The effect of the Cl− ion on the passive film on anodically polarized 304 stainless steel

The effect of the Cl− ion on the passive film on anodically polarized 304 stainless steel

Corrosion Science. Vol. 20, pp. 313 to 329 Pergamon Press Ltd. 1980. Printed in Great Britain. T H E E F F E C T O F T H E C1- I O N O N T H E P A S ...

967KB Sizes 7 Downloads 82 Views

Corrosion Science. Vol. 20, pp. 313 to 329 Pergamon Press Ltd. 1980. Printed in Great Britain.

T H E E F F E C T O F T H E C1- I O N O N T H E P A S S I V E FILM ON ANODICALLY P O L A R I Z E D 304 STAINLESS STEEL* M. E. CURLEY-FIORINOand G. M. SCHMID Department of Chemistry, University of Florida, Gainesville, FL 32611, U.S.A. Abstract--The differential capacity-potential behaviour of A1SI 304 was determined during anodic polarization in de-aerated KCI/NasSO4 solutions at pH 2.4 using the single pulse technique. The behaviour characterizing potentiostatic and galvanostatic polarization in non-pitting systems is paralleled during galvanostatic polarization in pitting solutions up to the maximum potential attained prior to breakdown. Characteristic potentiostatic potentials also agree well with galvanostatic potential arrests. The potentiostatic capacity-potential behaviour of the passive region is not influenced by pitting. No capacity peak is associated with the onset of pitting and similar film thicknesses are calculated in both pitting and non-pitting environments. INTRODUCTION

THE CORROSIONresistance of austenitic stainless steels is a result of the high degree of passive state stability conferred by the presence of chromium in amounts > 12~.1, z However, in some media, especially in halide solutions, this corrosion resistance is lost, as evidenced by the onset of intense local attack, i.e. pitting. The mechanism by which pits nucleate in the presence of chloride ion has been investigated in great detail. Excellent reviews are given by Kolotyrkin 3 and Szklarska-Smialowska. 4 Associated with the pitting phenomenon is an induction time, the time, at a given potential, which passes prior to breakdown of the passive film. The influence of the induction time is seen both in galvanostatic and potentiostatic polarization measurements. Previous work on the galvanostatic polarization of austenitic stainless steels in chloride-containing solutions has shown that systems which are not susceptible to pitting corrosion reach and maintain a positive steady state potential under the influence of an anodic current. In systems subject to pitting, the maximum potential attained is unstable. After some time, which is a function of current density and solution composition, a shift in potential to more active values is observed and pitting ensues. Similar results have been reported by several investigators. ~-9 During potentiostatic polarization, changing the potential to values just positive to the pitting potential at first results in a decrease in current density with time from an initially high value associated with double layer charging. 6 This current decrease represents the readjustment of the electrode-solution interface to maintain the passive condition. After a time interval, which depends on solution composition and potential, the current decrease is replaced by current oscillation signifying the pitting-induced breakdown of passivity. As the potential is made still more positive, the time for which passivity is maintained decreases until an induction period is no longer apparent. 1° The presence of this induction time associated with pitting suggests that a potential*Manuscript received 30 January 1978; in amended forms 2 April 1978 and 12 February 1979. 313

314

M.E. CURLEY-FIORINOand G. M. SCnMID

dependent, time-consuming change in an initial surface structure is occurring. One o f the mechanisms suggested to explain the onset o f pitting relies on the specific adsorption o f chloride ion. 11 I f chloride ion is indeed involved in the change in surface structure, its incorporation into the electrical double layer should be reflected in the differential capacity vs. potential behaviour o f the interface. A capacity m a x i m u m should occur prior to pitting. In addition, if the nature o f the initial, passivating film is the same in both pitting and non-pitting environments, the capacity-potential relationships observed in the passive region in both types o f solution should be similar, at least in the potential range in which the passive state is stable. It is also expected that the characteristic capacity and potential values determined during potentiostatic and galvanostatic polarization in non-pitting systems should be similar to those found during galvanostatic polarization in aggressive media prior to breakdown from a m a x i m u m potential which m a y lie in the passive, transpassive, or oxygen evolution region. The primary objective o f the experiments conducted in this study was, therefore, the evaluation o f the effect o f chloride ion on the nature o f the passive film formed on stainless steel during anodic polarization. This was approached through the determination o f the capacity-potential behaviour observed during potentiostatic and galvanostatic polarization in both pitting and non-pitting media. Differential capacity measurements have been successfully applied in the study o f stainless steels in both pitting and non-pitting media3 T M The nature o f the passive filmlS, 16-2° and anion adsorption ~-23 on iron have a l o been investigated. In addition, the characteristic potential peaks and potential plateaus defined during potentiostatic and galvanostatic polarization, respectively, were compared. EXPERIMENTAL METHOD The material investigated was stainless steel, AISI 304, provided by the United States Steel Corporation. Its composition was given as 0.03 C, 0.027 P, 1.10 Mn, 0.022 S, 0.43 Si, 9.26 Ni, 18.6 Cr, 0.39 Mo and 0.04 N (wt. ~). Bar stock was machined to cylinders with a diameter of 6 mm and a height of 9 mm. The cylinders were tapped, threaded, and mechanically polished at 2400 rpm with 400 followed by 600 grit emery paper. They were then degreased with spectral grade benzene in an ultrasonic cleaner, rinsed with triply distilled water, and stored in a closed polyethylene container until needed. Primary studies were carried out in solutions containing 0, 1.17 x 10-I, 9.97 × 10-I, 0.301, 0.508, and 1 M potassium chloride. Solution pH was adjusted to 2.4 with concentrated sulphuric acid. Sodium sulphate was added as required to maintain an ionic strength of one (0.318, 0.313, 0.284, 0.232, 0.147 and 0 M, respectively). Secondary experiments involved solutions of pH 2.4 containing 0.102M KCI with no NaISO4 added and 0.3M KC1 at pH 1.52. All chemicals used in solution preparation were of reagent grade. The water employed was distilled from alkaline potassium permanganate and then from a twostage Heraeus quartz still and collected in a two-litre Pyrex volumetric flask. Its maximum conductivity was 2 × 10-6 ~-~ cm-L The electrochemical cell was made of Pyrex and was of conventional design. A Luggin capillary connected the saturated calomel reference electrode (SCE) to the cell via two solution-lubricated mercury-seal stopcocks and a potassium chloride salt bridge. A 1 cms platinum flag auxiliary electrode was mounted in the cell in a separate compartment with a standard taper joint. For use in constant current polarization and capacitance measurements, a platinum gauze basket, approximately 100 emt in area (Engelhard Industries), was mounted concentric to the test electrode. The cell cap incorporated a 24/40 standard taper joint for holding the test electrode assembly which consisted of a stainless steel cylinder on a standard Stern-Makrides mount machined from KeI-F. The electrode area exposed to solution was approximately 2 cm2. Solutions were de-aerated with helium (99.99 ~) for a minimum of 8 h prior to use. Prepurification

Effect of the CI- ion on the passive film

315

and water saturation of the gas were accomplished by passing it through a 12 cm column of Linde 5 ,~ molecular sieve pellets into a gas wash bottle containing triply distilled water. The gas then flowed into a two-litre reservoir containing the solution and continued through a dispersion tube into the electrochemical cell. The cell gas outlet terminated in a gas wash bottle to prevent contamination of the cell contents with ambient air. Immediately before each experiment, the cell was washed with hot chromo-sulphuric acid cleaning solution and rinsed with triply distilled water. Gas pressure was then used to fill the cell with deaerated solution from the reservoir. The stainless steel sample was secured on the KeI-F holder, rinsed with triply distilled water and the solution to be studied, and immersed in solution. All samples were pretreated at - 0.700 V for 20 rain to reduce air-formed surface films. Solution stirring was accomplished with a magnetic stirring bar and the helium flow continued throughout the experiment. All solutions were at room temperature and all potentials are reported relative to the saturated calomel electrode. Current density and differential capacity were calculated using electrode geometric areas. The test electrode was grounded during all experiments.

Potentiostatic polarization Polarization was accomplished with a modified Harrar 24 potentiostat. After prepolarization at -- 0.700 V the potential of the test electrode was shifted anodically in increments. In regions where changes in potential caused significant changes in current density, steps of 20-30 mV were generally employed. In regions of approximately constant current, 50 mV steps were used. After an (arbitrary) time interval of 10 min, the current flowing in the auxiliary-test electrode circuit was determined. The differential capacity of the stainless steel-solution interface was determined as a function of potential using the single pulse method which excludes pseudo-capacitative interferences from the measurement. 25 A 100 ~ts, 5 mA pulse produced by triggering a Tektronix 114 pulse generator was applied to the platinum basket-test electrode circuit. The resulting potential-time transient was recorded on a Tektronix 549 oscilloscope operated in the storage mode at a sensitivity of 2 or 5 mV cm -1 and 2 or 5 izs cm -1. Differential capacity values were calculated from the linear segment of the transient slope during the initial 10 ~.s of the pulse. Since the transient was linear between 4 and 10 tzs, the slope determined between these two points was equivalent to that of a tangent drawn to the curve at t = 4 Vs. The maximum change in potential observed in this time span was approximately 2 mV and the differential capacity was assumed constant over this range. Prior to each experiment the pulse magnitude was calibrated using standard capacitors and resistors.

Galvanostatic polarization After potentiostatic prepolarization at -- 0.700 V for 20 rain the system was switched to galvanostatic control using a three-position toggle switch. A constant current was supplied to the platinum basket-test electrode circuit by a Hewlett Packard 881A power supply operated in the constant voltage mode in series with a bank of resistors. Anodic current densities ranged from 0.99 × 10-4 to 2.0 × 10-3 A cm -~. A new test electrode and fresh solution were employed for each current density and solution composition studied. The resulting potential vs. time curve was displayed on the recorder of a Beckman Electroscan 30. Potential values were monitored with a Keithley 610B electrometer. A study of differential capacity as a function of potential was also carried out during galvanostatic polarization by superimposing a square wave current pulse on the constant polarizing current. For small measuring times ( < 10 ~s) it was assumed that the change in potential arising from the constant current was negligible with respect to that caused by the pulse. Because of the rapid change in potential with time occurring in some potential regions during constant current polarization, difficulties were encountered in the correlation of capacity values and the potentials at which they were determined. A system was therefore developed which facilitated the sequential storage of pulse-produced transients and supplied a marker signal to the Eleetroscan recorder whenever the pulse generator was triggered (Fig. 1). Since the potential values recorded during constant current polarization reached 1.4 V while the marker pulse superimposed on this signal was only 50 mV, the cell signal was passed through a buffered amplifier and could be clipped at a preset value. The recorder sensitivity could therefore be adjusted to display the pulse marker while the entire potential vs. time curve remained on chart. EXPERIMENTAL

RESULTS

T h e p o s s i b l e r e s p o n s e s o f t h e s t a i n l e s s steel s a m p l e s t o p o t e n t i o s t a t i c p o l a r i z a t i o n a r e g i v e n s c h e m a t i c a l l y i n Figs. 2 a a n d 2b. T h e r e g i o n A B C i n b o t h r e p r e s e n t s t h e transition from the active to the passive state characteristic of this material in the

M.E. CURLEY-FIoRINOand G. M. SCHMID

316

solutions employed. An adjoining region of approximately constant current is then usually observed (Fig. 2a, CD). The increase in current on further polarization is caused by loss of passive state stability as the result of pitting (Fig. 2a, DE) ot of transpassivity (Fig 2b, FG). Characteristic potentials and currents determined from the

-15V

+SV ~ 2M~.-.~I00~r~

,

,

I- Shot MVI __ '

~

~

~-

m

Trig~l Output Pulse ° generator y I- Shot MV3 ] Summingampl. ,~_ " ^ =art

i

,

n

g

Trig. 2 Output

amp~

~

iVnel~ift (os¢lllolcopll)

~I

(;ell

FI~. 1. Block diagram of the circuit for capacity vs. potential data correlation. Io

.~ 3 -

E 8

91- lcrlL r 5-

E

7-

< .-=

7-

I o 5-

'

/

1

[CL']- 3.0 x I0-I M

A

[so:I- 0.232 M pH • 2 . 3 5

\

;6

-;4

-~.~

d.o o12 Potentiol, Vscc

~ i ~" 0.4

.....

i 0.6

Fro. 2a. Potcntiostatic current vs. potential behaviour (pitting, 0.301M KCI). ABC: transition from active to passive state; C D : constant current region; DE: pitting; E,: rest potential; Epp, =: primary passivation potential; Epit: pitting potential; ip: passive current density; icrlt. 1: critical current density.

Effect of the CI- ion on the passive film

317

3~

1 ~,

G " - ice's, '~

B

!

7-

5

A:.

[CL-]" 0.997 x I0 "p M [S0~,]-2.84 x I0-i M pH" 2.40

7 I--i c

] 4-

/

1

~

t,

E~ F" Dp, i ]

l

-0.6

-0.4

-0,2

010

012 Potential,

014

06

E.p,2 01,8

qO

[2

VscE

FIG. 2b. Potentiostatic current vs. potential behaviour (transpassive dissolution, 0.099M KCI). ABC: transition from active to passive state; C'F: constant current region; FG: transpassivity; E,: rest potential; Epp. ~: primary passivation potential; Err: potential at which transpassivity initiates; Epp, s: secondary passivation potential; i,: cathodic loop current density; ip: passive current density; i~,t,t: critical current density; icr,t, ~: secondary passivity current maximum.

potentiostatic polarization curves obtained in primary solutions are presented in Table 1. After the 20 min prepolarization period at - - 0 . 7 0 0 V net currents are cathodic with values between 2.1 × 10-3 and 5.3 × 10-3 A cm -2. As the potential is shifted anodic, the measured current decreases and becomes zero at the rest potential, E,. Polarization at potentials positive to Er results in a net anodic current which increases with potential to a maximum, signifying the onset of passivation. This maximum current density and its corresponding potential are the critical current density (i¢rit.l) and the primary passivation potential (Epp, 1), respectively. All values of icrit.t measured here are between 1.3 × 10-5 and 9.50 x 10-5 A cm -~. From its value at the maximum the current density then decreases to a low value ( ~ 10-e A cm -z) characteristic of the passive region, iu, and is approximately independent of potential. In some cases, priol to the attainment of the passive current value, a negative current loop is observed (Fig. 2b, CicC' ). The potential range in which it occurs, -- 0.3 to 0.0 V, is approximately independent of solution concentration. To evaluate ip without interference from the negative current loop measurements are taken at 0.3 V except in 0.508 and 1.0M KC1 where pitting occurs below this potential. Values of ip range from 5.2 × 10-7 to 1.29 × 10-6 A cm-L In solutions whose sulphate to chloride ion concentration ratio is greater than one ( > 0.3M chloride), a continuing shift to more anodic potentials results in the breakdown of the passive state with the onset of pitting. Two values are recorded for the potential at which pitting initiates, Epit. gpit (inc) is the most negative potential at

318

M . E . CURLEY-FIORINO and G. M. SCHMID

~X

~

II

II

II

II

II

~O

o

<. ~d I Q

Z

r.,4

~

°~

~x

b e~ o

~x

~O

Illlll

[,.

IIIlll

dd I I II e~

~x

~ ~ o

~se~

t~4~

o

Effect of the C1- ion on the passive film

319

which any increase in current density with time is observed within the 10 min waiting period. At potentials slightly positive to Epit 0~:) (0 to ~ 150 mV), an initial decrease in current followed by current spikes is usually observed. At more positive potentials an immediate increase in current density o c c u r s . Epi t (gt) is the potential obtained by extrapolation of the increasing pitting current density vs. potential curve back to the point at which it intersects the value of the passive current at 0.30 V. The graphically determined value of the pitting potential is always the more negative. In systems not subject to pitting attack ( < 0.3 M KC1) loss of passive state stability occurs with the onset of transpassive dissolution. The potential at which transpassivity initiates, Etr, is taken as the intersection of the computed Tafel line with the passive current density measured at 0.30 V. Err shifts negative from 0.671 to 0.575 V with increasing chloride ion concentration. Continued anodic polarization produces a slight current maximum as a result of secondary passivity. Neither the current magnitude at the maximum, icrit,~ (1 X 10-4 A cm-2), nor its associated potential, Epp.2 (0.95 V), is affected by solution composition at constant pH. The current then decreases to a minimum prior to oxygen evolution. A schematic representation of the dependence of capacity on potential is given in Fig. 3. After prepolarization at -- 0.700 V for 20 min capacity values of 40--60 Ffcm -z are typically observed. The one exception is in 0.5M KC1 (22 Ff cm-2). A definite increase in capacity with increasing potential occurs in the region of active dissolution of the metal surface. A single capacity peak (C~) is then described at a potential (E~) which is usually 0-40 mV more negative than the corresponding primary passivation potential. The capacity maximum ranges from 62 to 73 Ff cm -2

--c, .

/~

40-

Q. :30"

20-

"-'C"n"

-~.6

E

E

-d4

-~2

E, o'o

o!2 Potential,

E.~o , E3

o~4

o!~

o8

Er~z

,'o

,

Vsc E

FIG. 3. Potentiostatic capacity vs. potential behaviour. El, C1: potential and capacity of maximum in active region; E2, Ca: potential and capacity associated with approach to complete passivation; Eb: potential of break in passive region capacity; F_~ -- Emt,a,1: region of 1/C vs. E linearity; E~t~,a, C~a~.x: potential and capacity of minimum in passive region; Es, Cs: potential and capacity of maximum in transpassive region; E~n, 2, Cml., ~: potential and capacity of minimum in transpassive region; C-0.7o0: capacity at -- 0.700 V.

320

M.E. CURLEY-FIORINOand G. M. SCHMID

and is independent of solution composition. Although chloride ions are known to adsorb on iron ~1and steel, 12no curvature is observed in the potential transient used to determine capacity values at times less than 10 ~ts. This precludes the contribution of an adsorption pseudo-capacitance term to the interfacial capacity in this potential region. In the vicinity of the potential at which the electrode becomes completely passive, either a capacity peak or a sudden decrease in capacity is observed for all systems (E2, C2). Further anodic polarization results in a smooth decrease in capacity to 18-20 Izfcm -2 for 0M-0.3M KC1 solutions and 13-15 ~tfcm -~ for 0.5 and 1M KCI solutions. The capacity in the passive region then continues to decrease slowly until a potential of 0.28--0.31 V (Eb) is attained (in 0-0.3M KC1) or until measurements are terminated at potentials positive to the pitting potential (in 0.5 and 1.0M KC1). Polarization beyond 0.28 to 0.31 V causes a small fall in capacity followed by a slow decrease to a minimum value of 11-15 ~tfcm -~ (Cmin,1) at 0.6--0.65 V (Emin,l). Plots of 1/C vs. E in the potential region from E b to Emi,,1 are linear (Table 2). Slopes of 0.06, 0.09, 0.06 and 0.04 V cm 2 ~f-a are observed for 0, 1.17 × 10-3, 9.97 × 10-2 and 0.3M KCI, respectively. The corresponding intercepts and correlation coefficients obtained by linear regression analysis are also presented. The increase in current observed in the transpassive region is accompanied by an increase in capacity to a maximum (Ca) of 27-53 ~tf cm -2 at a potential of 0.850.90 V (Es). The capacity then decreases to a minimum of 19-22 ~tf cm -~ (Cmin. 1) at 1.20 V (Emi,, ~) and increases sharply at still more positive potentials. A brief survey of the effect of chloride ion concentration in the absence of sulphate ion on the polarization curve was carried out at pH 2.4 (HC1) in solutions containing 0.102M KC1. Characteristic potentials are included in Table 1. Polarization curves are qualitatively similar to those observed in primary solutions. However, no cathodic current loop is observed in this solution. In addition, in the absence of sulphate ion, pitting occurs. The effect of pH on polarization curve parameters was also examined in 0.3M KCI at pH 1.52 (Table 1). Sodium sulphate maintained the ionic strength at 1. Pitting occurs in all instances here. However, the change in capacity with potential for these secondary solutions corresponds closely to that described above for the primary solutions over the entire investigated potential range. The capacity values observed prior to and in the pitting region are similar to those seen in the same potential regions in non-aggressive solutions (11-20 ~tf cm -~) where the surface remains passive. Linear relationships are observed between 1/C and potential between E b and the potential at which polarization is terminated (Table 2). Polarization, continued for one experiment at pH 1.52, shows a capacity minimum at 0.6 V followed by an increase in capacity in the potential range corresponding to the transpassive region in non-aggressive solutions. No capacity increase is associated with the increase of current caused by the onset of pitting.

Galvanostaticpolarization After potentiostatic prepolarization at -- 700 V, control was switched to the galvanostatic circuit. The potential-time responses resulting from the application of anodic current densities from 1 × 10-4 to 2 × 10-s A cm -~ were observed. Only

Effect of the Ci- ion on the passive film

7T



I



C:;~

r~

(3 Z

Ig

Z

k-

Od

z 0

Od 7. Z

,¢ e.

I

r-i F" rt



O-

de~

odd~

321

322

M.E. CURLEY-FIORINOand G. M. SCHMID

primary solutions (constant ionic strength, p H 2.4) were investigated. Chloride ion concentrations were varied from 0-0.518M. A schematic representation of the current-induced potential transients is given in Fig. 4. The arrests observed and the stability of the maximum potential achieved for a given system are a function of the current density and solution composition employed. Representative data are presented in Table 3.

1.4

o -t- E°2

-2

f

i}0080-

g

050-

4020-0 I0--040-

140

I 200

I 360 4C)0 5~)0 600

Time,

s

F]o. 4. Gaivanostatic potential vs. time curves in the presence (l) and absence (2) of pitting. El: potential arrest associated with primary passivation; Ez: second potential arrest; EaA: potential arrest associated with transpassive dissolution; EaB: potential arrest associated with secondary passivity; Eo2: oxygen evolution potential. All systems studied exhibit an initial potential arrest near -- 0.4 V (//:1) as well as a second arrest at 0.0 V (//2). Potential values associated with arrests occurring at still more noble potentials as well as the stability of the maximum potential achieved are a function of current density and solution composition. In solutions which cause pitting (0.303 and 0.518M KC1), the maximum potential attained is unstable and a rapid decrease in potential to more active values occurs, despite the continued application of anodic current. Systems not susceptible to pitting attack (0M and 0.1M KCI) reach and maintain a constant maximum potential. The arrest observed near 0.8 V, EaA, occurs in solutions containing 0, 0.1 and 0.303M KCI and is visible at the recorder speeds employed (10-20 s in -x) for current densities of < 3.7 × 10 -4 A cm-L In 0.303M KCI this arrest is also seen in two of five experiments conducted at i = 1.0 x 10 -a A cm -2. In solutions containing 0.303M KCI polarized by current densities < 1.0 × 10 -a A cm -2, the potential of this plateau is the maximum achieved and breakdown to more active values ensues. The next arrest observed, Eaa, occurs in the vicinity of 0.85 V in 0.0M

Effect of the CI- ion on the passive film

323

TABLE 3. GALVANOSTATIC POTENTIAL ARRESTS.

(A era-*) × 10~ 0.99 0.99 0.99 0.99 1.9 1.9 1.9 1.9 3.8 3.8 3.8 3.9 9.9 9.9 10.1 9.9 20.0 20.0 19.7 20.0

[O1 × 10 ~ 0.0 1 . 0 0

3.03 5.18

[SO~'-I x 10 x 3.21 2.84 2.35 1.48

E~ (V, SCE) - -

------ -

---- -

- -

--

0.421 0.439 0.425 0.405 0.420 0.423 0.427 0.395 0.415 0.419 0.421 0.365 0.374 0.363 0.409 0.358 0.388 0.360 0.412

E, (V, SCE) -

----------- -

E,A (V, SCE)

EsB (V, SCE)

0.110 0.110

0.798 0.783*

0.840* --

0.125 0.150 0.125 0.130 0.075 0.045 0.010 0.09 0.140 0.04 0.10

0.447t 0.830 0.810 0.765t 0.471t 0.830 0.852 0.871t 0.508t ---0.683t --

-0.880 0.868t --0.890 0.902 --0.903 0.900 0.925 -0.920 0.952 0.905 --

- -

-- 0.08 0.20 0 . 1 5 5

-0.21

- -

-0.798~"

E, (V, SCE) ---1.145" ---1.26" 1.25" --1.36" 1.37" 1.30~r -1.40* 1.41 * 1.38t --

*Steady state. tMaximum E before breakdown.

(i > 0.99 × 10-4 A cm-t), 0.1M (i _> 1.93 x 10-4 A cm-~), and in 0.303M (i ~ 1.0 × 10-s A cm -2) KCI. In only two of five experiments at current densities of 1.0 x 10-s A cm -8 does breakdown occur from this plateau in 0.303M KCI. In all other cases, as with non-pitting systems, further polarization to oxygen evolution potentials occurs. The potential arrest associated with oxygen evolution, Eo2, is seen in 0M ( i > 1.94 × 10-4Acm-2), 0.1M (i_>3.72 × 10-'Acm-~), and 0.303M ( i > 1.0 × 10-8 A cm -2) KCI. Attainment of this plateau in 0.303M KCI is followed by pitting breakdown. Constant current polarization of specimens in 0.518M KCI initiates breakdown from potentials substantially below those observed in 0.303M KC1. Breakdown occurs from 0.447, 0.471, 0.508, 0.683 and 0.798 V during the initial imposition of 0.997 x 10-*, 1.93 × 10-4 , 3.89 × 10-4 , 0.995 × 10-3 and 2.02 × 10- 3 A c r e -~, respectively. The differential capacity of the stainless steel solution interface was determined as a function of potential during constant current polarization. A capacity vs. time curve for a system not subject to pitting (0 and 0.1M KCI) superimposed on its associated potential vs. time curve is presented in Fig. 5. Present results indicate that this curve also represents the behaviour in pitting systems. Steady state capacity values determined at the end of the 20 min prepolarization period range from 21 to 48 ~f cm -2 (C-~0o). In all systems subjected to galvanostatic polarization (0 to 0.518M KCI), a capacity peak, (71, is associated with the first potential arrest observed, Ev Solutions containing chloride ion exhibit a slightly greater peak magnitude (38-58 ~f cm -~) than those with none (35-48 ~f cm-2).

324

M.E. CURLEY-FIORINOand G. M. SCr~ID ~o 2 1.40 ~ ~

EO2

- 70

++oo,

,.,o.

"~E3B

I

-

-

0,50-

/~

020

.3+c,~oo I

- 0 10-

\E

- 0 40-

\

]~E2

C3

2

/

~

50

- 40 4~

C-7o0--~-30 C2

U

- 20

-10

E~

Time,

s

FIG. 5. Capacityvs. time behaviour during galvanostatic polarization. El : potential arrest associated with primary passivation; E~: second potential arrest; EaA: potential arrest associated with transpassive dissolution; EsB: potential arrest associated with secondary passivity; Eo=: oxygen evolution potential; C-0.700:capacity after 20 rain at - 0.700 V; Ca: capacity maximum associatedwith primary passivation; C~: capacity value at second potential arrest showing rapid capacity decrease from active to trampassive region; C8: capacity maximum in transpassive region; Co~: capacity maximum in oxygen evolution region. Capacity values observed during the rapid potential transition from the active to the transpassive state range from 12 to 22 ~f cm -a (C2) in 0 and 0.1M KCI. This represents a substantial decrease from the values in the active region. Similar capacity decreases occur prior to and during potential breakdown from the transpassive region in 0.303M KC1 The second capacity peak observed in systems containing 0 and 0.1M KCI (Ca) is associated with the potential arrest at 0.85 V (EaB). Peak values of 20-40 Ezfcm -~ are measured. The continued change in potential to oxygen evolution values is accompanied by a capacity peak (Co~) only when potentials > 1.3 V are attained. The maximum potentials achieved in these solutions are stable with respect to time. No pitting-induced potential breakdown occurs. Corresponding steady state capacity values range from 15 to 26 ~tfcm -z for final potentials _< 1.26 V and from 21 to 39 ~tf cm -a for final potentials > 1.34 V. In 0.303M KCI the potential arrests observed and the potential from which breakdown occurs coincide with arrests occurring in non-pitting systems. Capacity data, although not conclusive, suggest that the corresponding capacity-potential behaviour is also similar. A definite capacity increase is associated with potential arrests and/or maxima between 0.72 and 1.04 V. A capacity maximum (Con) occurs prior to breakdown from potentials in the oxygen evolution region (1.3-1.4 V). The magnitudes of the largest measured capacities are 15-24 and 28-63 tzf cm -2, respectively, in good agreement with those found in non-pitting systems in similar potential regions. The highest potentials (0.447-0.508 V) attained in 0.518M KCI for the three lower current densities employed are substantially more negative than any of the potential arrests characteristic of non-pitting systems. The most positive potentials

Effect of the Ci- ion on the passive film

325

reached at two higher current densities more closely approximate non-pitting values (0.683 and 0.798 V). The capacity vs. potential behaviour resulting from the application of 0.997 × 10-4 and 1.93 × 10-4 A cm -a exhibits a well-defined, smooth decrease in capacity from the prepassive region across the potential-time breakdown peak. Typical capacity values associated with these potential maxima range from 13 to 15 ~f cm -2. For 3.89 x 10-4 A cm -~ the number of data points available is too small to allow precise definition of the capacity vs. potential curve. However, the shape of the curve seems similar to that described above but with slightly higher capacity values (16 and 18 ~f cm -2 at potential maxima of 0.568 and 0.588 V, respectively). Since only two data points are available at the two higher current densities (9.95 × 10-4 and 20.2 × 10 .4 A cm-~), no conclusions can be drawn with respect to the presence or absence of a capacity peak associated with the potential-time maximum achieved prior to breakdown. However, capacity values determined for these higher potential maxima approach those observed in non-pitting systems. DISCUSSION The results of the present studies suggest that the initial passive films formed during anodic polarization of AISI 304 in both pitting and non-pitting environments are similar in both their electrical and chemical properties. This conclusion is based on: 1. Comparison of the potentials at which galvanostatic arrests occur in both environments; 2. Comparison of the potential arrests and peaks observed in both environments; 3. Comparison of the capacity vs. potential behaviour during galvanostatic and potentiostatic polarization in both environments; 4. Evaluation of film thicknesses formed during potentiostatic polarization in both environments. The values of galvanostatic potential arrests and maxima observed in both pitting and non-pitting environments are coincident. Specifically, in 0.1M KC1 solutions, no pitting occurs while samples subjected to constant current polarization in 0.303M KCI solutions undergo pitting corrosion. In both systems passivation, transpassive dissolution, secondary passivity, and oxygen evolution are observed at one or more current densities (Table 3). In 0.3M KC1, however, the chloride concentration is high enough to induce pitting. Noble potentials in the transpassive dissolution or oxygen evolution regions are then no longer necessary to maintain the imposed current density, and a decrease in potential to values associated with the pitting reaction occurs. Only the plateau associated with passivation is observed in 0.5M KCI solutions since pitting occurs from potentials considerably below those observed in 0.3M KCI (0.447-0.798 V vs. 0.765-1.390 V, respectively). Comparison of the characteristic potentials observed during galvanostatic and potentiostatic polarization as well as the corresponding capacity vs. potential behaviour both in pitting and non-pitting media also supports the conclusion that the initial passive film is similar in both environments. Excellent agreement exists between the low current density (0.99 × 10-4 A cm -~) value of the initial potential arrest, El, leading to the passivation of the alloy surface and the potentiostatic steady state primary passivation potential in all solutions investigated. The plateau at 0.8 V is

326

M.E. CURLEY-FIORINOfind G. M. SCHMID

attributed to the transpassive dissolution of chromium from the passive film since transpassive dissolution under potentiostatic conditions begins at approximately 0.6 V and terminates in secondary passivity at 0.95 V. The latter value is within the potential range of the second galvanostatic arrest observed in this region. At higher current densities a potential arrest corresponding to oxygen evolution also occurs. A capacity peak (CI) is associated with the active to passive transition under both galvanostatic and potentiostatic conditions in all solutions studied. A maximum in the capacity-potential relationship (Ca) is also associated with the onset of secondary passivity in both cases in solutions where this potential range is attained. During gaivanostatic polarization under conditions leading to oxygen evolution, an additional maximum is observed. No capacity peak appears to be associated with the timeconsuming alteration of the surface occurring prior to pitting under either galvanostatic or potentiostatic conditions. Additional support for the similarity of the films formed in pitting and non-pitting solutions is provided by the current density vs. potential and capacity vs. potential behaviour observed in the passive region during potentiostatic polarization. The current flowing across a passive metal-solution interface may have three components: an electron current flowing through the passive film, e.g. in the presence of a redox couple; an ionic current caused by the passage of ions through the film resulting in film growth; and an equivalent ionic current arising from dissolution of the passive film The low current densities observed in these studies in the passive region in systems not subject to pitting are independent of potential. In the absence of an electron current, this implies that the rate of film growth is independent of potential since, for short times ( < 1 h), the ionic current of film dissolution has been shown to be negligible with respect to the film growth current. ~e An increase in potential in the passive region must therefore induce a proportional increase in film thickness, resulting in the maintenance of a constant field across the film. For an 18-8 stainless steel in 0.SM H2SO4, Rahmel and Schwenk ~7have shown that the logarithm of the rate of passive film growth is inversely proportional to the thickness of the film, i.e. growth occurs in accordance with the inverse logarithmic law, i = A exp (BV/d) where i is the current density, V is the potential drop across the film, d is the film thickness, and A and B are constants. This behaviour satisfies the constant current density/constant field relation and can be explained in terms of the Mott-Cabrera ~s theory of high field film growth. This theory, originally derived for valve metals such as tantalum, has been extended to include the films formed on passive metals. 29 Growth occurs by the high field (ca. 108 V cm -1) conduction of metal cations through the film. The current-potential behaviour is therefore exponential rather than linear. If the presence of a constant current in the passive region implies a proportionality between field and passive layer thickness and if the interfacial capacitor can be represented as a parallel plate condenser with the passive layer as its dielectric, then a plot of reciprocal capacity (l/C) with respect to potential (E) should be linear. The contribution capacity to the total interfacial capacity is assumed to be small (see below). In the present studies, this theory does not apply to the entire potential range in which the passive current density remains constant or in the potential region

Effect of the Cl- ion on the passive film

327

preceding the break in capacity at about 0.3 V (Eb). However, in the potential range bounded by about 0.3 V and the potential at which the capacity minimum (Cmin,1) occurs or at which polarization is terminated, excellent correlation between the two functions is exhibited (Table 2) not only for solutions in which the passive state is stable but also under pitting conditions. The capacity of a parallel plate condenser is given by C = ep/0.113d where C is the capacitance in ~tf cm -2, ~ is the dielectric constant, p, the surface roughness factor and d, the distance between the capacitor plates in A. The constant, 0.113, has the units cm 2 ~f-1 A-1. The film thickness may therefore be calculated from the measured value of capacity at a given potential in the potential range in which the condition of linearity between I/C and E is satisfied. Since the measured capacities are not steady state values, only an approximation of the film thickness corresponding to a given potential is obtained. An analysis similar to that applied by Engell and Ilschner a° to iron in sulphuric acid can be applied to the capacity-potential behaviour observed in the present experiments (Table 2). Using the roughness factor of 1.2 reported for an iron surface mechanically polished through 600 grit emery, 31 the dielectric constant, 15.6, determined for films formed on AISI 304 in 0.5M sulphuric acid, la and applying a 2 A correction to account for the film-solution capacity contribution to the measured capacity, a° film thicknesses of 5.9 and 10.7 A at 0.30 and 0.60 V, respectively, are obtained for AISI 304 in 0.334M Na2SO4. The passive current density evaluated zt 0.30 V is 1.26 x 10-e A cm -2. The field required to maintain this rate of film formation is therefore (0.60o0.30)/(10.7-5.9) = 6.2 x 10-3 V A -1, or 6.2 x 10e V cm -1. Using the same approach, thicknesses of 5.98 A and 6.62 A are found in 1.17 x 10-2M and 9.97 x 10-2M KCI at 0.3 V, respectively. Field strengths of 5.4 x l0 s and 10.4 x 106 V cm -1 are calculated in conjunction with the corresponding thicknesses at 0.6 V (11.5 A and 9.5 A). Similar results are obtained in solutions in which pitting occurs. Linearity in the 1/C vs. potential relationship appears to be retained in spite of the imminence of pitting breakdown or an increase in current density due to the actual onset of pitting. At 0.3 V in 0.3M KCi (pH 2.4) the film is 6.2 A thick. Somewhat thicker films are found at pH 1.52 but the field of 5.7 x 106 V cm -1 calculated from thicknesses evaluated at 0.3 and 0.6 V (9.14 A and 14.4 A, respectively) agrees well with results in non-pitting systems. This similarity in film thicknesses found in solutions whose chloride content ranges from 0-1M indicates that the solubility of the film and therefore the chemical nature of the film on the alloy surface is the same in all cases. The analysis and comparison of film compositions by electrochemical techniques are limited since only the average properties of the interface can be measured. However, recent Auger depth profiles of the films formed on ferritic stainless steels support the conclusions of electrical and chemical similarity drawn here. Studies of the passive films at potentials in the passive region on alloys both immune and subject to pitting at higher potentials suggest that the presence of chloride in solution has no effect on passive film composition with respect to alloying elements and no difference in film

328

M . E . CURLEY-FIORINOand G. M. SCHMID

thickness is observed between films formed in 0.SM Na2SO4 (pH 3) and 1M NaCI (pH 3.45)) 2 In addition, chloride is restricted to the outermost layers of the film at all potentials studied)2, a3 However, Sugimoto and Sawada 34 using variations in peak area ratios interpret their XPS studies of the outer layers of films formed on Mo-containing 20Cr25NiFe alloys to suggest that films formed in 0.5M H2SO4 (pH 0.4) at 0 V are higher in Cr s+ and lower in Fe e+ than those formed in 1M HC1 (pH 0.02) at 0.3 V. In addition, the films formed in 1M HC1 at 0.3 V on 1 and 3 %Mo alloys which pit at this potential show a higher chloride/oxygen peak area ratio than that on 5 %Mo alloys which are immune (0.10, 0.11 and 0.07 respectively). The nature of the time-dependent phenomenon which leads to the pitting breakdown of passive films at potentials noble to the pitting potential remains to be investigated in more detail. The correspondence of the 1/C vs. E relationship observed here suggests one of two possibilities. Either the process leading to breakdown is localized and therefore not reflected in the measured capacitance which is an average property of the interface, or the process involves specific adsorption of chloride ion to a value greater than some critical value. In the latter case, a more complex analog of the interface in which the changes in the capacity of the electrical double layer are obscured by relatively small values of a film capacitance would have to be adopted. CONCLUSIONS

From the coincidence of potential arrests observed during galvanostatic polarization in both pitting and non-pitting solutions and the approximate agreement of the potential values of these arrests with characteristic potentiostatic potentials describing passivation, transpassivity, secondary passivity, and oxygen evolution, it is concluded that the initial film formed on the surface of AISI 304 stainless steel under the influence of anodic polarization is the same under all experimental conditions. This conclusion is further supported by the linearity observed in the I/C vs. potential relationship and similarity in film thicknesses occurring in both pitting and non-pitting environments. Film thicknesses are calculated by representing the interracial capacitance as a parallel plate condenser with the passive layer as its dielectric. Values of 5.97, 5.98 6.62 and 6.21 A were found at 0.3 V in 0, 1.17 × 10-3, 9.97 × 10-8 and 0.301M KCI, respectively. It therefore appears that pitting occurs through the perturbation of an initial surface film by chloride ion.

REFERENCES 1. H. H. UHLIG)Corrosion 19, 231 (1963). 2. K. OSOZAW^and H. J. ENGELL) Corros. Sci. 6, 389 (1966). 3. YA. M. KOLOTYRKIN, Corrosion 19, 261 (1963). 4. Z. SZKLARSKA-SMIAL~WSKA)Corrosion 27, 223 (1971). 5. G. M. SCHMXDand N. HACKERMAN,J. electrochem. Soc. 108, 6. W. SCHWENK, Corrosion 20, 129 (1964). 7. M. JANIK-CZACHORand Z. SZKLA~KA-SMIALOWSKA,Corros. 8. Z. SZKLARSKA-SMIALOWSKAand M. JAMK-CZACHOR,Corros. 9. Z. SZ~:LARSKA-SMIALOWSKAand M. JANIK-CzAcrIoR, Corros. 10. K. NOnE and R. F. TonlAS, Corrosion 20, 263 (1964).

741 (1961). Sci. 8, 215 (1968). Sci. 11,901 (1971). Sci. 7, 65 (1967).

Effect of the CI- ion on the passive film

329

11. YA. M. KOLOTYRKIN,J. electrochem. Soc. 10g, 209 (1961). 12. C. D. KIM and B. E. WILDE, Corrosion 28, 26 (1972). 13. P. POPAT and N. HACKERMAN,J. phys. Chem., Ithaca 65, 1201 (1961). 14. YA. M. KOLOTYRKIN,Z. Elektrochem. 62, 664 (1958). 15. V. S. KuzuB and G. G. KossYi, Zh. priM. Khim. 38, 104 (1965). 16. N. E. WISDOMand N. HACKERMAN,J. electrochem. Soc. 110, 318 (1963). 17. A. M. SUKHOTINand K. M. KARTASHOVA,Corros. Sci. 5, 393 (1965). 18. G. i . SCHMIDarid N. HACKERMAN,J. electrochem. Soc. 109, 1096 (1962). 19. R. MOSHTEV,Ber. Bunsenges. phys. Chem. 72, 452 (1968). 20. M. A. V. DEVANATHANand K. RAMAKRISHNAIAH,Electrochim. Acta 18, 259 (1973). 21. T. MURAKAWA,T. KATO, S. NAGAURAand N. HACKERMAN,Corros. Sei. 7, 657 (1967). 22. H. VAIDYANATHANand N. HACKERMAN,Electrochim. Acta 16, 2193 (1971). 23. T. MURAKAWAand N. HACKERMAN,Corros. Sci. 4, 387 (1964). 24. S. HARRAR, Analyt. Chem. 34, 1036 (1963). 25. J. S. RINEY, G. M. SCHMID and N. HACKERMAN,Rev. scient, lnstrum. 32, 588 (1961). 26. G. M. BULMANarid A. C. C. TSEUNG,Corros. Sci. 12, 415 (1972). 27. W. SCHWENKand A. RAHMEL,Electrochim. Acta 5, 180 (1961). 28. N. CABRERAand N. F. MOTT, Rep. Prog. Phys. 12, 163 (1948-49). 29. J. L. ORD and J. H. BARTLETT,J. electrochem. Soc. 112, 160 (1965). 30. H. J. ENGELLand B. ILSCHNER,Z. Elektrochem 59, 716 (1955) 31 S THIBAULT,J. M. GODOT, J. PAGETTI and J. TALBOT, Proc. Int. Cong. Metal. Corrosion, 4th, 1969, 594 (1972). 32. A. E. YANIV, J. B. LUMSDENand R. W. STAEHLE,d. electrochem. Soc. 124, 490 (1977). 33. M. OA CUNHABOLO, B. RONI~T, F. PONS, J. LE HERICYand J. P. LANGERON,J. electrochem. Soc. 124, 1317 (1977). 34. K. SOGIMOTOand Y. SAWADA,Corros. Sci. 17, 425 (1977).