You are on page 1of 11

Electrochimica Acta 52 (2007) 4363–4373

Electrochemical noise analysis of corrosion of pure aluminium in alkaline


solution in the presence of SO42− ion, NO3− ion and Na2S additives
Kyung-Hwan Na, Su-Il Pyun ∗
Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology,
373-1, Guseong-dong, Yuseong-gu, Daejeon 305-701, Republic of Korea
Received 1 July 2006; received in revised form 3 October 2006; accepted 15 December 2006
Available online 8 January 2007

Abstract
The electrochemical noises obtained from pure aluminium in an alkaline solution have been analysed as a function of sulphate, nitrate or
sodium sulphide concentration. A stochastic theory has been applied to the experimentally obtained electrochemical noises based upon the
Weibull distribution function. From the occurrence of two linear regions on one Weibull probability plot, the present analysis method permitted
us to distinguish between the dissolution of Al and the stochastic process concerning the series of nucleation, growth and detachment of hydrogen
bubbles in an engineering sense. Moreover, the conditional event generation rate and the average charge in each event were quantitatively determined
for the stochastic process of hydrogen bubbles as well as the dissolution of Al. Effects of sulphate, nitrate and sodium sulphide additives on the
corrosion of pure aluminium in an alkaline solution have been discussed in terms of the conditional event generation rate and the average charge
in each event.
© 2007 Elsevier Ltd. All rights reserved.

Keywords: Aluminium; Sulphate; Nitrate; Sodium sulphide; Electrochemical noise

1. Introduction In the microscopic point of view, the dissolution of Al can be


also regarded as one of the stochastic phenomena, being com-
The mechanism of corrosion of pure aluminium (Al) and Al- posed of a large number of independent events. In this respect,
based alloys in an aqueous alkaline solution has been studied the dissolution of Al (uniform corrosion) has been successfully
for the development of an anode in Al/air batteries [1,2]. In the distinguished from pit initiation in a neutral chloride solution in
corrosion of Al in an alkaline solution, it is well known that the our previous works [6,7].
dissolution of Al and hydrogen evolution are coupled through Since the fluctuations in potential or current associated with
their competition for surface sites, leading to the formation of a such stochastic phenomena as the dissolution of Al and the series
short-circuited corrosion cell at open-circuit potential [2,3]. of nucleation, growth and detachment of hydrogen bubbles can
Evolved hydrogen bubbles enhance mass transfer by locally be easily observed in the experimental data, electrochemical
stirring the electrolyte and increase the electrolyte resistance noise measurement has become a useful technique for moni-
during the electrochemical process [4]. Moreover, they diminish toring these stochastic phenomena. However, the conventional
the active area of the electrode by screening a part of the elec- analysis methods such as statistical analysis method [8,9] and
trode. Thus, the series of formation of hydrogen bubbles, their spectral analysis method [10–12] did not allow us to distinguish
growth and detachment may affect the electrochemical noise between these stochastic phenomena.
during the corrosion of Al in an alkaline solution. In addition, it In the present work, the electrochemical noises obtained
has been reported that this series of phenomena has a stochastic from pure Al in an alkaline solution were analysed as a function
nature [5]. of sulphate (SO4 2− ), nitrate (NO3 − ) or sodium sulphide (Na2 S)
concentration. For this purpose, the Weibull probability plot
was constructed first and then the conditional event generation
∗ Corresponding author. Tel.: +82 42 869 3319; fax: +82 42 869 3310. rate and the average charge in each event were quantitatively
E-mail address: sipyun@webmail.kaist.ac.kr (S.-I. Pyun). determined, based upon the stochastic theory combined with

0013-4686/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2006.12.029
4364 K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373

the shot noise theory. In addition, the effects of anion and Na2 S Table 1
additives on the corrosion of Al in an alkaline solution have The values of the Stern–Geary coefficient B determined from the potentiody-
namic polarisation curves
been discussed.
Solution Stern–Geary coefficient B [V]
2. Experimental 0.01 M NaOH 4.65 × 10−3
0.01 M NaOH + 0.01 M Na2 SO4 9.78 × 10−3
In this work, the experiments were done with a high-purity 0.01 M NaOH + 0.1 M Na2 SO4 1.28 × 10−2
Al rod of 6.35 mm diameter (99.999%, Aldrich Chemical Co., 0.01 M NaOH + 0.01 M NaNO3 1.24 × 10−2
0.01 M NaOH + 0.1 M NaNO3 1.40 × 10−2
Inc.). The specimen was set in a block of polyethylene and 0.01 M NaOH + 0.01 M Na2 S 3.27 × 10−3
the upper surface of the block was ground with silicon carbide 0.01 M NaOH + 0.1 M Na2 S 2.91 × 10−3
papers to 1500 grit, successively. In order to ensure a clean Al
surface before any exposure to the corroding electrolytic solu-
tion, the specimen was cleaned with ethanol after this polishing after 3 h of immersion. It can be seen in Fig. 2 that potential
procedure. The electrolytes used were 0.01 M NaOH solutions records are composed of discernable transients and they change
containing various Na2 SO4 , NaNO3 or Na2 S concentrations of with solution.
0, 0.01 and 0.1 M at room temperature. Shot noise is associated with current flow across a potential
The potentiodynamic polarisation experiment was performed barrier. It is due to the fluctuation of current around an average
in the applied potential range from −2.0 to 0 VSCE with a scan value resulting from the random emission of electrons. In other
rate of 0.5 mV s−1 by using an EG&G Model 263A Galvanos- words, each electron carries a discrete amount of charge inde-
tat/Potentiostat interfaced with a IBM-compatible computer. A pendently and produces a small current pulse that corresponds to
platinum gauze and a saturated calomel electrode (SCE) were an event [13,14]. According to this theory, the summation of all
used as the counter and reference electrodes, respectively. the current pulses produces the average corrosion current Icorr .
Electrochemical noise measurement was carried out within a Therefore, the average corrosion current Icorr is given by
Faraday cage with a Zahner IM6e impedance measurement unit
equipped with a Zahner NProbe: two identical Al specimens Icorr = qfn (1)
were galvanically coupled. After that, both the current between
where q is the average charge in each event and fn is the frequency
the two Al specimens and the potential of the Al specimens
of events.
against an SCE were simultaneously recorded with time for 5 h.
If we assume that shot noise is produced in the present system
The sampling interval used in this study was 0.05 s and each
[15–18], the frequency of events fn can be estimated from the
time record consisted of 2048 datapoints. 180 time records were
following equation:
analysed for each solution. The cumulative probability of the
frequency of events fn was calculated from the electrochemical Icorr B2
noises experimentally obtained based upon the shot noise theory. fn = = (2)
q ψE
It was then rearranged in the time domain to analyse the noises
based upon the stochastic theory. where B is the Stern–Geary coefficient and ψE is the low-
In order to compare the corrosion rate of Al in an alkaline frequency limit of power spectral density of potential.
solution in the presence of additives with that corrosion rate in Though the electrochemical process is not always a pure shot
the absence, weight loss was measured as a function of additive noise process, it has been revealed that it is also possible to
concentration. estimate fn by using Eq. (2) [15,17,18]. The cumulative prob-
ability of fn , F(fn ) was calculated by the following sequence
3. Results and discussion [6,7,17,18]:

3.1. Construction of Weibull probability plots (1) Arrange all calculated fn data in order from the smallest to
the largest.
Fig. 1 shows potentiodynamic polarisation curves on a (2) The cumulative probability can then be measured as
semilogarithmic scale for pure Al at a scan rate of 0.5 mV s−1 M/(N + 1), where M is the rank in the ordered fn data and N
in aqueous 0.01 M NaOH solution containing various SO4 2− , the total number of data.
NO3 − or Na2 S additive concentrations of 0, 0.01 and 0.1 M.
While the addition of SO4 2− or NO3 − did not exert any notice- The above sequence is based upon the mean rank approxima-
able effect on current density, the addition of Na2 S remarkably tion, which is the most widely used for calculating cumulative
increased both anodic and cathodic current densities. The value probability F(fn ). But the median rank (M − 0.3)/(N + 0.4) [19]
of the Stern–Geary coefficient B, which is necessary for the is sometimes also used when N is smaller than 10.
shot noise analysis, was obtained from polarisation curves. The The cumulative probability F(fn ) was experimentally deter-
resulting value of B is listed for each solution in Table 1. mined from the electrochemical noises based upon the shot noise
Fig. 2 shows examples of potential records in the time theory in aqueous 0.01 M NaOH solution containing various
domain, obtained from pure Al in aqueous 0.01 M NaOH solu- additive concentrations. The resulting values of F(fn ) are plot-
tion containing various SO4 2− , NO3 − and Na2 S concentrations ted against fn in Fig. 3. It was found that the distribution of
K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373 4365

Fig. 1. Potentiodynamic polarisation curves of pure Al with a scan rate of 0.5 mV s−1 in aqueous 0.01 M NaOH solution containing various (a) SO4 2− , (b) NO3 −
and (c) Na2 S concentrations.

fn shifted to a more high-frequency region with increasing analysed using stochastic approach based upon its stochastic
SO4 2− , NO3 − or Na2 S concentration. The positive shift of fn nature [20–24]. Accordingly, in order to quantitatively analyse
values means that the corrosion of Al occurs more rigorously the electrochemical noises and to clearly distinguish between
in an alkaline solution in the presence of additives. Moreover, it the dissolution of Al and the stochastic process concerning
is worthwhile to note that the values of fn ranged in 5 decades. the series of nucleation, growth and detachment of hydrogen
This range was in good agreement with the reported range of bubbles, a stochastic theory has been applied to the cumulative
2–5 decades in Cottis’ work [17,18]. probability F(fn ).
It looks as if F(fn ), calculated based upon the shot noise According to the stochastic theory, the cumulative probability
theory, allows a useful and reasonably intuitive categorization F(t) of a failure system can be introduced as a Weibull distribu-
of the type of corrosion [17,18]; however, it is not sufficient with tion function based upon a “weakest-link” model [25–28], which
F(fn ) obtained from the shot noise analysis to clearly distinguish is expressed as
between various components of stochastic phenomena, such as  m
−t
the dissolution of Al and the series of nucleation, growth and F (t) = 1 − exp (3)
detachment of hydrogen bubbles during the alkaline corrosion n
of Al in a certain frequency range. In this respect, it is worth- where m and n are the shape and scale parameters, respectively.
while to note that the shot noise process was mathematically m is a dimensionless parameter and n is expressed in sm .
4366 K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373

Fig. 2. Examples of potential records obtained from pure Al in aqueous 0.01 M NaOH solution containing various (a) SO4 2− , (b) NO3 − and (c) Na2 S concentrations
after 3 h of immersion.

The Weibull distribution function is one of the widely used By fitting Eq. (4) to the cumulative probability numerically
cumulative probability functions for predicting lifetime in relia- calculated, two parameters m and n can be determined from
bility tests. This is because it can easily approximate the normal the slope of the linear ln{ln[1/(1 − F(t))]} versus lnt plots
distribution, logarithmic normal distribution and exponential (Weibull probability plots) and from the intercept on the
distribution functions. Secondly, it is also possible to analyse ln{ln[1/(1 − F(t))]} axis, respectively.
data even when two or more failure modes are present at the Since Weibull distribution function is defined appropriately in
same time [25,26]. Finally, it is worthwhile to note that there are the time domain, the plot of the calculated F(fn ) against fn was
some research works [29,30] in which Weibull distribution was first transformed from the fn domain to the 1/fn time domain
successfully applied to the analysis of such shot noise process for the stochastic analysis. The Weibull probability plot was
as the interarrival distribution of data in the computer network- then constructed by fitting the calculated cumulative probabil-
ing. For these reasons, the Weibull distribution function has been ity, F(1/fn ), to the Weibull distribution function. The resulting
adopted for the analysis of F(fn ) in order to distinguish stochastic Weibull probability plots are given in Fig. 4.
phenomena. The plots showed satisfactorily good two linear regions in an
From rearrangement of Eq. (3) alkaline solution, irrespective of additive concentration. Accord-
   ing to the stochastic theory [26,27], the two linear regions on one
1 plot indicate that two mechanisms exist, depending upon fn . It
ln ln = m ln t − ln n (4)
(1 − F (t)) was reported that the alkaline corrosion of Al can be considered
K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373 4367

Fig. 3. Cumulative probability plots for the frequency of events, fn in aqueous 0.01 M NaOH solution containing various (a) SO4 2− , (b) NO3 − and (c) Na2 S
concentrations.

as the sum of the anodic partial reaction and the cathodic partial Al in an alkaline solution at open-circuit potential: one is the dis-
reaction given below [31,32]: solution of Al and the other is the stochastic process of hydrogen
bubbles. Accordingly, two linear regions in one Weibull proba-
Partial anodic reaction : Al + 4OH− = Al(OH)4 − + 3e− bility plot correspond to these stochastic processes, respectively.
E◦ = −2.59 VSCE (5) The linear region in the lower frequency range is owing to the
dominant stochastic process of hydrogen bubbles. On the other
Partial cathodic reaction : 3H2 O + 3e− = 3/2H2 + 3OH− hand, the linear region in the relatively higher frequency range
E◦ = −1.07 VSCE (6) is due mainly to the dominant dissolution of Al. It should be
emphasised that the fn s in the linear region of the lower frequency
From practical point of view, a couple of reactions (5) and range does not rigorously represent “pure” stochastic process of
(6) constitute a self-discharging corrosion process which causes hydrogen bubbles. They also include the dominant dissolution
the loss of Al fuel in Al-air battery in standby mode (namely at of Al in some degree.
open-circuit potential). In this case, it is reasonable to think that Recently, it was revealed [6,7] that this method produced use-
two stochastic processes readily occur during the corrosion of ful results from practical point of view for investigating pitting
4368 K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373

Fig. 4. Weibull probability plots (plots of ln{ln[1/(1 − F(1/fn ))]} vs. ln(1/fn )) in aqueous 0.01 M NaOH solution containing various (a) SO4 2− , (b) NO3 − and (c)
Na2 S concentrations.

corrosion of Al in a chloride solution in the presence of anions, cess of hydrogen bubbles, three linear regions in a hydrochloric
which were in good agreement with previous results reported by acid solution correspond to dominant uniform corrosion,
other researchers [33–36]. Accordingly, it is plausible to think dominant pitting corrosion, and the dominant stochastic process
that the fn s in the low frequency range may provide useful prac- of hydrogen bubbles. This implies that one linear region is
tical information on the stochastic process of hydrogen bubbles mainly related to one dominant stochastic phenomenon in an
in an engineering sense, when the contribution of uniform cor- engineering sense.
rosion to noise signal is overwhelmed by that contribution of the
stochastic process of hydrogen bubbles. 3.2. Stochastic analysis of electrochemical noise
In this regard, it is worth noting that two and three linear
regions were observed on one Weibull probability plot in a 3.2.1. Stochastic process concerning the series of
concentrated sulphuric acid solution and in a concentrated nucleation, growth and detachment of hydrogen bubbles
hydrochloric acid solution, respectively, as shown in Fig. 5. The values of shape parameter m and scale parameter n in
While two linear regions in a sulphuric acid solution correspond the Weibull distribution function were determined from Fig. 4
to dominant uniform corrosion and the dominant stochastic pro- for the stochastic process of hydrogen bubbles as a function
K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373 4369

Table 3
The values of the average charge in each event of the stochastic process con-
cerning the series of nucleation, growth and detachment of hydrogen bubbles,
and the dissolution of Al, calculated from electrochemical noises
Solution Hydrogen evolution Dissolution of
on Al [C] Al [C]

0.01 M NaOH 2.00 × 10−9 2.42 × 10−10


0.01 M NaOH + 0.01 M Na2 SO4 5.90 × 10−10 1.22 × 10−10
0.01 M NaOH + 0.1 M Na2 SO4 5.89 × 10−10 2.46 × 10−11
0.01 M NaOH + 0.01 M NaNO3 3.97 × 10−10 5.55 × 10−11
0.01 M NaOH + 0.1 M NaNO3 1.26 × 10−10 2.17 × 10−11
0.01 M NaOH + 0.01 M Na2 S 2.00 × 10−9 2.52 × 10−10
0.01 M NaOH + 0.1 M Na2 S 1.90 × 10−8 2.00 × 10−9

the generation probability of events in the next unit time dt for


the specimens in which events have not been yet generated when
t has elapsed.
The rate r(t) was quantitatively determined as a function of
additive concentration by inserting the values of the shape and
scale parameter m and n given in Table 2, into Eq. (7). The
resulting rate, r(t), is illustrated with time t for the stochastic
Fig. 5. Weibull probability plots (plot of ln{ln[1/(1 − F(1/fn ))]} vs. ln(1/fn )) in
aqueous 1 M H2 SO4 and 1 M HCl solutions. process of hydrogen bubbles in Fig. 6. It was observed that the
r(t) in the presence of SO4 2− , NO3 − or Na2 S was remarkably
of additive concentration. The resulting values of m and n are increased in value compared to that in the absence for a given
listed in Table 2. In reliability engineering [25], the failure rate time.
is of importance in predicting life cycle of a system. In our On the assumption that the amount of charge consumed
previous works [26–28], pit embryo formation rate was success- for hydrogen evolution should be equivalent to the amount of
fully determined in a neutral chloride solution by employing this charge consumed for the dissolution of Al in the alkaline cor-
failure rate concept. rosion of Al, the value of the average charge in each event
In a similar way, we introduced in the present work the was quantitatively determined for hydrogen evolution on Al
conditional event generation rate for the stochastic process of from the electrochemical noises. The resulting value is listed
hydrogen bubbles, r(t), defined in previous works [6,7], which in Table 3. The average charge in each event was decreased
is given by for the stochastic process of hydrogen bubbles by the addition
of SO4 2− or NO3 − . On the other hand, the average charge in
F  (t) m each event was increased for the stochastic process of hydrogen
r(t) = = t m−1 (7)
1 − F (t) n bubbles by the addition of Na2 S. This indicates that the aver-
age size of events becomes smaller for the stochastic process of
for the Weibull distribution function in an attempt to quan- hydrogen bubbles by the addition of SO4 2− or NO3 − whereas
titatively investigate the effect of additives on this stochastic the average size of events becomes larger by the addition of
process. In Eq. (7), F (t) is the derivative of F(t), indicating the Na2 S.
probability density function for the Weibull distribution. Fur-
thermore, 1 − F(t) in the denominator implies that r(t) is defined
based upon conditional probability. Physically, r(t)dt represents 3.2.2. Dissolution of Al
In a similar way, the values of shape parameter m and scale
parameter n in the Weibull distribution function were determined
Table 2
The values of the shape parameter m and the scale parameter n for the stochastic
for the dissolution of Al as a function of additive concentration,
process concerning the series of nucleation, growth and detachment of hydro- which are listed in Table 4. Fig. 7 demonstrates the resulting r(t)
gen bubbles determined from the linear slope in the low frequency region of with time t for the dissolution of Al. The r(t) in the presence of
ln{ln[1/(1 − F(1/fn ))]} vs. ln 1/fn plot SO4 2− , NO3 − or Na2 S was also remarkably increased in value
Solution Shape parameter Scale parameter compared to that in the absence for a given time.
m [−] n [sm ] Table 3 gives the value of the average charge in each event
0.01 M NaOH 0.20 8.21 × 10−2
for the dissolution of Al determined from the electrochemical
0.01 M NaOH + 0.01 M Na2 SO4 0.45 3.67 × 10−3 noises. The average charge in each event was decreased by the
0.01 M NaOH + 0.1 M Na2 SO4 0.46 3.57 × 10−3 addition of SO4 2− or NO3 − but it was increased by the addition
0.01 M NaOH + 0.01 M NaNO3 0.86 1.27 × 10−4 of Na2 S for the dissolution of Al. This implies that the average
0.01 M NaOH + 0.1 M NaNO3 1.04 2.10 × 10−6 size of events for the dissolution of Al becomes smaller by the
5.46 × 10−2
addition of SO4 2− or NO3 − whereas the average size of events
0.01 M NaOH + 0.01 M Na2 S 0.23
0.01 M NaOH + 0.1 M Na2 S 0.24 5.69 × 10−2
becomes larger by the addition of Na2 S.
4370 K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373

Fig. 6. Plots of the conditional event generation rate for the stochastic process concerning the series of nucleation, growth and detachment of hydrogen bubbles, r(t)
vs. t for Al in aqueous 0.01 M NaOH solution containing various (a) SO4 2− , (b) NO3 − and (c) Na2 S concentrations.

3.3. Corrosion rate

The corrosion rate for Al in an alkaline solution was estimated


Table 4
The values of the shape parameter m and the scale parameter n for the dissolution from the experimentally obtained electrochemical noises, poten-
of Al from the linear slope in the high frequency region of ln{ln[1/(1 − F(1/fn ))]} tiodynamic polarisation curves and weight losses as a function
vs. ln 1/fn plot of SO4 2− , NO3 − or Na2 S concentration. The resulting corrosion
Solution Shape parameter Scale parameter n rate is demonstrated in Fig. 8. The error bar on the plot desig-
m [−] [sm ] nates the 95% confidence interval. In this figure, the corrosion
rate calculated from the electrochemical noises was at most one
0.01 M NaOH 2.17 3.41 × 10−10
0.01 M NaOH + 0.01 M Na2 SO4 2.91 1.28 × 10−15 order of magnitude smaller in value than those corrosion rates
0.01 M NaOH + 0.1 M Na2 SO4 2.34 6.47 × 10−15 calculated from polarisation curves and weight losses.
0.01 M NaOH + 0.01 M NaNO3 2.43 3.14 × 10−13 Corrosion can occur in two different forms of microcells and
0.01 M NaOH + 0.1 M NaNO3 3.78 7.33 × 10−24 macrocells depending upon the different spatial location of local
0.01 M NaOH + 0.01 M Na2 S 2.40 6.71 × 10−12
anode and cathode [37–39]. Anodic and cathodic reactions are
0.01 M NaOH + 0.1 M Na2 S 3.04 6.26 × 10−15
immediately adjacent in microcells, whereas a net distinction
K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373 4371

Fig. 7. Plots of the conditional event generation rate for the dissolution of Al, r(t) vs. t for Al in aqueous 0.01 M NaOH solution containing various (a) SO4 2− , (b)
NO3 − and (c) Na2 S concentrations.

between anode and cathode is found in macrocells. Accord- tion of Al. According to the stochastic theory, mean time to event
ingly, electrochemical noise measurement appropriately senses generation is defined as [18]
the short-circuited corrosion current that originates mainly from  ∞  ∞  ∞ 
F (t)
macrocells where the anode and cathode are located in differ- tF  (t)dt = (1 − F (t))dt = dt (8)
ent electrodes that are galvanically coupled, resulting in smaller 0 0 0 r(t)
corrosion rate. Considering that the average frequency of events is inversely
By contrast, the corrosion rates calculated from polarisation proportional to mean time to event generation, it follows the
curves and weight losses include the short-circuited corrosion average frequency of events is proportionally related to r(t) and
current which is all provided by microcells as well as macrocells. hence, corrosion rate can be expressed to be proportional to
Moreover, the corrosion rate of Al was increased with increas- the product of r(t) and the average charge in each event. The
ing SO4 2− , NO3 − or Na2 S concentration. It is noted that the corrosion rate increased with the addition of SO4 2− or NO3 −
shot noise analysis by its nature yielded such a remarkable error and such an increase in r(t) is dominant over the decrease of
bound of corrosion rate. the average charge in each event. Furthermore, the corrosion
Corrosion rate just means the product of the average fre- rate remarkably increased with the addition of Na2 S, indicating
quency of events and the average charge in each event for the that it follows the increasing two factors of r(t) and the average
stochastic process of hydrogen bubbles as well as the dissolu- charge in each event.
4372 K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373

Fig. 8. Variation of the corrosion rate for Al in aqueous 0.01 M NaOH solution with the concentration of (a) SO4 2− , (b) NO3 − and (c) Na2 S.

The increase in corrosion rate in the presence of SO4 2− or It is well known that Na2 S acts as a hydrogen recombination
NO3 − was in contradiction to the previously reported results poison which enhances hydrogen permeation into such metal
[40], where corrosion rate of Al was reduced by the addition system as iron, carbon steel or Al alloy by producing H2 S [42].
of SO4 2− or NO3 − under the application of anodic potential. However, at the same time, it also produces hydroxide ions,
It seems that this difference may have its origin in the different resulting in more alkaline condition. Therefore, it is suggested
experimental environments. It should be kept in mind that a that the corrosion rate enhanced by the addition of Na2 S orig-
couple of the partial anodic reaction (Eq. (5)) and the partial inates from the fact that the contribution of the protection by
cathodic reaction (Eq. (6)) constitute a short-circuited corrosion the formation of H2 S is overwhelmed by the contribution of
cell, which is divided into microcell, and macrocell, in alkaline the chemical attack of hydroxide ions in a more severe alkaline
solution at open-circuit potential. solution to the dissolution of Al.
In an aqueous system, Na2 S is easily decomposed by the Hence, from the experimental findings, two corrosion-
following electrolysis reaction [41]: enhancing mechanisms have been proposed. On the one hand,
the corrosion rate increased with the addition of SO4 2− or
Na2 S + 3H2 O → 2H2 S + 2NaOH (9) NO3 − . This can be account for in the following way. The
K.-H. Na, S.-I. Pyun / Electrochimica Acta 52 (2007) 4363–4373 4373

increase in the conditional event generation rate overweighs the [3] S. Real, M.U. Macdonald, D.D. Macdonald, J. Electrochem. Soc. 135
decrease in the average charge in each event for the stochastic (1988) 2397.
process of hydrogen bubbles as well as the dissolution of Al. [4] F. Huet, M. Musiani, R.P. Nogueira, Electrochim. Acta 48 (2003)
3981.
On the other hand, the corrosion rate increased with the addi- [5] C. Gabrielli, F. Huet, M. Keddam, J. Appl. Electrochem. 15 (1985) 503.
tion of Na2 S. This arises from the increase in both factors of the [6] K.-H. Na, S.-I. Pyun, J. Electroanal. Chem. 596 (2006) 7.
conditional event generation rate and the average charge in each [7] K.-H. Na, S.-I. Pyun, H.-P. Kim, Corros. Sci. 49 (2007) 220.
event for the stochastic process of hydrogen bubbles as well as [8] F. Mansfeld, Z. Sun, Corrosion 55 (1999) 915.
the dissolution of Al in more severe alkaline environment. [9] C. Gouveia-Caridade, M.I.S. Pereira, C.M.A. Brett, Electrochim. Acta 49
(2004) 785.
[10] U. Bertocci, F. Huet, Corrosion 51 (1995) 131.
4. Conclusions [11] J.W. Isaac, K.R. Hebert, J. Electrochem. Soc. 146 (1999) 502.
[12] C. Gabrielli, F. Huet, R.P. Nogueira, J. Electrochem. Soc. 149 (2002) E71.
The electrochemical noises obtained from pure Al in an [13] F.R. Connor, Noise, Edward Arnold, London, 1982, p. 48.
alkaline solution have been analysed based upon the combined [14] H.W. Ott, Noise Reduction Techniques in Electronic System, John Wiley
& Sons, New York, 1988, p. 238.
stochastic theory and shot noise theory. From the occurrence of [15] R.A. Cottis, Corrosion 57 (2001) 265.
two linear regions on one Weibull probability plot, the anal- [16] R.A. Cottis, M.A.A. Al-Awadhi, H. Al-Mazeedi, S. Turgoose, Electrochim.
ysis method employed in the present work permitted us to Acta 46 (2001) 3665.
distinguish between the dissolution of Al in the relatively higher- [17] H.A.A. Al-Mazeedi, R.A. Cotis, Electrochim. Acta 49 (2004) 2787.
frequency region, and the stochastic process concerning the [18] J.M. Sanchez-Amaya, R.A. Cottis, F.J. Botana, Corros. Sci. 47 (2005) 3280.
[19] J.C. Fothergill, IEEE Trans. Electr. Insul. 25 (1990) 489.
series of nucleation, growth and detachment of hydrogen bub- [20] J. Rice, Adv. Appl. Prob. 9 (1977) 553.
bles in the relatively lower-frequency region in an engineering [21] J. Masoliver, Physica A 149 (1988) 395.
sense. [22] V. Palleschi, M.R. Torquati, Phys. Lett. Sect. A 138 (1989) 381.
Besides, the conditional event generation rate and the aver- [23] S.B. Lowen, M.C. Teich, Phys. Rev. A 43 (1991) 4192.
age charge in each event were quantitatively determined for the [24] I. Eliazar, J. Klafter, Physica A 360 (2006) 227.
[25] E.E. Lewis, Introduction to Reliability Engineering, John Wiley & Sons,
stochastic process of hydrogen bubbles as well as the dissolu- New York, 1987, p. 140.
tion of Al. In the case of the SO4 2− ion or the NO3 − ion, the [26] S.-I. Pyun, E.-J. Lee, Surf. Coat. Technol. 62 (1993) 480.
conditional event generation rate was increased by the addition [27] S.-I. Pyun, E.-J. Lee, G.-S. Han, Thin Solid Films 239 (1994) 74.
of anions for the stochastic process of hydrogen bubbles as well [28] J.-J. Park, S.-I. Pyun, Corros. Sci. 46 (2004) 285.
[29] P. Homble, W.P. McCormick, J. Appl. Probab. 32 (1995) 707.
as the dissolution of Al, but the average charge in each event was
[30] C. Nuzman, I. Saniee, W. Sweldens, A. Weiss, Comput. Networks 40 (2002)
decreased. Furthermore, both the conditional event generation 319.
rate and the average charge in each event were increased with [31] S. Zaromb, J. Electrochem. Soc. 109 (1962) 1125.
Na2 S concentration. The alkaline corrosion of Al was enhanced [32] S.-I. Pyun, S.-M. Moon, J. Solid State Electrochem. 4 (2000) 267.
by the addition of SO4 2− , NO3 − or Na2 S. [33] A.K.H. Bairamow, S. Zakipour, C. Laygraf, Corros. Sci. 25 (1985)
79.
[34] W.C. Moshier, G.D. Davis, Corrosion 46 (1990) 43.
Acknowledgements [35] C.B. Breslin, G. Treacy, W.M. Cornell, Corros. Sci. 36 (1994) 1143.
[36] K.C. Emregül, A.A. Aksüt, Corros. Sci. 45 (2003) 2415.
The receipt of a research grant for the 2-year period [37] C. Andrade, I.R. Maribona, S. Feliu, J.A. González, S. Feliu Jr., Corros.
2005/2006 from Korea Atomic Energy Research Institute Sci. 33 (1992) 237.
[38] S. Jäggi, H. Böhni, B. Elsener, Eurocorr 2001, Riva del Garda, Italy,
(KAERI) is gratefully acknowledged. Incidentally, this work
September 30–October 4, 2001.
was partly supported by the Brain Korea 21 project. [39] B. Elsener, Cem. Concr. Compos. 24 (2002) 65.
[40] S.-I. Pyun, S.-M. Moon, S.-H. Ahn, S.-S. Kim, Corros. Sci. 41 (1999)
References 653.
[41] J.E. Macintyre, Dictionary of Inorganic Compounds, Chapman & Hall,
[1] T.H. Nguyen, R.T. Foley, J. Electrochem. Soc. 126 (1979) 1885. London, 1992, p. 3645.
[2] C.B. Breslin, A.L. Rudd, Corros. Sci. 42 (2000) 1023. [42] H.-P. Kim, R.-H. Song, S.-I. Pyun, Br. Corros. J. 23 (1988) 254.

You might also like