Electrccbii~ Acta. 1970, Vol. 15, pp. 1865 to 1876. PWEUXIOU F%CSS.

Printed in Northern Ireland

THE ANODIZING OF ALUMINIUM IN

SULPHATE SOLUTIONS*
G. C. WOOD and J. P. O’SULLIVAN~
Corrosion Science Division, Department of Chemical Engineering, University of
Manchester Institute of Science and Technology, Manchester M60 lQD, England

Abstract-The morphology of films formed in H,SO,Na,SO, mixtures for different times at various
constant cds has been studied as a function of electrolyte concentration, pH and temperature, using
electron microscopy. At low pH, local preferential film dissolution competes with the film-forming
reaction, producing typical porous films whereas, as the pH is raised towards neutrality, pitting by
metal dissolution becomes a progressively more important competitive process, probably initiating at
the base of flaws in the iilm. For porous films, it is shown that the pore diameter, cell diameter and
barrier-layer thickness are all proportional to the formation voltage, there being evidence that pore
initiation and development are largely controlled by field-assisted oxide dissolution, with possible
lesser contributions from Joule heating effects. Severe anodizing conditions produce pore widening
and merging, as well as general roughening, at and near the outer surface but, although the film
appears less regular and compact, there is no evidence of major pore widening throughout its thickness.

R&am&-Etude par microscopic &ctronique de la morphologie des i%lmsform& dans les mdlanges
H,SO*-Na4S04 1 des temps diffkrents, pour diverses dens& de courant constantes en fonction de la
concentration de I’Gctrolyte, du pH et de la tempbrature. Aux bas pH, la dissolution locale p&f&en-
tielle du flms s’oppose & la r&action de formation du am, produisant des films poreux typiques tandis
que, quand le pH augmente vers la neutralit& l’attaque par dissolution du m&al devient progressive-
ment un processus comp&itif plus important, probablement initialement determinant des d6fauts dans
le film. Pour les films poreux, on montre que le diam&tre des pores, de la cellule et l%paisseur de la
couche front&e sent tous proportionnels 2tla tension de formation, mettant evidence que l%bauche
et le developpement des pores sont surtout contr&s par la dissolution de l’oxyde en pr&ence d’un
champ, avec la possibilitB de contributions moindres provenant de la chaleur par effet Joule. De
conditions d’anodisation strictes produisent un tlargissement et une transformation du pore, aussi
bien qu’une rugosit6 g&&ale, a et pr& de la surface extkrieure, mais, bien que le tirn apparaisse mains
r@ulier et compact, il n’y a aucune manifestation de pores importants dans 1’6paisseur.
Znsammenfassung-Die Morphologie von Filmen, die in HaS04--Na2SOo Mischungen sich bildeten,
ist als Funktion elektrolytischer Konzentration, pH und Temperatur unter Verwendung von Elektron-
enmikroskopie fiir verschiedene Zeiten bei verschiedenen konstanten cds untersucht worden. Bei
niedrigem pH tritt brtliche PrZiferenzl7lmaufl6sung in Wettbcwerb mit der filmbildenden Reaktion und
ergibt typische, poriise Filme, wahrend mit Vergrijsserung van pH ib Richtung Neutralitlt, Aushiihlen
durch MetallauflBsung ein fortschreitend wichtigerer, konkurrierender Vorgang wird, der wahrschein-
lich an der Basis der Fehler in dem Film beginnt. Fiir porijse Filme wird gezeigt, dass der Porendurch
messer, Zellendurchmesser und Sperrschichtsttike durchweg abh%ingig von der Formationsspannung
sind, wobei sich Anzeichen ergeben, dass Porenbeginn und Entwicklung zum grossen Teil von Feld-
unterstiitzter Oxydaulliisung kontrolliert wird, mit miiglicherweise geringerem Beitrag Joule’scher
W&mewirkung. Starke Anodisierbedingungen verursachen Porenerweiterung und Verschmelzung
sowie alIgemeine VergrBberung an und nahe der Aussenoberfl&he, aberes ergebcn sich keine An-
zeichen fiir griissere Porenerweiterung iiber die ganze Stlrke, obwohl der Film weniger regelm&sig
und kompakt erscheint.
INTRODUCTION
THE MORPHOLOGY of porous anodic films formed on aluminium by anodizing in
sulphuric acid has been investigated recently by several workers,l-ll using electron
microscopy. However, there is a paucity of quantitative data and no detailed,
integrated study of the influence of the various major anodizing variables on pore
initiation and the steady-state cell and pore diameters has been presented. The present
paper presents such data and gives consideration to anodic oxidation in sulphate
solutions of higher pH, where pitting, as well as pore formation, can compete with
f&n growth.
+ Manuscript received 9 September 1969.
7 Deceased.
1865
1866 G. C. WOOD and J. P. O’SULLIVAN

EXPERIMENTAL PROCEDURE
Spade electrodes, usually 2 x 1 x O-1 cm, of 99.99% aluminium (Cu O-002,
Fe OXKl4, Si 0@03 wt%) and occasionally zone-refined aluminium (Ca 10, Cu 10,
Mn 2, Mg 2 ppm by wt), were electropolished individually in a perchloric-acid/ethanol
bath, washed and dried. Anodic films were formed at constant cd on the individual
specimens in stirred solution within f0.5”C of the stated temperature. Voltage/time
curves were obtained on a fast recorder.
Films were formed in the following solutions, prepared from Analar grade
reagents and distilled water :
O-5 M HZS04 (pH O-3); 1.0 M H&O, (pH 0%); l-5 M HZS04 (pH - 0.15);
mixtures of 1.0 M H&O4 and 1-O M Na2S0, with pH I - 1,2-O, 4-O and 6-7 respectively.
The voltages attained in sulphuric acid anodizing at cds up to 100 mA/cm* were
generally less than 30 V and usually in the range 10-25 V, although they were as high
as 35 V in sodium sulphate solution. Consequently the cell and pore dimensions were
considerably smaller than with phosphoric acid anodizingl*lsJs and were thus less
amenable to detailed quantitative assessment by electron microscopy. The standard
conditions fqr many experiments were anodic oxidation at 25 mA/cm* in 1.5 M H&O,
at 2O”C, which established 18 V across the electrolytic cell. The cathodic overpotential
and II? drop contributed only O-4 V, which may be neglected by comparison with
errors in measurements from electron micrographs, so that the voltage across the
electrolytic cell may be taken as the forming voltage.
Thin films stripped from the aluminium substrate, and shadowed carbon replicas
of film surface, under-sides and fracture sections, were prepared and examined in the
electron microscope by methods described elsewhere.lZ*14 Fitted specimens were
examined in a Cambridge Stereoscan scanning electron microscope.
RESULTS AND INTERPRETATION
l$?uence of cd, pH and concentration upon cell-voltage/time curves
The voltage/time curves, Fig. 1, for anodizing at different cds under the standard
conditions were reproducible to within 10%. At 50 and 80 mA/cm” the voltage
reached a maximum in 2-3 s, corresponding to the onset of major pore development
in the barrier layer. The steady voltage was reached after 10-15 s, corresponding to
the establishment of the steady-state pore distribution. At 10 and 25 mA/cm* barrier-
layer thickening was slower, the forming voltage only reaching its much shallower and
barely detectable maximum after l&l2 s. The slopes seemed linear for 2-3 s in each
case, being greater at higher cds due to the higher fields producing more rapid growth.
The subsequent departure from linearity corresponded to some film dissolution and,
although it was only detected after 2-3 s it seems reasonable to assume that dissolution
under the field was occurring from the onset of growth under the constant cd con-
ditions. The curves for 10 and 25 mA/cm2 clearly illustrate that dissolution was
occurring while the voltage, and thus the general barrier-layer thickness, was still
increasing.
Corresponding curves for anodizing at 10 mA/cm2 in HzS0JNazS04 mixtures of
constant anion strength at 2O“C, Fig. 2, show that increase in pH towards neutrality
raised the peak voltage and the final steady-state cell voltage. The curve for the pH 1.1
solution, which produced solely a porous film, was similar to that for 1.5 M H&SO,
 The anodizing of aluminium in sulphate solutions 1867

FIG. 1. Voltage/time curves for the early stages of anodizing at constant cd in l-5 M
sulphuric acid at 20” C.
a, 10 mAjcma; b, 25 mA/cm’; c, 50 mA/cm’; d, 80 mA/cm~_

0 IO 20 30 40 50

Time, s

FIG. 2. Voltage/time curves for the early stages of anodizing at 10 mA/cm* in

H$O,/Na,SO* solutions at 20°C.
All solutions were M in SO1’- and had pH values of a, 1-l; b, 2-O; c, 4-O; d, 6-7.

(shown later), except that a slightly sharper peak was achieved. The other curves were,
however, partly determined by the pitting reaction which competed with, and in
certain cases predominated over, pore initiation.
Figure 3 shows that the peak voltage and final steady voltage were both higher the
more dilute the sulphuric acid, under conditions where pore formation but no pitting
occurred. In l-5 M acid the voltage rose to a steady value without passing through a
significant maximum, indicating that it possessed a greater dissolving power (in the
presence of the field) that could keep pace more easily with the competing oxide
growth, a trend also observed with decreasing cd, Fig. 1.
1868 G. C. WOOD and J. P. O’SULLIVAN

Time, s

FIG. 3. Voltage/time curves for the early stages of anodizing at 10 mA/cms in sulphuric
acid at 20°C. a, 1.5 M; b, 1.0 M; c, 0.5 M.

General film morphology

Figure 4 displays the main morphological features of porous film growth under
relatively mild conditions. The film surfaces reproduced the characteristic electro-
polished finish of the metal surface, Fig. 4a. Pores were not regular in size or packing
on these outer surfaces, although greater regularity was observed in the film sections
where the fracture plane changed level slightly, Fig. 4e. Measurements of pore
diameter obtained from sections tended to depend on the nature of the fracture
sections, so no such values are included. Evidently the pores generally ran through
the entire film, Figs. 4b-d, except for the barely resolvable barrier layer, Fig. 4d,
with only occasional pore termination or branching. The bottom of the film revealed
an approximate but imperfect close-packed, hexagonal distribution of cells, Fig.
4f. Upon long exposure times, particularly for low cd (low voltage) films, general
dissolution tended to obscure the typica porous pattern by causing pore merging and
surface roughening.

Infsuenceof anodizing time

Figure 5a presents histograms illustrating that the distribution of pore diameters
on the outer surfaces for several anodizing times was normal in each case, in general
agreement with Zalivalov et al.8 400 measurements were made on each plate from
surface replicas, using a travelling microscope. With increased anodizing time
(and hence film thickness), the range covered by the diameter distribution increased
and the number of pores with the most frequent diameter decreased, due to chemical
dissolution by the solution of the pore walls very near the outer surface. Figure 5b
shows clearly that the most frequent diameter at the film surface increased with
anodizing time due to dissolution by the acid and this must always be considered
when evaluating pore dimensions measured on Clm surfaces. Extrapolation of the
curve in Fig. 5b to zero anodizing time gave 120-140 A for the pore diameter, prior
to enhancement by continued exposure. Figure 5b also essentially gives outer pore
diameter as a function of film thickness because, despite the outer pore enlargement,
FIG. 4. Typical replica electron micrographs of films formed by anodizing aluminium
in 1.5 M acid at 20°C.
a, Surface of 45-pm film formed at 25 rnA/cms (18 V). b, Section of film shown in (a).
c, Film shown in (a), displaying surface and section near outer surface. d, Film shown
in (a), displaying section near barely resolvable barrier layer and pore base pattern.
e, Section of 45-pm film grown at 80 mA/cm’ (23-5 V), showing where fracture section
changes level, so displaying some pores in section. f, Bottom surface of film similar
to that shown in (a). g, Surface of film grown for 140 min at 10 rnA/cms (14.3 V)
showing pore merging.
FIG. 6. Replica electron micrographs of films formed at 25 mA/cm2 in 1.5 M sulphuric
acid at various temperatures for 57 min. Surfaces of films grown at a, 0°C; b, 27°C;
c, 31°C; d, 35°C. Sections of the films at the oxide/solution interface grown at: e, 0°C;
f, 31°C.
FIG. 7. Transmission electron micrograph of film grown for 30 s at 25 mA/cma in
in l-5 M sulphuric acid at 20°C.

FIG. 8. Transmission electron- micrographs of films grown for 45 s at 10 mA/crrP in

H,SOJNa,SO, solutions, M in SO,‘-, at 20°C.
a, pH 6-7; b, 4-O; c, pH 2-O; d, pH 1-l.
FIG. 9. Scanning electron micrographs of pitted regions of a specimen anodized for
5 min at 10 mA/cma in H,SO,-Na,SO, solution, pH 2.0, 20°C.
a, Plan view at low magnification, showing several pits. b, Tilted view of single pit,
showing anodic film curling up at the edge of the pit.
 The anodizing of aluminium in sulphate soIutions 1869

Pore diometer, a

3
250-

I I I I I I
IO 20 30 40 50 60 3
Anodizing time, min

FIG. 5-a. Distributions of pore diameters on the surfaces of films grown at 25 mA/cm’
in 1.5 M sulphuric acid at 20°C for a, 15 min; b, 25 min; c, 35 min; d, 45 min; e, 57
min, from a total of 400 pores (data from replicas of film surfaces).
b. Average pore diameter 011 the surfaces of films grown at 25 mA/cmB in l-5 M
sulphuric acid at 20°C as a function of anodizing time.

macroscopic film thinning was almost negligible under these particular conditions,
the effective film growth rate being approximately O-88 pm/min.

hjkmce of solution temperature

Forming voltages, and film thicknesses determined by metallographic cross-
sections, are presented in Table 1 for films formed in 57 min at 25 mA/cma in 1.5 M
sulphuric acid at various temperatures, conditions sufficient to produce a 50 pm
film at 20°C. Clearly, increasing the temperature reduced the bulk film thickness
(reproducible to within 5 %), and also the barrier-layer thickness because the cell
voltage fell.
Replicas from film surfaces showed that pores produced at 0°C were difficult to
resolve, due to the general smoothness of the surface, Fig. 6a. Cracks visible to the
naked eye tended to appear as the films warmed up to room temperature. Pores
1870 G. C. WOOD and J. P. O’SULLIVAN

TABLE 1. FORMING VOLTAGE, FILER THICKNESS AND PORB DIAMETER AS A FUNCTION

OF TEMPERATURE FOR ANODIZING FOR 57 MKN AT 25 ld/CIll’ M l-5 M &SOP.
PORE DIAMETERS FROM SURFACE REPLICAS

Temperature Voltage Film thickness

“C V pm pore dk=e”
2: 23
18 53.6
50.1 245
170

27 15 45-o 300
31 14 39.2 -
35 13 26% -

formed at 20°C have already been shown, Fig. 4a, but at 27°C the pore diameter
increased and adjacent pores were merging due to the increased dissolution, Fig. 6b.
At 31 and 35°C greatly enhanced bulk dissolution at the outer surface of the film
occurred, producing rough surfaces obscuring discrete pores. Cracks, not visible
optically after removal from the anodizing bath, appeared in the film surfaces when
they dried out, due to the collapse of the more porous outer regions.
In section, the cells were larger at O”C, Fig. 6e, than at 31”C, Fig. 6f, consistent
with the forming voltages established, although the pore dimensions at the outer film
surface were increasing with temperature through this range, Table 1. Additionally,
the cell structure became more irregular at the higher temperature. Figure 6f indicates
that the more severe solution had a generally disruptive effect on the film structure
but increased the actual pore diameter only at the immediate film surface. As seen
below, the reverse effect should be true next to the barrier layer, where the pore
diameter is directly proportional to the voltage established.

Influence of cd
Films were formed for 30 s at various constant cds under otherwise standard
conditions, the forming voltage being allowed to take up its equilibrium value in each
case. Transmission electron micrographs of such films, Fig. 7, enabled an easy
estimation of the pore diameters before any appreciable surface widening occurred.
Pore diameter distributions similar to those in Fig. 5a were obtained and the values
corresponding to the peak of these distributions are compared with forming voltage
in Table 2. Evidently, contrary to previous theories, l but in complete agreement with
detailed studies of phosphoric acid anodizing, l2 the pore diameter was directly pro-
portional to the forming voltage, the pore-diameter/forming-voltage ratio being
9-10 A/v.

T&LB 2. PORE AND CELL DIAMETERS AND PORE POPULATION AS A FUNCTION OF FORMING
VOLTAGE (MEASUREMENTS FROM TRANSMISSION ELECTRON MICROORAPHS). STANDARD
ANODIZING CONDITIONS

Current Cell Pore Pore-diameter/ Pore Cell Cell-diameter/

density voltage diayter forming-voltage population forming-voltage
mA/cm= v A/V number/cmB diaYter &V

;: 17.8
14.5 160
145 10.1 X1.3
61.1 x loo
10’ 400
460 28.0
25-9
50 19.6 180 ;:; 45.2 x 10D 535 27.3
80 23.5 215 9.1 41.7 x 100 585 24.9
 The anodizing of alutium in sulphate solutions 1871

Pore counts were also made on transmission micrographs of 25 cm2 area, corre-
sponding to about 100-150 pores. Several estimates were made in this way for each
specimen and the mean value taken. By assuming a cylindrical oxide cell about each
pore, an estimate could be made of the cell diameter from the number of pores (and
thus cells) on a given film area. The assumption of a cylindrical cell gave a cell diameter
intermediate between those of the inscribed and escribed circles of the hexagonal cell,
leading to a possible error of not greater than 8%. ‘Again, the cell diameter was
proportional to the forming voltage, J the mean ratio was 26.5 A/V, Table (2),
compared with 27-7 for films formed in phosphoric acid.12 The difference may well
have been within experimental error.

Inzuence of solution pH and concentration

Films were formed at 10 mA/cm2 in the 1 M H2S04/NaZS04 mixtures (constant
anion strength, variable but controlled pH) at 20°C for 45 s for transmission electron
microscopy and for 5 mm for optical and scanning electron microscopy. Figure 8
shows that as the pH was increased from 1.1 the films became increasingly barrier-
type. At pH 6.7 the smooth film merely replicated the metal surface, Fig. 8a, but as
the pH was reduced the films became rougher, due to dissolution by the solution.
At pH above 1.1 the chief reaction occurring after the steady-state voltage had been
reached was pitting by dissolution of the aluminium metal. No pitting was observed .
at pH 1.1.
Figure 9a is a plan scanning electron micrograph of a pitted region, whereas
Fig. 9b presents a tilted view of an individual pit. The oxide film can be seen curling
up around the edges of the pit and the metal is left apparently in direct contact with
the solution. No pitting occurred at and below pH l-l.

TABLET. VARIATION
OF FORMING VOLTAGE AND PORE DIAMETER WITH ACID CONCENTRATION
PROM TRANSMISSIONELECTRONMICROGUPHS). ANODIZING AT 10 mA/cm*
(MEASUREMENTS
AND 20°C
Solution Pore diameter/
concentration Solution Forming voltage Pore diameter forming-voltage
M PH V BL &V
0.5 0.3 23.5 230 9.8
1.0 0.06 19.0 180
1.5 -0.15 14.3 145

Transmission electron micrographs were also obtained for films anodized for 45 s
at 10 mA/cm2 in 0.5, 1-O and 1.5 M sulphuric acid at 2O”C, yielding the pore data in
Table 3. The pH values were calculated using activity data, on the assumption that
HSO, was produced by the ionization reaction. The aim was to maintain constant
pH and change the HSO, ion content; however the two factors cannot be isolated,
so the results are best interpreted on the basis of increasing suIphuric acid concen-
tration. An increase in the aggressive nature of the solution clearly reduced the
barrier-layer thickness, because the forming voltage fell. The pore size also decreased,
which would be expected on the basis of the pore diameter being proportional to
forming voltage. The constant of proportionality appeared to be slightly higher than
calculated in the earlier experiments, although pore diameter showed a definite
relationship to cell voltage.
 1872 G. C. Woor, and J. P. O’SULLIVAN

DISCUSSION
Pore and pit initiation
In considering pore initiation in the initial barrier layer, it is pertinent to compare
the present work at constant cd with that of Hoar and Yahalom6 at constant voltage.
They deduce from current/time transients that the initial step of barrier-layer growth
is essentially the same in solutions giving either barrier or porous films. Pore initiation
is field-inhibited and is only observed in acids after a few seconds, when the field has
fallen to a low value. In the present work, deviation from the linear voltage/time
curve at constant cd also occurs after only a few seconds but it seems likely, from the
roughening of the surface detectable in the electron micrographs, that dissolution
starts from the outset but is not readily detectable electrically. The average field is,
of course, constant at constant cd, although it seems probable that locally enhanced
values will be achieved at certain structural irregularities. Whether the roughening
is due to uneven growth or dissolution as a result of local field variations, or to differ-
ences in film structure and hence solubility, cannot yet be stated. However, in phos-
phoric acid anodizing, l2 the final pore pattern results from a merging of such regions
of surface roughening (‘nuclei’), which appear related to the metal sub-grain bound-
aries and possibly other sub-structural features (genuinely related to the metal or
produced by electropolishing), and subsequent concentration of current at the
. locally thin regions. It does not appear necessary under constant cd conditions to
invoke concepts of field-inhibited, localized proton entry to explain the hexagonal
pore distribution.
Considering Fig. 1, and bearing in mind that films growing at 80 mA/cms are
thickening about 8 times as fast as those growing at 10 mA/cm2, even though the
barrier layers stay essentially constant in thickness with time, it follows that film
dissolution at the pore bases must be much faster in the former case. This is a strong
indication that field-assisted dissolution 16,26of the oxide is occurring at the pore bases,
as suggested by Hoar and Mott,la since the solution conditions are nominally identical
in each case. Local Joule heating, which would be higher at higher cds, is an alter-
native explanation but, although it may be a contributory factor, it is not considered
to be the sole effect. Thus, it is perhaps unlikely that stagnant solution would be
permitted in the very shallow pores of the thin films examined by transmission, when
the solution is vigorously agitated. Furthermore, Hunter and Fowlels have calculated
that, for anodizing at 20 mA/cm2 in 1.5 M sulphuric acid at 21”C, the actual dis-
solution rate at the pore bases must be 3725 AImin, compared with a chemical
dissolution rate of about 0.84 AImin under these conditions. They estimate that this
would require 50% acid at its boiling point of 124°C at the pore bases to maintain
dissolution at the required rate. Although Applewhite et aZ1s*2ohave measured
overheating of the metal (0.1 mm wire) and hence of the film, by 2-3°C at 10 mA/cma,
5-6”C at 20 mA/cm2 and 20°C at 80 mA/cm2 for barrier-type films growing at 200 V
in ammonium borate solutions, and have indeed suggested that temperature increases
cause pore growth in the rather different case of borate anodizing, the temperature
rise seems insufficient to be the major factor in the present case, even ahowing for
local heating effects at pore bases, if in fact these could occur. Finally Murphy and
Michelsonl’ have pointed out that if such high temperatures existed at the pore bases,
it would be unlikely that minor changes in bulk solution temperature would produce
significant changes in film morphology, which they are observed to do. It could be
argued that film material produced at high cd is more amenable to dissolution,
 The anodizing of aluminiumin sulphate solutions 1873

perhaps due to differences in anion incorporation. However, data for oxalic, sulphuric
and phosphoric acids under various conditions show the cell and pore A/V ratios
to depend very largely on voltage and much less on anion content or hydrogen
bonding. l2 Effects of anion incorporation and hydrogen bonding upon the formation
field and ti-dissolution rate would therefore appear of secondary importance to the
massive effect of field-assisted dissolution of contaminated alumina, particularly if
the temperature is also somewhat enhanced at the pore bases.
The absence of a peak in the voltage/time curve, the lower initial barrier-layer
thickness and lower steady-state barrier-layer thickness for anodizing in 1.5 M acid,
Fig. 3, must be due to the increased dissolving power of the acid under the field, as
compared with more dilute acid. The films are forming at the same rate and under
the same field in all three acids used, so the margin between the growth and dissolution
reactions in less in the more concentrated acid.
As the pH in the H2S04/NaS04 mixtures increases, pitting competes with pore
formation and eventually replaces it as the phenomenon limiting barrier-layer thick-
ening. Thus at pH 1.1 there is no pitting, at pH 2-O and 4-O some surface roughening
or pore initiation as well as pitting occurs, whereas at pH 6.7 pitting occurs in an
essentially barrier-type film. According to the work of Richardson and Wood21 on
pitting in halide solutions, pitting occurs by metal dissolution at the bases of flaws in
the anodic film, eventually undermining the film, as shown in Fig. 9. Pitting does not
occur here in the highly acidic solutions partly because the pores, occupying about
10 % of the total specimen area, pass much of the current, preventing it being con-
centrated in the flaws and enlarging then into pits. More especially, they prevent the
voltage imposed across the film, and particularly the electrode potential of the metal
at the flaw bases, and the cd concentrated there, rising sufficiently high for the flaws
to propagate quickly into pits. It is interesting that a high steady-state voltage is
maintained even under reasonably severe pitting conditions; presumably this is a
compromise voltage, sensed by the instrument, between the voltage across the film
and that down the pits. Some film thickening may also be continuing. If very dilute
spent sulphuric acid is used,22 or anodic oxidation is performed in 3 % w/v ammonium
tartrate solution containing 0.01 M sodium chloride,23 the voltage drops to a much
lower level, due to extensive pit development whilst the voltage is initially high.
Excursion to higher pH in the absence of chloride produces spongy films without
discrete pits or pores.24
Curves for electropolished 99.99% and zone-refined aluminium were similar in
the various H2S04/NaZS04 solutions. However, when the metal was not electro-
polished the cell voltage remained at a high value for 5 min, which would appear to
show that the physical state of the surface, controlling the number of mechanical
flaws,21 rather than the purity per se, influenced the pit initiation most significantly.
It is noteworthy that many of the weak organic acids tend to give high voltage
films with relatively large pores and pitting in certain circumstances.

Generaljilrn morphology
The model of thick porous anodic fdms due to Keller ef all is confirmed in principle
but not in certain important details. In particular, the steady-state pore diameter is
proportional to the forming voltage, in the same way as the barrier layer and cell
diameter. No evidence of pores with star-shaped cross-sections has been found.
1874 G. C. WOOD and J. P. O’S~LL.~AN

Increased solution aggressiveness, due to higher acid concentration or temperature,

and proloyqjed exposure, progressively enlarge the pores by simple dissolution where
they emerge at the film surface, until eventually they tend to merge and the surface
becomes rough. Whilst pore-diameter measurements from film sections are neither
reproducible nor exact, there was never any indication of pore widening with increas-
ing distance from the metal, confirming one previous suggestion* but refuting
others.1°*15 This means that pore-wall dissolution occurs significantly only at the
extreme outside of the film even in the more aggressive solutions. The pores are
therefore “trumpet-shaped” with the main section of virtually uniform diameter.
Probably the solution in the very narrow pores becomes relatively saturated with
AP+ ions, so dissolution is slow except at the pore bases, where field-assisted dissolu-
tion is active in enhancing dissolution, and at the outer surface where fresh acid is
readily supplied.
Despite no increase in pore diameter, the more severe solution does lead to a
spongier, more distorted and irregular film, Fig. 6, which may in turn affect the bulk
dissolution rate. This aspect is consistent with the model of Murphy and Michelsonl’
in the sense that the “internal surfaces”, surrounding the micro-crystallites containing
anions and hydrogen-bonded material, would probably be more open, explaining the
superior dyeing characteristics etc, but would have no great effect on the mechanism
of growth, which is determined at the pore bases. This general “loosening” ofmaterial
in the cell walls, producing “open” structures, could permit collapse and cracking on
drying, as was indeed observed. Ir spectra indicated that all the films are a relatively
open array of largely anhydrous alumina crystallites, invisible in the electron micro-
scope but readily permeated by molecular water, the surfaces of the crystallites
containing hydroxyl ions or groups and probably acid anions, and the system being
hydrogen-bonded.ll

Influence of formation conditions on film parameters

Considering the different anodizing conditions and techniques, the modest agree-
ment between the present cell diameters and those of Keller et, uZ,l Paolini et uP”
and Booker et L-@ may be considered reasonable. The visible porosity for an 18 V
film in the present work is 12.1% (pore diameter 160 A and cell diameter 460 A),
in close agreement with the figure of 13 % quoted by Booker et al.
Keller et al, using a massive, unreliable extrapolation, obtained indirectly a pore
diameter of 120 A for 15 M sulphuric acid at lO”C, which they assumed to apply to
any film grown in sulphuric acid, regardless of voltage. Cosgrove16 calculates pores
of similar dimensions from his adsorption studies for anodic films formed at various
voltages in 15 M sulphuric acid at 19°C. Paolini et allo used adsorption techniques
on films grown in 20% w/w acid at 20°C. They find pores generally larger than 120 A
and suggest that this was due to film dissolution. Zalivalov et UP measured pore
diameters directly from the surfaces of films grown at 25 and 50 mA/cm2 (22 and
25 V respectively) in 4 N sulphuric acid at 0-5”C, and at 10 mA/cm2 (10 V) in the same
acid at 2O”C, claiming 120 A for the most common pore diameter in each case. Each
study appears to invoke, or be influenced by, the assumptions of Keller et al. Booker
et aZ2measured pore diameters directly from film surfaces and by a relatively accurate
sectioning method, yielding 150-200 A from sections of fihns grown at 50 r&&m2
 The anodizing of aluminiumin sulphate solutions 1875

in 17-5 v/v acid at 20°C and 250-300 A on the surfaces of films anodized at 100 mA/
cm% for 1 h. This last figure compares with 250 A after 1 h in 1.5 M sulphuric acid
at 25 mA/cm2 here. In the present work, for anodic oxidation at 25 mA/cm* (17*8-
18 V) in 1.5 M acid, the pore diameter from extrapolation to zero time of the surface
replica results yields 120-140 A, whilst transmission micrographs of thin films give
160 A. These “ideal” values, avoiding enlargement effects by subsequent electrolyte
contact, agree quite well with the earlier results.
The lack of dependence of pore diameter (measured on film surfaces) upon anodiz-
ing conditions found by Zalivalov et UP is surprising. It may be explicable because
they used low cds, and hence voltages, in aggressive conditions but higher voltages in
their hard anodizing conditions. Thus, pore enlargement due to chemical dissolution
on the surface in the first case may have occurred to match the pore sizes in the second
case, which would be determined more by the higher forming voltage.
There is, in fact, some evidence of the dependence of pore diameter on forming
voltage in the data of Cosgrove16 and particularly of Paolini et aLlo Thus, before
eventually introducing their truncated cone pore model they calculated their adsorption
data and pore counts from electron microscopy on the basis of a cylindrical pore with
no taper. For anodizing for 30 min in 20 % w/w sulphuric acid, values of 140, 15 1
and 165 A are deduced for the pore diameter at 10, 15 and 25 mA/cm2 respectively
(corresponding to forming voltages of 11.9, 14.1 and 16.1 V). Pore-diameter/voltage
ratios of 11.7, 10.7 and 10-2 are therefore obtained, which agree moderately well
with the present study.
In the present study, higher ratios of 9.8, 9.5 and 10.1 A/V for 0.5, 1-O and 1.5 M
sulphuric acid respectively are found than for various cds in 1.5 M acid. This must
be related to the influence of acid concentration and cd on the steady-state barrier
layer thickness. It is shown elsewherea that the barrier-layer thickness, determined
by the interplay of field-assisted film growth and field-assisted film dissolution at
the pore bases, determines both the cell and pore diameters_

Acknowledgeme~fs-Thanks are due to Professor T. K. Ross for the provision of facilities and for
his interest, and to Mr. B. Vaszko for assistance with the electron microscopy.

REFERENCES
1. F. KELLER, M. S. HUNTER and D. L. ROBINSON,J. efectrochem. Sot. 100,411 (1953).
2. C. J. L. BOOKER, J. L. Wool and A. WALX-I, Br. J. uppl. Phys. 8,347 (1957).
3. T. A. RENSHAW, J. electrochem. SOC. 108, 185 (1961).
4. H. GINSBERG and K. WEFERS,MetaN. 17, 202 (1963).
5. P. CSOKAN, Trans. Inst. Metal Finish. 41,51 (1964).
6. T. P. HOAR and J. Y AHALOM, .K eiectrochem. Sot. 110, 614 (1963).
7. R. W. FRANKLIN and D. J. STIRLAND, J. electrochem. Sot. 110, 262 (1963).
8. F. P. ZALIVALOV, M. N. TY~KINA and N. D. TOMA~HOV,Zh. rass. fiz. khim 35,429 (1961).
9. H. AKAHORI, J. Electron. Microsc. Japan 10, 175 (1961).
10. G. PAOLINI,M. MASOERI,F. SACCHI and G. PAGANELLI,J. electrochem. Sot. 112,32 (1965).
11. J. P. O’SIJLLI~AN, J. A. HOCKEY and G. C. WOOD, Trans. Faraday Sot. 65, 535 (1969).
12. J. P. O’SULLIVAN, Thesis, University of Manchester (1968).
13. G. C. WOOD, J. P. O’SULLIVAN and B. VASZKO, J. electrochem. Sot. 115, 618 (1968).
14. G. C. WOOD and J. P. O’SULLIVAN, J. electrochem. Sac., 116, 1351 (1969).
15. L. A. Cos~~ove, J.phys. Chem. 60,385 (1956).
16. T. P. HOAR and N. F. MOTT, J. Phys.Chem. Solids 9,97 (1959).
17. J. F. MURPHY and C. E. MICHELSON,Proc. Co@ on Anodizing Aluminium, Nottingham, 1961,
p. 83. Aluminium Development Association, London (1962).
18. M. S. HUNTEIIand P. E. FOWLE, J. electrochem. SOC. 101, 515 (1954).
19. F. R. APPLEWHITE, J. S. LL. LEACH and P. NEUFELD, Corros. Sci. 9, 305 (1969).
1876 G. C. WOOD and J. P. O'SULLIVAN

20.J. S. LL. LEACHand P. N~FELD, Corms. Sci. 9,413 (1969).

21. J. A. Fu-SON and G. C. WOOD, Corros. Sci. 10,313(1970).
22. G. C. WOOD and G. BAILXY, unpublished work.
23. A. J. BROCK, Thesis,University of Manchester (1965).
24. A. J. BROCK and G. C. WOOD, Electrochim. Acta 12,395 (1967).
25. J. P. O’SULLIVAN and G. C. WOOD, Proc R. Sot., in press.
26. hf. NAGAYAMA and U. TAMURA, Elecbochim. Acta 12, 1097 (1967); 13, 1773 (1968).