PH On Copper
PH On Copper
PH On Copper
Received 15.08.2007
Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry were used to investigate the oxide
layer formed on a copper disc electrode and the changes that took place when treated potentiostatically in
the range of -0.3 V to 0.9 V in aqueous buffer solution of pH 9.2. Equivalent circuits were used to model
the response of the electrode, initially at equilibrium to an applied potential. These circuits were proposed
for different potential ranges in order to illustrate the Cu/oxide/electrolyte system and its properties in
terms of 2 interfaces. A criterion for the applicability of equivalent circuit models was discussed. Changes
in the film/metal interface as a function of potential were probed at 30 mHz from Nyquist plots. Diffusion
coefficient calculated for the ionic movement in the film at 2 potential values using EIS data was of the order
of 10 −9 cm 2 s −1 .
Key Words: Copper, electrochemical impedance spectroscopy, cyclic voltammetry, passive film, interfaces.
Introduction
Corrosion of copper occurs in presence of oxygen when in contact with electrolytes. 1 Thin film formed over
copper in anodic conditions in neutral, 2 weakly acidic, or alkaline aqueous 3 media has attracted considerable
interest of researchers studying corrosion, electro catalysis, and double layer structures. 4 Passive layers on
copper metal have been widely investigated 5−8 and several surface analytical and electrochemical tools have
been used. 9−13 Some of the obvious factors affecting the thickness and structure of the oxide film are pH,
potential, and the time of contact with the aqueous environment.
∗ Corresponding author
739
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
740
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
The current peak A 1 was due to the electroformation of Cu 2 O barrier layer (Eq. 1) and the anodic formation
of a complex hydrous CuO layer (Eqs. 3 and 2), which was overlaid upon the barrier layer and gave a duplex
structure to it. The latter layer was responsible for the decrease in anodic dissolution of copper. The 2
cathodic peaks were unquestionably due to electroreduction of CuO and Cu(OH) 2 to Cu 2 O, and Cu 2 O to Cu
respectively. 20
The anodic peak appeared broader and dragged in the anodic direction. The appearance of more than
one anodic peak in the cyclic voltammogram was pH and sweep rate dependent. 10,16 That only one anodic peak
observed was not surprising because of the relatively high scan rate used in this study. Even a scan rate of 20
mV s −1 did not resolve anodic peaks. 20 Strehblow and Titze 10 obtained 2 well resolved anodic peaks with a
distinct broad shoulder to higher potential peak at 0.1 mV s −1 scan rate. However, in the same system (pH 9.2
borate buffer) at 20 mV s −1 sweep rate although 2 anodic peaks were observed, the higher potential peak was
weaker and dragged. Therefore, the observed anodic peak in the present study is assigned to electroformation
of Cu 2 O and complex CuO/Cu(OH) 2 .
741
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
Equilibrium redox potentials for the formation of Cu 2 O from copper and CuO, Cu(OH) 2 from Cu 2 O
(Eqs. 1, 2, and 3) showed that in the first potential region the formation and growth of the inner layer of Cu 2 O
and the outer layer of CuO/Cu(OH) 2 overlaying the Cu 2 O barrier layer occurred as in Figure 2. Capacitive
circuit (inset Figure 3), which characterized a diffusion process represented the behavior of the modified film. In
the third potential region diffusion controlled kinetic phenomenon changed to a charge transfer process especially
at 0.4 and 0.7 V. In this region CuO oxide layer grew using Cu 1+ ions from the inner Cu 2 O layer and then
Cu(OH) 2 from CuO at higher potential.
120 -0.1 V
100 -0.2 V
80
0.0 V
2
Zʹʹ kOhm.cm
60 -0.3 V
40 Q1
Q2
20 0.1 V Rpor
Rsol Rcf
0
Z ʹ kOhm.cm
2
Figure 2. Nyquist plots for the initial stage of passivation of copper surface.
It is to be pointed out that EIS is a potentiostatic technique and prior to EIS measurements at each
potential step, the electrode was given sufficient time for stabilization.
Complex plane plots, Z vs. Z , are shown in Figure 2 for potentials of the initial stage. A complete semicircle
was not observed; instead a highly reactive arc was obtained, which rose steeply from the higher frequency
end. At 0.0 V, structural transformations began in the film and were enhanced at 0.1 V. The Nyquist plot
of 0.1 V potential became almost parallel to the Z axis at low frequencies. The dissolution also affected the
homogeneity of the surface leading to incoherent response of different species present to the applied signal.
The behavior of the film in relation to its continuously varying structure on application of potential is
best represented by fitting the impedance data (Table) to the circuit in the inset of Figure 2. Better agreement
was achieved between theoretical and experimental results by including frequency dependent constant phase
elements (CPEs), Q 1 and Q 2 in the circuit. A CPE was introduced instead of pure capacitance and Warburg
impedance. In the proposed circuit CPE represented interfacial capacitance Q 1 in parallel with a sub-circuit
involving charge transfer resistance (R ct ). The circuit represented the characteristics of the underlying Cu 2 O
layer and the duplex layer.
742
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
In general CPE appeared due to a distribution of relaxation times arising from inhomogeneities present
at the microscopic level in the oxide phase and at the oxide/electrolyte interface. The static disorder such as
porosity, 23 a random mixture of insulator and conductor described by the effective medium approximations
as percolation, 24 an interface that can be described by fractal geometry, 25 an RC transmission line, 26,27 or
dynamic disorder, such as diffusion, are contributors to CPE. The impedance of a CPE is defined by Eq. (4).
When n = 1, Q becomes equivalent to a true capacitance (ideal case of no dispersion); for n = 0, Q becomes
equivalent to a resistance. The value of n = 0.5 corresponds to diffusion of a reactant in a Faradaic process and
Q becomes equivalent to Warburg impedance. At intermediate values 1 > n > 0.5, Eq. (4) describes frequency
behavior of a constant phase element, which contrary to Warburg impedance is not dependent on the presence
or absence of redox reactions. Magnitude of the exponent n indicates the extent of surface homogeneity. Poor
surface homogeneity is related to some type of pore structure permeating solvent through it.
The electrolyte/film interface was represented by Q 1 and film/metal interface by Q 2 . R por and R ct
represent film resistance and charge transfer resistance, respectively. Similarly n 1 represents inhomogeneity of
the electrolyte/oxide interface and n 2 represents homogeneity of the film/metal interface. The rate of corrosion
was controlled by the charge transfer resistance R ct and the undergoing structural changes of the passive layer
were indicated by Q 1 and Q 2 . The Q 2 and R ct combination characterized a higher time constant and therefore
it was connected to the processes occurring at lower frequencies.
The Q 1 and R por parallel combination characterized a lower time constant and it was connected to the
processes occurring at higher frequencies. It reflects the properties of the oxide/electrolyte interface. It refers
to the double layer capacitance and resistance of electrolyte in the pores of the oxide film. The contribution to
743
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
the overall impedance from the porous outer layer with pores filled with electrolyte would be small when the
outer layer is thin. In such a case the charge transfer resistance dominated the impedance.
Theoretically, Eqs. (5) and (6) provided the criteria that should be met for the experimental data to
be resolved into 2 parts. Each time constant gave rise to a semicircle when the 2 time constants were quite
far apart from each other so that the phenomena represented by them were not interacting and these can be
studied independently. 28
and
Here τt and τp stand for time constants at low frequency and high frequency, respectively. Appearance of one
semicircle (incomplete) in Nyquist plots of Figure 2 indicated that 2 time constants did not differ from each
other widely as required. The Nyquist plots contained information about the solution/film and the film/metal
interfaces and apparently it was unlikely that a second semicircle would be observed due to film resistance
and capacitance at frequencies greater than 100 kHz. The observed semicircle therefore represented a charge
transfer semicircle, although incomplete, which gave information about the rate of corrosion occurring through
the passive film. The diameter and the maximum reactance (Z ) at low frequencies gave R ct and Q 2 strictly in
accordance with the equivalent circuit model used. The R por and Q 1 strongly influenced the oxide/electrolyte
interface.
The shape of a wide open capacitive arc was characteristic of a capacitor charged via CPE or Warburg
impedance due to semi infinite diffusion of the charging species.
R ct values were higher than R por and inversely related to exchange current thus it measured the rate of
electron transfer reaction. Higher values of R ct in the -0.3 V to 0.1 V potential range indicated the presence of
Cu 2 O barrier layer with some CuO/Cu(OH) 2 formed 8 at later potentials (0.0, 0.1 V) from Cu 2 O. Apparently,
the film was compact and protective. The capacitance values in this initial potential stage may be rationalized
in terms of the charge transfer processes leading to conversion of Cu to Cu 1+ and Cu 2+ , and Cu 1+ to Cu
2+
. Low capacitance value (Q 2 ) in this potential stage suggested near perfect oxide covering of metal 29 with
cuprous layer. At -0.3 V, the n 2 value corresponds to that of an ideal capacitor indicating coverage of the
electrode surface by well organized Cu 2 O component of the bulk oxide film. The decrease in n 2 values with
increasing applied potential suggested that the film is losing its initial homogeneity due to reactions of Eqs. (2)
and (3).
Q 1 , n 1 , R por parameters characterized the oxide film/electrolyte interface. The Q 1 value first decreased
followed by a slight increase. At -0.3 V, Cu 2 O is formed and this aspect is reflected in the Q 1 , n 1 values.
With further development of Cu 2 O layer at -0.2 and -0.1 V potentials, the interface became more organized
and homogenous with the removal of dislocations, edges present. Changes in the interface structure start taking
place at 0.0 V and continues at +0.1 V as reflected by corresponding Q 1 , n 1 . At this potential, copper also
dissolved in the form of HCuO −
2 species in buffer of pH 9.2.
10
Such low values of Q 1 suggested negligible
contribution of charging current as also noticed in the case of chromium. 30 This trend together with low values
of R por suggested that the outer layer is thin with pores filled with little electrolyte.
744
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
Middle stage
Best fit Nyquist plots are shown in Figure 3 for 0.2 and 0.3 V. An incomplete circle at high frequencies interacted
with a “reactive tail” at low frequencies at an angle of 45 ◦ with the Z axis. The shape was characteristic of a
diffusion controlled process, which in this case was responsible for the movement of ions and accumulation of
charge in the bulk film. At 0.3 V, the Nyquist plot is a circular arc at the high frequency side which changed
to an arched line at low frequencies indicating the presence of a CPE. Both impedance (Z , Z ) values have
increased as compared to that at 0.2 V potential indicating more compactness and rigidity of the film by diffusion
of ionic species into the duplex layer.
30 Exp 0.3V
Fit
25
Exp
20
Fit
2
0.2V
Z ʹʹ kOhm. cm
15
10
Q1
5
Rpor Q2
Rsol
0
0 10 20 30 40 50 60 70 80 90
Zʹ kOhm. cm
2
Figure 3. Nyquist plots for the second stage of passivation of copper surface.
The proposed circuit (inset of Figure 3) represented diffusive character of the film. It consisted of an
electrolyte/film interface denoted by Q 1 in parallel with R por (film resistance) and another for film/metal
interface Q 2 . Kinetic parameter, diffusion coefficient was calculated using the following equation: 31
√ √
S = RT /n2 F 2 c 2D; S = 1/(Q2 2) (7)
The diffusion coefficient value is of the order of 10 −9 cm 2 s −1 and suggested a protective film where ion
movement was low and potential dependent. According to point defect model the oxidation of metal atoms to
form ions at the metal/oxide (inner) interface generated a driving force to form anion vacancies. 32,33
where Cu Cu (m), a copper atom in a regular metal site, Cu Cu (ox), a copper cation in a regular site of the oxide
film, V ..o , a positively charged oxygen vacancy, and e represented the electron. At the oxide/electrolyte (outer)
interface anion vacancies became occupied by anions:
745
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
where O o (ox), an oxygen anion in a regular site of the oxide film and H + (aq), the hydrogen ion in the aqueous
electrolyte. Cation vacancies, which were formed by dissolution of cations into the electrolyte diffused towards
the inner interface:
CuCu (ox) = Cu+ (aq) + VCu (ox) (10)
Cu + (aq), a positively charged copper cation in the aqueous electrolyte, V Cu (ox), a negatively charged cation
vacancy in the oxide, and V Cu (m), a neutral vacancy in a regular metal site. 34
CuCu (ox) = Cu+
duplex + VCu(ox) (12)
where Cu +
duplex showed the ion that moved into the duplex film and V Cu(ox) , vacancy created in the Cu 2 O
layer. 35
The interfacial capacitance (Q 1 ) of the oxide/electrolyte interface slightly decreased on increase in
potential. However, constant n 1 values showed that the interfacial homogeneity is retained. The Q 1 values are
typical capacitance values observed for an oxide covered surface of metal. 29 An increase in R por at 0.2 V and
then further to a maximum value at 0.3 V suggested that the Cu 1+ ions from the inner Cu 2 O layer diffuse and
accumulated in the duplex layer. In this buffer solution, 0.2 V and 0.3 V are those applied potentials, where
compressive stress developed in the crystals causing ejection of Cu 1+ ions into the duplex film. 36 For the same
reason Q 2 increased but n 2 decreased. The small n 2 values were characteristic of the solid electrolyte 37 and
electrochemical systems. 34,38 The increased resistance caused blocking of the percolation of the molecules into
the passive film although the film appeared to be amorphous in nature (low n 2 value) and behaves like solid
electrolyte 37 .
Final stage
Best fit Nyquist plots in Figure 4 for the 0.4 V – 0.9 V potentials range showed a skewed semicircle at 0.4 V.
At this potential a charge transfer process owing to the presence of Cu 1+ occurred that resulted in the CuO
growth.
At 0.5 and 0.6 V the skewed semicircle transformed into a rising curve, almost linear at the low frequency
end suggesting diffusion of OH − opposite to Cu 1+ due to increased porosity.
The change of Nyquist plots along with Z and Z suggested that the charge transfer process appeared to
have slowed down considerably and the diffusion process took over. At 0.7 V again a skewed semicircle appeared
which gradually opened up as the potential increased to 0.9 V. Corresponding Z and Z values for the peaks
were respectively 4.2 and 2.7 kΩ cm 2 for 0.4 V and 7.6 and 4.8 kΩ cm 2 for 0.7 V. The maxima in the skewed
semicircles in the Nyquist plots for 0.4 V and 0.7 V were observed at 1.08 Hz and 294 mHz, respectively. The
increase in the Z values at these potentials may be correlated with the molar volume increase on transformation
of underlying Cu 2 O to CuO and Cu(OH) 2 at oxide/metal interface due to diffusion. Formation of Cu(OH) 2
is expected at 0.7 V since thermodynamically this compound is formed at a higher potential. 1,8 w
746
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
0.6 V
15
0.5 V
2
10
Z ʹʹ kOhm. cm
0.9 V
5 0.8 V
0.7 V
0.4 V
0
0 5 10 15 20 25
Z ʹ kOhm. cm
2
Figure 4. Nyquist plots for the final stage of passive copper surface.
At 0.8 V, the semicircle structure of the plot is still retained, but the maximum now shifted further to
a lower frequency (212 Hz). At this potential it is assumed that formation of hydroxide still continued leading
to change in the structure of the oxide/electrolyte interface. At 0.9 V, the shape of the Nyquist plot changed
toward the low frequency end. The maximum may be located somewhere near 30 mHz. Apparently, potential
increase modified the amorphous structure of the film. First, film thickness increased with the formation of
CuO from Cu 2 O, and later conversion to Cu(OH) 2 produced wide pores in the film, eventually, facilitated the
movement of ions from solution to metal surface. Passivity of the film was thus broken down, i.e. trans-passive
region was reached with filling of pores with solution. Aluminum oxide films were also porous when formed
but they possessed a barrier type layer underneath. 19,20 In case of copper the increased potential destroyed the
passive film.
The influence of potential change on the metal/oxide and oxide/electrolyte interfaces was reflected in the
values of the best fit parameters also. The growth of the duplex film was accompanied by the formation of
CuO, which restarted at 0.4 V after diffusion of Cu 1+ stopped into the duplex film. The formation of CuO
islands from the oxidation of Cu 2 O increased the roughness of the surface 39 as also shown in the scanning
tunneling microscopy (STM). 40 The resistance R por increased and attained a maximum value of 10.4 kΩ cm 2
at 0.5 V due to CuO growth on the electrode surface which had a smaller molar volume (12.4 cm 3 mol −1 ) than
Cu 2 O (23.9 cm 3 mol −1 ), thus made the film compact. Simultaneously, Q 1 and n 1 values increased. With
further increase in potential, both the R por and Q 1 decreased, which may be due to the variation in the film
texture. At 0.7 V conversion of CuO to Cu(OH) 2 through a charge transfer process slightly decreased R por
and increased Q 1 and n 1 , indicating trapping of charged species, due to the dissolution of Cu(OH) 2 having
molar volume of 29.0 cm 3 mol −1 .
The charge transfer process continued at 0.8 V and 0.9 V also causing further decrease in R por and
increase in Q 1 . Dissolution of Cu(OH) 2 may also be a factor at these potentials that affected the value of n 1 .
747
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
According to the band structure model of semiconductors, crossing of valence band and fermi level above
0.308 V for Cu 2 O inject an acceptor level, i.e. Cu 2+ states, which finally leads to the formation of CuO and
Cu(OH) 2 as the potential increased to 0.708 V. According to Strehblow, any further potential increase would
be located in the Cu(II) oxide/electrolyte interface. 9 Therefore, it is expected that potential increase to 0.7 V
would give hydroxide of copper according to the reaction: 41
CuO + 2OH − →
← Cu(OH)2 + O
2−
(13)
The Q 2 values together with n 2 values reflected the potential dependence of the film structure.
140 120
120 100
100
80
2
2
80
Zʹ kOhm. cm
C^10 F. cm
60
60 -5
40
40
20
20
0 0
Potential / V vs SCE
Figure 5. Variation of Z and capacitance (C) with potential at 0.03 Hz.
Conclusion
The corrosion process strongly depended upon the applied potential. In the first potential stage, the film was the
most stable and its formation and growth over the polycrystalline copper surface followed a consistent pattern
in the negative potential regime.
748
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
The second stage acted like the demarcation line between passivation and trans-passivation stages.
Diffusion coefficient calculated in conjunction with impedance spectra showed that film was protective with
difficult ion movement.
In the final potential stage the film structure was severely affected by increasing potential. So electrolyte
supersaturated with Cu 2+ ions adjacent to the metal surface did not cause precipitation of oxides rather direct
oxide growth seems very likely.
References
1. Pourbaix, M. In Atlas of the Electrochemical Equilibria in Aqueous Solution; Chap. 4, pp. 384-392, Pregmon Press,
New York, 1966.
4. Kautek, W.; and Gordon II, J. G. J. Electochem. Soc., 1990, 137 , 2672-2677.
5. Halliday. J. S. Trans. Faraday Soc. 1954, 50 , 171-178.
8. Shoesmith, D. W.; Rummery, T. E.; Owen, D.; Lee, W. J. Electrochem. Soc. 1976, 123 , 790-799.
9. Strehblow,H.-H.; Collisi. U.; Druska, P. International Symposium on Control of Copper Alloys Oxidation, 1992,
Rouen, Tranrvied, 6.-8. 7.
11. Strehblow, H.-H.; Maurice, V.; Marcus. P. Electrochim. Acta. 2001, 46 , 3755-3766.
12. Ambrose, J.; Barradas, R. G.; Shoesmith, D. W. J. Electroanal. Chem. 1973, 47 , 47-64.
13. Droog, J. M. M.; Alderliesten, C. A.; Anderliesten, P. T.; Bootsma G. A. Electroanal. Chem. . 1980, 111 , 61-70.
14. Diard, J.-P.; Le Canut, J.-M.; Le Gorrec, B.; Montell. C. Electrochim. Acta. 1998, 43 , 2469-2483.
15. Diard, J.-P.; Le Canut, J.-M.; Le Gorrec, B.; Montell.C. Electrochim Acta. 1998, 43 , 2485-2501.
16. Babic, R.; Metikos-Hukovic, M.; Jukie, A. J. Electrochem. Soc. 2001, 148 , B146-B151.
19. Hitzig, J.; Juttner, K.; Lorenz, W. J.; Paatsch, W. J. Electrochem. Soc., 1986, 133 (5), 887-892.
20. Milosev, I.; Metikos- Hukovic, M. Electrochim Acta, 1997, 42 , 1537-1548.
21. Boukamp, B. A. EQUIVCRT software manual, University of Twente Enschede, The Netherlands, 1993.
22. Gabrielli, C. Identification of Electrochemical Processes by Frequency Response Analysis, Technical report number
004/83, Solartron Instruments, England.
749
A study of growth and breakdown of..., A. NASEER, A. Y. KHAN
750