Electrochimica Acta 46 (2001) 3879– 3886

www.elsevier.com/locate/electacta

Electrochemical methods in corrosion on petroleum

industry: laboratory and field results
L.A.C.J. Garcia a, C.J.B.M. Joia a, E.M. Cardoso b, O.R. Mattos b,*
a
Petrobras R&D, CENPES/DIPLOT/SEMEC, Cidade Uni6ersitária, Rio de Janeiro, RJ, Brazil
b
Laboratório de Corrosão Professor Manoel de Castro, EE/PEMM/COPPE, Uni6ersidade Federal do Rio de Janeiro,
Caixa Postal 68505, CEP 21945 -970 Rio de Janeiro, RJ, Brazil

Received 31 May 2000; received in revised form 8 February 2001

Abstract

This paper presents results concerning the use of a multitask cell coupled with a semi-industrial autoclave. A very
good correlation was found between electrochemical measurements performed with the multitask cell and the
three-electrode probe introduced in one of the autoclave’s flanges. Using the permeation technique, a short time
response system was obtained with the multitask cell. The observed performance of this system was consistent with
the external one attached at the flange. It was observed that neither H2S nor cyanide alone increase the corrosion
process in the alkaline wet H2S environment. The coupled action of these two contaminants is necessary to increase
the hydrogen permeation through the steel wall of the equipment. As a consequence of this work, the multitask cell
was successfully transferred to a PETROBRAS fluid catalytic cracking light end recovery unit. © 2001 Elsevier
Science Ltd. All rights reserved.

Keywords: Hydrogen permeation; Multitask cell; H2S corrosion; Petroleum industry; Corrosion monitoring

1. Introduction sion kinetic at acceptable levels. In the 1990s, heavy

CO2 and H2S are claimed to be the main corrosion
agents present in petroleum plants. In particular, H2S
becomes very dangerous due to its role in the corrosion
process, which generates H0 at the steel surface that
may penetrate into the steel and originate many cracks
and fractures in industrial equipment.
Until the 1970s, only light oil with less than 1000
ppm of nitrogen was usually processed at the petroleum
catalytic cracking units around the world. The anticor-
rosion treatment in the fluid catalytic cracking (FCC)
light end recovery unit was based on water injection to
dilute the aggressive agents, thus maintaining the corro-
* Corresponding author. Tel.: + 55-21-560-8832; fax: + 55-
21-290-6626.
E-mail addresses: cardoso@metalmat.ufrj.br (E.M. Car-
doso), omattos@metalmat.ufrj.br (O.R. Mattos).
petroleum has also been cracked. Recently, offshore
reserves in Brazil present a total nitrogen content of
4000 ppm [1], typical of the highest concentration an-
nounced in the world.
During the cracking process, heavy petroleum gener-
ates a great concentration of CN− and some H2S.
Wilhelm [2] classifies the aggressiveness of a petroleum
by the HS− and CN− concentrations generated during
the catalytic cracking. HS− can be generated by H2S or
NH4HS dissociation. During the H2S attack of the
steel, for a certain range of pH, a film of Fex Sy is
formed at the steel surface, giving rise to a slow corro-
sion kinetic. The presence of CN− causes the removal
of the superficial film by the following reaction:
FeS+ 6CN − = Fe(CN)46 − + S2 − (1)
In the light end recovery unit, different classes of
petroleum are usually processed and the aggressiveness

0013-4686/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 3 - 4 6 8 6 ( 0 1 ) 0 0 6 7 5 - 2
3880 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886

Fig. 1. Autoclave system used for electrochemical measurements. (A) Upper flange, (B) lateral flange, (b1) half-permeation cell, and
(b2) three-electrodes probe.

of the system may change very quickly. Under such
conditions, online information about the kinetic of the
corrosion process is essential.
In a previous paper [3], results concerning the use of
a half-permeation cell coupled to PETROBRAS equip-
ment have been presented. It was demonstrated that it
is possible to detect operational disturbances that give
rise to high hydrogen permeation currents. The main
disadvantage of this cell is related to the long time
interval needed to detect the permeation current. This is
due to the high thickness of the equipment steel wall.
The main purpose of this paper is to present the
results of different electrochemical tests performed with
a semi-industrial autoclave, with the ultimate goal of
monitoring a real plant. Results concerning a multitask
cell already installed in this plant are also shown.
2. Experimental
The autoclave system is shown in Fig. 1A and B. The
purpose of using such a complex system is to simulate
the real situation occurring in the plant and also to
perform several electrochemical tests simultaneously.
The autoclave body, made of Hastelloy-C276, has four
flanges and an inner volume of 150 l. This autoclave
allows one to work with pressures up to 40 atm. The
autoclave’s upper flange (Fig. 1A) has an online pH-
meter, adapted for high-pressure work, and injection
pumps, for the addition of chemical products, such as
H2S, cyanide and H2O2. As can be seen in Fig. 1B, a
conventional half-permeation cell was attached on the
flange wall. A three-electrode probe was also inserted in
this flange to measure impedance and linear polarisa-
tion. In this probe, the working electrode was carbon
steel ASTM A516Gr60 and the reference and the auxil-
iary ones were made of Hastelloy-C276. A by-pass for
a multitask system consisting of a fast response perme-
ation cell (Fig. 2) was also adapted. The multitask
system has a classical electrochemical set-up that allows
two- or three-electrode cell configuration.
All electrochemical measurements were performed
with a digital multi-channel system from Gamry Instru-
ments, controlled by a microcomputer. The following

Fig. 2. Multitask cell used for electrochemical measurements. (A) General view, (B) inner view, (b1) fast response permeation cell,
and (b2) electrochemical probes.
 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886
Fig. 3. Nyquist diagrams, NACE TM 0177. Frequencies in Hz.
Bench and multitask results.
operational steps were sequentially done for the start-
up of an experiment:
1. 150 l of solution were prepared and transferred to
the primary reservoir and deaerated with nitrogen
for at least 96 h.
2. The detection side of the steel membranes used in
the permeation cells (autoclave and multitask) were
nickel-plated (30 mm) [3].
3. An external permeation cell was fixed on the flange
and the three-electrode probe inserted into the
autoclave.
4. The by-pass of the autoclave was installed and the
multitask cell was set-up.
5. The autoclave was evacuated down to 10 − 3 torr
and then nitrogen was introduced for at least 24 h.
6. The working solution was transferred to the auto-
clave without any contact with air and bubbled with
nitrogen for at least 24 h.
After deaerating, chemical products to be tested,
H2S, cyanide or H2O2, were transferred to the autoclave
by specific injection pumps. Two working solutions
were tested:
Fig. 4. Hydrogen permeation current obtained with the multi-
task cell: the effect of H2S without cyanide.
3881
1. NACE TM 0177 solution (0.5% acetic acid [HAc];
5% NaCl) was used with the permeation and the
impedance cells of the autoclave. The results were
compared with those obtained with the classical
laboratory cells [3].
2. The working solution simulating the corrosion ag-
gressiveness of the FCC light end recovery unit was
prepared with the following composition: 2 M of
NH4OH mixed with 2 M of (NH4)2SO4 plus 100
ppm of NaCl. This solution, once saturated with
H2S, is adequate to match the corrosion process
occurring in the unit [2]. In this paper, only the
results with the autoclave kept at 1 atm are pre-
sented. Other tests using 10 atm are under way.
3. Results and discussion
3.1. Laboratory results
The first step of this work was to validate the auto-
clave set-up, using solution 1 described above. All
permeation curves are presented as a function of the
permeation coefficient (P (current/cm)), equal to the
current density (J (current/cm2)) normalised by the
width of the sample (L (cm)). The first test showed that
the permeation results obtained using the multitask and
the classical Devanathan cells were similar, showing
equivalent permeation coefficients for the same times.
In Fig. 3, impedance diagrams for a classical three-elec-
trode cell and the multitask one are shown. A polarisa-
tion resistance around 200 V/cm2 was obtained for both
systems and two capacitive loops are present in the
diagrams. For the multitask cell, the high frequency
capacitive loop was more clearly depicted. This differ-
ence detected by the impedance is a minor point consid-
ering the purpose of showing the equivalence between
the two cells. Indeed, the multitask set-up imposes a
longer exposure time of the working electrode to the
atmosphere before the start-up, which could be the
source of the above difference between the two systems.
Once it was verified that laboratory classical cells and
the autoclave electrochemical systems gave similar re-
sults [3], solution 2 was tested. The following results
were obtained using solution 2. Results concerning the
hydrogen permeation current using the multitask cell
are shown in Fig. 4. As it can be seen, the injection of
H2S does not provide an increase in the permeation
current, as it always occurs for an acid medium [3]. In
fact, for basic pH, a film of Fex Sy is formed at the steel
surface and practically no hydrogen permeation is de-
tected. However, once cyanide is injected into the sys-
tem, the current increases until cyanide is completely
consumed. Actually, reaction (1) removes the sulphide
film, increasing the corrosion process and thus the
permeation current. When the cyanide is consumed, the
3882 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886
film is again formed and the permeation current
decreases.
In Fig. 5, the effect of cyanide can be analysed. Once
the permeation current related to the test solution was
Fig. 5. Hydrogen permeation current obtained with the multi-
task cell: the effect of cyanide without H2S.
Fig. 6. Hydrogen permeation current obtained with the multi-
task cell: the effect of H2O2 inlet.
Fig. 7. Permeation current vs. cyanide concentration.
measured, 4000 ppm of cyanide was injected. It is
possible to see that cyanide did not increase the perme-
ation current as observed for H2S. Later, H2S was
injected for 30 min and then the current started to
increase, until all cyanide was consumed. A new injec-
tion of H2S accelerated the current decrease.
From Figs. 4 and 5, one may notice that the individ-
ual presence of cyanide or H2S is not able to increase
the permeation process. This is a very important result,
because it is empirically adopted in the operation of the
unit. These results also confirmed that the multitask cell
could be used as a tool for online monitoring at field
plants. Indeed, the detection time for different perme-
ation current peaks with the multitask cell is always
shorter than that with the classical half-permeation one,
installed in the autoclave flange. It was noted that with
1.7 mm, the thickness used in the multitask cell, only a
few hours are necessary to detect the current peak
whereas almost 20 h are needed for a 15 mm thickness.
Once a disturbance occurs in the unit, a practical
procedure adopted in the Petrobras plant is to inject
H2O2. Fig. 6 shows that the increase of the current was
drastically interrupted by the H2O2 injection, showing
the efficacy of the multitask cell to detect the changes
that occurred on the permeation process.
The following procedure was adopted to obtain Fig.
7: once the Fex Sy film was formed, a precise quantity of
cyanide was injected and the peak of the permeation
current was plotted. It can be observed that the current
is zero until 27 ppm and then starts to increase with the
cyanide concentration. This result confirms another
important practical aspect, well known in the UFCC
plants: the maximum concentration of cyanide that
does not cause corrosion problems in the presence of
H2S is 20 ppm, being the practical threshold value. A
very near value to the empirical one was verified in Fig.
7 (27 ppm).
Impedance diagrams presented a total correlation
with the permeation process. Fig. 8 shows the diagrams
obtained with the multitask cell and solution 2 at
different test times. For 1.5 h, the diagram presents a
classical behaviour normally detected for passive elec-
trodes with an infinite value of RP. Actually, for this
time a very low permeation current is detected (see Fig.
5). For 190 h (Fig. 8B), the diagram shows at least two
capacitive loops and the RP is very well defined around
450 V/cm2. In this case, the permeation current is very
high and the passive film was totally destroyed by the
cyanide. Later, for 310 h (Fig. 8C), the current is again
very low (Fig. 6), and the impedance displays an infinite
value of RP. For 350 h, the diagram shown in Fig. 8D
is very similar to the one presented in Fig. 8B. A finite
value of RP was measured and the diagram represents
the dissolution process caused by the attack of the
cyanide onto the sulphide film.
 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886 3883

Fig. 8. Nyquist diagrams at different immersion times, multitask cell, NH4OH 2M+(NH4)2SO4,2M solution. (A) 1.5 h. (B) 190 h,
(C) 310 h, (D) 350 h.

The same qualitative results were obtained with the
probe cell inside the autoclave (Fig. 9). In Fig. 9A the
diagram with infinite RP represents the condition where
a protective film is covering the electrode surface. Fig.
9B is correlated to the dissolution process imposed by
the cyanide.
The effect of H2O2 is shown in Fig. 10. In Fig. 10A
the impedance diagram for 310 h can be seen. In this
case, the sulphide film is present at the electrode surface
and the RP is infinite. For 360 h (Fig. 10B) the condi-
tion is the opposite: cyanide is attacking the electrode
surface and RP is finite. In Fig. 10C, the impedance
diagram 3 h after the injection of H2O2 is presented.
This diagram displays an infinite RP due to the role of
the H2O2 over the corrosion attack.
The last point to be emphasised concerns the use of
linear polarisation or the so-called noise resistance to
monitor the corrosion process. As it was possible to
observe by the impedance results, the difference be-
tween a corrosive system (presence of the couple cya-
nide/H2S) and a passive one (presence of sulphide film)
was the shape of the diagram. In the last case, an

Fig. 9. Nyquist diagrams at different immersion times, autoclave, NH4OH 2M+(NH4)2SO4, 2 M solution. (A) 312 h representing
the same conditions as Fig. 8C; and (B) 360 h representing the same conditions as Fig. 8D.
3884 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886

infinite RP is seen; in the other case, the RP is finite.
However, sometimes the same value is found even if a
low frequency is used to measure the RP, for example 3
mHz in diagrams B and C in Fig. 10. This result points
out the fact that it could be very difficult to monitor the
aggressiveness of petroleum inside this unit using only
the RP value. Indeed, using the concept of RP, whatever
the experimental technique, the same values can be
obtained in both conditions: corrosion and passivation.
3.2. Field results
In Fig. 11, results obtained with the multitask cell
working in the field conditions are presented. Fig. 11A
shows the first hours after the start-up. As can be seen,
the cell is very sensitive, and presents a very fast
response. Indeed, the few changes introduced in the first
2 h concerning the inlet solution were immediately
detected by the cell. After 2 h, the cell has been lined
with the unit. Fig. 11B presents the hydrogen current
plot, where it is possible to see an average value of 0.2
mA/cm1, suggesting that the system is not under aggres-
sive conditions [3]. Fig. 12, obtained with the half-per-
meation cell confirms this non-aggressive condition [3].
The electronic equipment used to perform the elec-
trochemical measurements was placed in an air-condi-
tioned room 20 m apart from the field cells, due to
safety precautions. This condition could make it impos-
sible to measure the impedance due to noise and the
cable’s impedance problems. Noise was minimised by a
very careful shield system and fortunately the required
frequency range was not high enough to be influenced
by the cable’s impedance. Measurements for the same
dummy cell placed at different locations, one time in
the air-conditioned room and the other time placed
around the field cells were done, and the same RC
behaviour was obtained for both cases, proving that
impedance could be measured in such experimental
conditions.
In Fig. 12 typical impedance plots for the multitask
cell at field conditions and in the autoclave are pre-
sented. A capacitive loop is seen around 16 Hz followed
by a linear behaviour in low frequencies. The same
qualitative diagrams were obtained and an opened plot
is seen in the low frequency range for both conditions.
This opened diagram, as previously discussed, repre-
sents a non-aggressive medium in which the electrode is
covered by a surface film of Fex Sy. This fact is corrob-
orated by the permeation curves presented in Fig. 11
where a low value has been detected for hydrogen
permeation. The difference noted in the high frequency
capacitive loop in Fig. 12 can be attributed, as previ-

Fig. 10. Nyquist diagrams, NH4OH 2 M+ (NH4)2SO4, 2 M solution with H2S. (A) Open loop representing the presence of a passive
film (Fex Sy ), (B) the effect of cyanide inlet, and (C) the effect of H2O2 inlet.
 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886 3885

Fig. 11. FCC light end recovery unit; water drain from the high pressure vessel, multitask cell. (A) First 10 h after start-up, and (B)
Typical permeation curve.

ously done, to a few differences which occurred during 4. Conclusions

the start-up of each experiment. Actually, the time in
which the working electrode was in contact with the
atmosphere is not the same due to experimental factors.
Finally, it is expected that an increase in the corrosion
attack (operational disturbance) will give rise to an
increase in the hydrogen permeation and thus a closed
impedance diagram will be obtained.
The results presented in this paper have shown that
the multitask cell was able to detect the kinetic effects
caused by cyanide, H2S and H2O2. It was seen that H2S
or cyanide alone were not able to increase the perme-
ation current. Only the coupled effect of these contam-
inants can increase the corrosion process in the FCC
3886 L.A.C.J. Garcia et al. / Electrochimica Acta 46 (2001) 3879–3886

Fig. 12. Nyquist diagrams, field and laboratory results, multitask cell. Frequencies in Hz.

light end recovery unit. Results presented in this paper
allowed the installation of the multitask cell at the field
plant, as an online control. Finally, it was seen that it is
very difficult to monitor the corrosion process by the
RP value, whatever the technique used to measure this
parameter.
Acknowledgements
The authors thank the financial support of CNPq,
CAPES, FAPERJ, FINEP and FUJB Brazilian agen-
cies. The authors are also grateful to the PETROBRAS
that allows the present results to be published.
References
[1] L.A. Correa, W. Baptista, L.A. Ferreira, Emprego do
peróxido no controle da corrosão por H2S e CN− em meio
aquoso, Relatório Interno Petrobras, Brasil, 1998.
[2] S.M. Wilhelm, D. Abayarathana, Inhibition of Absorption
of Hydrogen by Steel in Wet H2S Refinery Environments,
Corrosion, paper 449, USA, 1992.
[3] C. Azevedo, P.S.A. Bezerra, F. Esteves, C.J.B.M. Joia,
O.R. Mattos, Hydrogen permeation studied by electro-
chemical techniques, Electrochimica Acta 44 (1999) 4431.