Wear mechanisms of anodes

Q

WEAR

MECHANISMS

lxw-l6s6,S9 53.ro+o.M ,989. Per&amon Press ptc.

OF ANODES

F. BECK University -GH- Duisburg, FB 6-Elektrochemie, Lotharstr. 1, D-4100 Duisburg 1, F.R.G. (Received 2 August 1988; in revised form 9 November

1988)

Abstract-In industrial anode processes, the starting materials dissolved in aqueous electrolytes are oxidized at inert anodes. In most cases, metal anodes with an oxidic surface layer are employed. According to their fabrication, chemical, thermal, anodic and ceramic oxide coatings can be distinguished. This allows a wide variation with respect to electrocatalytical behaviour. Passivation phenomena play a distinct role in the formation and stabilization of these electrodes as well as in their long term behaviour. Wear mechanisms of industrial anodes in terms of ion migration through the oxide layer, depassivation and transpassive dissolution are comprehensively discussed from this viewpoint.

1. INTRODUCTION The long term stability of electrodes for electrolysis and electrosynthesis is the most important property, which must be realized for any industrial application. Any electrode wear may lead to contaminated products, it changes the dimensions of the electrode construction and increases its resistance, and it causes additional material and labor costs due to the need for its periodical replacement. Metal electrodes are preferred due to their valuable electrical and mechanical properties. In the following, only metals will be discussed as electrode substrates, while alternative materials, namely graphite, carbon, carbonaceous materials and conducting polymers will be outside the scope of this presentation. As cathodes, they offer no large problems due to their inherent thermodynamic stability in the prevailing potential regions. However, the opposite is the case for metal anodes. Especially under polarization, they operate far away from thermodynamic equilibrium. It has been known for a long time that non-metallic coatings of the metals can provide a satisfying solution of the problem. The most simple case is a metal in its passive state. The passivity of metals is generally considered in connection with corrosion minimization for metallic materials[l]. The model of an oxidic surface film as the essential feature, going back to Schiinbein and Faraday[2], is widely accepted today. Some striking parallels can be found concerning passive metals and passive anodes. The chemical solubility of the passive oxide in a given medium must be low in order to prevent continuous dissolution and reformation of the passive oxide film. Non-porous layers provide much better protection than porous layers. Optimal match between metal base and oxide films is a necessary precondition for a good adherence. Figure 1 schematically represents the geometrical situation. The oxide film has no constant stoichiometry. A region rich in metal adjacent to the metal and rich in oxygen at the surface must be expected. Under current flow only electrons should be transferred at the phase boundary. The partial current densities for the metal and oxygen ions, constituting

irCe

ce

Me

Fig. 1. Layer model of a passivated metal anode.

the oxide lattice, should approach zero, which means:

. . le- %_h&=+,J02-.

(1)

If this condition is fulfilled, the electrochemical stability of the anode will be excellent. On the other hand, anode wear can be defined as the sum of ion transfer reactions (2a) and (2b): Me”+(ox) H20(aq)-2H*

+ Me’+(aq), (aq) + 0’~

(2a) (ox).

(2b)

According to Fig. 2, three mechanisms are possible: (A) metal ions cross the oxide film and enter through phase boundary 2 into the solution; (B) both metal ions and oxygen invade the oxide film, whose thickness will increase; (C) the oxide film, generated initially from the metal, dissolves “chemically”, the metal depassivates. The rate of these ionic processes can be determined by the rate of ion transport in the film or by the rate of ion transfer at the phase boundary 2. Both are activated processes with an electrochemical barrier. A quantitative treatment was given by Cabrera and Mott[3] and by Vetter and Gom[4], respectively. More recently, Kirchheim combined both models to a general treatment[S]. Passivated metals as anodes are fabricated by anodie “formation” (anodizing) of the metal substrate. Table 1 gives five examples for industrial electrolytic processes, where such types of anodes are employed (examples 6-10). Obviously, two other modes of 811

812

F. BECK

I

A

L>+- cM”

(OX)

--I

-+-Me’+

0--

I 2e‘

2

M&

Me” 0--

Moe f

'X

sol

Fig. 2. Schematic representation of anodic wear of Me/Ox-composites via ion and electron transport in an oxide layer at a metal electrode (inference (1)) and ion transfer reactions at the interphase oxide-lectrolyte (2): (A) anodic metal ion transfer without thickness change of oxide; (B) anodic thickness growth of oxide; (C) “chemical” dissolution of oxide.

Table

no

1. Oxide

anodes

for electrochemical

Process

processes fabrication

type

l-3; 4, 5 and 610 (cf text)

Anode

constitute

process

three modes

Oxide anode

2 3

Brine electrolysis Chlorate electrosynthesis Metal winning (low pH) Bromide electrolysis

Cl - -Kl, Cl - +H,O+CIO, H,O+O, Br-+1/2 Br,

4 5

Perchlorate electrosynthesis Ozone electrosynthesis

Cl - + H,O-tClO, H20-+OJ

Ti/PbO,, C/PbO,

y-MnO,-electrodeposition Chromic acid regeneration Peroxydisulfuric acid/H,O, Water electrolysis (high pH) Adiponitrile process (Monsanto)

Mn++MnO, Cr”+-+H,Cr04 H,SO,-+H$,O, H,O+O, H,O+O,

Ti/TiO,, Pb/PbO, Pb/PbO, Pt/PtO, Ni/NiO,, Ti/Co,O, Fe/FeOOH

1

6 7 8 9 10

fabrication

a dominant

of anode

Ti/RuO,, Ti(Pd)/RuO,

TiO,,

SnO,

+ IrO,,

Ta/RuO,, TiO, Ti/PbO,,

TiO,

Pb/PbO,

deliver today anode materials, which play role in large scale processes.

I. Ceramic oxide layers (examples i-3). Titanium was used in most cases as a substrate. After the introduction by Henry Beer (1966), cf.[6, 71, these “activated titanium anodes” were revolutionary in the chloralkali industry, and within 15 years nearly the total of graphite anodes, traditionally employed in industrial chlorine production, were substituted by these “DSAs”. II. Electrodeposited oxide layers (examples 4 and 5). If titanium is used as a base material, total passivation and activation in the course of anodic deposition must be prevented. Several strategies have been worked out,

QW-

In contrast to passivated metals, these composite electrodes normally bear non-identical metal components in the oxide layer A and in the base metal B, cf: Fig. 3. In addition, an interlayer B’ and a surface layer A’, originating from the base and/or oxide layer and from the electrochemical interaction with the electrolyte, respectively, must be considered. A high electronic conductivity of the oxide layer is necessary to avoid excessive Ohmic voltage drops in the film. For very thin layers, the precondition is not

44-

z0

Fig. 3. Layer

model

of an oxide-composite TiO, + RuO,).

so severe, as it can be easily judged AU = jpd.

anode

(eg Ti/

from Ohm’s law: (3)

For d=l nm and j=O.l Acm-‘, AU=1 mV corresponds to a resistivity of p = lo5 Rem. On the other hand, electronic conductivity in case of a protecting layer is not important at all; corrosion protection should be even improved in the case of insulating films. For industrial anodes, the necessary level of specific conductivity depends on the thickness of the oxide layer. Most oxides of technical relevance have a metal like conductivity (PbO,, RuCi,). But there are

813

Wear mechanisms of anodes

even means for the application of oxides of poor as it will be shown conductivity (TiOz, Cr,O,); later on. The variation in stoichiometry and oxidation state of a metal oxide of the general formula MeO,H, provides unique means for: -matching the adequate composition for a good adherence at the phase boundary to the metal; -constituting active catalytic centers at the surface of the anode; -providing redox centers of an adequate redox potential to realize heterogeneous redox catalysis, cf: Section 4. In this way, an oxide anode is much more flexible than a metal electrode. Only reductants can be oxidized at blank metal anodes, eg Cr2+ at mercury or gold anodes or H,,at platinum, but this more academic case will be neglected in the following. The oxides are generally brittle materials of poor mechanical properties. Pure bulk material is to be excluded from these grounds. The thin oxide layer on top of a metal carrier helps greatly to overcome this problem. Last not least, economic limitations must be considered. Here again, the composite electrode greatly facilitates the introduction of even expensive electrode materials. Table 2 demonstrates this with a few examples. In the following, wear mechanisms of metal/oxide anodes will be discussed in detail. Experience from passivity of metals will be involved for the interpretation of the results. The electrolytes will be normally aqueous. They are compatible with the components of an oxide anode. It is believed that even non-oxidic materials as nitrides, carbides, silicides etc[9] undergo a formation in aqueous electrolytes, leading finally to an oxidic surface layer A’.

2. CHEMICAL WEAR QF CERAMIC OXIDE LAYERS The great importance of activated titanium anodes in modem brine electrolysis justifies starting with this category. The fabrication mode resembles CVD. Metal chlorides in aqueous or alcoholic solution are “painted” or “brushed” onto the etched titanium substrate, dried and fired in the air (400-600”C). A

Table 2. Area specific material costs for various metal/ oxide anodes. Prices per kilogram: Pt = 18,200 $ kg-‘; Ru -2,lOOS kg-‘; Ag=203$ kg-‘; Ti=35 % kg-’ (as at 31 October 1988) Anode 1. 2. 3. 4. 5. 6.

Pt-sheet, 25 pm Ti (1 mm)/Pt (2.5 pm) Ti (1 mm)/PtO,, 6 g Pt me2 Ti (1 mm)/RuO,, 6 g Ru me2 Pb(2mm), 1% Ag Pb(Zmm),6%Sb

costs, $ m - = 11,200 1,300 300 175 120 loo

thermal decomposition, eg: RuCl;+02-+RuOz+3/2Cl;,

(4)

yields the oxide layer. Reactive sputtering of metal powders in the presence of oxygen or plasma spraying of oxide powders or sintering of emails are alternatives. A RuO,/TiO, layer has a high electronic conductivity. RuO, is a metal-like conducting oxide (due to the weak bonding of electrons by the oxygen, which constitutes the valence band), while TiO, is a poorly conducting n-semiconductor (due to the heavy bonding of electrons by the oxygen in this case), cf[lO]. In spite of the identical crystal lattice @utile) with nearly coinciding cation radii (64 pm), both oxides form an oxide mixture rather than a mixed oxide[ll]. Nevertheless, the black oxide layer adheres perfectly, it is chemically and electrochemically stable, and in addition it constitutes an excellent electrocatalyst for the chlorine electrode. The RuO, “floods” the TiO? of the valve metal Ti with electrons, thus overcoming its rectifying properties upon anodic polarization. Such ceramic layers of ruthenium dioxide applied to Ti sheet electrode, constitute an excellent chlorine anode under brine electrolysis conditions. Layers containing 6 g Ru mm2 and stabilized by TiO,, Sn02[12] and others yield a service life of 10 years in diaphragm cells at 0.4 Acme2. This means a specific charge of about 35 000 Ah cm-‘. The corrosion current density is as low as 7 x 10T9 A cme2, and Equation (1) is absolutely fulfilled. The RuO,-surface layer changes its oxidation state upon cyclic polarization. The CV-curves (in acid) are relatively featureless[ 13-171. A small oxidation peak immediately prior to the anodic rise is observed (U,= 1.3 V- at pH 0), which is more pronounced in 1 M NaOHr161. The redox nrocess is accomoanied bv proton tra&p&t[l 3, 151. Ii must be concluded that anodic chlorine evolution at this electrode is an electrochemical step.: RuOOH e----RuO,

+H+ +e-

(5)

in conjugation with a chemical step, generating the product and regenerating the tri-valent state. c

Ru02+Cl-+H+--+

RuOOH + l/2 Cl,

(6)

Tafel slopes obtained by steady-state current-voltage curves are also interpreted by such a mechanism[ 181, which we call heterogeneous redox catalysis[19], and which will be further discussed in Section 4. Both oxidation states, RuOl and RuOOH or Ru,O,,, which are present at the surface of an operating anode, represent extremely insoluble species. As Ru02 is an excellent catalyst, the chemical step EEquation (611 must be fast, leading to a relatively low RuO,-surface concentration. The wear mechanism seems to be a very slow chemiqal dissolution of active material under anodic conditions, eg under the influence of active chlorine. The microkinetics is of electrochemical nature, cj Fig. 2c. Practical long term experiments with periodical crystallographic micrographs of cross sections of the DSA exhibit a continuous consumption of the

F. BECK

814

active layer[20]. This is confirmed by Ru-determination in the electrolyte. We applied defined surface concentrations for RuO, on a Ti sheet, varying in the range oftc,=0.5 mg Ru m-* to 50 mg Ru m-*[21]. The service life in terms of Qa = jr as a chlorine anode increased with increasing cA according to a relationship: Qa = Kck=.

(7)

An extrapolation to C~ = 6 g Ru (as RuO,) per m’, which is standard for industrial chlorine anodes, leads to Q,=7 x lo6 Ccm-’ or 2000 Ah cm- 2, in good accordance with the results in Fig. 13. They are inferior to industrial DSAs due to the absence of other stabilizing binary and ternary oxide components.

3. WEAR OF PASSIVE METAL ANODES BY ION MIGRATION THROUGH THE OXIDE tiiiM The mechanisms of ion migration through the oxide lilm were already discussed in Section 1 and Figs 2a and b. In case of ion transport in the film to be the rate determining step, the chemical energy barrier between the stable positions in the oxide lattice is modified by an electric flow field vector E, which is introduced by the flowing current density j or by the application of a voltage AU at the layer thickness d[3]:

(p=resistivity of the material). Equation (8) corresponds to Equation (3). The electrochemical energy barrier becomes asymmetrical and ion migration starts in field direction. Obviously, high fields are only possible at passive layers with a relatively high resistivity. The present view on this problem tends to relatively disordered, nearly amorphous oxide films with a medium resistivity[22]. The charge transfer reaction at the oxide-electrolyte interface itself may become rate determining. In this case, the chemical energy barrier between the two phases is modified by an electric field vector E’, which is established by the Galvani voltage Acp across the thickness of the Helmholtz layer d,: (9) In the case of iron(III)oxide (y-Fe,O,), which makes up the passivating layer on iron, the transport of the cations through the film plays the decisive role. It is known from early measurements[23] that the corrosion current density j,,,, in the passive region is independent on the potential up to the beginning oxygen evolution. The oxide layer thickness increases in the passive region from 2 to 5 nm, cJ[24]. With increasing pH (from 0 to lo), j,,,, decreases from 10 to 0.01 ~Acm-*. However, it increases again in strong alkaline solution due to the transpassive dissolution of iron as ferrate(VI). Presumably, j,,,, increases under all conditions in the region of oxygen evolution. However, corrosion of an iron-oxygen anode in slightly alkaline solution is so small that it could be used under these conditions.

Recently, carbon steel is also used in a large scale electroorganic process, the adiponitrile process of Monsanto Corp. The aqueous electrolyte in the undivided cell has a neutral pH value, containing 10% Na2HP0,, 2% Na,B,O,, 0.5% Na, EDTA, 0.4% quaternary ammonium salt (for the cathode reaction) and an emulsion of the organic reactants[25]. Some proprietary additives depress further the steady-state corrosion of this cheap Fe/y-Fe,O, anode -to an acceptable level. Fe3+ ions migrate through the film and the iron base is consumed. But the rate determining step must be located at the oxide-electrolyte phase boundary, and it is greatly influenced by the composition of the electrolyte. This point will be further discussed in Section 5. The oxygen overvoltage of the anode is nearly as low as with Ti/RuO,[26], and this helps greatly to circumvent the anodic oxidation of the organic materials. It should be mentioned that the iron sheet electrodes are bipolar and that the cathodic side is plated galvanically with a 1OOpm Cd-layer. Service life is in the order of one year. The depletion of 1 mm iron after that time corresponds to an average corrosion current density for Fe3+ of 0.1 mA cm-‘, making up 0.1-0.01 % of the total current. Up to this point, the anode wear mechanism for two large scale inorganic and organic processes was discussed in detail. The anode materials, Ti/RuO, and Fe/y-Fe,O,, play a relatively exclusive role alone for these electrochemical processes. This is not the case for another anode type, namely the passivated lead anode in dilute sulfuric acid, which is a lead dioxide/lead composite. This anode has been very well known for a long time. No large scale, but many small scale applications are known[27]. The anode is not absolutely stable due to the porous nature of the PbO, layer. But even in case of a non-porous coherent PbO, layer, the migration of oxygen anions through the PbO, lattice, which has an oxygen deficiency of about 1%, must be allowed for. This mechanism was recently investigated by Pohl et aZ.[28, 29). It was found that 02- migration is the rate determining step and that the two phase boundary reactions are in equilibrium. A parabolic law of thickness growth:

dL

k

dt

L'

-=-

was found with the help of chronoamperometric curves (Aq = 0.2 V). Figure 4 gives an example. This confirms the model of Fig. 2b. Upon anodic polarization to perform oxygen evolution or electroorganic oxidation at rates up to several 100 mA cm-‘, the electronic current through the layer is complemented by a continuously flowing small oxygen ion current j- through the oxide film, which consumes the lead meta1[30]. It makes up about O.Ol-O.l% of the total current, c$ Fig. 4. It is clear from this special situation that adherence of the oxide will fail. In practice, a continuous shedding of powdery PbO, material is found at Pb/PbO, synthesis anodes. Some industrial developments to overcome these problems have been started in the last years. The Ti/PbO, anode avoids the problem of Pb corrosion, but a rectifying TiO, interlayer is growing in the course of time[31, 321. This is not surprising at all in

Wear mechanisms of anodes

0

1

I

I

40

60 ++P,,0-3

I

I

120

160

I

*-l/2

Fig. 4. Anodic formation of a PbO, layer on Pb in 4.5 M H,SO, after previous anodic polarization (AU =q=O.2 V). Current density i us c-“~. From the slope, D= 5.6 x lo- I6 cm’ s- 1 is derived. After[28]. the light of t’he above mentioned oxygen ion migration mechanism. Even during storage in the absence of any electrolyte, an insulating interlayer of PbO can develop in the course of a solid state reaction: PbO,

+Ti+PbO+TiO.

(11)

This resembles the so called thermal passivation of the Pb/PbOl electrode for the lead acid accumulator. It was tried to improve the Ti/PbO, anode by the application of a modified TiO, layer on Ti prior to the PbOl electrodeposition. A very thin platinum or palladium deposit was useful. Doping with oxides of the transition metals[33] was another successful method. In the case of RuO,/TiO,, a low concentration of Ru must be applied, otherwise only oxygen evolution is observed instead of PbO,-deposition from acid Pb(II) salt electrolytes[34]. PbO, has been also electrodeposited onto graphite rods to form a C/PbO, composite electrode. Again, oxygen ions will penetrate into the oxide layer and may attack the surface of the graphite. This will hamper the adhesion between graphite and Pb02.

1(a)

815

Lead dioxide is a strong oxidant. So it is not a wonder that in some cases, anodic oxidation is performed via heterogeneous redox catalysis. Cyclohexanonoxime is an example. PbO, oxidizes the organic compound. The lower valent oxide is reoxidized electrochemically. Cyclic current voltage curves for a Pb/PbO, anode in 1 M H,SO, with increasing concentrations of cyclohexanonoxime are shown in Fig. 5[35]. The increasing current amplification in the anodic branch clearly indicates redox catalysis. The oxidation potential coincides with the formation potential of PbO,. The re-reduction peak of the PbO, layer decreases along the same way due to the consumption of the oxide layer by the chemical step. However, surprisingly enough, in many other cases as for the anodic oxidation of alcohols[35], PbO, behaves as an inert electrode material, and the oxidation potential is well above the formation potential of the lead dioxide. The rate of chemical oxidation of alcohols by PbO, under these conditions is negligibly small. .

4. ANODE DISSOLUTION

WEAR BY TRANSPASSIVE OF CERAMIC OXIDE LAYERS

We have fabricated activated Ti anodes with a ceramic chromium(III)oxide layer, mostly with additional TiO,: Ti/Cr,O,, TiO,. The conditions were quite in analogy to the Beer anode Ti/RuO,, TiO,. However, firing in air was at 650°C rather than 450°C to meet the higher decomposition temperature of CrCl,[19,36]. The porous layers had a typical thickness of 1 pm. Figure 6 shows a slow cyclic current voltage curve in 1 M H,SO, as a full line. The current rises at very positive potentials, which is due to the formation of CrO, (index s means surface bound species): (Cr,O,),

+ 3H,O

-2(Cr0,),+6H+ ‘-

+6e-.

1(b)

Fig. 5. Cyclic current voltage curves at a Pb/PbO, electrode with 10 mVs_’ scan rate at 25°C. Seventh cycles are shown. The composition of the magnetically stirred electrolytes were: (a) 1 M H,SO,; (b) 1 M H,SO,-0.03 M cyclohexanone oxime; (c) 1 M H,SO,-O.l M cyclohexanone oxime; (d) 1 M H,SO,-O.3 M cyclohexanone oxime. After[35].

(12)

F. BECK

816

1: ; :/

The chemical follow up reactions lead to the following rate equation, assuming reaction orders of unity in the rate determining step:

1 M IP I 1 M HzSOk

j = zF

I !

i ' '

60 t

:’

t

: : ,

CO1(kc,,,+ k,),

j = zFk,[A] If k, & (k, + kc,.,), the overvoltage:

1

l + k,/(k, + kc,>,)

the current rises exponentially j=zFk,[A].

Fig. 6. Current voltage curved for standard Ti/Cr,O,, TiO,basic curve, 1 M H,SO,;--anode, v.=5mVs-‘: curve in the presence of an oxidizable substrate, 1 M H,SO,. After[19].

The in situ anode product dissolves under deprotonation: (CrO,),

+ H,O -

L

will be hydrated

and

+H+.

(13)

3(CH&CO

+ 3H20.

(14)

A large current amplification is observed in Fig. 6, which is again indicative for a heterogeneous redox catalysis [Equations (12) and (14)]. Thus, both chemical steps compete for the electrochemically generated Crop, but only one closes the catalytic cycle. In the steady state, a balance at the branching point gives[19, 371: v,=vc+v,,

(15)

The current density is thus given by: j = zFv, = zFk,([A]

- CO]),

(16)

where [A] is the sum of both surface concentrations of the reduced [R] and oxidized [0] form of the chromium oxide:

CA1= L-RI+ WI, and k, is the electrochemical rate constant usual dependency on overvoltage ‘I: k,= k,.,exp(azFq/RT).

(17) with the (18)

(20) with (21)

On the other hand, for high overvoltages, k,p(k, + kc, J, and a mixed reaction limiting current density is dekved from the general Equation (20): jlim =

zFCAl(k,+ kd.

(22)

Under these conditions, the ratio of the overall current density j to the wear current density j, is given by Equation (23), assuming here a reaction order of Y with respect to the substrate: j/L = (k, f k&)/k,

HCrO,

The peak type current-voltage curve is due to the limited chromium oxide inventory in the thin ceramic layer. On scanning back the potential, no reducible CrO, species can be detected. If this measurement is repeated in the presence of isopropanol (dotted curve), a chemical reaction with (CrO,). will compete with reaction (13) under generation of acetone and regeneration of Cr(III), [or Cr(IV)] species at the surface, which enter again into the electrochemical step [Equation (12)]: L 2(CrO,), + 3(CH,),CHOH (Cr,O,),+

(19)

where c,., is the concentration of the organic substrate immediately at the surface of the electrode. Experimentally, v=O.4 was found for isopropanol[37]. A combination of Equations (16) and (19) leads to a general kinetic expression:

= I+ c,,, k/k,.

(23)

The Cr,O, inventory at the anode corresponds to an area-specific charge Q*,o and we obtain another expression for jw: Q*.o =j&

(24)

where z is the service life of the anode. A combination of the last two equations leads to:

The factor in brackets gives the turn over factor of the oxide layer. If k-0, the layer dissolves quantitatively. If kp k,, the oxide layer is regenerated again and again. In case of isopropanol, turn over factors of up to 400 were realized. A strong stabilization of the Cr,O,/TiO, oxide layer by the ternary component Sb,O, was possibleC19, 381. Galvanostatic experiments show a distinct rise of electrode potential after a time z[19]. However, the simple z/j-relationship following Equation (25), which is independent from Y, is not observed. We find: z=Kj-"

(26)

with 1= 1.27 . . . 1.43 (instead of 1)[19]. This looked like a passivation by a rectifying TiO, interlayer at the end of service life, cf: Section 7. However, Cr,03 anodes with Pt or Au as alternative substrates showed essentially the same behaviour[19], cf: Fig. 7. From this we draw the conclusion that it is the consumption of the Cr,O, anode itself, which limits finally the service life. A further discussion will be found in[19, 37, 381. A potentiodynamic measurement of the basic curve as in Fig. 6 reveals that the peak current density increases with increasing voltage scan rate u,. However, conversion of the layer dropped down from 90% to only 10% at the highest scan rates[37]. This

Wear mechanisms of anodes

817

I Q,I

log j /mAcm-* Q”‘dl

1

C cm-’

1/ ,

$0

Fig. 7. Double logarithmic plot of life time z us current densityj for galvanostatic test of standard Cr,Os electrodes on alternative substrates 0-O Au (sheet) (A= - 1.32); O-O Pt (mesh) (,I=; - 1.37); x-x Ti (sheet, for comparison) (A= - 1.37). Electrolyte: 1 M H,SO,, 1 M isopropanol. After[l9].

0

1

, 2

rv:r1’7/cyi’l-1’7 3

4

Fig. 8. Potentiodynamic anodic dissolution of a Ti/Cr,O,, TiO, electrode in 1 M H,SO,. Plot of charge Q,, vs l/(v,)‘/‘. QA,o= 0.602 C cm -’ (z = 3). After[37].

interesting behaviour could be explained quantitatively in terms of the formatiofi of an insulating 00, layer on the Cr,O,. A rapid formation of CrO, leads to its accumulation at the surface rather than to its consumption by the two chemical follow up reactions. A space charge is established, which absorbs a large amount of the applied voltage. A relationship:

Qa=Kv,'/',

(27)

predicted by our model, was verified experimentally, CJ Fig. 8. From the slope K, a mobility of the charge carriers as low as 6 x low4 ~rn~V-‘s-~ is derived. Thus ions (protons 7) rather than electrons are involved in the process. It should be mentioned that all oxidation processes at a Ti/Cr,O, anode proceed in full analogy to the Westheimer mechanism for the chemical oxidation with chromic acid in solution[19, 391. Recently, we found an analogous wear mechanism at oxide cathodes from the type Ti/TiO,[40] or Ti/VO,[41]. Figure 9 demonstrates that low valency states are reversibly generated at a cathodically polarized Ti/TiO, electrode, cJ[42], quite contrary to thermally or anodically grown TiOz layers. The TiOOH state enters into a catalytic reduction of nitroaromatics[40,42]. The electrode was stable, turn over factors of more than lo6 were observed. We tried to do the same for the TiO state in connection with the reduction of carbonyl compounds[43]. However, in this case the electrode corroded, presumably uiu dissolution: TiO+2Ht-kTiZ++H20,

(28)

which is the cathodic analogy to the transpassive dissolution. A complication of the mechanism is due

Fig. 9. Cyclic voltammetric curves at a ceramic Ti/TiO, cathode (oia Ti (OBu),), 1 mol TiO, cm-‘, in 1 M H,SO,, v.=SO mV s-l. Negative end potentials: cycles l-5: U.- 1.50 v; cycles 6-10: u,= - 1.65 v. The c’urve at the bottom is for comparison with a thermal TiO, layer on Ti (400°C, air) v,= 50 mVs_‘. After[42].

to rapid homogeneous decomposition acid[46]: TiZ++H+LTi3++1/2H2.

of TP+

in (29)

F. BECK

818

If the Ti/ RuO, (TiO,) anode is operated as an oxygen anode, eg in 1 M H,SO,, a heavy corrosion is observed and the active layer is lost in relative short periods, A reaction path in terms of heterogeneous redox catalysis is proposed[45]. The electrochemical step is: RuO,+~H,O+RUO,+~H++~~-, while the chemical

(30)

log j ImAcme2

20 Y

60 “C

30

10

step is:

RuO,-+RuO,+O,.

(31)

As RuO, is volatile, it leaves the electrode surface by an extent which is described by the kinetic parameters as above. The oxide layer can be efficiently stabilized by the introduction of IrO,, which lowers the corrosion rate of RuO, by nearly two orders of magnitude[46].

3

1 OS

0.5

5. ANODE WEAR BY TRANSPASSIVE DISSOLUTION OF PASSIVE FILMS

Fig. 10. Semilogarithmic

The redox-catalytic mechanism for the oxygen evolution of RuO,-oxygen anodes, discussed at the end of the previous section, was already put forward for Fe, Co and Ni (in alkali) in the 192Os[47, 483. An iron anode Fe/Fe,O, is oxidized to the hexavalent state: Fez0,+3Hz0+2Fe03+6H++6t~. It decomposes

in a chemical FeO,*L

(32)

step to yield oxygen:

FeO,

+ l/20,.

(33)

In strongly alkaline solution, a complexation to yield the ferrate(V1) ion is proceeding in addition and iron dissolves transpassively[49]: k2 (FeO,),+20H--FeO:-+H,O.

(34)

The wear rate again is determined by the ratio k,/kz. It was possible to measure the relative rates at the rotating iron disc/platinum ring electrode[50]. Partial current voltage curves have been derived from this, cf: Fig. 10. They show that the oxygen evolution is favoured at high current densities and at low temperatures. Both reactions have a rate constant in the order

Cr- Metal base

Fig. 11. Kinetic

,

I

0.6 plot

of

0,7 steady-state

current voltage curves (Z) and of partial current curves for 0, and Fe(W) for a Fe (99.5%) anode NaOH at 20 and fiO”C. After[SO].

over

all

voltage in 50%

of 1 SK’. Homogeneous decomposition of ferrate(VI), yielding ferrate(II1) and O,, is slower by a factor of 10’ (at 25°C). By this way, ferrate(V1) can be prepared through electrosynthesis with medium to high current efficiencies. In slightly alkaline, neutral or acid solutions, where Fe(V1) was also detected in the passive layer by the chromate method[51], the rate constant of reaction (33) seems to be much higher, and the wear mechanism oia Equation (34) seems to be negligible in comparison to that already discussed in Section 3. On the other hand, Ni/NiO, or Ni/Co,O, anodes, developed as low overvoltage oxygen anodes in alkaline water electrolysis[52], may also be deactivated by a slow transpassive dissolution of the oxide layer. We tried to employ the Cr,O, layer on passive chromium or FejCr alloys for a heterogeneous redox catalysis as outlined in Section 4[53]. However, this attempt failed totally. The reaction scheme in Fig. 11 for a Cr/Cr,Os anode could not be verified in the

Electrode surface

scheme for the transpassive anodic dissolution of chromium (alloys) in the presence isopropanol. Index s depicts surface bound species. After[53].

of

Wear mechanisms of anodes

819

120

oL-----2

4

IO

20

40

ICCJ

j/mAmi*

Fig. 12. Weight loss per cm* (Am,) after galvanostatic passage~ofQ,,=626.4Ccm-zforV2Aelectrodesin(0, A, 0) 1M H,SO,; (0, A, n ) 1 M H,SO.,/I M isopropanol. The current density was varied in the range of l-100 mAcmm2.

Dashed line:quantitativedissolutionCr(0) + C!r(VI).

IO'

sense of a regenerative cycle. We observed for galvanostatic polarization of, for instance, stainless steel (Fig. 12), a more or less quantitative transpassive dissolution of the Cr as Cr(VI). The results were essentially the same for 1 M H,SO, in the presence or in the absence of isopropanol. The mass losses m,, for low current densities were above the theoretical value for Cr-rCr(VI), which is due to the participation of iron dissolution. With chromium foil, the theoretical value for m, was measured at all current densities. These interesting findings seem to indicate that the transpassive dissolution of chromium does not go through the intermediate Cr,O, in this case. It starts with the metal directly. The thermodynamic driving force for: Cr+4H20-rHCr0~+7H++6e-,

(35)

U,,, = + 0.30 V (us she) is much stronger as for: dr,O, + H,O-r2HCrO,

+8H+ +6e-,

(36)

where &+, is + 1.18 V, very close to the potential of transpasstve dissolution, Uu= 1.25 V (working potential at 1 mAcm- ‘). At a Ti/Cr,O, anode, the working potential was about 1.7 V. 6. UNDOPING

OF A PASSIVE INTERLAYER

According to Fig. 3, an interlayer B’ between the base B and the ceramic oxide layer A must be assumed for ceramic or galvanic oxide-composite anodes. As valve metals (Ti, Ta, Zr _. . ) are normally used for the substrate B, the anodically polarized interlayer B’, which is a n-type semiconductor, will act as a rectifier at backward bias. Ti/RuO, or Ti/Cr,O, should be a sandwich Ti/TiO,/RuO, or Ti/TiO,/Cr,O,. However, in the normal case no problem arises due to the fact that the interlayer will be heavily doped with the material providing the active layer. To gain further insight into the mechanism of anode wear under these conditions, tests at extremely high current densities up to 100 A cm-* were performed[21]. The time r until breakdown of the electrode was appreciably lowered by this way. We found that Q,, was not constant, as

20

io

Bo

io

Fig. 13. Service life T of a Ti/RuO, anode at high current densities j. Semilogarithmic plot. After[21].

expected. It decreased strongly from 2000 Ah cm -’ at 1 Acn-* to 3Ahcrn2 at 100Acm-2. The semilogarithmic plot in Fig. 13 shows an exponential decay of r withj. This behaviour can be rationalized with the migration of Ru*+ ions out of the layer due to the high flow field E, defined in Equation (8). Finally, the undoping process is completed, and the TiO,-barrier layer is reestablished. A combination with the exponential equation for a field enhanced migration of cations in an ionic crystal[3]: j+ =z+FcZv,exp

(37)

(where e =ion concentration, 1=jump width, vu = thermal frequency and E=electric field strength) leads to the following relationship: nA logs=log~-j.

D

0.432,Fl 2RTic

(38)

An evaluation of the slope of the experimental curve in Fig. 13 gives K= 10m4 S cm-‘. From this we conclude that the interlayer is a RuO,-doped TiO,, indeed, whose specific conductivity is in between the metal like conductor RuO, (K= 1O+4Scm-‘) and the nsemiconductor TiO, (K= lo-‘* Scm-I). This mechanism is a special case of anode wear for the Ti/RuO,-anode, as discussed in Section 2.

7. ANODE WEAR BY DEPASSIVATION THE BASE METAL

OF

Industrial anodes from the type Ti/RuO, or Pb/PbOs carry a porous active metal oxide layer. Thus the metal base electrode is not fully protected by it. It is a thin, coherent interlayer of metal oxide, which acts as a protecting film.

F. BECK

820

If the anolyte contains species, which attack the passive layer, a very dangerous situation is established. The species may cross the active layer through pores or flaws, and they finally may destroy the passive interlayer. The base metal becomes active and dissolves. Often this corrosion is localized as pitting corrosion. Adherence of active layer is badly damaged. Ti/RuO, is an anode, which is easily destroyed under the following conditions. --In the presence of fluoride. ---In the presence of even traces of bromide, leading to heavy pitting corrosion[54]. In this case, Ta is a suitable anode base metal, CJ Table 1. ---In the presence of carboxylic acids. These may be generated by anodic oxidation of organic precursors. -In the presence of methanol[55]. However, the addition of 10% water inhibits this anodic attack totally. From Pb/PbO, as a classical anode, which works optimally in diluted sulfuric acid, it is known that it breaks down under the following conditions. -In the presence of nitrate[56], chloride in acid solutions. -In the presence of nitrite.

perchlorate

or

The Pt/PtO, anode and anodes made of the other platinum group metals appreciably depassivate in electrolytes with a high concentration of Cl-, CN- or S2 - [8, 633. It is clear from these considerations that the composition of the anolyte feed must be thoroughly controlled in order to realize long service life of a given composite material.

8. FORMATION OF PASSLVATING (1NTER)LAYERS This behaviour is of special importance for active metal electrodes, which are used in primary and secondary batteries. On discharge, passivation would shift the potential of the negative so much into the positive direction that cell voltage would break down. As passivation needs a critical current density, a common strategy is to develop highly porous battery electrodes. This is the case for instance for Pb/PbO, in sulfuric acid. However, even for this electrode in lead dissolving electrolytes such as HBF, or HClO,, the discharge curves exhibit a decreasing capacity with increasing discharge current density, as it is shown in[57]. This is quantitatively explained in terms of a semiconducting PbO, layer, which develops on top of the dissolving PbO,. A space charge is established in this layer. When it has grown to a sufficient thickness, voltage drop is so large in this layer that cell voltage breaks down. The exponent 1= 1.33, found experimentally for the discharge time/discharge current density relationship as in Equation (26) was verified theoretically by our model. Zinc electrodes (Zn/ZnO) in KOH are very interesting systems. Once again, a passivation is found, which increases with increasing current densities. In _ this case, the product j,/r was constant, showing a Sand type relationship[58]. We also found a strong passivation of zinc in 10 M H,SO,. Figure 14 shows some galvanostatic cycling curves with thin zinc layers on a glassy carbon base. By changing the electrolyte, eg using amine salt hydrogen sulfate, passivation was circumvented, and cycling of the electrode was pos-

,

t/h

6.5

lb

lb

t/h

,

2’0

Fig. 14. Galvanostatic alternating polarization of a Zn layer on glassy carbon, 1 mA cm-*: (a) 10 M H,SO,, ZnSO, sat., dZn- 11.3 pm; (b) 1 mol morpholinium hydrogensulphate/3 mol H,O, ZnSO, sat., d,, = 37.3 pm. After[59].

Wear mechanisms

of anodes

sible with a minimum of corrosion[59]. Similar experience was found with lithium electrodes, eg Li/LiCl. The porous salt layer may become too densely, leading to the so called voltage delay[60]. The formation of passivating inrerlayers in the Ti/RuO,, TiO,, Pb/PbO, (thermal passivation) and Ti/PbO, systems were discussed in Sections 4 and 6.

821

122 (1836). 3. N. Cabrera and N. F. Mott, Rep. Prog. Phys. 12, 267 (1948). 4. K. J. Vetter and F. Gom, Electrochim. Acta 18, 321 (1973). 5. R. Kirchheim, Electrochim. Acta 32, !619 (1987). 6. H.+L Beer, G.D.R. Patent 55 223 (1965); Belgium Patent 710 551 (1967). 7. G. Bianchi, P. Gallone, A. Nidola and V. de Nora, F.R.G. Patent. 1814567 (1968). 8. F. Beck, Elektroorganische Chemie, p. 97, Verlag Chemie, 9. CONCLUDING REMARKS Weinheim (1974). 9. A. T. Kuhn and H. Shalaby. Electrochim. Acta 25, 745 The chemical and electrochemical stability of a (1980). metal anode, which ensures a long service life, is 10. J. W. Schultze and S. Mohr, Dechem-Moncgr. 98, 231 realized by an appropriate composition of the oxidic (1981). 11. W. A. Gerrard and B. C. H. Steele, J. appl. Electrochem. 8, passive layer. Electronic conductivity should exceed 417 (1978). the ionic contribution by orders of magnitude. How12. K. J. O’Leary, U.S. Patent 3776 834 (1973). ever, in many systems not the transport of ions 13. K. Doblhofer, M. Metikos, 2. Ogumi and H. Gerischer, through the film, but the ionic charge transfer reaction Ber. Bunsenges. phys. Chem. 82, 1046 (1978). at the outer phase boundary is the rate determining 14. S. Trasatti and G. Buzzanca, J. electroanal Chem. 29, step. A good adherent, non-porous passive film is App. l(l971). another important precondition. The level of elec15. T. Arikado, C. Iwakura and H. Tamura, Electrochim. tronic conductivity should be as high as possible, not Actn 22, 513 (1977). only to meet the above mentioned condition, but also 16. L. D. Burke and 0. J. Murphy, _ J. electroanal. Chem. 109, 199 (1980). to prevent the establishment of high electric flow fields, 17. A. Bandi, I. Vartires, A. Mihelis and C. Hainarosie, where ions may become mobile. The description of the J. efectroanal. Chem. 157, 241 (1983). oxide layer as a semiconductor[l, lo] or as a “chemi18. L. I. Krishtalik, Electrochim. Acta 26, 329 (1981). conductor”, with many energy levels occuring in the 19. F. Beck and H. Schulz. Electrochim. Acta 29,1569 (1984). band gap[61, 621, is presently under discussion. 20. A. Nidola, in Electrodes of Conductive Metal Oxides Redox transitions at the surface of the oxide layer (Edited by S. Trasatti), pp. 627659, Elsevier, Amsterdam enter into the electrode reactions in terms of hetero(1981). geneous redox catalysis. Thus the catalytic properties 21. F. Beck. Cowos. Sci.. in nress. of the electrode depend highly on the geometric and 22. J. Kruger, Cowos. Sk, ii press. electronic quality of the anodically generated high 23. K. Heusler, K. F. Bonhoeffer and K.-G. Weil, cited in: H. Gerischer, Angew. Chem. 70,288 (1958); cjI N. Sato, K. valency oxides. Moreover, a new understanding of Kudo and T. Noda, Corros. Sci. 10, 785 (1974). many wear phenomena is derived therefrom.’ While 24. H. Kaesche, Die Korrosion der Metalle, Springer, Berlin tri- and tetravalent states are usually very stable (eg 119791. \-RuO,/RuOOH), oxides in their highest oxidation 25. Ch. R. Campbell, D. E. Danly and W. H. Miiller, U.S. states as CrO,, FeO, or RuO,/RuO, are Lewis acids Patent 3 830 712 (1972). and they can leave the electrode by hydration‘or even 26. D. E. Danly, Hydrocarb. Process. 60, 161 (1981). vaporisation. On the other hand, low valent oxides as of Lead, 27. A. T. Kuhn (Editor), The Electrochemistry TiO or VO, which are present at the cathode, act as Academic Press, New York (1979). Lewis bases, which dissolve either. The competing 28. J. P. Pohl and J. Zschoche, Dechem.-Monogr. 109, 297 (1987). oxidation reaction (or cleavage under 0, generation) 29. J. P. Pohl and W. Schendler, J. Pwr Sources 13, 101 should be as fast as possible in order to minimize wear (1984). of electrode materials. 30. B. N. Kabanow, E. S. Weisberg, I. L. Romauowa and E. According to Figs 1 and 3, the base material of the V. Krivolapova, Electrochim. Acta 9, 1197 (1964). oxide composite anode is a metal. It works in a region 31. D. Wabner, H. P. Fritz and R. HUD, Chem.-lng.-Techn. of extreme thermodynamic instability. The oxide layer 49, 329 (1977). acts therefore in addition as a corrosion protecting 32. J. Thanos, H. P. Fritz and D. Wabner, J. appl. Electrocoating for the substrate. As most industrial anode them. 14, 389 (1984). 33. F. Beck, F.R.G. Patent 2 023 292 (1970). processes are operating in aqueous electrolytes, this 34. F. Hine, M. Yasuda, T. Iida, Y. dgata and K. Hara, task is appreciably facilitated. It leads back to the Electrochim. Acta 29, 1447 (1984). corrosion aspect of these system@, 22, 241, which is 35. F. Beck and W. Gabriel, J. electroanal. Chem. 182, 355 usually prevailing (1985). 36. F. Beck and H. Schulz, Ber. Bunsenges. phys. Chem. 88, Acknowledgement-Financial support of our work by 155 (1984). Deutsche Forschungsgemeinschaft is gratefully acknow37. F. Fleck and H. Schulz, J. electroanal. Chem. 229, 339 ledged. (1987). 38. H. Schulz and F. Beck, J. OQQf. Electrochem. 17,914 (1987). 39. H. Schulz and F. Beck, Angew. Chem. Inc. Ed. Engl. 24, 1049 (1985). REFERENCES 40. F. Fkck and W. Gabriel, Angew. Chem. Jnt. Ed. Engl. 24, 771 f1985). 1. R. P. Frankenthal and J. Kruger (Editors), Pnssioity of Metals, The Electrochemical Society, Princeton, NJ 41. W. Gabriel, PhD Thesis, University of Duisburg (1985). 42. F. Beck, W. Gabriel and H. Schulz, De&em.-Moaogr. (1978). 102, 339 (1986). 2. Ch. F. Schiinbein and M. Faraday, Phil. Mag. 9, 53, 57, ~I

F. BECK

822

43. F. Beck and E. Zimmer, unpublished results. 44. G. S. Forbes and L. P. Hall. J. Am. chem. Sot. 46, 385 (1924). 45. R. Katz, H. J. Lewerenz and S. Stucki, J. electrochem. Sot. 130, 825 (1983). 46. R. K&z and S. Stucki, Electtwchim. Artu 31, 1311 (1986). 47. G. Grube and H. Gmelin, Z. Elektruchem. 26,459 (1920). 48. G. Grube, 2. Elelltrochem. 33, 389 (1927). 49. F. Beck, R. Kaus and M. Oberst, Electrochim. Acta 30, 173 (1985). 50. F. Beck and R. Kaus, unpublished results. 51. H. Uhlig and T. O’Connor, J. electrochem. Sot. 109,781 (1962). 52. D. E. Hall, J. electrochem. Sot. 132, 41 (1985). 53. F. Beck, H. Schulz and B. Jansen, Electrochim. Acta 31, 1131 (1986).

54. T. R. Beck, J. electrochem. Sot. 120, 1310 (1973). 55. L. D. Burke, J. F. Healy and 0. J. Murphy, J. appl. Electrochem. 13, 459 (1983). 56. D. E. Danlv and C. R. Camobell. in Technioues of Chemistry, vol. 5, Part III (Edited by &L L. Weinberg an& B. V. Tilak), Wiley, New York (1982). 57. F. Beck, J. electrochem. Sot. 129, 1880 (1982). S8. N. A. Hampson and M. J. Tarbox, J. electrochem. Sot. 110, 95 (1963). 59. F. Beck and I. Litzenberger, unpubhshed results. 60. E. Peled, J. electrochem. Sot. 126, 2047 (1979). 61. B. D. Cahan and C. T. Chen, J. electrochem. Sot. 129,921 (1982). 62. B. D. Cahan, Corros. Sri., in press. 63. J. Llopis, J. M. Tordesillas and M. Muniz, Electrochim. Acta 10, 1045 (1965).