Progress in Organic Coatings 37 (1999) 91±98

Accelerated tests for the prediction of cut-edge

corrosion of coil-coated architectural cladding
Part I: cyclic cabinet salt spray
R.L. Howard, S.B. Lyon*, J.D. Scantlebury
Corrosion and Protection Centre, UMIST, PO Box 88, Manchester M60 1QD, UK

Received 26 March 1999; received in revised form 17 August 1999; accepted 20 August 1999

Abstract

Various accelerated cabinet tests have been used for the evaluation of the cut-edge corrosion of coil-coated architectural cladding. These
include the conventional ASTM B-117 method (5% continuous NaCl spray), the standard Prohesion test (0.35% (NH4)2SO4 ‡ 0.05% NaCl
wet/dry spray) as well as modi®ed wet/dry spray tests using a relatively dilute arti®cial acid rain solution, shallow specimen incline angles
with variations in the ratio of the wet and dry periods. Comparisons with outdoor exposure samples reveal that the B-117 test shows
unrealistic corrosion morphology with the most realistic cut-edge corrosion given by the modi®ed tests. Furthermore, with all types of wet/
dry test specimen-to-specimen variation is considerably reduced compared with the continuous salt spray test. In the novel wet-dry tests the
acceleration factor appears consistent with the number of wet and dry cycles however, the overall acceleration factor is limited and, hence,
test times are extended. It is concluded that, over a 1000 h test time, the standard Prohesion test appears to show the best combination of
realism and acceleration. # 1999 Elsevier Science S.A. All rights reserved.

Keywords: Accelerated tests; Edge corrosion; Coil-coated; Cladding

1. Introduction ing to recent industry surveys [1,2]. Damage on the top coat
typically results from foot traf®c on roofs, incorrect ®xing
Coil-coated cladding is a widely used roo®ng and siding and joining practice and careless installation. Interestingly,
material. It is typically organically-coated galvanised steel, vertical sidings installed at the same time as roofs, generally
with additional protection provided by an organic primer perform much better.
coating and top coat. Pro®le and panel manufacturers roll- Standard accelerated tests employ a continuous 5% NaCl
form the coil-coated strip-steel and guillotine it into shorter spray solution, such as that used to date in the ASTM B-117
lengths for installation. The cutting operation produces bare and similar standards. However, in the 1960's, Harrison and
(uncoated) edges of steel and zinc which are directly Tickle recognised the bene®t of including ammonium sul-
exposed to the atmosphere. Subsequent galvanic corrosion phate for improved prediction of outdoor performance over
of the steel (cathode) and zinc (anode) occurs at the cut-edge sodium chloride solution [3]; thus a solution of 0.25% NaCl
and results in the recession of the zinc coating and con- and 3.5% (NH4)2SO4 has become known as Harrison's
sequent loss of the interface between the organic coating and solution [4]. Timmins [5] employed a diluted solution based
the cut-edge. This eventually leads to unsightly peelback of on Harrison's formula (0.4% (NH4)2SO4 ‡ 0.05% NaCl) in
the organic coating and the development of red rust on the a cyclic test (3 h wet/1 h dry) to provide improved prediction
steel at the cut-edge. Cut-edge corrosion is worst on shallow- over the salt spray test.
pitched roofs with poor drainage and accumulated dirt at the Later, Lyon and co-workers [6] used an arti®cial acid rain
cut-edge, because of the longer time-of-wetness under these solution, shallow specimen incline angles (5±158) and wet/
conditions. The corrosion of cut-edges is one of the major dry cycles with an atmospheric test cabinet. The evaluation
failure modes for coil-coated architectural cladding accord- of this cyclic cabinet test has continued in the present study
to include comparisons of the performance of a range of
factory-produced materials in the ASTM B-117 salt spray,
*
Corresponding author. Tel.: ‡44161-200-4846; fax: ‡44161-200-4865 Prohesion and other cyclic cabinet tests. Also, materials
E-mail address: [email protected] (S.B. Lyon) (Zn±55Al) tested at severe marine and industrial/marine

0300-9440/99/$ ± see front matter # 1999 Elsevier Science S.A. All rights reserved.
PII: S 0 3 0 0 - 9 4 4 0 ( 9 8 ) 0 0 0 4 8 - 4
92 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98

outdoor exposure sites have been examined to further assess in the centre of the exposed cut-edge of each specimen; all
the realism of the accelerated tests. other specimen edges were well-masked by wax. This visual
assessment of undercutting was performed with the aid of a
2. Experimental metallurgical stereo microscope and a graduated ruler. For a
continuous blister front, the disbonded area was given as the
The accelerated test conditions are summarised in Table 1. disbonded distance from the edge multiplied by the width
Two experimental cyclic cabinet tests were performed with along the edge. Discrete blisters were typically circular or
the second test modi®ed to increase the acceleration of edge semi-circular and thus the disbonded area was calculated
undercutting. The details of the materials (obtained from from the blister diameter. The accuracy of these measure-
various commercial suppliers) and the accelerated tests to ments are estimated to be 5 mm2.
which each were subjected are shown in Table 2. Samples A, Zn±55Al samples (A) were selected for the examination
B and C are all PVC Plastisol-coated, but with different of the morphology of metal coating corrosion. The mechan-
types of metal coating. Samples D and E are both polyester- ism of corrosion may be seen by examining the corrosion
coated, hot-dip galvanised (HDG) but with the materials morphology of the dendritic aluminium-rich and interden-
obtained from different suppliers. Finally, samples C, D and dritic zinc-rich regions, whereas the corrosion morphology
E are all HDG but with different top coats and zinc : steel of the zinc coatings (HDG) does not provide such a distinc-
area ratios. All specimens were sheared with the same tion [7]. Therefore, a comparison of the corrosion morphol-
guillotine from the top coat side downwards, and measured ogy for Zn±55Al provides an indication of whether the
140 mm  70 mm, with the base edge 70 mm wide. The mechanism of edge corrosion outdoors is duplicated in
conversion coatings on all the samples were similar (chro- accelerated tests. The microstructure of the Zn±55Al
mate) and all primers contained strontium chromate inhi- (wt%) alloy coating, before exposure to atmospheric corro-
bitor. At least four specimens of each type of material sion, was revealed by etching for approximately 60 s in a
(replicates) were tested and the results for all the specimens Nital solution.
are shown in the results graphs. Cross-sections of corroded Zn±55Al specimens were
Edge disbondment (undercutting) was measured by sum- prepared in order to examine the morphology of metal
ming the disbonded areas within an area of 15 mm  30 mm coating corrosion from the cut-edge after both accelerated

Table 1
Accelerated test conditions

Test type Cyclic cabinet ASTM B-117 salt spray Prohesion

1 2

Test duration (h) 1500 1500 1000 1000

Total wet time (h) 250 500 1000 500
Number of cycles 500 1500 500
Wet : dry ratio (min) 30 : 150 20 : 40 ± 60 : 60
Incline angle (8) 5 15 75 75
Solution composition (ppm) Ionic strength: 342 NaCl 50,000 Ionic strength: 7500
H2SO4 31.85 pH ˆ 6.5±7.2 NaCl 500
(NH4)2SO4 46.20 (NH4)2SO4 3500
Na2SO4 31.95 pH ˆ 5.2
HNO3 15.75
NaNO3 21.25
NaCl 84.85
pH ˆ 3.5

Table 2
Details of coil-coated materials and the tests employed for each material

Sample Steel Metal coating Primer Topcoat Tests used

A 0.6 mm 20 mm: Zn±55Al 6 mm acrylic 150 mm PVC Plastisol Cyclic cabinet: 1 and 2;
(``Aluzinc'') salt spray, Prohesion
B 0.6 mm 20 mm: Zn±5Al 8 mm (not known but 140 mm PVC Plastisol Cyclic cabinet: 1; salt
(``Galfan'') probably acrylic) spray, Prohesion
C 0.6 mm 20 mm: Zn (HDG) 8 mm acrylic/phenolic 200 mm PVC Plastisol Cyclic cabinet: 1, salt
spray, Prohesion
D 0.53 mm 20 mm: Zn (HDG) 5 mm epoxy 20 mm Polyester Ð Supplier A Cyclic cabinet: 1 and 2,
salt spray, Prohesion
E 0.6 mm 30 mm: Zn (HDG) 6 mm epoxy 26 mm Polyester Ð Supplier B Salt spray, Prohesion
 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98 93

tests and outdoor exposure. At least four cross-sections were plier A (sample D) compared to similar material from
prepared for each specimen shown in this paper in order to supplier B (sample E) in both ASTM B-117 salt spray
provide a representative result. A cut-off wheel with a SiC and Prohesion tests.
abrasive wheel was used to cut specimens perpendicular to Discrete (individual) blisters are produced at the cut-edge
the cut-edge. Thereafter, the cut surface was ground (4000 of PVC Plastisol-coated Zn±55Al in the cyclic cabinet and
SiC grit) and polished to a 1 mm ®nish and examined with an Prohesion tests. This corrosion morphology is clearly com-
AMRAY 1810 Scanning Electron Microscope (SEM). parable to 2 years severe marine exposure (Fig. 3). Extensive
red edge rust is produced by the Prohesion test for all
3. Results Plastisol-coated substrate types (HDG, Galfan and Zn±
55Al) with somewhat less edge rust on polyester-coated
3.1. Experimental cyclic cabinet tests HDG (Fig. 4).
The microstructure of the Zn±55Al metal coating consists
The results for the cyclic cabinet tests are shown in (Fig. 1) of aluminium-rich dendrites and zinc-rich interdendritic
and Table 3 for a number of samples of each material. There regions. Cross-sections of PVC Plastisol-coated Zn±55Al
is a large variation in the amount of undercutting for reveal the preferential dissolution of zinc after cyclic cabinet
nominally identical materials. A comparison of (Fig. 1a and Prohesion tests, and areas of complete coating removal,
and b) reveals that an increase in undercutting is produced by i.e., removal of zinc and aluminium after the salt spray test
increasing the number of wet/dry cycles from 500 to 1500. (Fig. 5). PVC Plastisol-coated Zn±55Al specimens exposed
This is con®rmed by the summary of results in Table 3, outdoors reveal preferential zinc corrosion (Fig. 6). Under-
which also reveals that the amount of undercutting produced cutting on these and polyester-coated HDG specimens has
in test 2 is approximately three times that produced in test 1, been measured as approximately 1  0.5 mm from the cut-
which corresponds to the threefold increase in the number of edge.
wet/dry cycles. The decrease in time required for complete
drying of the test specimens was accomplished by increasing 4. Discussion
the incline angle from 58 to 158 to the horizontal and by
improving the air ¯ow during the drying cycle. In addition, 4.1. Appearance of cut-edge corrosion
the solution ¯ow rate was increased in order to accelerate the
wetting of the samples. Red edge rust is noticeably absent from all the specimens
tested in the ASTM B-117 salt spray test. Red rusting of the
3.2. Salt spray and Prohesion test results steel substrate at the cut-edge results from loss of protection
by the sacri®cial coating (zinc or zinc alloy). This may occur
Undercutting measurements for the materials tested in the either, when there is complete loss of contact between steel
ASTM B-117 salt spray and Prohesion tests are shown in and the sacri®cial zinc or zinc/aluminium coating due to
(Fig. 2) (note the vertical scale is twice that in Fig. 1) and drying of the electrolyte, or when recession of the zinc from
also in Table 4. Greater undercutting is produced in salt the cut-edge increases the distance between zinc and steel so
spray than Prohesion. Interestingly, greater undercutting is that the resistance of the electrolyte path between the metals
observed on the polyester-coated HDG material from sup- is too high for steel to be cathodically polarised. The

Table 3
Area of paint disbonded (undercutting) for coil-coated materials after cyclic cabinet tests (mean, standard deviation and data range shown)

Sample Material Cyclic cabinet 130 min Cyclic cabinet 220 min
wet/150 min dry wet/40 min dry

A Zn±55Al (Plastisol) 2  4 (0±10) 30  20 (15±50)

B Zn±5Al (Plastisol) 45  9 (30±55)
C HDG (Plastisol) 40  14 (25±60)
D HDG (Polyester) Supplier A 32  5 (29±40) 95  40 (65±143)

Table 4
Area of paint disbonded (undercutting) for coil-coated materials after salt spray and Prohesion tests (mean, standard deviation and data range shown)

Sample Material Salt spray Prohesion

A Zn±55Al (Plastisol) 170  170 (20±330) 65  20 (45±90)

B Zn±5Al (Plastisol) 230  110 (150±390) 190  135 (30±360)
C HDG (Plastisol) 130  100 (10±250) 130  60 (75±180)
D HDG (Polyester) Supplier A 200  120 (40±320) 105  45 (55±150)
E HDG (Polyester) Supplier B 50  25 (30±85) 30  13 (20±45)
94 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98

Fig. 1. Degree of undercutting for modified cyclic cabinet tests: (a) test 1: wet/dry ˆ 30 min/150 min (1500 h); (b) test 2: wet/dry ˆ 20 min/40 min
(1000 h).

concentrated NaCl solution employed in the salt spray test approximately 20 more concentrated than the 10 arti®-
increases the protection distance of zinc and the absence of cial acid rain solution. The salt spray test, without wet/dry
drying cycles excludes the possibility of only steel being cycles, does not produce discrete blisters on Zn±55Al. The
covered by electrolyte. other feature observed on Zn±55Al is preferential corrosion
Walter reported that, in cyclic dip tests using 100 ppm of zinc which is observed in the cyclic cabinet, Prohesion
NaCl solution, increasing the drying time promotes the tests and in outdoor exposure, but not in the continuous salt
formation of discrete blisters rather than continuous blisters spray test. This is likely to be due to relatively concentrated
[7]. In the present study, both 10  arti®cial acid rain 5% NaCl solution employed in this test, than in the absence
solution and dilute Harrison's solution (the Prohesion test of wet/dry cycles. Possibly, the more soluble, less protective
solution) produce discrete blisters in cyclic cabinet spray corrosion products which are produced in 5% NaCl lead to
tests. This is despite the fact that the Prohesion solution is corrosion of aluminium as well as zinc. In contrast, the more
 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98 95

Fig. 2. Degree of undercutting for standard cabinet tests: (a) ASTM B-117 salt spray (1000 h); (b) Prohesion test: wet/dry ˆ 60 min/60 min (1000 h).

dilute solutions used in the cyclic cabinet and Prohesion tests blisters, such as the nucleation of blisters at sites of zinc
are able to corrode only zinc, but not aluminium, from the corrosion.
Zn±55Al coating.
A further question arises, which is the correlation between 4.2. Material performance
preferential zinc corrosion and discrete blisters, which
appear to occur together. Clearly, the individual blisters A comparison of the undercutting produced by the experi-
on PVC Plastisol-coated Zn±55Al which are upwards of mental cyclic cabinet tests reveals a number of differences
0.5 mm in diameter, do not correlate with individual areas of between the various types of materials. For PVC Plastisol-
preferential zinc corrosion, which have dimensions in coated materials, Zn±55Al exhibits less undercutting than
micrometers. However, it is conceivable that there is a causal Zn±5Al and HDG (Table 3). Polyester-coated HDG (Sup-
relationship between preferential zinc corrosion and discrete plier A) exhibits similar undercutting to PVC Plastisol-
96 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98

mance in both salt spray and Prohesion tests (Table 4). The
material from supplier B has a zinc : steel area ratio (i.e.,
anode : cathode area ratio) some 50% larger than that from
supplier A. However, the performance advantage is a factor
of 3±4 improved. Thus, the presence of the thicker organic
coating contributes signi®cantly to its superior performance.
For the ASTM B-117 salt spray test, the polyester-coated
HDG material from supplier B exhibits less undercutting
than all the other materials. The large variability in under-
cutting for the salt spray test precludes further comparisons
between the materials tested. For the Prohesion test, PVC
Plastisol-coated Zn±55Al exhibits less undercutting than
PVC Plastisol-coated Zn±5Al and Zn (HDG). This may
be due to the increased protection of aluminium corrosion
products on the Zn±55Al alloy coating compared to zinc
corrosion products on HDG (Zn).

4.3. Corrosion rates

The ratios of maximum to minimum undercutting area for

four specimens of each material type tested in salt spray and

Fig. 3. Blistering at the cut-edge of Plastisol-coated Zn±55Al after: (a)

modified cyclic cabinet test 2 (1000 h); (b) Prohesion test (1000 h); (c)
severe marine exposure (2 years).

coated HDG, despite the thicker coating of the latter mate-

rial. Fig. 4. Red edge rust produced by the Prohesion test on Plastisol-coated
The two polyester-coated HDG materials, from suppliers substrates: (a) Zn±55Al; (b) Zn±5Al; (c) Zn (HDG); (d) HGD-polyester-
A and B, exhibit signi®cantly different undercutting perfor- coated.
 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98 97

Fig. 4 (Continued ).

Table 5
Ratio of maximum to minimum undercutting (mm2/mm2) for the
accelerated tests
Fig. 5. Cross-section micrographs of Plastisol-coated Zn±55Al after: (a)
Sample Salt Prohesion Cyclic Cyclic modified cyclic cabinet test 2; (b) Prohesion test; (c) ASTM B-117 salt
spray cabinet 1 cabinet 2 spray test.
A: Zn±55Al (Plastisol) 17 2 10 3
B: Zn±5Al (Plastisol) 3 12 2
C: HDG (Plastisol) 19 2 2 The amount of undercutting which is produced by an
D: HDG 8 3 1 2
accelerated test should be similar to that produced in outdoor
(Polyester-supplier A)
E: HDG 3 2 exposure during the guarantee period of the coil-coated
(Polyester-supplier B) material, i.e., between 15 and 20 years [2]. Walter reported
undercutting of 14±17 mm for silicone-modi®ed polyester-
coated HDG and Zn±55Al after 10 years exposure at the
Prohesion tests are shown in Table 5. In general, there is Bellambi Point severe marine exposure site [7]. The outdoor
greater scatter in the results for the salt spray test than the specimens examined in the present study (Figs. 3 and 6)
Prohesion test, indicated by the larger ratio of maximum to were exposed at severe marine and industrial/marine expo-
minimum undercutting values for the former test. The sure sites for 2 and 5 years respectively. These exposures
amount of undercutting produced in cyclic cabinet test 2 produced 1  0.5 mm of undercutting, which equates to an
is approximately three times greater than that produced in undercutting area of 30  15 mm2 over a 30 mm length of
cyclic cabinet test 1, i.e., with a threefold increase in the cut-edge. The ASTM B-117 salt spray and Prohesion tests
number of wet/dry cycles. are currently used by industry because they are capable of
98 R.L. Howard et al. / Progress in Organic Coatings 37 (1999) 91±98

and that produced in the salt spray and Prohesion tests after
1000 h.
Greater amounts of undercutting are produced in the salt
spray and Prohesion tests than the cyclic cabinet tests. The
main difference between the cyclic wet/dry tests, namely
Prohesion and cyclic cabinet test 2, is the greater solution
concentration of the former. Despite this more concentrated
solution, Prohesion produces realistic cut-edge corrosion
features; namely, red edge rust for all specimens, and for
PVC Plastisol-coated Aluzink, discrete blisters and prefer-
ential zinc corrosion.

5. Conclusions

1. The ASTM B-117 salt spray test produces unrealistic

coating degradation and a large variation in the results
for duplicate specimens of the same material.
2. The Prohesion test offers a good compromise between
realism and adequate acceleration of atmospheric
corrosion, with considerably less variation in results
than those of the ASTM B-117 salt spray test.
3. The use of an artificial acid rain solution in cyclic wet/
dry accelerated tests does not appear to provide a
substantial improvement in the realism of corrosion
compared with the Prohesion test. Furthermore, the
Prohesion test produces greater acceleration of corrosion
Fig. 6. Cross-section micrographs of Plastisol-coated Zn±55Al after
compared with cyclic tests using artificial acid rain
outdoor exposure: (a) 2 years at severe marine site; (b) 5 years at
industrial/marine site. solution.

producing suf®cient undercutting and general corrosion over Acknowledgements

a period of 1000 h to predict the performance over the
guarantee period. Therefore, accelerated tests of 1000± The authors wish to thank the industrial collaborators on
1500 h duration which produce a minimum of approxi- this project, TAC Metal Forming, Courtaulds Nippon Paints
mately 15 mm of undercutting from the cut-edge are and SSAB, for their advice and assistance with the acquisi-
regarded by industry as acceptable for the prediction of tion of materials and outdoor exposure data. The ®nancial
outdoor performance. The salt spray and Prohesion tests support of the Engineering and Physical Sciences Research
produce between 10 and 450 mm2 of undercutting for the Council is gratefully acknowledged.
range of materials tested in the present study (Table 4).
These areas of undercutting were measured along 30 mm of References
the cut-edge, therefore, the equivalent undercutting distance
from the cut-edge is between 0.3 and 15 mm. [1] J.A. Cox, Kempster, R. Bassi, A survey of performance of organic-
The amount of undercutting produced in cyclic cabinet coated metal roof sheeting, Building Research Establishment Report,
Watford, 1993.
test 1 (Table 3), is similar to that exhibited by the specimens [2] P.A. Ryan, R.P. Wolstenholme, D.M. Howell, Durability of Cladding,
examined in the present investigation after 5 years exposure Bath Press, 1994.
at an industrial/marine exposure site (i.e., 1  0.5 mm). [3] J.B. Harrison, T.C.K. Tickle, J. Oil Col. Chem. Assoc. 458 (1962)
Therefore, the cyclic cabinet tests are able to produce 571.
undercutting equivalent to that produced in outdoor expo- [4] J.B. Harrison, J. Oil Col. Chem. Assoc. 62 (1979) 18.
[5] F.D. Timmins, J. Oil Col. Chem. Assoc. 62 (1979) 131.
sure after 5 years. Cyclic cabinet test 2 (20 min wet/40 min [6] S.B. Lyon, J.D. Scantlebury, J.B. Johnson, in: G.S. Haynes, K.
dry) produced slightly more undercutting than cyclic cabinet Tellefson (Eds.), Cyclic Cabinet Corrosion Testing, ASTM STP
test 1 (30 min wet/150 min dry). However, this is consider- 1238, ASTM, Philadelphia, PA, 1994.
ably less than that produced at a severe marine exposure site [7] G.W. Walter, Corros. Sci. 355±8 (1993) 1391.