Reaction intermediates of ethanol electro-oxidation on platinum investigated by SFG spectroscopy

Journal of Catalysis 302 (2013) 67–82

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis

journal homepage: w ww.elsevier.com/locate/jcat

Reaction intermediates of ethanol electro-oxidation on platinum

investigated by SFG spectroscopy

Janaina F. Gomes ^a, Kleber Bergamaski ^b,1, Melissa F.S. Pinto ^b,2, Paulo B. Miranda ^b,

⇑

^aInstituto de Química de São Carlos, Universidade de São Paulo, C.P. 780, São Carlos, SP 13560-970, Brazil

^bInstituto de Física de São Carlos, Universidade de São Paulo, C.P. 369, São Carlos, SP 13560-970, Brazil

a r t i c l e i n f o

a b s t r a c t

Article history:

Although electrochemical oxidation of simple organic molecules on metal catalysts is the basic ingredient

of fuel cells, which have great technological potential as a renewable source of electrical energy, the

detailed reaction mechanisms are in most cases not completely understood. Here, we investigate the eth-

anol–platinum interface in acidic aqueous solution using infrared–visible sum frequency generation

(SFG) spectroscopy and theoretical calculations of vibrational spectra in order to identify the intermedi-

ates present during the electro-oxidation of ethanol. The complex vibrational spectrum in the ﬁngerprint

region imply on the coexistence of several adsorbates. Based on spectra in ultra-high-vacuum (UHV) and

electrochemical environment from the literature and our density functional theory (DFT) calculations of

vibrational spectra, new adsorbed intermediates, never before observed with conventional infrared (IR)

Received 16 November 2012

Revised 25 February 2013

Accepted 27 February 2013

Keywords:

SFG spectroscopy

DFT

Electro-oxidation

Ethanol

Platinum

spectroscopy, are proposed here:

g

²-acetaldehyde,

g

²-acetyl, ethylidyne, monodentate acetate, methoxy,

tertiary methanol derivative, COH residue,

g

²-formaldehyde, mono and bidentate formate, CH₃and CH₂

Adsorbed intermediates

residues. In addition, we present new evidences for an ethoxy intermediate, a secondary ethanol deriv-

ative and an acetyl species, and we conﬁrm the presence of previously observed adsorbates: a tertiary

ethanol derivative, bidentate acetate, and CO_ad. These results indicate that the platinum surface is much

more reactive, and the reaction mechanism for ethanol electro-oxidation is considerably more complex

than previously considered. This might be also true for many other molecule-catalyst systems.

1. Introduction

permeability through polymeric membranes and also for leaving

less toxic by-products [2]. However, before they become widely

Electrochemical oxidation of simple organic molecules on metal

catalysts is at the heart of fuel cells, which have great technological

potential as a renewable source of electrical energy. Direct alcohol

fuel cells, especially those using ethanol as a fuel, are becoming a

very attractive alternative energy source for fossil fuels due to sev-

eral advantages: (i) easier and safer transportation and storage of

liquid fuels when compared to gases; (ii) the theoretical mass en-

ergy density of alcohols such as methanol and ethanol is high and

comparable to gasoline [1,2], and the thermodynamic efﬁciency of

fuel cells is much higher than internal combustion engines; (iii)

ethanol can be mass-produced by renewable sources, is less pollu-

tant than fossil fuels, and can share the available gasoline distribu-

tion infrastructure [3]; (iv) ethanol is a good candidate for

polymeric electrolyte membrane fuel cell (PEMFC) due to its low

available, a few technological issues need to be overcome. In regard

to the electrochemical reaction, the main problems that need to be

addressed are the high overpotential for the oxidation reaction due

to the presence of poisoning adsorbed intermediates at low poten-

tials and the formation of partially oxidized products. The solution

to them necessarily demands a thorough understanding of the de-

tailed reaction mechanisms, which are in most cases not com-

pletely understood. Therefore, the possible application of ethanol

as a fuel for direct alcohol fuel cells continues to motivate research

aimed at developing a mechanistic understanding of the electro-

oxidation of ethanol.

Fundamental studies of ethanol electro-oxidation in acid media

were mostly performed on platinum and its alloys [4–8]. Among

various Pt-based binary catalysts, PtASn has been reported as the

most effective for the electro-oxidation of ethanol [4,9]. Palladium

has been studied less actively in acid media due its relatively low

performance. On the other hand, in alkaline media, palladium ap-

pears to be quite active for ethanol oxidation [10]. Palladium-based

alloys in alkaline media have also received signiﬁcant attention

[11], even though the ethanol oxidation on a pure palladium elec-

trode has not yet been fully understood.

⇑

Corresponding author.

E-mail address: miranda@ifsc.usp.br (P.B. Miranda).

Current address: Departamento de Química, Universidade Federal de Sergipe,

1

Rod. Marechal Randon, s/n, São Cristóvão, SE 49100-000, Brazil.

2

Current address: Departamento de Física – ICEB, Universidade Federal de Ouro

Preto, Campus Morro do Cruzeiro, Ouro Preto, MG 35400-000, Brazil.

http://dx.doi.org/10.1016/j.jcat.2013.02.024

68

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

Concerning the ethanol oxidation on Pt electrodes, although it is

detection and identiﬁcation of other species, including those previ-

ously suggested by IR spectroscopy, is still lacking.

well-known that it follows parallel reaction pathways leading to

acetaldehyde, acetic acid and CO₂[9,12–15], the mechanism of this

reaction remains unclear. There is not yet a consensus concerning

the adsorbed intermediates that lead to the formation of these

products. From Differential Electrochemical Mass Spectrometry

(DEMS) investigations, it was concluded that the CO₂can originate

from either the alcohol or the methyl groups of the ethanol [12,16].

However, the available data related to the dissociation of the CAC

bond are contradictory. On one hand, Willsau and Heitbaum [12]

suggested that the adsorbed intermediate contains the intact

CAC bond of ethanol. On the other hand, Bittinscattaneo et al.

[16] proposed that the ethanol adsorbs mainly dissociatively in

form of one carbon atom species. In 1994, the Fourier Transform

Infrared (FTIR) and DEMS results published by Iwasita and Pastor

[17] showed that the presence of an acetyl residue could not be

discarded. This partially agrees with the idea of Willsau and Heit-

baum [12]. Nevertheless, it was demonstrated that the acetyl was

not the only adsorbed species formed [17]. Adsorbed CO was

clearly evident. In addition, it was proposed that the secondary

alcohol derivative (PtACHOHACH₃) could be the precursor for

the formation of the adsorbed CO via an acetyl species. However,

In this work, we will present the ﬁrst SFG spectra of ethanol in

acidic medium at a polycrystalline platinum surface over a wide

spectral range (1000–3300 cm^ꢀ1). The main purpose of the present

work is to identify adsorbed intermediates of the ethanol oxidation

on Pt using SFG spectroscopy and DFT calculations of their vibra-

tional spectra. In contrast to previous SFG experiments, our results

indicate the coexistence of several adsorbates. In addition to pre-

senting new evidences for previously suggested intermediates

and conﬁrming the presence of adsorbates already observed by

IR and SFG spectroscopies, we identify new adsorbed intermedi-

ates. As an aid to the reader, Fig. 9 (at the end of Section 3.4) dis-

plays the structure of potential adsorbates that will be discussed

in the text.

2. Methodology

2.1. Experimental section

Similar to previous work [25,28], the SFG experiments were

performed in a three electrode spectro-electrochemical cell with

an IR and visible transparent 2 mm CaF₂window attached to the

top of the cell. The working electrode was a polished polycrystal-

line platinum disk 12 mm in diameter. A platinized platinum wire

and a reversible hydrogen electrode were used as the counter and

reference electrodes, respectively. A movable piston supports the

working electrode. A platinum wire connected to the working elec-

trode passes through the piston and keeps the electric contact. Be-

fore each experiment, the working electrode was annealed in a H₂

ﬂame for approximately 15 min, cooled to room temperature in an

inert N₂atmosphere and then protected with a drop of water to

prevent contamination before being transferred to the spectro-

electrochemical cell. The working electrode was introduced at

open circuit potential to the spectro-electrochemical cell contain-

ing 0.1 M ethanol and 0.5 M HClO₄and subsequently polarized at

0.05 V. SFG spectra were taken in the region between 1000 cm^ꢀ1

and 3300 cm^ꢀ1, at potentials increasing from 0.05 V to 1.1 V vs.

RHE (0.05 V steps), waiting about 1 min after setting each poten-

tial. The complete range of wavenumbers was scanned in four dif-

ferent experiments. During the acquisition of the SFG spectra, the

working electrode was pressed against the CaF₂window to obtain

a thin liquid ﬁlm in order to minimize the absorption of the infra-

red beam by the electrolytic solution. The maximum attenuation of

the reﬂected IR beam in the spectral range from 1000 to 2100 cm^ꢀ1

occurs at 1650 cm^ꢀ1, and it varied from 60% to 90% (crossing the

liquid ﬁlm twice), which corresponds to a single pass attenuation

of ꢁ37–68%. The SFG spectra were not corrected for this effect.

From these measurements, the thickness of the thin electrolyte

no direct evidence of the m(CAOH) band of a secondary alcohol

[18] was observed in that work [17]. Based on Electrochemically

Modulated Infrared Reﬂectance Spectroscopy (EMIRS), different

models for the adsorbed species were discussed, including the

adsorption of an ethoxy species [19]. Acetaldehyde [20], acetic acid

[21], and acetyl [22] were also proposed as adsorbed intermediates

of ethanol oxidation.

Undoubtedly, the studies described above have contributed

enormously to the elucidation of the complex mechanism of the

ethanol oxidation on platinum. However, it is important to empha-

size that in all these studies, the electro-oxidation of ethanol was

investigated by linear spectroscopic techniques. The disadvantage

of using this kind of spectroscopic probe is that it is impossible

to completely discriminate features from adsorbates and contribu-

tions from bulk species. Eliminating the contribution from the bulk

is critical when the adsorbed intermediates and the bulk products

contain the same functional groups, as for the partially oxidized

products acetaldehyde and acetic acid. In such cases, it is difﬁcult

to differentiate between bulk and adsorbed species and hence dif-

ﬁcult to make unambiguous identiﬁcations of adsorbed intermedi-

ates. Sum-Frequency Generation (SFG) spectroscopy can be used as

a powerful tool to obtain vibrational spectra of adsorbed species

without contribution from the bulk [23,24]. Since SFG is a sec-

ond-order non-linear optical process that occurs only at media

where the inversion symmetry is broken, it is allowed only at the

interface between two centrosymmetric materials, making it pos-

sible to detect adsorbed species without any contribution from

the bulk. Because of this major advantage, SFG spectroscopy has al-

ready been used by several groups to investigate the electrochem-

ical interface, notably on Pt electrodes. However, these studies are

usually limited to vibrations with high dipole moment and polariz-

ability derivatives, such as adsorbed CO or CN [25–32]. One excep-

tion is the formation of PtAOH detected by SFG spectroscopy [33].

Very recently, Kutz et al. [34] have investigated ethanol electro-

oxidation on polycrystalline Pt using SFG spectroscopy. They only

detect adsorbed CO and acetate ions (and (bi)sulfate ions, for the

case of sulfuric acid supporting electrolyte). Furthermore, using

¹³C-labeled ethanol, they demonstrate the cleavage of the CAC

bond upon electrochemical adsorption. It should be also men-

tioned the very recent SFG study by Braunschweig et al. [35] of ace-

tic acid electrochemical adsorption onto Pt and Au electrodes,

where they also found adsorbed acetate ions, as in Ref. [34].

Although these studies have further advanced our understanding

of the electro-oxidation of simple molecules on Pt electrodes, the

layer is estimated to be 2.65 0.75 lm. However, the stronger IR

absorption of water in the OH stretch range attenuated most of

the IR energy, making the SFG spectra unreliable above

3200 cm^ꢀ1(or between 2350 cm^ꢀ1and 2700 cm^ꢀ1, for experi-

ments with deuterated water). From 1250 cm^ꢀ1to 1000 cm^ꢀ1

,

the IR transmittance decreases from 95% to 60% due to bulk IR

absorption of the 2 mm CaF₂window. Our results were also not

corrected for this effect.

The SFG spectrometer (Ekspla, Lithuania) consists of a ﬂash-

lamp-pumped Nd⁺³:YAG laser that generates 25 ps pulses at

1064 nm with a repetition rate of 20 Hz. A 532 nm beam is pro-

duced by second harmonic generation (SHG) and a tunable IR beam

(2.5–10

l

m, bandwidth ꢁ3 cm^ꢀ1) is produced by an optical para-

metric generator/ampliﬁer (OPG/OPA) followed by a difference-

frequency generation (DFG) stage. Before the SFG experiments,

the alignment of beams (temporal and spatial overlap of beams

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

69

on the sample and SFG signal detection) was optimized on a refer-

ence sample of intense nonresonant signal (quartz or ZnS), which

was then replaced by the sample of interest. The p-polarized

532 nm and IR beams overlapped temporally and spatially on the

sample with spot sizes and angles of incidence of ꢁ1.0 mm and

ꢁ0.5 mm, and 61° and 55°, respectively. The p-polarized SFG signal

of the platinum–solution interface was normalized to the IR pulse

energy, which was monitored before reaching the sample and re-

corded simultaneously with data acquisition. Each data point in

the SFG spectrum is an average of 125 shots. The energies of visible

for heavy elements. This basis was chosen for this study because

it incorporates relativistic effects for Pt (and other heavy metals),

thus describing more efﬁciently their properties, and it is also com-

patible with the elements of ethanol derivatives (H, C, O). In addi-

tion, reports in the literature on platinum complexes using B3LYP/

LanL2DZ level showed that it is able to reproduce with satisfactory

agreement spectroscopic experimental data [43–46].

Infrared and Raman spectra calculations were carried out for all

simulated adsorbates, since the SFG selection rules require that the

vibrational modes must be active in both spectra. Despite the

broad range of calculated spectra (up to 3500 cm^ꢀ1), we will dis-

cuss only the range from 1000 to 1500 cm^ꢀ1, which is the most

prominent in the experimental spectra and is the most sensitive

to molecular structure (vibrational ﬁngerprint region).

and IR pulses were about 40 lJ and 50 lJ, respectively.

2.2. Theoretical section

We have used the density functional theory (DFT) approach to

perform calculations of vibrational spectra to support peak assign-

ments of the experimental SFG spectra. For that purpose, we have

compared the calculated vibrational spectra of several potential

intermediates adsorbed on a Pt cluster to the observed SFG reso-

nances. This demanded a high accuracy in the calculation of vibra-

tional frequencies, since the SFG peaks were quite closely spaced

(ꢁ20–50 cm^ꢀ1). The Pt cluster was constructed using the crystallo-

graphic data reported by Waseda et al. [36] for FCC platinum with

space group Fm-3m, with cell parameters a = b = c = 392.42 pm and

3. Results and discussion

3.1. Vibrational spectra of adsorbates

Before exploring the vibrational spectra of adsorbates of the

ethanol oxidation on Pt in acidic medium, the cyclic voltammo-

gram of the polycrystalline platinum electrode in 0.5 mol L^ꢀ1

HClO₄+ 0.1 mol L^ꢀ1ethanol is introduced in Fig. 2. In the positive-

going scan, the oxidation of ethanol begins close to 0.40–0.50 V

and the current peaks at 0.90 V and 1.30 V. In the negative-going

sweep, the current starts to increase at 0.90 V, reaches a maximum

at 0.65 V, and then decreases to nearly zero at 0.40 V. These results

are in accord with those previously observed for 0.1 mol L^ꢀ1

ethanol over polycrystalline platinum in acidic medium [20].

Fig. 3 shows the SFG spectra (from 1000 cm^ꢀ1to 3300 cm^ꢀ1) of

the 0.5 mol L^ꢀ1HClO₄+ 0.1 mol L^ꢀ1ethanol–Pt interface at 0.30 V

(ﬁlled circles), below the onset of the ﬁrst oxidation current peak.

In these spectra, different features assigned to adsorbed species

can be clearly observed. The most pronounced features are at

a

= b = c = 90°. In this conﬁguration, the distance between the Pt

atoms is 277.5 pm [36]. The cluster was arranged to model the

Pt(111) surface, with the ﬁrst layer consisting of ten platinum

atoms and the second layer having six platinum atoms, as shown

in Fig. 1. The third layer was discarded because it has minimal

inﬂuence on properties under study, as well as to minimize the

computational effort in electronic structure calculations. The plat-

inum cluster was maintained ﬁxed and we carried out fully relaxed

geometry optimizations of the ethanol-derived adsorbates, which

were attached to the central Pt atom of the ﬁrst layer. Calculations

also showed that, within the extent and arrangement of our

Pt(111) model, the position of a second bond for bidentate adsor-

bate has minimal effect on spectroscopic assignments.

1056 cm^ꢀ1

,

1084 cm^ꢀ1

,

1122 cm^ꢀ1

,

1168 cm^ꢀ1

,

1214 cm^ꢀ1

,

1270 cm^ꢀ1, 1408 cm^ꢀ1, 1432 cm^ꢀ1, and 2056 cm^ꢀ1. There are also

three broad features, one between 1310 cm^ꢀ1and 1380 cm^ꢀ1, an-

other between 1450 cm^ꢀ1and 1625 cm^ꢀ1, and a third around

1736 cm^ꢀ1. Based on spectra taken at other potentials (data shown

in Fig. 5, Section 3.2), these features are most likely comprised of

overlapping bands, at 1330 cm^ꢀ1, 1348 cm^ꢀ1and 1370 cm^ꢀ1for

the ﬁrst, and at 1475 cm^ꢀ1, 1496 cm^ꢀ1and 1546 cm^ꢀ1for the sec-

ond. An overlapping peak at 1288 cm^ꢀ1and a weak feature at

1030 cm^ꢀ1can also be distinguished. The SFG spectra from

2100 cm^ꢀ1to 3300 cm^ꢀ1is relatively more complex to analyze,

The potential adsorbates investigated were the following: CO,

secondary and tertiary ethanol derivative, acetyl, bidentate acetate,

ethoxy and

g

²-acetaldehyde. The interactions between these

derivatives of ethanol and the Pt(111) model surface have been

studied using density functional theory (DFT) approach, with the

hybrid functional B3LYP associated with effective core potential

(ECP) type basis set [37], using the GAUSSIAN 03 program [38].

In the B3LYP hybrid functional scheme, the nonlocal Hartree–Fock

(HF) approach is mixed into the energy functional of the general-

ized gradient approximation (GGA), in which are included three

parameters of Becke [39–41] for exchange energy estimation and

the Lee–Yang–Parr correlation energy [42]. The Lan2DZ basis in-

cludes Dunning–Huzinaga full double zeta (DZ) basis functions

for the ﬁrst row and Los Alamos effective core potentials (ECPs)

2.0

1.5

1.0

0.5

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

E (V) vs. RHE

Fig. 2. Cyclic voltammogram of a polished polycrystalline platinum electrode in

0.5 mol L^ꢀ1HClO₄(dashed line) and 0.5 mol L^ꢀ1HClO₄+ 0.1 mol L^ꢀ1ethanol (solid

Fig. 1. Side view of the Pt(111) cluster model.

line); scan rate = 0.10 V s^ꢀ1

.

70

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

since it contains weak and broad features interfering with a few

narrow peaks. Among them, we clearly observe a broad band from

2650 cm^ꢀ1to 3200 cm^ꢀ1, which may be assigned to OH stretches,

as will be discussed below. In order to verify that the observed

bands are indeed vibrations of adsorbates on Pt derived from eth-

anol, we have scanned the same frequency ranges with a solution

containing only the electrolyte (0.5 mol L^ꢀ1HClO₄). As seen in

Fig. 3 (empty circles), most of the SFG peaks vanish, except for a

prominent peak at 1408 cm^ꢀ1, a band between 1445 cm^ꢀ1and

1490 cm^ꢀ1, and weak bands in the region 2650–3300 cm^ꢀ1. Since

there is no organic molecule in solution, these resonances must

come from water or the electrolyte. The latter bands may be due

to OH stretches of water molecules or the hydronium ion (H₃O⁺)

[47–50] interacting with the Pt surface by hydrogen bonding (not

chemisorbed). However, the peak at 1408 cm^ꢀ1and the broad fea-

ture at ꢁ1475 cm^ꢀ1cannot be attributed to perchlorate, water, or

hydronium. We will show below that they are not due to molecular

adsorbates on Pt. Therefore, except from these two spurious peaks,

all other features of the SFG spectra in Fig. 3 are due to adsorbed

intermediates of the electro-oxidation of ethanol on Pt. This sur-

prising result implies that there are a lot more molecular adsor-

bates than previously thought, revealing a very complex reaction

mechanism and making a deﬁnite assignment of all vibrational

bands a formidable task. After presenting the dependence of the

SFG spectra on electrochemical potential in Section 3.2 and SFG

spectra with isotopically labeled reagents in Section 3.3, we will

discuss in more detail a tentative assignment of these peaks in Sec-

tion 3.4. For now, it sufﬁces to say that peaks in the ﬁngerprint

range (from 1000 cm^ꢀ1to 1500 cm^ꢀ1) are due to coupled vibra-

tions of the molecular skeleton of adsorbates, such as CAC and

CAO stretches, and CH, OH, and CACAO angular deformations

[18]. A broad band at ꢁ1700 cm^ꢀ1is due to the hydrogen bonded

C@O stretch of adsorbates [9,15], while the intense peak at

2056 cm^ꢀ1is the well known C„O stretch of linearly bonded CO

on Pt [13,16,19].

Fig. 3. SFG spectra (polarization combination PPP) of the 0.5 mol L^ꢀ1HClO₄(empty

circles) and 0.5 mol L^ꢀ1HClO₄+ 0.1 mol L^ꢀ1ethanol (ﬁlled circles)–polycrystalline

platinum interface at 0.30 V vs. RHE.

Since the peaks in the frequency range from 1000 cm^ꢀ1to

1500 cm^ꢀ1are so numerous and appear to be regularly spaced, at

ﬁrst one could imagine that they could result from interference ef-

fects in the thin liquid ﬁlm sandwiched between two reﬂective sur-

faces: the CaF₂window and the Pt electrode. As mentioned above,

this can be ruled out by the spectrum for the pure electrolyte/Pt

interface shown in Fig. 3 (empty circles), where most of the reso-

nances disappear, except for the narrow peak at 1408 cm^ꢀ1and a

broad feature at ꢁ1475 cm^ꢀ1. Additional evidence for the absence

of interference effects in the thin liquid ﬁlm is the comparison of

CaF₂/ ethanol + HClO₄/ Pt

A

the SFG spectrum taken with a 15 lm thick Teﬂon spacer between

the window and the Pt electrode with that SFG spectrum taken

without any spacer [51]. It is observed that they present similar

peak positions, although with considerably reduced SFG intensities

CaF₂/ HClO₄/ Pt

B

for the spectrum taken with a 15 lm thick Teﬂon spacer due to IR

absorption by water (broad OH bending at ꢁ1650 cm^ꢀ1and OH

libration below 900 cm^ꢀ1) in this thicker electrolyte layer. If the

spectra resulted from optical interference effects, the peak spacing

C

CaF₂/ air / Pt

should have been Dm = 1/(2ndcosc), where c is the beam angle with

respect to the surface normal within the liquid ﬁlm, n is the elec-

air / Pt

D

trolyte refractive index, and d is the liquid layer thickness (in

air / CaF₂

E

cm). Since the inverse relation of

served, interference cannot explain the observed SFG spectra be-

low 1500 cm^ꢀ1

Dm and spacer thickness is not ob-

1000

1100

1200

1300

1400

1500

.

We will now turn to the origin of the SFG bands at 1408 cm^ꢀ1

and ꢁ1475 cm^ꢀ1. As will be shown in Section 3.2, the intensity of

these signals remains constant as function of the applied potential.

This behavior is quite different than that observed for all other

bands in the SFG spectra shown in Fig. 3. In order to investigate

whether these peaks are originated at the CaF₂/solution interface,

Fig. 4 displays the SFG spectra for the following interfaces: CaF₂/

IR wavenumber (cm^-1)

Fig. 4. SFG spectra from 1000 cm^ꢀ1to 1500 cm^ꢀ1of the following interfaces: CaF₂/

0.1 mol L^ꢀ1ethanol + 0.5 mol L^ꢀ1HClO₄/polycrystalline platinum at 0.05 V vs. RHE

(A), CaF₂/0.5 mol L^ꢀ1HClO₄/polycrystalline platinum at 0.05 V vs. RHE (B), CaF₂/air/

polycrystalline platinum (C), air/polycrystalline platinum (D) and air/CaF₂(E). The

SFG spectra A, B, C, D, and E have the same y-axis scale, and they are offset along

this axis for clarity.

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

71

0.05 V

*

0.05 V

(l)

0.3

0.0

(j)

(i)

(g)

(f)

0.2

0.0

(b)

(e)

(a)

(k)

*

(h)

(d)

(c)

1400 1500 1600 1700 1800

1000 1100 1200 1300 1400 1500

0.3

0.3 V

0.0

0.2

0.0

1400 1500 1600 1700 1800

0.3

0.5 V

1000 1100 1200 1300 1400 1500

0.0

1400 1500 1600 1700 1800

0.6 V

0.2

0.0

0.3

0.7 V

0.0

1000 1100 1200 1300 1400 1500

1400 1500 1600 1700 1800

0.9 V

0.2

0.0

0.3

0.9 V

0.0

1000 1100 1200 1300 1400 1500

IR wavenumber (cm^-1)

1400 1500 1600 1700 1800

IR wavenumber (cm^-1)

0.4

0.2

0.0

0.05 V

3

0.1 V

2000 2200 2400 2600 2800 3000 3200 3400

0.4

0

3

1700 1800 1900 2000 2100 2200

0.3 V

0.2

0.0

0.3 V

2000 2200 2400 2600 2800 3000 3200 3400

0.4

0

3

1700 1800 1900 2000 2100 2200

0.5 V

0.2

0.0

0.5 V

0

3

2000 2200 2400 2600 2800 3000 3200 3400

0.4

1700 1800 1900 2000 2100 2200

0.7 V

0.2

0.0

0

3

2000 2200 2400 2600 2800 3000 3200 3400

0.4

1700 1800 1900 2000 2100 2200

0.9 V

0.8 V

0.2

0.0

0

2000 2200 2400 2600 2800 3000 3200 3400

IR wavenumber (cm^-1)

1700 1800 1900 2000 2100 2200

IR wavenumber (cm^-1)

Fig. 5. SFG spectra (1000–3300 cm^ꢀ1) of the (0.1 mol L^ꢀ1ethanol + 0.5 mol L^ꢀ1HClO₄)/Pt interface as a function of the applied potential. The potential is indicated in the SFG

spectra. Peaks labeled with letters are listed in the text. ^ꢂThe SFG peak at 1408 cm^ꢀ1and the SFG band at ꢁ1475 cm^ꢀ1are not related to any molecular vibrational transition

(see Section 3.1).

(0.1 mol L^ꢀ1ethanol + 0.5 mol L^ꢀ1HClO₄)/Pt at 0.05 V vs. RHE (A),

CaF₂/0.5 mol L^ꢀ1HClO₄/Pt at 0.05 V vs. RHE (B), CaF₂/air/Pt (C),

air/Pt (D), and air/CaF₂(E). It is important to emphasize that the

acquisition of the SFG spectra for the CaF₂/air/Pt, air/Pt, and air/

CaF₂interfaces was performed in this sequence, on the same com-

ponents, and immediately after a rigorous cleaning of the CaF₂

window (with KMnO₄/KOH and H₂O₂/H₂SO₄solutions, followed

by copious rinsing with Milli – Q water) and H₂-ﬂame-annealing

the Pt electrode. For the air/CaF₂interface, the visible pulse energy

not due to any organic contamination and are also not originated

from molecular resonances of adsorbates from the solution. In-

stead, they only appear when the CaF₂window is in close contact

with the Pt electrode. A plausible explanation would be that these

bands are due to adsorbed species on the CaF₂window, most likely

the OH bending mode of surface CaOH groups, since hydroxyapa-

tite – Ca₅(PO₄)₃(OH) – presents a vibration just above 1400 cm^ꢀ1

[52]. These vibrations would only become apparent when in con-

tact with a rough Pt surface due to local enhancement of the opti-

cal ﬁeld on the metal surface asperities, which would signiﬁcantly

increase the SFG sensitivity. However, this explanation is also un-

likely, since the wavenumber of this band did not redshift in pres-

ence of deuterated reagents (as shown in Fig. 7, Section 3.3). The

origin of these signals at 1408 cm^ꢀ1and ꢁ1475 cm^ꢀ1is a very

interesting question and clearly more experimental and theoretical

work is needed to clarify this issue. Here, we will not investigate

this further, since this is not directly relevant to the present study.

A more detailed discussion of possible origins of these features can

was increased from 40 to 900 lJ to increase the experimental sen-

sitivity. It can be noted in Fig. 4 that the peak at ꢁ1408 cm^ꢀ1and

the broad band between 1445 cm^ꢀ1and 1490 cm^ꢀ1are present

in the SFG spectra of the CaF₂/ethanol + HClO₄/Pt, CaF₂/HClO₄/Pt,

and CaF₂/air/Pt interfaces. Nothing is evidenced in the spectra of

the air/Pt and air/CaF₂interfaces.

The presence of these bands in the spectrum for the CaF₂/air/Pt

(in the absence of the electrolytic solution!) and their absence for

the air/Pt and air/CaF₂interfaces clearly show that these signals are

72

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

potential steps from 0.05 V up to 0.90 V (0.05 V steps) in the wave-

number range between 1000 cm^ꢀ1and 3300 cm^ꢀ1

(a)

(b)

.

The SFG spectrum taken at 0.05 V presents more clearly some

bands in the wavenumber range between 1000 cm^ꢀ1and

1500 cm^ꢀ1that are less evident in the spectra taken at relatively

more positive potentials. Particularly in the wavenumber region

between 1000 cm^ꢀ1and 1440 cm^ꢀ1, SFG bands are observed at:

(a) 1030_ꢀc₁m^ꢀ1, (b) 1056 cm^ꢀ1, (c) 1084 cm^ꢀ1, (d) 1122 cm^ꢀ1, (e)

1168 cm

, , , , (i)

(f) 1214 cm^ꢀ1(g) 1270 cm^ꢀ1(h) 1288 cm^ꢀ1

1330 cm^ꢀ1, (j) 1348 cm^ꢀ1, (k) 1370 cm^ꢀ1, and (l) 1432 cm^ꢀ1

.

In general, it can be noted in Fig. 5 that the intensity of the

majority of the bands evidenced between 1000 cm^ꢀ1and

1800 cm^ꢀ1and between 2100 cm^ꢀ1and 3300 cm^ꢀ1decreases with

increasing potential from 0.05 V to 0.90 V vs. RHE. Differently, the

intensity of the band assigned to the C„O stretching mode of the

linear-bonded CO molecules (ꢁ2050 cm^ꢀ1) gradually increases un-

til 0.50 V and then starts to decrease. Most of the SFG signals in the

wavenumber range from 1000 cm^ꢀ1to 3300 cm^ꢀ1are signiﬁcantly

reduced at 0.90 V. As previously mentioned in Section 3.1, the

intensity of the peak at 1408 cm^ꢀ1and the broad band at

ꢁ1475 cm^ꢀ1(not from adsorbates) remains nearly constant with

the increasing potential. The broad bands at ꢁ1500 cm^ꢀ1and

ꢁ1740 cm^ꢀ1are also reduced with increasing potential, but are

still noticeable at 0.9 V.

It is noteworthy to mention that the frequency of the C„O

stretching vibration of the linear-bonded CO molecules noticeably

blueshifted as function of the increasing potential from 0.1 V to

0.9 V as a result of the vibrational Stark effect [53,54], back-dona-

tion mechanism [55], and dipole–dipole coupling effect [28,56].

Due to the high complexity of the SFG spectra ranging from

1000 cm^ꢀ1to 1800 cm^ꢀ1and from 2100 cm^ꢀ1to 3300 cm^ꢀ1, it is

unclear whether the bands other than that at ꢁ2050 cm^ꢀ1are

shifting with the electric ﬁeld, although any shift may be at most

a few wavenumbers.

In order to quantitatively evaluate the effect of the applied po-

tential on the SFG resonances and the nonresonant background

from the Pt electrode, the SFG spectra between 1000 cm^ꢀ1and

1500 cm^ꢀ1and between 1700 cm^ꢀ1and 2200 cm^ꢀ1of the

(0.1 mol L^ꢀ1ethanol + 0.5 mol L^ꢀ1HClO₄)/Pt interface (data pre-

sented in Fig. 5) were ﬁtted to the standard Lorentzian model:

Fig. 6. (a) Normalized amplitudes of the nonresonant background and SFG

resonances between 1000 cm^ꢀ1and 1500 cm^ꢀ1, and from 1700 to 2200 cm^ꢀ1(from

the theoretical ﬁts of the SFG spectra presented in Fig. 5) as a function of the applied

potential. (b) Amplitudes of the various SFG resonances at low potentials.

0.25

1214 cm^-1

1190 cm^-1

0.20

0.15

0.10

0.05

0.00

0.05 V

ꢀ

2

ꢀ

X

A_q

ꢀ

I_SFG/ Be^i/

þ

ð1Þ

ꢀ

x_IR

ꢀ

x_qþ iC_q

q

0.80 V

400

where A_q,

x

_q, and C_qare the strength, resonance frequency and line

width of the qth vibrational mode, while / and B are the phase and

amplitude of the nonresonant contribution, respectively. In order to

reduce the number of adjustable parameters and reduce their

uncertainty, all SFG spectra in a given frequency range were ﬁtted

0

200

600

800

1000

time (s)

simultaneously with the same /,

x_q, and C_q(only B and A_qwere al-

Fig. 7. Intensity of the SFG signals at 1190 cm^ꢀ1(gray line + gray circles) and

1214 cm^ꢀ1(black line + black squares) as function of time. The red-dashed lines

indicate the exact time of the potential step from 0.05 V to 0.80 V, and back to

0.05 V. (For interpretation of the references to color in this ﬁgure legend, the reader

is referred to the web version of this article.)

lowed to vary with potential, except for the CO peak at ꢁ2050 cm^ꢀ1

,

which also has a potential-dependent frequency). The SFG spectra

between 1700 cm^ꢀ1and 2200 cm^ꢀ1were ﬁtted with two resonant

signals at ꢁ1750 and ꢁ2050 cm^ꢀ1. For the frequency range between

1000 cm^ꢀ1and 1500 cm^ꢀ1, the data were ﬁrst corrected for the

absorption of the CaF₂window (see Section 2.1), since this would

make it difﬁcult to ﬁt with a constant nonresonant background plus

be found elsewhere [51]. Hereafter, we will no longer mention

these bands in the discussion, since they are not related to adsor-

bates on the Pt surface.

resonances. It was then ﬁtted with ten different peaks at 1033 cm^ꢀ1

1056 cm^ꢀ11086 cm^ꢀ11127 cm^ꢀ11169 cm^ꢀ11218 cm^ꢀ1

,

1271 cm^ꢀ1, 1344 cm^ꢀ1, 1409 cm^ꢀ1, and 1457 cm^ꢀ1. Some of these

peaks have contributions from more than one resonance, but they

were considered to be a single peak to simplify the theoretical ﬁts,

therefore reducing parameter uncertainty. However, peak strengths

in SFG spectroscopy are proportional to ðA_q=C_qÞ. As a result of this,

the relative amplitudes A_qfrom the theoretical ﬁts of the SFG

resonances between 1000 cm^ꢀ1and 1500 cm^ꢀ1are not accurate,

3.2. Effect of the applied potential on the SFG spectra

Fig. 5 shows a set of SFG spectra for the (0.1 mol L^ꢀ1ethanol

+0.5 mol L^ꢀ1HClO₄)/Pt interface as a function of the applied poten-

tial. The SFG spectra were taken during the ﬁrst series of increasing

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

73

because they also depend on the bandwidth C_q. The amplitude vs.

potential behavior is however truthful within the error bars. To ﬁt

the two broad bands in the frequency range between 1450 cm^ꢀ1

and 1800 cm^ꢀ1, it was necessary to include four broad and overlap-

ping peaks, leading to a very large parameter uncertainty. There-

fore, they are unreliable and will not be presented or discussed

further. Fig. 6a presents the amplitudes of the SFG resonances be-

tween 1000 cm^ꢀ1and 1500 cm^ꢀ1, at ꢁ1750 cm^ꢀ1and the CO peak

at ꢁ2050 cm^ꢀ1as a function of the applied potential, normalized

to their values at low potentials. Since the peaks from 1000 cm^ꢀ1

to 1400 cm^ꢀ1have the same potential dependence within the ﬁt-

ting uncertainty, they are represented with an average curve la-

oxidation of the adsorbed species begins (above 0.50 V). In this

way, the reaction would occur dynamically and the intensity of

the bands would remain constant as function of the increasing po-

tential. In contrast to this expected behavior, we observe a progres-

sive decrease in the band intensities. This result could be

interpreted in three different ways:

(i) The gradual decrease in the SFG band intensities in the oxi-

dation region could be explained by the consumption (at rel-

atively high potentials) of ethanol in the thin layer of

electrolytic solution between the platinum electrode and

the CaF₂window, which would then not be available to

replenish the adsorbed species after reaction. In this picture,

we consider that the rate of ethanol transport from the bulk

to the thin layer of electrolytic solution is negligible in com-

parison with the rate of oxidation of the adsorbed species,

leading to the ethanol exhaustion in the thin layer.

(ii) Another possibility is that the electrochemical surface reac-

tion rate is much faster than the diffusion rate of ethanol

from the thin layer of electrolyte to the electrode surface.

In this case, the steady state concentration of adsorbates

would be very low, being below the sensitivity limit of our

SFG spectra.

beled as ‘‘low

x’’. It also displays the potential dependence of the

nonresonant background obtained from the spectra between

1700 cm^ꢀ1and 2200 cm^ꢀ1, since it is determined more accurately

in that frequency range. Fig. 6b presents the A_qfor the various res-

onances at low potentials, to show their relative magnitudes.

The quantitative behavior of the resonant amplitudes as func-

tion of the applied potential (as seen in Fig. 6a) shows that the

intensity of the CO band increases up to about 0.50 V and then de-

creases, while the low frequency peaks initially stay constant, but

then decrease with increasing potential above 0.70 V vs. RHE. In

addition to that, it can be observed in Fig. 6 that the intensity of

the nonresonant background also decreases with increasing poten-

tial. However, the decrease in the nonresonant contribution is

more gradual than that of the low frequency resonant signals, indi-

cating that their potential dependence is not determined by the

nonresonant contribution, giving further support to the conclusion

that these multiple low frequency peaks are not due to interfer-

ence effects [51].

(iii) Alternatively, as the oxidation of the adsorbed intermediates

begins, these species are removed from the surface and the

free platinum sites are partially reoccupied by different spe-

cies that are not SFG active in the spectral range we have

explored. In this case, these species would poison the plati-

num surface and act as an ethanol oxidation inhibitor.

Surprisingly, the major part of the adsorbed intermediates of

the ethanol electro-oxidation reaction over platinum in acidic

medium is formed at 0.05 V. These results show that the platinum

surface is highly reactive, even at potential as low as 0.05 V. In the

potential range between 0.05 V and 0.40–0.50 V, the small varia-

tion of the faradaic current (Fig. 2) shows that adsorbed species

are not forming the ﬁnal products of the reaction. However, in this

potential region, there is a gradual increase in the amplitude for

the band related to the C„O stretching vibration of the linearly-

bonded CO molecules. This suggests that the CO layer may be

formed from the progressive consumption of part of the adsorbed

species formed at 0.05 V, although the reduction of amplitudes for

the low frequency SFG peaks (Fig. 6) is not obvious (given the large

uncertainties), perhaps because there are so many adsorbed spe-

cies that only a small fraction of each needs to be converted to

CO. Therefore, it seems that the adsorbed species formed at

0.05 V are partially transformed into CO with the increasing

potential.

Considering the scenario (i) above, we now estimate the time re-

quired to consume most of ethanol molecules in the thin layer vol-

ume by electrochemical oxidation at 0.80 V. The initial current

density in a chronoamperometry curve for potentials in the ethanol

oxidation peak (0.8 V) is of the order of 1 mA/cm². [34] Therefore,

the faradaic charge that passes through the working electrode (area

ꢁ1 cm²) during one second is ꢁ1 mC, which corresponds to

1.04 ꢃ 10^ꢀ8mol of electrons. According to Wang et al. [57], the

product yield of the 0.1 mol L^ꢀ1ethanol oxidation reaction on plat-

inum in acidic medium (over a full potential cycle) is equivalent to:

1.3%, 55% and 44% for carbon dioxide, acetaldehyde and acetic acid,

respectively. Considering these product yields and the number of

electrons produced in each reaction (12 e^ꢀ, 2 e^ꢀand 4 e^ꢀfor etha-

nol ? CO₂, ethanol ? acetaldehyde and ethanol ? acetic acid,

respectively), the average number of electrons produced per oxi-

dized ethanol molecule is 3. With that, 1.04 ꢃ 10^ꢀ8mol of electrons

will be produced in the electro-oxidation of 3.47 ꢃ 10^ꢀ9mol of eth-

anol molecules. For estimating the electrolytic solution volume, we

At around 0.40–0.50 V, we observe the onset of the faradaic cur-

rent (Fig. 2). Thus, from this potential, the adsorbed species are re-

moved from the surface, originating the reaction products. In

agreement with our results, Wang et al. [57] demonstrated on

the basis of Differential Electrochemical Mass Spectrometry

(DEMS) data that the formation of acetaldehyde (m/z = 29) begins

at 0.40 V vs. RHE, just at the onset potential of the faradaic current.

Moreover, they observed that acetaldehyde is produced in a wide

range of potentials, including the reserve cycle, whereas the carbon

dioxide (m/z = 22) is produced only in the region of potentials be-

tween 0.50 V and 0.90 V vs. RHE [57]. This is exactly the potential

take the thin layer thickness as 1 lm, which yields a volume of

1 ꢃ 10^ꢀ4cm³. At 0.1 mol L^ꢀ1, the amount of ethanol present in the

thin layer is 1 ꢃ 10^ꢀ8mol. Therefore, this crude estimate indicates

that it would take about 3 s to consume all the ethanol in the thin

layer of solution between the electrode and the CaF₂window. We

will describe below an experiment designed to investigate the time

evolution of the SFG spectra upon a potential step, which may ad-

dress this issue.

Fig. 7 shows the intensity of the SFG signals at 1190 cm^ꢀ1and

1214 cm^ꢀ1as function of time. The curves were acquired simulta-

neously, with the OPA tuned back and forth between the two fre-

quencies, and the time interval between two subsequent points in

each curve is about 3 s. In this experiment, the electrode potential

was changed from 0.05 V to 0.80 V and then, after almost 500 s,

from 0.80 V back to 0.05 V. The SFG signals at 1190 cm^ꢀ1and

1214 cm^ꢀ1correspond to a valley between two nearby resonances

and to one of the intense resonances shown in Fig. 5, respectively.

On one hand, the SFG signal at 1190 cm^ꢀ1serves as a guide for the

range where we observe a decrease in the CO peak at 2056 cm^ꢀ1

,

supporting the picture that CO₂is produced mostly from further

oxidation of adsorbed CO.

In the oxidation region (between 0.5 V and 0.90 V), we do not

observe a dynamic equilibrium between the adsorbed species

and the species in the solution. We could expect that the free plat-

inum sites would be replaced by other ethanol molecules as the

74

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

nonresonant signal of the platinum electrode at 0.05 V and 0.80 V.

On the other hand, the signal at 1214 cm^ꢀ1, related to adsorbed

intermediates of the ethanol oxidation over platinum, monitors

the lifetime of the adsorbed species when the potential is changed

from 0.05 V to 0.80 V and also the time required to re-establish the

layer of adsorbed molecules when the potential is changed from

0.80 V to 0.05 V.

It can be noted in Fig. 7 that the intensity of the SFG signals at

1190 cm^ꢀ1and 1214 cm^ꢀ1remains constant as function of time at

0.05 V and quickly decreases when the potential is changed from

0.05 V to 0.80 V. The abrupt change in the SFG signal is consistent

with an instantaneous jump, within the temporal resolution of the

measurement (ꢁ3 s). The ratio between the intensities of the SFG

signal at 0.05 V and 0.80 V is equal to 2 and 6.5 for the nonresonant

sponds to a reactivation of the electrode for ethanol oxidation?

Above 0.90 V, it is possible that the adsorbed reaction intermedi-

ates are different than those observed at relatively lower poten-

tials. The conversion of part of the adsorbed intermediates

(formed at 0.05 V) to other adsorbed species, in particular linear-

bonded CO, is a proof that the reaction mechanism depends on

the applied potential. Additional evidences were reported by Wang

et al. [57] and Iwasita [58]. They investigated the volatile products

(acetaldehyde and carbon dioxide) of the ethanol electro-oxidation

over platinum using DEMS and they observed that only acetalde-

hyde is formed above 1.10 V. In this potential region, probably

the adsorption of ethanol molecules and the CAC bond breaking

are inhibited by the presence of other species adsorbed on the plat-

inum surface. As a result, the intermediates are weakly adsorbed

(little residence time at the surface) and quickly transformed into

partially oxidized products. In such a manner, we do not observe

the SFG bands associated with the reaction intermediates in the

potential range between 1.10 V and 1.30 V due to their low stea-

dy-state surface coverage, below the sensitivity of our SFG appara-

tus. Therefore, we are not able to explore these adsorbed species

and offer a mechanistic view of the Pt reactivation at 1.10 V. It is

important to mention that Ohmic drop effects could affect the re-

gion of faradaic current, above 0.5 V vs. RHE, increasing the applied

potential necessary to attain the desired electrode potential E.

Therefore, when we use 0.9 V in the thin ﬁlm geometry, it may

actually correspond to a lower electrode potential vs. RHE. We

did not correct our results for the ohmic drop. However, neglecting

this correction does not affect our qualitative analysis on the for-

mation/reactivity of adsorbed intermediates of the ethanol elec-

tro-oxidation on Pt.

Finally, we discuss the thermodynamic implications of our ﬁnd-

ings. At ﬁrst, it seems quite surprising that several adsorbates

coexist simultaneously at low potentials, as implied by the com-

plex SFG spectra in the 1000–1400 cm^ꢀ1range. This would require

that many adsorbates would have similar adsorption free energies,

so that their equilibrium surface coverages are comparable. How-

ever, this unlikely conclusion assumes thermodynamic equilibrium

between the adsorbates and ethanol in solution. If the adsorption

energies are high enough when compared to the thermal energy,

the adsorption would be ‘‘irreversible’’ (at a ﬁxed potential), and

the surface coverages would not be determined by thermodynamic

equilibrium, but rather by the kinetics of adsorption and reaction.

In this way, all adsorbates that have high adsorption energies

would have appreciable coverage, despite any difference in their

energy values.

signal of the platinum (1190 cm^ꢀ1

) and the SFG signal at

1214 cm^ꢀ1, respectively. When the potential is changed from

0.80 V to 0.05 V, the original situation is instantaneously retrieved

(again, within the temporal resolution of the measurement, ꢁ3 s).

This experiment was repeated in triplicate and the obtained results

were in accord with those presented in Fig. 7. Also, the same

behavior was observed for the SFG signal at 1120 cm^ꢀ1

.

Although the fast decrease in the SFG signal at 1214 cm^ꢀ1is

consistent with the hypothesis of ethanol consumption and our

estimate of its timescale, the similarly fast recovery of this signal

would not be expected, since it would take place in the much

slower timescale of the diffusion of ethanol from the bulk solution

into the thin layer of electrolyte. On the other hand, if our estimate

of the timescale for ethanol consumption in the thin layer is too

short, we should then have observed a gradual decrease in the

SFG signal upon increasing the potential from 0.05 V to 0.8 V. Thus,

we can rule out the exhaustion of the ethanol molecules in the thin

layer of electrolyte as the reason for the decrease in the intensity of

the most of the SFG bands at the wavenumber range between

1000 cm^ꢀ1to 3300 cm^ꢀ1from 0.50 V to 0.90 V (Fig. 5).

We believe the most appropriate explanation is given by sce-

nario (iii) above, that is, the adsorbed intermediates from ethanol

oxidation are gradually oxidized with increasing potential, and

the free platinum sites are not re-occupied only by ethanol mole-

cules, but also by other species that poison the catalyst surface,

suppressing the dynamic oxidation reaction via adsorbed interme-

diates. This is in fact what is observed in chronoamperometry

experiments, where a marked reduction in the oxidation current

is observed at a ﬁxed potential within the ethanol oxidation peak

[34]. However, these poisoning species are not detected by SFG

spectroscopy, probably because their vibrations are below

1000 cm^ꢀ1(e.g., PtAO). As the potential increases, the relative cov-

erage of ethanol adsorbates to the poison species decreases, and at

0.90 V, most of the adsorbed intermediates were completely re-

moved from the platinum surface (very weak SFG signals in

Fig. 5), and the faradaic current starts to decrease in the cyclic vol-

tammogram shown in Fig. 2. The immediate recovery of the SFG

signal when the potential is changed from 0.8 V back to 0.05 V

shows that the poison is readily reduced and leaves the Pt surface

ready for ethanol readsorption. Additionally, the reduction of the

SFG peak intensities could also be due to a low dynamic coverage

of the intermediates during the oxidation reaction, as a result of a

faster oxidation turnover with respect to local ethanol diffusion

and adsorption (scenario (ii)). However, this cannot be the only

mechanism to explain the reduction in SFG signal at 0.8 V, since

this would imply on a constant oxidation current, which is incon-

sistent with chronoamperometry experiments.

3.3. Results with isotopically labeled ethanol

In order to assist the peak assignment of the SFG spectra, espe-

cially in the ﬁngerprint region, we have performed experiments

with deuterium labeled reagents (water and ethanol) and also

¹³C labeled ethanol. Fig. 8 shows the SFG spect_ꢀra₁from 1000 cm^ꢀ1

to 1800 cm^ꢀ1and from 2100 cm^ꢀ1to 3300 cm of the following

interfaces: (0.1 mol L^ꢀ1CH₃CH₂OH + 0.5 mol L^ꢀ1HClO₄)/Pt and

(0.1 mol L^ꢀ1CD₃CD₂OD/D₂O + 0.5 mol L^ꢀ1HClO₄/D₂O)/Pt. All SFG

spectra presented in Fig. 8 were taken at 0.3 V vs. RHE. SFG spectra

ranging from 1000 cm^ꢀ1to 1500 cm^ꢀ1of the (0.1 mol L^ꢀ1CH₃CH_2-

OH + 0.5 mol L^ꢀ1HClO₄)/Pt interface and (0.1 mol L^ꢀ1CH₃A¹³CH_2-

OH + 0.5 mol L^ꢀ1HClO₄)/Pt interface at 0.05 V vs. RHE are

presented in Fig. 9. Table 1 summarizes the wavenumbers of the

main bands evidenced in the SFG spectra presented in Figs. 8 and

9. It also includes possible band assignments that will be discussed

in Section 3.4.

At this point, another interesting question appears: if we do not

observe most of the SFG bands related to the adsorbed intermedi-

ates of the ethanol electro-oxidation reaction on platinum above

0.90 V, how could we explain the second oxidation peak (at around

1.30 V) in the cyclic voltammogram shown in Fig. 2, which corre-

As seen in Fig. 8, the SFG spectrum for CD₃CD₂OD + HClO₄/Pt

presents most of the bands observed in the wavenumber region

between 1000 cm^ꢀ1and 1450 cm^ꢀ1for the CH₃CH₂OH + HClO₄/Pt

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

75

0.4

*

CH₃CH₂OH + HClO₄

0.2

0.1

0.0

0.2

*

0.0

0.4

1500

1600

1700

1800

1000

1100

1200

1300

1400

*

CD₃CD₂OD + HClO₄

0.2

0.1

0.0

0.2

0.0

*

1000

1100

1200

1300

1400

1500

1600

1700

1800

IR wavenumber (cm^-1)

0.4

0.2

0.0

0.4

0.2

CH₃CH₂OH + HClO₄

0.0

0.4

2600

2800

3000

3200

2100

2200

2300

2400

2500

0.4

0.2

0.0

CD₃CD₂OD + HClO₄

0.2

0.0

2100

2200

2300

2400

2500

2600

2800

3000

3200

IR wavenumber (cm^-1)

Fig. 8. SFG spectra (1000–1800 cm^ꢀ1and 2100–3300 cm^ꢀ1) of the following interfaces: (0.1 mol L^ꢀ1CH₃CH₂OH + 0.5 mol L^ꢀ1HClO₄)/Pt and (0.1 mol L^ꢀ1CD₃CD₂OD in

D₂O + 0.5 mol L^ꢀ1HClO₄in D₂O)/Pt at 0.30 V vs. RHE. ^ꢂThe SFG peak at 1408 cm^ꢀ1and the SFG band at ꢁ1475 cm^ꢀ1are not related to any molecular vibrational transition (see

Section 3.1).

spectrum, except the SFG bands at 1288 cm^ꢀ1, 1348 cm^ꢀ1and

1370 cm^ꢀ1. From 1450 cm^ꢀ1to 1800 cm^ꢀ1, the main change in

the CD₃CD₂OD + HClO₄/Pt spectrum compared to the CH₃CH_2-

OH + HClO₄/Pt interface spectrum is the absence of the band at

attached to the OH group of the ethanol (a carbon), although the

partial overlap between the displaced bands (¹³C) and the original

bands (¹²C) may allow a contribution from species containing only

¹²C (the b carbon of ethanol).

around 1496 cm^ꢀ1

.

In the wavenumber range between

2100 cm^ꢀ1and 2500 cm^ꢀ1for the deuterated solution (only the

HClO₄was protonated), it can be noted a broad band (from

2100 cm^ꢀ1to 2350 cm^ꢀ1centered at ꢁ2250 cm^ꢀ1), which may be

due to the OD stretch of water molecules interacting with the

Pt surface by hydrogen bonding, that is, the same band at 2700–

3300 cm^ꢀ1for CH₃CH₂OH + HClO₄/Pt but red shifted upon deuter-

ation. Finally, between 2500 cm^ꢀ1and 3300 cm^ꢀ1, the CD₃CD₂

OD + HClO₄/Pt spectrum presents relatively less intense bands than

the CH₃CH₂OH + HClO₄/Pt spectrum, since the H abundance in the

solution is much smaller. In presence of CH₃A¹³CH₂OH (as shown

in Fig. 9), bands in the wavenumber range between 1000 cm^ꢀ1

and 1500 cm^ꢀ1are red shifted with respect to the band positions

for CH₃CH₂OH + HClO₄/Pt.

The results with isotopically labeled ethanol are extremely

important for the identiﬁcation of the adsorbed intermediates of

the ethanol electro-oxidation on Pt in acidic medium. On one hand,

the bands present in both CH₃CH₂OH/Pt and CD₃CD₂OD/Pt spectra

deﬁnitively correspond to CAC and CAO vibrational modes of the

adsorbed intermediates. On the other hand, the bands that disap-

pear in the presence of deuterated ethanol and D₂O are surely re-

lated to CAH and OAH vibrational modes. The ¹³C results

indicate that most of the adsorbed intermediates detected between

1000 cm^ꢀ1and 1500 cm^ꢀ1present the carbon atom initially

3.4. Tentative Assignment of the Vibrational Bands and Identiﬁcation

of Reaction Intermediates

Based on the results with isotopically labeled ethanol, which

were presented in the last section, we have discussed only the gen-

eral assignment of the vibrational SFG bands observed in the pres-

ent work. The detailed discussion of them is relatively more

complex. For this reason, we need support from literature and the-

oretical calculations.

In order to have supporting information for the assignment of

the observed peaks in the SFG spectra for the ethanol solution/Pt

interface (listed in Table 1), we theoretically evaluated the infrared

and Raman spectra of some of the possible adsorbed intermediates

of the ethanol electro-oxidation on Pt. The species considered

here were: carbon monoxide, a secondary ethanol derivative

(PtACHOHACH₃), a tertiary ethanol derivative [(Pt)₂@COHACH₃],

acetyl, bidentate acetate, ethoxy and

g

²-acetaldehyde. Although

the accurate computation of IR and Raman intensities is notori-

ously difﬁcult, the theoretical results for fundamental frequencies

usually show satisfactory agreement (within a few tens of cm^ꢀ1

)

with the experimental data [44,59]. As a benchmark calculation,

we have computed the infrared and Raman spectra for an isolated

ethanol molecule to compare them with experimental results for

76

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

Table 1

Wavenumber of the main bands observed in the SFG spectra shown in Fig. 8 and Fig. 9.

CH₃CH₂OH

CH₃¹³CH₂OH on

Pt/cm^ꢀ1(this

work)

Adsorbates (deuterated

adsorbates) on metal surfaces

[Ref.]/cm^ꢀ1

Assignment [Ref.]^*,a,b

Proposed adsorbed intermediate

(CD₃CD₂OD) on Pt/

cm^ꢀ1(this work)

1030 (1024)

1056 (1050)

1022

1030 [73,74,91]

1055 (1070) [69] 1045^*

1068^*

1135 (1115) [72], 1127^*

1120–1160 [79]

1155^*1163 [80]

m

(CAO) [82–84,91]

Methoxy [82–84,91], ethoxy [73,74]

Bidentate acetate [69], secondary ethanol derivative^*

Tertiary ethanol derivative^*

(CAC) +

m(CAO) [73,74]

1048

m

q

(CAC)

(CH₃) +

m(CAC) +

m

(CAO)^*

1084 (1082)

1122 (1116)

1078

1114

m

(CAC) + d(OH)^*

m

(CAC) [72]

g

²-Acetaldehyde [72], acetyl^*, ethylidyne [79]

(CAC) +

q

(CH₃)^*

_s(OACAC) +

(CAC) [80]

m

(CAC) [79]

(CH₃)^*

1168 (1162)

1160

d(OH) +

m

q

Tertiary ethanol derivative^*ethylidyne [80]

m

1214 (1210)

1270 (1270)

1204

1250

1200 [85]

(CAOH) [85]

(CAOH)

Tertiary methanol derivative [85]

COH residues [85,86]

1270 [85,86]

1288 (not seen)

1330 (1326)

1266

1310

1262, 1280^*

d_ip(CH) +

q

(CH₃)^*

Secondary ethanol derivative,^{* 2}-acetaldehyde^*

g

1320 [87–90], 1360^*

m

_s(COO^ꢀ) [87–90]

Bidentate formate [87–90], bidentate acetate^*

Ethoxy [92], acetyl^*

(CAC) +

m

_s(COO^ꢀ)^*

1348 (not seen)

1370 (not seen)

1326

1346

1350 [92], 1353^*

1370 [72–74,78],^*

x (CH₂) [92]

d_s(CH₃)^*

d(CH₃)

methoxy [78], ethoxy [73,74],

g

²-acetaldehyde [72],

bidentate acetate,^*secondary and tertiary ethanol

derivative^*

1432 (1430)

1430

1415–1420 [78]

m

_ip(CAO) [78]

g

²-Formaldehyde [78],

²-acetyl

g

²-acetaldehyde [78]

1496 (not seen)

1546 (1546)

1499 [71]

d_s(CH₂) [71]

Ethoxy [71], CH₂(ads)

Acetyl [77]

1565 (1595) [77]

1735 [81], 1745[87]

2055–2070 [13,16,19]

m

(C@O)

1736 (1736)

Monodentate acetate [81], monodentate formate [87]

2056 (2100–2350)

m

(C„O)

(OD)

Adsorbed CO [13,16,19]

Adsorbed D₂O

2700–3300 (weak)

2700 [49,50] (broad) 2670 [48],

2900 [47] (broad)

m

(OH)

Adsorbed H₂O, H₃O⁺

*

Wavenumbers and assignments from theoretical calculations carried out in the present work.

a

Notation used for mode description:

m, stretching; d, deformation;

x, wagging; q, rocking.

b

Subscripts indicate the type of mode: s – symmetric, as – asymmetric, op – out-of-plane vibration, ip – in-plane vibration.

(liquid) ethanol [60–62] and check the accuracy of our method. The

theoretical results are shown in Tables 2 and 3.

Table 2 presents experimental and calculated wavenumbers

corresponding to ethanol vibrations. In addition, the differences

between experimental and calculated wavenumbers (considering

the scale factor or not) are also presented in Table 2. In both cases,

it can be noted that the differences between experimental and cal-

culated wavenumbers are not uniform for low and high wavenum-

ber regions. By analyzing these differences without the scale factor,

it can be seen that the calculation obtains frequency values closer

to those observed experimentally in the region corresponding to

the CAC and CAO vibrations. Therefore, in the following discussion

of SFG peak assignments (Table 1) we have not used the scale fac-

tor in comparisons between experimental and calculated wave-

numbers for vibrations involving CAC and CAO stretches. On the

other hand, in the wavenumber regions corresponding to CH vibra-

tions, the unscaled calculation overestimates the experimental fre-

quencies and the scaled calculated wavenumbers show good

agreement with experiment. In this way, for CH vibrations, we

have used the scaled calculated frequencies when comparing to

experimental peaks listed in Table 1. It is important to mention

that for OH vibrations, the calculation estimates wavenumber val-

ues very far from those observed experimentally (considering the

scale factor or not). This is obviously due to the fact that hydrogen

bonding of the OH group to other ethanol or water molecules is not

taken into account in our calculations. Therefore, theoretical fre-

quencies which involve OH vibrations will not be used for compar-

isons with experiments.

Table 2

Experimental results and theoretical predictions for ethanol. Only the resonances that

are intense in both IR and Raman spectra are listed.

Experimental

[60–62]

Theoretical

Assignment^b,cD_exp.

D_-

exp.

-

wavenumbers unscaled

unscaled

-

scaled

(cm^ꢀ1

)

(scaled)^a(cm^ꢀ1

)

882

1051

1094

1320

1383

896 (860)

m

_s(CCO)

_a(CCO)

_s(CO)

ꢀ14

+17

0

+50

ꢀ66

+22

+58

+44

+101

ꢀ8

1034 (993)

1094 (1050)

1270 (1219)

1449 (1391)

d(COH)

(CH₂),

x

d_s(CH₃)

1454

2877

2890

2928

2974

ꢁ3350

1502 (1442)

2963 (2844)

2989 (2869)

3017 (2896)

3082 (2959)

3089 (2965)

3794 (3642)

d_as(CH₃)

ꢀ48

ꢀ86

+12

+33

+21

+32

+15

+9

m_s(CH₂)

m_a(CH₂)

m_s(CH₃)

m_a(CH₃)

m_s(OH)

ꢀ99

ꢀ89

ꢀ108

ꢀ115

ꢀ444

ꢀ292

As shown in Table 3, there is satisfactory agreement between

the theoretical results for CO on Pt (2002 cm^ꢀ1) with the experi-

mental data (2056 cm^ꢀ1), especially considering that this vibration

is coverage- and potential-dependent, effects which have not been

a

b

c

Scale factor 0.96 [93].

Notation used for mode description:

Subscripts indicate the type of mode: s – symmetric and as – asymmetric.

m

, stretching; d, deformation.

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

77

3.4.1. Water, hydronium and OD (bands between 2300 cm^ꢀ1and

3300 cm^ꢀ1

Table 3

Theoretical predictions for CO, secondary and tertiary ethanol derivative, acetyl,

bidentate acetate, ethoxy and

²-acetaldehyde adsorbed on Pt(111). Only the

resonances that are intense in both IR and Raman spectra are listed. Underlined

values will be considered in comparison with experimental peaks (see text for

details).

)

g

The broad band from 2650 cm^ꢀ1to 3200 cm^ꢀ1may be due to

OH stretches of water molecules or hydronium ions (H₃O⁺) [47–

50] interacting with the Pt surface by hydrogen bonding (not

chemisorbed). This band was previously reported in SFG studies

[49,50] and assigned to the OH stretching vibration of the water

molecules in the double layer. Indeed, the OH stretching vibration

frequency of the water molecules in the double layer may be rela-

tively different compared to that observed for liquid water

(ꢁ3400 cm^ꢀ1) [63]. However, we disagree with this interpretation

for two reasons: (i) the proﬁle of the SFG band related to water in

the double layer has been previously reported [23,24] and it is not

exactly the one we observe; (ii) the dependence of the band inten-

sity with the potential is not the one expected for water in the dou-

ble layer. From a negatively charged surface (at low potential) to a

positively charged surface (at high potential), we should have ob-

served a decreasing, and then increasing, band intensity, which is

also not the case.

Adsorbates

Theoretical

Assignment^b,c

wavenumbers unscaled

(scaled)^a(cm^ꢀ1

2002 (1922)

)

CO

m(C„O)

Secondary ethanol

m

q

(CAC) + d(CH₃) + d(OH)

1035 (994)

1045 (1003)

1188 (1140)

(CH₃) +

m

(CAC) +

m

(CAO)

Strong d_op

(CH) + d(OH) + weak

Strong d(OH) + d_op

q

(CH₃)

1229 (1180)

1315 (1262)

(CH) + weak

Strong d_ip(CH) + weak

(CH₃)

q(CH₃)

q

d_s(CH₃)

1430 (1373)

1493 (1433)

1499 (1439)

Strong d_as(CH₃)

d_as(CH₃)

Tertiary ethanol

q_op(CH₃) + d(OH)

q_ip(CH₃) + d(OH)

1017 (976)

1066 (1023)

1068 (1025)

1203 (1155)

1426 (1369)

1037 (996)

1127 (1082)

1409 (1353)

1753 (1683)

3.4.2. Adsorbed intermediates previously proposed at electrochemical

environment with basis on electrochemical evidences (ClO^ꢀ₄, tertiary

ethanol derivative, bidentate acetate, and CO)

m(CAC) + d(OH)

d(OH) +

d_s(CH₃)

m_s(OACAC) + q(CH₃)

The band at 1120 cm^ꢀ1is extensively reported in the literature

[14,20,64,65] and has been assigned to the asymmetric stretching

of perchlorate anions in solution. Nevertheless, in the present

work, this band presents atypical behavior as a function of the po-

tential (data shown in Fig. 5), compared to previous work [66]. Spe-

ciﬁcally, the SFG spectra show that the band at 1120 cm^ꢀ1is

present even at potential as low as 0.05 V. Further, its intensity de-

creases with positive going potential, in contrast to what we can

expect for an anion [66]. In fact, another band at 980 cm^ꢀ1has

been observed with SFG spectroscopy and was assigned to ab-

sorbed perchlorate anions [33]. Therefore, the feature at

1120 cm^ꢀ1is more likely to be related to another adsorbed species.

The tentative assignment of this feature will be presented later in

Sections 3.4.3 and 3.4.4.

Acetyl

q

(CH₃)

(CAC) +

Strong d_s(CH₃)

m

q(CH₃)

m

(C@O) + d_as(CH₃)

_s(COO^ꢀ) +

(CAC)

_as(COO^ꢀ) + d(CH₃)

Bidentate acetate

Ethoxy

m

1360 (1306)

1407 (1351)

1428 (1371)

Strong

m

d_s(CH₃)

m

q

s

x

(CAC) +

(CH₃)

(CH₂)

d_s(CH₃)

d_as(CH₃)

d_s(CH₂) + weak d_s(CH₃)

m(CAO)

1027 (986)

1115 (1070)

1271 (1220)

1370 (1315)

1419 (1362)

1498 (1438)

1508 (1448)

According to our calculations, the bands at 1084 cm^ꢀ1and

1168 cm^ꢀ1correspond, respectively, to the

the d(OH) + m_s(OACAC) + q(CH₃) modes of a tertiary ethanol deriv-

m

(CAC) + d(OH) and

g

²-acetaldehyde

m

q

(CAC)

1101 (1057)

1159 (1113)

1333 (1280)

1425 (1368)

3024 (2903)

(CH₃) + q_op(CH)

ative, (Pt)₂@COHACH₃. The latter mode has a strong contribution

of the OH bending vibration. For this reason, the scaled theoretical

wavenumber in Table 3 is more appropriate than the unscaled one.

However, the wavenumber shift in presence of deuterated reac-

tants is only 6 cm^ꢀ1(Table 1), not as important as we should ex-

pect for a mode including a hydrogen atom. With basis on our

calculations, we explain this conﬂicting behavior in terms of the

geometry of the adsorbed tertiary ethanol intermediate and the

important contribution of the OACAC stretching to the mode at

1168 cm^ꢀ1. The OH bond is almost parallel to the platinum surface.

Then, although the OH bending is very strong, it contributes not

much to the observed peak intensity, according to the infrared

selection rule that allows only those vibrations with dipole

changes perpendicular to the surface to be observed. Consequently,

the character of OACAC stretching mode, which does not involve

any H atom, prevails over the character of OH bending. Therefore,

deuteration of the OH group would decouple it from this mixed

vibrational mode, without affecting its frequency signiﬁcantly. Ter-

tiary ethanol derivative was previously proposed by Iwasita and

Pastor based on a band at 1257 cm^ꢀ1assigned to its CAOH stretch

[17]. Iwasita and Pastor mentioned that this frequency was slightly

higher than that one expected for tertiary ethanol in solution be-

tween 1205 cm^ꢀ1and 1125 cm^ꢀ1. Therefore, we agree with the

possible formation of this intermediate, but disagree on the assign-

ment of the band at 1257 cm^ꢀ1(see Section 3.4.4).

strong d(CH) + weak

d_s(CH₃)

q(CH₃)

m_s(CH₃)

a

Scale factor 0.96 [93].

b

Notation used for mode description:

, rocking and sci, scissoring.

Subscripts indicate the type of mode: s – symmetric, as – asymmetric, op – out-

m, stretching; d, deformation; x, wagging;

s

, twisting; q

c

of-plane vibration, ip – in-plane vibration.

included in the calculations. This validates the methodology em-

ployed in the present work. However, the most important conclu-

sion from the theoretical results is that no single adsorbate (nor a

small set of them) can explain the very complex vibrational spectra

of Fig. 8 in the 1000–1400 cm^ꢀ1range, particularly knowing that

most of these bands are CAC or CAO stretches (as discussed in Sec-

tion 3.3). We must conclude that it is originated from several

adsorbates being present on the Pt surface, even at low potentials.

The implications of this ﬁnding will be discussed later.

Next, we will discuss in detail the tentative assignments of the

SFG bands observed in the present work. These band assignments

will be based on our theoretical calculations and on electrochemi-

cal and UHV studies previously reported in the literature. We orga-

nize the band assignments in following Sections 3.4.1–3.4.4.

78

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

The band at 1330 cm^ꢀ1can be attributed to symmetric

(COO^ꢀ) +

(CAC) mode of bidentate acetate, according to our the-

oretical predictions. Usually, the symmetric

(COO^ꢀ) vibration

cal environment, the SFG spectra also provide us with evidence of

m

bands at 1056 cm^ꢀ1and 1288 cm^ꢀ1, also never before reported in

electrochemical studies. These vibrational modes can be related

to the secondary ethanol intermediate (PtACHOHACH₃). According

to our calculations, the bands at 1056 cm^ꢀ1and 1288 cm^ꢀ1could

m

mode of adsorbed acetate ions is at relatively higher wavenumbers,

near 1420 cm^ꢀ1, as reported in an FTIR study on the adsorption of

acetate at the basal planes of platinum single-crystal electrodes

[67]. Recently, the Wieckowski group has also observed acetate

vibrations at ꢁ1410 cm^ꢀ1on Pt and Au electrodes with SFG spec-

troscopy, with bandwidths from 30 to 50 cm^ꢀ1[35]. Based on the

usual frequency of adsorbed acetate ions, acetate was previously

proposed as an adsorbed intermediate of the ethanol electro-oxi-

dation on Pt in acidic medium [13,68]. This species is reported to

be formed when acetic acid, one of main soluble products of the

ethanol oxidation, becomes adsorbed on Pt. In these studies

correspond to the [

weak (CH₃)], respectively. Previously, Iwasita and Pastor [17]

have already proposed that ethanol was adsorbed through the

-carbon after the rupture of a CAH bond to form an adsorbed

q(CH₃) + m(CAC) + m(CAO)] and [strong d_ip(CH) +

q

a

secondary alcohol derivative. However, no evidence for the pres-

ence of a secondary ethanol intermediate in the electrochemical

environment has been provided in that work [17].

Finally, the formation of acetyl, which was also previously pro-

posed as an intermediates of ethanol oxidation in electrochemical

environment (without any clear experimental evidence) [17,68], is

[13,68], the intensity of the band related to the symmetric m

(COO^ꢀ)

mode of bidentate acetate increases as function of the increasing

potential, just the opposite tendency we observe in the present

study. These differences concerning the wavenumber and potential

tendency may be related to the fact that in our study, bidentate

acetate species are formed on the surface (by the recombination

of the residues from ethanol dissociation on the platinum surface).

This situation is particularly different with respect to that one in

which acetate adsorbs on platinum as a consequence of the elec-

trostatic attraction between the positively charged surface and

acetate anions in solution (from acetic acid ionization). The wave-

number redshift we verify in the present study may be due to the

stronger interaction between the platinum surface and oxygen

atoms of the acetate formed in-loco than that one existing between

the platinum surface and adsorbed acetate ions that came from

solution, in which the negative charge is expected to be more delo-

calized between the carbon and oxygen atoms. The unusual poten-

tial tendency we observe here is attributable to the consumption of

the bidentate acetate, which is a precursor for the formation of

other adsorbed intermediates and/or reaction products. Besides

the band at 1330 cm^ꢀ1, the band at 1056 cm^ꢀ1can be assigned

to the CAC stretching mode of bidentate acetate. This assignment

is in accord with UHV studies of acetate adsorption on Al(111)

[69] and Cu(100) [70]. Therefore, based on the evidences of the

bands at 1330 cm^ꢀ1and 1056 cm^ꢀ1, we suggest that the bidentate

acetate is one of the adsorbed intermediates of the ethanol electro-

oxidation over platinum in acidic medium.

consistent with the presence of the bands at 1122 cm^ꢀ1

,

1348 cm^ꢀ1, and 1546 cm^ꢀ1. Based on our calculations, the bands

at 1122 cm^ꢀ1and 1348 cm^ꢀ1can be assigned to [

m

(CAC) + (CH₃)]

q

and strong d_s(CH₃) modes of an acetyl species, respectively. The

band at 1546 cm^ꢀ1was formerly assigned to the C@O stretching

mode of acetyl intermediates, which have been detected following

the adsorption of acetaldehyde on Pd(111) in UHV [77]. Our calcu-

lation shows that this vibrational mode should occur close to

1750 cm^ꢀ1. However, given the fact that our calculation does not

take into account neither the electrochemical potential nor the ac-

tive site of platinum, we believe that the band at 1546 cm^ꢀ1is

more likely to be associated with the C@O of an acetyl. Once

C@O is directly bonded to the negatively charged platinum surface,

the

p

electrons involved in the double bond, which are more mo-

electrons shared in a single bond, can resonate be-

vable than the

r

tween platinum, carbon, and oxygen atoms. In this case, the

interaction between carbon and oxygen atoms has a character of

a 1.5 bond and its CAO stretch may occur at around 1550 cm^ꢀ1

(such as the typical asymmetric stretching mode of COO^ꢀ).

3.4.4. Adsorbed intermediates of ethanol electro-oxidation proposed in

the present work for the ﬁrst time (

ethylidyne, monodentate acetate, methoxy, tertiary methanol

derivative, COH residue,

²-formaldehyde, mono and bidentate

formate, CH₃and CH₂residues)

The formation of

²-acetaldehyde (in the

the C and O atoms interact with the Pt surface through the car-

bonyl orbital) as an adsorbed intermediate of the ethanol oxida-

g g

²-acetaldehyde, ²-acetyl,

g

²-geometry, both

As mentioned before in Section 3.1, the strong band at

2056 cm^ꢀ1has been widely reported in the literature [13,16,19]

and is attributed to the C„O stretching mode of CO molecules lin-

early bonded to the Pt electrode.

p

tion on Pt in acidic medium is compatible with the presence of the

bands at 1122 cm^ꢀ1, 1288 cm^ꢀ1, and 1432 cm^ꢀ1. Here, we attribute

the bands at 1122 cm^ꢀ1and 1288 cm^ꢀ1to the CAC stretching and

3.4.3. Adsorbed intermediates previously proposed at electrochemical

environment without any clear experimental evidence (ethoxy,

secondary ethanol derivative and acetyl)

CH in-plane bending vibrations of the g

²-acetaldehyde, respec-

tively, which are theoretically predicted to be at around

1100 cm^ꢀ1and 1280 cm^ꢀ1(see Table 3). The band at 1432 cm^ꢀ1

The SFG bands at 1030 cm^ꢀ1, 1348 cm^ꢀ1, and 1496 cm^ꢀ1can

has been assigned to the CAO stretching mode of the g

²-acetalde-

correspond to the OACAC stretching mode [

m

(CAO) and

m

(CAC)

hyde in previous studies in electrochemical [68] and UHV [78]

environment. However, in the wavenumber range theoretically

investigated, we did not observed the band corresponding to the

coupling], CH₂wagging and CH₂deformation of ethoxy species,

respectively. This result is in excellent agreement with theoretical

calculations for ethoxy adsorbed on Mo(110) [71] and also in

agreement with previous UHV studies of ethoxy adsorbed on

Rh(111) [72], Cu(100) [73], Mo(110) [74], and Cu(111) [75]. For-

merly, ethoxy has been suggested as an intermediate of ethanol

oxidation on Pt at electrochemical environments, but only on the

basis of a band at 2920 cm^ꢀ1, in the CAH stretching region [17]. Gi-

ven the well-known difﬁculty of unambiguously assigning bands in

the CAH stretching region [76], the current observation of the SFG

bands at 1030 cm^ꢀ1, 1348 cm^ꢀ1and 1496 cm^ꢀ1, which are also

supported by our theoretical calculations, provides additional evi-

dence of the presence of ethoxy.

CAO stretching mode of the

be quite weak in the SFG spectrum since the CAO bond should be

nearly along the surface. In a general sense, the presence of g²

g

²-acetaldehyde, and indeed, it should

-

acetaldehyde suggested by the three bands discussed above is in

excellent agreement with an UHV study about the acetaldehyde

adsorption on Rh(111) [72] and with some results previously re-

ported by Davis and Barteau [78]. In an UHV study about the spec-

troscopic identiﬁcation of intermediates in ethanol decomposition

on Pd(111), Davis and Barteau [78] observed the

g

²-acetaldehyde

formation at 200 K. An additional interpretation for the band at

1432 cm^ꢀ1is to be associated with the CAO stretching mode of

In addition to providing the new and coherent evidences of the

presence of ethoxy species adsorbed on platinum in electrochemi-

g

²-formaldehyde species. There is a good correlation between this

band at 1432 cm^ꢀ1and that observed at 1420 cm^ꢀ1by Davis and

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

79

CH₃-¹³CH₂OH

CH₃CH₂OH

are also compatible with our SFG results and could be the precur-

sors to C„O formation.

The band at 1030 cm^ꢀ1matches the CAO stretching mode of a

methoxy intermediate. This frequency is similar to that found for

methoxy on Cu(110) [82], Ag(110) [83], and Pt(111) [84] surfaces

in UHV studies.

0.2

0.1

0.0

We also suggest the formation of a (Pt)₃„COH intermediate

(COH residues) by considering the band at 1270 cm^ꢀ1, which can

be attributed to its CAOH stretching mode, according to Iwasita

and coworkers [85,86]. The band at 1214 cm^ꢀ1is close to the OH

bending of a tertiary methanol derivative, (Pt)₂@CHOH, which

was previously suggested for methanol oxidation on Pt in acidic

medium, on the basis of a band at around 1200 cm^ꢀ1[85]. How-

ever, this peak does not shift signiﬁcantly upon deuteration, so it

cannot be an OH bending. Nevertheless, it may be assigned to

the CAOH stretch of a tertiary methanol derivative, since its fre-

quency is between those of COH residues (1270 cm^ꢀ1) and of the

OACAC stretching of tertiary ethanol derivative (1168 cm^ꢀ1).

Furthermore, as seen by the fact that the bands at 1736 cm^ꢀ1

and 1330 cm^ꢀ1correspond to characteristic frequencies of the

C@O stretching mode of modentate formate [87] and symmetric

COO^ꢀstretching mode of bidentate formate [88–90], we may also

consider the possible formation of these species in the early stages

of the ethanol oxidation on Pt.

1000

1100

1200

1300

1400

1500

IR wavenumber (cm^-1)

Fig. 9. SFG spectra (1000–1500 cm^ꢀ1) of the following interfaces: (0.1 mol L^ꢀ1

CH₃CH₂OH + 0.5 mol L^ꢀ1HClO₄)/Pt (ﬁlled circles) and (0.1 mol L^ꢀ1CH₃A¹³CH_2-

OH + 0.5 mol L^ꢀ1HClO₄)/Pt (empty circles) at 0.05 V vs. RHE. The SFG peak at

1408 cm^ꢀ1and the SFG band at ꢁ1475 cm^ꢀ1are not related to any molecular

vibrational transition (see Section 3.1). The spectrum with regular ethanol has been

displaced vertically for clarity of presentation.

Barteau [78] for methanol decomposition on Pd(111) under UHV

conditions. In addition to the probable formation of

hyde or

²-formaldehyde, we cannot exclude the possibility of

the presence of a

²-acetyl species since the CAC stretching mode

of a

²-acetyl intermediate may be quite close to the CAO stretch-

ing mode of a

²-acetaldehyde, at around 1430 cm^ꢀ1. Although

evidence of formation of

²-acetyl as an intermediate of the elec-

g

²-acetalde-

g

Finally, the bands at 1370 cm^ꢀ1and 1496 cm^ꢀ1are compatible

with the presence of residues like CH₃and CH₂directly attached to

the Pt surface. These bands at 1370 cm^ꢀ1and 1496 cm^ꢀ1can be

assigned to the CH₃bending of PtACH₃and CH₂bending of

(Pt)₂ACH₂, respectively.

g

tro-oxidation of ethanol is lacking in the literature, it is very likely

that the reaction pathway toward the CO and subsequent CO₂for-

mation may involve an adsorbed intermediate with both carbon

atoms attached to the surface, which may be the precursor of these

species. Therefore, based on our results, we tentatively propose

Table 1 also summarizes the assignment of the vibrational

bands and the identiﬁcation of the reaction intermediat_ꢀes₁.

Taken together, the bands at 1030 cm^ꢀ1, 1056 cm^ꢀ1, 1084 cm

,

1122 cm^ꢀ1

1330 cm^ꢀ1

,

1168 cm^ꢀ1

1348 cm^ꢀ1

,

1214 cm^ꢀ1

1370 cm^ꢀ1

,

1270 cm^ꢀ1

1432 cm^ꢀ1

,

1288 cm^ꢀ1

1496 cm^ꢀ1

that

g

²-acetyl is an active adsorbed intermediate of the ethanol

1546 cm^ꢀ1, and 1736 cm^ꢀ1in SFG spectra and the predictions from

oxidation on Pt in acidic medium.

the theoretical calculations conﬁrm that ethoxy (1030 cm^ꢀ1

,

The bands at 1122 or 1168 cm^ꢀ1can also correspond to the CAC

stretching mode of an ethylidyne-like adsorbate. In UHV, the CAC

stretching mode of an ethylidyne adsorbed on Pt(111), from the

1348 cm^ꢀ1and 1496 cm^ꢀ1), a secondary ethanol derivative –

PtACHOHACH₃(1056 cm^ꢀ1and 1288 cm^ꢀ1) and acetyl species

(1122 cm^ꢀ1, 1348 cm^ꢀ1and 1546 cm^ꢀ1), previously suggested at

electrochemical environment based on ambiguous evidences, are

active intermediates of the ethanol oxidation on Pt in acidic med-

ium. In addition, some of these bands are also compatible with the

presence of never before proposed two-carbon atoms intermedi-

ethylene decomposition, was characterized by

a band at

1120 cm^ꢀ1[79]. Nevertheless, this mode assignment for adsorbed

ethylidyne was based on a comparison with an ethylidyne tricobalt

nonacarbonyl complex, as ﬁrst outlined by Skinner et al. [80], that

assigned a band at 1163 cm^ꢀ1to the CAC stretching mode of the

ethylidyne. Therefore, we cannot exclude the possibility of any

one of these bands (at 1122 or 1168 cm^ꢀ1) to be related with the

CAC stretching mode of an ethylidyne-like adsorbate.

ates, such as:

g ,

²-acetaldehyde (1122 cm^ꢀ11288 cm^ꢀ1and

1432 cm^ꢀ1),

g

²-acetyl (1432 cm^ꢀ1), ethylidyne (1122 cm^ꢀ1or

1168 cm^ꢀ1), monodentate acetate (1736 cm^ꢀ1), and also 1-carbon

atom intermediates, for instance: methoxy (1030 cm^ꢀ1), tertiary

The band at 1736 cm^ꢀ1can be attributed to the C@O stretch of

an adsorbed monodentate acetate species. This band assignment is

in accord with that reported in an UHV study on the adsorption of

acetic acid on Pd(111) [81] with the formation of monodentate

acetate. In earlier investigations on the electrochemical oxidation

of ethanol on Pt in acidic medium by conventional IR spectroscopy,

the band at ꢁ 1740 cm^ꢀ1was attributed to the C@O stretching

vibration of the carbonyl group of the acetaldehyde and acetic acid

in solution [9,15,68]. However, this cannot be the origin of the

band in the SFG spectroscopy, since it is not sensitive to molecules

randomly oriented in solution.

methanol derivative (1214 cm^ꢀ1), COH residue (1270 cm^ꢀ1), g²

-

formaldehyde (1432 cm^ꢀ1), monodentate formate (1736 cm^ꢀ1),

bidentate formate (1330 cm^ꢀ1), CH₃(1370 cm^ꢀ1), and CH₂

(1496 cm^ꢀ1) residues. Moreover, we have conﬁrmed the presence

of intermediates previously documented with conventional IR

spectroscopy:

a

tertiary ethanol derivative (1084 cm^ꢀ1and

1168 cm^ꢀ1), bidentate acetate (1330 cm^ꢀ1), and CO_ad(2056 cm^ꢀ1).

Based on the results discussed above, a simpliﬁed scheme of the

electro-oxidation of ethanol on platinum in acidic solution is pro-

posed in Fig. 10.

It is generally suggested that the ethoxy and the secondary eth-

anol derivative formation is the initial step of the mechanism of

ethanol electro-oxidation on platinum. As mentioned by Iwasita

and Pastor [17], the saturated alcohols have two active centers

Therefore, based on our experimental results in conjunction

with our theoretical calculations and supporting information from

the literature, we suggest for ﬁrst time that

acetyl, ethylidyne, and monodentate acetate can also be adsorbed

intermediates of the ethanol electro-oxidation over platinum in

acidic medium. All of these species contain two carbon atoms

chain. In addition to them, we cannot ignore the possible formation

of adsorbed intermediates containing only one carbon atom, which

g -

²-acetaldehyde, g²

for adsorption on metal catalysts: the OH and the a-carbon (alco-

holic carbon). Thus, the ethanol molecule would adsorb either

through the oxygen atom or through the carbon atom after the

rupture of either the OAH (1) or the CAH (2), respectively.

80

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

Fig. 10. Proposed intermediates of ethanol oxidation on platinum electrode in acidic medium. The geometry of the adsorbed intermediates is not considered in this scheme.

CH₃CH₂OH þ Pt ! Pt ꢀ OCH₂CH₃ðethoxyÞ þ H^þþ 1e^ꢀ

ð2Þ

tion is high enough so that the adsorbate coverage and diversity is

governed by adsorption/reaction kinetics, and not by thermody-

namic equilibrium. We believe the latter is more likely, since

adsorption free energies should be quite sensitive to which atom

bonds to Pt (C or O) and to the number of bonds per molecule (sin-

gle-, double-, or triple-bonded to the surface).

In Fig. 10, additionally to the adsorbed intermediates of the

ethanol oxidation on Pt in acidic medium, the well-known reac-

tion products (CO₂, acetaldehyde, and acetic acid) and potential

ones (formaldehyde and formic acid) are also represented. Based

on our results, it is clear that the presence of complex adsorbed

intermediates (with an intact CAC bond, and not only CO) up to

the onset potential of the ethanol oxidation (0.5 V vs. RHE)

should lead to partially oxidized products, such as acetaldehyde

and acetic acid.

CH₃CH₂OH þ Pt ! Pt

ꢀ CHOHCH₃ðsecondary ethanol derivativeÞ

þ H^þþ 1e^ꢀ

ð3Þ

The presence of many other adsorbed intermediates, which should

therefore be produced by the reaction of the species formed in (1)

and (2), implies that the Pt surface is much more reactive than pre-

viously though. This coexistence of several adsorbates with compa-

rable surface coverages (all detectable with SFG spectroscopy,

whose signal is proportional to the square of the surface density)

is in fact quite remarkable. It either means that all these adsorbates

have very similar adsorption free energies, or the barrier for desorp-

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

81

theoretical work to conﬁrm the spectral assignments proposed

herein and to elucidate the detailed reaction pathways for elec-

tro-oxidation of several organic molecules, especially at high

potentials, where the low coverage prevented us from getting a

vibrational spectrum of the interfacial species.

As a perspective for future work, we intend to investigate by

SFG spectroscopy the adsorption and oxidation of ethanol at higher

concentrations and temperatures (relevant to fuel cell operation)

not only Pt-based catalysts, but also on other materials, such as

Au and Pd, in acidic and alkaline media. This would contribute to

the understanding of the inﬂuence of ethanol concentration, tem-

perature, electrode material, and pH on the reaction mechanism.

4. Conclusion

We have investigated the adsorbed intermediates of ethanol

oxidation on platinum using SFG spectroscopy and DFT calcula-

tions. Generally, the effect of the applied potential on the reaction

mechanism showed that in the pre-oxidation region (0.05–

0.50 V), with increasing potential, the species formed at 0.05 V

are partially converted to other adsorbed species, particularly

to CO linearly bonded to the platinum surface. Also, we demon-

strated that in the potential range between 0.50 V and 0.90 V,

the adsorbed intermediates from ethanol oxidation are gradually

oxidized with increasing potential and the free platinum sites

are not re-occupied only by ethanol molecules, but also by other

species. These species have strong interaction with the platinum

surface and, as a result, they block the catalyst active area and

inhibit the dynamic ethanol oxidation reaction (via adsorbed

intermediates). We suggest that above 0.90 V, the ethanol

adsorption and the CAC bond breaking are inhibited by the

increasing presence of oxygenated species adsorbed on the plat-

inum surface. In this manner, the adsorbed intermediates are

weakly bonded and quickly transformed into partially oxidized

products (with intact CAC bond) in such a way that the steady-

state superﬁcial concentration of the adsorbed molecules decreases

below the detection limit of our apparatus.

Acknowledgments

We dedicate this manuscript to the memory of Prof. Francisco C.

Nart. The authors gratefully acknowledge Professor Wolf Vielstich

and Prof. Teresa Iwasita for the helpful discussions and guidance.

We thank Hilton B. de Aguiar and L. Jay Deiner for assistance with

experiments and discussions during the initial stages of this work.

The authors also thank FINEP, CNPq, CAPES, and FAPESP for ﬁnan-

cial support of this work. M.F.S.P. and K.B. acknowledge CAPES/

CNPq (process number PNPD0052086) and FAPESP (process num-

ber 05/02284-0) for postdoctoral fellowships, respectively.

The observed vibrational spectrum in the 1000–1400 cm^ꢀ1

range is very complex, with at least 11 peaks (most of them as-

signed to CAO or CAC stretches), implying on the coexistence of

several adsorbates, even at low potentials. It is at variance with

those recently reported by Kutz et al. [34] for ethanol electro-oxi-

dation on polycrystalline Pt, since they only detect adsorbed CO

and acetate ions (and (bi)sulfate ions, for the case of sulfuric acid

supporting electrolyte). We attribute this discrepancy to different

surface preparation and SFG data acquisition methodologies: they

clean the surface by electrochemical potential cycling and measure

References

[1] J.M. Leger, S. Rousseau, C. Coutanceau, F. Hahn, C. Lamy, Electrochim. Acta 50

(2005) 5118.

[2] C. Lamy, A. Lima, V. LeRhun, F. Delime, C. Coutanceau, J.M. Leger, J. Power

Sources 105 (2002) 283.

[3] J. Goldemberg, Science 315 (2007) 808.

[4] E. Antolini, E.R. Gonzalez, Catal. Today 160 (2011) 28.

[5] D.M. dos Anjos, F. Hahn, J.M. Leger, K.B. Kokoh, G. Tremiliosi, J. Braz. Chem. Soc.

19 (2008) 795.

[6] J.F. Gomes, B. Busson, A. Tadjeddine, G. Tremiliosi, Electrochim. Acta 53 (2008)

6899.

[7] L. Jiang, A. Hsu, D. Chu, R. Chen, Int. J. Hydrogen Energy 35 (2010) 365.

[8] G.R. Salazar-Banda, K.I.B. Eguiluz, M.M.S. Pupo, H.B. Suffredini, M.L. Calegaro,

L.A. Avaca, J. Electroanal. Chem. 668 (2012) 13.

[9] F. Vigier, C. Coutanceau, F. Hahn, E.M. Belgsir, C. Lamy, J. Electroanal. Chem.

563 (2004) 81.

SFG spectra during a potential sweep using a 25 lm thick electro-

lyte layer, while we clean by H₂ﬂame annealing, adsorb ethanol

from solution onto a pristine Pt surface held at low potentials

and take the SFG spectra at ﬁxed potentials with a window-elec-

trode gap of 2–4 lm. It is interesting to note that although our cyc-

[10] G.F. Cui, S.Q. Song, P.K. Shen, A. Kowal, C. Bianchini, J. Phys. Chem. C 113 (2009)

15639.

lic voltammograms are similar to those in Ref. [34], the SFG spectra

disagree, suggesting that voltammetry may not be as sensitive as

SFG spectroscopy to ascertain surface quality and cleanliness.

Our theoretical methodology has satisfactorily supported

experimental measurements and this mutual agreement provided

additional insights concerning some of the adsorbed intermediates

involved at electro-oxidation of ethanol. Our results show that eth-

oxy, secondary ethanol derivative and acetyl, previously suggested,

but never before unambiguously documented, are active interme-

diates of ethanol oxidation in acidic medium. In addition, we report

new spectroscopic evidences (in electrochemical environment)

[11] C. Bianchini, P.K. Shen, Chem. Rev. 109 (2009) 4183.

[12] J. Willsau, J. Heitbaum, J. Electroanal. Chem. 194 (1985) 27.

[13] X.H. Xia, H.D. Liess, T. Iwasita, J. Electroanal. Chem. 437 (1997) 233.

[14] L.W.H. Leung, S.C. Chang, M.J. Weaver, J. Electroanal. Chem. 266 (1989) 317.

[15] T. Iwasita, B. Rasch, E. Cattaneo, W. Vielstich, Electrochim. Acta 34 (1989)

1073.

[16] B. Bittinscattaneo, S. Wilhelm, E. Cattaneo, H.W. Buschmann, W. Vielstich, Ber.

Bunsen. Phys. Chem. 92 (1988) 1210.

[17] T. Iwasita, E. Pastor, Electrochim. Acta 39 (1994) 531.

[18] G. Socrates, Infrared and Raman Characteristic Group Frequencies, Wiley-

Interscience, New York, 1980.

[19] B. Beden, M.C. Morin, F. Hahn, C. Lamy, J. Electroanal. Chem. 229 (1987) 353.

[20] R. Holze, J. Electroanal. Chem. 246 (1988) 449.

[21] J.W. Shin, W.J. Tornquist, C. Korzeniewski, C.S. Hoaglund, Surf. Sci. 364 (1996)

122.

[22] M. Heinen, Z. Jusys, R.J. Behm, J. Phys. Chem. C 114 (2010) 9850.

[23] P.B. Miranda, Y.R. Shen, J. Phys. Chem. B 103 (1999) 3292.

[24] Y.R. Shen, P. Natl, Acad. Sci. USA 93 (1996) 12104.

[25] S. Baldelli, N. Markovic, P. Ross, Y.R. Shen, G. Somorjai, J. Phys. Chem. B 103

(1999) 8920.

that are consistent with the presence of

yl, ethylidyne, monodentate acetate, methoxy, tertiary methanol

derivative, COH residues,

²-formaldehyde, mono and bidentate

g g

²-acetaldehyde, ²-acet-

g

formate, CH₃and CH₂residues. Moreover, the results obtained in

this work conﬁrm the presence of previously observed ethanol oxi-

dation intermediates: tertiary ethanol derivative, bidentate ace-

[26] S. Baldelli, J. Phys. Chem. B 109 (2005) 13049.

[27] M. Tadjeddine, J.P. Flament, A. Le Rille, A. Tadjeddine, Surf. Sci. 600 (2006)

2138.

tate, and CO_ad

.

Therefore, in addition to showing for ﬁrst time that the reaction

mechanism of the electro-oxidation of ethanol on Pt is much more

complex than believed so far, we have also contributed for the

understanding of the reaction mechanism by proposing new inter-

mediates of the ethanol oxidation on Pt, among them some inter-

mediates that have been previously proposed only in UHV

environment and some intermediates that we propose here for

the ﬁrst time. This should stimulate both experimental and

[28] J.F. Gomes, B. Busson, A. Tadjeddine, J. Phys. Chem. B 110 (2006) 5508.

[29] K.A. Friedrich, W. Daum, C. Klunker, D. Knabben, U. Stimming, H. Ibach, Surf.

Sci. 335 (1995) 315.

[30] C. Klunker, M. Balden, S. Lehwald, W. Daum, Surf. Sci. 360 (1996) 104.

[31] R.L. Behrens, A. Lagutchev, D.D. Dlott, A. Wieckowski, J. Electroanal. Chem. 649

(2010) 32.

[32] G.Q. Lu, A. Lagutchev, D.D. Dlott, A. Wieckowski, Surf. Sci. 585 (2005) 3.

[33] W.Q. Zheng, A. Tadjeddine, J. Chem. Phys. 119 (2003) 13096.

[34] R.B. Kutz, B. Braunschweig, P. Mukherjee, R.L. Behrens, D.D. Dlott, A.

Wieckowski, J. Catal. 278 (2011) 181.

82

J.F. Gomes et al. / Journal of Catalysis 302 (2013) 67–82

[35] B. Braunschweig, P. Mukherjee, R.B. Kutz, A. Wieckowski, D.D. Dlott, J. Chem.

Phys. 133 (2010) 8.

[36] Y. Waseda, K. Hirata, M. Ohtani, High Temp. - High Press. 7 (1975) 221.

[37] W.R. Wadt, P.J. Hay, J. Chem. Phys. 82 (1985) 284.

[56] B.N.J. Persson, R. Ryberg, Phys. Rev. B 24 (1981) 6954.

[57] H. Wang, Z. Jusys, R.J. Behm, J. Phys. Chem. B 108 (2004) 19413.

[58] T. Iwasita, J. Braz. Chem. Soc. 13 (2002) 401.

[59] Y. Zhang, J. Zhao, G. Tang, L. Zhu, Spectrochim. Acta A 61 (2005) 697.

[60] B. Schrader, W. Meier, Raman/IR Atlas, Verlag Chemie, Weinheim, 1974.

[61] J.-P. Perchard, M.-L. Josien, J. Chim. Phys. 64 (1968) 1834.

[62] J.-P. Perchard, M.-L. Josien, J. Chim. Phys. 64 (1968) 1856.

[63] B.A. Sexton, Surf. Sci. 94 (1980) 435.

[64] B. Rasch, T. Iwasita, Electrochim. Acta 35 (1990) 989.

[65] L.W.H. Leung, M.J. Weaver, J. Phys. Chem. 92 (1988) 4019.

[66] R. Holze, W. Vielstich, Electrochim. Acta 33 (1988) 1629.

[67] A. Rodes, E. Pastor, T. Iwasita, J. Electroanal. Chem. 376 (1994) 109.

[68] M.H. Shao, R.R. Adzic, Electrochim. Acta 50 (2005) 2415.

[69] J.G. Chen, J.E. Crowell, J.T. Yates, Surf. Sci. 172 (1986) 733.

[70] B.A. Sexton, Chem. Phys. Lett. 65 (1979) 469.

[71] P. Uvdal, A.D. Mackerell, B.C. Wiegand, J. Electron Spectrosc. 64–5 (1993) 193.

[72] C.J. Houtman, M.A. Barteau, J. Catal. 130 (1991) 528.

[73] B.A. Sexton, Surf. Sci. 88 (1979) 299.

[38] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,

J. Montgomery, J. A., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar,

J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A.

Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,

M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox,

H.P. Hratchian, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E.

Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y.

Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S.

Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K.

Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J.

Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L.

Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M.

Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A.

Pople, GAUSSIAN 03, Revision-B.4, in, Gaussian, Wallingford., 2004.

[39] A.D. Becke, Phys. Rev. A 38 (1988) 3098.

[74] D.A. Chen, C.M. Friend, Langmuir 14 (1998) 1451.

[40] A.D. Becke, J. Chem. Phys. 98 (1993) 1372.

[41] A.D. Becke, J. Chem. Phys. 96 (1992) 2155.

[75] S.C. Street, A.J. Gellman, J. Chem. Phys. 105 (1996) 7158.

[76] P. Uvdal, M.K. Weldon, C.M. Friend, Phys. Rev. B 50 (1994) 12258.

[77] J.L. Davis, M.A. Barteau, J. Am. Chem. Soc. 111 (1989) 1782.

[78] J.L. Davis, M.A. Barteau, Surf. Sci. 235 (1990) 235.

[79] R.G. Windham, B.E. Koel, J. Phys. Chem. 94 (1990) 1489.

[80] P. Skinner, M.W. Howard, I.A. Oxton, S.F.A. Kettle, D.B. Powell, N. Sheppard, J.

Chem. Soc. Farad. T. II (77) (1981) 1203.

[42] C.T. Lee, W.T. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785.

[43] E. Alonso, J.M. Casas, J. Fornies, C. Fortuno, A. Martin, A.G. Orpen, C.A. Tsipis,

A.C. Tsipis, Organometallics 20 (2001) 5571.

[44] A. Bakalova, H. Varbanov, S. Stanchev, D. Ivanov, F. Jensen, Int. J. Quantum

Chem. 109 (2009) 826.

[45] B. Giese, G.B. Deacon, J. Kuduk-Jaworska, D. McNaughton, Biopolymers 67

(2002) 294.

[46] Y. Zhang, L. Zhang, H.B. Tao, X.J. Sun, L.G. Zhu, Spectroc. Acta Pt. A-Molec.

Biomolec. Spectr. 59 (2003) 493.

[81] R.D. Haley, M.S. Tikhov, R.M. Lambert, Catal. Lett. 76 (2001) 125.

[82] B.A. Sexton, A.E. Hughes, N.R. Avery, Surf. Sci. 155 (1985) 366.

[83] Q. Dai, A.J. Gellman, Surf. Sci. 257 (1991) 103.

[84] B.A. Sexton, Surf. Sci. 102 (1981) 271.

[47] M. Falk, P.A. Giguere, Can. J. Chem. 35 (1957) 1195.

[85] T. Iwasita, F.C. Nart, J. Electroanal. Chem. 317 (1991) 291.

[86] T. Iwasita, F.C. Nart, B. Lopez, W. Vielstich, Electrochim. Acta 37 (1992) 2361.

[87] C. Xu, D.W. Goodman, J. Phys. Chem. 100 (1996) 1753.

[88] Y.X. Chen, A. Miki, S. Ye, H. Sakai, M. Osawa, J. Am. Chem. Soc. 125 (2003) 3680.

[89] M. Endo, T. Matsumoto, J. Kubota, K. Domen, C. Hirose, J. Phys. Chem. B 105

(2001) 1573.

[90] F.o. Hahn, B. Beden, C. Lamy, J. Electroanal. Chem. Interfacial Electrochem. 204

(1986) 315.

[91] T.H. Ellis, H. Wang, Langmuir 10 (1994) 4083.

[48] M. Okumura, L.I. Yeh, J.D. Myers, Y.T. Lee, J. Phys. Chem. 94 (1990) 3416.

[49] F. Dederichs, K.A. Friedrich, W. Daum, J. Phys. Chem. B 104 (2000) 6626.

[50] K.A. Friedrich, W. Daum, F. Dederichs, W. Akemann, Z. Phys, Chem. 217 (2003)

527.

[51] J.F. Gomes, K. Bergamaski, M.F.S. Pinto, P.B. Miranda, Optics Express, submitted

for publication.

[52] A. Holmgren, L.M. Wu, W. Forsling, Spectroc. Acta Pt. A-Molec. Biomolec.

Spectr. 50 (1994) 1857.

[53] D.K. Lambert, Solid State Commun. 51 (1984) 297.

[54] C. Korzeniewski, R.B. Shirts, S. Pons, J. Phys. Chem. 89 (1985) 2297.

[55] S. Holloway, J.K. Norskov, J. Electroanal. Chem. 161 (1984) 193.

[92] T. Kratochwil, M. Wittmann, J. Kuppers, J. Electron Spectrosc. 64–5 (1993) 609.

[93] N.s.r.d.n. NIST Computational Chemistry Comparison and Benchmark

Database, Release 14, in: R.D. Johnson III, Ed., NIST, Gaithersberg, MD, 2006.

Article Doi

Reaction intermediates of ethanol electro-oxidation on platinum investigated by SFG spectroscopy

DOI: 10.1016/j.jcat.2013.02.024

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.jcat.2013.02.024

Full text of DOI:10.1016/j.jcat.2013.02.024

Products guided by the article

R&D Labs maybe for 124-38-9

Relevant to this article

Hot Product