PERGAMON
Electrochimica Acta 44 (1998) 559±571
Copper dissolution in bromide medium in the absence and
presence of hexamethylenetetramine (HMTA)
A.G. Brolo 1, M.L.A. Temperini *, S.M.L. Agostinho
Instituto de QuõÂmica da Universidade de SaÄo Paulo, C.P. 20780, 01498-970 SaÄo Paulo, Brazil
Received 3 April 1998
Abstract
The anodic electrochemical behavior of copper in bromide medium, in the absence and presence of
hexamethylenetetramine (HMTA), has been studied, using electrochemical and spectroelectrochemical techniques.
The copper dissolution in the absence of HMTA was dependent on mass transport of bromide ions. A porous CuBr
®lm was formed at positive overpotentials. The ®lm growth mechanism followed the nucleation, growth and
superposition (NGS) model, with instantaneous nucleation and a di€usion controlled growth of a three dimensional
®lm. In the presence of the inhibitor, a passivant Cu/HMTA/Brÿ ®lm was observed. This ®lm has slow growth
kinetics, favored by mass transport. The inhibitory e€ect was observed for the potential range between 0 mV and
+800 mV (SCE). A maximum inhibitory eciency, near 100%, was observed for 15 mM HMTA. The nature of the
passivant ®lm was con®rmed by in situ surface-enhanced Raman scattering (SERS) and SEM/EDX measurements.
# 1998 Elsevier Science Ltd. All rights reserved.
Keywords: Corrosion inhibitor; Copper corrosion; Hexamethylenetetramine (HMTA); Surface Raman spectroscopy; SERS
1. Introduction
Copper corrosion in di€erent media and conditions
has been extensively studied [1]. The role of anions,
such as sulfate [2], ¯uoride [3], chloride [4, 5],
bromide [6], thiocyanate [7], carbonate [8] and
bicarbonate [9], on the copper anodic dissolution and
passivation has received special attention. Recently, the
dependence of the mechanism of copper corrosion on
the pH has also been reported [10]. It is well established that the copper dissolution process yields either
Cu+ or Cu+2 derivatives, depending on the nature of
the electrolyte. For instance, the monovalent cation is
stabilized in halide (except Fÿ) and pseudohalide
* Corresponding author. Fax: +55 11 815 5579; e-mail:
[email protected]
1
Current address: University of Waterloo, Department of
Chemistry, Waterloo, ON, N2L3G1, Canada.
media; and Cu+2 compounds are more likely to be
found for copper corrosion in sulfate and carbonate/
bicarbonate media [1].
The mechanism for copper dissolution in bromide
medium has been established in a series of papers by
Brossard and co-workers [11±14]. In the absence of
any CuBr precipitate on the electrode surface, the dissolution occurs according to the following steps:
ÿ
Cu ‡ 2Brÿ „ CuBrÿ
2…i † ‡ e 1st: step
ÿ
CuBrÿ
2…i † ÿ4CuBr2…s† 2nd: step:
Where the subscripts i and s are related to interfacial
and solution species, respectively. The di€usion transnd
port of CuBrÿ
step) is the rate determining step in
2 (2
this situation [13]. However, as the anodic current density becomes large enough, CuBr precipitates on the
electrode surface.
0013-4686/98/$19.00 # 1998 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 3 - 4 6 8 6 ( 9 8 ) 0 0 1 7 9 - 0
560
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
The mechanism for copper dissolution in the presence of CuBr ®lm, showed below, is di€erent, since
CuBr reacts with Brÿ forming soluble complex anions,
2ÿ
such as CuBrÿ
2 and CuBr3 [14].
Cu ‡ Brÿ „ CuBr…i † ‡ eÿ 1st: step
ÿ
Cu ‡ nBrÿ „ CuBr1ÿn
n…i † ‡ e 2nd: step
1ÿn
1ÿn
CuBrn…i
† ÿ4CuBrn…s† 3rd: step:
The di€usion of CuBr1n ÿ n species into the solution
(3rd step) is the rate determining step. The anodic current eventually reaches a limiting value, which is equal
to the dissolution rate of the CuBr precipitate accumulate on the electrode surface [14].
The mechanism for copper dissolution in bromide
medium is similar to the one in chloride medium [15].
However, the formation of Cu(II) have been reported
for copper dissolution in chloride medium [16, 17] at
relatively lower overpotentials than for copper dissolution in bromide [12].
Several inhibitors for copper corrosion have been
widely studied mainly in those media where the copper
dissolution mechanisms are known [18]. Therefore, various works can be found describing the action of inhibitors for copper in Clÿ [19±22] and SOÿ2
[23±25]
4
media. However, despite the mechanism for copper
dissolution in bromide medium has been determined, it
is still dicult to ®nd studies about the action of corrosion inhibitors in this medium [26].
In a previous work, we reported the e€ects of hexamethylenetetramine (HMTA) as an inhibitor for copper dissolution in bromide medium [27]. We showed
that HMTA is an e€ective corrosion inhibitor for copper in bromide medium containing either Fe(III) or O2
as oxidants. Corrosion potential measurements indicated that the anodic process was not completely di€usion controlled even in the absence of the inhibitor.
Potentiostatic and potentiodynamic anodic polarization curves for copper dissolution in bromide solutions containing HMTA were shifted to more
negative potentials due to the complexing action of
HMTA on Cu(I) ions. The total anodic current
became zero at more positive potential values due to
the formation of the passivant ®lm. The Tafel analysis
showed a decrease in the (dE/dlogI) slope as the
HMTA concentration increased. The oxidation mechanism was quasi-reversible while the CuBrÿ
2 species
were the only oxidation product, and it was not
a€ected by the presence of HMTA. The inhibitory
action was observed for [HMTA] r 3.0 mM. The
anodic process was not di€usion controlled in the
entire polarization curve for these HMTA concentrations. The passive ®lm was characterized by ¯uor-
escence spectroscopy as being a Cu(I)/HMTA/Brÿ
complex.
The objectives of this work are to understand some
aspects of the mechanism for CuBr and Cu/HMTA/
Brÿ ®lms formation during the copper dissolution process, and further characterize the nature of the passivant ®lm.
2. Experimental
2.1. Electrodes
The working electrode was a 0.50 cm diameter rotating disk electrode built with an electrolytic copper
plate (99,9% purity). It was mechanically polished
with 300, 400 and 600 mesh wet emery-paper and
rinsed with double distilled water, prior to use. A
cylindrical polytetra¯uorethylene sleeve was pressure
®tted onto the rotating disk electrode. A complete
description of the electronic equipment used is given
elsewhere [28]. A platinum foil was used as the auxiliary electrode, and a saturated calomel electrode (SCE)
as the reference electrode.
2.2. Solutions
The electrolyte solutions were prepared from analytical grade reagents and double distilled water. HMTA
was reagent grade and it was used without further
puri®cation. The pH values of the electrolyte solutions
containing HMTA were adjusted to 3.0 with HBr aqueous solution. Bubbling of the solution with N2 was
carried out for 30 min before the measurements, and a
weak stream was maintained above the solutions
during the experiments. All experiments were performed at (25 2 1)8C.
2.3. Electrochemical instrumentation
The electrochemical system used was a PAR (potentiostat±galvanostat) model 273 coupled to an IBM personal computer which contain the data acquisition and
the data treatment software. The curves were recorded
on a HP 7090A plotter.
2.4. SEM and EDX instrumentation
Environmental scanning electron microscope model
E-3 was used for SEM analysis. The electron gun was
operating at 20 kV. The cathode ®lament current was
set to 1.37 A. The instrument is equipped with an
Oxford link EDX analyzer model 6580 and silicon
detector. In contrast to an older EDX analyzer, this
model is supplied with a state of art thin electron win-
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
561
Fig. 2. Cyclic voltammograms for a stationary copper electrode in 0.1 M KBr, pH = 3 at scan rate = 10 mV/s; (A) [HMTA] = 6
mM; (B) [HMTA] = 15 mM.
dow ATW2. This thin window allows the quantitative
elemental analysis of light elements such as nitrogen.
2.5. Raman instrumentation
In situ Raman spectra from the copper electrode
were obtained with a Renishaw 1000 Raman microscope system equipped with a Peltier-cooled CCD
detector and a holographic notch ®lter. A macropoint
accessory, with 40 mm of focal length, was used to
redirect the laser light to the electrode surface. The
spectroelectrochemical cell has been described
elsewhere [29].
A 35 mW Melles Griot He±Ne laser provided excitation at 632.8 nm. The laser power was ca. 6 mW at
the sample.
Fig. 1. Cyclic voltammogram for a stationary copper electrode in 0.1 M KBr, pH = 3 at scan rate = 10 mV/s; in the
absence of HMTA.
3. Results and discussion
A typical cyclic voltammogram for a stationary Cu
electrode immersed in 0.1 M KBr solution, pH = 3.0,
is presented in Fig. 1. Two anodic peaks, labeled A1
and A2, and two cathodic peaks, C1 and C2, were
observed. According to Brossard [11], the formation of
soluble CuBrÿ
2 species occurs before the peak A1. The
®rst anodic peak (A1) is due to the formation of a
CuBr precipitate on the electrode surface. After this
®rst saturation the electrochemical reaction continues
to occur through the porous ®lm and a second anodic
peak, labeled A2, is observed (Fig. 1). The CuBr formation is still the predominant process in the A2
region [12]. The cathodic peak C1 is related to the reduction of these surface species. The cathodic peak C2
was not reported before [12].
Figs. 2A and 2B show the voltammograms for the
copper dissolution in acidic bromide medium (pH = 3)
in the presence of 6.0 and 15.0 mM HMTA (stationary
electrode), respectively. For 6 mM HMTA (Fig. 2A),
at sweep rate equal to 10 mV/s, the voltammogram
was similar to the one obtained without the inhibitor
(Fig. 1), except by the presence of an anodic pre-peak,
App (identi®ed in Fig. 2A). The App peak was
strongly dependent on the HMTA concentration. The
current ratio, IApp/IA1, was higher for more concentrated HMTA solutions (Fig. 2B). The passivant e€ect
was also more evident for copper dissolution in bromide medium (pH = 3) containing 15 mM of HMTA
(Fig. 2B). The cathodic region observed in Fig. 2B presents several peaks indicating the reduction of di€erent
kinds of ®lms.
562
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
Table 1
Surface chemical analysis (by EDX). The results are
presented in mol fraction (X)
Region
Nitrogen
Bromine
Copper
®lm
Patches
0.0691
Ð
0.0776
0.0420
0.8365
0.9333
The results observed in Figs. 1 and 2, suggest that
the pre-peak, App, corresponds to the Cu(I)/HMTA/
Brÿ passive layer formation, and the peak A1 corresponds to the CuBr ®lm. The Cu(I)/HMTA/Brÿ ®lm is
formed at low positive overpotentials [27], and the
CuBr ®lm is formed either through the pores of the
®rst ®lm or on surface regions not completely covered
by the HMTA-containing ®lm. The increase of the
ratio IApp/IA1 with the increase of the HMTA concentration (Figs. 2A and 2B) agrees with this assignment
(more HMTA in solution seems to favor the formation
of the species at App).
SEM measurements were performed in order to con®rm these voltammetric results. The SEM image,
obtained from a Cu electrode submitted to a similar
condition as presented in the anodic part of Fig. 2A,
showed a ®lm on the electrode surface with few
patches. This result indicate that part of the electrode
was not covered by the inhibitory ®lm. Elemental
chemical analysis on the ®lm and inside the patches
were performed. The results are summarized in Table 1.
The analysis of the ®lm showed a fair amount of nitro-
Fig. 3. Cyclic voltammograms for a stationary copper electrode in 0.1 M KBr, pH = 3.0 at several reversal scanning
potential values (Er). Scan rate = 10 mV/s; in the absence of
HMTA.
gen, which indicates the presence of HMTA. On the
other hand, chemical analysis of the patches showed
mainly Cu and Br.
Fig. 3 shows the cyclic voltammograms for a
stationary copper electrode in 0.1 M bromide medium
(pH = 3), in the absence of inhibitor, at several reversal scanning potentials values, Er (indicated in Fig. 3),
at di€erent positive limits. In the cathodic region, the
peak C1 was observed at ÿ190 mV for Er= ÿ 100 mV.
At this Er value, the formation of the soluble species
(CuBrÿ
2 ) is expected [12, 27]. The cathodic peak C1
shifted to more negative potential values as the Er
became more positive (Fig. 3). When peak C1 reached
ca. ÿ400 mV (for Err + 100 mV), there were no
further signi®cant changes in both its potential and its
current values. This is an indication that the CuBr
®lm, formed in the surface at positive potentials (A1
region), is reduced at C1. The cathodic peak C2
appeared only after the Er values reached +200 mV.
The current of the cathodic peak C2 increased as the
Er values become more positive than +200 mV. Two
anodic peaks, A1 at ca. +30 mV and A2 at ca. +400
mV, were observed in the anodic part of the voltammograms. The ®lm breakdown, probably due to a pitting process [30], was noticed (Fig. 3) when the Er
values were made more positive than +600 mV. The
evidence for this ®lm disruption was the high values of
anodic current observed in the reversal part of the voltammogram (cathodic sweep) for Er values higher than
+600 mV (Fig. 3). The localized attack was con®rmed
by SEM. The SEM images from a Cu electrode submitted to a sweep scan to +600 mV, in the conditions
presented in Fig. 3, showed pits of ca. 20±50 mm of diameters. Similar pit formation processes have been
observed for copper dissolution in chloride and sulfate
media [31].
The observations from Fig. 3 indicate that the copper anodic dissolution products (in 0.1 M bromide
medium and pH = 3 in the absence of HMTA) formed
in A1 and A2 are reduced in C1 and C2, respectively.
This results contradicts a previous work [12] which
claims that the species formed in both A1 and A2 are
reduced together in the cathodic region C1. However,
cathodic potentials studied in Ref. [12] were not negative enough to allow the observation of the peak C2. It
seems evident that the peak A1 is related to the formation of CuBr ®lm which was reduced at C1. This
hypothesis is in agreement with the experimental data
presented here and with other prior works [11±14, 27].
The nature of the peaks A2 and C2 are unclear, and it
will be discussed in the following paragraphs.
Following the mechanism for copper oxidation in
chloride medium, one can suggest that the anodic peak
A2 is due to the formation of Cu(II) ions [16].
Nevertheless, the oxidation of the CuBr ®lm is not the
main reaction that occurs at the anodic peak A2. This
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
conclusion is supported by the following observations:
®rst of all, the anodic peak A2 (at potential more positive than A1, see Fig. 1) was clearly related to the
cathodic peak C2 that appears at potentials more
negative than C1 (Fig. 3). Therefore, the specie formed
in A2 was kinetically more stable than the species
formed in A1. Besides, the Cu(II) species are generally
more soluble than Cu (I) derivatives, therefore, the
peak C2 should decrease when the electrode was
rotated (RDE experiments), which was not observed.
The Pourbaix diagram for copper in 1 M bromide
medium [26] predicts that the formation of Cu(II)
starts at potential more positive than +800 mV versus
SCE. Finally, the Cu(II) ions are blue in aqueous medium and this color was noticed only when the potential
was swept to values more positive than those showed
in Figs. 1 and 3.
Another possibility is that the peak A2 is due to a
copper oxide formation. Copper oxides are very stable
and the ®lm could remain in the surface even at very
negative potentials. However, the formation of copper
oxides is not expected for copper dissolution in acidic
medium free of oxygen. In fact, the equilibrium potential-pH diagram (Pourbaix diagram) for copper in
chloride medium shows that, at lower pH, the oxidation of CuCl should lead to the formation of Cu2+
ions instead of copper oxides [32]. The same is
observed in the Pourbaix diagram for copper in 1 M
bromide solution [26].
The mechanism for copper dissolution in bromide
medium (presented in Section 1) suggest that CuBr is
the predominant species in both regions A1 and
A2 [12]. Therefore, there are two possible explanations
for the nature of the anodic peak A2 and the cathodic
peak C2 (not considering the possibility of Cu(II) and
copper oxides formation). First of all, one can consider
that a CuBr ®lm (type I) is formed in A1. This type I
®lm must be porous and permits the reaction to continue to occur through the pores, leading to the formation of another kind of CuBr ®lm (type II) with
Fig. 4. Cyclic voltammograms for a stationary copper electrode in 0.1 M KBr, pH = 3, [HMTA] = 15 mM at several
reversal scanning potential values (Er). Scan rate = 10 mV/s.
563
Fig. 5. E€ect of the sweep rate on the cyclic voltammogram
for a stationary copper electrode in 0.1 M KBr, pH = 3 and
15.0 mM HMTA. The sweep rates are indicated in the ®gure.
di€erent morphology, which is responsible for the
anodic peak A2. The type II ®lm is more stable and it
is reduced at potentials more negative than the cathodic peak C1. Another possibility takes into account
that A2 is related to the limiting current of the CuBr
®lm dissolution (3rd step in the mechanism for dissolution of a copper electrode covered with CuBr presented in Section 1). In this case, the C2 peak must be
connected to the reduction of the species possibly generated during the localized attack of copper.
Fig. 4 presents the voltammograms for the dissolution of a stationary copper electrode in 0.1 M bromide (pH = 3) in the presence of 15 mM HMTA, at
several Er values. One can observe from Fig. 4 that the
cathodic region become complex as the Er values were
made more positive than the potential of the anodic
peak A1. The complex cathodic region suggest the reduction of mixed ®lms formed in the anodic sweep.
Fig. 5 shows the dependence of the cyclic voltammograms for copper in bromide medium (pH = 3), in the
presence of 15 mM HMTA, on the sweep rate. It can
be noticed that the inhibitor was more ecient at low
sweep rates (1 mV/s). Almost total passivation was
observed for the voltammogram at 1 mV/s (the anodic
current is close to zero after the anodic peaks, Fig. 5).
The ratio of the anodic currents IApp/IA1 was higher
for the voltammogram obtained at 1 mV/s. The ratio
IApp/IA1 decreased as the sweep rate increased. The
small value of the ratio, IApp/IA1, for high sweep rates,
suggest that the specie responsible for the peak App
can not be well formed at fast potential scans. The
inverse relationship between IApp/IA1 and the sweep
rate indicates a slow kinetics for the formation of the
species responsible for the App peak. In fact, a slow
formation rate for the Cu(I)/HMTA/Brÿ ®lm has been
demonstrated in our previous work [27].
Fig. 6 shows the dependence of the cyclic voltammograms of copper in 0.1 M bromide medium (pH = 3)
containing 15.0 mM HMTA on the stirring of the solution. As it can be observed in Fig. 6, a more ecient
564
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
Fig. 6. E€ect of mass transport on the voltammogram for a
copper electrode in 0.1 M KBr, pH = 3 and 15 mM HMTA.
The rotation frequencies are shown in the ®gure.
protective ®lm is formed on the electrode which was
rotated at 16 Hz. The results presented in Fig. 6 indicates that the mass transport favors the inhibition process. In the voltammogram under stirring (f = 16 Hz),
only one cathodic peak at ca. ÿ360 mV and one
anodic peak at ca. ÿ104 mV was observed. Besides,
peak A1 did not appear in the voltammogram under
stirring. These results suggest that, for the system
under stirring, only a compact Cu/HMTA/Brÿ ®lm is
formed. In contrast, the voltammogram of the nonstirred solution (f = 0 Hz in Fig. 6) presents several
cathodic peaks due to the reduction of mixed ®lms,
and the anodic current increased as the potential
become more positive than +200 mV.
Chronoamperometry can be used for ®lm formation
mechanism studies. A theoretical model for multiple
nucleation phenomena followed by di€usion controlled
growth of three dimensional center was developed [33].
The nucleation process can either be progressive or
instantaneous [33]. This nucleation, growth and superposition (NGS) model has been successfully applied to
several systems [34, 35].
Chronoamperograms were obtained by stepping the
potential from ÿ1.0 V (Einitial) to di€erent positive potential limits (E®nal). Fig. 7 shows the chronoampero-
grams for copper in 0.1 M bromide medium (pH = 3),
in the absence of HMTA, obtained at several E®nal
values (the E®nal values are indicated in Fig. 7).
An exponential current decrease until a constant residual value was obtained for low overpotentials
(E®nal= ÿ 100 mV in Fig. 7). This behavior is expected
for an oxidation process with soluble products controlled by mass transport [36]. It has been demonstrated that the soluble CuBrÿ
2 is the dominant species
at this overpotential [11, 12, 27]. When the E®nal values
were made more positive than +100 mV, the chronoamperograms presented the characteristic shape
expected in a ®lm formation process. For
E®nal> + 100 mV, a increase in the current was
observed at the beginning of the chronoamperogram
(initial transient). After a time tm, the current reaches
a maximum value originating a current peak (Im),
from which the current starts to decrease. This kind of
result characterizes the classical behavior of a ®lm formation process by NGS mechanism [33].
The chronoamperograms for copper in 0.1 M bromide (pH = 3), in the presence of 15 mM HMTA, are
presented in Fig. 8. The chronoamperograms in the
presence of HMTA can be divided in two distinct sets.
The ®rst set (set I) comprise the chronoamperograms
obtained for E®nalR + 400 mV (Fig. 8). These curves
presented an initial increase in the current, followed by
complete passivation (current drops to zero). On the
other hand, a second set of chronoamperograms (set
II) was obtained for E®nal values more positive than
+400 mV. In this case, the shape of the chronoamperometric curves are very similar to the ones obtained
in Fig. 7 (in the absence of HMTA). All chronoamperograms obtained in the presence of 15 mM HMTA
(Fig. 8), presented values for Im lower than the one
observed in the absence of the inhibitor (Fig. 7). The
maximum current peaks were also reached at lower tm
values in the presence of 15 mM HMTA. The set I-
Fig. 7. Chronoamperograms for a stationary copper electrode in 0.1 M KBr, pH = 3. Einitial= ÿ 1.0 V; E®nal values are indicated
in the ®gure.
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
565
Fig. 8. Chronoamperograms for stationary copper electrode in 0.1 M KBr, pH = 3 with 15 mM HMTA. Einitial= ÿ 1.0 V; E®nal
values are indicated in the ®gure.
type chronoamperograms in Fig. 8 (for E®nalR + 400
mV) show rapidly decrease in the current to a value
practically equal to zero after the Im peak, indicating
that, in the presence of HMTA, the formed ®lm was
not porous. This ®lm completely blocked the surface
and prevent the copper oxidation reaction. For
E®nalr + 600 mV (set II curves in Fig. 8), the chronoamperograms were similar to the ones without the inhibitor, suggesting that there was a competition
between the formation of CuBr and Cu(I)/HMTA/Brÿ
®lms. The porous CuBr ®lm was formed preferentially,
even in presence of HMTA, when E®nalr + 600 mV,
due to the slow kinetics of the Cu/HMTA/Brÿ passive
®lm formation. These results are in accordance with
those obtained by cyclic voltammetry.
The mechanism for the ®lm formation can be veri®ed by comparing the experimental chronoamperogram with theoretical curves, for progressive and
instantaneous nucleation, considering a NGS-type
mechanism, using the expressions [33]:
2
I2
t
t
ˆ 1:9542
1 ÿ exp ÿ 1:2564
tm
tm
I 2m
for instantaneous nucleation, and
2 2
I2
t
t
ˆ 1:2254
1 ÿ exp ÿ 2:3367
tm
tm
I 2m
for progressive nucleation.
The experimental (from Fig. 7, for copper in 0.1 M
bromide, pH = 3, without HMTA) and the theoretical
Fig. 9. Comparison between the NGS theoretical models and experimental results for copper in 0.1 M KBr, pH = 3;
E®nal= + 200 mV. (a) experimental data; (b) calculated chronoamperogram, using a NGS model with instantaneous nucleation; (c)
calculated chronoamperogram, using a NGS model with progressive nucleation.
566
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
Fig. 10. Comparison between the initial transients of chronoamperograms for copper in 0.1 M KBr (pH = 3.0). (a) absence of
HMTA; (b) presence of 15 mM HMTA. E®nal= + 200 mV.
curves relating to dimensionless parameters (I/Im)2 and
(t/tm) are presented in Fig. 9. Curve A shows the experimental chronoamperogram for the system without
inhibitor (from Fig. 7), while curves B and C are the
calculated results for instantaneous and progressive
nucleation process, respectively. The result in Fig. 9
clearly shows that the CuBr ®lm formation occurs
through a NGS mechanism. The nucleation is instantaneous, and the growth is controlled by di€usion.
This result agrees with previous work which consider
that the ®lm formation is controlled by the mass transfer of bromide ions [11]. The electrocrystalization of
other copper dissolution products have also been
found to follow the NGS mechanism [37]. The di€usion coecient of bromide was also determined from
the NGS model (see Refs. [34] and [35] as examples of
the procedure). The average value for the di€usion
coecient of bromide was ca. 1.6 10ÿ5 cm2/s, which
was in fair agreement with voltammetric determinations (1.8 10ÿ5 cm2/s in Ref. [11]).
Unfortunately, a similar quantitative treatment was
not possible for the chronoamperograms in the presence of HMTA (Fig. 8). Despite the curves showed in
Fig. 8 for E®nalr + 600 mV (set II) being very similar
to the ones obtained in the absence of inhibitor, a critical analysis of these curves becomes dicult due to the
supposedly presence of a second ®lm of CuBr, which
competes with the formation of Cu/HMTA/Brÿ. The
shape of the chronoamperograms obtained in the presence of HMTA for E®nalR + 400 mV (set I in Fig. 8)
does not correspond to the NGS model with di€usion
controlled growth. These curves suggest a two dimensional nucleation-superposition process rather than
three dimensional. In this case, the Cu/HMTA/Brÿ
nuclei are formed and grow parallel to the surface,
forming a compact, non-porous, layer. According to
the voltammetric data presented before, the nuclei
growth is controlled by di€usion. As the superposition
occurs, the surface is completely blocked, leading to
immediate decrease in the total current to zero.
Another interesting feature appears when the chronoamperograms in the absence (Fig. 7) and in the presence of HMTA (Fig. 8) are compared at short time
intervals. Fig. 10 presents this comparison for
E®nal= + 200 mV. According to Fig. 10, for t < 0.05
s, the anodic current in the presence of the inhibitor
was higher than the current obtained in the absence of
HMTA. This behavior con®rms the complexing e€ect
of HMTA observed before [27], and the low rate for
the passive ®lm formation. The current values presented in Fig. 10 are practically constant in both
curves for t>0.4 s.
The electric charges can be extracted from the
chronoamperograms, and the eciency y of the inhibi-
Fig. 11. Dependence of inhibitory eciency o HMTA on the applied potential (E®nal). The electric charges were calculated using
the chronoamperometric measurements presented in Figs. 7 and 8.
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
tor can be calculated using the following expression:
yˆ
Q0 ÿ QHMTA
:
Q0
Where Q0 and QHMTA are the electric charges in the
absence and presence of HMTA, respectively.
Fig. 11 shows the dependence of the eciency of the
inhibitory action (y values), obtained (by the integration of the chronoamperograms) for 6.0 and 15
mM HMTA, on the ®nal potential (E®nal). A much
higher eciency for more concentrated HMTA solutions (15 mM HMTA) up to +400 mV was observed.
For E®nal values more positive than +400 mV the inhibitor eciency decreases sharply. For HMTA 6.0
mM, the inhibiting eciency increases with the E®nal,
and above +400 mV it falls down to approximately
zero. These results con®rmed the existence of a compe-
567
tition between the CuBr and Cu/HMTA/Brÿ ®lms at
higher overpotentials. The negative y at zero volts for
6.0 mM (Fig. 11) is related to the complexing e€ect of
the inhibitor [27].
More information about the nature of the passivant
®lm was obtained by surface Raman scattering experiments. This technique also provided some insights
into the understanding of the copper oxidation in the
presence of the inhibitor.
The normal coordination analysis and vibrational
assignment for HMTA have been performed [38]. SER
spectra of HMTA adsorbed on a silver electrode have
also been reported [39]. The in situ surface-enhanced
Raman (SER) spectra from a copper electrode
immersed in a solution containing 1 mM HMTA (0.1
M KBr, pH = 3) at di€erent potentials (before and
after the ®lm formation) are presented in Fig. 12, and
Fig. 12(a).
568
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
Fig. 12. Surface-enhanced Raman spectra of a copper electrode immersed in 0.1 M KBr + 1 mM HMTA solution at several potentials. Region 1: from 300 to 1500 cmÿ1. Region 2: from 2700 to 3100 cmÿ1. Applied potentials: (a) ÿ1.2 V; (b) ÿ0.4 V; (c) +0.4 V.
the vibrational frequencies and assignments are presented on Table 2.
Several striking di€erences can be observed in the
spectra presented in Fig. 12 for di€erent applied potentials. The monoprotonated cation (HMTAH+) is the
predominant species in solution throughout our experiments (pH = 3), considering that HMTA is a moderate base ( pKa=6.2). However, at very negative
potential (E = ÿ 1.2 V, Fig. 12a), the SER spectrum
closely resembled the one observed for adsorbed
HMTA on metallic surface [39], with two strong peaks
at 778 and 1048 cmÿ1. The protonation of one of its
nitrogen dramatically decreases the basicity of other
nitrogen in the HMTA molecule. Being so, the monoprotonated HMTA can only adsorb to the electrode
surface by electrostatic interactions (not by acid±base
reaction). The amount of speci®c adsorbed bromide at
ÿ1.2 V is expected to be small, which avoids the possibility of an ion-pair interaction. Therefore, the HMTA
molecule interacts directly with the electrode surface at
this potential. The SER spectrum (Fig. 12b) at ÿ400
mV (before the copper oxidation) shows an overall
shift and splitting in several peaks. Besides, the 778
and 1048 cmÿ1-bands' intensities decreased and new
strong bands at 794, 1022 and 1061 cmÿ1 appeared. A
strong ca. 1026 cmÿ1 band is a ®ngerprint for the
monoprotonated HMTA cation [39]. These spectral
changes suggest that, as the potential was made more
positive, the amount of speci®cally adsorbed bromide
anions increased, which made the ion-pair interaction
between HMTAH+ and Brÿ possible. The adsorption
of HMTA and HMTAH+ before the copper oxidation
are illustrated in Fig. 13.
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
569
Table 2
Vibrational frequenciesa (in cmÿ1) and assignment for HMTA adsorbed on the copper electrode
E = ÿ 1.2 V
453
514
704
778
797
(m)
(m)
(w)
(s)
(m, sh)
996 (w)
1012 (w)
E = ÿ 0.4 V
E = + 0.4 V
150 (b, s)
150 (b, vs)
360 (w)
443 (m)
500 (m), 512 (m)
692 (sh, w), 700 (w)
760 (m)
792 (s)
848 (m), 854 (m)
888 (m)
959 (m), 970 (m)
996 (b, w), 1014 (m)
1025 (m)
1053 (m)
1060 (m), 1066 (m)
1177 (s)
1200 (s)
1250 (m), 1260 (m)
1272 (w)
1302 (m), 1324 (m)
1340 (s)
1354 (m), 1357 (m)
1394 (m)
1438 (m)
443
500
696
764
794
810
886
970
(m)
(m), 516 (w)
(m)
(b, s)
(sh, s)
(m)
(b, m)
(b, m)
1236 (m)
1022
1049
1061
1190
1205
1254
1327 (w)
1341 (m)
1306 (b, w)
1340 (b, w)
1371
1455
1474
1571
2896
1391 (b, w)
1450 (b, w)
1048 (vs)
(m)
(m)
(w)
(w)
(b, s)
2932 (s)
2959 (s, sh)
(b,
(s)
(s)
(b,
(b,
(b,
s)
w)
w)
w)
2907 (sh, s)
2948 (s)
2971 (sh, s)
2874
2900
2942
2969
3000
(b, w)
(b, w)
(s), 2951 (m)
(s), 2994 (sh)
(s), 3006 (sh, s)
Literature
Assignment
378c
456b
512b
688b
783b
n (Cu±Br)
n16 (F1)
n10 (E)
n25 (F2)
n24 (F2)
n4 (A1)
818b
[911]
960b
1011b
1026b
1050b
n23 (F2)
n15 (F1)
comb.
n22 (F2)
n9 (E)
n3 (A1)
[1179]
n14 (F1)
1240b
[1280]
[1346]
1350b
1381b
[1401]
1458b
1458c
[1460]
[2869]
2873c
2883c
[2952]
2955c
n21 (F2)
n5 (A2)
n13 (F1)
n8 (E)
n20 (F2)
n12 (F1)
n7 (E)
n19 (F2)
n2 (A1)
n6 (E)
n18 (F2)
n1 (A1)
n11 (F1)
n17 (F2)
a
Intensities: v Ð very; w Ð weak; m Ð medium; s Ð strong; b Ð broad; sh Ð shoulder.
Frequencies from solution spectrum presented in Ref. [39].
c
Frequencies from Ref. [38]. Calculated frequencies (from Ref. [38]) are presented in square
brackets. The assignment is presented considering a Td symmetry.
b
Fig. 12c presents the in situ spectrum of the Cu/
HMTA/Brÿ ®lm formed at +0.4 V. The splitting of
several bands is more evident in this spectrum. HMTA
is a cage-like compound with Td symmetry. The coordination of one of its nitrogen provoke a decrease in
the overall symmetry to C3v. Using the correlation
table technique [40], it is possible to predict that the
degenerated F1 and F2 modes should split into A2+E
and A1+E, respectively. The A2 modes are Raman
inactive under C3v symmetry. The spectrum presented
on Fig. 12c is consistent with a monocoordinated
HMTA complex. The bands at 760, 1060, 1066 and
2931 cmÿ1 are probably from HMTAH+.
4. Conclusions
The results presented in this work for the copper
oxidation in bromide medium, in the absence of the in-
hibitor, indicate a di€usion controlled process at low
anodic overpotentials, while there is the formation of
soluble products. At higher overpotentials, a porous
CuBr ®lm is formed. The CuBr is the main product
for the copper oxidation in bromide medium for overpotentials as high as +800 mV (versus SCE). Two
anodic peaks were observed in the voltammogram,
which may correspond to two types of CuBr ®lms with
distinct morphology. Evidence from localized dissolution of the CuBr ®lm at potentials more positive than
+200 mV was observed by cyclic voltammetry and
SEM. The CuBr ®lm formation follows a growth
mechanism that corresponds to a NGS model, with an
instantaneous nucleation and a di€usion (of bromide
ions) controlled growth.
The HMTA inhibits the copper oxidation process in
bromide medium through the formation a Cu(I)/
HMTA/Brÿ ®lm. This ®lm is formed at anodic potential values more negative than the potentials where the
570
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
HMTAH ‡ „ HMTA ‡ H ‡ 2nd: step
ÿ
CuBrÿ
2 ‡ HMTA ÿ4Cu…HMTA†Br ‡ Br 3rd: step
Cu ‡ Brÿ „ CuBr ‡ eÿ 4th: step:
For low overpotentials and slow sweep rates, the
steps 1, 2 and 3 are important. In this case, the ®rst
step seems obvious, because, according to our SERS
results, the bromide ions are directed adsorbed to the
electrode surface. The equilibrium presented in step 2
is also observed from the SERS experiments. At low
overpotentials, and slow sweep rates, the soluble
CuBrÿ
2 can either di€use to the bulk solution or react
with HMTA to form a protective ®lm. The fourth step
became important at higher overpotentials. In these
cases (more anodic potentials or fast sweep rates), the
CuBr ®lm formation start to compete with the Cu/
HMTA/Brÿ. Evidences for multiple ®lm formation in
these situations was observed by cyclic voltammetry,
chronoamperometry and SEM/EDX.
Fig. 13. Models for the adsorption of HMTA and HMTAH+
on a copper electrode in 0.1 M Brÿ medium before the copper
oxidation. (a) without speci®c bromide adsorption (E = ÿ 1.2
V); (b) with speci®c bromide adsorption (E = ÿ 0.4 V).
CuBr ®lm is formed, indicating a complexing e€ect of
HMTA on Cu(I) ions. The inhibitory e€ect of HMTA
on copper dissolution was observed for potential
values between 0 and +400 mV for [HMTA] = 6 mM
and between 0 and +800 mV for [HMTA] = 15 mM.
For [HMTA] = 15.0 mM, the inhibitory eciency was
near 100% in the potential range between 0 and +400
mV (versus SCE). The passive HMTA ®lm has a slow
growth rate when compared to the kinetics of formation of the CuBr ®lm. In fact a competition between
both types of ®lm occurs. The mixture of these ®lms
on the electrode surface was determined by SEM/EDX
measurements. The inhibitory HMTA ®lm formation
is also strongly dependent on mass transport. The
SERS results showed that a HMTAH+±Brÿ ion-pair is
speci®cally adsorbed on the electrode surface prior to
the oxidation. Therefore, the deprotonation of the
HMTA cation is a mandatory step before the ®lm formation. This deprotonation process may be related to
the slow kinetics of formation for the inhibitory ®lm.
Considering that the CuBrÿ
2 formation is the ®rst step
for the oxidation, the copper dissolution process in the
presence of the inhibitor may follow the steps presented below:
ÿ
Cu ‡ 2Brÿ „ CuBrÿ
2 ‡ e 1st: step
Acknowledgements
This work was supported by FAPESP and CNPq.
The authors thank for the grant of research fellowships. A.G. Brolo also thanks Dr. Marek
Odziemkowski for the help with the SEM measurements and Professor Donald E. Irish to allow access to
the Raman instrumentation.
References
[1] W.H.
Smyrl,
Comprehensive
Treatise
of
Electrochemistry, J.O'M. Bockris and R.E. White (Eds.),
Vol 4, Ch. 2, Plenum Press, New York, 1981, p. 97.
[2] I.T.E. Fonseca, A.C.S. Marin, A.C. Sa, Electrochim.
Acta 37 (1992) 2541.
[3] E. Potvin, M. Drogowska, H. Menard, L. Brossard,
Can. J. Chem. 65 (1987) 2109.
[4] F.K. Crundwell, Electrochim. Acta 37 (1992) 2707.
[5] C. Deslouis, B. Tribollet, G. Mengoli, M.M. Musiani, J.
Appl. Electrochem. 18 (1988) 374.
[6] D.B. Hibbert, S. Richards, V. Gonzales, Corrosion Sci.
30 (1990) 367.
[7] P.A. Kilmartin, G.A. Wright, J. Chem. Soc. Faraday
Trans. 91 (1995) 4403.
[8] M. Drogowska, L. Brossard, H. MeÂnard, J. Electrochem.
Soc. 139 (1992) 39.
[9] Y.P. Feng, S.K. Sinha, C.A. Melendres, D.D. Lee,
Physica B 221 (1996) 251.
[10] Y. Feng, K.-S. Siow, W.-K. Teo, K.-L. Tan, A.-K.
Hsieh, Corrosion 53 (1997) 389.
[11] R.L. Brossard, J. Electrochem. Soc. 131 (1984) 1847.
[12] R.L. Brossard, Can. J. Chem. 62 (1984) 1112.
A.G. Brolo et al. / Electrochimica Acta 44 (1998) 559±571
[13] R.L. Brossard, Can. J. Chem. 62 (1984) 36.
[14] R.L. Brossard, G.M. Raynaud, Can. J. Chem. 63 (1985)
720.
[15] H.P. Lee, K. Nobe, J. Electrochem. Soc. 133 (1986)
2035.
[16] H.P. Lee, K. Nobe, J. Electrochem. Soc. 131 (1984)
1236.
[17] S.F.L.A. Costa, S.M.L. Agostinho, J. Electroanal. Chem.
296 (1990) 51.
[18] V. Brusic, M.A. Frischi, B.N. Eldridge, F.P. Novak, F.B.
Rush, B.M. Rush, G.S. Frankel, J. Electrochem. Soc.
138 (1991) 2253.
[19] S.L.F.A. Costa, S.M.L. Agostinho, J. Electroanal. Chem.
296 (1990) 51.
[20] S.L.F.A. Costa, S.M.L. Agostinho, J.C. Rubin,
Corrosion 45 (1989) 472.
[21] D. Tromans, J.C. Silva, Corrosion 53 (1997) 17.
[22] A.M. Breccaria, C. Bertolotto, Electrochim. Acta 42
(1997) 1361.
[23] M. Beier, J.W. Schultze, Electrochim. Acta 37 (1992)
2299.
[24] P. Schweinsberg, S.E. Bottle, V. Otieno-Alego, T.
Notoya, J. Appl. Electrochem. 27 (1997) 161.
[25] M.R. Vogt, W. Polewska, O.M. Magnunssen, B.J. Behm,
J. Electrochem. Soc. 144 (1997) L113.
[26] T. Aben, D. Tromans, J. Electrochem. Soc. 142 (1995)
398.
571
[27] A.G. Brolo, M.L.A. Temperini, S.M.L. Agostinho, J.
Electroanal. Chem. 335 (1992) 83.
[28] H.C. Chagas, Tese de Livre Docencia, Instituto de
QuõÂ mica, USP, Brazil, 1987.
[29] A.G. Brolo, B.D. Smith, D.E. Irish, J. Mol. Struct. 405
(1997) 29.
[30] H. Bohni, Langmuir 3 (1987) 924.
[31] G. Mankowski, J.P. Duthill, A. Giusti, Corrosion Sci. 39
(1997) 27.
[32] U. Bertocci and D.R. Turner, Encyclopedia of
Electrochemistry of the Elements, A.J. Bard (Ed.), Vol.
II, Ch. II(6), Dekker, New York, 1974, p. 383.
[33] B. Scharifker, G. Hills, Electrochim. Acta 28 (1983) 879.
[34] R.T. Carlin, W. Crawford, M. Bersh, J. Electrochem.
Soc. 139 (1992) 2720.
[35] C.L. Hussey, X. Xu, J. Electrochem. Soc. 138 (1991)
1886.
[36] A.J. Bard and L.R. Faulkner, Electrochemical Methods
± Fundamental and Applications, Willey, New York,
1980.
[37] B. Scharifker, R. Rugeles, J. Mozota, Electrochim. Acta
29 (1984) 261.
[38] J.E. Bertie, M. Solinas, J. Chem. Phys. 61 (1974) 1666.
[39] J.R. Bertolino, M.L.A. Temperini, O. Sala, J. Mol.
Struct. 176 (1988) 113.
[40] E.B. Wilson, Jr., J.C. Decius and P.C. Cross, in:
Molecular Vibrations ± The Theory of Infrared and
Raman Vibrational Spectra, McGraw Hill, NY, 1955.