1990-sato-an_overview_on_the_passivity_of_metals.pdf

Corrosion Science, Vol. 3 I, pp. 1-19, 1990 Printed in Great Britain

0010-938X/90 $3.00+0.00 © 1990 Pergamon Press pie

AN O ~ g n V I v E W O N T H E P A S S M T Y O F ] E T A L S NORIO SATO

Electrochemistry Laboratory, Faculty of Engineering, Hokkaido University, Sapporo, 080 Japan

ABSTRACT A review is made of some essential subjects on the passivtion process, the passive film, and the film breakdown. Passivation of metals results from the formation of a condensed phase of continuous oxide layer on the metal surface. The electrochemical stability of passivated metals depends not only on the chemical property but also on the electronic property of the passive films. The electronic avalanche breakdown of passive film occurs when the film is thick, while the ionic breakdown emerges with thin oxide films in the presence of aggressive anions. The film breakdown is followed either by repassivation or by pitting corrosion depending on the degree of metal salt enrichment at the film breakdown sites.

KEYWORDS Passivation kinetics, semiconductors

passive

oxide

films,

passivity

breakdown,

1. INTRODUCTION Most engineering metals and alloys oxidize and frequently passivate even in aqueous conditions to form thin surface oxide films, the stability of which is much greater than the metal itself. The phenomenon of metal passivation was discovered by Keir (i) as far back as 1790, who observed that metallic iron in concentrated nitric acid became suddenly in the altered state (the passive state) after violent metal dissolution had occured in the fresh state(the active state). Since then, a number of electrochemists and materials scientists have been associated with the histry of metallic passivity. Recent years have seen a considerable progress in developing knowledge of passivity. New optical, electronoptical and electrochemical surface analysis techniques have allowed us to disclose the molecular nature of passive oxide films even several monolayers thick formed on passivated metals (2-4). In the field of semiconductors the passivity is also one of the subjects of practical importance not only in their corrosion resistance but also in the stability of their surface electronic property. The electronic stability is usually determined by the electron energy level distribution, whereas the corrosion stability is mainly associated with the chemical energy levels of constituents in the surface layer of solid

2

N. SATo

i

!

!

!

i

Fe/HsPO4. pH1.85

6 5

s -3

,, 2

-4

2

-5 I

0

0.4

I

I

1.2

1.6

ELECTRODE POTENTIAL

Fig. 1

Jl'

t

0.8

0

E (V, NHE)

Anodic polarization curve (current i vs. potential E) of iron in phosphoric acid solution and thickness L of passive oxide films as a function of potential (5).

. . . . . . . . . .

2

-1

~L -2

-3 -1.30

-1.25

-1.20

-1 15

-1 10

ELECTROOE POTENTIAL E ~ )

Fig. 2

Anodic current vs. potential curves calculated for different values of an interaction energy parameter ~ between the metal hydroxide ad-ions, O k ~ 6, on the metal surface: Other rate parameters for metal dissolution and oxide formation are fixed (6).

materials. Thus we need the knowledge of both atomic and electronic properties of the surface layer for the better understanding of the passivity of metals and semiconductor.

Ovc~iew

METAL

PRECIPITATE

/ ANION-SELECTIVE

Fig. 3

2.

3

FILM

ELECTROLYTE

\ CATION-SELECTIVE

A bipolar ion-selective corrosion precipitate film retarding anodic ion transport because of its rectification effect (7). ~)= positive fixed charge, ~)= negative fixed charge.

PASSIVATION OF I~TALS AND SENICONDUCTORS

The nature of metal passivation in aqueous electrolytes can be illustrated most clearly by the anodic polarization curve (current vs. potential plot) of metals in the active-passive transition region, in which the anodic dlssolution(oxldation) current passes through a maximum and decreases steeply with increasing anodlc polarization, as shown for iron in Fig. 1 (5). In the passive state of iron a thin surface film of iron oxide is formed whose thickness increases with anodlc potential. Passivation kinetics may well be developed by a two-dimentional phase transition model (6), in which a dilute phase of isolated metal hydroxide ad-ions(adsorbed ions) is transformed into a condensed phase of continuous oxide layer on the metal surface. The elementary steps may be represented as follows, _

-ne M(solid)V

M ~+

M~+ ~ (ad)

(aq) M~Jxide)

(1)

where step 1 is the oxidative hydrolysis forming adsorbed metal hydroxide ions, step 2 the dissolution or desorption of hydrated metal ad-ions into the aqueous electrolyte phase, and step 3 the two-dimentional transformation forming a condensed phase of metal oxide layer. The phase transformation occurs at a certain potential when the interaction or bridge-formatlon energy between neighboring metal hydroxide ad-ions is greater than a threshold value, and the metal dissolution rate is considerably diminished by the complete coverage of a surface oxide layer, as schematically shown in Fig. 2 (6).

N. SATO

4

I

1.0

Si 1 40~ KOH. 60"C, dark 1.5

0.8

.... n-typm

R '~ 0.6

i 0.4

f

./"

..........

/'kl/

1.0

,.J

0.5

0.2

0

-1.0

-0.5

0

+0.6

+1.0

0

ELECTRODE POTENTIAL E I (V, NHE)

Fig. 4

Anodic polarization curve (current i vs. potential E) of < 1007 oriented p-type and n-type silicon electrodes in 40% KOH aqueous solution at 60eC and thickness L of anodic oxide films as a function of potential (8).

The passivation kinetics is also closely related to the rate constants of step 1 and step 2, both of which are controlled by the potential of the electrical Helmholtz double layer at the metal-electrolyte interface. The overall rate constant of step 2 (metal ion dissolution) is further controlled by the anodic ion migration across a concentrated metal-salt electrolyte layer or a porous precipitate layer on the metal surface. If this interracial electrolyte or precipitate layer is of bipolar character consisting of an anion-selective inner layer on the metal side and a cation-selective outer layer on the bulk electrolyte side, as shown in Fig. 3, the anodic ion migration is hindered because of its rectification effect, and hence the passivation will be accelerated. This is the passivation induced by the bipolar nature of an ion-selectlve interracial layer (7). Anodic passivation of semiconductors emerges in much the same manner as that of metals. Figure 4 shows the anodic polarization curve of p-type silicon in basic solution (8). In the active state silicon dissolves in the form of divalent soluble cation, Si(OH) + or Si(OH)2, while a surface layer of quadravalent silicon dioxide, Si02, is formed in the passive state (8). In general anodic semiconductor passivtion proceeds with electrochemical reactions involvin K the valence band holes and/or the conduction band electrons, and hence the passivation behavior differs with different types and concentrations of dopants. Silicon is known to passivate readily at hiEh concentration of p-type dopants with the following reactions, +2h +

~ S i ( O H ) 2 ( a q

)

Si(solid)--T--~ Si(ad) 3__.~_~__SiO2(oxide) +2h +

(2)

Overview

5

OXIDE I ELECTROLYTE

METAL I OXIDE

/

!

~INNER

I

~1

OUTER

LAYE R'4~I

L.AYER

- I I

I A', OH"

!

I O ~-

2.0

0.10

~, 0 . 0 8

OIAI I I I I

g o.os '~ O . 0 4 ul

:/-

~ 0.02 ,s 0

O 1.0

m

la 10

20

I 40

FILM THICKNESS

Fig. 5

3.

0

PIA] I 30

R 50

l 60

7O

0

L (nm)

Depth profile of O/A1 and P/A1 mole ratio across an anodic oxide film consisting of an anion-incorporating outer layer and a dehydrated inner layer, and ion transport during anodic oxide formation on aluminum (9).

PASSIVE

OXIDE

FILMS

In most cases, the passive film grows with increasing anodic potential a~ the rate of I ~ 3 nm/V, which corresponds to an electric field ixlO~--~lO V/cm in the film (Figures I and 4). For most engineering metals and alloys the thickness of spontaneously formed passive films are less than .~everal nano-meters in the potential region where water is thermodynamically stable. ~le passive film on iron in aqueous solutions grows up to about 5 nm at potentials prior to the onset of oxygen evolution. If the passivated metals are those such as iron, nickel and cobalt on which anodic oxygen evolution takes place, there is virtually no further thickenin~ of the film in aqueous electrolyte. For refractory metals such as aluminum and niobium on which no anodic oxygen evolution proceeds, the film can be £rown to thicker than iO0 nm by increasing the polarization up to a critical hi,heat limit of potential at which the film breakdown occurs. The film growth proceeds by the place-exchange and/or field-assisted ion transport. Oxygen ions migrate toward the metal/oxide interface and react with metal ~ons to form an inner part of the oxide, while metal ions mi£rate toward the oxide/electrolyte interface and react with oxygen ions and other anions present in the electrolyte to form an outer anionincorporating part of the oxide (Fig. 5). The ratio of the outer anionincorporating layer thickness L(out) to the total film thickness L(t) = L(out) + L(in) will be equal to the tranport number T(M) of the metal ion migratin~ thrmugh the film (g). The incorporation of electrolyte anions takes place presumably only when the film grows at the oxide/electrolyte interface, and their concentration in the films is greater when the electrolyte concentration ratio [anion] / [OH] at the film surface is greater (I0). For a given CS 31-B

6

N. SAxo

Ti / 0.1M HzSO4

50

-3 40

|jj, J 4"S~' /

-4

/"

i --

/"

3o

-5

v 20

"J

t0

0

2

4

6

8

111

ELECTRODE POTENTIAL E (V. RHE)

FiK. 6

Anodic polarization curve (current i vs. potential E) of titanium in sulfuric acid solution and thickness L of anodic oxide films as a function of potential (12). i d = anodic metal dissolution current.

MO

MnOm

Fig. 7

MnOm

(A)

(B)

Two groups of oxide films on metals, (A) network modifiers forming a multi-layered film and (B) network formers with a single-layered film (13,14).

bulk electrolyte pH there will be a linear relationship between the logarithum of the impressed anodic current and the pH at the film surface, provided the acidity [ H + ] ( s ) at the surface is greater than that [ H+ ] (b) of the bulk electrolyte (I0). The incorporated anion concentration, therefore, will increase as the anodic current is increased. Generally, when an anodie oxide film is formed at constant ,~ot~ti~1, an initial large anodic current gradually decreases with

Overview

/ 0.~ ~', -5

7

pH8A2

-%

O

I \ , A

/

. . . .

l_

.L'/ t

~.4

I

0

I

f T ~ I

/

Z I

0.4

7

TI

0.8

. =

IT

1.2

~ECTR~E~NT1ALE~.NHE)

Fig. 8

Anodic polarization curve (current i vs. potential E) of cobalt in borate solution and thickness L of passive oxide films as a function of potential (15). Linn and Lout are the thicknesses of an inner CoO layer and an outer Co304 or Co203 layer.

increasing film thickness and reaches its steady state current. The incorporated anion concentration, therefore, is greater in the initial stage than in the final stage of oxide formation, provided the oxide formation and lisso]ution are simultaneously taking place. Generally. the anodic passive oxide films are not crystalline when they are thin, but turn frequently to being partially crystalline when they become thick for some reasons such as internal compressive stresses created in the film during its growth (I0). Since crystallization is slower in the presence of impurities (II), the crystalline part may develop most readily at the inner part of amorphous anodic oxides (Fig. 5). Figure 6 shows the thickness-potential relationship of the anodic oxide film on titanium suggest~n~ the development of a crystalline part in the thick film at high potentials (12). In general the presence of a crystalline part may impair the electronic, electrical and corrosionresisting properties of the film. The passive oxide films may be divided into two groups characterized by the structure and composition of the resulting oxides, which have been labeled network(glass) forming oxides and network modifiers (13,14). The former is exemplified by Si, AI, Ti, Zr and Mo and the latter Fe, Ni, Cu and Pt. Metals of the network modifiers passivate by forming primarily an oxide film of low or intermediate oxidation state frequently terminated by a stable oxide film of its highest common oxidation state (FIE. 7). For examples, at relatively high anodic potentials, iron passivates w~th a bilayer film of Fe/FeqO~Fe20 q , copper with and cobalt with Th~ ~ilm c~mposition depends, of course, on the potential and the ~ h e r valent oxide is formed when the potential is more anodic, as s h o ~ for cobalt in Fig. 8 (4,15). In most cases the lower valent oxide of a network modifier is less corrosion-resistant than their higher valent oxides. Metals of network forming oxides are characterized by the formation of passive oxide films

Cu/Cu20/CuO,

Co/CoO/Co~3.

8

N. SATO ANION-SELECTIVE LAYER

M % E¢

Ev

EF

METAL

Fig. 9

FILM

ELECTROLYTE

Bipolarity of passive oxide films in their ion-selectlve and semiconductive properties.

in one oxidation state of the highest AI/AI203 and Ti/TiO 2 (Fig. 6) (12).

common oxidation

state;

Si/Si02,

According to Fehlner-Mott (13), the network forming oxides grow by inward oxygen diffusion and hence form a dehydrated compact film containing no anions incorporated from the electrolyte, whereas the network modifiers £row by outward metal ion diffusion and hence form an anionincorporating, less protective film. There are of course exceptions; the anodic passive film of iron of network modifier appears to grow by inward oxygen diffusion (16). From the standpoint of selective ion permeability, the passive oxide film may be characterized by a bipolar structure (7) consisting of excess metal ions or oxygen ion vacancies in the inner layer, which provide the positive fixed charge with an anion-selective property, and excess oxygen ions or met~l ion vacancies in the outer layer, which provide the negative fixed charge with a cation-selective property, as shown in Fi E. 9. This type of bipolar ion-se]ective oxide films, which acts as an ionic current rectifier, blocks the anodic ion transport across the film and prevents the anodic metal corrosion. Furthermore, the negative fixed charge in the outermost layer prevents aEEressive anions from incorporating into the film and enhances the corrosion resistivity. From electronic view points the passive films are either insulators or semiconductors. Most of the passive oxide films behave as n-type semiconductors(Fe, T i ~ 9 Sn, 2~b,_ 3 W,etc.) with a very high donor concentration N = I0 ~ I0 cm . A few of them are p-type semiconductors(Ni, Cr, Cu,etc.) and insulators(Al, Hi,etc.). Recent photo-electrochemical studies have allowed us to obtain the flat band

Overview

+

9

+

p

+/I

<

ELECTRODE POTENTIAL E

Fig.lO

Electron energy band bending and Helmholtz layer potential ~ u at the oxide/electrolyte interface as functions of electrode potential (21). •

ii

potential Efb , the bandgap energy Eg, and the donor or acceptor concentrations of the passive oxide films grown on metals and alloys in different electrolytic environments. It appears that thin amorphous oxide films on metals differs in their electronic properties from the corresponding crystalline bulk oxides. In many cases the bandEa p enerKy of passive oxide films can deviate from that of bulk oxides by up to 10% (17). For amorphous oxide films with a high concentration of localized donor and acceptor states, the term bandgap should be replaced by the mobility gap, which is delimited by two sharp mobility edge Ec and Ev in the conduction band and valence band, respectively. It is in fact the mobility gap which can be determined from photo-electrochemical measurments of passive oxide films. For simplicity, however, we use in this paper the term band~ap even for amorphous films. For thin oxide films the thickness may be insufficient to accommodate the space charge associated with the metal/oxide contact, and hence the possible space char£e in the film may not mean much compared to the surface charge on the metal, ~(m') and on the oxide, ~(ox), in determJnin£ the electrode potential. At the oxide/electrolyte interface there arises an electrical Helmholtz double layer, which separates the surface charge ~ ( o x ) on the oxide from the excess ionic charge at the plane of closest approch of hydrated ions to the oxide surface in the electrolyte. In most cases the Helmholtz layer potential ~(H) is determined by the acid-base dissociation of the surface hydroxyl groups, MOH(acld site)

~

~MOH(base site)

~

~ MO- ~ H ~ aq EMH20*+ OH&

(3)

IO

N. SATO

and thus represented as a function of electrolyte pH. ~+(H)

= RT (pHpzc - pH) 2F

{4)

where pHpzc is the point-of-zero-charge pH at which ~ ( H ) = O. The Helmholtz potential ~ ( H ) is independent of the electrode potential as far as the surface charge (ox) is constant. Figure I0 shows the electronic energy band diagrams and the Helmholtz layer potential for thin n-type, i-type and p-type oxide films on metals as a function of potential. When the passive oxide films are anodically or cathodically polarized, the band bends upwards or downwards in the film creating a potential drop 4 E equal to the applied bias potential, while the Helmholtz layer potential ~ ( H ) remains constant. This mode of potential distribution is maintained as long as the Fermi level remains within the bandgap between Ec and Ev at the oxide/electrolyte interface. As we further increase the applied bias potential, the bandbending becomes so great that the Fermi level may be lower than Ev or higher than Ec at the outermost surface layer of the film. This corresponds to the degeneracy of electron energy levels in the outermost surface layer of the film. If the passive films are sufficiently thin for electron tunnelling, the positive holes or negative electrons will accumulate at the densely degenerated energy levels in the outermost layer and contribute to the surface charge <~(ox) of the oxide films. The degeneracy of electron energy levels thus converts the outermost oxide surface layer from a semiconductor phase to a metal-like layer, which is capable of accommodating an excess charge to alter the Helmholtz layer potential ~(H). In the potential region of degeneracy, therefore, the Helmholtz layer potential is no longer constant but changes linearly with the electrode potential. The critical bias potential beyond which the film surface are degenerated is greater for n-type films than for p-type films in the anodic polarization and the reverse is realized in the cathodic polarization, as shown in Fig. i0. It Js worth noting that the degeneracy emerges when the Fermi level at the film surface is located not only in the conduction and valence bands but also at the dense surface states within the bandgap. It is evident that the critical bias potential is greater, when the band~ap is ~reater. For passive oxide film on iron with Eg = 1.6"-2.6 eV (]8), the film thickness at the anodic critical bias potential does not exceed 2 " ~ 3 n~ ( I nm/V) allowing the electron tunnelling to occur and hence the f~im surface can be degenerated. For anodic oxide films on aluminum, whose bandgap is greater than 6 eV, the film thickness exceeds the electron tunnelling distance before the anodic potential reaches the critical bias potential and hence the film surface can not be degenerated into a metal-like phase. Electrochemical reactions such as anodic oxygen evolution and cathodic hydrogen evolution frequently take place involving electron transfer on the passive oxide films. The electron transfer across passive oxide films takes pleace by three different mechanism, (i) the conduction band (Ca) mechanism, (ii) the valence band (VB) mechanism, and (iii) the direct and resonance tunnelling mechanism in which localized states within the bandgap play an important role. For instance, the CB mechanism may be applicable to the hydrogen electrode reaction and the VB mechanism to the oxygen reaction and oxidative oxide dissolution.

Overview

i

I 1

,

M.

OX.

EL.

I I

-2

6 o -4

,

tl...v4/ M.

OX.

i EL. Ec

~ ~ l . 3 e V 1.6eVEv

1/I /

o -4

Etp /

- -

1.3V

I I ~

Fe / 0.SM H;tSO4 I

O

Fig.ll

I

Oi.5 1.0 115 ELECTRODE POTENTIAL E (V. NHE)

2.0

Anodic polarization curves of iron and nickel in sulfuric acid solution and electron energy band structures of passive oxide films at the flatband potential Efb (21). Etp is the transpassivation potential.

There are a few other models apparently more sophisticated than a simple semiconducting model of passive films. A p-n junction model simply assumes a bilayered structure consisting of an n-type semiconducting layer on the metal side and a p-type layer on the electrolyte side, which may correspond to the bipolar fixed charge model mentioned above (Fig. 9). The p-n junction model further extends into a p-i-n junction model (ig), which assumes a three-layered structure with an intrinsic semiconducting layer between an inner n-type layer and an outer p-type layer. A chemlconductor model (20) similar to the p-i-n junction has assumed the presence of an insulating dielectric layer of stoichiometric oxide sandwiched between the semiconducting non-stoichiometric regions associated with excess metal ions or lower valent metal ions on the metal side and metal ion vacancies or high valent metal ions on the electrolyte side. Anodic dissolution of passivated metals takes place through the passive oxide films present on the metal surface, and its rate is apparently represented by the dissolution rate i d of passive oxide films, which is controlled by the Helmholtz layer potential ~(H) at the film/electrolyte interface. As long as ~ ( H ) is constant, id remains constant. At the steady state the film thickness remains constant and the dissolution proceeds by the metal ion transfer across the Helmholtz laver,

12

N. SATO

300

(~ ZJrI 0.IN Hz~04 ] = 8mA'¢m~

s .-,--4D

m--~ 200

>

0

/

~

__/

)

/

• :

electronic Ixeakdown

m : m.ch.cld br.mkdow~

TIME S

Fig.12

Growth curve (voltage vs. time) for anodic oxide films grown at constant current on zirconium in (I) IN HoSO A solution and (2) O.IN H~SO. solu~io~ (22). 4 e = onset of electronic-Dreakdown, m = onset of mechanical breakdown.

Mn+(oxide)

~

M ~+ aq

(5)

and its rate is a function of E~(H) independent of the electrode potential degenerate states of the passive oxide observations that the dissolution rate in acid solution is dependent on pH potential in the passive state.

and hence of electrolyte pH but in the potential region of nonfilm. This is in agreement with of passivated metals such as iron but independent of the electrode

Figure 11 shows the anodic polarization curves of iron and nickel in H SO. solution and the electron energy band diaErams of the passive oxide f~Im@ at the flatband potential (21). It appears that the passive film on iron, which is an n-type oxide, is electrochemically stable against anodic polarization from the flat band potential Ef1~, because of a relatively large energy Eap between the Fermi level E~ and the valence band edEe Ev within which the fi]m cannot be degenerated. The passive film on nickel, which is of p-type oxide with a narrow energy gap between E F and Ev, is relatively unstable against anodic polarization from E b and is subject to a potentlal-dependent transpassive dissolution in t~e potential region where the film is deEenerated. There is another mode of anodlc film dissolution in which the oxidation valency of metal ions or oxygen ions increases, i.e. the oxidative dissolution. The transpassive dissolution of Cr and Ti belonEs to this mode, Cr203 + 5H20 + 6h + = 2Cr~9~- + 1OH + (6) TiO 2 + 4h +

~

Ti 4+ + 02

(7)

Overview

FILM

METAL

i .2 _j

/

_l

13

ELECTROLYTE

A

ut

o -2 -3

VOLTAGE

Fig.13

Electron enerKy band diagram of anodic oxide formation, and qualitative variation of ionic current i. for oxide formation and i ip for anion- incorporation and i for eIectronlc avalanche current with anodlzatlon voltage at constant current (23).

where Cr(III) ions in the passive film on chromium are oxidized to soluble Cr(IV) ions, and oxygen ions in the passive film on titanium are oxidized to oxygen molecules, both leading to anodic destruction of the film. This mode of oxidative dissolution of passive films is thermodynamically possible when the Fermi level of passivated metals is lower than the electron level of the oxidative dissolution reaction. The reaction proceeds by either the valence band mechanism or the electron tunnelling mechanism.

4.

BRE~)O~N

OF PASSIVE OXIDE FILMS

Breakdown of passive oxide films occurs in the presence of electrolyte anions, particularly, aggressive anions such as halide anions. Three different modes of film breakdown may be distinguished; (i) mechanical breakdown, (2) electronic breakdown, and (3) ionic breakdown. The mechanical breakdown introduces microscopic cracks into the film, which is caused by large internal stresses associated with the film, e.g. a growing film(Zr0p 400nm thick) during anodic oxidation of zirconium in aqueous electrol~te at low anion concentration (10,22) (Fig. 12). The electronic breakdown, which is caused by an electronic avalanche current, occurs presumably only with thick oxide films. During anodic oxidation processes, the electrolyte anions enter the anodic oxide as donor impurities at a constant rate i and provides localized electron energy levels within the bandga p of tk~ oxide. Under the high anodic field, these impurity centers can be ionized by the Poole-Frenkel mechanism releasing a current of electrons into the oxide conduction band. These electrons, once in the conduction band, are accelerated by C$ 31-¢

14

N. SATO

METAL/AQUEOUS ELECTROLYTE

METAL ATCONSTANTPOTENTIAL

addRlanof hldlde anion

ELECTROOEPOTENTIAL E

Fig.14

7tt~t

Anodic polarization curve of passivable metals in the presence of halide anion showing the onset of pittinE at potential Epit and anodic current vs. time curve showinK the incubation time "Cpit for the onset of pitting at constant potential.

/ \0 / \

/o M

M

0

M

M

\ / \ / \ / 0

poly-

0 dentate

0 anion

CI

CI

Cl

Cl

mono-

dentate

/ ,/ M M , /,

Fig.15

L/

0 2"

CI

,/ , M M / ,/ Cl anion

Cl °

Poly-dentate anions forming a b r i d g e - O bonding and mono-dentate anions destroying a bridge-bondlng between the metal ions in metal oxides.

the anodic field leading to avalanche multiplication by an impact ionlgation mechanism (23) (Fig.13). The film thickness must be thick enough for these electrons to 6gain ~ufficlent enerEy for the impact ionization; a typical field 5xlO Vcmcombined with the mean free path of electrons i00 nm gives electrons 50 eV for I00 nm thickness. The

Overview

15

ratio of the electronic avalanche current i to the total anodic current i increases with the anodic bias potential~ as schematically shown in Fig. 13 (23). As a final step, the film breakdown will occur when the electronic avalanche current reaches a critical size beyond which the highly localized discharge results in a local destruction of the film. No electronic avalanche breakdown occurs in thin oxide films, which however, frequently suffer an ionic breakdown in the presence of a ~ r e s s i v e anions such as chlorides ions. The ionic film breakdown in aqueous electrolytes is usually followed by either repassivation or local pitting dissolution at the breakdown sites. In the anodic polarization curve of passivated metals the onset of pitting is characterized by a specific pit initiation potential, Epit, which usually decreases with increasing chloride anions (Fig. 14). The pitting is also characterized by a incubation time -~pit before its breakout at constant potential (Fig. 14). We note that the pitting is a localized metal dissolution which is an entirely different process from the preceding film breakdown. The ionic film breakdown starts with the competition for adsorption, at the film/electrolyte interface, between hydroxyl ions and aggressive anions such as chloride ions. The next process is somewhat controversial. The local adsorption of chloride ions may produce a spot of chloride salt or transitional chloride complex accelerating local film dissolution, or introduce metal ion vacancies which migrate to form a vacancy condensate at the metal/oxide interface, or induce a mechanical stress sufficient for microcrack formation to create embryos leading to the local film breakdown. It is worth noting that aggressive anions such as chloride ion are mono-dentate anions with only one covalent bonding coordinate and destroy the bridge-bonding formed by a poly-dentate oxygen ion between the metal ions in the oxides (Fig. 15). Whatever the mechanism of film breakdown, the final stage is to expose a small area of the fresh metal surface to the electrolyte. In most cases the breakdown preferentially occurs at surface defect sites such as nonmetallic inclusions where the film is defective. The anodic metal dissolution at the breakdown sites creates a local concentration of metal ion and chloride ion which migrates from the bulk electrolyte. This is a mass transport perturbation in the electrolyte diffuse layer near the metal surface. According to the linear stability theory, the perturbation will grow when the perturbation wave length ~ exceeds a marginal wave length ;~,*, the magnitude of which depends on the various conditions of metal-electrolyte systems. We may simply assume that the marginal wave length ~.* decreases with increasing anodlc potential (24,25), as shown in Figure 16. When the perturbation wave length7%, is smaller than ~.*, the local concentration of metal chloride decays and the breakdown sites will soon be repassivated by the local film formation. When the perturbation wave length 7k is greater than ~%.*, however, the local chloride concentration grows giving rise to a local acidification and hence preventing the repassivation of the breakdown sites. When the growing local metal chloride concentration exceeds a critical value characteristic for the metal-electrolyte systems, the pitting dissolution proceeds steadily (24,26). Apparently, the mass transport perturbation wave length 7%. is related to the size of film breakdown where the local metal dissolution takes place and hence to the size of surface defects, which is independent of anodic

16

N. SATO

2 MASS T R A N S P O R T PERTURBATION DUE TO PASSIVE FILM B R E A K D O W N Z uJ

Breakdown sites grow into pitting

u, N

0

Breakdown sites repassIvate E~ ELECTRODE POTENTIAL

Fig.16

E

Marginal perturbation wavelength ?t* in mass transport for the initiation of pitting at film breakdown sites, which is assumed to decrease with anodic potential, and surface defect size ~, which will determln the size of film breakdown and hence the mass transport perturbation wave l e n g t h ~ t a t the film breakdown sites (24).

0

-1

t S %% %

-2

6

-4

-5 -2 -8 0' 0.4 0.8 0 :8 o'. 2 ' ' ELECTROOE POTENTIAL E (V, NHE)

Fig.17

1.0

Critical film breakdown size (pit embryo size) ~ for the onset of steady pitting as a function of potential, and anodic current vs. potential curve for anodic dissolution current i and pitting current ipi t (27).

Overview

etch pit

17

polish pit

/2 ELECTRODE POTENTIAL

Fig.18

E

Two types of pitting; etch-pitting in the active state at low potentials and polish-pitting at high potentials where electropolishing takes place of the pit surface.

potential (Fig.16). From the potentlal-dependent ~.* and potentialindependent 2%, we can assume that there is a critical potential beyond which 7t is larger than 7t" (24). This critical potential corresponds to the pitting potential below which the film breakdown sites soon repassivate and beyond which a film breakdown leads to a steady pitting. If the size of film breakdown increases, the perturbation wave length will increase and then the critical pitting potential will decrease (Fi E . 16). Figure 17 shows the relationship between the film breakdown size (pit embryo size) and the critical pitting potential for a stainless steel in sulfuric acid solution containing chloride ion (24,27). It is apparent that the critical pitting potential becomes more positive as the film breakdown size is smaller. Configurationally, there are two types of pitting corrosion; one is the etch pit exposing crystal facets and the other the polish pit of semispherical shape (Fig. 18). The etch pitting proceeds when the pit surface is in the active state at low potentials, whereas the polish pitting proceeds when the electropolishing, which is a mode of the transpassive dissolution, takes place on the pit surface usually covered with a salt layer at high potentials. The film breakdown and the initiation of pitting corrosion are stochastic processes. Accordingly, the characteristic quantities such as the pitting potential and incubation time are observed to scatter in statistial fashions. The distribution of the pitting potential obeys the normal probability function (28) and the pit generation at constant potential may be regarded as a Markov process (29), as shown in Fig. 19. Stochastic birth and death processes are assumed to operate in the pitting corrosion of stainless steels. The pit generation rate is found to increase with anodic potential, while the pit repassivation rate is independent of potential (28).

18

N. SATO

!

18Cr S T E E L I 3 . 5 % NaCI o.~ 9 9 I

18Cr

J/

~ ro m 0n-

50

IO

,.Y

Jo

s:

.~

.)s

el I

0.4 PITTING

-'

18Cr - 2Mo

i

I

I

0.5

0.6

0 .7

POTENTIAL

Ep~ (V. NHE)

-1

n. o.

\ i l I 5 10 15 PiT G E N E R A T I O N TIME INTERVAL

-2

Fig.19

5.

o~ I 20 AT (re|n)

25

Cumulative probability P~ in normal probability scale for pitting potential Epit of 18 Cr stainless steels in 35% NaCI solution (28) and cumulative probability Pz for pit generation time interval 4 T at constant potential of 18 Cr-8 Ni stainless steel in 0.2M HaCI solution containing 0.1M Ha2S04 (29).

CONCLUDING

RENARKS

Phenomenologically, the anodic passivation of metals is the same as that of semiconductors. But the points of interest differ in these two groups of solids. The passivity of metals mostly concerns the corrosion protection and hence the interest is focused on the chemical stability of the passive films, whereas the electronic stability of the surface and hence the electronic e n e r ~ levels such as the surface states of the passivation films are of basic interest in the field of semiconductors. Fundamentally, however, the ionic and electronic stability of the passive films are equally important in improving the corrosion-resistivity and electrical durability of the materials. This review has dealt with only a few limited subjects. Among other subjects, the alloy passivity is of fundamental and practical interest, in which some alloying elements are enriched in the passive film. This is compared with the anodic passivation of compound semiconductors which produces a passivation film whose composition is frequently not proportional to that of the semiconductors.

Overview ~

~

19

S

i) J.Keir, Fhll. Trans., 80, 359(1790). 2) " P a a s l v l t y o f ~ t a l s " , Ed. Frankenthal and Kruger, The Electrochem. Soc. Inc., Princeton, New Jersey (1978). 3) "PMsivity of metals and Semiconductors", Ed. M. Froment, Elsevier, Amsterdam (1983). 4) N. Sato and G. Okamoto, "CA)Iprehemsive T r e a ~ s e o f E l e c t ~ s i ~ " Vol.4, pp. 193-245, Ed. Bockris, Conway, Yeager, and White, Plenum Pub. Corp. (1981). 5) N. Sato, K. Kudo, and T. Noda, Z. Phys. Chem., N.F. 98, 271(1975). 6) G.L. Griffin, J. ElectI~)chem. Soc., 131, 18(1984). 7) M. Sakashlta and N. Sato, COITosion, 35, 351(1979); Reference 2), pp. 379(1978). 8) R.L. Smith, J. Electrx3anal. Chem., 238, 103(1987); R.L. Smith, B. Kloeck, and S.D. Collins, J. Electzx)chem. ~ . , 135, 2001(1988). 9) H. Takahashl, F. Fujimoto, H. Konno, and M. N~ayama, J. E l e ¢ ~ h e m . Sot., 131, 1856(1984). I0) J.S.L. Leach and B.R. Pearson, Colnm0si~ Science, 28, 43(1988). ii) K. Shimizu, G.E. Thompson and G.C. Wood, T ~ n Solid Films, 77, 313(1981); 85, 53(1981). 12) T. Ohtsuka, M. Masuda, and N. Sato, J. Electrochem. Soc., 132, 787(1985). 13) F.P. Fehlner and N.F. Mott, Oxidation of Netals, 2, 59(1970) 14) T.L. Barr, J. Phys. Chem., 82, 1801(1978). 15) N. Sato and T. Ohtsuka, J. E l e c t I ~ e m . Soc., 125, 1735(1978). 16) R. Goetz, D.F. Mitchell, B. MacDougall, and M.J. Graham, J. Elec~hem. Sot., 134, 535(1987). 17) J.W. Schultze and L. Elfenthal, J. Electrcmna].. Chem., 204, 153(1986). 18) T. Ohtsuka, K. Azumi and N. Sato, J. I~*ysique, C A ) I I ~ e C10, 191(1983). 19) A. Pinkowski, Werksl~ffe umd Koz-mmsion, 37, 526(1986); Z. Phys. Chem., Leipzig, 266, 904(1985); 267, 718(1986). 20) C.T. Chen and B.D. Cahan, J. Electz~w~hem. Soc., 129, 17, 474, 921(1982). 21) N. Sato, J. ElectIx)chel. Soc., 129, 255(1982). 22) F.Di Quarto, S. Piazza, and C. Sunseri, J. Electzx)ehem. Sos., 131, 2901(1984). 23) J.M. Albella, I.Monters and J.M. Martinez-Duart, ElectI~w:hlm. Acta, 32, 255(1987). 24) N. Sato, Oo*~x)sion Science, 27, 421(1987). 25) T. Okada, Eleetroehlm. Acta, 33, 389(1988). 26) Y. Hisamatsu, "Passivity and Its Breskdown on Iron and I I ~ Base Alloys", p.99, NACE, Houston, (1976). 27) N. Sato, J. E l e c ~ h e m . Sot., 129, 260(1982). 28) T. Shibata, Trans. Iron Steel Inst. Jpn. 23, 785(1983). 29) N. Sato, J. Electrochem. Sos., 123, 1197(1976).