You are on page 1of 11

Topics in Catalysis Vol. 40, Nos. 14, November 2006 ( 2006) DOI: 10.

1007/s11244-006-0113-7

111

Electrocatalysis for the direct alcohol fuel cell


verine Rousseau, Christophe Coutanceau, Jean-Michel Leger, and Claude Lamy* Fabrice Vigier, Se
Laboratory of Electrocatalysis, Universite de Poitiers-CNRS, UMR n6503, 40 avenue du Recteur Pineau, 86022 Poitiers, France

The basic principles of a direct alcohol fuel cell are rst presented. Low temperature fuel cells (working between ambient temperature and 80120 C) need improved catalysts to reach performance levels sufcient for practical applications, particularly for the electric vehicle and for portable electronic devices. This is the case of proton exchange membrane fuel cells (PEMFC) and of direct alcohol fuel cells (DAFC) for which the kinetics of the electrochemical reactions involved (oxidation of reformate hydrogen containing some traces of carbon monoxide, oxidation of alcohols, reduction of oxygen) is rather slow. Basic understanding of electrocatalysis is then examined, showing how to increase the reaction rate both by the nature and the structure of the catalytic electrode and by the electrode potential. Finally the most used Pt-based electrocatalysts to activate the electrode reactions occurring in a direct ethanol fuel cell (DEFC) are discussed on the basis of electrochemical, spectro-electrochemical and fuel cell experiments. KEY WORDS: direct alcohol fuel cell; ethanol electro-oxidation; IR reectance spectroscopy; reaction mechanisms.

1. Introduction Discovered in England in 1839 by Sir William Grove [1,2], the fuel cell is an electrochemical device, which transforms directly the heat of combustion of a fuel (hydrogen, natural gas, methanol, ethanol, hydrocarbons, etc.) into electricity. The fuel is electrochemically oxidized at the anode, without producing any pollutants (only water and/or carbon dioxide are rejected in the atmosphere), whereas the oxidant (oxygen from the air) is reduced at the cathode. This process does not follow the Carnots theorem, so that higher energy eciencies are expected: 4050% in electrical energy, 8085% in total energy (electricity + heat production). Whereas pure hydrogen or an hydrogen-rich gas as a fuel for polymer electrolyte membrane fuel cell (PEMFC) allows to obtain higher electric eciency and performance than alcohols, the clean production, storage and distribution of hydrogen are still strong limitations for the development of such techniques [3,4]. In this context, the use of hydrogen carriers like alcohols (methanol, ethanol, etc.), in a direct alcohol fuel cell (DAFC), appears then advantageous for two main reasons: they are liquids (easy storage) and their theoretical mass energy density is rather high, close to that of gasoline [5]. The case of alcohols (mainly methanol and ethanol) has often been considered [612], but as the consequence of the acidic environment of the ionomeric conducting membrane and of the low working temperatures of DAFCs (80120 C), a rather poor kinetics of electro-oxidation is observed with platinum, which is quite impossible to get over due to its catalytic properties to promote the CH bond cleavage during the rst
* To whom correspondence should be addressed. E-mail: claude.lamy@univ-poitiers.fr

adsorption steps. In order to improve the reaction kinetics it is necessary to develop new pluri-functional catalysts, called electrocatalysts because they activate electrochemical reactions, leading to increase the rate of fuel oxidation and oxygen reduction. This is the eld of electrocatalysis, i.e. the heterogeneous catalysis of electro-chemical reactions by the electrode material [1315]. In this paper, after recalling the working principles of a DAFC, the basic principles of electrocatalysis will be presented. Furthermore some electrocatalysts for the oxidation of the fuel (alcohols such as methanol and ethanol) will be discussed. Finally some examples of direct ethanol fuel cell (DEFC) electrical performance will be presented.

2. Experimental 2.1. Catalyst preparation The preparation method of the Pt based catalysts dispersed onto a carbon support (such as Vulcan XC72) is based on the synthesis of colloidal precursors using the procedure described by Bo nnemann et al. [16], but slightly modied. The preparation method was described elsewhere [17,18]. Briey, the synthesis method is as follows. The rst step consisted in the synthesis of the reducing agent by mixing a stoichiometric amount of tetra(alkyl)ammonium bromide (Nalk4)+Br) and potassium triethylhydroborate K+(BEt3H)) in tetrahydrofuran (THF) as solvent. After elimination of the precipitated KBr, a solution of tetra(alkyl)ammonium triethylhydroborate (Nalk4)+(BEt3H)) was obtained which reduced the metallic salts according to the following reaction, written in the case of platinum:
1022-5528/06/11000111/0 2006 Springer Science+Business Media, Inc.

112

F. Vigier et al./Direct alcohol fuel cell

PtCl2 2 Nalk4 BEt3 H ! PtNalk4 Cl 2 2 BEt3 H2 :

(a)

In this way, the platinum nanoparticles are stabilized by (Nalk4)+Cl), which acts as a surfactant protecting the metal particles by its long alkyl chain. The colloid particles were then adsorbed on Vulcan XC 72, previously treated for 4 h at 400 C under nitrogen to clean it from catalyst poisoning impurities (e.g. sulphur), in order to obtain a catalyst loading in the range of 3060 wt.% based on the metal content. Before using them as electrocatalysts, the organic surfactant shell of the supported colloid catalysts must be removed by thermal treatment under air atmosphere at 300 C. 2.2. Catalyst characterization Characterizations of the catalyst particles so obtained were carried out by dierent physical techniques. Transmission electron microscopy (TEM images recorded with a Phillips CM 120 microscope with a resolution of 0.35 nm) was used to observe the morphology of their surface and to determine the particle size distribution. Figure 1 gives a TEM image (gure 1a) and the size distribution (gure 1b) of a typical catalyst for ethanol electro-oxidation, i.e. PtSn, with a Pt to Sn atomic ratio (90:10), supported on Vulcan XC72 (30 wt.%). An estimation of the particle composition in dierent zones was realized by energy dispersive analysis of X-rays (EDX), as shown in gure 2 for a 30% PtSn(90:10)/ XC72 catalyst. X-ray diraction (XRD), recorded with a Brucker D5005 diractometer tted with an AR monochromator and a scintillation detector, was used to determine the catalyst structure and evaluate the particle size (gure 3). Typical results of such characterizations are given in table 1 for Pt/C and PtSn/C catalysts. 2.3. Electrochemical measurements The catalyst is deposited on a 0.071 cm2 surface area vitreous carbon (VC) rotating electrode according to the following procedure as described previously [19]. Briey, 3 lL of an ink made of 2.5 mL of ultra pure water, 25 mg of the catalytic powder and 0.5 mL of Naon 5 wt.% in water/aliphatic alcohol solution (Aldrich), are deposited on the VC electrode using a micro syringe. The electrode is then dried in an oven at 70 C for 30 min. The electrochemical set-up consists of a Voltalab PGZ 402 potentiostat controlled by a computer, a Radiometer Unit CTV 101 Speed Control and a Radiometer BM-EDI 101 rotating disk electrode (RDE). The counter electrode was a vitreous carbon plate and a reversible hydrogen electrode (RHE) was used as reference electrode. Electrochemical measurements were carried out in a N2 saturated 0.5 M H2SO4 electrolyte at a 20 mV s)1 sweep rate and at 298 K.
300

(b)
250

Number of particles

200 150 100 50 0 0 1 2 3 4 5 6

Particle size / nm
Figure 1. (a) TEM image of a Vulcan supported PtSn(90:10)/XC72 catalyst (with a metal loading of 30%) prepared by the colloidal method. (b) Particle size distribution of a PtSn(90:10)/XC72 catalyst (based on the observation of 1005 particles). Mean diameter: 2.4 0.3 nm.

2.4. In situ infrared reectance spectroscopic measurements In order to better understand the overall mechanism of alcohol oxidation on the dierent electrocatalysts, in situ IR reectance spectroscopic studies were performed. With this technique it is possible to observe the adsorbed species at the electrode surface and intermediates and reaction products formed at its vicinity. IR spectra were obtained with a Bruker IFV 66 spectrometer with data acquisition techniques, allowing to perform single potential alteration infrared spectroscopy (SPAIRS) and substractively normalized interfacial Fourier transform infrared reectance spectroscopy (SNIFTIRS) experiments, as previously described in details [20].

F. Vigier et al./Direct alcohol fuel cell

113

(a) 500
400

C (support) Cu PtM

Cu (grille)

300 200 100 0 0

Cu

PtL

SnL SnL

PtL

10

12

Energy / keV

surface area were carried out with a Globe Tech test bench. The cell voltage E and power density P versus current density j curves were recorded using a high power potentiostat (Wenking model HP 88) interfaced with a variable resistance in order to x the current applied to the cell and with a PC to apply constant current sequences and to store the data. A high performance liquid chromatograph (HPLC) (Dionex CHROMELEON) tted with an isocratic pump, an auto-sampler, an UV detector and a refractometer detector was used to analyse quantitatively the reaction products at the outlet of the anode side of the DEFC, as it was described in a previous work [21].

Counts (a.u.)

(b)
Element Zone n1 Zone n2 Zone n3 Pt 89.7 91.1 89.7 Sn 10.3 8.9 10.3

3. Results and discussion 3.1. Working principles of the DAFC The principles of the fuel cell will be illustrated with a DEFC taken as a typical example (gure 4). The electrochemical cell consists in 2 electrodes, an anode and a cathode, which are electronic conductors, separated by an electrolyte, a proton exchange membrane (PEM), which is an ionic conductor (thanks to migration and diusion of protons). At the anode (negative pole of the cell) the electro-oxidation of ethanol takes place as follows: CH3 CH2 OH 3 H2 O ! 2 CO2 12 H 12 e E0 1 0:085 V vs SHE; 2 whereas the cathode (positive pole) undergoes the electro-reduction of oxygen, i.e.: O2 4 H 4 e ! 2 H2 O E0 2 1:229 V vs SHE; 3 where E0 i are the standard electrode potentials versus the standard hydrogen (reference) electrode (SHE). This corresponds to the overall combustion reaction of ethanol in oxygen: CH3 CH2 OH 3 O2 ! 2 CO2 3 H2 O; 4

Figure 2. (a) EDX analysis of a 30% PtSn(90:10)/XC72 catalyst. (b) Atomic composition measured at dierent sample zone.

Pt3Sn Pt
Intensity / a.u.

Pt Pt-Sn(90:10) Pt-Sn(80:20) Pt-Sn(75:25) 30 40 50 60 70 80 90 2 theta / degree

Figure 3. X-ray diraction (XRD) patterns of PtSn/XC72 catalysts with dierent atomic composition. Comparison with pure Pt.

2.5. Fuel cell tests and product analyses The preparation of electrodes and membrane electrode assemblies (MEA) were described elsewhere. Fuel cell tests in a single cell with 5 or 25 cm2 geometric

Table 1 Values of the physical and electrochemical parameters of PtSn/C catalysts with dierent composition Electrocatalyst 30% 30% 30% 30% 60%
a b

dTEM/nm 2.4 2.4 2.7 2.9 2.8

Dispersion a/% 44 45 41 38 40

dXRD/nm 2.1 1.9 2.1 2.0 3.4

Atomic composition (EDX)

Sb/m2 g)1 39 24 15 8 11

Pt/XC72 PtSn(90:10)/XC72 PtSn(80:20)/XC72 PtSn(75:25)/XC72 PtSn(90:10)/XC72

89.7/10.3 80.4/19.6 74.1/25.9 89.8/10.2

The dispersion was evaluated from the particle size measured on TEM images. The true surface area S was calculated from the hydrogen adsorption region in a cyclic voltammogram recorded at 50 mV s)1 on the electrocatalyst in contact with the supporting electrolyte alone (0.1 MHClO4).

114

F. Vigier et al./Direct alcohol fuel cell

I
We = - G G =8 = 8.0 .0 kWh/kg k Wh/kg

A
12e-

Anode
elec electronic troni cconductor conduc to r + catalyst ca ta lyst +

12eElectrolyte
protonic protonic conductor

+
Cathode
elelectronic conductor + catalyst

C2H5OH + 3 H2O

Oxygen 2CO2+12H++1 2e12H+ 12e-+12H++3O2 (Air)

I
6 H20

2 CO 2

Figure 4. Schematic principle of a direct ethanol/oxygen fuel cell (DEFC).

with the thermodynamic data, under standard conditions: DG0 1325 kJ mol1 ; DH0 1366 kJ mol1 ; 5 This gives a standard electromotive force (emf) at equilibrium: E0 eq DG0 1325 103 nF 12 96485
0 E0 2 E1 1:144 V;

ohmic drop Re j in the electrolyte and interface resistance Re, and mass transfer limitations for reactants and products (gure 5 in the case of an H2/O2 fuel cell). The cell voltage can thus be expressed as follows: Ejjj E2 jjj E1 jjj
0 E0 2 gc E1 ga Re jjj

E0 eq

jga j jgc j Re jjj:

with F = 96485 C the Faraday constant and n = 12 the number of electrons exchanged per molecule for complete oxidation to CO2. The protons produced at the anode crossover the membrane, ensuring the electrical conduction inside the electrolyte, whereas the electrons liberated at the anode reach the cathode (where they reduce oxygen) through the external circuit producing an electrical energy, 0 Wel nFE0 eq DG ; with a mass energy density We = )DG0/(3600 M) % 8 kWh kg)1, where M = 0.046 kg is the molecular weight of ethanol. The energy eciency, under reversible standard conditions, is dened as the ratio between the electrical energy ()DG0) and the heat of combustion ()DH0) at constant pressure, i.e.: erev eq We1 DH0 DG0 TDS0 1325 0:97 1 0 1366 DH DH0

where the overvoltages ga (>0 for an anodic reaction, i.e. the oxidation of the fuel) and gc (<0 for a cathodic reaction, i.e. the reduction of the oxidant) take into account both the slow kinetics of the electrochemical reactions (activation polarization) and the limiting rate of mass transfer (concentration polarization). In addition the oxygen electrode experiences a mixed potential due to ethanol crossover, which shifts its potential towards more negative potentials, thus increasing the cathode overvoltage [8]. Therefore, for a DEFC working at 0.5 V at 50 mA cm)2 with complete oxidation to CO2, the energy eciency would be:
2 ecell

C H5 OH=O2

nexp FEjjj DH0 rev eeq eE eF 0:97 0:437 1 0:424;

However under working conditions, with a current density j, the cell voltage E(j) decreases greatly as the result of three limiting factors: the charge transfer overvoltages ga and gc at both electrodes due to a rather low reaction rate of the electrochemical processes (g is dened as the dierence between the working electrode potentials Ei and the equilibrium potential Eeq i ), the

5 0:437 and since the potential eciency is eE 10::14 nexp the faradaic eciency is eE nth 1; for complete oxidation to CO2. This is quite similar to that of the best thermal engine (Diesel engine). But, if the reaction process stops at the acetic acid stage, which involves the transfer of 4 electrons (instead of 12 for complete oxidation), the eciency will be reduced by one third, reaching only 0.14. It appears that the only way to increase signicantly the overall energy eciency is to increase eE and eF , i.e. to decrease the overvoltages g and the ohmic drop Re j,

F. Vigier et al./Direct alcohol fuel cell

115

1,200 1,000 0,800 0,7 0,600 0,400 0,200 0,000

Eeq = 1.23 V Charge transfer overvoltage Ohmic losses R e j Mass transfer overvoltage

E/ V

0,25 0,4

0,5

0,75 j/A cm -2

0,9 1

1,25

1,5

Figure 5. Theoretical cell voltage versus current density E(j) curves determined from equation (8). j j0 = 10)8 A cm)2, Re = 0.3 W cm2, jl = 1.2 A cm)2; j0 = 10)8 A cm)2, Re = 0.15 W cm2, jl = 1.3 A cm)2; m j0 = 10)6 A cm)2, Re = 0.15 W cm2, jl = 1.4 A cm)2. j0 is the exchange current density for oxygen reduction and jl the limiting current density resulting from mass transfer limitations.

since erev eq is given by the thermodynamics (one can increase it slightly by changing the pressure and temperature operating conditions). The decrease of |g| is directly related to the increase of the rate of the electrochemical reactions occurring at both electrodes. This is typically the eld of electrocatalysis, where both the action of the electrode potential and of the catalytic electrode material will synergistically increase the reaction rate v. Indeed, the current intensity I is proportional to the rate of reaction v, i.e. I = nFv. For heterogeneous reaction v is proportional to the surface area A of the interface, so that the kinetics of electrochemical reactions is better dened by the intrinsic rate vi = v/A and the current density j = I/A = nFvi. Introducing the exponential behavior of the rate constant with the electrochemical activation energy, i.e.: DG DG 0 a nFE;
^

10

Figure 6. Activation barrier for an electrochemical reaction: K is the decrease in activation energy due to the electrode catalyst and anFE that one resulting from the electrode potential E.

which comprises 2 terms, the rst one, DG 0 , being the chemical energy of activation, and the second one, anFE, the electrical part of the activation energy. This last term is a fraction a1 (0 a 1) of the total electrical energy coming from the applied electrode potential E (gure 6). One obtains, in the theory of absolute reaction rate, for a rst order electrochemical reaction (the rate of which is proportional to the reactant concentration ci): j nFvi nFkT;E ci nFk0 ci e
DG 0 RT

tion of the electrode potential E, and an exponential function of the chemical activation energy DG 0 . By modifying the nature and structure of the electrode material, one may decrease DG 0 ; thus increase j0, as the result of the catalytic properties of the electrode. This leads to an increase of the reaction rate j. 3.2. Electro-oxidation of ethanol at platinum catalysts Ethanol oxidation has been extensively studied at platinum electrodes [2224]. Iwasita and Pastor [23,24] have shown that ethanol can undergo C-adsorption on platinum: Pt CH3 CH2 OH ! Pt CHOHCH3 H e 12 In a recent work, the analysis of the reaction products at the outlet of the anode compartment of a DEFC tted

anFE RT

j0 e

anFE RT :

11

This last equation contains the 2 essential activation terms met in electrocatalysis, i.e. an exponential func1

a is called the charge transfer coecient

116

F. Vigier et al./Direct alcohol fuel cell

with a Pt/C anode showed that only acetic acid (AA), acetaldehyde (AAL) and carbon dioxide (CO2) could be detected by HPLC [21]. Several mechanisms were proposed to explain the formation of the reaction products at platinum electrodes [2234]. Depending on the electrode potential, acetaldehyde (AAL), acetic acid (AA), carbon dioxide (CO2) and traces of methane (CH4) are observed, the main products being AAL and AA (table 2), the latter being considered as a nal product because it is not oxidized under smooth conditions. According to electrochemical and spectro-electrochemical studies, the reaction mechanism given in gure 7 is generally admitted. In this mechanism, the rst reaction product coming from the dissociative adsorption of ethanol at platinum surface is acetaldehyde, which requires only the transfer of 2 electrons per ethanol molecule. Acetaldehyde has to
Table 2 Chemical yields in AA, AAL and CO2 for Pt/C and PtSn (9:1)/C catalysts during the electro-oxidation of ethanol for 4 h at a constant current density (8 mA cm)2 for Pt and 32 mA cm)2 for PtSn) in a 25 cm2 surface area DEFC working at 80 C Electrocatalyst Chemical yield in AA/% 33 77 Chemical yield in AAL/% 47 15 Chemical yield in CO2/% 20 8

60 wt.% Pt/C XC72 60 wt.% PtSn (9:1)/ C XC72

readsorb to complete its oxidation either into acetic acid or carbon dioxide with methane production at low potentials (E < 0.2 V versus RHE). To complete the oxidation reaction leading to both these species, extra oxygen atom is needed, which has to be brought by activated (adsorbed) water molecules at the platinum surface. The study of ethanol electro-oxidation by in situ infrared reectance spectroscopy may allow observing the adsorbed intermediate species and the reaction products, what can give some interesting information concerning the reaction mechanism. Figure 8 displays the IR spectra of the species coming from ethanol adsorption and oxidation recorded at a Pt/C electrode. Considering the SNIFTIR spectra (gure 8a), an absorption band located close to 2050 cm)1 related to linear bonded CO species, COL, appears as soon as the third potential modulation between 200 and 400 mV versus RHE and disappears in the potential modulation region between 1000 and 1200 mV versus RHE [3537]. This indicates that platinum is able to break the CC bond of ethanol and to adsorb CO species at relatively low potentials. On the other hand, looking at the SPAIR spectra (gure 8b), one can see that an absorption peak appears close to 2345 cm)1 revealing the formation of CO2 at potentials higher than 0.6 V versus RHE. As soon as CO2 was formed, the amount of adsorbed CO decreases as it is shown in gure 9, which indicates that this latter species is strongly adsorbed at the platinum surface and blocks the active sites for potentials lower

CH3

Ethanol

HO H

OH2

H3O+

H3C

OH2

H3O+ O

CH3 H

Acetaldehyde
0. .8 E/V(RHE)

1e-

1e-

Acetaldehyde
O

CH 3 H OH 2 H3O
+

Adsorbed acetyl
CH3 O C OH

CH3

O
OH

Acetic acid

1e-

H3O + +1e-

2H 2O

Adsorbed acetyl
H3C C O

CH3 H

CH4 E < 0.2 V (RHE)

O C O

OH2

H3O+

O C O E > 0.5V (RHE)

1e-

Figure 7. Mechanism of ethanol electrooxidation at platinum surface in acid medium.

F. Vigier et al./Direct alcohol fuel cell

117

(a)
0.2 %
0-200 mV vs. RHE 100 -300

(b)
3.0 %
A

(A) 100 mV
D

(B) 300 mV (C) 500 mV (D) 550 mV

Absorban sorbance / %

200-400 300 -500 400-600 500-700 600-800 700 -900 800-1000 900 -1100 1000 -1200

Absorban bsorbance / %

(E) 600 mV (F) 650 mV (G) 700 mV (H) 800 mV

(I) 1000 mV

CO2 COL C=O 1000 1500 COL 2000 2500

(J) 1200 mV (K) 1400 mV

1000

1500 2000

2500 3000

3000

Wavenumber / cm-1

Wavenumber / cm-1

Figure 8. IR spectra of the species coming from ethanol adsorption and oxidation on a 30% Pt/XC72 electrode. 0.1 M HClO4 + 0.1 M C2H5OH; T = 25 C. (a) substractively normalized interfacial Fourier transform infrared reectance (SNIFTIR) spectra; (b) single potential alteration infrared (SPAIR) spectra [Eref = 50 mV versus reversible hydrogen electrode (RHE)].

that 0.6 V versus RHE, but is oxidized into CO2 as soon as platinum is able to activate (by chemisorption) water molecules according to the bifunctional theory of electrocatalysis [38], as follows: Pt H2 O ! Pt OH H e ; 13

Pt CO Pt OH ! 2 Pt CO2 H e 14 In the SPAIR spectra, an absorption peak can also be seen at 1725 cm)1, appearing at potentials between 0.65 and 0.8 V versus RHE, and which is related to adsorbed

carbonyl species from acetic acid or acetaldehyde [23,39,40]. At potentials higher than 0.8 V versus RHE a broad peak located close to 2600 cm)1 appears, which is assigned to the (CH) vibration mode of acetic acid or to the OH stretching mode of the carboxylic group overlapped by the CH stretching mode of the methyl group [41]. The formation of acetic acid can be described according to the following equation: Pt H2 O ! Pt OH H e ; 15

Pt CHO CH3 ! Pt CO CH3 H e ; 16

Pt CO CH3 Pt OH ! 2 Pt CH3 COOH:


COL band CO2 band

17

Absorbance / % (a.u.)

The spectro-electrochemical study of the adsorption and oxidation of ethanol and the HPLC analyses of reaction products underline the necessity to activate water molecules at lower potentials in order to increase the activity of the catalyst and the selectivity towards either acetic acid or CO2 formation, which means the improvement of the potential eE and faradic eF eciencies. To perform this, the modication of platinum by other metal is necessary. 3.3. Electro-oxidation of ethanol at plurimetallic catalysts

200

400 600 800 1000 1200 1400 Potential / mV vs. RHE

Figure 9. Correlation between the CO infrared peak intensity and the CO2 peak intensity for a 30% Pt/XC72 catalyst; values taken from the single potential alteration infrared (SPAIR) (CO2) and the substractively normalized interfacial Fourier transform infrared reectance (SNIFTIR) spectra (CO) of gure 8.

Several added metals were investigated to improve the kinetics of ethanol oxidation at platinum based electrodes, including ruthenium [26,27], lead [28] or tin [29,30]. Amongst them, tin appeared to be very promising. Figure 10 shows the polarization curves of ethanol electro-oxidation recorded at a slow sweep rate (5 mV s)1) on dierent platinum based electrodes. The

118

F. Vigier et al./Direct alcohol fuel cell

50 40

j / mA mgPt

Pt/XC72 Pt-Sn (90:10)/XC72 Pt-Sn (80:20)/XC72 Pt-Ru (80:20)/XC72 Pt-Mo (80:20)/XC72

30 20 10 0 0.0 0.1 0.2 0.3 0.4 0.5

E / V vs. RHE
Figure 10. Electro-oxidation of ethanol on Pt/C (full line) and dierent Pt-based (dashed and dotted lines) catalysts with 0.1 mgPt cm)2 loading. 0.1 M HClO4 + 1 M C2H5OH; 5 mV s)1; 3000 rpm; 20 C.

PtSn(0.9:0.1)/C displays the best activity in the potential range from 150 to 500 mV versus RHE, as it gives higher oxidation current density than the other catalysts. The role of tin can also be pointed out by analyzing the distribution of reaction products at the anode outlet of a fuel cell tted with a PtSn(0.9:0.1)/ C anode and from in situ IR reectance spectroscopy measurements. Table 2 indicates that alloying platinum with tin led to an important change in the product distribution: increase of the AA chemical yield and decrease of the AAL and CO2 chemical yields. The presence of tin seems to allow, at lower

potentials, the activation of water molecules and the oxidation of AAL species into AA. In the same manner, the amount of CO2 decreased, which can be explained by the need of several adjacent platinum atoms (3 or 4) to realize the dissociative adsorption of ethanol into CO species, via breaking the CC bond. In the presence of tin, dilution of platinum atoms can limit this reaction. The eect of tin, in addition to the activation of water molecules, may be related to some electronic eects (ligand eects) on the CO oxidation reaction [42]. Figure 11 displays the SNIFTIR and SPAIR spectra recorded at a PtSn(0.9:0.1)/XC72 electrode. The COL band (gure 11a), at the second potential modulation between 100 and 300 mV versus RHE, appears at a potential 100 mV lower that at a platinum catalyst and disappears at a potential close to 500 mV versus RHE (what is 600 mV lower than at platinum). This fact is very important, because it indicates that the presence of tin atoms close to platinum atoms favored the dissociative adsorption of ethanol at low potentials and led to changes in the electronic structure of platinum allowing weakening of the PtCO bond (ligand eect [42]). In the SPAIR spectra (gure 11b), one can see that CO2 is detected as soon as 500 mV versus RHE, which is a potential 150 to 200 mV lower than that at a platinum electrode. On the other hand, the relative intensity of the signal is very low compared to that recorded at a platinum electrode, indicating, in agreement with the analysis of the reaction products, that the production of CO2 is smaller at such electrodes. The dissociative adsorption of ethanol is favored at lower potentials, but in lower amounts at PtSn catalysts. Moreover, conversely to

-1

(a)

0-200 mV vs.RHE 100-300

(b)
A

(A) 100 mV
Absorbance / % 0.25 % Absorbance / % 3.0 %
C E

(B) 300 mV (C) 500 mV (D) 600 mV (E) 650 mV

200-400 300-500 400-600 500-700 600-800

(F) 700 mV (G) 750 mV (H) 800 mV (I) 900 mV

CO COL L
1000 1500

700-900
3000 1000

C? O =O C O

CO CO 22

(J) 1000 mV

2000 2500 Wavenumber/ cm-1

1500 2000 2500 Wavenumber/ cm-1

3000

Figure 11. IR spectra of the species coming from ethanol adsorption and oxidation on a PtSn(90:10)/XC72 electrode. 0.1 M HClO4 + 0.1 M C2H5OH; T = 25 C. (a) substractively normalized interfacial Fourier transform infrared reectance (SNIFTIR) spectra; (b) single potential alteration infrared (SPAIR) spectra [Eref = 50 mV versus reversible hydrogen electrode (RHE)].

F. Vigier et al./Direct alcohol fuel cell

119

what happens at a Pt/C electrode, the appearance of the CO2 absorption band in SPAIRS experiments is not directly correlated with the disappearance of the COL absorption band as determined in SNIFTIRS experiments (gure 12). It appears that with a PtSn/C catalyst, the intensity of the COL band decreases from 300 mV versus RHE, whereas a CO2 band is clearly detected only at 650 mV versus RHE. This means that the small amount of CO2, which is likely formed at potentials close to 350400 mV versus RHE resulting from the oxidation of adsorbed CO, is not detected under our experimental conditions. This is probably due to the too small amount of CO2, coming only from a monolayer of adsorbed CO, which remains below the limit of detection of the SPAIRS experiments. At potentials higher than 600 mV versus RHE, the direct oxidation of ethanol and the breaking of the CC bond do occur at this catalyst leading to form CO2, without formation of adsorbed CO. This step should be related to the presence of acetaldehyde and acetic acid as reaction products as it can be seen from the IR spectra. The dissociation of acetaldehyde does not lead to the formation of strongly adsorbed CO, and its further oxidation to CO2 appears to be easier with the PtSn/C electrocatalysts. 3.4. Fuel cell tests Fuel cell tests (determination of the cell voltage E and power density P versus the current density j) were carried out in a single DEFC with 5 or 25 cm2 active surface area electrodes made with several Pt-based electrocatalysts (gure 13). The pressure applied to the

800 700 600 Cell voltage / mV 500 400 300 200 100 0 0 20 40 60 80 100 120 140 160 Current density / mA.cm-2
Pt Pt-Sn(80:20) Pt-Ru(80:20) Pt-Mo(80:20)

30

Power density / mW.cm-2

25 20 15 10 5 0 0

Pt Pt-Sn(80:20) Pt-Ru(80:20) Pt-Mo(80:20)

20

40 60 80 100 120 140 160 Current density / mA.cm-2

Figure 13 Fuel cell characteristics of a direct ethanol fuel cell (DEFC) recorded at 110 C. Inuence of the nature of the bimetallic catalysts (30% loading). Anode catalyst: 1.5 mg cm)2; cathode catalyst: 2 mg cm)2 (40% Pt/XC72 from E-TEK); membrane: Naon 117; ethanol concentration: 1 M; pEtOH = 1 bar; pO2 = 3 bar.

CO L band CO 2 band

200

400

600 800 1000 1200 Potential / mV vs. RHE

1400

Figure 12. Correlation between the CO infrared peak intensity and the CO2 peak intensity for a 30% PtSn(90:10)/XC72 catalyst. Values taken from the single potential alteration infrared (SPAIR) (CO2) and the substractively normalized interfacial Fourier transform infrared reectance (SNIFTIR) spectra (CO) of gure 11.

anodic compartment (pfuel = 1 bar) was usually lower than that applied to the cathodic one (pO2 = 3 bar), in order to limit the ethanol crossover through the Naon membrane. The use of platinum alone as anode catalyst leads to poor electrical performance, the open circuit voltage (OCV) being lower than 0.5 V and the maximum current density reaching only 100 mA cm)2, leading to a maximum power density lower than 7 mW cm)2 at 110 C. The addition of Ru, and above all of Sn to Pt in the anode catalyst leads to greatly enhance the electrical performance of the DEFC by increasing the OCV up to 0.75 V, which indicates that the bi-metallic catalyst is less poisoned by adsorbed species coming from ethanol than the Pt/C catalyst. In the latter case, current densities up to 150 mA cm)2 are reached giving a maximum power density greater than 25 mW cm)2, which means electrical performances four times higher than those obtained with Pt/C. The increase of the electrical performance indicates that the bi-metallic catalyst is more active for ethanol electro-oxidation than the Pt/C catalyst. Both these results are consistent with the

Absorbance / % (a.u.)

120

F. Vigier et al./Direct alcohol fuel cell

electrochemical and spectro-electrochemical studies, and analyses of the reaction products discussed earlier in this paper. With the best electrocatalyst, i.e. PtSn(0.9:0.1)/ XC72, the eect of temperature on the cell voltage E and power density P versus current density j characteristics, is shown in gure 14 at 50, 70, 90, 100, and 110 C, respectively. It appears clearly that increasing the temperature greatly increases the performance of the cell, from a maximum power density close to 5 mW cm)2 at 50 C, to 25 mW cm)2 at 110 C, i.e. almost ve times higher. This fact conrms the diculty of oxidizing ethanol at low temperatures and the necessity to work at temperatures higher than 100 C to enhance the electrode kinetics and, thus, the performance of a DEFC. Finally the performance of a DEFC with an anode containing a higher amount of platinum (i.e. 60 wt.% instead of 30 wt.% for the previous experiments) were determined with 3 Naon membranes, N117, N115, and N112 of dierent thickness (180, 125, and 50 lm, respectively) to investigate the eect of ethanol crossover through the membrane. Figure 15 shows the ethanol crossover as measured by following the ethanol concentration in a second cell compartment, containing initially no ethanol, separated by the Naon membrane from the rst one containing 1 M ethanol. The crossover rate through N117 is about twice lower than that through N112. The better behaviour of N117 is conrmed in gure 16 showing the comparative electrical characteristics of a DEFC having one of the 3 Naon membranes and an anode with a higher platinum loading (60% PtSn(90:10)/XC72): the DEFC with N117 displays the highest OCV (0.8 V) and leads to a power density of 52 mW cm)2.

Ethanol concentration / mmol.L-1

300 250 200 150 100 50 0 0

Nafion 117 Nafion 115 Nafion 112

3 4 5 Time / hour

Figure 15. Behavior of dierent Naon membranes. Ethanol permeability measurements of dierent Naon membranes. T = 25 C; 1 M C2H5OH.

800 700

50 40 30 20 10 0

600 500 400 300 200 100 0 0 40 80 120 160 200

Current density / mA.cm-2


Figure 16. Fuel cell characteristics of a direct ethanol fuel cell (DEFC) recorded at 90 C with a 60% PtSn(90:10)/XC72 catalyst with 3 mgPt cm)2 loading for dierent Naon membranes; pEtOH=1 bar; pO2 = 3 bar.

800 700 600

E / mV

500 400 300 200 100 0 0 20 40 60 80 100 j / mA cm-2 120 140

20 15 10 5

0 160

Figure 14. Fuel cell characteristics of a 25 cm2 direct ethanol fuel cell (DEFC) recorded with a 30% PtSn(90:10) catalyst. Inuence of the working temperature. Anode catalyst: 1.5 mg cm)2 (30% PtSn(9:1)/ XC72); cathode catalyst: 2 mg cm)2 (40% Pt/XC72 from E-TEK); membrane: Naon 117; ethanol concentration: 1 M; pEtOH = 1 bar; pO2 = 3 bar.

P / mW cm-2

B C 50 50C 70C 70 CC 90C 90 DC C 100 100 E C 110 C 110 F C

30 25

4. Conclusion This paper presented the main features concerning the direct oxidation of ethanol in a fuel cell. To achieve the best electric eciency, it is necessary to develop electrocatalysts able to oxidize completely ethanol into CO2 with a high reaction rate. However, in a strongly acidic medium, as in a DEFC using a PEM, platinum is necessary to realize the dissociative adsorption of ethanol, but is easily poisoned by adsorbed species like CO. To reduce the poisoning eect, it is necessary to add to platinum other metals, like tin. PtSn catalysts display higher activity for ethanol electro-oxidation, in an electrochemical half-cell as well as in a DEFC. However, the selectivity towards CO2 was decreased as well as selectivity towards acetaldehyde, whereas that towards acetic acid increased. Therefore, it seems that the challenge to increase simultaneously the faradic eciency eF (selectivity of the catalyst towards complete oxidation)

Power density / mW.cm-2

Nafion 112 Nafion 115 Nafion 117

60

Cell voltage / mV

F. Vigier et al./Direct alcohol fuel cell

121

and the potential eciency eE (reduction of catalyst poisoning) has to be abandoned. At least, one goal could be to achieve a total selectivity of the electrocatalysts towards the production of acetic acid, which is a liquid easy to manage, but provides the third of the faradic eciency (eF = 0.33), and thus decreases the overall eciency and power density by one third.

References
[1] [2] [3] [4] [5] [6] [7] [8] W.R. Grove, Phil. Mag. 14 (1839) 127. W.R. Grove, Phil. Mag. 15 (1839) 287. J.W. Gosselink, Int. J. Hydrogen Energy 27 (2002) 1125. R. Stro bel, M. Oszcipok, M. Fasil, B. Rohland, L. Jo rissen and J. Garche, J. Power Sources 105 (2002) 208. ger, J. Phys. IV 4 (1994) C1. C. Lamy and J.-M. Le C. Lamy, A. Lima, V. Le Rhun, F. Delime, C. Coutanceau ger, J. Power Sources 105 (2002) 283. and J.-M. Le E. Peled, T. Duvdevani, A. Aharon and A. Melman, Electrochem. Solid State Lett. 44 (2001) A38. ger and S. Srinivasan, in: Modern Aspects of C. Lamy, J.-M. Le Electrochemistry, Vol. 34, eds. J.OM. Bockris and B.E. Conway (Plenum Press, New York, 2000), ch. 3, pp. 53117. ger, J. Appl. Electrochem. 31 C. Lamy, E.M. Belgsir and J.-M. Le (2001) 799. T. Iwasita-Vielstich, in: Advances in Electrochemical Science and Engineering, Vol. 1, eds. H. Gerischer and C.W. Tobias (VCH Verlag, Weinheim, 1990) p. 127. A. Hamnett, Catal. Today 38 (1997) 445. K.-Y. Chan, J. Ding, J. Ren, S. Cheng and K.Y. Tsang, J. Mater. Chem. 14 (2004) 505. J.OM. Bockris and A.K.N. Reddy, Modern Electro-chemistry, 2 (Plenum Press, New York, 1972) 1141. G.P. Sakellaropoulos, in: Advances in Catalysis, eds. D.D. Eley, H. Pines and P.B. Weisz (Academic Press, New York, 1981) p. 218. A.J. Appleby, in: Comprehensive Treatise of Electrochemistry, Vol. 7, eds. B.E. Conway, J.OM. Bockris, E. Yeager, S.U.M. Khan and R.E. White (Plenum Press, New York, 1983) pp. 173239. H. Bo nneman, W. Brijoux, R. Brinkmann, E. Dinjus, T. Joussen and B. Korall, Angew. Chem. Int. Engl. 30 (1991) 1312. ger and C. Lamy, L. Dubau, C. Coutanceau, E. Garnier, J.-M. Le J. Appl. Electrochem. 33 (2003) 419.

[9] [10]

[11] [12] [13] [14]

[15]

[16] [17]

[18.] C. Lamy, S. Rousseau, E.M. Belgsir, C. Coutanceau and J.-M. ger, Electrochim. Acta 49 (2004) 3901. Le [19] F. Gloaguen, F. Andolfatto, R. Durand and P. Ozil, J. Appl. Electrochem. 24 (1994) 861. [20] A. Kabbabi, R. Faure, R. Durand, B. Beden, F. Hahn, J.-M. ger and C. Lamy, J. Electroanal. Chem. 444 (1998) 41. Le ger, J. Power [21] S. Rousseau, C. Coutanceau, C. Lamy and J.-M. Le Sources (2005 in press). [22] R.A. Rightmire, R.L. Rowland, D.L. Boos and D.L. Beals, J. Electrochem. Soc. 111 (1964) 242. [23] T. Iwasita and E. Pastor, Electrochim. Acta 39 (1994) 531. [24] T. Iwasita and E. Pastor, Electrochim. Acta 39 (1994) 547. [25] C. Lamy, A. Lima, V. Rhun, F. Delime, C. Coutanceau and J.-M. ger, J. Power Sources 105 (2002) 283. Le [26] J. Wang, S. Wasmus and R.F. Savinell, J. Electrochem. Soc. 142 (1995) 4218. [27] J. Souza, F.J.B. Rabelo, I.R. Moraes and F.C. Nart, J. Electroanal. Chem. 420 (1997) 17. [28] H. Hitmi, PhD Thesis (University of Poitiers, 1992). ger and C. Lamy, J. Appl. Electrochem. 29 [29] F. Delime, J.-M. Le (1999) 1249. ger, J. Appl. Electrochem. 31 [30] C. Lamy, E.-M. Belgsir and J.-M. Le (2001) 799. [31] A. Rezzouk, PhD Thesis (University of Poitiers, 1994). [32] M. Bonarowska, A. Malinowski and Z. Karpinski, Appl. Catal. A: General 188 (1999) 145. [33] A.K. Aboul-Gheit, M.F. Menoufy and A.K. El-Morsi, Appl. Catal. 61 (1990) 283. ger, C. Lamy and R.O. Lezna, [34] H. Hitmi, E.-M. Belgsir, J.-M. Le Electrochim. Acta 39 (1994) 407. [35] B. Beden, C. Lamy, A. Bewick and K. Kunimatsu, J. Electroanal. Chem. 121 (1981) 343. [36] K. Kunimatsu, J. Electroanal. Chem. 140 (1982) 205. ger, [37] B. Beden, F. Hahn, S. Juanto, C. Lamy and J.-M. Le J. Electroanal. Chem. 225 (1987) 215. [38] M. Watanabe and S. Motoo, J. Electroanal. Chem. 60 (1975) 275. [39] S.C. Chang, L.W. Leung and M.J. Weaver, J. Phys. Chem. 94 (1990) 6013. [40] T. Iwasita, B. Rasch, E. Cattaneo and W. Vielstich, Electrochim. Acta 34 (1989) 1073. [41] B. Rasch and T. Iwasita, Electrochim. Acta 35 (1990) 989. [42] P. Liu, A. Logadottir and J.K. Nrskov, Electrochim. Acta 48 (2003) 3731.

You might also like