Corrosion Science 52 (2010) 848858

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Pitting of zinc: Observations on atmospheric corrosion in tropical countries

I.S. Cole a,*, W.D. Ganther a, S.A. Furman a, T.H. Muster a, A.K. Neufeld b
a
CSIRO Materials Science & Engineering, Highett, Victoria 3190, Australia
b
BlueScope Steel Ltd., Port Kembla, New South Wales 2505, Australia

a r t i c l e i n f o a b s t r a c t

Article history: A ve-nation study has investigated the mechanisms and rates of the atmospheric corrosion of zinc and
Received 31 July 2009 steel in tropical regions in Australia, Thailand, Indonesia, Vietnam and The Philippines. For the study, 18
Accepted 6 November 2009 exposure sites encompassing severe marine, marine, severe industrial, industrial, marine/industrial,
Available online 11 November 2009
urban and highland environments were established across the countries. At each location, zinc and steel
plates were exposed for periods of three months and one year, and measurements were taken of a wide
Keywords: range of surface-response and climatic parameters, including gaseous SOx and NOx, airborne salinity, rel-
A. Zinc
ative humidity (RH) and temperature, rainwater composition, surface temperature and time of wetness
A. Steel
C. Pitting corrosion
(TOW). Exposed plates were used to determine mass loss, the nature of corrosion products (using FTIR
C. Atmospheric corrosion and SEM-EDS) and the morphologies of corrosion layers (via SEM-EDS). Regression analysis indicated that
C. Oxide coatings the prime factors controlling zinc corrosion rate were climate (temperature and rainfall) and surface-
response (TOW), and surprisingly not pollutant levels, despite signicant variation in SOx levels across
the sites. SEM studies indicated the presence of pitting below the oxide layers on zinc, particularly those
plates exposed at marine and other sites with relatively low SOx levels. In contrast, no pitting was
observed (or pits had very low aspect ratios) in the specimens exposed at sites with high SOx levels.
The possible processes leading to the observed damage patterns are discussed.
Crown Copyright 2009 Published by Elsevier Ltd. All rights reserved.

1. Introduction
A study of the atmospheric corrosion of zinc and steel has been
made across ve countries (Australia, Thailand, Indonesia, Vietnam
and The Philippines), with the aim of gaining an understanding of
the relationships between pollutants and the corrosion of steel and
zinc in tropical countries, and in particular to assess:
1. The forms of damage observed on zinc in diverse environments.
2. The role of atmospheric pollutants on the corrosion rate of zinc.
The study found that oxide layers of varying thickness formed
on zinc specimens in the diverse environments, and that signicant
pitting occurred beneath the layers that were thick. While it is gen-
erally considered that the atmospheric corrosion of zinc occurs by
general corrosion, in fact there have been observations of pitting in
the atmospheric exposure of zinc. Dunbar and Showak [1]
observed shallow pits (i.e. diameter greater than depth) in atmo-
spheric exposures of high-purity zinc, but found that while the
depth of pits increased with time, their aspect ratio (depth/width)
decreased.
* Corresponding author. Tel.: +61 3 92526000; fax: +61 3 92526244.
E-mail address: Ivan.Cole@csiro.au (I.S. Cole).
On the other hand, pitting has frequently been observed in zinc
that has been immersed, particularly in distilled [2] or hot water
[3], but also in other solutions [4]. From their observations of pit-
ting in distilled water, Evans and Davies [2] argued that pitting
was associated with gravitational sinking of solid corrosion prod-
ucts. They found that when corrosion products precipitated and
settled in a crevice on a zinc surface, an oxygen-depleted area
was created between the deposited product and the zinc surface.
This oxygen-depleted area then acted as a local anode and pro-
moted the initiation of pitting. Similar observations were noted
by Feitknecht [5] who noted that electrochemically polished zincs
would remain shiny and uncorroded for weeks upon immersion
into salt solutions. For zinc immersed in solution, it was found that
pit density increased with the halide ion concentration in the solu-
tion [26]. Muster et al. [7] observed that rolled zinc that had been
treated to develop an enlarged grain size developed pits when ex-
posed to saline solution.
The occurrence of pitting under the three varying solutions is
due to different mechanisms. In distilled water, the solution resis-
tance is very high and thus the local anodes and cathodes are close
together and localized. The cathodic zones close to the anodic
zones subsequently become passivated due to oxide growth pro-
moted by the high pH (at the cathodic zones) [2]. Acid dissolution
of oxides does not occur, as the hydroxide produced at the cathode
neutralizes the acid produced at the anodes. In hot water, whole

0010-938X/$ - see front matter Crown Copyright 2009 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2009.11.002
 I.S. Cole et al. / Corrosion Science 52 (2010) 848858
surfaces become passivated and pitting is associated with the
breakdown in this passive lm. Furthermore, when corrosion oc-
curs through a passive lm, it is promoted by the acidication of
pits due to the formation of zinc hydroxides [2] as indicated below:
Zn ! Zn2 2e 1 zinc specimens exposed at tropical sites across Australia and South
Zn2 2H2 O ! ZnOH2 2H 2 East Asia, with the aim of highlighting those environments where
When the anodes and cathodes are well separated (in pure water),
acidication may spread out from the anodes and dissolve pre-
existing oxides, thus promoting general corrosion. However, where
an oxide layer is thick, acidication will remain localized at the ini-
tial oxide defect and pitting can occur. This is the basis of the pit
theory of Galvele and co-workers [711], whereby anodic processes
lead to acidication, and a sufciently acidic solution is required to
prevent repassivation. However, as cations move towards cathodic
sites to ensure electroneutrality, acidic species will be transported
out of a pit, and thus the maintenance of a pit depends on the bal-
ance between the corrosion current and transport processes. If the
corrosion current is such that more acidic species are produced
(by Eqs. (1) and (2)) than are transported out of a pit, then pitting
will continue. On the other hand, if the corrosion current is insuf-
cient to counter the transport of acidic species and neutralizing
reactions within the pit, then the pH will rise and repassivation will
result. The number of ions transported out of the pit with depends
on the transport length (i.e. pit depth), and thus Galvele [9,10] for-
mulates his solution species concentration dependence (including
[H+]) as a function of x.i, where x is the pit depth and i is the current
density.
In chloride-containing solutions, pitting only occurs on passiv-
ated surfaces [57], with specic adsorption of chlorides onto the
passive lm (under appropriate potentials) leading to local break-
down of the lm. If the chloride concentration is high, general lm
breakdown and uniform corrosion can occur.
Thus the pitting of zinc has been reported previously, however
it has not been reported to be taken into account for models of the
atmospheric corrosion of zinc [1214], even though such models
do take into account the formation and dissolution of oxide layers.
Turning to the effects of atmospheric pollutants on zinc corro-
sion, it has long been established that industrial atmospheres can
cause enhanced corrosion [13,1517], with many studies in the
northern hemisphere observing a strong correlation between cor-
rosion rate and atmospheric SO2 level. Indeed, there is a theoretical
basis for such a relationship, as the absorption of SO2 should pro-
mote the acidication of moisture lms. More recent work [18
20] has highlighted the importance of nitric oxide and ozone in
the corrosion process, and it has been postulated that these gases
have a synergistic effect with SO2. Work by Graedel [14] and Far-
row et al. [19] has highlighted the role of the various chemical
reactions that occur in the aqueous phase between sulphur prod-
ucts and water. Work by Cole [21] indicates that much of the syn-
ergistic effect of other atmospheric gases may be attributed to their
role in oxidizing S(IV) to S(VI), and that alkali precursor gases (such
as NH3) may have a contrary effect to SO2. Thus, the pH and aggres-
sivity of moisture layers depends on a wider range of factors than
simply SO2, so that atmospheres with the same SO2 may have con-
siderable variation in their corrosivity [21].
The form of zinc degradation in polluted tropical environments
is particularly interesting, as in such environments, industrial and
marine aerosols may have acid pHs. A review by Cole et al. [22] has
highlighted that industrial or marine aerosols that had been acidi-
ed by the absorption of SO2 or acidic gases, could have pH levels
of from 1.5 to 3.5. Thus the deposition of aerosols are likely to lead
to the localized dissolution of oxides. Graedel [14] has highlighted
that zinc oxides and hydroxides are amphoteric and that, under
acidied rainwater (or aerosols), such oxides are likely to dissolve.
849
One of the aims of the present study was to investigate whether
generalized corrosion still occurs in situations where localized
oxide dissolution may occur and, if pitting does occur, how the
local environment inuences it.
This paper will examine the patterns of damage observed on
the prime mechanism of corrosion (below oxide layers) is pitting,
and providing insites into how environmental factors may promote
pitting.
Developing an accurate understanding of the mechanism of zinc
corrosion is of signicant importance in enhancing the perfor-
mance of galvanising steel both through the development of more
resistant alloys and passivation systems and in developing better
materials selection tools. Given an accurate understanding of
how the physical environment promotes the degradation of oxide
layers and subsequently the underlying zinc metal, it should be
possible to develop intervention strategies that restrict processes
promoting this degradation. These strategies could include alloy
changes, new conversion coatings or additions to naturally or arti-
cially developed oxide layers. Further accurate materials perfor-
mance predicts, such as corrosion models [12], performance tests
which may be embedded in both standards and industrial guide-
lines can ensure that the right coating is used in the right environ-
ment. Resolving the issue of whether zinc corrodes by a pitting or
general mechanism and how this relates to the external environ-
ment is not sufcient to deliver detailed knowledge for alloy
design or performance evaluation but it is a necessary step.
2. Experimental
2.1. Exposure conditions
The location and the characteristics of each exposure sites are
given in Table 1. The monsoonal periods that affect the studied re-
gions are operative from May to September/October in Thailand,
Indonesia and Vietnam, June to August in The Philippines, and
December to February in Australia.
At each site, airborne salinity was measured over three month
periods using a standard wet candle technique according to ISO
9223, while NOx, SOx and HNO3 were measured using a passive
sampling technique [23]. Time of wetness (TOW) was measured
using gold grids mounted on zinc specimens, using the procedures
developed by Ganther et al. [23]. Relative humidity and tempera-
ture was measured using integrated humidity and temperature
sensor (Vaisala). Rainfall was measured using electronic tipping
bucket rain gauges. Rainwater pH was measured monthly on col-
lected rainwater using an electronic pH meter (Hanna). All micro-
climatic and pollutant parameters were measured at a height of
approximately 1.5 m above ground level.
Mass loss data were determined from 1  100  150 mm zinc
specimens with the composition given in Tables 2 and 3  100 
50 mm (density 7.764) mild steel (containing Cu) specimens with
the composition given in Table 2. In general, exposure congura-
tions were in accordance with ASTM G1, with one exception
rather than the plates being faced north or south, they were faced
in the direction of the prevailing sea wind at each site.
2.2. Fourier transform infra-red (FTIR) spectroscopy
Infra-red analysis was carried out using a Bruker Equinox 55
FTIR spectrometer. The infra-red spectra of oxidation products
formed on plate specimens were obtained using a Praying Mantis
(Harrick) diffuse reectance cell, using polished zinc as a back-
ground. Seasonal and annual exposure specimens were subjected
850 I.S. Cole et al. / Corrosion Science 52 (2010) 848858

Table 1
Locations and characteristics of Australian and South East Asian exposure sites.

Location Type Latitude/longitude Salient features

Australia
Cowley Beach Severe marine 17410 S; 146060 E On beach facing SE
Walkamin Highland 17080 S; 145250 E On tableland NE of Cowley Beach
Philippines
Cabuyao Industrial 14100 N; 121070 E Large industrial centre
Bicutan Urban 14300 N; 121020 E Metro Manila
Bagiou Highland 16300 N; 120330 E 1500 m above sea level
Thailand
Bangkok Urban/industrial 13400 N; 100300 E On building near major road
Rayong Industrial 12410 N; 101190 E 500 m from sea
Phrapradaeng Industrial 13390 N; 100340 E Adjacent to river 10 km from sea
Phuket Marine 7500 N; 98200 E On seashore
Indonesia
Lembang Highland 648.60 S; 10736.60 E Highland central Java, 62.7 km from coast, 1100 m high
Jebus Marine 139.60 S; 10519.80 E Bangka Island off Sumatra, 10 m from coast
Mentok Industrial/marine 24.80 S; 10510.80 E Bangka Island off Sumatra, 25 m from coast
Gresik Coastal/industrial 79.60 S; 11240.80 E West Java, 300 m from coast
Cikampek Inland rural 624.60 S; 10727.60 E Java, 25 km from coast
Vietnam
Hanoi Urban 21010 N; 105520 E 100 km from coast
Hue Urban 16280 N; 107360 E 5 km from coast
Nha Trang Marine 12200 N; 109000 E 10 m from coast
Ho Chi Minh City Urban/industrial 10460 N; 106430 E 56 km from coast

Table 2
Composition (mass %) of metal specimens.

Metal C P Mn Si S Ni Cr Mo Cu Ti Fe Cd Pb Sn Al Zn
Zinc 0.064 0.006 0.003 0.002 0.001 0.001 <0.001 Remainder
Steel 0.166 0.004 0.602 0.057 0.002 0.285 0.13 0.01 0.235 Remainder

to FTIR examination using the techniques discussed by Neufeld and
Cole [24].
2.3. SEM analysis and mapping
Scanning electron microscopy (SEM) imaging and mapping, and
energy-dispersive spectroscopy (EDS) were performed using a
Philips XL30 FESEM with a LinkISIS X-ray analysis system (Oxford
Instruments). EDS was carried out with a beam current of 5 kV. All
samples were mounted using conductive carbon tape.
3. Results
3.1. Mass loss and environmental conditions
Table 3 presents the pollutant parameters, TOW and mass loss
data for 1-year exposures at all sites. The ISO-TOW data are de-
rived from the ISO 9225 approximation of TOW, which is when
the temperature is greater than 0 C and RH is greater than
80%. The rainwater pH values given represent the average pH of
rainwater samples for a year. The mass loss is reported as an
average mass loss as this is standard in the eld, although as re-
ported later in this paper signicant pitting is observed so that
the reported average may not reect the most common or modal
mass loss.
3.1.1. Surface analysis
Species identication was made through a comparison of
spectra obtained after seasonal and annual exposures with those
obtained by Neufeld and Cole [24]. As shown in Table 4, the species
identied were always zinc hydroxy carbonate, chloride or
sulphate (only anions are listed in the table, thus CO3 represents
zinc hydroxy carbonate). SEM was also used to examine cross-sec-
tions of specimens, which, as will be discussed later, invariably dis-
played pitting under an oxide layer. These oxide layers were found
to be either relatively uniform or, in the case of marine exposures,
consist of regions with very different thicknesses. In agreement
with FTIR analysis, EDS indicated that the major elements at all
locations were Zn, O and an additional anion (as dened in Table
4), which was either carbon, sulphur or chloride. The observed ra-
tios in the EDS elemental peaks were consistent with those previ-
ously observed for zinc hydroxy carbonate, chloride or sulphate
[24].
3.2. SEM examination of exposed surfaces
A number of characteristic patterns were observed during the
examination of corroded surfaces. A typical formation on speci-
mens exposed to marine environments was the pedestal, as previ-
ously reported by Neufeld and Cole [25] and presented in Fig. 1.
The diameter of these morphologies were typically about
100 lm, however, sizes ranged from 50 to 250 lm. Their distribu-
tion on the surfaces of specimens was very regular, and in some
cases there were up to three pedestals in an area of 500 
500 lm. EDS analysis indicated that the surface of a pedestal
contained Zn, O and C in ratios typical of zinc hydroxy carbonate.
In other regions, locally dissolved oxide layers (again with an ele-
mental composition of Zn, O and C, consistent with zinc hydroxy
carbonate) were clearly evident, as demonstrated in Fig. 2. Locally
dissolved pits through the oxide with diameters of 25100 lm
were frequent on specimens at all marine exposures (Cowley
Beach, Phuket, Jebus and Nha Trang).
For specimens exposed at industrial sites (Phrapradaeng,
Mentok, Gresik and Bangkok), pedestal formation was not evident
 I.S. Cole et al. / Corrosion Science 52 (2010) 848858 851

Table 3
Climatic, pollutant and mass loss data from tropical sites.

Site Exposure period (dd/mm/yy) Days exposed Season at start of exposure Mass loss ISO-TOW (% time) TOW measured (% time)
(lm/year)
Steel Zinc
Bagiou 12/07/9816/07/99 369 End dry 13.0 2.1 80 94
Bicatun 16/07/9802/09/99 413 End dry 37.1 1.1 45 39
Cabuyao 07/07/9808/07/99 366 End dry 41.3 1.8 52 65
Phrapradaeng 24/06/9825/06/99 366 Wet 40.1 1.5 30 61
Rayong 30/06/9828/06/99 363 Wet 24.9 2.6 33 74
Phuket 02/07/9822/06/99 355 Wet 30.9 1.9 53 67
Bangkok 23/06/9825/06/99 367 Wet 18.0 1.2 33 34
Cowley Beach 20/10/9802/11/99 378 End dry 285.5 7.6 85 94
Walkamin 19/10/9801/11/99 378 End dry 6.8 0.4 71 53
Lembang 15/03/9904/04/00 386 End wet 6.4 0.9 62 49
Jebus 18/03/9910/04/00 389 End wet 37.0 2.0 70 65
Mentok 17/03/9911/04/00 391 End wet 40.1 2.7 64 70
Gresik 19/03/9905/04/00 383 End wet 30.3 2.5 50 64
Cikampek 16/03/9907/04/00 388 End wet 28.6 2.4 52 99
Hanoi 16/01/9820/01/98 365 Wet 34 1.8 61 85
Hue 16/01/9820/01/98 365 Wet 22 2.4 69 88
Nha Trang 16/01/9820/01/98 365 Wet 16 0.7 49 60
Ho Chi Minh 16/01/9820/01/98 365 Wet 36 1.3 47 43

SO2 (lg/m3) Max. ave. daily temp. (C) Ave. temp. (C) Ave. RH (%) Rainwater pH Salinity(mg/m2.day) Rainfall (mm/year)
Bagiou 2 39.0 22 86 5.1 3 3283
Bicatun 21 39.7 27 78 4.2 6 2408
Cabuyao 14 27 75 4.6 6 2408
Phrapradaeng 68 38.2 30 73 6.0 6 1530
Rayong 15 38.1 30 70 6.0 23 1022
Phuket 4 38.0 30 77 5.9 87 2133
Bangkok 21 33.7 29 70 5.6 6 1530
Cowley Beach 0.8 28.7 25 80 5.4 400 3335
Walkamin 0.5 27.7 21 70 4.7 8 1037
Lembang 4 37.6 22 77 5.6 11 2062
Jebus 1 34.9 27 85 6.5 79 2475
Mentok 50 37.0 28 81 4.9 39 2475
Gresik 25 39.6 29 74 5.4 24 2043
Cikampek 19 39.6 28 73 4.8 25 2112
Hanoi 6 27.1 23.5 82 8 1779
Hue 0.5 29.3 25.2 84 6 2256
Nha Trang 12 30.7 26.6 79 13 1127
Ho Chi Minh 14 32.0 27 82 4 1494

Table 4
FTIR of seasonal and annual exposures (In SEM column gures in bold indicate dominant element).

Location Summer Autumn Winter Spring Annual SEM

Thick oxide Thin oxide Pit
Cowley Beach CO3/Cl CO3/Cl CO3/Cl CO3/Cl CO3/Cl C S, Cl Cl
Walkamin CO3 CO3 CO3 CO3 CO3 C C
Cabuyao CO3/SO4 CO3/SO4 CO3/SO4 CO3/SO4 CO3/SO4 C, S, Cl S
Bicutan SO4 CO3/SO4 CO3 CO3/SO4 CO3/SO4 C, S S
Bagiou CO3 CO3 CO3 CO3 CO3 C C
Bangkok SO4/CO3 CO3/SO4 CO3/SO4 SO4/CO3 CO3/SO4 C, S C, S
Rayong CO3/Cl CO3/SO4 SO4/Cl CO3/SO4/Cl CO3 C, S C, S
Phrapradaeng SO4 CO3/SO4 CO3 SO4/CO3 CO3/SO4 C, S C, S
Phuket CO3/SO4/Cl CO3/SO4 CO3 CO3/SO4/Cl CO3/Cl C C C
Lembang CO3 CO3 C, S S
Jebus CO3 CO3 CO3 CO3 CO3 C C Cl
Mentok CO3/SO4 CO3/SO4 CO3/SO4 CO3/SO4 CO3/SO4 C, S S, C
Gresik SO4/CO3 SO4 SO4 SO4 SO4/CO3 C S, C
Cikampek CO3 CO3 SO4 C, S C, S
Hanoi CO3 CO3 CO3 SO4 CO3 C S
Hue CO3 CO3 CO3 C C
Nha Trang Cl/CO3 CO3/Cl CO3/Cl CO3 C, Cl Cl, S Cl, S
Ho Chi Minh CO3/SO4 SO4/CO3 SO4/CO3 CO3/SO4 CO3/SO4 C, S C, S

and surfaces appeared more regular. However, a number of
features were apparent, including islands of sulphate-rich crystal-
line lattice structures, and other formations of dimensions 50
200 lm. Examples of such formations are presented in Fig. 3;
they consist of built up rims with interior zones that contain corro-
sion nodules in a general area of light oxide covering and ne cra-
ters (15 lm). Both the rims and the oxide nodules at the centre of
the formation (shown in Fig. 3) had the elemental composition Zn,
O, C and S, and were likely to be a mixture of zinc hydroxy carbon-
ate and sulphate.
852 I.S. Cole et al. / Corrosion Science 52 (2010) 848858

Fig. 1. Pedestal formation observed on specimen exposed at Phuket.

Fig. 4. Cross-section through predestal formation on specimen exposed at Cowley

Beach.

Fig. 2. Dissolution pits observed on specimen exposed at Nha Trang.

Fig. 5. Cross-section through dissolution zone on specimen exposed at Nha Trang.

Fig. 3. S-containing formations on specimen exposed at Bangkok.

3.2.1. Cross-sections
Cross-sections of specimens exposed to marine conditions
exhibited very irregular oxide layers containing both pedestals
(Fig. 4) and dissolution areas (Fig. 5). Furthermore, the oxide layers
were frequently cracked and contained numerous pores. An inter-
esting observation was that large pits (often nearly occluded) were
frequently observed below or adjacent to a pedestal formation, an
example of which is shown in Fig. 6. EDS analysis of the cross-sec-
tion shown in Fig 6 indicated that the pedestal had an elemental
composition of Zn, O and C in ratios consistent with zinc hydroxyl
carbonate, while the occluded pit contained Zn, O and Cl, and the Fig. 6. Cross-section of pit below pedestal on specimen exposed at Cowley Beach.
 I.S. Cole et al. / Corrosion Science 52 (2010) 848858 853

lighter material capping the pit contained only Zn. Fig. 5 shows a
cross-section through a dissolution zone in which the thick oxide
layer contained Zn, O and C, while the thin oxide layer contained
Zn, O and S and Cl.
For specimens exposed at non-marine sites, oxide layer surfaces
were much smoother, without any indication of signicant pores
or cracking. However, a number of features were apparent from
the cross sectional observations. Small dissolution zones of around
15 lm were frequently observed in the oxide, whilst in other
areas oxide had built up to thicknesses of the order of 50 lm or
more. Cross-sections revealed pits with both low (<0.7) and rela-
tively high (0.7 to >2) aspect ratios. Fig. 7 shows a low aspect ratio
pit in a specimen from Phrapradaeng (industrial), while Fig. 8 and 9
show two different high aspect ratio pits in specimens from
Bicutan (urban) and Bangkok (urban/industrial) respectively.
SEM observations of cross-sections of specimens exposed for
Fig. 9. High aspect ratio pit on specimen exposed at Bangkok.
one year are summarized in Table 5, using the key in Table 6 to
dene the signicance of the ratings. In all at least 10 pits were
examined on each specimen. In determining the pit depth in Table Table 5
6, the depth is taken as the distance from the current metal surface SEM observations from cross-sections.
to the bottom of the pit. Where there has been signicant general
Site SO2 (lg/m3) Range of oxide Form Average pit
corrosion the current metal surface may differ signicantly from thickness of pit depth (lm)
the original metal surface.
Phrapradaeng 68 1 1 2.8
Mentok 40 1 1 8.8
3.2.2. Regression analysis Gresik 30 2 1 9.6
On the basis of the data in Table 3, regression analysis was used Bangkok 21 1 2 19.2
Bicutan 21 1 2 30.4
to derive the best parametric models to describe the variation in Ho Chi Minh 20 2 1 2.8
zinc and steel corrosion rates. In addition to the variables listed Cikampek 19 2 1 2.9
in the table, the average hydrogen ion concentration in lm in Cabuyao 14 1 2 6
rainwater ([H+]conc) was derived and used in the regression Hanoi 5 1 3 14
Lembang 4 2 2 11
Bagiou 2 1 2 15
Walkamin 0.5 1 2 5
Hue 0.5 2 2 6
Salinity
(mg/m2 day)
Nha Trang 13 4 3 27
Jebus 47 4 3 15
Rayong 23 3 2 5
Phuket 87 3 2/3 10
Cowley Beach 400 5 3 20

Table 6
Key to ratings used in Table 5.

Class Pit form (aspect ratio) Pit depth (lm) Oxide thickness (lm)
1 Very shallow pit (<0.7) (P) < 5 (T) < 5
2 Sharper pits (0.7 to >2) 2<P<5 5 < T < 10
Fig. 7. Low aspect ratio pit on specimen exposed at Phrapradaeng. 3 Nearly occluded pit 5 < P < 10 10 < T < 20
4 10 < P < 20 20 < T < 80

analysis. The analysis was undertaken for marine sites, industrial

sites, urban and highlands sites and various combinations. The
combinations are used as the number of sites in each category is
quite small and so more signicant models can be derived if cate-
gories are combined. The models with the highest degree of t are
given below. In the analysis the variables were normalised so that

T n T=20; max T n max T=20; Rn R=1000;

TOWn TOW=100; H n 107  H 

For zinc:
Using severe marine and marine sites (as dened in Table 1)
u
M dSal

where M is the mass loss (in lm/year), Sal is the annual salinity and
Fig. 8. High aspect ratio pit on specimen exposed at Bicutan. d and u are constants.
854 I.S. Cole et al. / Corrosion Science 52 (2010) 848858

Using industrial sites The dose functions determined for both zinc and steel for the
v locations in the present study have vastly different dependencies
M dTOWn
than determined for European exposure studies [1517], particu-
where TOWn is the normalised annual % time of wetness and v is a larly as industrial pollutants such as SO2 do not appear to be dom-
constant. inant in the present work.
Using marine, highland and urban sites or using marine, high-
land, urban and marine/industrial sites
u 4. Discussion
M dSal Rbn TOWvn
where Rn is the normalised annual rainfall and b is a constant. Given the available evidence and present data, an attempt was
For the three groups of: (1) industrial, highland and urban sites, made to dene the sequence of events and links between pollutant
(2) all sites except marine sites and (3) all sites, the best model is deposition and surface degradation. Plates are likely to be wet for
around 50% of the time at non-marine sites, and for considerably
M dT an Rbn TOWv 3 more than this time at marine sites. Work by Cole et al. [2628]
where Tn is the normalised average daily temperature and a is a has indicated that periods of wetness at non-marine sites are dur-
constant. ing the night and during rain spells. During dry days, plate surfaces
The values of the constants and the R-squared adjusted coef- should attain quite high temperatures, so that moisture lms or
cient of the model are given in Table 7. droplets will quickly evaporate. However, due to the high levels
For steel: of hygroscopic salts at marine locations, moisture layers will per-
For marine environment sists for longer periods despite high surface temperatures [26].
Deposition onto surfaces may be either by aerosol deposition,
u
M dSal gaseous deposition onto a moisture lm, gaseous deposition onto
a dry surface, or deposition by fog or rain droplets. Models devel-
For marine, urban and highland sites
oped by Cole et al. [29] have shown that in all cases except gaseous
u
M d max T gn Sal TOWvn deposition onto a dry surface, the chemical nature of an aerosol or
moisture lm or droplet will depend on the nature of the gaseous
where max Tn is the normalised average maximum daily tempera-
species present in the atmosphere, and their gaseous/aqueous
ture and g is a constant.
phase interactions. For example, although gaseous SO2 levels con-
For marine, marine industrial, urban and highland sites
trol the chemistry of moisture lms or droplets, it is also heavily
u
M dSal TOWv inuenced by oxidizers of SO2 in solution and by alkali precursors
(e.g. NH3).
For industrial and benign and highland sites and for all sites except Data on rainwater chemistry and particulate chemistry for the
marine current sites has been published previously [29]. In addition, Table
M dSal kH n e T /n RHkn 4 9 presents some pertinent molar ratios for the particulates col-
lected at each site. A key conclusion of this earlier data were that
+
where [H ]n is the normalised average hydrogen ion concentration particulates in the high gaseous SO2 locations (such as Phraprada-
in water and e and k are constants eng) had SO4:NH4 ratios around 1.5, while particulates in the lower
For all sites gaseous SO2 locations (Cikampek or Lembang) had SO4:NH4 ratios
close to or below 1. The SO4:NH4 ratio indicates the balance be-
M dSal kH n e T / Rb
tween SO2 absorption, which acidies an aerosol, and NH3 absorp-
The value of the constants and the R-squred adjusted values for the tion, which neutralizes an aerosol. When the ratio is around 1, the
steel model are given in Table 8. No meaningful data could be de- effect of these absorption processes will be balanced, and thus
rived for the steel mass loss in the industrial sites only. aerosol pH would be controlled by other factors (such as CO2

Table 7
Regression coefcients and R-squared values (adjusted) for models of zinc mass loss in different categories of exposure sites.

Sites used v b a u d R-Sq (Adjusted)

Industrial 0.71 7.1 91.4
Marine 0.86 7.4  104 98.0
Marine, urban, highland 1.72 0.82 0.21 0.71 94.4
Marine, marine/indusrial, urban, highland 2.1 0.37 0.26 1.95 79.5
Indusrial, urban, highland 0.71 0.93 2.34 0.33 82.0
All except marine 0.91 0.70 2.88 0.44 78.2
All 1.04 0.87 2.54 0.52 71.7

Table 8
Regression Coefcients and R-squared values (adjusted) for models of Steel mass loss in different categories of exposure sites.

Sites Used v / b c u e g k d R-sq. (adjusted)

Marine 1.3 3.8  103 96.8
Marine, urban, highland 2.6 0.54 1.3 1.05  103 97.7
Marine, marine/indusrial, urban, highland 2.5 0.54 288 92.0
Industrial, highland and urban 5.0 3.4 0.17 1.8 63 65.7
All except marine 5.0 3.3 0.17 4.0 49 71.0
All 3.2 0.67 0.42 0.15 3.3 62.3
 I.S. Cole et al. / Corrosion Science 52 (2010) 848858
Table 9
Analysis of particulate composition.
Site SO4:NH4 % SO4 from sea water
Phrapradaeng 1.34 3.5
Rayong 1.00 2.2
Bangkok 1.45 0.6
Cikampek 1.18 0.1
Bandung 0.82 0.2
Cabuyao 0.5 10
Baguio 0.4 12
Bicutan 1.1 25
Walkamin 71
Cowley Beach 100
Jebus 100
absorption), while at a ratio >1, SO2 absorption will be stronger,
thus producing an acidied aerosol.
Analysis of elemental ratios for a given particulate chemistry
can also indicate what percentage of the sulphate present is likely
to have come from a marine source, and what percentage is from a
land-based source. This analysis follows standard atmospheric sci-
ence procedures [30] and uses Na ion concentration as a marker for
a marine source and so calculates the excess sulphate over what
would be expected given the Na ion sulphate ratio in seawater.
The excess is then assumed to have a non-marine source. Not sur-
prisingly, the sulphates in the particulates collected at the marine
sites Cowley Beach and Jebus had a 100% marine origin, while the
sulphates in the particulate collected at Walkamin, Bicutan, Bagiou
and Cabuyao had a mixed marine and land origin. Rainwater
showed a similar variation in SO4:NH4, consistent with the rainwa-
ter pH values which were all above 4.2. However, these ratios are
not included here, as our earlier data [29] may have underesti-
mated NH4+ levels in the rainwater samples, due to the volatility
of NH3 and the time delays involved in the sample analysis from
the South East Asian eld surveys.
The impact of dose functions on corrosion phenomena can be
interpreted in light of the above discussion. While the acidicat-
ion of particulate (and aerosols) and rain by SO2 absorption still
occurs, it is partially neutralized by NH3 absorption, which may
account for the fact that SO2 is not a signicant variable in either
the zinc or steel dose functions. For steel, the [H+] concentration
in rainwater allows the effect of acidication to be taken into ac-
count in the dose function in a manner that incorporates the sul-
phate/ammonia absorption balance. Interestingly, in the zinc
dose function there is no parameter that directly reects indus-
trial pollutants. In addition to the role of ammonia in neutralizing
acidied aerosol or water, there may be two other reasons for
this:
1. As will be discussed later, the mode of corrosion changes in
response to industrial pollutant levels, with a more localized
mode of attack occurring at locations with lower gaseous SO2
concentrations. This change in mode may tend to increase the
rate of attack at the lower gaseous SO2 concentrations, decreas-
ing the overall dependence on SO2.
2. The dose function for zinc shows a strong positive dependence
on rainfall and temperature, which is hypothesized to dominate
over the inuence of industrial pollutants. For example, the
increase in corrosion with increasing rainfall may arise from
the role of rainfall in dissolving corrosion products. In fact, it
was observed that continuous oxide layers had formed on all
plates before the end of the seasonal exposures (see seasonal
FTIR data in Table 4), and that these oxides were predominately
zinc hydroxy carbonate, with some zinc hydroxy sulphate and
zinc hydroxy chloride in industrial and marine locations,
respectively. Data from Graedel [14] indicates that zinc hydroxy
855
carbonates, and to a lesser extent zinc hydroxy sulphates,
would be reasonably insoluble in most of the rainwaters
sampled in this survey. However, dissolution of the zinc
hydroxy chloride would be expected, as evidenced in the SEM
observations and by the rarity of compounds identied in the
FTIR spectra from plates exposed at marine sites.
SEM analysis of cross-sections revealed three forms of
corrosionshallow pits, sharp pits and necked or occluded
pitsthat appeared to be related to environment type, i.e. high
SO2, moderate SO2, and marine environments.
Specimens from all marine sites (Cowley Beach, Phuket, Jebus,
Nha Trang) showed a common set of effects. All specimens exhib-
ited nearly occluded deep pits, which (except for Phuket speci-
mens) incorporated signicant chloride (probably zinc hydroxy
chloride). In some areas, oxide layers were thick and were found
to have signicant carbon (probably zinc hydroxy carbonate), how-
ever, in other areas the oxide was thinner, particularly in associa-
tion with dissolution patches. The oxide under these patches
may have been either sulphur and chloride or carbon.
Two signicant features were present in the marine site speci-
mens: pedestals and dissolution pits. The width of the pedestals
observed varied from 50 to 300 lm, and oxide thicknesses ranged
from 15 to >50 lm. Interestingly, pedestals exhibited a very lay-
ered internal structure (almost like tree rings) that centred on
the entrance to a nearly occluded pit. Galvele [9] proposed a model
of zinc pitting in which pitting is promoted by acidication at the
anode at the bottom of holes in the oxide layer (the cathode occurs
away from the hole). Pedestal formation could be promoted by a
similar separation of anodic and cathodic regions as illustrated in
Fig. 10. As anodic activity occurs at the bottom of a pit and cathodic
activity occurs at the oxide/droplet interface, a sharp pH gradient
will be established through a defect hole in the oxide layer. Zinc
ions (or Zn(OH)+) will migrate away from the anode (while chlo-
ride ions should migrate to the anode) through the defect in the
oxide layer to precipitate as zinc hydroxide or hydroxy carbonate
at the droplet/oxide interface. Alkaline pH adjacent to the cathodic
zone will promote CO2 absorption, thus favouring the precipitation
of this carbonate. The process should be self-reinforcing, as the
build-up of zinc carbonate layers will increase the physical separa-
tion of the anodic and cathodic zones reinforcing the pH gradients,
assuming that solution resistance is low enough to maintain a suf-
ciently high current exchange. This would then lead to the lay-
ered structures of the pedestal formations observed in the
micrographs.
As indicated in Figs. 2 and 6 from Nha Trang, the localized dis-
solution of oxides was commonly observed at marine sites. The
dissolution pits had diameters from 10 to 100 lm, and while the
material apparently being dissolved was rich in C (probably zinc
carbonate), the material beneath the dissolution zones was often
rich in chloride and sulphur. Recently, work reviewed by Cole
et al. [22] and in part developed by Keene et al. [32] has high-
lighted that marine aerosols may absorb both SO2 and acid gases
(H2SO4, HCl etc.), and so may develop pH values below 3 (depend-
ing on aerosol size). As highlighted by Cole et al. [22], such acidied
aerosols (particularly if sulphate-rich and thus buffered by bisul-
phate/suphate reaction) would be capable of dissolving apprecia-
ble zinc hydroxide layers. This may explain the size range of the
dissolution pits observed in the current study, which was consis-
tent with that observed for aerosol impacts [33], and why the
remaining oxide was rich in chloride and sulphur.
Specimens from high SO2 environments (Phrapradaeng, Men-
tok, Gresik) also exhibited a common set of effects, with low aspect
ratio and shallow pits, and thin to moderate oxide layers (always
<10 lm). The oxide layers generally contained C and S, while the
shallow pits always contained C and S.
856 I.S. Cole et al. / Corrosion Science 52 (2010) 848858
Fig. 10. Schematic of Galvele like model applied to corrosion in marine environments.
In contrast, samples from sites with low levels of gaseous SO2
and low marine aerosol concentrations (Cabuyou, Hanoi, Lembang,
Bagiou, Walkamin, Hue) exhibited relatively deep and sharp pits.
In general, the oxide layers were rich in C, but the oxide on samples
from Lembang also contained S, and those from Cabuyou also con-
tained S and Cl. The pits were also rich in C, except those in sam-
ples from relatively higher gaseous SO2 environments (Cabuyou,
Hanoi), which were rich in S.
Samples from a number of sites with moderate gaseous SO2 lev-
els (Bangkok, Bicutan, Ho Chi Minh, Cikampek) exhibited either the
low aspect ratio shallow pits observed for the high SO2 environ-
ments, or the sharp deep pits observed for the low SO2
environments.
Thus, the results indicate that shallow pits occurred in the high
SO2 environments and that sharp or necked pits occurred in the
moderate or low SO2 environments or marine locations. In the
industrial environments where shallow pits occurred, S was found
throughout the oxide, but where sharp pits occurred it was found
almost exclusively at the bottom of the pits.
Historical models developed by Sato [34] and Galvele [9,10] may
be relevant to the above observations. In discussing iron-group met-
als, Sato indicated that there are two modes of pitting corrosion. One
is pitting dissolution in the polishing state, which proceeds at more
noble potentials, and the other is in the active state, which proceeds
at less noble potentials. Increased ion concentrations and reduced
cathodic areas stabilize the region of polishing state stability, whilst
larger cathodic regions and lower pH promote the active state. Fol-
lowing Satos model, if an acidic droplet rich in sulphate lands on a
zinc plate that is already covered with an oxide lm, then the high
sulphate level and reduced cathodic area (due to diffusion limits
by the oxide) may promote the polishing state and shallow pits will
result. In contrast, when the sulphate concentration is lower, the
general corrosion zone passivates as sulphate precipitates out of
solution. Corrosion will only recommence if a localized area of higher
susceptibility (e.g. ground boundary) is attacked, in which case the
microstructural feature will promote a sharp and very localized
pit, which will lead to a lowering of the local pH and active localized
anodic dissolution of the metal. However, at some stage pitting will
cease either because of the depletion of the active species or due to
diffusion limits.
As mentioned earlier, Galveles model [9,10] is based on the
separation of the anodic and cathodic areas, and acid-promoted
dissolution at the anodic zone. Pitting will progress when there
is sufcient anodic current exchange at the bottom of a pit to
maintain a pH level that is conducive to the dissolution of any pro-
tective oxide, and when the pit is of sufcient depth that any ion
migration will not dissipate this pH difference. Galvele calculates
that pitting will occur for zinc in a solution of pH 9, when the prod-
uct of the pit depth and the current density exceeds 106 A/cm, and
further, that the current density should be 1 A/cm2 so that pitting
is possible when pit depth exceeds 102 lm. However, Galveles
current density of 1 A/cm2 appears rather high and thus larger pits
depths would be required for more moderate current densities.
However, there are a number of reasons why Galveles original
model cannot be directly applied to pitting under an oxide layer
promoted by either acid aerosol or acid rain deposition. Firstly,
the pH of a deposited aerosol is likely to be between 5 and 6
(assuming that oxide dissolution alters the original acidic pH of
the aerosol) not 9 as in Galveles example. Secondly, zinc hydroxy
carbonate and zinc hydroxy sulphate may passivate the pit surface,
and thus acidication would have to maintain a pit pH below 5 to
prevent precipitation of these oxides and not 8.7 as estimated by
Galvele on the basis of preventing the precipitation of zinc oxide.
Despite the lack of applicability of Galveles particular exam-
ples, the basic premise of his model can explain the general obser-
vations reported in this paper. This is illustrated in Figs. 11 and 12,
respectively, for moderately and signicantly acidied aerosols. If
an acidied aerosol of moderate pH (say 4.05.5), and that may
have formed in a low to moderate SO2 environment, lands on a sur-
face covered in zinc hydroxy carbonate or zinc sulphate, the acid-
ied solution will not be sufcient to dissolve the oxide layers of
P5 lm observed on metals exposed in mild environments. How-
ever, it may cause localized dissolution at defects in an oxide lm
and promote the linking of already established cracks in the oxide,
so that a localized oxide hole is established (see Fig. 11). Anodic
activity can then commence at the bottom of the pit, causing acid-
ication, while cathodic activity will occur at the oxide/liquid
interface. The depth of the oxide hole, being the oxide thickness,
will be sufcient to retain the pH differential between the pit bot-
tom and the cathodic areas that exist external to the pit. As in the
discussion on marine aerosols, Zn2+ or Zn(OH)+ ions may migrate
out of the hole and will subsequently precipitate in the neutral
pH zones close to the entrance to the pit. As these areas will be
higher in pH, CO2 absorption will be enhanced and the
 I.S. Cole et al. / Corrosion Science 52 (2010) 848858
Fig. 11. Schematic of Galvele like model applied to corrosion from in moderately
acidied aerosols.
precipitation of zinc carbonate is expected, while zinc sulphates or
sulphates may precipitate in the pits as they dry out. However, if a
strongly acidied drop (pH < 3), such as may occur in a high SO2
environment, falls on a hydroxy carbonate or hydrosulphate sur-
857
Fig. 12. Schematic of corrosion process caused by highly acidied aerosols.
face, then signicant oxide dissolution (such as indicated in the
surface micrograph of the Bangkok specimen shown in Fig. 3)
may occur across the whole of the oxide lm under the aerosol
drop (see Fig. 12). Anodic and cathodic areas are likely to be estab-
lished within the freshly exposed surface, and thus Galvele mech-
anism cannot operate as diffusion will rapidly balance out any pH
variation. Thus the Galvele mechanism does appear to explain the
observed morphology of pitting, however, more detailed studies of
the electrochemistry of pit formation are required before its pre-
cise (if any) role in controlling such pitting can be determined.
A knowledge of the exact mechanism whereby Zinc corrodes in
different environments is a pre-requisite to developing more resis-
tant zinc coatings. For example if a Galvele like mechanism does
drive pit formation in moderate environments then strategies to
limit the formation of through oxide defects could be a useful
direction for research. This could include reducing the porosity of
the initial oxide that forms on zinc coatings.
This study was undertaken on zinc samples, rather than gal-
vanised steel (which is the industrially signicant material) as
microstructural and chemical variations can be more readily con-
trolled in zinc specimens. The corrosion rate of zinc and galvan-
ised steel are likely to be similar (except initially if the
galvanised steel has undergone a chromate passivation process,
858 I.S. Cole et al. / Corrosion Science 52 (2010) 848858
or at the later stage of corrosion when corroding pit encounters
the intermetallic zone at the steel/zinc interface). A typical gal-
vanised coating used in Australia and South East Asia is Z275
(AS 1650 [35]) which has 275 g/mm2 of zinc coating or approxi-
mately 19.4 lm on zinc on each side of the steel plate. The pit
depths reported in Table 5 for a number of marine (Nha Trang
and Cowley Beach) and industrial sites (Bangkok and Bicutan) ap-
proach or exceed this coating thickness. Thus it is possible that
red rust could form on Z275 galvanised steel in these locations
in periods not much longer than a year. General corrosion rates
are much lower, for all sites except Cowley Beach, it will take
in excess of 4 years to decrease the coating thickness to 50% of
the original. Thus the observation of pitting on zinc has signi-
cant implications both to the predicted life of zinc and to any
strategies to enhance zinc coatings.
5. Conclusions
A study undertaken in tropical regions in ve nations measured
the mass loss of exposed zinc and steel specimens, as well as a
range of climatic and pollutant factors. The study found that:
1. Regression models derived to explain the variation in zinc cor-
rosion rate did not show a dependence on atmospheric SO2
level.
2. Although a denitive conclusion could not be reached, possible
reasons for this lack of dependence include:
the relatively high pH of rainwater (and probably aerosols)
due to the effect of other gaseous species; and
the different mechanisms of attack observed in high and low
SO2 environments, which may lead to relatively higher cor-
rosion rates in the lower SO2 environments due to enhanced
localized attacked.
Further, the study found that pitting of zinc could occur under
oxides when exposed to marine and relatively mild environments
(mild industrial, rural and urban). In contrast, pitting was not ob-
served when zinc was exposed to high SO2 environments.
References
[1] S.R. Dunbar, W. Showak, in: W.H. Ailor (Ed.), Atmospheric Corrosion, John
Wiley & Sons, New York, 1982, pp. 529552.
[2] U.R. Evans, D.E. Davies, Corrosion 8 (5) (1952) 165.
[3] P.T. Gilbert, Sheet Met. Ind. OctoberDecember (1948).
[4] K.F. Lorking, J.E.O. Mayne, in: Proceedings of the First International Congress on
Metallic Corrosion, London, April 1961, pp. 144146.
[5] W. Feitknecht, Chem. Ind. (1959) 11021109.
[6] D.E. Davies, M.M. Lotlikar, Br. Corros. J. 1 (1966) 149.
[7] T. Muster, W.D. Ganther, I.S. Cole, Corros. Sci 49 (2007) 20372058.
[8] A.D. Keitelman, S.M. Gravano, J.R. Galvele, Corros. Sci. 24 (1984) 535.
[9] J.R. Galvele, J. Electrochem. Soc. 123 (1976) 464.
[10] J.R. Galvele, Corros. Sci. 21 (1981) 551.
[11] S.M. Gravavo, J.R. Galvele, Corros. Sci. 24 (6) (1984) 517.
[12] I.S. Cole, D.A. Paterson, W.D. Ganther, Corros. Eng. Sci. Technol. 38 (2003) 129
134.
[13] J.W. Spence, F.H. Haynie, in: Corrosion Testing and Evaluation: Silver
Anniversary Volume ASTM STP 1000, ASTM, Philadelphia, 1990, pp. 208224.
[14] T.E. Graedel, Corros. Sci. 38 (12) (1998) 2153.
[15] L. Atteraas, S. Haagenrud, in: W.H. Ailor (Ed.), Atmospheric Corrosion, John
Wiley & Sons, New York, 1982, pp. 873891.
[16] M. Benarie, F.L. Lipfert, Atmos. Environ. 20 (10) (1986) 1947.
[17] J. Tidblad, A.A. Mikaliov, V. Kucera, in: Proceedings of the UN/ECE Workshop
on Quantication of Effects of Air Pollutants on Materials, Berlin, Germany,
2427 May 1998, p. 77.
[18] J.E. Svensson, L.G. Johansson, J. Electrochem. Soc. 140 (1993) 2210.
[19] L.A. Farrow, T.E. Graedel, C. Leygraf, Corros. Sci. 38 (12) (1998) 2181.
[20] S.E. Haagenrud, J.F. Henriksen, in: C. Sostrom (Ed.), Durability of Building
Materials and Components 7, vol. 1, E&F Spon, London, 1996, pp. 210218.
[21] I.S. Cole, in: S.W. Dean, G. Hernandez-Duque Delgadillo, J.B. Bushman (Eds.),
Marine Corrosion in Tropical Environments, ASTM STP 1399, ASTM,
Philadelphia, 2000, pp. 3347.
[22] I. S Cole, N. Shahzad, A. Kanta, M. Venkatraman, Int. Mater. Rev. 54 (3) (2009)
117135.
[23] W. Ganther, P. Norberg, I.S. Cole, in: Proceedings of the Thirteenth
International Corrosion Conference, Melbourne, Australia, 1996.
[24] A.K. Neufeld, I.S. Cole, Corrosion 53 (10) (1997) 788.
[25] A.K. Neufeld, I.S. Cole, in: Proceedings of the Second International Symposium
on Pits and Pores: Formation, Properties, and Signicance for Advanced
Materials, as a Part of the 198th Meeting of The Electrochemical Society,
Phoenix, Arizona, 2227 October 2000, paper F10449.
[26] I.S. Cole, W.D. Ganther, Corros. Eng. Sci. Technol. 43 (2) (2008) 156.
[27] I.S. Cole, D.A. Paterson, Corros. Eng. Sci. Technol. 41 (1) (2006) 67.
[28] I.S. Cole, W.D. Ganther, Corros. Eng. Sci. Technol. 41 (2) (2006) 161.
[29] I.S. Cole, A.K. Neufeld, P. Kao, W.D. Ganther, L. Chotimongkol, C. Bharmornsut,
N.V. Hue, S. Bernardo, in: Proceedings of the Fourteenth International
Corrosion Congress, Cape Town, South Africa, 27 September to 1 October
1999, paper 265.2 (CD-ROM).
[30] J.N. Mphepya, C. Galy-Lacaux, G. Held, J.J. Pienaar, J. Atmos. Chem. 53 (2006)
169183.
[32] W.C. Keene, R. Sander, A.P. Pszenny, R. Vogt, P.J. Crutzen, J.N. Galloway, J.
Aerosol Sci. 29 (3) (1998) 339.
[33] I.S. Cole, D. Lau, D.A. Paterson, Corros. Eng. Sci. Technol. 39 (3) (2004) 209.
[34] N. Sato, Corros. Sci. 37 (12) (1995) 1947.
[35] AS 1650-1989, Hot-Dipped Galvanized Coatings on Ferrous Articles, Standards
Australia, Sydney, 1989.