www.elsevier.com/locate/corsci

Corrosion Science 46 (2004) 1969–1979

Corrosion of nickel-containing stainless steelin concentrated sulphuric acid

Yanxu Li, M.B. Ives *, K.S. Coley, J.R. Rodda

Walter W. Smeltzer Corrosion Laboratory, McMaster University, 1280 Main Street West,

Hamilton, ON, Canada L8S 4L7

Received 7 February 2003; accepted 9 October 2003

Available online 9 April 2004

Abstract

The corrosion of UNS S30403 stainless steel was investigated in 93.5 wt% sulphuric acid at

temperatures from 40 to 100 �C. Time-dependent free corrosion potential measurements

demonstrate that the steel is subject to spontaneous active–passive oscillation. Studies on the

effect of temperature and electrode rotation speed indicate that the oscillation is activation and

transport controlled over different segments of the oscillations. It is concluded that the

oscillation is caused by reversible changes in the exchange current for the cathodic reduction of

sulphuric acid, depending on the presence or absence of nickel sulphide on the steel surface.

� 2004 Published by Elsevier Ltd.

Keywords: Stainless steel; Polarization; Acid corrosion; Passivity

1. Introduction

Nickel-containing austenitic stainless steels type UNS S31600 (Ni 12.0 wt%) and

alloy 825 (Ni 41.7 wt%) were first reported to exhibit a spontaneously oscillating

activation–passivation sequence when immersed in concentrated sulphuric acid [1].

The phenomenon has been confirmed by further studies [2–8]. The oscillation isconsidered to be provoked by the reproducible formation and dissolution of an

intermediate phase in the acid [2–8]. Understanding the nature of the intermediate

phase is critical to any interpretation of mechanisms. This phase has been argued to

be sulphates [2–6] or nickel sulphides [7,8]. Its formation is rapid but its dissolution is

* Corresponding author. Fax: +1-905-528-9295.

E-mail address: ives@mcmaster.ca (M.B. Ives).

0010-938X/$ - see front matter � 2004 Published by Elsevier Ltd.

doi:10.1016/j.corsci.2003.10.017

1970 Y. Li et al. / Corrosion Science 46 (2004) 1969–1979

much slower, thus determining the cyclic nature of the open circuit potential [2–8].

The present study describes an instant blackening film which exists only for a short

period of time believed to correspond to the formation of an intermediate phase. The

present study has enabled the investigation of a nickel sulphide model [7,8] based on

observations on the effects of temperature and acid velocity (electrode rotation)

conditions.

2. Experimental

Full details of the experimental setup for the electrochemical measurements were

presented previously [7–9]. Hollow cylindrical steel specimens were incorporated intoa standard rotating cylinder electrode (RCE) with dimensions: Inner diameter 0.7

cm, outer diameter 1.2 cm, height 0.8 cm. The alloy was UNS S30403 plate (Fe 71.6

wt%, Cr 18.1 wt%, Ni 7.9 wt%, C less than 0.01 wt%, and other elements less than 1

wt% total). The outer surface of the RCE was wet-ground by 600 grit emery paper to

a uniform surface finish. The sulphuric acid was made from BDH reagent grade acid

(Cl� 0.2 ppm, NO�3 0.5 ppm, NHþ

4 2 ppm, SO2 2 ppm, Pb 1 ppm, Fe 0.2 ppm, As

0.01 ppm, Hg 5 ppb). Temperature of sulphuric acid was controlled within ±0.5 �Cby a Digi-Sense temperature controller.

A typical three-electrode system was adopted for the electrochemical analyses. A

Koslow mercurous sulphate (Hg/Hg2SO4) electrode (MSE) with a practical potential

of 0.513 VH was adopted as reference electrode (RE). This electrode is more stable

than the platinum electrode typically used in electrochemical studies of hot con-

centrated sulphuric acid. A spiral platinum wire was used as the counter electrode

(CE). Electrochemical measurements were controlled by EG&G software Model 352

through an EG&G Princeton Applied Research 273 Potentiostat/Galvanostat.

As there is a rapid potential change during the active–passive transition of theoscillation (see Fig. 1(a)), resolution of the potential–time curves during the short

time at active (‘‘spike’’) potentials required particularly fast potential sampling rates.

In this study a set of measurements were made with sampling to 0.08 s per point, to

ensure precise acquisition of potential changes.

3. Results

3.1. Time-dependent corrosion potential curves and polarization curves of UNS S30403

in 93.5 wt% H2SO4

Fig. 1(a) presents the potential oscillation of UNS S30403 steel superimposed on

the forward scanning potentiodynamic polarization of the steel at 1000 rotation per

minute (rpm) in 93.5 wt% H2SO4 at 60 �C. The reproducible periodical potentialoscillation develops after an initial potential variation which is found to depend on

the particular sample condition when measurement is started. The peak potential is

0.02 VMSE and the spike potential is )0.58 VMSE. The average period of oscillation is

0 1 2 3 4-0.8

-0.6

-0.4

-0.2

0.0

0.2-6 -5 -4 -3 -2 -1 0

Peak

Spike

Potential oscillation of steel at 60°C

E (

V/M

SE)

E (

V/M

SE)

t (h)

Potentiodynamic polarization of steel

0 1 2 3 4 5-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

Passivation

Activation

Passivation induction

t (s)

(a)

(b)

Fig. 1. (a) Oscillation of free corrosion potential of UNS S30403 RCE (1000 rpm) steel in 93.5 wt% H2SO4

at 60 �C at 1.2 s/point superimposed on the anodic polarization of the steel; (b) potential spike of UNS

S30403 RCE (1000 rpm) stainless steel in 93.5 wt% H2SO4 at 60 �C at 0.08 s/point.

Y. Li et al. / Corrosion Science 46 (2004) 1969–1979 1971

about 1.1 h. The steel is in a passive state at the peak potential and is in an active

state at the spike potential.

The details of a spike during the oscillation are shown in Fig. 1(b). Apparently

there are three processes occurring sequentially within the spike: A rapid activation,

a relatively slow passivation induction (from )0.58 to )0.5 VMSE), and a rapid

passivation. The spike duration is defined as the total time for these three processes.

As soon as a fresh specimen was exposed to the acid, bubbles were found to beproduced immediately and the sample surface became black. An instant ‘‘blacken-

ing’’ at the surface was observed corresponding to each spike on the potential curve.

The ‘‘blackening’’ film or cloud immediately disappeared when the potential rose

and the sample surface became shiny again.

1972 Y. Li et al. / Corrosion Science 46 (2004) 1969–1979

3.2. Effect of temperature on potential oscillation

Fig. 2(a) and (b) present the time-dependent potential curves of UNS S30403

stainless steel (RCE 1000 rpm) in 93.5 wt% H2SO4 at 40, 60, 80, and 100 �C at a

sampling rate of 1.2 s/point. Fig. 2(c)–(f) present the active spikes sampled at 0.08

s/point. The oscillation period and spike duration decrease significantly withincreasing temperature.

Fig. 2. (a,b) Effect of temperature on potential of UNS S30403 RCE (1000 rpm) stainless steel in 93.5 wt%

H2SO4 at 1.2 s/point, and (c–f) effect of temperature on spike duration of UNS S30403 RCE (1000 rpm)

stainless steel in 93.5 wt% H2SO4.

Y. Li et al. / Corrosion Science 46 (2004) 1969–1979 1973

3.3. Effect of velocity on potential oscillation

Fig. 3(a) shows the potential of the UNS S30403 RCE in 93.5 wt% H2SO4 at

60 �C at 1.2 s/point at 0, 500, 1000, and 1500 rpm. The oscillation period decreases

monotonically with increasing velocity. During the experiment at 0 rpm a bright

yellowish film was observed on the surface. This was found to be partly insoluble in

Fig. 3. Effect of rotation speed on (a) potential of UNS S30403 RCE stainless steel and (b–e) spikes in 93.5

wt% H2SO4 at 60 �C.

1974 Y. Li et al. / Corrosion Science 46 (2004) 1969–1979

distilled water after it was removed from the acid. However, no such film was seen at

1000 and 1500 rpm.

Fig. 3(b)–(e) resolve the active spikes to 0.08 s/point at each condition. The

rotation speed does not have a significant effect on the spike duration it being from

2 to 3 s at all speeds.

3.4. Corrosion of UNS S30403 and temperature and velocity effects

It can be concluded from the experimental results that temperature has a signif-

icant effect on oscillation period and the activation–passivation transition spikeduration, while rotation speed has an effect only on the passive part of oscillation

period, and not on the active spike width. Apparently the whole process is alter-

natively under both activation and transport control, but the passivation induction

period, defined in Fig. 1(b), is solely a thermally activated process since it is not

dependent on electrode rotation rate.

Fig. 4 shows an Arrhenius plots for the passivation induction time at 1000 rpm.

The total oscillation period at 1000 rpm is also presented in a similar way. The

passivation induction appears to be a single thermally activated process, with anactivation energy of 75 kJ/mol, while the overall oscillation period does not obey

Arrhenius behaviour suggesting it is not governed by a single process. The effect of

mass transport (Section 3.3) must be responsible for the difference. The velocity

dependency, indicating mass transport control, becomes more significant when the

temperature rises above 80 �C. This temperature effect is also consistent with a

transition from chemical to mass transport control.

3.5. Effect of dissolved metal ions on potential oscillation

The effects of dissolved Ni2þ ion and Fe3þ on the oscillations were also studied by

adding NiSO4 and Fe2(SO4)3 salts into the acid. Fig. 5 illustrates the effect of Ni2þ on

-10

-8

-6

-4

-2

0

2

4

0.0026 0.0028 0.003 0.0032 0.0034

1/T (K)

ln(1

/t) (

s)

Spike duration - temperatureOscillation period - temperatureLinear fit (Spike duration)

100°C

80°C60°C 40°C

Fig. 4. The Arrhenius plot of temperature effect on the periods of oscillation and passivation induction.

-0.60 1 2 3 4

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

200 ppm Ni2+

500 ppm Ni2+

Pure acid

E (

V/M

SE)

t (h)

Fig. 5. Effect of 200 ppm and 500 ppm Ni2þ impurities on corrosion potential of UNS S30403 RCE (1000

rpm) in 93.5 wt% H2SO4 at 60 �C.

0 1 2 3 4-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2 pure acid 36 ppm Fe3+ 50 ppm Fe3+

E (

V/M

SE)

t (h)

Fig. 6. Effect of 36 and 50 ppm Fe3þ impurities on corrosion potential of UNS S30403 RCE (1000 rpm) in

93.5 wt% H2SO4 at 60 �C.

Y. Li et al. / Corrosion Science 46 (2004) 1969–1979 1975

the corrosion potential. A Ni2þ concentration of 200 ppm slightly decreases the

oscillation frequency compared with the oscillation without any Ni2þ. However, the

steel potential becomes stable in the passive state (at least for the duration of the test)

when there is 500 ppm Ni2þ in the acid. The same stable potential is also obtained

when there is more than 36 ppm Fe3þ in the acid (Fig. 6).

1976 Y. Li et al. / Corrosion Science 46 (2004) 1969–1979

4. Discussion

4.1. Model for spontaneous potential cycling

The above results permit a model for the potential oscillation phenomenon in

nickel-containing stainless steels, summarized in Fig. 7. The model builds on theconcepts first outlined by Kish [7,9].

Since this active–passive process occurs spontaneously, the active and passive

states of the steel might be induced by changes in the cathodic reaction. The

reduction reaction on the bare steel surface might appears to be too slow to polarize

the steel beyond the passivation potential thereby maintaining the anodic reaction of

the steel in the active state range. At this potential the nickel sulphide may be

produced as an intermediate phase on the steel surface [7,8]. The reduction on this

phase may then be sufficient fast to spontaneously passivate the steel. The reductionreactions on the bare steel and on the nickel sulphides are simplified in Fig. 7 and the

mechanism based on this model is discussed below.

4.1.1. Active dissolution

When a steel sample with a well polished surface is first exposed to acid the

governing cathodic reactions are hydrogen evolution and the reduction of sulphuric

acid by a reaction such as:

H2SO4 þ 8Hþ þ 8e� ! H2Sþ 4H2O ð1Þ

The consequent mixed potential results in active anodic dissolution of the alloy. One

of the dissolving species is Niþþ which reacts with the hydrogen sulphide to form anickel salt. The salt has been analyzed to be nickel sulphide in a previous ex situ XPS

study under the same conditions [8,9]. The active reactions are heralded by the black

film/cloud observed at the metal surface only in this potential region. It takes some

-7 -6 -5 -4 -3 -2 -1 0-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

Reduction of sulphuric acid on NiS

Reduction of sulphuric acid on 304 steel

E (

V/M

SE)

log I (A)

Fig. 7. Model of potential oscillation.

Y. Li et al. / Corrosion Science 46 (2004) 1969–1979 1977

time (1–3 s) before the nickel sulphide nucleates as a solid phase on the alloy surface

or before a passive film is formed. This process is entirely surface reaction-controlled,

as has been discussed in Section 3.4, and is dependent on temperature but inde-

pendent of transport from the acid.

4.1.2. Passivation

Once sufficient nickel sulphide is nucleated, the cathodic reduction of sulphuricacid (probably to SO2 at this potential [9]) now takes place on the sulphide,

H2SO4 þ 2Hþ þ 2e� ! SO2 þ 2H2O ð2Þ

thereby increasing the exchange current and producing a mixed potential in the

passive range. It is probable that the sulphide surface is quite irregular, providing an

effective surface area which is much greater than the geometric area of the electrode.

The passive film so formed at this potential is expected to be a passivating chromium

compound (oxide).

4.1.3. Loss of passivity

At the passive potential, nickel sulphide formation is not thermodynamicallypossible, so the unstable sulphide itself will be dissolved. From experimental

observations, it appears that this dissolution is slow. It depends on temperature and

the efficiency of transport rate of (nickel) ions into the bulk acid. This latter is

dependent on both the rotation rate of the electrode (Fig. 3(a)) which determines the

ion transportation and the concentration of dissolved Ni2þ ions in the solution

(Fig. 5) which inhibits the dissolution of the nickel sulphide. The dissolution causes a

decrease in the effective cathodic reduction area, as well as a change back to a less

active cathodic surface type, reducing the cathodic current and the consequent mixedpotential. The peak potential declines steadily during this process.

Support for this model is provided by an additional observation, depicted in

Fig. 8. If a sample is potentiostatically maintained at a passive potential for 6 h and

then switched to ‘‘open circuit’’, the potential drops quickly down to the active spike

in 10 min. In Fig. 8, zero time is set at the point when the open circuit potential

begins to decrease from the passive range. However, if the sample is immersed in the

acid with no applied potential––the normal condition for these experiments––the

dashed line clearly shows that the potential requires 1.4 h to decrease to the nextactive spike. Subsequent oscillations for the two samples are similar, indicating that

the difference lies in the conditions imposed at the beginning of the experiment.

The difference in the decay of the peak potential between the normal initial

condition and the polarized condition is a consequence of the role played by an

intermediate phase. In case A the intermediate phase is able to nucleate at a lower

potential and takes some time to dissolve at the peak potential. In case B any pre-

existing intermediate phase would have more than sufficient time to dissolve when

the potential was held for 6 h at 0 VMSE. This result indicates that the intermediatephase has a significant effect on holding the potential in the passive range and

strongly suggests that the duration of the passive peak depends on the time for

dissolution of that phase.

-1.0 -0.5 0.0 0.5 1.0 1.5 2.0-0 6

-0 5

-0 4

-0 3

-0 2

-0 1

0 0

0 1B

A

E (

V/M

SE)

t (h)

A: Normal Open Circuit Potential B: Open Circuit Potential after Polarization

.

.

.

.

.

.

.

.

Fig. 8. Open circuit potential of static UNS S30403 stainless steel in 93.5 wt% sulphuric acid at 50 �Cunder two conditions: normal condition and after 6 h potentiostatic polarization at 0 VMSE.

1978 Y. Li et al. / Corrosion Science 46 (2004) 1969–1979

4.1.4. Re-activation

Eventually, sufficient nickel sulphide is removed from the surface that there isenough bare alloy for the dominant cathodic process to be the reduction of acid on

bare steel, and the potential returns to the active range. The cycle continues thus.

4.2. Effect of dissolved Fe3þ on the potential oscillation

When there are oxidizing impurities in the acid, for example, Fe3þ, the total

cathodic reduction becomes a combination of two or more reactions; the potentialoscillation will change accordingly.

The redox reaction of Fe3þ/Fe2þ in aqueous solutions at 25 �C can be charac-

terized by the following Nernst equation,

Fe3þ þ e� $ Fe2þ; E ¼ 0:771þ 0:0591 logð½Fe3þ=½Fe2þÞ

The standard redox potential of Fe3þ/Fe2þ (0.771 VSHE, or 0.258 VMSE) is much

higher than the passive peak potential (0.02 VMSE). The redox potential will vary

with the activities of Fe3þ and Fe2þ ions and the temperature. It is reasonable to

assume the redox potential to be close to or higher than 0.771 VSHE when there is

more Fe3þ than Fe2þ in the acid. It is likely that the reduction of the Fe3þ can

maintain the mixed potential in the passive range once the steel is passivated. The

situation is presented in Fig. 6 with over 36 ppm ferric ion in the acid. A reasonableexplanation is that the exchange current density of the reduction of Fe3þ on the steel

surface should be sufficient to hold the potential once the steel has reached the

passive range.

Y. Li et al. / Corrosion Science 46 (2004) 1969–1979 1979

5. Conclusions

The current investigation has enabled the interpretation of the spontaneous po-

tential oscillation mechanism of UNS S30403 stainless steel in 93.5 wt% sulphuric

acid at different temperatures. The following conclusions have been drawn there-

from:

1. The oscillation is caused by reversible changes in the exchange current for the

cathodic reduction of sulphuric acid, depending on the presence or absence of

nickel sulphide on the steel surface.

2. The nucleation of the nickel sulphide is solely a thermally activated process,

whereas the overall oscillation process is a combination of both activation and

transport controlled steps.

3. As a consequence, the oscillation is not only affected by redox reactions (such asFe3þ/Fe2þ) in the solution which changes the mixed potential, but also the concen-

tration of Ni2þ in the solution which inhibits the nickel sulphide dissolution.

Acknowledgements

This research was supported by the Natural Sciences and Engineering Research

Council of Canada under a Cooperative Research and Development Grant con-

tributed to by a consortium comprising Cecebe Technologies Inc., Cominco Ltd.,

Falconbridge Limited, Inco Limited, Kubota Metal Corporation, NORAM Engi-

neering and Constructors Ltd., Noranda Inc.

References

[1] R.M. Kain, P.E. Morris, Paper No. 149, Corrosion/76, NACE, Houston, TX, 1976.

[2] Y.S. Chang, Periodic Active–Passive Corrosion Behaviour, Ph.D. Dissertation, Cambridge University,

1984.

[3] M.H.W. Renner, Corrosion Behaviour of Stainless Steel and Nickel Alloys in Hot Concentrated Static

and Flowing Sulphuric Acid, Ph.D. Dissertation, University of Teesside, 1992.

[4] M.H.W. Renner, Paper No. 01342, Corrosion/2001, NACE International, Houston, TX, 2001.

[5] R. Matsuhashi, Corrosion Engineering 36 (1987) 531.

[6] R. Matsuhashi, Corrosion Engineering 39 (1990) 89.

[7] J.R. Kish, M.B. Ives, Passivity of Metals and Semiconductors, Proceedings of the Eighth International

Symposium, Electrochemical Society, Pennington, NJ, 2000, p. 915.

[8] J.R. Kish, M.B. Ives, J.R. Rodda, Corrosion mechanisms of nickel stainless steel in concentrated

sulphuric acid, Corrosion, submitted for publication.

[9] J.R. Kish et al., Journal of the Electrochemical Society 147 (2000) 3637.