.
6. In all cases but one, slightly more ferric ion was reduced than ex-
pected, resulting in a higher than expected stoichiometry. This obser-
vation initially suggested that some of the platinum was oxidized to
Pt(II) and/or Pt(IV). However, with glyoxal, for example, 4.92 mmol
of Fe(III) was reduced, 4.45 mmol of which could be accounted for
by complete oxidation of the glyoxal. If the rest of the iron oxidized
the Pt(0), it would correspond to complete oxidation of the Pt(0) to
Pt(IV), (4.92–4.45)/4 ϭ 0.12 mmol of Pt(0) Ǟ Pt(IV) but 24.3 mg
Pt(0) и (1 mmol/195.09 mg) ϭ 0.12 mmol. Furthermore, substantial
amounts of platinum black remained after the reaction. Within this
experimental uncertainty, this stochiometry does correspond to com-
plete oxidation of these substrates.
to carbon dioxide can be substituted for this reformer using
the same cell scheme (9). Moreover, replacement and re-
covery of the catalyst is simpler and less expensive for
liquid reformers than for conventional fuel cell stacks (7).
Our results suggest that it is possible to develop redox fuel
cells that oxidize a variety of fuels for both production of
energy and for removal of contaminants from waste
streams; such work is in progress in our group.
5
1
1
1
ACKNOWLEDGMENTS
This work was supported in part by the ONR and ARPA. SHB thanks
the National Science and Engineering Research Council of Canada for
a Postdoctoral Fellowship.
1
7. Vinke, P., de Wit, D., de Goede, A. T. J. W., and van Bekkum, H.,
in ‘‘New Developments in Selective Oxidation by Heterogeneous
Catalysis’’ (P. Ruiz, and B. Delmon, Eds.), p. 1. Elsevier, New
York, 1992.
REFERENCES
1
2
3
. Sachtler, W. M. H., Catal. Rev. 4, 27 (1970).
. Kilty, P. A., and Sachtler, W. M. H., Catal. Rev. 10, 1 (1974).
. Langer, S. H., J. Appl. Electrochem. 23, 745 (1993).
18. McNicol, B. D., J. Electroanal. Chem. 118, 71 (1981).
19. Watanabe, M., Uchida, M., and Motoo, S., J. Electroanal. Chem. 199,
311 (1986).