Effect of alloying elements on the contact resistance and the passivation behaviour of stainless steels

Effect of alloying elements on the contact resistance and the passivation behaviour of stainless steels

Corrosion Science 44 (2002) 635±655 www.elsevier.com/locate/corsci E€ect of alloying elements on the contact resistance and the passivation behaviou...

1MB Sizes 6 Downloads 64 Views

Corrosion Science 44 (2002) 635±655

www.elsevier.com/locate/corsci

E€ect of alloying elements on the contact resistance and the passivation behaviour of stainless steels J.S. Kim a,*, W.H.A. Peelen a, K. Hemmes a, R.C. Makkus b a

Faculty of Applied Sciences, Delft University of Technology, Rotterdamseweg 137, 2628 AL Delft, The Netherlands b Netherlands Energy Research Foundation (ECN), P.O. Box 1, 1755 ZG Petten, The Netherlands Received 7 January 2000; accepted 15 June 2001

Abstract The e€ects of alloying elements (Cr, Mo) on the passivation and the transpassive transition behaviour of various commercial stainless steels, which are candidate materials for the bipolar plates in polymer electrolyte membrane fuel cell, were investigated via a contact electric resistance (CER) technique with graphite counter part. The contact resistance, too high for a semi-conductor ®lm with high donor density, is not the inherent ohmic resistance of the passive ®lm. The CER potentials decrease linearly with the pitting resistance equivalent number (PREN). The passive ®lms of stainless steels with high PREN are more readily removed compared to those with low PREN in terms of the charge necessary to remove the passive ®lms. The change in transpassive transition behaviour of the stainless steels from that of iron to that of chromium went on steeply with an increase in the PREN. Ó 2002 Published by Elsevier Science Ltd. Keywords: Contact electric resistance; Passive ®lms; Stainless steels; Passivation; Transpassive transition; Pitting resistance equivalent number; Chromium; Molybdenum; Polymer electrolyte membrane fuel cell; Bi-polar plate

1. Introduction The polymer electrolyte membrane fuel cell (PEMFC) is a feasible alternative for the internal combustion engine and also as a power source for portable electronics *

Corresponding author. Fax: +31-15-278-6730. E-mail address: [email protected] (J.S. Kim).

0010-938X/02/$ - see front matter Ó 2002 Published by Elsevier Science Ltd. PII: S 0 0 1 0 - 9 3 8 X ( 0 1 ) 0 0 1 0 7 - X

636

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

such as lap-top computers, camcorders, mobile phones etc. because of its low weight and its lower operating temperature relative to the other fuel cells [1±3]. In developing the PEMFC, cost reduction is a priority. One of the most expensive parts in PEMFC stacks is currently the graphite bi-polar separator plate. In order to reduce the cost, it is necessary to develop low cost separator plates [4±7]. Apart from low price, any separator must have two other important features: high corrosion resistance and good electric conductivity. Stainless steels satisfy these requirements. The high corrosion resistance of stainless steels is primarily due to the passive ®lms formed on their surfaces. Ironically, however, the passive ®lm is also suspected of causing increased ohmic losses in fuel cell stacks. The contact between carbon papers and bi-polar plates of PEMFC are current paths and therefore the increase in contact resistance causes eciency loss. The contact resistance in this case might be attributed to the formation of the passive ®lm. Thus, when selecting a stainless steel bi-polar plate, both the conductivity and the corrosion resistance with respect to the passive ®lm must be considered. Corrosion resistance of stainless steels and the characteristics of the passive ®lm depend strongly on the content of alloying elements such as chromium [8,9], molybdenum [10,11] and nitrogen [12,13]. A newly developed contact electric resistance (CER) technique has been reported to be very convenient and useful in measuring the contact resistance or possibly the DC resistance of the passive ®lm in the solution [14,15]. High temperature passive ®lms of AISI 316 stainless steel, pure Ni and Ni base alloy were studied with this technique by Saario et al. [14]. Piippo et al. [15] investigated the e€ect of applied potential and immersion time on the resistance of the passive ®lm on Cu. For iron, chromium and their alloys, only a few studies [16± 18] so far have been made using the technique. In the study on the passivity of Cr, Fe±12%Cr and Fe±25%Cr alloy [18], they made a brief comment about the e€ect of Cr on the resistance of the passive ®lm: In short, the increase in the resistance is more gradual and occurs at the more positive potential as the Cr content of the materials becomes less. However, extensive studies of the e€ects of Cr on the contact resistance of the passive ®lm, not to mention the e€ects of the other alloying elements of stainless steels such as Mo and N, have not been made. In this study, we investigate the e€ect of the alloying elements on the passivation behaviour of stainless steels via the DC contact resistance of their passive ®lms and polarization behaviours. The ultimate purpose of this study is to o€er fundamental information on the passivation behaviour of candidate bi-polar separator plate materials for PEMFC, i.e. stainless steels, in the simulated environments. 2. Experimental procedure 2.1. Principle of contact electric resistance technique The CER measurement technique was developed by Saario et al. [14]. The principle of the CER technique is very straightforward as detailed in Fig. 1. Two specimen surfaces are brought into contact and then separated repeatedly with a

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

637

Fig. 1. Principle of CER measurement and the schematic experimental setup.

chosen frequency. When they are disconnected and exposed to the environment, the passive ®lm is formed and grown on the surfaces. When the two surfaces are in contact and a direct current is passed through the specimens, the DC electric resistance of the system can be determined according to Ohm's law by measuring the voltage drops. Because of high conductivity of bulk metals, the voltage drop occurs mainly through the passive ®lms or possibly through the interface of the metal and the passive ®lm. Fig. 1 shows the method to measure the electric contact resistance (Rc ) under the potentiostatic condition. We use a shunt resistance (R2 ) and a controlled DC applied current (ia ). Without the passive ®lm, Rc is equal to the resistance of bulk metal (Rm ). Once the passive ®lm forms, Rc is the sum of the resistance the passive ®lm (Rf ) and the resistance of bulk metal (Rm ). Electric current in contact (ic ) is

638

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

ic ˆ

R2 ia R2 ‡ Rc

…1†

Then the potential drop across the ®lm in contact state (Vc ), measured via a voltmeter, is expressed as follows: Vc ˆ ia

R 2 Rc R 2 ‡ Rc

…2†

In this study we used a 10 X shunt resistance and applied a 2 mA DC current for the measurement; accordingly the voltage di€erence between two samples is at most 20 mV. Thus, the charge transfer, which includes charge transfer from the oxide to solution, ionic conduction through the solution and charge transfer from solution to the oxide, is extremely dicult and the electrical resistance between two electrodes through the electrolyte (Re ) is very high. This means that electrical resistance in the separated situation (Rs ), which is the sum of Re and Rm , is much higher than R2 . In consequence, the potential in the separated state (Vs ) equals: Vs ˆ ia

R 2 Rs  i a R2 R2 ‡ Rs

…3†

From Eqs. (2) and (3) Vc …R2 ‡ Rc † Vs ˆ R2 Rc R2

…4†

Consequently, the contact resistance can be calculated from the contact voltage (Vc ), the separation voltage (Vs ) and the shunt resistance (R2 ). Rc ˆ

Vc Vs

Vc

R2

…5†

Once the passive ®lm forms, Rc is almost equal to the resistance of the passive ®lm (Rf ) because the electric resistance of the metal, i.e. stainless steel in this study, is much smaller than that of the passive ®lm. 2.2. Contact electric resistance measurement setup Before starting the CER measurement, we need to determine a reference position called ``home position'' in order to apply a controlled displacement to the specimen. The position of the upper sample of the pair is controlled by the step motor located on the upper part of the machine. As the motor gives downward displacement to the sample via a spring, the two samples become closer. When the samples come into contact with each other, the CER value being recorded drops abruptly. We normally choose the position where the CER value starts to drop as home position. Once the CER value drops and reaches a low plateau, it remains nearly the same even with further downward force by the step motor. The spring between the step motor and the sample makes it possible to control the mechanical pressure at the interface of the specimens and therefore keeps the passive ®lms from collapsing.

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

639

Fig. 2. CER values of a 304 LN sample as a function of a displacement from the home position, in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

Fig. 2 is showing the CER response of a 304 LN stainless steel sample with the relative displacement from the home position. To measure the CER, we applied ‡20,000 step displacement (3.26 lm) versus the home for the separated position and 10,000 steps for contacted position. The displacement here represents the value in terms of the step motor. Negative displacement cannot be made and it means that the two samples are already in contact and under pressure. In this condition, the force directed to the contact surfaces would be 4.47 N, resulting in a nominal contact pressure of 1.42 MPa for a typical contact area diameter of 2.0 mm. Comparing this value with a typical value for the compressive strength of oxides (>1000 MPa), it is clear that the surface ®lm will not be crushed in a CER measurement. Once the specimen starts moving, the home position has no more meaning, but only the contact and the separate positions are important. Positions are referred to as the position in terms of the step motor. Though it is dependent on the passive ®lm examined, the distance between the contact and the separate position is normally within a few lm range. In order to detect the variation of the CER as a function of the variables such as applied displacement, applied potential and exposure time to environment, the CER measurement needs to be done repeatedly with a speci®c frequency. The contact time and separation time can also be controlled. In determining those times, the stability of the data being recorded must be checked. It is also necessary to consider how fast the ®lm grows, which is determined by a material composition, the environment, the applied potential, the potential scan rate etc. 2.3. Materials and electrolyte In order to investigate the e€ect of the major alloying elements like Cr, Mo and N, 11 stainless steels were chosen according to their pitting resistance equivalent number (PREN) [19,20] values and a graphite sample was used for comparison. The stainless

640

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

steels chosen encompass austenic, ferritic and duplex stainless steels. The PREN, represents the degree of pitting resistance of stainless steels and is de®ned as: PREN ˆ %Cr ‡ 3:3%Mo ‡ k%N

…6†

k is reported to be 16 or 30, and here we choose 16 [19,20]. Though this formula is normally applied for austenitic stainless steels or duplex stainless steels, it was applied for the ferritic stainless steels in this study. The compositions of the selected materials were listed in Table 1, and the PREN values of the materials are represented in Fig. 3. The PREN values of the chosen materials are between 18 and 45. The specimen speci®cation for the CER measurement is delineated in Fig. 4. All the stainless steel samples were cut from hot rolled plates, and the graphite from a bar. The contact surface was cut and machined to be parallel to the rolled surface. All the stainless specimens for the CER measurements were plates more than 10 mm thick and a graphite specimen from a 5-mm diameter round bar. In order to simulate a typical PEMFC environment, we carried out the CER measurements in pH 4.8 aqueous solution with sulphuric acid (H2 SO4 ) and sodium sulphate (Na2 SO4 ) at room temperature. The solution was prepared by mixing 10 4 M H2 SO4 with 0.5 M Na2 SO4 . 2.4. Contact electric resistance measurement procedure The CER was measured as a function of applied potential being increased from 1500 mV (SCE) to above the apparent transpassive transition potential at a rate of 0.5 mV/s. A standard calomel electrode (SCE) and a platinum electrode were used as respectively a reference and a counter electrode. All potentials in this paper are represented versus SCE. In order to remove the air-formed oxide and clean the surface of the specimen, we applied 1500 mV for 10 min. The contact position was set at the point 0.815 lm (5000 steps by the motor) apart downward, and a separate Table 1 Chemical compositions (wt.%) of the stainless steels (Fe balanced) Alloy/UNS number

Cr

Mo

N

Ni

Mn

Si

C

Others

304L/S30403 304LN/S30453 316L/S31603 316LN/S31653 317L/S31703 E-brite/S44627

18.24 19.8 17.33 16.52 18.09 26.3

± 0.4 2.09 2.08 3.07 0.99

0.04 0.124 0.06 0.119 0.072 0.0128

10.17 8.8 10.62 10.15 13.68 0.14

0.58 0.45 0.43 0.56 0.43 0.23

21.97 20.9 28.9

2.99 4.44 3.83

0.16 0.035 0.0076

5.85 24.45 2.42

0.59 0.49 0.11

0.021 0.034 0.024 0.03 0.024 0.01 (max) 0.024 0.024 0.02

± Cu, Cu, Cu, Cu, Cu,

SAF 2205/S31803 904L/N08904 AL29-4-2/S44800 SAF 2507/S32750

25.08

3.82

0.3

6.86

1.15 1.8 1.61 1.77 1.86 0.1 (max) 1.49 1.67 0.1 (max) 0.81

0.21

0.022

AL-6XN/S08367

20.83

6.22

0.23

23.75

0.33

0.024

0.23

Co Co Co Co Ni

Cu, Co Cu, Co Cu Cu, Co, Nb, Al, Sn, Ti Cu

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

641

Fig. 3. The PREN values of the stainless steels investigated.

Fig. 4. Speci®cation of the CER specimen used.

position at the point 3.26 lm apart upward from the home position. Samples were hold for 5 s at each position. The measurements involved not only self-contact measurements for 11 stainless steels and graphite but also stainless steel versus graphite contact measurements. Self-contact means the contact between the same materials, for example 304 L versus 304 L or a graphite versus a graphite. In this paper, the analyses are performed for the stainless steel/graphite contact because of the reliability of measurements. The details are to be explained below. In the PEMFC, bi-polar plates are often in contact with carbon paper. To simulate this situation, we chose graphite as a material to measure the resistance of the passive ®lm and measured the contact resistance between stainless steels and graphite.

3. Results and discussion Fig. 5 shows the contact resistance responses of 304 LN and 316 LN stainless steels, measured by the self-contacted method as a function of applied potential in

642

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

Fig. 5. CER responses, which was measured via self-contact, and current responses with applied potential for (a) 304 LN (b) 316 LN stainless steels in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

pH 4.8 acid solution at room temperature. For convenience, we used the geometric contact area (0.2 cm2 ) when calculating the CER values. The current response is also represented in each graph. The contact resistance of 304 LN starts to increase at about 1.2 V and reaches a plateau value from about 0.3 V until it plunges at about 1.15 V. The measured corrosion potential was about 0.24 V. 316 LN shows similar behaviour to that of 304 LN in its contact resistance response except for di€erences in the potential values where the resistance reaches plateau and drops. In spite of clear plots in Fig. 5, the results from the self-contact measurements were considerably scattered. Fig. 6 shows the CER responses for 304 L stainless steel recorded under the same conditions. Such a scattering in the measurements might be associated with instability or partial breakdown of the passive ®lm caused by the consecutive contact-separation processes with the other stainless sample. The contact with graphite could provide a more stable environment to the passive ®lms on stainless steels compared with self-contact because graphite is much softer than stainless steels.

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

643

Fig. 6. CER responses of 304 L stainless steels, obtained from the repeated measurements via self-contact, in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

Fig. 7. CER, which was measured via self-contact, and current responses versus applied potential for graphite in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

Fig. 7 shows contact resistance of graphite as a function of applied potential, which was measured for the self-contact of graphite. The contact resistance drops near the open circuit potential and maintained stable until it increases again near the transpassive transition potential. Comparing the anodic polarization responses, the changes in the CER response are likely due to evolution of gases: hydrogen around the corrosion potential and oxygen at the transpassive transition potential. The contact resistance at the stability region was a few X cm2 . Fig. 8 shows the contact resistance responses of 304 LN and 316 LN stainless steels measured being in contact with graphite. The behaviour is basically similar to the self-contact measurements in Fig. 5. However, in these heterogeneous combinations, the data were less scattered.

644

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

Fig. 8. CER, which was measured via graphite contact, and current responses versus applied potential for (a) 304 LN (b) 316 LN stainless steels in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

Contact resistance responses and current responses with applied potential for 11 di€erent stainless steels are delineated in Fig. 9. For the graphite±stainless steel contact measurements, however, it is still impossible to compare absolute CER values of the passive ®lms among the di€erent stainless steels because the values at the passive potential range di€ers up to more than 10 times even for the same sample. This is con®rmed from Fig. 10 which shows the CER values of 11 samples which were measured potentiostatically under 0.3 V. The 11 stainless steels are represented on the x-axis in terms of the PREN. The CER values look independent of the composition, i.e., the PREN, and the order of the magnitudes are also quite di€erent from the order that can be made from the Fig. 9. Though the CER measurement has a limitation in terms of absolute values, it is quite applicable to the analysis in terms of the potential values where changes in the CER values took place. Fig. 11 shows the typical CER responses of 304 LN stainless steel on linear and log scale as a function of the applied potential. The contact resistance starts to increase at the potential less than 1.2 V as shown in log plot. Then, the contact resistance increases in two stages; in the ®rst stage the resistance increases slowly and at

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

645

Fig. 9. CER and current responses as a function of applied potential for 11 di€erent stainless steels in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

second stage it increases very steeply. According to the Pourbaix diagram of Cr [21], 1.2 V is very near the potential where Cr(OH)3 hydroxide forms on chromium surface in water at 25°C. Thermodynamically the Cr2 O3 can be formed on Cr at a

646

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

Fig. 10. CER values of 11 stainless steels measured under potentiostatically at 0.3 V in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

Fig. 11. CER responses of 304 LN stainless steel in pH 4.8 sulphuric acid ‡ sodium sulphate in linear as well as log scale.

little higher potential than 1.2 V, but recent studies [22,23] suggests the Cr2 O3 will not be formed especially at low potentials. Then, as the ®lm grows the resistance increases slowly until it starts to increase abruptly at about 0.5 V. As is well known, the formation of the passive ®lm on the stainless steel is accompanied by dominant dissolution of iron. This phenomenon also can be easily understood from the Pourbaix diagram of Fe in aqueous solution [24]. The oxide formation potential of Fe, which is around 0.3 to 0 V (SCE) depending on the concentration of Fe2‡ ions, is higher than that of Cr. Rapid iron dissolution might well trigger the change in composition, structure and thickness of the passive ®lm on stainless steels thereby a€ecting the CER values. CER starts to increase steeply from about 0.5 V and reaches a plateau around 0.1 V. Subsequent plummet of the CER value is followed at the potential around 1.5 V.

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

647

There are some noticeable features in the CER responses that indicate that the CER might not be just ohmic resistance of the passive ®lm, which is known as a semi-conductor [25,26]. The CER values in the passive state, about 100 X cm2 or more, are not realistic. These values, which convert to be at least about 5  108 X cm under the assumption that the thickness of the ®lm is 2 nm, are too high for a resistance value of the semi-conducting ®lm with a lot of intrinsic defects [25,26]. According to Castro and Vilche [25], the donor density of an Fe±12%Cr alloy is 1:2  1022 cm 3 . Jacobs [26] also reported that donor density values of Fe±20%Cr alloy are 2  1022 ±6  1022 cm 3 and the calculated resistivities are as low as 2:5  10 4 ±3:9  10 4 X cm within the passive region. Such a discrepancy between the CER values and previous reported resistivities of the passive ®lm might be explained in terms of Schottky barriers which would be formed at the interfaces of metal/ passive ®lm or passive ®lm/graphite. Accordingly, the resistance measured this study should be considered literally as the ``contact resistance'' of the system, including the stainless steel and the passive ®lm, rather than ohmic resistance of the passive ®lm itself. In other words, the CER values measured in this study supposedly represent the nature of the contact accompanied by the formation of the passive ®lm rather an inherent property of the passive ®lm. This is discussed below. In order to compare the CER as a function of composition, we compare the CER by measuring the potentials where the contact resistance reached 10 X cm2 . The potentials are plotted as a function of Cr content and the PREN values of stainless steels in Fig. 12. Corrosion potentials are also plotted in both the graphs. There is no conspicuous di€erence in the potential values with a change of Cr content or the PREN. Though it was reported that as the Cr content of stainless steels increases, the CER increases [18], such a trend could not be con®rmed in this study. Transpassive ®lm dissolution process of stainless steel includes dissolution of the oxide/hydroxide of chromium and subsequently that of iron. It has been reported that the transpassive dissolution of Cr proceeds via the oxidation process from Cr(III) to Cr(VI) ions in the passive ®lm [27], which causes an increase in the anodic current on the polarization curve. Therefore, in the polarization responses in Fig. 9, the transpassivation dissolution of Cr is identi®ed by the current peaks around 0.7 V in most occasions. The contact resistances of the stainless steels remain high without a change even at potentials well above the apparent transpassive transition potential. As shown in Fig. 9, the values begin to decrease only above the potential where the whole ®lm dissolves into the solution. In Fig. 13, the variation of the potential where the CER drops to half the plateau value is delineated as a function of Cr content and the PREN. The PREN values involve the e€ect of Mo and N as well as Cr. The two graphs show similar behaviour except that Fig. 13(a) has one or two deviated points. Fig. 13(b), without the deviation of Fig. 13(a), shows that the potential where the ®lm is removed clearly decreases as the PREN increases. Transpassive transition potentials of the stainless steels, evaluated from the polarization responses and delineated in Fig. 14, decrease with the alloying elements in a similar way. The two deviated points in Figs. 13(a) and 14(a) represent 904 L and AL6XL stainless steels, both of which contain a relatively small amount of chromium but much molybdenum. Figs. 13 and 14 show

648

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

Fig. 12. The potential where contact resistance gets larger than 10 X cm2 as a function of (a) Cr content and (b) the PREN, measured in pH 4.8 sulphuric acid ‡ sodium sulphate solution in room temperature. Corrosion potentials were also plotted.

that molybdenum as well as chromium shifts the transpassive transition region to lower potentials. Furthermore, considering the factor 3.3 before molybdenum in the de®nition of the PREN, it could be said that the in¯uence of molybdenum should be greater than chromium in determining the transpassive transition behaviour. Fig. 15 shows a linear relation between the transpassive transition potentials and the CER decrease potentials, which were evaluated respectively from the polarization and CER responses. The slope of the plot is <1. For the stainless steels with high PREN, the CER plummet potentials are lower than transpassive transition potentials and vice versa. This means that as the PREN increases, actual removal of the protective ®lm of the stainless steel, identi®ed from the drop of the CER value, becomes faster and hence occurs before the apparent increase of the transpassive

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

649

Fig. 13. CER plummet potential as a function of (a) Cr content and (b) the PREN of the stainless steels, in pH 4.8 sulphuric acid ‡ sodium sulphate solution in room temperature.

transition current on the polarization response. Schmuki et al. [28] suggest that the formation of Cr(VI) in the passive ®lm on pure Cr begins at a much lower potential than the apparent transpassive transition potential and the dissolution of the passive ®lm occurs when a sucient fraction of the outermost layer of the ®lm has been converted to Cr(VI). They also suppose that the oxidation of Cr from Cr2 O3 to Cr(VI)film is considerably faster than the subsequent dissolution of Cr(VI)film to Cr(VI)aq: It is questionable as to why the passive ®lms on the stainless steel with high PREN values are removed faster than those with low PREN values. Fig. 16 shows the charges, each of which was integrated from 0.75 V to the CER decrease potential on the polarization. 0.75 V is the potential where the transpassive transition begins on pure Cr, which was estimated from the polarization response of pure Cr. Practically,

650

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

Fig. 14. Transpassive potential as a function of (a) Cr content and (b) the PREN of the stainless steels, in pH 4.8 sulphuric acid ‡ sodium sulphate solution in room temperature.

whatever potential we choose around this potential makes little di€erence because the passive current is much smaller than the transpassive dissolution current. As was noticed above, the CER decrease potential represents the value where the protective ®lm is lost. The examples with excessively high or low passive current are excluded from the comparison. Consequently, included are 316 L, 317 L, E-brite, SAF 2205, 904L, AL 29-4 and SAF 2507. The passive current values of the stainless steels are within 3  10 4 to 4:5  10 4 A/cm2 . It is showing that as the PREN increases, the charges decrease. Though all the charges are not from the ®lm dissolution, it might show that the passive ®lms of stainless steels with high PREN are eciently removed in terms of charge ¯ow compared to those of stainless steels with low PREN. The PREN of the stainless steel increases, the critical amount of Cr(VI), and possibly Mo(VI) necessary to trigger the ®lm dissolution readily attainable. Once

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

651

Fig. 15. CER plummet potential versus transpassive potential in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature.

Fig. 16. Charges integrated from 0.75 V to the CER plummet potential on each polarization response of the stainless steels.

dissolution starts, the passive ®lms on stainless steels with high PREN are removed eciently, i.e. by relatively smaller amount of charge ¯ow compared to those on stainless steels low PREN. Pure iron and chromium are included in Fig. 17, which shows that the transpassive transition potentials of the stainless steels become closer to that of chromium as the PREN increases and to that of iron as the PREN decreases. To make a comparison, we assume that the PREN value of chromium (1 0 0) and that of iron (0) are also valid, though the PREN equation (Eq. (6)) is generally applicable only to

652

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

stainless steels. What is remarkable is, as noted above, that it is not the Cr content, but rather the PREN that is related to the decrease of the transpassive transition potential. Fig. 17 indicates that the properties of the passive ®lm, most probably the composition, changes to that of chromium from that of iron as the PREN value increases. This also shows that molybdenum and chromium are incorporated into the passive ®lm and a€ect the transpassive transition behaviour of the stainless steels. The decrease in transpassive transition potentials is quite steep and it occurs mainly within the narrow range of the PREN not more than 35. This behaviour agrees with the result of Asami et al. [29]. They measured the composition of the passive ®lm on Fe±Cr alloys in 1 M H2 SO4 deaerated solution by XPS, and concluded that the main constituent of the passive ®lm on Fe±Cr alloys changes suddenly from iron oxyhydroxide to chromium oxy-hydroxide when the Cr content of the alloys exceeds 13 at.% (12.2 wt.%). Their results are plotted in Fig. 18. We recalculated the Cr contents of the alloys from atomic percentage to weight percentage and represented them in terms of the PREN. Returning to PEMFC application as bipolar plates it is clear that the values, which are about 100 X cm2 , are much higher than the contact resistance of a real PEMFC system. In a real PEMFC system, the contact resistance between the backing carbon paper and the bi-polar plates were measured to be less than 0.1 X cm2 . Furthermore, as was noticed above, the measured CER values here are too high for the passive ®lms with high defect density, whose resistance was previously evaluated to be 10 10 X cm2 [26] from their defect concentration. Accordingly, the conduction process between the graphite and the stainless steel in this study cannot be an ohmic process. Selecting stainless steels for the separator plates, the resistance of the passive ®lm itself might not be the matter of concern.

Fig. 17. Transpassive potential as a function of (a) Cr content and (b) the PREN of the stainless steels. Pure chromium (PREN ˆ 100) and pure iron (PREN ˆ 0) are also included.

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

653

Fig. 18. Compositional analysis of the passive ®lm on Fe±Cr alloys by Asami et al. [29], plotted as a function of the PREN.

4. Conclusions From the CER measurements in pH 4.8 sulphuric acid ‡ sodium sulphate solution at room temperature for 11 di€erent commercial stainless steels covering austenitic, ferritic and duplex stainless steels, which are candidate materials for bi-polar plates of PEMFC and cautiously selected in terms of their PREN values, we conclude as follows: 1. The CER measurement via the graphite/stainless steel contact provides more reliable experimental results compared to the self-contact measurement of the stainless steel. 2. The CER decrease potentials and the transpassivation potential linearly decreases as the PREN increases, which shows that not only Cr but also Mo plays an important role in transpassivation dissolution behaviour of the passive ®lm on stainless steels. 3. It was con®rmed that the transpassive transition of stainless steels changes to be closer to that of chromium from that of iron as their PREN values increase and the change occurs steeply within a narrow PREN range of about 35. 4. Once the dissolution starts, the passive ®lms on stainless steels with high PREN are removed by relatively smaller amount of charge ¯ow compared with those with low PREN. 5. The contact resistance values of the stainless steels investigated were excessively high for a semi-conductor ®lm with high donor density. Therefore, the conduction process between graphite and the stainless steel in this study cannot be explained in terms of ohmic contact but rather perhaps in terms of the Schottky contact [30].

654

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

Acknowledgements The authors acknowledge the instrumental support of CORMET testing systems. We also wish to appreciate the helpful advice from Dr. Piippo and Mr. Ihonen for the experimental technique and the data analyses. References [1] S. Hirano, J. Kim, S. Srinivasan, Electrochimica Acta 42 (1997) 1587. [2] J. Kim, S.M. Lee, S. Srinivasan, C.E. Chamberlin, J. Electrochem. Soc. 142 (1995) 2670. [3] A. Heinzel, M. Zedda, R. Nolte, T. Meyer, H. Schmidt, Alternative fuel cell con®gurations as battery replacement, Proceedings of 1998 Fuel Cell Seminar, Palm Springs, California, 16±19 November, 1998, p. 686. [4] C.E. Reid, W.R. Merida, G. McLean, Results and analysis of a PEMFC stack using metallic bi-polar plates, Proceedings of 1998 Fuel Cell Seminar, Palm Springs, California, 16±19 November, 1998, p. 603. [5] C. Zawodzinski, M.S. Wilson, S. Gottesfeld, Corrosion testing of stainless steel fuel cell hardware, Proceedings of 1998 Fuel Cell Seminar, Palm Springs, California, 16±19 November, 1998, p. 647. [6] K. Ledje€-Hey, F. Mahlendorf, O. Niemzig, A. Trautmann, Electronically conducting composite materials as bipolar plates for PEM fuel cells, Proceedings of 1998 Fuel Cell Seminar, Palm Springs, California, 16±19 November, 1998, p. 570. [7] D.N. Busick, M.S. Wilson, Low cost composite materials for PEFC bi-polar plates, Proceedings of 1998 Fuel Cell Seminar, Palm Springs, California, 16±19 November, 1998, p. 632. [8] K. Osozawa, H.J. Engell, Corros. Sci. 6 (1966) 389. [9] J. Horbath, H.H. Uhlig, J. Electrochem. Soc. 115 (1968) 791. [10] K. Sugimoto, Y. Sawada, Corros. Sci. 17 (1977) 425. [11] M. Sakashita, N. Sato, in: R.P. Frankenthal, J. Kruger (Eds.), Passivity of Metals, The Electrochemical Society, Princeton, NJ, 1978, p. 479. [12] J.J. Eckenrod, C.W. Kovach, in: Properties of Austenitic Stainless Steels and Their Weld Metals, ASTM STP 679, ASTM, Philadelphia, Pennsylvania, 1979, p. 17. [13] J.E. Truman, M.J. Coleman, K.R. Pirt, Br. Corros. J. 12 (1977) 236. [14] T. Saario, T. Laitinen, J. Piippo, Mater. Sci. Forum 289±292 (1998) 193. [15] J. Piippo, T. Saario, T. Laitinen, M. Bojinov, J. Hinttala, Mater. Sci. Forum 289±292 (1998) 429. [16] M. Bojinov, I. Betova, R. Raiche€, G. Fabricius, T. Laitinen, T. Saario, Passivation mechanism of iron in concentrated phosphoric acid, Proceedings of EUROCORR'98, Utrecht, The Netherlands, Septermber 28±October 1, 1998. [17] M. Bojinov, G. Fabricius, J. Ihonen, T. Laitinen, T. Saario, The e€ect of temperature on the passive and transpassive behaviour of chromium in neutral aqueous solutions, Proceedings of EUROCORR '98, Utrecht, The Netherlands, September 28±October 1, 1998. [18] M. Bojinov, G. Fabricius, T. Laitinen, K. Makela, T. Saario, G. Sundholm, Conduction mechanism of the anodic ®lm on Cr and Fe±Cr alloys in H2 SO4 solution, Proceedings of EUROCORR '98, Utrecht, The Netherlands, September 28±October 1, 1998. [19] J. Charles, Structure and mechanical properties of duplex stainless steels, 4th International Conferences on Duplex Stainless Steels, paper KI, Glasgow, Scotland, TWI, 1994. [20] J.S. Kim, H.S. Kwon, Corrosion 55 (1999) 512. [21] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, NACE, Houston, Texas, 1974, p. 262 (English edition). [22] C. Sunseri, S. Piazza, F. Di Quarto, J. Electrochem. Soc. 137 (1990) 2411. [23] V. Maurice, W.P. Yang, P. Marcus, J. Electrochem. Soc. 141 (1994) 1352. [24] M. Pourbaix, in: Atlas of Electrochemical Equilibria in Aqueous Solutions, NACE, Houston, Texas, 1974, p. 312 (English edition).

J.S. Kim et al. / Corrosion Science 44 (2002) 635±655

655

[25] E.B. Castro, J.R. Vilche, Electrochim. Acta 38 (1993) 1567. [26] Luc Jacobs, Characterisation of the passive ®lm on iron-chromium alloys by elipsometry and electrochemical impedance spectroscopy, Ph.D. Thesis, Technische Universiteit Delft, The Netherlands, 1995. [27] S. Haupt, H.H. Strehblow, Corros. Sci. 37 (1995) 43. [28] P. Schmuki, S. Virtanen, A.J. Davenport, C.M. Virus, J. Electrochem. Soc. 143 (1996) 3997. [29] K. Asami, K. Hashimoto, S. Shimodaira, Corros. Sci. 18 (1978) 151. [30] M. Lazzarino, T. Ozzello, G. Bratina, L. Vanzetti, J.J. Paggel, L. Sorba, A. Franciosi, J. Cryst. Growth 159 (1996) 718.