Materials and Corrosion 2012, 63, No. 9999 DOI: 10.1002/maco.

201206655 1

Application of automated corrosion sensors for monitoring

the rate of corrosion during accelerated corrosion tests
T. Prosek*, N. Le Bozec and D. Thierry

An automated corrosion monitoring system using the electrical resistance

technique was applied for assessment of the corrosivity towards carbon steel
and zinc in different phases of a complex accelerated corrosion test recently
introduced by VDA, an association of German car makers. It comprises salt spray,
wet, dry, and freezing phases. The developed small and battery-driven
atmospheric corrosion loggers provided high sensitivity allowing for sub-
angstrom (<1010 m) measurements of corrosion depth and good accuracy. The
actual corrosion rate was affected by the exposure history due to a limited rate
of wetting/drying and oxygen and ion transport to the reaction interface under a
layer of corrosion products. The hysteresis was particularly strong for carbon
steel. Except the freezing phase, the steel corrosion rate varied in a narrow
range from 0.2 to 0.6 mm/h. For zinc, the corrosion rate varied from 0.001 to
0.1 mm/h in particular phases of the cycle with the maximum in the salt spray
phase. Seventy-five percent of the metal corroded in the salt spray phase and in
the following drying period representing only 13% of the total test time. The
obtained data suggest that the proposed test cycle allowed for rather efficient
drying of the zinc surface, which is believed to be crucial for the formation of
corrosion products with certain protective ability observed also in field
conditions.

1 Introduction Several techniques for corrosion monitoring by optical

means have been proposed. A metal clad optical sensor is formed
Whereas there is a relatively wide range of products for corrosion from a standard multimode optical fiber where the plastic
monitoring in aqueous media, the availability of means for real- cladding is replaced by a metal film. The metal cladding increases
time corrosion monitoring in the atmosphere is limited. A quartz radiation losses, since the fiber is no longer a wave-guide in the
crystal microbalance (QCM) measures corrosion through metal-coated region. As the metal oxidizes, light losses decrease
changes in the resonance frequency of a piezo electric quartz and the output power from the optical fiber increases [1, 5].
crystal coated with a thin metallic layer, the resonance frequency However, there is not yet any commercial product based on the
being a function of the mass of the crystal [1, 2]. This technique, optical methods on the market.
based on evaluation of mass gain of the metal coating during the Ultrasonic testing is a well-established non-destructive
corrosion process, can be applied to monitor corrosivity under technique used to control the corrosion state of metallic
mild atmospheric conditions [3]. A disadvantage of this method is structures. It consists in generating an acoustic wave that
that it is equally sensitive to any mass gain, e.g. from dust and propagates in the tested material and in recording reflections
other contaminants, as to the formation of corrosion products. returning from the surface of the objects or from internal
Further, water is adsorbed on the surface, and its amount depends heterogeneities. The accuracy in field conditions is usually
on the relative humidity, metal, surface roughness and surface limited to a few hundreds of micrometers on corroded materials
film present [4]. These facts limit the technique for applications in [6], but better results can be obtained in laboratory conditions
low corrosive environments. [7, 8]. The drawback of the technique is that it is sensitive to the
ultrasonic properties of specific corrosion products [8]. Further, it
can be applied only for steel objects.
An interesting alternative to the traditional coupon technique
T. Prosek, N. Le Bozec, D. Thierry
French Corrosion Institute, 220 rue Pierre Rivoalon, 29200 Brest and methods discussed above is the electrical resistance (ER)
(France) technique described for instance in standards ASTM G96 and
E-mail: [email protected] ISO 11844–2 [9, 10]. The principle of the method is that ER of a

www.matcorr.com wileyonlinelibrary.com ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2 Prosek, Le Bozec and Thierry Materials and Corrosion 2012, 63, No. 9999

measuring element made from the material of interest increases

as its cross sectional area diminishes due to corrosion. In practice,
two such elements are built into a probe. One is exposed to the
corrosive environment and corrodes whereas the other is
shielded and thus protected from corrosion. The resistance is
measured at the same time and resistivity changes due to
temperature difference are compensated. Based on the initial
cross-sectional area of the exposed element, the cumulative metal
loss at the time of reading is determined. The benefits of the ER
technique include its versatile nature; the technique can be used
in practically any kind of environment with any material
corroding about uniformly.
The ER monitoring system AirCorrTM has been developed
within European projects [11–13]. It exists in two basic variants
designed for low corrosive environments where the sensitivity is Figure 1. Schematic drawing of corrosion sensor (left) and photograph
crucial and for highly corrosive applications where robustness of Fe-250 mm sensor (right)
and water tightness are of the utmost importance. The former
system was applied for monitoring in indoor cultural heritage
premises where sub-angstrom (<1010 m) sensitivity was exposure, Rsens and Rref are the actual ERs of the sensor and
reached using thin copper, silver, lead, iron and bronze sensors reference tracks, and Rsens,init and Rref,init are the initial ERs of the
with thicknesses from 50 to 800 nm [14, 15]. The latter set-up was sensor and reference tracks. The calculation is based on the ERs
used on operating vehicles in order to assess the corrosion measured as a potential difference along the track through which
aggressiveness towards steel and zinc as a function of locations on defined currents pass [18, 19].
vehicle and thus possibly adapt the corrosion protection [16]. The The changes in ER are recorded by a precise miniature
corrosion loggers were successfully applied also in extreme electronic logger. For purpose of this study, watertight AirCorr O
corrosive environments represented by outdoor locations on logger of 100  65  37 mm3 with a tightness of IP 65 designed
operating ships such as a container carrier navigating between for highly corrosive environments was used, see Fig. 2. The
Europe and China [17]. attached sensor is protected with a robust polyurethane casting
The aim of this study was to monitor the corrosion rate of and can be replaced when reaching the end of the life. The
carbon steel and zinc during a complex accelerated corrosion test cadence of the measurement can be adjusted from minutes, to
including wet, dry and salt spray phases. Hence, the influence of hours, to even days if necessary in order to match the sensitivity of
the different phases on the corrosion rate during the accelerated the measurement with the anticipated corrosion rate. It was set to
corrosion test could be studied. 5 min in this study. Details about reproducibility, sensitivity and
accuracy of the technique can be found in reference [14].

2 Experimental 2.2 Accelerated corrosion test

2.1 AirCorr corrosion loggers Among accelerated corrosion tests that are used by the
automotive makers, the so-called N-VDA test was selected to
The concept of the applied corrosion monitoring system is
simple: the electronic unit measures and registers the change
over time of ER of a thin metal track applied on an insulating
substrate. If the metal corrodes, the cross-sectional area of the
track decreases and the ER increases. A part of the metal track is
protected by an organic coating or tape and, thus, serves as a
reference to compensate for resistivity changes due to varying
temperature. A schematic drawing and photograph of the sensor
is shown in Fig. 1. This geometry ensures high sensitivity to
changes in the ER due to metal corrosion.
The corrosion depth (CD) of the metallic sensor is calculated
according to equation

 
Rref ;init Rref
CD ¼ tinit  (1)
Rsens;init Rsens

where tinit is the initial thickness of the reference metallic track, Figure 2. Photograph of watertight AirCorr O logger with mounted
which is assumed equal to the sensor track in the beginning of Fe-250 mm sensor for heavy corrosive environments

ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.matcorr.com

Materials and Corrosion 2012, 63, No. 9999 Application of automated corrosion sensors for monitoring 3

Figure 3. Sub-cycles of N-VDA test

evaluate the robustness of AirCorr O loggers. It should be 3 Results and discussion

mentioned that the N-VDA test, which is currently in develop-
ment by a working group of the VDA (German vehicle producers 3.1 Zinc
association) and the VDEh (German steel producers association)
as VDA 233-102 shall replace the test standard VDA 621-415 [20], Records of CD registered by two loggers with fresh sensors, L1
indeed known to fail in simulating on vehicle exposures [20–23]. and L2, during three consecutive week cycles, C1, C2, and C3, are
The N-VDA test was selected for this study because it includes a plotted in Fig. 4. Logger L2 was removed during the second cycle
variety of technical phases such as a salt spray phase (NaCl and logger L1 in cycle C3.
1 wt%, pH neutral), wet and dry cycles at several temperatures up CD records from initial cycles C1 differed somewhat from
to 50 8C and a freezing phase at 15 8C. It consists of three sub- those from the following cycles. A layer of corrosion products
cycles A, B, and C, with duration of 24 h each, as shown in Fig. 3. formed on the fresh sensor surface. After the initial corrosion
In a week cycle, the sub-cycles are ordered as A–B–A–C–A–B–B product build-up in the first sub-cycles B and A corresponding to
or B–A–C–A–B–B–A. A typical duration of the test is 6 weeks. about 1.5 mm in CD, the corrosion rate dropped.

2.3 Experimental procedure

Since the corrosion logger is independent of any power supply, it

was placed directly in a test chamber. Two steel sensors with the
metal track layer thickness of 250 mm (Fe-250mm) and two zinc
sensors of 25 mm (Zn-25mm) were consequently exposed.
Loggers were placed with the longer sensor side parallel to the
ground and with the protected part above the sensing part. In this
position, water cannot accumulate on the sensing part. The
sensor was oriented at 208 to vertical. In parallel, three zinc
(99.99%) and carbon steel coupons were exposed to the chamber
at the same angle. On completion of the exposure, collected and
stored data were downloaded from the logger to a computer by a
non-contact inductive data reader and evaluated using dedicated
software. The mass loss of zinc and steel coupons was measured
according to ISO 8407 after respective removal of corrosion Figure 4. Record of CD measurements with Zn-25 mm sensors in
products in glycine (250 g/L C2H5NO2) and Clark’s (1 L HCl, N-VDA; L1, L2: Two parallel loggers/sensors; C1, C2, C3: 1st, 2nd, and 3rd
20 g Sb2O3, 50 g SnCl2) water solutions at room temperature. week cycle of the test

www.matcorr.com ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

4 Prosek, Le Bozec and Thierry Materials and Corrosion 2012, 63, No. 9999

Table 1. Changes in CD in sub-cycles of N-VDA measured with Zn-25 mm sensors (in mm)

Logger/cycle Sub-cyclea Full cycle

B1 A1 C A2 B2 B3 A3

L1/C1 0.2 1.4 0.2 0.8 0.1 0.1 0.8 3.6

L2/C1 0.3 1.1 0.1 0.8 0.1 0.1 0.9 3.4
L1/C2 0.1 0.8 0.1 1.0 0.3 0.2 0.8 3.3
L2/C2 0.2 0.7 0.1 – – – – –
L1/C3 0.1 1.0 0.2 – – – – –
Average 0.1b 0.8b 0.1 0.9 0.1 0.1 0.8 3.4
Standard deviation 0.0b 0.1b 0.1 0.1 0.1 0.1 0.1 0.1
a
Index gives an order of the sub-cycle in a week cycle;
b
L1/C1 and L2/C1 are excluded because of the initial formation of corrosion products.

The CD measured in three complete cycles L1/C1, L2/C1, Anyway, it can be concluded that the two measurement
and L1/C2 was 3.6, 3.4, and 3.3 mm, respectively. It gives the techniques gave comparable results.
average CD of 3.4  0.1 mm. The relative standard deviation of Changes in CD per sub-cycle are given in Table 1. It is evident
4% shows a very good reproducibility of the measurement. The that the most corrosive phases were sub-cycles A comprising the
obtained CD can be compared to the total mean zinc CD surface contamination with a solution of 1 wt% sodium chloride
estimated from mass loss of zinc coupons exposed in the test for in a form of salt spray. The average increase in CD recorded in
6 weeks. It was 16.3  2.4 mm, which corresponds to 2.7 mm/ sub-cycles A in the 2nd and 3rd cycle of N-VDA (C2, C3) was
week if a constant rate of the corrosion process is expected from 0.8  0.1 mm. It was significantly lower in sub-cycles B and C, i.e.
cycle 1 to cycle 6. The sensors showed 27% higher CD than 0.1  0.1 mm. Although a part of sub-cycle C is a freezing phase,
coupons. It is not surprising because the ER technique measures the sub-cycles B and C were similar in terms of the corrosivity
in its principle a maximal depth of attack, whereas the coupon towards zinc.
technique assesses an average depth. The maximal depth of The sub-cycles can be analyzed separately in detail as it is
corrosion is important from the practical point of view and shown below for B1 and A1. CD, temperature and relative
often followed as the main parameter defining corrosion humidity records from the first B-sub-cycle, B1, are seen in Fig. 5.
degradation of automotive materials [21, 24, 25]. In reality, the The relative humidity and temperature records are from the 1st
difference between the coupon and ER measurements is probably cycle C1. There was no principal difference between the RH and
even smaller due to the fact that the corrosion rate of zinc usually T-values measured in week cycles C1, C2, and C3. Corrosion of
tends to somewhat decrease in time due to a weak inhibiting zinc initiated within less than an hour after the start of the test.
effect of formed corrosion products and the CD in the initial The initial layer of corrosion products formed by corroding
cycles was thus probably higher than that calculated as 1/6 of the 0.2 mm of zinc at 25 8C and at 70% RH. In two cases, further
total in 6 weeks. Further difference in corrosion properties of bulk activation was observed when the temperature reached 40 8C at
and thin film materials may come from different metallurgy. the same RH.

Figure 5. CD, temperature, and relative humidity measurements in Figure 6. CD, temperature, and relative humidity measurements in
sub-cycle B1 with Zn-25 mm sensors sub-cycle A1 with Zn-25 mm sensors; dashed line show desired RH

ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.matcorr.com

Materials and Corrosion 2012, 63, No. 9999 Application of automated corrosion sensors for monitoring 5

Table 2. Corrosion rate of zinc measured in periods with constant T and

RH (in mm/h); average values for sensors with a developed layer of
corrosion products

Conditions Sub-cycle Order of sub-cycle Average

T [8C] RH [%] 1 2 3

35 100a A 0.123 0.110 0.107 0.113

50 50 A 0.002 0.003 0.001 0.002
50 90 A 0.020 0.015 0.012 0.016
35 95 A 0.013 0.013 0.011 0.012
25 70 B 0.004 0.016 0.004 0.008
50 70 B 0.002 0.001 0.001 0.001
50 90 B 0.010 0.007 0.009 0.009
35 95 B 0.007 0.011 0.007 0.008 Figure 7. Record of CD measurement with Fe-250 mm sensor in N-VDA
50 90 C 0.003 – – 0.003 in modification A–B–A–C–A–B–B (L1); C1, C2: 1st and 2nd week cycle of
35 95 C 0.004 – – 0.004 the test
a
Salt spray phase.

A similar record for sub-cycle A1 is seen in Fig. 6. The steel did not initiate immediately. It took most of the first sub-
corrosion rate increased immediately after the application of cycle of 24 h to initiate the corrosion process, even when it was the
sodium chloride solution by spraying. During the salt spray sub-cycle A with the salt spray phase in L1. Consequently, the first
phase, it kept at 0.11  0.01 mm/h in sub-cycles A1, A2, and A3 on and second cycle C1 and C2 in L1 differed in total CD. It was 51
pre-corroded surfaces. Thus, the process was reproducible. The and 81 mm, respectively. CD measured in the first week in L1 and
corrosion rate only decreased slightly in the following phase when L2 differing in the order of sub-cycles was however identical,
the temperature increased from 35 to 50 8C. It dropped when the 51 mm.
relative humidity got below about 75% (ranging from 78 to 71% The CD measured using the automated sensors can be
for particular cycles and sensors). It was relatively low in the rest compared to that calculated from mass loss of carbon steel
of the sub-cycle. coupons. They were exposed for 1 and for 6 weeks in L2. The
Corrosion rates of zinc were extracted from the changes in mean CD was 32  3 and 254  21 mm, respectively. When the
CD in periods when both temperature and relative humidity were depth obtained in 6 weeks is recalculated to 1 week, it gives
constant. They are given in Table 2. Obviously, there is a tendency 42  4 mm. The maximal depth of corrosion attack was measured
to induce a higher corrosion rate at elevated temperatures and on the coupons exposed for 1 week using a micrometer gauge. It
higher relative humidity. However, variations in the corrosion was found to be 80  30 mm. Obviously, the depth measured
rate measured at identical RH and T in sub-cycles A, B, and C can by the sensors was higher than the mean depth calculated from
be observed. For example, the corrosion rate at 50 8C and at 90% the mass loss and within the maximal depth measured using a
RH was 0.016  0.004, 0.009  0.002 and 0.003 mm/h in the gauge.
respective sub-cycles. This emphasizes the fact that the rate of the CDs recorded in particular sub-cycles in L1 and L2 tests are
corrosion process, taking place on the metal surface under a layer given in Tables 3 and 4. Two principal observations can be made
of corrosion products does not immediately reflect the actual when the data are compared to those obtained for zinc. First,
climatic conditions. In particular, the drying and wetting process there were smaller differences in corrosion rates measured in
can be slow. It may take significantly longer than an hour until particular sub-cycles for steel than for zinc. The average change in
equilibrium is established between the atmosphere and the
corroding system. The actual amount of water, chloride and other
species available at the metal surface that control the surface
chemistry and the corrosion process was besides the actual
climatic and pollution characteristics affected also by the
exposure history.

3.2 Carbon steel

Steel sensors of 250 mm were exposed to test cycles comprising

sub-cycles A–B–A–C–A–B–B (L1) and B–A–C–A–B–B–A (L2).
The aim was to investigate if an order of sub-cycles affects the
total CD of carbon steel. Two week cycles of L1 were performed,
denominated as L1/C1 and L1/C2. A single week cycle was
carried out for L2, L2/C1.
Records of CD measured in L1 and L2 are displayed in Fig. 7 Figure 8. Record of CD measurement with Fe-250 mm sensor in N-VDA
and in Fig. 8. It is interesting to notice that the corrosion of carbon in modification B–A–C–A–B–B–A (L2)

www.matcorr.com ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

6 Prosek, Le Bozec and Thierry Materials and Corrosion 2012, 63, No. 9999

Figure 9. CD, temperature, and relative humidity measurements in Figure 10. CD, temperature, and relative humidity measurements in
sub-cycles A with Fe-250 mm sensors; dashed line show desired RH sub-cycles B with Fe-250 mm sensors; dashed line show desired RH

CD in sub-cycles A, B, and C was 13  5, 8  3 and 4  3 mm. products on steel contain both Fe2þ and Fe3þ species, which
Thus, the most corrosive sub-cycle A caused about 3-fold larger makes the system less dependent on the actual availability of
increase in the CD than the least corrosive sub-cycle C and the oxygen on the surface. Fe3þ species can reduce to Fe2þ replacing
difference to sub-cycle B was of only 40%. For zinc, 6-fold thus oxygen in the depolarization reaction [26].
difference was observed between sub-cycles A and both B and C. Changes in CD and relative humidity and temperature
Second, the corrosivity in particular sub-cycles of the same type records for sub-cycle A are plotted in Fig. 9. Corrosion rates in
varied according to their order. For example, the first, second and periods characterized by constant RH and temperature are given
third sub-cycle B in L1/C1 gave changes in the CD with the in Table 5. Excluded are first A-sub-cycles for L1/C1 and L2/C1
standard deviation of 40%. Both these observations indicate that where lower CDs were measured due to incomplete activation of
the atmospheric corrosion of carbon steel was even less sensitive the surface. The corrosion rate was about 0.4 mm/h in the salt
to actual bulk climatic conditions than that of zinc under spray phase at 35 8C. It increased sharply during a drying phase at
conditions of this study. The strong hysteresis of the corroding 50 8C when the relative humidity dropped under about 83–87%. A
system is expected to be due to slow changes in the conditions at corrosion rate of about 2 mm/h was registered. It decreased when
the metal/corrosion product interface. Both the drying of RH reached 73–76%. Thus, the maximal corrosion rate of steel
corrosion products and transport of activators towards and from was found after the salt spray cycle at the elevated temperature of
the interface are slow processes. In addition, the corrosion 50 8C and the relative humidity between 75 and 85%. At around
80% RH, the surface is covered with a thin and continuous layer
of water solution that enables for efficient transport of oxygen and
Table 3. Changes in CD in sub-cycles of N-VDA in modification L1
measured with Fe-250 mm sensor (in mm)

Cycle Sub-cyclea Table 5. Corrosion rate of carbon steel measured in periods with
constant T and RH (in mm/h); average values for sensors with a
A1 B1 A2 C1 A3 B2 B3 Full cycle developed layers of corrosion products

C1 2 4 10 5 14 8 8 51 Conditions Sub-cycle Order of sub-cycle Average

C2 14 9 14 7 15 12 10 81
T [8C] RH [%] 1 2 3
a
Index gives an order of the sub-cycle in a week cycle.
a
35 100 A 0.3 0.4 0.4 0.4
50 50 A 0.2 0.2 0.3 0.2
Table 4. Changes in CD in sub-cycles of N-VDA in modification L2 50 90 A 0.5 0.6 0.7 0.6
measured with Fe-250 mm sensor (in mm) 35 95 A 0.4 0.5 0.7 0.5
25 70 B 0.2 0.2 0.2 0.2
Cycle Sub-cyclea 50 70 B 0.3 0.3 0.2 0.3
50 90 B 0.3 0.3 0.3 0.3
B1 A1 C A2 B2 B3 A3 Full cycle 35 95 B 0.4 0.5 0.4 0.5
50 90 C 0.4 – – 0.4
C1 1 5 1 13 7 5 19 51 35 95 C 0.3 – – 0.3
a a
Index gives an order of the sub-cycle in a week cycle. Salt spray phase.

ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.matcorr.com

Materials and Corrosion 2012, 63, No. 9999 Application of automated corrosion sensors for monitoring 7

Excluding the salt spray phase, it varied in a rather narrow range

from 0.2 to 0.6 mm/h, whereas a variation from 0.001 to
0.012 mm/h was recorded for zinc. Even more striking difference
between the materials is seen when the corrosivity within the salt
spray phase is compared to the rest of the test. For carbon steel, it
was only slightly elevated and periods with higher actual
corrosion rates could be identified. For zinc, the salt spray phase
constituted clearly a period of extreme changes in CD. In total,
75% of zinc corroded in this phase that took only 13% of the total
test time. The corrosion rate of zinc was from one to two orders
higher in the salt spray phase than in any other part of the test.
When the external conditions changed, the activity of species
participating in the corrosion process at the carbon steel/
corrosion products interface only slowly adjusted, which was
reflected by gradual changes in the corrosion rate. In contrary,
zinc tended to corrode in distinct steps caused by activation of the
corrosion reaction in presence of an elevated amount of chloride
Figure 11. CD, temperature, and relative humidity measurements in ions or when the relative humidity or temperature increased
sub-cycle C with Fe-250 mm sensors; dashed line show desired RH above certain limits. Subsequently, free chloride ions were
removed from the system by binding them in stable corrosion
ionic species [27]. The oxygen diffusion path is longer at higher products such as simonkolleite [28] and zinc got into a passive-
RH and ionic transport is disrupted at lower RH. The corrosion like state characterized by a significant drop in the corrosion rate.
rate in sub-cycle A was the lowest at 50 8C and 50% RH when it In the salt spray phase, the permanent supply of chloride ions led
reached 0.2 mm/h. The highest corrosion rate was obtained at to extreme corrosion rates.
50 8C and at 90% RH. It is in agreement to earlier measurements This observation is in line with previously published results
in a similar accelerated test using the ultrasonic technique [8]. on poor prediction of zinc mass loss or time to red rust
It is interesting to note that the corrosion rate increased appearance for galvanized products by the normal salt spray test,
systematically from the first to the third sub-cycle, see Table 5. It NSST [21]. Although the corrosion stability of zinc coatings is at
indicates that not only the corrosion products formed on the least 20-fold higher than that of carbon steel in real service
surface had no protective ability but the corrosion process was conditions, comparable mass losses for zinc and carbon steel are
accelerated underneath. It can be expected that local acidification often recorded in NSST. This can be understood in view of
and separation of anodic and cathodic locations was more the current results showing the very high zinc corrosion rate in
efficient under a thicker layer of iron corrosion products. the salt spray phase. It must be pointed out that the chloride
No dramatic changes in the corrosion rate can be seen in the concentration in NSST as well as in VDA621-415 test is 5 wt%
sub-cycle B (Fig. 10, Table 5). It decreased when the relative compared to 1 wt% in N-VDA. Any test that should reasonably
humidity dropped from 95 to 70%, then it started to increase mimic real service conditions has to involve shorter periods of salt
when the temperature increased from 25 to 50 8C and decreased contamination or lower salt loads and sufficiently long drying
again at 35 8C. The highest corrosion rate was observed at 95% periods. This is in line with the recent evolution of accelerated
RH and 35 8C. It is interesting to note that the third B-sub-cycle corrosion tests for automotive materials such as the replacement
was systematically less corrosive than the second one. It can be of VDA 621-415 by the N-VDA (or VDA 233-102) [20].
explained by a larger distance from sub-cycle A with the salt spray Although the actual corrosion rate of carbon steel was not
phase. particularly elevated in the salt spray phase, the chloride
During the freezing phase in sub-cycle C, the corrosion rate contamination of the surface affects the corrosion process
was nil, see Fig. 11. It decreased sharply during cooling when the significantly. It is obvious when corrosion rates measured under
temperature got below about 15 8C. The corrosion process re- identical climatic conditions but in sub-cycles with different time
initiated when the relative humidity crossed 75% at 50 8C. distance from the salt spray phase are compared. For example, the
Corrosion of the sensor used in cycle L2 initiated even later, when corrosion rate at 50 8C and at 90% RH in the sub-cycle A and in
the relative humidity increased to 95%. This can be understood in the following sub-cycle B (B2) starting 14 and 38 h after the salt
view of the probabilistic nature of corrosion leading to somewhat spray phase were 0.6 and 0.3 mm/h, respectively. Thus, the
different corrosion behavior of parallel identical samples. chloride contamination clearly accelerated the corrosion process
of carbon steel. As discussed above, carbon steel showed a larger
3.3 Comparison of carbon steel and zinc degradation hysteresis and any equilibrium at the metal/corrosion product
and implications for test development interface was built up relatively slowly.
Zinc reacted more readily to changes in the exposure
Comparison of the data on corrosion rates in different phases of conditions. There were periods when the corrosion rate of zinc
the test with constant conditions for zinc and carbon steel in dropped as low as 0.001 mm/h. It was shown that relatively dry
Table 2 and Table 5 shows that the actual corrosion rate of carbon periods are crucial for formation of stable corrosion products on
steel was less sensitive to conditions in the test chamber. zinc with certain protection ability. Such conditions are naturally

www.matcorr.com ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

8 Prosek, Le Bozec and Thierry Materials and Corrosion 2012, 63, No. 9999

met in service conditions and tests with permanently high Framework Programme, contract number 226539. All members
wetness generally fail to predict zinc performance in field. of the project team are thanked for their contribution. Fréderic
Although further analyses are necessary to complete the study, Ledan is thanked for valuable experimental support.
the available data indicate that the N-VDA test might correctly
simulate real exposure conditions and provide reasonable
prediction of the performance of galvanized materials. 5 References
For both carbon steel and zinc, changes in the corrosion rate
were often detected when the relative humidity dropped below or [1] L. Sjögren, On-line corrosion monitoring of environments
increased above about 75%. It well corresponds to the for electronic equipment, Report 2002:2, Swedish Corrosion
deliquescence point of sodium chloride, which is 75–76% in Institute, Stockholm, Sweden 2002.
the given temperature interval [29–31]. The deliquescence point is [2] ASTM B 808-97, Standard test method for monitoring of
the relative humidity when sodium chloride forms saturated atmospheric corrosion chambers by quartz crystal micro-
solution in equilibrium to humid air. Below this threshold level, balances, ASTM International, West Conshohocken, USA
sodium chloride is solid and thus most chloride ions are removed 2003.
from the system. Diluted solution forms above the deliquescence [3] S. Zakipour, C. Leygraf, Br. Corros. J. 1992, 27, 295.
point with a lower chloride concentration and higher film [4] M. Forslund, C. Leygraf, J. Electrochem. Soc. 1997, 144, 113.
thickness. [5] L. Falat, US patent 5 411 890, (1995).
The automated corrosion logger appears to be an interesting [6] O. Baltzersen, A. Solstad, A. Daaland, J. K. Holberg, NDTnet
device for understanding the effect of technical phases (salt 2005, 10.
contamination, level of temperature or relative humidity, duration [7] M. Ström, G. Ström, G. Strannhage, presented at EURO-
of each phase, variation between temperature and relative CORR 2006, Maastricht, The Netherlands, September 25–28,
humidity, etc.) in accelerated corrosion tests, which may help 2006.
in the development of such tests. [8] G. Rannou, D. Thierry, N. Le Bozec, presented at COR-
ROSION 2010, NACE, San Antonio, USA, March 14–18,
2010, Paper 10170
[9] ASTM G96, Standard Guide for On-Line Monitoring of
4 Conclusions Corrosion in Plant Equipment (Electrical and Electrochemi-
cal Methods), West Conshohocken, USA: ASTM Inter-
An automated corrosion logger allowing for real-time measure- national 2008.
ments of carbon steel and zinc corrosion rate was used in a cyclic [10] ISO 11844-2, Corrosion of metals and alloys – classification
accelerated corrosion test, N-VDA cycle. This test was selected as of low corrosivity of indoor atmospheres – Part 2: Determi-
it offers the advantage of alternating salt spray and wet and dry nation of corrosion attack in indoor atmospheres, Inter-
cycles at temperatures from 15 up to 50 8C. national Organization of Standardization, Genève, Suisse,
From the results, the following conclusions may be 2005.
drawn: [11] Automated corrosion sensors as on-line real time process
control tools (CORRLOG), Co-operative Research Project,
 The ER technique proved to be a very sensitive tool for studying 6th Framework Programme, Contract No. 0182 07, 09/2005-
the corrosion kinetics of the metals. It was possible to follow 02/2008.
changes in the corrosion rate in order of nanometers per hour. [12] Protection of cultural heritage by real-time corrosion monitor-
The obtained results were well reproducible and comparable to ing (MUSECORR), Collaborative Project, 7th Framework
mass loss measured by metal coupons. Programme, Contract No. 2265 39, 06/2009-05/2012.
 The actual corrosion rate was not fully reflecting climatic [13] www.musecorr.eu, accessed 22/03/2012.
conditions in the test chamber. It was affected by the exposure [14] T. Prosek, F. Dubois, M. Kouril, B. Scheffel, Y. Degres,
history due to limited rates of wetting/drying and oxygen and M. Jouannic, M. Taube, M. Dubus, V. Hubert, D. Thierry,
ion transport to the reaction interface under a layer of corrosion presented at 18th ICC, Perth, Australia, November 20–24,
products. The hysteresis was particularly strong for carbon 2011, Paper 436.
steel. Except the freezing phase, the steel corrosion rate varied [15] M. Dubus, M. Kouril, T.-P. Nguyen, T. Prosek, M. Saheb,
J. Tate, Stud. Conserv. 2010, 55, 121.
in a narrow range from 0.2 to 0.6 mm/h.
[16] L. Vaissière, M. Ragazzini, G. Despert, N. Le Bozec,
 For zinc, 75% of the metal corroded in the salt spray phase and
D. Thierry, presented at EUROCORR 2011, Stockholm,
during following drying period while it represented only 13%
Sweden, September 4–8, 2011, Paper 4631.
of the total test time. The corrosion rate varied from 0.001 to [17] N. Le Bozec, D. Thierry, P. Le Calvé, J.-P. Pautasso, presented
0.1 mm/h in particular phases of the cycle. at EUROCORR 2011, Stockholm, Sweden, September 4–8,
 During drying phases, the corrosion rate usually dropped when 2011, Paper 4629.
the relative humidity decreased below about 75%, i.e. close to [18] T. Prosek, M. Kouril, L. R. Hilbert, Y. Degres, V. Blazek,
the deliquescence point of sodium chloride. D. Thierry, M. Ø. Hansen, Corros. Eng. Sci. Tech. 2008, 43,
129.
[19] T. Prosek, D. Thierry, M. Kouril, Y. Degres, presented
Acknowledgements: We gratefully acknowledge the financial at CORROSION 2008, NACE, New Orleans, USA,
support by the European Commission under the Seventh March 16–20, 2008, Paper 08296.

ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.matcorr.com

Materials and Corrosion 2012, 63, No. 9999 Application of automated corrosion sensors for monitoring 9

[20] F. Beier, K.-H. Stellnbberger, S. Geisler, presented at EURO- [25] N. Le Bozec, A. Le Gac, D. Thierry, Mater. Corros. 2012, 63,
CORR 2009, Nice, France, September 6–10, 2009, Paper 408.
8412. [26] M. Stratmann, K. Hoffmann, Corros. Sci. 1989, 29,
[21] N. Le Bozec, N. Blandin, D. Thierry, Mater. Corros. 2008, 59, 1329.
889. [27] M. Stratmann, K. Bohnenkamp, T. Ramchandran, Corrros.
[22] N. Le Bozec, A. Legac, D. Thierry, presented at 4th Inter- Sci. 1987, 27, 905.
national Seminar in the field of Automotive Corrosion, Stock- [28] T. Prosek, D. Thierry, C. Taxén, J. Maixner, Corros. Sci. 2007,
holm, Sweden, May 5–6, 2010. 49, 2676.
[23] B. Rendhal, N. Le Bozec, presented at Eurocorr, Nice, France, [29] T. Prosek, A. Iversen, C. Taxén, D. Thierry, Corrosion 2009,
September 6–10, 2009, Paper 8395. 65, 105.
[24] T. Prosek, N. Larché, M. Vlot, F. Goodwin, D. Thierry, Mater. [30] A. Apelblat, J. Chem. Thermodyn. 1993, 25, 63.
Corros. 2010, 60, 412. [31] L. Greenspan, J. Res. Nat. Bur. Stand. 1977, 81A, 89.

(Received: April 27, 2012) W6655

(Accepted: May 20, 2012)

www.matcorr.com ß 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim