Study of the Passive Layer on Steel by Hydrogen Permeation
T. Casanova and J. Crousier
Laboratoire de Physico-chimie des matériaux.
Université de provence, 13331 Marseille cedex 3, France
email: corelec@newsup.univ-mrs.fr
Introduction
Several papers deal with hydrogen permeation through steel using the Devanathan two-compartment cell [1], mostly in the aim of determining the hydrogen diffusion coefficient. The steel membrane was either palladium plated on both sides [2-5], or on the detection side [2, 6, 7], or on the input side [3 - 5], or without Pd coating [3, 8-10]. In theses last cases, the effect of an iron oxide film on the exit side was also studied. The aim of this study was to study the modifications of the oxide layers, which exist on both the input and the exit sides of a permeation lamella, under hydrogen flux. The oxide layers were built under well-controlled electrochemical conditions, and their stability could be checked by the stability of the passive current.
In a Devanathan cell, the two sides of the steel membrane are submitted to two different solutions and different polarisations, and therefore a preliminary determination of the steel behaviour in NaOH solutions was necessary. In this study, the input side of the cell was filled with a 3.75M NaOH solution, this concentration being the one that will be used for further experiments (mostly plating experiments). Hence the steel behaviour was studied in this normally aerated solution. The input cell contains the working electrode (one side of the steel membrane) and a platinum counter-electrode. After long time cathodic polarisation, the NaOH solution was oxygenated by oxygen produced at the anode, where water oxidation occurred. Hence the electrochemical behaviour of steel was also studied in an oxygenated solution. After permeation through the iron membrane, the exit hydrogen was ionised following the scheme in figure 1, in a 0.1M NaOH deaerated solution, and hence building an oxide layer in this solution was also studied.
Some authors [2, 11, 12] report on the reduction, or quasi-reduction, of the passive film formed on the input side of an iron membrane, the detection side being covered by a layer of electrodeposited palladium. It was found that for long time cathodic polarisation, the hydrogen permeation increased with time due to the higher permeability of the membrane after reduction of the passive film. This last result asks questions about the stability of the oxide layer formed on the exit side of the iron lamella under hydrogen transport through the oxide layer. Hence the stability of the exit passive layer was checked.
Experimental procedure
Passivation of steel
The steel behaviour was tested by cyclic voltammetry in deaerated 0.1M NaOH, aerated 3.75M NaOH and oxygenated 3.75M NaOH solutions. All solutions were made of Milli-Q water (Millipore). In fact the oxygenated solution was the solution used for 5 days in the input cell in which strong oxygen evolution was observed at the platinum counter-electrode. After stabilisation of the rest potential, voltammograms were run; the sweep started from the open circuit potential towards anodic potentials. For these preliminary tests, a bulk iron electrode polished before experiments in the usual way, and a classic three-electrode cell were used. All the potentials were recorded with respect to a saturated calomel electrode.
Devanathan cell
The scheme of the Devanathan cell is shown in Figure 2. The membrane consisted of mild steel 100 µm thick. The free area of the lamella, which was submitted to permeation was 10 cm2. We observed that small permeation areas lead to a dispersion of the results, but by using 10 cm2 samples, the permeation currents were perfectly reproducible. The sheet was cleaned by dipping it in 50% HCl, and then rinsed just before putting it in position in the cell.
Figure 2: Devanathan cell
Before filling up the detection cell, the 0.1M NaOH solution was carefully deaerated in a separated apparatus by bubbling argon for several hours, and then siphoned under argon to the cell. Light argon bubbling continued during the experiment. In this way, the detection side of the membrane was in contact with air and then with an already deaerated solution. Following the results obtained by the preliminary experiments on bulk iron, the passive layer was formed by an anodic polarisation until the background anodic current became constant at 0.03 ± 0.02 µA cm-2. Then the 3.75M NaOH solution was admitted in the input cell and the entry side of the membrane was immediately cathodically polarised for generating hydrogen. Current changes with time on the detection side were registered to obtain the hydrogen permeation transients. For long time cathodic polarisation, some works found that hydrogen permeation increased with time [2, 11, 12]. This increase was due to the higher permeability of the membrane after reduction of the passive layer on entry side. This result asks questions about the stability of the oxide layer formed on the exit side under hydrogen transport. The stability of the passive layer was checked.
Results
Passivation of iron in NaOH solutions
0.1M NaOH solution
The solution was carefully deaerated and the only reduction reaction was water reduction, whose thermodynamic potential was about -1057 mV/sce, at pH = 14. The potential changes with time indicate spontaneous passivation and potential stabilisation at -380 mV. In an oxidising solution, the spontaneous passivation (or chemical passivation) is due to the presence of oxidising species and the corrosion potential stabilises in the passive region. In a deaerated solution, not only there is not any oxidising species, but the thermodynamic potential of water reduction is too active to intervene in the oxidation process. The spontaneous passivation curve can be only explained by an air formed oxide, which upon immersion gives a very protective hydrated oxide. Figure 3 shows the current changes with time for the formation of the passive layer under -180 mV. Thickening of the passive film is evidenced by the decrease with time of jpass . A steady state was reached with a background current on the nA scale. A literature review shows that the anodic oxide film built in NaOH solution has a thickness of some nanometers, for instance in 0.1M NaOH the film is d=2.8 nm [13]. An experiment conducted by polarising a sample immediately after polishing showed that the passive layer cannot be built, the current keeping a relatively high value, which confirms the hypothesis of a passive layer built from the air-formed oxide. To check the stability of the oxide layer, the polarisation was maintained for 30 h without observing modifications of the current density, showing that the passive layer was quite stable for time long enough to estimate that if a break of the passive layer occurs, it will be due to modifications bring by hydrogen permeation through the oxide film.
Figure 3: Intensity-time curve
for passivation under anodic polarisation at -180 mV
3.75M NaOH solution
Figure 4 reports the free-potential changes with time for both aerated (curve 2) and oxygenated (curve 1) solutions, showing that passivation is easier in an oxygenated solution. In both cases, the curve exhibits two potential plateaux at -850 and -400 mV, preceded by a short plateau at -1200 mV; in the aerated solution an other plateau arises at -1000 mV. Cyclic voltammograms for the same solutions are reported in figure 5. The voltammogram for the aerated solution presents one anodic peak and several humps. On the reverse scan it shows two well-formed reduction peaks. The curve for oxygenated solution is similar but does not show sharp peaks. The small value of the anodic currents is in keeping with a passivation process. This result is in accordance with either a passive film consisting of different oxides (or hydroxides), or a multi-step formation of the passive layer.
It has been said that the air-formed film can be reduced by cathodic polarisation [14, 15], but other results [5, 11, 16] have shown that an oxide layer was never completely removed, and that a porous layer remained on the surface. This film could be completely removed in presence of EDTA [2]. After polishing of the iron electrode, it was immersed immediately in the NaOH solution (rest potential -700 mV) or left in air for several min. before immersion (rest potential -500 mV). Several cathodic polarisation curves were run to determine whether cathodic reduction of the air-formed oxide occurs. As soon as immersed in NaOH solution, the steel electrode was polarised for several minutes or hours, either potentiostatically at a potential cathodic to the thermodynamic potential of water, or galvanostatically with various cathodic currents, and then left in free corrosion. Whatever the type of polarisation and its duration, the potential was -1.2 V after switching off the polarisation, and then increased rapidly to stabilise at -1.070 V. That result shows that the oxide film was rapidly reduced but also rapidly built.
figure 4: Rest potential versus time for oxygenated (curve
1) and aerated (curve 2) 3,75 M NaOH solutions
Figure 5: Cyclic voltammograms
for oxygenated (curve 1) and aerated (curve 2) 3,75 M NaOH solutions.
Scan rate 5 mV s-1.
The Devanathan cell
The exit side of the membrane must be polarised at a potential high enough to oxidise the hydrogen atoms before their recombination, which would give hydrogen bubbles (see scheme in Fig. 1), i.e., to obtain a steady state adsorbed hydrogen concentration equal zero. The free potential stabilised at about -380 mV, and the choice of the ionisation potential was not very large, since it must be higher than the free potential. A small overpotential will give a potential high enough to ensure ionisation of all hydrogen atoms which permeated and in the present paper the exit side of the membrane was polarised at -180 mV. Some results have also shown that the passive film on steel was modified by hydrogen permeation [4, 17]. Modifications were also observed after hydrogen evolution on titanium oxide [18]. The modifications were explained by a partial reduction of the oxide to a lower oxidation state by hydrogen, followed by chemical dissolution and film loading with hydrogen species. The stability of the oxide layer on the exit was tested by experiments run for several days to check the effect of hydrogen permeation through the oxide. Hydrogen permeation was carried out without breaking for five days. After switching off, the potential of the exit side of the membrane was still about -380 mV, and slightly higher when the permeation went on for more than five days. It seems that hydrogen permeation through the oxide layer did not reduce the exit side oxide layer. Atomic hydrogen is a strong reducing agent and since the oxide layer on the input side was rapidly reduce by hydrogen evolution onto the oxide surface, the fact that the oxide was not reduced by atomic hydrogen through the oxide layer was unexpected. This result asks a question about the type of hydrogen species which was moving through the oxide.
The inset in figure 6 shows a typical transient response for hydrogen permeation under galvanostatic polarisation of the input side of 10 mA cm-2. The plot shows an increase in current during a time which depends, for a given metal, on the thickness of the lamella. Then a steady state current J_ was reached. An experiment was run to show the influence of the input current on the hydrogen oxidation current. Figure 6 shows the transients obtained for 0.4, 1 and 2 mA cm-2; for higher input currents Jï stopped increasing. Identical steady state currents were obtained regardless the overpotential applied at the exit side, showing that an overpotential of 200 mV is a suitable value.
An expression for the steady state permeation current Jï in terms of the steady state concentration of hydrogen atom q on the surface has been reported [6] showing that the permeation current is linked to the hydrogen coverage and that limiting adsorption of hydrogen on the surface could reduce the hydrogen embrittlement. It is worth noticing that modifications of hydrogen permeation can be observed in presence of organic species capable of adsorbing on the surface, and acting as competitor with adsorbed hydrogen for sites on the permeation surface. These observation suggest that modifications of the oxide state of the input side could be detected by the values of the permeation current.
If the permeation was continued for longer times, more than three hours, the current increased again to stabilise after ten hours (Fig. 7). To explain the rise in oxidising current after the stabilisation at Jï, two hypotheses arise after the literature review: this high current density can be due either to a modification of the passive film on the detection side or to the reduction of the oxide on the input side, and some experiments were run to support one of this hypothesis.
Figure 6: Permeation transients for in 0.4, 1 et 2 mA cm-2. Inset: typical current- time transient.
Figure 7: Permeation transient for Jin = 10 mA cm-2
for 15h
Exit side - As shown above, the oxide layer on the exit side is stable even on a duration of one week under hydrogen permeation (the rest potential was still -380 mV), and the first hypothesis can be ruled out. However the stability of this passive film asks a question. Actually the Devanathan cell gives only the amount of hydrogen which exits as non-charged particles, and obtaining an oxidation current means that the atomic hydrogen goes out the steel lamella and its oxide layer. Since atomic hydrogen is a powerful reducing agent, there is a contradiction between atomic hydrogen passing through the passive layer and this passive layer being stable for several days under hydrogen permeation. The question "where does the charge transfer take place" [19] is of first importance and the answer can only be: charge transfer occurs at the metal / oxide interface according to the hypothesis that hydrogen diffuses in the passive film as positively charged particles [20].
Input side - Since it has been shown above that the increase in hydrogen permeation is due to a process which occurs on the input side, a thoroughly study of modifications of the oxide layer on the input side, by hydrogen evolution, was carried out. After stabilisation of the exit current at about 2.5 µA, under polarisation of the input side at 10 mA cm-2, i.e., after 12 h under hydrogen permeation (see Fig. 7), perturbations were applied in the input cell. In the following the applied current for hydrogen formation on the input side is referred to J in , and the current resulting for hydrogen oxidation to J out .
* If hydrogen evolution was switched off for 1 or 2 s: by switching on, J out was immediately restored at 2.5 µA.
* If hydrogen evolution was maintained under small cathodic current (0.1 mA cm-2) to prevent oxidation: by switching on J in at 10 mA cm-2, jout was immediately 1 µA and then started again to increase up to the previous permeation current was restored. A short current plateau at about 1 µA was sometimes observed in the transient.
* If hydrogen evolution was switched off for several min.: by switching on J in at 10 mA, J out was 0.5 µA and started again to increase.
From that it appears that three permeation currents were observed, which are roughly: 0.5, 1 and 2.5 µA cm-2, which indicates the oxidation states of the input side of the lamella. When 0.5 µA was detected at the exit side, the passive film on the input side acts as a barrier which hinders H permeation, this barrier was progressively reduced by hydrogen evolution and 1 µA reflects a reduction step. When the permeation current jout was 2.5 µA, the passive layer was reduced as much as possible.
In an attempt to find a correlation between the three permeation currents observed and the oxidation states of the input side of the lamella, the potential of the input side was measured after current perturbations. Under 10 mA cm-2, the potential was -1600 mV. As soon as hydrogen evolution was switched off, the potential reached -1200 mV and from that started increasing to trace a plateau at -800 mV. As soon as the current (10 mA cm-2) was switched on, the potential was again -1600 mV. When the input current was switched off the potential reached -1200 mV and then -800 mV. As soon as the potential reached -800 mV, the input current for hydrogen evolution was switched on, resulting in an output current of 1µA which confirms the results above. The potential -800 mV is due to the formation of a prepassive layer which already protects the iron substrate. If the formation of the passive layer occurs by a multi-step mechanism [17] and therefore its reduction also occurs by several steps, these steps correspond to the potential plateaux determined above (Fig. 4).
Here again two hypotheses can be proposed to explain the increase
in permeation current as the oxide film on the entry side was
reduced. A small diffusion coefficient of hydrogen in an oxide
layer [5] could explain the small current observed during the
first time of the experiments. An other explanation could be related
to the coverage by adsorbed hydrogen. The kinetics of the hydrogen
evolution reaction plays an important role in electro-permeation
as it determines the coverage of the electrode. It is generally
accepted that H evolution occurs in two steps implying the intermediate
(FeH)ad. The coverage by H adsorbed depends
on the rate determining step. If hydrogen adsorption is the rate
determining step qH
ª 0, if hydrogen desorption
(i.e., gaseous hydrogen formation) is the rate determining step
qH ª
1. Since the kinetics depends on the chemical nature of
the substrate, every modifications of the steel oxide will modify
the kinetics of hydrogen evolution and therefore the coverage
by H adsorbed.
Conclusions
Several conclusions can be drawn:
* Steel in NaOH solution was spontaneously passivated and therefore the study of hydrogen permeation through an iron lamella started with a multilayer: oxideout / Fe / oxidein .
* The input oxide film was reduced under hydrogen evolution, and there is a direct relation between the oxidation state and the permeation current.
* The oxide film on the exit side is not reduced by hydrogen permeation
and the passive state is preserved even under hydrogen permeation
for several days. This result indicates that hydrogen oxidation
on the exit state takes place at the metal / oxide interface and
not at the oxide / solution interface.
References
1. M.A.V. Devanathan and Z. Stachurski, Proc. Roy. Soc., A270, 90 (1962)
2. T. Zakcroczymski and z. Szklarska-Smialowska, J. Electrochem. Soc., 132, 2548 (1985)
3. Su-il ryun and R.A. Oriani, Corros. Sci., 29, 485 (1989)
4. Rak-hyun song, Su-il pyun and R.A.. Oriani, J. Electrochem. Soc., 137, 1703 (1990)
5. P. Bruzzoni and R. Garavaglia, Corros. Sci., 33, 1797 (1992)
6. B.S. Chaudhari and T.P. Radhakrishnan, Corros. Sci., 30, 1219 (1990)
7. J. Flis and T. Zakroczymski, Corrosion, 7, 530 (1992)
8. A. Turnbull, M. Saenz de Santa Maria and N.D. Thomas, Corros. Sci., 29, 89 (1989)
9. G. Razzini, L. Peraldo Bicelle and B. Scrosati, Electrochim. Acta, 38, 89 (1993)
10. A. Turnbull, Corros. Sci., 34, 1751 (1993)
11. Z. Szkalarska-Smialowska, T. Zakcroczymski and C.J. Fan, J. Electrochem. Soc., 132, 2543 (1985)
12. T. Zakcroczymski , Scrip. Metall., 19, 521 (1985)
13. S. Haupt and H.H. Strehblow, Corros. Sci., 29, 163 (1989)
14. O.J. Murphy, T.E. Pou, J. O'M. Bockris and L.L. Tongson, J. Electrochem. Soc., 131, 2785 (1984)
15. Rak-hyun Song, Su-il Pyun and R.A. Oriani, Electrochim. Acta, 36, 825 (1991)
16. A. Hugo-le-Goff,J. Flis, N. Boucherit and J. Wilinski, J. Electrochem. Soc., 137, 2684 (1990)
17. M.E. Armacanqui and R.A. Oriani, Corrosion, 44, 696 (1990)
18. R.M. Torresi, O.R. Camara and C.P. de Pauli, Electrochim. Acta, 32, 1357 (1987)
Modifications of passive films, The European Federation of Corrosion, Ed. P. Marcus, B. Baroux and M. Keddam, 155 (1993)
19. P. Bruzzoni and E. Riecke, Corros. Sci., 34, 1597 (1994)
20. R. Ruetschi and G. Giovanoli, J. Electrochem. Soc., 135,
2663 (1988)