Incorporation and mobility of zinc ions in anodic alumina films

Incorporation and mobility of zinc ions in anodic alumina films

thin ELSEVIER o Thin Solid Films 292 (1997) 150-155 Incorporation and mobility of zinc ions in anodic alumina films H. Habazaki b, X. Zhou "~, K. S...

654KB Sizes 0 Downloads 24 Views

thin ELSEVIER

o

Thin Solid Films 292 (1997) 150-155

Incorporation and mobility of zinc ions in anodic alumina films H. Habazaki b, X. Zhou "~, K. Shimizu °, P. Skeldon ~, G.E. Thompson ~, G.C. W o o d a a

Corrosion and Protectio~z Centre, University of Manchester hzstitute of Science and Technology, P.O. Box 88, Manchester M60 IQD, UK b Institute for Materials Research, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendal 980-77, Japan c Department of Chemistry, Keio University, 3-14-1 Hiyoshi, Yokohama 223, Japan Received 11 March 1996; accepted I 1 June 1996

Abstract

The migration rate of zinc ions in growing anodic alumina films has been determined as part of a systematic study for understanding the ionic transport processes in anodic films under a high electric field. An A1-0.2 at% Zn alloy, about 35 nm thick, sputter-deposited onto an electropolished, high purity aluminium substrate has been anodized at a constant current density to various voltages at high current efficiency. During anodizing of the alloy, zinc atoms are accumulated in a layer of alloy, about 2 nm thick, just beneath the anodic film as a consequence of prior oxidation of the aluminium atoms. No zinc ions are incorporated into the anodic aIumina film during anodizing of the alloy. When the alloy film is almost totally consumed by anodizing, zinc atoms in the enriched alloy layer are oxidized and incorporated immediately into the anodic film as a result of the presence of an air-modified, electropolishing film on the aluminium substrate. During further anodizing, zinc ions migrate outwards; the migration rate of the zinc ions is about 2.3 times that of A13+ ions. The migration rate of the zinc ions is slower than that of the Cu 2+ ions incorporated into the anodic films during anodizing of dilute AI-Cu alloys, although the strengths of the single Zn2+ -O and Cu 2+-O bonds and the radii and valences of Zn 2+ and Cu 2+ ions are similar. Keywords: Aluminium; Anodic oxidation; Zinc ions

1. Introduction Barrier-type amorphous anodic films, formed on valve metals, their alloys and semiconductor materials, are of considerable interest both for practical applications, such as capacitor production, and, fundamentally, for understanding the complex ionic transport processes within the anodic films during their growth in aqueous solutions at ambient temperature. The anodic films develop both at the film/electrolyte interface, by migration of cations outwards, and at the metal/ film interface, by the migration of anions inwards. For anodizing of aluminium at 100% current efficiency, about 40% of the film thickness is formed at the film/electrolyte interface, the remainder forming at the metal/film interface [ 1 ], over a wide range of current density and hence electric field strength [ 2]. In order to explain these findings, a cooperative mechanism of transport of cations and anions in amorphous alumina films is needed [ 3 ]. It is well known that anodic alumina films are usually contaminated with species derived from the particular anodizing electrolytes employed, e.g. boron species derived from borate electrolyte [ 4]. Contaminant species can also be incorporated into anodic films from substrates by the anodizing of aluminium alloys. The species incorporated from either the

electrolyte or alloy reveal a range of mobilities. Furthermore, the mobilities of various implanted ions have been examined previously using Rutherford backscattering spectroscopy (RBS) [5], although precise migration rates of many species were not determined as a result of their faster migration than aluminium ions and their direct ejection into the electrolyte at the film/electrolyte interface during anodizing [5]. Elucidation of the relationship between the mobilities of the individual incorporated species and the physical and/or chemical properties of the species can assist the understanding of the ionic transport mechanisms in anodic films. Consequently, it is important to investigate the mobility of a wide range of species incorporated into anodic alumina films. Recently, the present authors have found that anodic oxidation of various dilute aluminium alloys such as AI-Cu [6], AI-Ti [7], A I - W [8] and AI-Zn [9] alloys proceeds in two stages. In the initial stage, prior oxidation of the aluminium atoms and the accumulation of the alloying element atoms in the alloy at the alloy/film interface occur. Utilizing this prior oxidation of aluminium atoms and the interfaciat enrichment of the alloying element atoms, a novel approach in determining the precise mobility of the contaminant ions has been developed. Thus, a dilute aluminium alloy film, of thickness about 30-40 nm, is sputter-deposited on to an electropolished

0040-6090/97/$17.00 Copyright © 1997 Elsevier Science S.A. All rights reserved PHS0040-6090(96)09006-2

H. Habazaki et ai. / T h i n Solid Films 292 (1997) 150-155

aluminium substrate and subsequently anodized [ 10 ] ; during anodizing of the alloy film, atoms of the alloying element are accumulated in a thin alloy layer, about 2 nm thick, just beneath the anodic film. The atoms in the layer enriched in the alloying element are oxidized and incorporated immediately into the anodic film when the original alloy layer is consumed fully by oxidation and the alloy/film interface reaches the air-modified film of the electropolished aluminium substrate. From the positions of the enriched layer of alloying element ions in the anodic film, at particular anodizing voltages, the migration rate of the ions in anodic alumina can be determined. Using this approach, the migration rate of copper ions [ 10 ] has been determined precisely. The present article focuses on the migration rate of zinc ions in anodic alumina.

I00

. . . . AI-0.2

151

I at%

- Anodized

80

-

' ' ' ' Zn/Aluminium

at

50

I 10

T

A

~

'

~

'

'

I

m "2

. . . .

J

> 60 :~ 4o

2O 0

, ~~

t

r

T

~

I ~~, 20

Time

T

r 30

,

r

r

q 40

/ s

Fig. l. Voltage-time response of the AI-0.2 at% Zn alloy film sputterdeposited onto 99.99% aluminiumduring anodizingat 50 A m-z to 90 V in 0.01 M ammoniumpentaborateelectrolyteat 293 K.

3. Results

2. Experimental

3.1. Voltage-time response

Al-0.2 at% Zn alloy films were prepared by d.c. magnetron sputtering using a Microvac 350 system. An AI-0.4 at% Zn alloy disc, 38 mm in diameter, was used as the target. The composition of the alloy films was confirmed by RBS to an accuracy of about 5%. The reduced content of zinc in the alloy film, compared with the alloy target, is possibly due to the relatively lower sputtering rate of zinc. The alloy films were deposited onto 99.99% pure aluminium substrates which had been electropolished at 20 V for 5 min in ethanolperchloric acid electrolyte at less than 283 K. After evacuation to 5X 10 .5 Pa, sputter deposition was performed in 3 × 10- ~ Pa argon at a constant current of 200 mA for 5 min. During deposition, the substrates were rotated about the central axis of the chamber in order to achieve a relatively uniform film thickness. Specimens of about 10 .4 m 2 were anodized at a constant current density of 50 A m -2 in stirred 0.01 M ammonium pentaborate electrolyte at 293 K to selected voltages of 40, 60 and 90 V. An aluminium sheet was used as the counter electrode. Electron transparent sections of the anodized specimens were obtained using a Reichert-Jung Ultracut E ultramicrotome. The encapsulated specimens were trimmed initially with glass knives, and suitably thin transverse sections less than 10 nm thick were prepared by final sectioning approximately normal to the alloy/film interface of the specimens with a diamond knife. The sections were examined in aJEOL TEM 2000FX II transmission electron microscope operating at 120 kV. The anodized specimens were analyzed by RBS using a 2.0 MeV alpha particle beam supplied by the Van de Graaff accelerator of the University of Paris. The beam current and diameter were about 60 nA and 0.5 mm respectively. Scattered particles were detected at 150 ° to the incident beam direction. The data were analyzed using the RUMP program [ 11 ], with scaling of the stopping power for oxygen in accord with recent RBS analyses [ 12].

Fig. 1 shows the voltage-time response of the Al-0.2 at% Zn alloy superimposed on electropolished aluminium during anodizing at a constant current density of 50 A m - 2 to 90 V in 0.01 M ammonium pentaborate electrolyte at 293 K. Following a voltage surge of about 2 V at the beginning of anodizing, caused by the presence of the air-formed film at the alloy surface, the voltage increases linearly to 47 V with a slope of 2.3 V s - 1, which is similar to that for anodizing high purity aluminium, at high current efficiency, under the selected conditions. At 47 V a second voltage surge of about 2-3 V is observed, indicating the presence of an air-modified electropolishing film, about 2-3 nm thick, sandwiched between the sputter-deposited alloy film and the aluminium substrate [ 10]. Thus, detection of this second voltage surge indicates that the sputter-deposited A1-Zn alloy film is consumed totally by anodizing to 47 V. Assuming a nm V-1 ratio of 1.2 [ 13] and a Pilling-Bedworth ratio of 1.61 [ 14] for anodic alumina formed on the dilute AI-Zn alloy, the average thickness of the original alloy film is estimated to be 35 nm. The second voltage surge is reproducible to about 1 V between different specimens indicating that the average thicknesses of the deposited alloy layers were similar, to within about 1-2 nm. The original alloy surface was probably relatively rough, similar to sputter-deposited At-0.4 at% Cu alloy films, caused by the development of facets during sputtering [ 10 ]. Following the second voltage surge, the voltagetime response is linear with a slope similar to that for the initial region of anodizing, below 47 V, which is indicative of further growth of film at a high current efficiency.

3.2. Transmission electron microscopy (TEM) A transmission electron micrograph of an ultramicrotomed section of the specimen anodized to 60 V is shown in Fig. 2. The section reveals the main features of the anodic film also disclosed by the examination of sections of specimens anodized to 40 and 90 V. The anodic film has relatively flat and

152

(1997) 150-155

H. Habazaki et al. /Thin Solid Films 292

, (a) ~ .

==

AI-0.2 at% Zn AIIoy/Alumlnlum Anodized to 40 V

,~/A,

"-"'::'%'~>:

4000

8

Voids Layer - 0

t

/

Simulated '

I

r

1 O0

I

I

P

\1

{', t

200

I ,";t"

.t l,

. ,

"'o']l \ 't~, "-?';","

300

400

Channel

Fig. 2, Transmission electron micrograph of an ultramicrotomed section of the A1-0,2 at% Zn alloy film sputter-deposited on to 99.99% electropolished, high purity aluminium after anodizing at 50 A m - 2 to 60 V in 0.01 M ammonium pentaborate electrolyte at 293 K.

parallel metal/film and the film/electrolyte interfaces. The thickness of the film is 74 + 3 nm, indicating a nm V - 1 ratio of 1.23 + 0.05, similar to that for films formed on high purity aluminium under the same conditions, namely 1.2 [ 13]. The anodic film is relatively featureless and of uniform appearance, typical of amorphous alumina, except for the presence of a light band, apparently comprised of fine voids at a location about 10% of the film thickness measured from the metal/film interface. The light band probably contains argon bubbles, which may be incorporated into the material at, or near, the interface between the alloy film and the underlying aluminium substrate during initial sputtering. Thus, bubbles were not observed in the film formed to 40 V since the alloy/ film interface is then separated from the aluminium substrate by a residual layer of alloy. The bubbles were also evident in the film formed to 90 V and appear to be immobile in the anodic film during anodizing [ 10]. When a thin layer of Al-0.9 at% Cu alloy, sputter-deposited on to an electropolished aluminium substrate was anodized to completion, followed by further anodizing of the aluminium substrate in a similar manner to the experiment here, a dark band of film material, about 3 nm thick enriched in copper species was observed directly in ultramicrotomed sections as a result of the enhanced electron scattering from the incorporated copper ions [10]. However, no zinc enriched layer of film material was observed in the films formed on the present more dilute alloy specimens because of the relatively low concentration of zinc in the alloy, and hence, relatively low concentration of zinc ions in the film.

3.3. Rutherford backscattering spectroscopy Fig. 3 shows experimental and simulated RBS spectra for the specimen anodized to 40 V. Prior oxidation of aluminium and the accumulation of zinc in the alloy, just beneath the anodic film, are known to occur during anodizing of an A10.23 at% Zn alloy to 200 V [9], with no incorporation of zinc ions into the anodic film. In agreement with this result, the energy of the peak caused by the scattering of alpha particles from zinc nuclei contained in the enriched alloy layer

30

. (b)

i

,

i , I ' I AI-0.2 at% Zn Altoy/Alumimum Anodized to 40 V

~20

Simulated for

s

,o v

, "i , j z

Simulated for As-Dep, Film

• """

300

320

340 360 Channel

380

,

400

Fig. 3. (a) Experimental and simulated Rutherford backscattering spectra for the AI-0.2 at% Zn alloy film sputter-deposited on to 99.99% electropoIished, high purity aluminium after anodizing at 50 A m-'- to 40 V in 0.01 M ammonium pentaborate electrolyte at 293 K. (b) Region of spectra of (a) for elastic scattering from zinc in greater detail, together with the simulated spectrum for the as-deposited alIoy film.

for the AI-0.2 at% Zn alloy anodized to 40 V is lower than that for scattering from zinc nuclei at the specimen surface, as evidenced by comparison with the simulated spectrum for the as-deposited alloy, 35 nm thick, of composition 0.2 at% Zn (Fig. 3(b) ). The simulation for the anodized specimen, assuming an anodic film, of 48 nm thickness, consistent with a nm V - t ratio of 1.2, a layer of alloy, of 2 nm thickness, of composition 3 at% Zn at the alloy/film interface and an A10.2 at% Zn alloy film, of 5 nm thickness, is in good agreement with the experimental spectrum (Fig. 3(a)). The assumed thickness of the enriched alloy layer is based on direct observation, by TEM, of the zinc-enriched layer in a more concentrated alloy [9]. From the voltage-time response, given previously, zinc ions are expected to be incorporated into the anodic film at about 47 V. The presence of zinc ions in anodic films formed to 60 and 90 V is confirmed by the RBS spectra shown in Figs. 4-6. The energy of the peak as a result of scattering from zinc ions, now contained in the enriched layer of anodic film material, is higher for the 90 V film than for the 60 V film (Fig. 4). Hence, the distance from the film/electrolyte interface to the layer enriched in zinc ions is greater in the 60 V film than in the 90 V film; revealing the greater mobility of Zn a+ ions than A13+ ions in the thickening anodic alumina film. The simulated spectra for the 60 and 90 V films correspond closely with the measured spectra (Figs. 5 and 6) : the positions and compositions of the bands enriched in zinc ions

H. Habazaki et al. / Thin Solid Fihns 292 (1997) 150-155

I

J

I

'

I

I

AI-0•2 at% ZnlAluminlum

6O

g "• .



40

. °,*,'°',.',•o

,,o,,, ,°°

90 V

, ,',' , ,, • , ,. ,,,,,=,,.',°.

•, ,,

°,.,

20 "o°'

'

".

,. ,, ,°.,

;

o

','

320

•.



°

I

,• •,

60 V

,

°,,.

....

,..'. ~"""'r•.''•'•.,.'.•'. T "'.

340

360

380

400

Channel

Fig, 4. Experimental Rutherford backscattering spectra of the A1-0.2 at% Zn alloy film sputter-deposited on to 99.99% electropolished, high purity aluminium after anodizing at 50 A m -'- to 60 and 90 V in 0.01M ammonium pentaborate electroIyte at 293 K.

8000

6000

°

8

LA oo,, ,,ooov b



-

i

i

,

"'":"

:

,

'

'

J

>

,,...

4000

°

I

'

'

'

"

~

/

Simulated

-

i

i

100

i

q

~ l

i

'~

200

{-: ....

/~/Zn /\

P,",',:"","',~ ~'~

"c',,

i

300

400

Channel

Fig. 5. Experimental and simulated Rutherford backscattering spectra of the AI-0.2 at% Zn alloy film sputter-deposited on to 99.99% electropolished, high purity ahiminium substrate after anodizing at 50 A m -a to 60 V i n 0.01 M ammonium pentaborate electrolyte at 293 K.

8000 •

. . . . f?~

6000 ~ ' ' " . ' , :

== 8

" 4000~

-= ooo

i ', , , ' i . . . . A I . 0 . 2 at% Zn A l l o y / A l u m i n i u m

A

..... Simulated

100

(0.6+0.4×

/ z n 2 + ) ( 1 - --"~)

in which Vo is the voltage at which the zinc ions are incorporated into the film, that, from Fig. 1, is about 47 V. The equation is derived by considering growth of an anodic alumina film at a high Faradaic efficiency with transport numbers Of A13 + and O a- ions of about 0.4 and 0.6, respectively, and thus, formation of about 40% and 60% of the film thickness at the film/electrolyte and metal/film interfaces, respectively. Fig. 7 reveals a linear change in the normalized distance of the band enriched in zinc ions with the reciprocal of the film formation voltage. From the slope in Fig. 7, the relative migration rate of zinc ions, with respect to A13 ÷ ions, is 2.3 + 0.2.

4. Discussion

4.1. Enrichment of zinc in the alloy at the alloy/film interface

"-. ' " . " . ,

/

and the total thicknesses of films, formed to different.voltages, assumed for the simulations, are summarized in Table 1 with the precision of the results indicated. In these simulations, the thickness of the band enriched in zinc ions is 3 nm, based on an original thickness of zinc-enriched alloy of 2 nm and a Pilling-Bedworth ratio of 1.61 [14]. This estimate of the thickness of the zinc-enriched band of film material is consistent with direct observations of the thickness of the copperenriched band of film material evident in similar studies of AI-Cu alloy [ 10]. For outwardly mobile zinc ions with a migration rate relative to A13 + ions, Izn~.+, the ratio, r, the distance of the band enriched in zinc ions from metal/film interface to the total film thickness at a particular voltage, V, can be expressed as follows [ 10]: r=

'

IxS° . . . . . Experimental

0

I

/At

~

*~ 2000

~

AI-0.2 at% Zn A l l o y / A l u m i n i u m

_

I53

I

/

I

/

I'~ . . .

200

300

A

.', 400

Channel

Fig. 6. Experimental and simulated Rutherford backscattering spectra of the Al-0.2 at% Zn alloy film sputter-deposited on to the 99.99% electropolished, high purity aluminium after anodizingat 50 A m - 2 t o 9 0 Vin0.01 Mammon±urn pentaborate electrolyte at 293 K.

It is known that anodic oxidation of sputter-deposited dilute alloys, such as A1-1.9 at% W [8], or mechanically• polished, dilute bulk aluminium alloys, such as A1-0.9 at% Cu [6], proceeds in two stages. In the initial stage, both the prior oxidation of aluminium and the progressive accumulation of the alloying element in a thin layer of alloy, 1-2 nm thick, just beneath the anodic film takes place. In the second stage, both aluminium and the alloying element ions are incorporated into the film at the alloy/film interface in virtually their alloy proportions in the presence of a steady-state enriched alloy layer. The transition between the two stages

Table 1 The thickness of anodic films and the position and composition of the zinc-containing band in the anodic films Anodizingvoltage(V)

Filmthickness(nm)

No~aiizedpositionofthezincb~d,r

Compositionofthezincb~d[Zn]/([A1]+[Zn])at%

40 60 90

48+3 73±3 103±3

0 0,29±0.03 0.73±0.03

3.0±0.5 3.0±0.5 3.0±0.5

154

H. Habazaki et al. /Thin Solid Fihns 292 (1997) 150-155

t

QE

~

I

'

I

0.6

z~ ~a ~ 0.4 t~

"-0.2

o.o

0,012

O.01Q V" ~ (Voltage)" /

0.02

Fig. 7. Variation of the distance, r, of the band of zinc-enriched film material, normalized with respect to the total film thickness, from the metal/film interface with the reciprocal of the film formation voltage, V, for the AI0.2 at% Zn alloy film sputter-deposited on to high purity, electropolished aluminium after anodizing at 50 A m -2 in 0.01 M ammonium pentaborate electrolyte at 293 K.

occurs when the necessary enrichment of the alloying element for oxidation of the alloying element atoms is achieved. Thus, in the second stage, the average enrichment of the alloy does not change significantly with an increase in the anodizing voltage; hence there is no further net accumulation of alloying element atoms and accordingly aluminium and the alloying element atoms are oxidized at the alloy/film interface and enter the anodic film in proportion to the concentrations in the main alloy material. The critical average compositions for the enriched alloy layer are dependent upon the particular alloying element and the alloy composi.tion. Thus, the enrichments of the alloying element in the steady-state layer beneath the anodic film on Al-0.9 at% Cu [6] and A l - l . g a t % W [8] alloys correspond to 5.6 × 1019 and 2.3 × 1019 atoms m -a respectively. In a similar manner, the anodic oxidation of dilute AI/Zn alloys proceeds in two stages [9]. The prior oxidation of aluminium on an electropolished AI-0.9 at% Zn alloy is terminated upon achieving an average enrichment of zinc; determined experimentally, corresponding to about 4.6X t019 atoms m -a in a thin layer of alloy immediately beneath the anodic film. For a sputter-deposited A1-0.23 at% Zn alloy, of similar composition to the alloy employed in the present study, the measured enrichment of zinc at the alloy/ film interface is about 1.8 × 1019 atoms m -2 following anodizing to 200 V, and there is no significant incorporation of zinc ions into the anodic film to this voltage [9]. Accordingly when the A1-0.2 at% Zn is anodized to 47 V, with accumulation of about 4.2× 1018 zinc atoms m -2, estimated from the thickness of alloy consumed, assuming an alloy density of 2.7 g cm - 3, no 'incorporation of zinc ions into the anodic film is expected. However, for the present specimens the accumulated zinc atoms cannot retreat further with the alloy/ film interface following anodizing to voltages exceeding 47 V, due to the presence of the air-modified electropolishing film above the electropolished aluminium substrate. Thus, the zinc atoms are oxidized and incorporated into the anodic film at a much lower voltage than for a bulk alloy of similar

composition. The zinc ions incorporated into the film migrate outwards at a uniform rate with respect to the A13+ ions. To a reasonable approximation, the ordering of ions is conserved during growth of anodic alumina, thus the zinc enriched band remains evident as the film thickens under the high electric field. From Table t, the amount of zinc in the enriched band is about 3.3× 1018 Zn atoms m -2, assuming anodic film material of usual density 3.1 Mg m - 3, about 20% lower than the calculated enrichment of the alloy following total consumption of the alloy layer. The reduced amount of zinc in the enriched band of the film may indicate some spreading of the band or a loss of zinc ions from the band with increasing distance of migration, as might be anticipated from the mechanism of ionic transport within amorphous anodia oxides [ 15]. However, this spreading and loss cannot be calculated from present knowledge of the transport process.

4.2. Mobility of zinc ions in anodic alumina films Foreign species incorporated into anodic alumina films reveal a range of mobilities under a high electric field and have been used as tracers or, in the case of immobility, markers for understanding the ionic transport processes [4]. The varied mobility may be related partly to the charge of the individual species, present in growing anodic films, which may be effectively positive, negative or neutral. For example, tungsten species incorporated from tungstate electrolyte migrate outwards, as a result of the transformation of tungstate anions to cationic species in the growing alumina film [4]. Boron and phosphorus species incorporated into films from borate and phosphate electrolytes are immobile and migrate inwards, respectively, indicating the presence of neutral and negatively charged species, respectively [4]. For other cationic species, such as molybdenum [ 16], tungsten [ 8 ] and zirconium [ 17 ] species incorporated from alloy substrates, a range of migration rates relative to A13 + ions is revealed. Recently Shimizu and Kobayashi [18] found a good correlation between the relative migration rates of such species with respect to A13 + ions and their single metaloxygen bond energies. They also found a good correlation between the relative migration rates of species associated with weak single metal-oxygen bonds, compared with the A13 + O bond, and their ionic radii and valencies [ 18]. These findings suggest strongly that the rupture of metal--oxygen bonds is an important rate-determining step in the ionic migration process in amorphous anodic alumina films. From the present study, the relative migration rate of zinc ions is about 2.3 times that of AI 3+ ions. This migration rate is determined relatively precisely from the positions in the film for the thin layer of zinc-enriched film material at particular formation voltages. Using a similar approach, the relative migration rate of Cu a+ ions in a growing alumina film is about 3.2 times that of A13+ ions. Further, In 3+ ions implanted deeply into anodic alumina migrate about 1.8 times as fast as A13+ ions [5]. Unpublished studies of the present authors, using extended X-ray absorption spectroscopy sug-

H. Habazaki et al. /Thin Solid Fihns 292 (1997) 150-155

gest that zinc is present in the anodic film as Zn 2+ ions. Interestingly Cu e +, In 3 ÷ and Zn 2 ÷ ions, having similar single metal-oxygen bond energies of about 44 kcal m o l - 1 and ionic radii of 0.06-0.08 nm [ 19], are expected, from the correlations found by Shimizu and Kobayashi [ 18], to have similar migration rates in anodic alumina. Apparently, the different migration rates of these ions in anodic alumina cannot be explained solely in terms of the strengths of the single m e t a l - o x y g e n bonds, ionic radii or valences, and additional factors which influence the migration rates of incorporated ions need to be considered.

5. Conclusions

1. The anodic oxidation of an A l - 0 . 2 at% Zn alloy film, about 35 nm thick, sputter-deposited on to high purity, electropolished aluminium, results initially in prior oxidation of aluminium atoms, with zinc atoms accumulating in a thin layer o f alloy, just beneath the anodic film. 2. After anodizing to 47 V, the alloy film is consumed and zinc atoms accumulated in the enriched alloy layer are then oxidized and incorporated into the anodic film. The concentration of zinc atoms in the enriched alloy layer, about 4;< 1018 atoms m -2, compares with a measured critical composition for incorporation of zinc ions into the anodic film grown on a thick layer of alloy of more than 1.8 × 1019 atoms m -2. 3. The oxidation of zinc atoms at low alloy enrichment is assisted by the presence of the air-modified electropolishing film sandwiched between the alloy and aluminium substrate, which prevents the retreat of zinc atoms with the a l u m i n i u m / a n o d i c film interface. 4. The zinc ions, present in a thin band of film material, migrate outwards during anodic oxidation at about 2.3 times the rate of A13 + ions.

I55

Acknowledgements The authors thank the Engineering and Physical Sciences Research Council for the award of Research Assistantships to X. Zhou. The assistance of Professor G. AmseI and Dr C. Ortega in the use of the Van de Graaff at the University of Paris is gratefully acknowledged.

References [ 1] F. Brown and W.D. Mackintosh, J. Eiectrochem. Soc., 120 (1973) 1096. [2] G.E. Thompson, Y. Xu, P. Skeldon, K. Shimizu, S.H. Han and G.C. Wood, Phil. Mag. B55 (1987) 651. [3] L. Young and D.J. Smith, J. Electrochem. Soc., 126 (I979) 765. [4] P. Skeldon, K. Shimizu, G.E. Thompson and G.C. Wood, Thin Solid Fihns, 125 (1985) I27. [5] W.D. Mackintosh, F. Brown and H.H. Plattner, J. [email protected] Soc., 121 (1974) i281. [6] H. Habazaki, M.A. Paez, K. Shimizu, P. Skeldon, G.E. Thompson, G.C. Wood and X. Zhou, Surf Interface Anal., 23 (1995) 892. [7] H. Habazaki, K. Shimizu, P. Skeldon, G.E. Thompson, G.C. Wood and X. Zhou, Corros. Sci, in press. [ 8] H. Habazaki, K. Shimizu, P. Skeldon, G.E. Thompson and G.C. Wood, Phil. Mag. B., 73 (1996) 445, [9] X. Zhou, H. Habazaki, K. Shimizu, P. Skeldon, G.E. Thompson and G.C. Wood, Corros. Sci., in press. [ 10] H. Habazaki, X. Zhou, K. Shimizu, P. Skeldon, G.E. ThomPson anti G.C. Wood, Electrochim. Acta, submitted. [ 11] L.R. Doolittle, Nucl. Instr. and Meth., B9 (1985) 344. [12] J.C. Cheang Wong, Jian Li, C. Ortega, J. Siejka, G. Vizkekethy and Y. Lemaitre, Nucl. Instr. and Meth., B64 (1992) 169.' ' [ I3] A.C. Harkness and L. Young, Can. J. Chem., 44 (I966) 2409. [ 14] J.P.S. Pringle, Electrochem. Acta., 25 (I980) i423. [ 15] H. Habazaki, K. Shimizu, P. Skeldon, G.E. Thompson and G.C.Wood, Phil. Mag. B, 73 (I996) 297. [ 16] H. Habazaki, P. Skeldon, K. Shimizu, G.E. Thompson and G.C. Wood, Corros. Sci., 37 (i995) 1497. [ 17] H. Habazaki, K. Shimizu, P. Skeld0n, G.E. Thompson and G.C. Wood, J. Phys. D: Appl. Phys., 28 (i995) 2612. [ 18] K. Shimizu and K. Kobayashi,J. Surf. Finishing. Soc. Jpn., 46 (I995) 402. [19] CRC Handbook of Chemistry and Physics, (Ed. D. R. Lide) CRC Press, 1995.