Intra-Laboratory

Commercial Test
Chloride on Steel
By Shuang-Ling Chong, Ph.D., Federal Highway Administration, and Yuan Yao

Fig. 1: Injection of extraction

fluid into the patch
Photos courtesy of the authors

B
illions of dollars are spent each year for bridge reha-
bilitation and maintenance due to corrosion problems.
The corrosion process is caused mainly by moisture,
oxygen, and electrolytes. The most common electrolytes
occurring in the natural outdoor environment are chlo-
rides, sulfates, and nitrates. Of these three electrolytes, chlorides have
the greatest effect on the corrosion of steel. Therefore, the coating perfor-
mance depends largely on chloride concentrations existing in the exposure
environment.1 High concentrations of chloride on steel bridge surfaces
cause early coating failures such as rust, coating delamination, etc. The chlo-
ride threshold limit causing failure varies with coating type and service envi-
ronment; for example, zinc-rich primers can tolerate a higher chloride concen-
tration than barrier primers without sacrificing performance significantly.2 For
cases involving immersion, such as Navy ships and similar applications, it is nec-
essary to specify and achieve low chloride levels before painting. Chloride levels are

42 JPCL August 2003 PCE www.paintsquare.com

Assessment of
Kits for Quantifying
Surfaces
and Muriel Rozario, Soil and Land Use Technology, Inc.
set at 35 g/cm2.3 Therefore, assurance of
steel surface chloride at a specified low
level becomes a critical step before any
painting job.
In 1991, the Federal Highway Ad-
ministration (FHWA) completed a
contract research study entitled
Effect of Surface Contaminants
on Coating Life. In it, numer-
ous chloride test methods were
investigated and compared; the
study investigated the influence
of soluble salts on the perfor-
mance and lifetime of protective
coatings on steel bridges.4
Subsequently, new chloride test kits
have been developed; these test kits
have not been evaluated with regard to
measurement reliability.5 Currently, three Fig. 2: Chloride extraction using
popular commercial chloride test kits are
used for determining chloride concentration
on steel surfaces. These tests are the swab test
(manufacturer X), patch test (manufacturer Y),
and sleeve test (manufacturer Z); their detectors
are ion detection strip, bottle titration liquids, and
ion detection tube, respectively. Unfortunately, re-
sults from these tests have been inconsistent and
highly variable. These discrepancies are believed to
derive from both different extraction efficiencies and
detection sensitivities in the tests as well as operator
variability. Hence, FHWA investigated the variations and
limitations of these test methods in order to establish tech-
www.paintsquare.com JPCL August 2003 PCE
niques that can be used to obtain reliable and ac-
curate field test results. The current practice
allows a four-hour maximum exposure
(aging) of bridge surfaces to the atmos-
phere after abrasive blast cleaning and
before painting. The laboratory ex-
periments in this study were there-
fore designed to calculate the per-
cent of chloride recovery from both
freshly-doped steel panels and
doped and aged steel panels. More
importantly, this research aimed to
establish guidelines for performing
field tests that would yield accurate
and precise results.
Experimental Procedures
All the test panels were SSPC-SP 10 prepared
steel panels of 15 cm x 10 cm x 0.32 cm
the sleeve test (6 in. x 4 in. x 18 in.) and were doped with
chloride by the same operator. Four different lev-
els of chloride concentration, namely, 3, 5, 9, and
30 g/cm2, were applied to the steel panels in ac-
cordance with Mitschkes technique.6 The chloride
tests were performed under three different conditions:
(1) within one minute after the panels were doped, (2)
after aging doped panels at moderate humidity
(57 3% relative humidity (RH) at 37 2 C (99 3 F)
for 4 hours, and (3) after aging doped panels at high hu-
midity (78 3% RH) and at 37 2 C (99 3 F) for 4 hours.
The humidities in cases (2) and (3) were established by the
saturated salt solution of magnesium nitrate hexahydrate and
43
 Fig. 3: Chloride recovery from freshly doped steel surfaces by different extraction-analytical kits

2
Chloride Recovery, %

a. Mean values are shown for patch extraction

b, Error bar = 1 standard deviation
c. Extracted area: Swab (150 cm2), Patch (12.25 cm2), Sleeve (10 cm2)

potassium chloride at 37
Table 1: Number of Test Panels for Elements
2 C (99 3 F) in a vac-
of Chloride Test Matrix
uum oven, respectively.
All the extractions and the Technique Concentration of Chloride Total #
chloride measurements (microgram per sq centimeter) of Panels
were made at ambient 3 5 9 30
No aging
conditions.
swabbing 9 9 9 9 36
Three operators at the patch 9 9 9 9 36
Paint and Corrosion Labo- sleeve 9 9 9 9 36
ratory at the Turner-Fair- Aging at 37 C & 57% RH
bank Highway Research swabbing 9 9
Center (TFHRC) of FHWA patch 9 9
participated in this study sleeve 9 9
Aging at 37 C & 78% RH
in order to investigate the
swabbing 9 9
intra-laboratory repeata- patch 9 9
bility of the current chlo- sleeve 9 9
ride test kits. Each test Totals for concentrations 27 27 27 81 162
was performed in tripli-
cate by each of three operators; all the tests were conducted on the test panels in a vertical posi-
tion. The complete test matrix used in this study is shown in Table 1. To identify the best test kit or
any extraction and detection combination test to achieve optimum extraction efficiency, different de-
tection techniques were used to determine the chloride concentration for each extraction method
(Table 2). Each extraction and each detection technique are described in detail below. Boiling water ex-
traction of clean steel panels was performed to determine the background chloride level; all the reported
chloride data have been corrected by this trace amount of chloride contamination on the blasted steel.

Extraction Method
For each extraction/detection scenario, three measurements were obtained for each of three tests for each
chloride concentration. De-ionized (DI) water used in all the extractions and dilutions had conductivities
lower than 2 microsiemens.
Swab Testa surface area of 150 cm2 (15-cm long x 10-cm wide steel panel) was extracted with a total of 15

44 JPCL August 2003 PCE www.paintsquare.com

 Fig. 4: Chloride recovery from steel surface doped with 30 g/cm2 after different aging conditions

ml of DI water for cotton swab-

bing; the test employed is sim-
Chloride Recovery, %

ilar to the method in SSPC-TU

4 section 3.4.7 Only one-third
of each of the four cotton balls
used for swabbing was wetted
with DI water to avoid any
dripping; the fifth cotton ball
was used for absorbing all the
remaining liquid on the steel
panel. Each cotton ball was
held with a pair of tweezers
during the operation. The ex-
tracted solution was divided
into separate portions so that
a. Swab extraction solution: pH = 6.7.
b. Patch extraction solution: pH = 3.9 four different detection tech-
c. Sleeve extraction solution: pH = 4.2 niques (chloride electrode,
d. Mean values are shown for patch extraction.
e. Error bar = 1 standard deviation chemical titration, ion detec-
tion strip, and ion detection
tube) could be used to deter-
Table 2: Extraction and Detection mine the chloride concentra-
Combination Test Matrix tion for each doped chloride
level.
Extraction Chloride Ion Ion Detection Titration Ion Detection
Patch Test8For each patch
Method Electrode Strip Tube
Swab test X X X X covering an area of 12.25 cm2,
Patch test N/A X X X 1.5 ml of extraction fluid was
Sleeve test N/A X N/A X injected into the patch after the
N/A: not applicable. Test was not performed due to ion interference by extraction fluid patch was glued securely on
the steel surface (Fig. 1). Two-
thirds of the fluid was then withdrawn and re-injected to produce vigorous mixing. The patch was rubbed
with the finger for one minute to favor chloride solubility. Another 1.5 ml of extraction fluid was subse-
quently used to rinse the area in the same manner. The two extractions were combined and the resulting ex-
tract (3 ml) was titrated with reagents included in the kit. Another set of extractions was conducted for chlo-
ride determination by an ion detection strip and an ion detection tube. A chloride ion electrode was not used
in this test because it did not provide reasonable readings for chloride due to the interference by the acidity of
the extraction fluid (pH = 3.9) and the presence of other ions.
Sleeve TestThe entire 10-ml extraction fluid was poured into a sleeve that was then firmly attached to the steel
panel. The free end of the test sleeve was lifted and held upright with one hand to allow the extraction fluid to
come in contact with the test surface; the other hand was used to massage the solution through the test sleeve
against the steel surface for 2 minutes (Fig. 2). When extraction was completed, the test sleeve was removed. To de-
termine chloride concentration, both ends of an ion detection tube were snapped off with a metal snapper and the tube
was inserted into the test sleeve. After about 2 minutes, solution migrated to the top of the ion detection tube and a
reading in parts per million (ppm) was recorded at the color break point, i.e., between the orange color zone and the
white color zone. Frequently, a suction bulb using gentle squeezing was needed to initiate the solution flow into the tube.

Chloride Detection Method

The accuracy of each lot of all the detectors was verified with known chloride standards before their use.
Chloride Ion ElectrodeThe chloride ion electrode provides the most accurate reading for chloride concentration and is the
www.paintsquare.com JPCL August 2003 PCE 47
best estimate of the chloride
extracted by each aqueous so- Table 3: Comparison of Three Chloride
lution. Prior to chloride ion Extraction Test Methods
electrode measurement, a cal-
ibration curve of millivolts Swab Test Patch Test Sleeve Test
System type Open Closed Closed
(mv) versus ppm was ob-
Area, cm2 150 12.25 10
tained for the chloride ion Extraction fluid, ml 15 3 10
electrode using six chloride Area (cm2)/volume
solutions with known concen- (ml) 10 4.1 1
trations of 2, 5, 10, 25, 50, and pH value of fluid DI water, pH = 6.7 Acidic, pH = 3.9 Acidic, pH = 4.2
100 ppm. Chloride concentra-
tion of extracted solutions
was adjusted by dilution to Table 4: Chloride Concentration Determined by Swab
the calibration curve range as Test Using Different Detection Techniques
necessary; the final chloride Initial Chloride Chloride Ion Ion Detection Titration Ion Detection
concentration (ppm) of the Concentration Electrode Strip Tube
extracts was calculated from 3 g/cm2 2.6 (0.2)a 2.1 (0.2) 4.2 (0.5) 2.2 (0.1)
the mv reading and the cali- 5 g/cm2 4.7 (0.7) 3.8 (0.3) 5.1 (1.7) 4.2 (0.4)
bration curve. 9 g/cm2 8.8 (0.4) 8.2 (0.4) 8.5 (2.2) 7.7 (0.5)
30 g/cm 2 30.8 (2.5) 28.5 (1.0) 31.1 (0.0) 27.4 (1.1)
Ion Detection StripNo di-
lution is needed for this detec- a: standard deviation
tion method since the ion de-
tection strip covers a large range of chloride concentrations. However, the largest
flaw for the strip method is that no reading can be obtained for chloride concen-
trations below 30 ppm.
TitrationThis detection technique using bottle titration liquids in the test kit gives
only a range of chloride concentration; no exact value can be obtained. Dilutions are
needed for the extracts from the swab test to provide a narrower reading of concentra-
tions and more accurate readings in the lower range scale of the chart provided. The re-
ported values in this study were the mean value of the minimum and the maximum read-
ings found from the chart on the basis of titration data. Even though the titration method
generally can generate an exact value, no such values are available from the chart provided in
the test kit.
Ion Detection TubeThe ion detection tube is limited to a maximum extract concentration of
50 ppm. It is essential to keep chloride readings on the scale; dilution was made for the solutions
extracted from large panel areas such as those subjected to
the swab test or from the panels that were doped with Table 5: Percent of Apparent Chloride Recovery
high chloride concentrations. from Freshly Doped Panels for Swab Test
Table 3 summarizes different parameters used in the by Different Detection Techniques
three extraction test methods: system type, area, extrac-
tion fluid volume, ratio of area to fluid volume, and pH Initial Chloride Ion Ion Detection Titration Ion
Chloride Electrode Strip Detection
value of extraction fluid.
Concentration Tube
% of Apparent Chloride Recovery
Results and Discussion 3 g/cm2 86.7 (6.7)a 70 (6.7) 140 (16.7) 73.3 (3.3)
Comparing Different Analytical Methods 5 g/cm2 94.0 (14) 76.0 (6.0) 102 (34) 84.0 (8.0)
for Constant Extraction Method 9 g/cm2 97.7 (4.4) 91.1 (4.4) 94.4 (24.4) 85.6 (5.6)
The chloride concentrations determined by different com- 30 g/cm2 103 (8.3) 95.0 (3.3) 104 (0) 91.3 (3.7)
binations of extraction methods and detection techniques a: standard deviation

www.paintsquare.com JPCL August 2003 PCE 49

 are described in the following
Table 6: Chloride Concentration Determined by Patch paragraphs. The chloride cont-
Test Using Different Detection Techniques amination level on the SSPC-SP
10 steel surface before contam-
Initial Chloride Titration Ion Detection Ion Detection
Concentration Strip Tube ination was found to be 0.74
Chloride Concentration Detected, g/cm2 g/cm2; this value was the av-
3 g/cm2 2.5 (0.7)a ND 1.5 (0.4) erage of the chloride concentra-
5 g/cm2 4.5 (0.7) ND 3.2 (0.3) tions obtained by boiling two
9 g/cm2 10.7 (1.2) 6.2 (0.6) 5.7 (0.5) clean SSPC-SP 10 steel panels
30 g/cm2 30.4 (7.4) 22.7 (0.9) 19.3 (1.1) with a standard deviation of
a: standard deviation ND: not detectable 0.03 g/cm2. This value was
then subtracted from the mea-
sured value to get the actual
Table 7: Percent of Apparent Chloride Recovery chloride concentration in each
from Freshly Doped Panels by Patch Test test.
Swab TestThe most com-
Initital Titration Ion Detection Ion Detection
Chloride Strip Tube plete results were obtained by
Concentration the swab extraction method
% of Apparent Chloride Recovery that enabled the use of all four
3 g/cm2 83.3 (23.3)a ND 50.0 (13.3) detection techniques because
5 g/cm2 90.0 (14.0) ND 64.0 (6.0) the DI water extracts could be
9 g/cm2 119 (13.3) 68.9 (6.7) 63.3 (5.5) analyzed without problems.
30 g/cm2 101 (24.7) 75.7 (3.0) 64.3 (3.7) The test results for freshly
a: standard deviation ND: not detectable
doped panels are presented in
Table 4.
The extraction efficiency by the cotton swab should be the value measured by
the chloride ion electrode that is supposed to provide the most accurate measure-
ment. Such values are generally higher than those obtained by the other three de-
tection methods. The chloride concentrations obtained by titration are sometimes
high because the reported mean value does not represent the actual value but gives
only an estimate. The apparent chloride recovery by each detection technique is pre-
sented in Table 5. It was calculated by the measured value divided by the doped value
and then multiplied by 100.
The test results suggest that chloride recovery by the swab test for all the concentrations
are very high; however, detection sensitivity generally increases with increasing chloride
concentration. The low sensitivity at low chloride concentra-
Table 8: Chloride Concentration Determined by tions is caused by the presence of fewer chloride ions that can
Sleeve Test Using Different Detection Techniques be measured by detectors, especially with the ion detection
strip and ion detection tube. High chloride recovery values im-
Initial Ion Detection Tube Ion Detection Strip ply easy extraction of chloride from steel and no interference
Chloride with chloride determination in water by all four detection
Concentration
methods. Overall, these test results essentially verify the relia-
Chloride Concentration Detected, g/cm2
bility and sensitivity of the test kits.
3 g/cm2 0.7 (0.4)a ND
5 g/cm2 3.3 (0.6) ND
Patch TestOnly three detectors could be employed for chlo-
9 g/cm 2 8.4 (1.1) ND ride measurements in the patch test using acidic extraction flu-
30 g/cm2 32.1 (6.3) 33.4 (0.0) id because reasonable readings were not obtainable with the
a: standard deviation ND: not detectable
chloride ion electrode. All the test results are listed in Table 6.
Furthermore, no readings could be observed on the ion detection
50 JPCL August 2003 PCE www.paintsquare.com
 Fig. 5: Percent of chloride recovery from freshly doped panels using acidic fluid and DI water
Chloride Recovery, %

Acidic Fluid DI Water

Extraction Fluid
a. Error bar = 1 standard deviation

strip for the extracted solutions of steel panels doped with
2
3 and 5 g/cm . The percent of apparent chloride recovery
for all the tests is shown in Table 7.
The chloride recoveries by the ion detection strip and the
ion detection tube in the patch test were much lower than
those obtained by titration; the values were low even for
the 30 g/cm2 chloride level. The former two detection
methods both convert silver chromate impregnated on the
strip or on the tube packing into silver chloride in the pres-
ence of chloride ion. It is known that chromate ion changes
It is clear that chloride concentrations be-
low 30 ppm (i.e., 30 g/cm2 in the sleeve
test) are not detectable as indicated by the
ion detection strip. On the other hand, the
ion detection tube was capable of detecting
chloride concentrations extracted from panels
doped with concentrations ranging from 3 to 30
g/cm2, but the recoveries were low for low chlo-
ride levels. The recovery for doped chloride of 3
g/cm 2 was found to be only 23.3 13.3% demon-

to dichromate in the presence of mineral acid (H ); the + strating the low sensitivity and high variability of the
acidic medium of the patch extraction fluid may have re- ion detection tube for low chloride levels. The high
duced the formation of silver chloride. These two detection readings obtained at high chloride concentrations sug-
methods produce lower than actual chloride readings for gest that the sensitivities of ion detection strips and ion
the patch test; therefore, the titration technique is the only detection tubes are not affected by the acidic extraction
reliable detection method for the patch test based on the fluid provided in the sleeve test.
reported mean values.
Sleeve TestFor the extracts ob- Comparing Extraction
Table 9: Percent of Apparent Chloride Efficiency for Different
tained from the sleeve test, only the
Recovery from Freshly Doped Panels by
ion detection tube and ion detection Extraction-Analytical Kits
Sleeve Test Method
strip were employed to measure Figure 3 summarizes all the
chloride concentration. The chlo- chloride test results with error
Initial Ion Detection Ion Detection
ride concentration and percent of Chloride bars shown for each measured
Tube Strip
apparent chloride recovery results Concentration value in each scenario. This fig-
are displayed in Tables 8 and 9, re- % of Apparent Chloride Recovery ure includes data from three test
spectively. It should be noted here 3 g/cm 2 23.3 (13.3) a ND methods for four different chlo-
that the apparent chloride electrode 5 g/cm 2 66.0 (12.0) ND ride concentrations. As seen in
9 g/cm 2 93.3 (12.2) ND this bar graph, the swab test
and titration method did not give re-
2
alistic readings; therefore, their re- 30 g/cm 107 (6.3) 111 (6.0) generated the most reproducible
sults are not shown. a: standard deviation ND: not detectable data with recoveries ranging

www.paintsquare.com JPCL August 2003 PCE 53

 Fig. 6: Percent of chloride recovery from doped/aged panels using acidic fluid and DI water
Chloride Recovery, %
Acidic Fluid
a. Error bar = 1 standard deviation
from approximately 70 to 95%. The patch test also recov-
ered a high amount of chloride but with much larger er-
rors. The recovery by the sleeve test decreased with de-
creasing chloride concentration and showed moderate
errors.
Comparing Extraction Efficiency for Different
Extraction-Analytical Kits for Aged Panels
In reality, steel surfaces are exposed to atmospheric condi-
tions for a period of time (up to 4 hours) prior to painting.
For this reason, two aging conditions were included in the
testing to investigate the effect of aging or weathering.
Only a chloride concentration of 30 g/cm2 was used for
this study. Each extraction test used the chloride detection
method that was recommended for that test. When the
doped panels were aged at high temperature (37 C) for 4
hours under two different humidity conditions (57% RH
and 78% RH), the chloride recovery varied to a different
degree. The test results (Table 10) showed that recovery did
not differ much under the conditions of 57% RH and 37 C
as compared to that from freshly-doped panels, but they
decreased significantly at 78% and 37 C for all three tests.
Figure 4 provides a better illustration of the the aging effect
with error bars marked on each measured value. The test
results show that exposure in a hot and humid environ-
ment can significantly reduce the chloride extractable by
the current test kits.
It is likely that very low levels of rust formed after ex-
posing the steel panel at 78% RH and 37 C for 4 hours,
making the entrapped chloride more difficult to extract and
54
DI Water
Extraction Fluid
reducing chloride recovery. These test
results demonstrate that surface chloride
concentration should be determined as
soon as possible after abrasive blasting;
any delay, especially after exposure to high
humidity and temperature, will result in low-
er than actual chloride values.5 Early failures
could occur if painting decisions were made on
the basis of these chloride values.
Effect of Acidity of Extraction Fluid
on Chloride Recovery from Patch Test
To investigate the effect of the acidity of the extrac-
tion fluid on chloride recovery, the patch test was con-
sidered the best one to be used for this study. The fol-
lowing test series was conducted for comparing
extraction efficiency for the same test using different flu-
ids, because two variations of the patch test are being em-
ployed in the coating industry: (1) acidic fluid8 using titra-
tion for detection and (2) DI water using ion detection strip
for detection. The chloride concentration of 30 g/cm2 was
doped on test panels in all cases and only two operators
conducted the tests. All tests were performed in triplicate.
The tests were performed on the freshly doped panels and
doped, aged panels separately. Three detectors were used:
titration, ion detection tube, and ion detection strip. In
each test, acidic fluid and DI water were used separately
for extraction to compare their extraction efficiency.
Freshly Doped PanelsChloride recovery from freshly
doped panels extracted with acidic fluid was 105% when
JPCL August 2003 PCE www.paintsquare.com
 measured using the titration
method; when DI water was used Table 10: Percent Chloride Recovery from Steel
as the extraction fluid, the recov- Surfaces Doped with 30 g/cm2 Chloride after Aging
ery decreased to 86.0%. Chloride
recovery from the freshly doped Aging Conditions Swab Testa Patch Testb Sleeve Testc
panels extracted with acidic fluid % of Chloride Recovery
and measured with the ion detec- No aging
(freshly doped) 95.0 (3.3)d 101 (24.7) 107 (6.3)
tion tube and ion detection strip
Aged at 37 C
were about 70% for the acidic flu- and 57%
id and DI water, with the former RH for 4 hours 80.7 (7.5) 97.7 (6.7) 99.3 (4.9)
showing no enhancement when Aged at 37 C
compared with the titration and 78% RH
method. for 4 hours 43.6 (4.5) 79.9 (6.7) 60.4 (6.5)
Freshly Doped and then Aged a: Ion detection strip as detection method
PanelsIt is more difficult to ex- b: Titration as detection method
c: Ion detection tube as detection method
tract chloride from doped and d: Standard deviation
aged panels with DI water than
with acidic fluid. Using titration,
the recoveries with acidic fluid
and DI water were 79.8% and Table 11: Chloride Extraction Efficiency
62.0%, respectively. With the ion from Steel Surfaces Doped with 30 g/cm2
detection tube test, the recoveries Chloride by Acidic Fluid and DI Water in
with acidic fluid and DI water Patch Test
were 61% and 40.3%, respective-
Detector for Acidic Fluid DI Water
ly. With the ion detection strip,
Chloride Ions
the recoveries with acidic fluids Chloride Concentration,
and DI water were 76% and g/cm2 (% of Chloride Recovery)
38.7%, respectively. These test re- Freshly doped panel:
sults indicate that acidic fluid can Titration 31.6 (105) 25.6 (86.0)
extract up to 30 to 50% more chlo- Ion detection tube 19.4 (64.7) 21.7 (71.0)
Ion detection strip 22.7 (75.7) 21.4 (71.3)
ride than DI water (Figs. 5 and 6
Freshly doped and then
and Table 11). In fact, all the read- aged panel:
ings obtained by ion detection Titration 24.0 (79.8) 18.6 (62.0)
tube and ion detection strip are Ion detection tube 18.3 (61.0) 12.1 (40.3)
lower than the actual chloride val- Ion detection strip 22.8 (76.0) 11.6 (38.7)
ues as mentioned earlier. There-
fore, it is clear that acidic fluid can
extract much more chloride for fresh-
ly doped and then aged panels.

Strengths and Weaknesses of Chloride

Extraction and Analytical Methods
Swab Test Using Ion Detection Strip as
DetectorThe swab had the highest chlo-
ride recovery due to the largest ratio of area
to extraction fluid, that is 10 as compared to
4.1 and 1 for the patch test and sleeve test, re-
spectively. More chloride ions are present per
milliliter of extracted solution from the swab test
56 JPCL August 2003 PCE www.paintsquare.com
 Table 12: Strengths and Weaknesses of Three
Chloride Extraction and Detection Test Methods

Extraction
Method Swab Test Patch Test Sleeve Test
Chloride
concentration High (10)a Medium (4.1) Low (1)
Loss of Yes No, if patch No, if patch
extraction fluid adhered to adhered to
steel firmly steel firmly
Threshold limit 3 g/cm2 ~ 1 g/cm2 ~ 5 g/cm2
Detection
Method Ion detection strip Titration Ion detection tube
Reproducibility High (above 30 ppm) Low (only minimum and High (above 9 ppm)
of detection maximum values given)
method
Detection > 30 ppm > 1 g/cm2 > 5 ppm
sensitivity
Detection range 30600 ppm 150 g/cm2 150 ppm
a: Extracted area/extraction fluid volume

such that accuracy increases. However, water dripped easily during swabbing, partic-
ularly at the overhead areas of steel structures, resulting in lower chloride values.
The ion detection strip can detect chloride concentration only above 30 ppm. Never-
theless, the large extraction area compensates for this detection limit; it was found that 3
g/cm2 was about the lowest limit if extraction was performed properly.

Patch Test with Titration as DetectorThe patch test provides a closed extraction system
that prevents fluid evaporation and loss. However, there are two other possible extraction flu-
id losses: (1) if the patch is not adhered to the steel surface firmly, and (2) if the syringe is in-
serted improperly in the extraction procedure, causing leaks. Loss of some extraction fluid will
generate inaccurate chloride data.
If the titration end point (i.e., color change) is at the borderline between two drops, the number of
drops obtained by different operators or even by the same operator in triplicate tests will result in dif-
ferent chloride concentrations, as indicated by larger error bars in general. There is one additional step
for consideration, i.e., the extracted solution containing mercury has to be saved for hazardous waste dis-
posal.

Sleeve Test with Ion Detection Tube as DetectorThis test is also a closed system without much risk of flu-
id loss. However, there is possible extraction fluid loss if the sleeve does not adhere to the steel surface firm-
ly. Loss of some extraction fluid will generate unreliable chloride data. This sensitivity would be increased if
the extraction volume is reduced or extracted area is increased.
The ion detection tube is not very sensitive below 5 ppm.
To summarize, the strengths and weaknesses of the test methods are organized in Table 12.

Summary and Conclusions

The efficiency of three commercial methods for extracting chloride from contaminated steel surfaces was investi-
gated and compared. Each test was repeated three times by each of three operators.
The swab extraction method provides high chloride recovery in a laboratory setting and good reproducibility from
steel surfaces doped with chloride ranging in concentrations from 3 to 30 g/cm2. Liquid loss in this open system
www.paintsquare.com JPCL August 2003 PCE 59
is a major problem; extreme care is
needed to perform this test properly.
The patch extraction and detection kit
also provides high chloride recovery. How-
ever, only minimum and maximum values
are given; the actual values are not known.
The maximum value, however, can be used as
a guideline before a decision on a painting job
is to be made. Acidic fluid is found to be more ef-
fective to extract chloride than de-ionized water
for the patch test.
The sleeve extraction and detection kit provides a
high recovery for chloride concentrations of 9 and 30
g/cm2; however, the recovery at concentrations of 3
and 5 g/cm2 is poor due to the low sensitivity and un-
clear color separation at the low reading end of the ion de-
tection tube.
After chloride-contaminated steel panels are aged at 37 C
and 57% RH for 4 hours, the extraction recovery differs
only slightly from that for freshly doped panels, and recov-
eries are all close to 100%. However, when humidity reach-
es 78% RH at 37 C, the extraction efficiency decreases sig-
nificantly.
Chloride ion electrodes, titration, ion detection strips,
and ion detection tubes all can be used to measure chloride
ions in the extracts from the swab test. Both the ion detec-
tion strip and the ion detection tube are applicable to the
sleeve test, whereas titration is the only suitable detection
method for the patch test.
Guidelines and Recommendations for Performing
Chloride Extractions Efficiently and Accurately
It is important that operators obtain good training on
each extraction method using steel panels that are freshly
doped with known chloride concentration before perform-
ing field tests.
It is essential to verify the accuracy of a detector in each
batch with known chloride standards before performing
field tests.
It is critical to prevent loss of water solution during swab
extraction. Each of the four cotton balls used for extraction
should be soaked with an amount of water that wets no
more than one-third of the cotton ball. An effort should be
made to mix cotton balls with water thoroughly after ex-
traction and then to squeeze out all extracts.
For the patch test, acidic fluid instead of DI water is rec-
ommended. This will result in higher extraction efficiency;
however, this restricts the method of detection to titration.
60 JPCL August 2003 PCE
Sensitivity of tests will be increased if large areas are ex-
tracted and smaller amounts of extraction fluid are used
for extractions.
The operator should conduct extractions as soon as pos-
sible after the blast because exposure of blast-cleaned steel
to high humidity can affect the level of chloride measured.
References
1. P. Ault, J. Ellor, J. Repp, and B. Shaw, Characterization
of the Environment, FHWA Publication, FHWA Publi-
cation No. FHWA-RD-00-030, 95 pp., August 2000.
2. M. Morcillo, S. Feliu, J.C. Galvan, and J.M. Bastidas,
Some Observations on Painting Contaminated Rusty
Steel, JPCL (September 1987), pp. 3843.
3. B.P. Alblas and Ing A.M. van Londen, The Effect of
Chloride Contamination on the Corrosion of Steel Sur-
faces: A Literature Review, PCE (February 1997), pp.
1625.
4. B. Appleman, S. Boocock, R. Weaver, and G. Soltz,
Effect of Surface Contaminants on Coating Life,
FHWA Publication, FHWA Publication No. FHWA-RD-
91-011, 256 pp., November 1991.
5. S. Flores, J. Simancas, and M. Morcillo, Methods for
Sampling and Analyzing Soluble Salts on Steel Sur-
faces: A Comparative Study, JPCL (March 1994), pp.
7683.
6. H. Mitschke, Effects of Chloride Contamination on Per-
formance of Tank and Vessel Linings, JPCL (March
2001), pp. 4956.
7. B. Appleman, J. Rex, and T. Sowers, SSPC-TU 4, Field
Methods for Retrieval and Analysis of Soluble Salts on
Substrates, SSPC Painting Manual, Vol. 2, Eighth Edi-
tion (Pittsburgh, PA: SSPC, August 1998).
8. ISO Method 8502-10:1999, Preparation of Steel Sub-
strates before Application of Paints and Products
Tests for the Assessment of Surface CleanlinessPart
10: Field Method for the Titrimetric Determination of
Water-Soluble Chloride.
www.paintsquare.com