A New atmospheric corrosion rate monitor—development and evaluation

A NEW ATMOSPHERIC CORROSION RATE MONITORDEVELOPMENT AND EVALUATION F. MANSFELD: S. L. JEANJAQUETand M. W. KENDIG Rockwell International Science Center, Thousand Oaks, CA 91360, U.S.A. and

D. K. ROE Poland

State University, Portland, OR, U.S.A.

(First received29 April 1985and receiwdfir publicuth 3 October 1985) Abstract-A research program has been carried out in which a new instantaneous atmospheric corrosion rate monitor (ACRM) has been developed and evaluated, and equipment has been constructed which will allow the use of many sensors in an economical way in outdoor exposures. The ACRM was tested in flow chambers in which relative humidity and gaseous and particulate pollutant levck can be controlled. A relatively inexpensive electronics system for control of the ACRM and measurement of atmospheric corrosion rata was designed and built. Calibration of deterioration rates of various metallic and nonmetallic materials with the response of the ACRMs attached to these materials was carried out.

statistical analysis has been carried out in which the factors that affect the response of ACMs and the Since atmospheric corrosion can be assumed to be an repr~ucibility of ACM data have been determined electrochemical process, it is quite lo&al to USC (Mansfeld,. 1981, 1982). These factors included enelectrochemical techniques for monitoring atmosvironmental parameters such as relative humidity, t, pheric corrosion phenomena. These techniques have and SO1 concentration, heat of the metal used to the important advantages that the atmospheric corroassemble an ACM, location in the test cell, aging in sion behavior can be monitored continuously and that outdoor exposure, etc. correlations can be established with atmospheric paraThe new atmospheric corrosion rate monitor meters which are recorded at the same location. The (ACRM) which was developed and tested in this development of corrosion sensors for atmospheric program is based on the Sereda’s concept (Sereda et al., corrosion studies has been relatively slow when com1982) which is the basis for ASTM G84 ‘Standard pared to application of corrosion sensors to other Practice for M~su~ng the Time-of-Wetness on corrosion phenomena The approach of several groups Surfaces Exposed to Wetting Conditions as in that have been using electrochemical sensors in atmosAtmospheric Corrosion Testing.’ The geometry of this pheric corrosion research has been reviewed recently sensor promises a much faster response to rapidly by Mansfeld (1980-1982). The main activities have changing atmospheric conditions than the more bulky been in Canada by Sereda (1960, 1965, 1974) who ACMs. While the ASTM document describes only the developed aconcept to monitor the time-of-wetness t,; approach to measure time-of-wetness, the present in Scandinavia, where Kuccra (1974, 1981) and effort has focused on using the same sensor to Haagenrud (1980) have applied electrochemical sendetermine atmospheric corrosion rates which can be sors to measurements oft, and atmospheric corrosion used to determine the corrosivity of test sites, follow rates; in the U.S.S.R., where Mikhailovskii and co- corrosion episodes such as acid rain, give an indirect workers are carrying out an extensive program in measure of the corrosion behavior of different mateatmospheric corrosion research (1973, 1980); and in rials of construction, etc. The new ACRM is a twothe United States, where Mansfeid and co-workers electrode system with zinc electrodes which can be have carried out systematic atmospheric corrosion attached to the metal of interest. The time-of-wetness research using electrochemical techniques. In these t, would be a by-product of corrosion rate measurestudies, the atmospheric corrosion monitor (ACM), ments in outdoor exposure, since corrosion activity which is an electrochemical sensor, has been used to occurs only during t,. monitor the time-of-wetness (Mansfeld, 1976-1981) Atmospheric corrosion sensors can be designed as and atmospheric corrosion rates in outdoor exposure galvanic cells such as Cu/Zn, Cu/steel, etc., or as oneand laboratory studies. Most recently, a detailed metal, two-electrode cells such as steel/steel Zn/Zn, etc. Roth approaches have been used earlier by Mansfeld and co-workers (Mansfeld, 1976,1977,1979, 1980, 1981). IIn the present program a Zn/Zn sensor ‘Present address Dept. Materials Scicncc, University of Southern California, Los Angetes, CA 90089-0241,U.S.A. was US& to which a 30mV potentiostatic pulse is 1179

1180

F.

MANSFELD et al.

applied. In this case, the corrosion current density i_ can, in principle, be calculated from the Stern-Geary equation (Mansfeld, 1976): B

i,,,, = %

=

24x130 2.3AE x A

=kl

‘Or

where

Corrosion rates are calculated from i, using Faraday’s law for zinc. Application of Equation (I) requires the knowledge of the factor B, which depends on the anodic Tafel slope b, in the case of diffusion control of the cathodic reaction (b, -+ mf. A is the area of one of the two electrodes. For an anodic Tafel slope ofb, = 60 mV, AE = 30 mV and A I= 1 cm’, k N 1.75. Since the parameter B can change with environmental conditions and with time, and because its exact value often is unknown for a given application, a theoretical value often is used for B (or k) in Equation (1). Theoretical values of B range from B = 6.5 mV to 52.1 mV, although B = 13 mV to 26 mV seems to be most common. The electrochemical me~urement carried out here is, therefore, basically a dete~ination of the polarization resistance R,. In this work the ACRN was developed and tested using laboratory facilities in which temperature, r.h., and gaseous and particulate pollutant levels can be controlled. A relatively inexpensive electronics system for data logging and control (ACRMDL) of the sensors was developed and tested and it was determined whether the corrosion sensor response can be used as a proxy for the atmospheric corrosion behavior of a variety of materials such as gafvanixed steel, weathering steel, marble, and latex or oil base house paints.

2. THE NEW ATMOSPHERIC MONITOR

CORROSION

RATE

(ACRM)

The design of the new miniature atmospheric corrosion rate monitor (ACRM) is based on the sensor used by Sereda et al. (1982) which is now the basis of ASTM G84. The advantage is its small size which makes it possible to attach it to a material of interest and monitor the time-of-wetness t, for this material in its micr~I~ate and determine a (indirect) measure of the ins~nta~us corrosion rate of this material. Since the sensor output only reflects the corrosion rate of the sensor material, in this case zinc, conversion factors have to be established to cakulate corrosion rates for other materials. An alternative approach would be to calculate a corrosivity index for a given microclimate based on the ACRM corrosion rate data.

2.1. Design of the ACRM and measurement principle The design of the ACRM is shown schematically in Fig 1. A photograph of the sensor part taken after a corrosion test is shown in Fig. 2. An important

improvement over Sereda’s original design (Sereda et & 1982) is the use of solid zinc instead of zinc-plated copper which avoids the problem of porosity of the zinc layer after extended exposure periods and accelerated corrosion due to the Zn/Cu couple which would also cause errors in the measurement. The thickness of the zinc sensors used in this project is 0.05 mm; it is bonded to an epoxy laninate board of 0.25 mm. The rcnsor ‘fingers’ are 7.8 mm long, 0.6 mm wide and are +ioed 0.2 mm apart (Fis. 1). The active part of the sensor measures 19 mm x 12.3 mm, the remainder is used for ~onnectioll of the current amp~~er, capacitors and leads which connect the sensor to the control unit. The current measuring circuit is not used when the measurement is controlled by the ACRMDL. The measurement principle is different from that of Sereda et al. (1982) who used a galvanic couple &r/Au or Cu/Au to measure 1,, whereas in the present approach the polarization resistance technique is used to measure instantaneous corrosion rates. Initially, a computerized system which had been developed earlier (Kendig et al., 1983, 1985) was used to evaluate the response of the ACRMs to variations of the environmental conditions. Figure I shows the electronic circuit for measurement of the current flow as a result of application of a + 30 mV pulse. This potential pulse is produced by the D/A card in the multiprogrammer of the computerized system which controls the experiment and is described in more detail elsewhere (Kendig et al., 1983, 1985; Mansfeld, 1985). A schematic is shown in Fig. 3. The potential pulse is applied between the INPUT and COMMON terminals of the sensor as shown in Fig. 1. The current I is determined from the output Y of the amplifier as I = V/R,, where R, is the measuring resistor. The polarity of the applied 30 mV emf is changed at the end of the 2 min pulse. Sequential measurements of the currents on as many as eight sensors are made during each pulse using relay csrds in the multiprogrammer with the output from i!le relay cards being measured by an A/D. The current-time curves are stored on cassettes for later analysis. 2.2. Design and test of the data (ACRMDL)

logging system

The ACRMDL is used for control of the ACRMs and data collection. It incorporates a small computer (Radio Shack TRSSO Model PC-3) programmed in Basic language which has sufficient memory to log4608 measurements from six sensors over a 4-day period. A printer/plotter which is part of the ACRMDL is used to provide hard copies of the results as log R, vs time in tabular and graphical form. Results may also be transferred to cassette tape or to another computer via an RS-232C interface. The specifications and operational characteristics of the ACRMDL are described elsewhere (Mansfeld. 1985). The experimental data have been transferred from the ACRMDL to a VAX 1l/780 computer for

1181

New atmospheric corrosion rate monitor

-"s

+vs

Fig. 1. Design of ACRM and electronic circuit.

storage and further processing (analysis, display) of the data. This data transfer is also necessary since only a

limited amount of data can be stored on the ACRMDL. It also prevents accidental loss of data. In order to check the operational procedure for the use of the ACRMDL, a test series was performed in which r.h. was kept at 95% for 7 h followed by r.h. = 65 % overnight. After 1 h at r.h. = 95 % the base plate of the test cell to which the sensor was attached was cooled 3°C vs room temperature except on the first day where it was 4°C. Cooling was maintained for 5 h. Two hours after the start of the test SO1 was introduced at 0.5 ppm for 1 h, followed by 1 ppm for 1 h and 0.5 ppm for 1 h. The SO1 was then turned off. This cycle was carried out for 5 days. Figure 4 gives an example for four sensors with different pretreatments for the first and third day. The inverse polarization resistance l/Rpis plotted which is directly proportional to the corroston rate of the ACRM [Equation (l)]. This type of plot shows the changes of corrosion rates during periods of accekrated corrosion more clearly than log Riplot, while the corrosion behavior during the periods of low corrosion activity cannot be recognixed too well.

3. RESPONSE TO VARIATIONS IN ENVIRONMENTAL CONDITIONS

3.1. Eflect of pretreutment pollutants

and exposure

to single

on ACRM response

Research at this laboratory and by others has shown (Mansfeld, 1982) that the chemical nature of the corrosion product layer has a great influence on the critical r.h.-level at which condensation occurs. The time-of-wetness is therefore strongly affected by the chemistry of the corrosion product layer and one can assume that the corrosion rate and its changes with time depend also on this chemistry. It was therefore necessary to evaluate the effects of different pretreatments on sensor response and stability. The sensors were pretreated with 1 mM NaCI, water or were untreated. CaC12, deionized Pretreatment following cleaning in ethanol consisted of application of a known volume of deionized water, 1 mM NaCl or 1 mM CaCls to which a small amount of alcohol was added to improve wettability. One set of sensors had no pretreatment after cleaning. The liquids placed on the surface of the sensors were atlowed to dry out and the sensors were kept overnight at 45 y/. r.h.

F. MANSFELDet al.

.

‘~.L-S’&_i&‘i~~

~__I*.

Fig. 2. Photograph of sensor part of ACRM after corrosion test.

They were then exposed to a r.h-cycle of 65 x-80 % -95 %-80x-65 % for 1 h each. All experiments were carried out at room temperature (21 f l”C), except when the sensors were cooled (see Section 4). Figure 5 shows the response of eight ACRMs to r.h. after pretreatment with CM&. A pronounced effect of r.h. is seen with fairly similar response of the sensors. The R,,-values are fairly large since only a very small current flows during exposure of a dry surface to r.h. in the absence of pollutants; similar results were observed for the NaCl pretreatment. For Hz0 pretreatment and without a pretreatment, the current tlow was very low and often below the detection limit of the system. This set of data was taken, therefore, more as a qualitative

indication of the necessity of an effective, reproducible pretreatment. Table 1 explains the test sequence for a test series in which the effect of sensor pretreatment and nature of pollutant were evaluated. For each subset of this series, eight ACRMs were pretreated according to the four possibilities Ia-d and then subjected to the sequence under II in Table 1. Two sensors (e.g. Nos. 1 and 2) were coated with carbonaceous particulate matter (CPM) to an extent which was equivalent to 0.5~ exposure in an urban environment. These two sensors and two additional (e.g. Nos. 3 and 4) were then exposed for 2 h at r.h. = 90% and Rp-values were determined continuously. As shown in Rg. 6a, the SOz

1183

New atmospheric ixrrosion rate monitor

ADSl7JH

* CW*y;R MAO TAPE

c

A/D

c

‘,

::%

+

R?ii AMP ADS52

c

D/A LOQIC PULSE

--f---

-I-

Fig. 3. Block

diagramof measurement system.

0.8

9

.

100

9

1.6

q

;

ii

7

slo

0.8

0.4

0

0

2

4

6 1 moutsl

8

10

12

Fig. 5. Response of eight ACRMs to r.h. after pretreatment with 1 mM C&l2 (numbers refer to sensor number).

tbl

Fig.4. Response of four ACRMs with different prctreatments

to

changes in conditions.

environmental

Ieve was increased to 1 ppm for 1 h and then changed stepwiseto 0.1.0.5 and 1.0ppm for 1 h each (step II.2). The SO, was then turned off and after another hour at r.h. = 90%, the r.h. was reduced to 65% overnight. Figure 6a shows the response of these four ACRMs

from steps XI.1to II.3 in Table 1.The next morning two additional sensors (e.g.Nos. 5 and 6) were added to the test cell and the cycle repeated for NO1 (steps 11.4and 11.5)(see Fig. 6b).On the last day the pollutant was 0,. The design of this test serieswas such that sensors Nos. l-4 were exposed sequentially to SO1, NOr and O,, sensors Nos. 5 and 6 to NO1 and Or, while sensors Nos. 7 and 8 experienced only exposure to OS. A comparison of the corrosion behavior of sensors Nos. 1 and 2 with sensors Nos. 3 and 4 should allow an

F. MANSFELD et

1183 Table 1. Test schedule 1. Effect of pretreatment dist.

a.

HI0

b. NaCl (1 mM) c. CaC& (1 mM) d. None Cycle r.h. = 65-80-95x

and back, 1 h each

II. Effect of ~1Iutants 1. Sensors Nos. 1,2 with pretreatment a, + CPM Sensors Nos. 3.4 with pretreatment a, r.h. = W%, 2 h 2. [SO,] = hi + low + stepwise to hi, 1 h each 3. r.h. = 90% for I h, followed by r.h. = 65% overnight 4. Add sensors Nos. 5, 6, r.h. = 90% for 1 h 5. Repeat 2. and 3. for NO2 6. Add sensors Nos. 7, 8, r.h. = 90% for 1 h 7. Repeat 2. and 3. for 0, Same for pretreatments b-d (sensors Nos. 9-32) Particulate matter is applied only in Step 1

estimate of the effect of CPM on the corrosion behavior. SO;: has a strong effect on the corrosion rate of the zinc sensor as can be seen in Fig. 6a, where the

c4clrn3RH

l

oi.

polarization resistance decreased almost immedtntely when the SO2 level was increased to I ppm. This effect is not due to increased conductivity of the surface electrolyte (as could be argued). as can be seen from the almost instantaneous increase of R, when the SOI level is decreased to 0.1 ppm. NO2 also affects corrosion rates as can be seen from the changes of R, following the changes of the NO2 level. Ox, on the other hand, seems to have a passivating effect and no clear changes of R, with ozone level can be detected (Fig. 6-z).Similar behavior was observed for the NaCl pretreatment. For the Hz0 pretreatment the current flow was very low and often under the detection limit of the system. It has to be considered that for R, = 10’ ohm, the current I equals AEJR, = 30 mV’ lob9 ohm = 30nA. AI such low current levels, electrical noise problems affect the accuracy of the measurement severely. Integration of the R;‘-time curves for the time when a given pollutant was present gives the corrosion loss (in s ohms- ‘f shown in Table 2. Since the parameter B [Equation (I)] is not known for the present experimental conditions, it is preferable to use the integrated corrosion loss as a qualitative measure of the corrosion effects as shown in Table 2. Coating with

sew-22%!!

80%

1.0

9

8 0”

z 0.5 _ t;i

7

11”

i llML,*31

8

(b)

0 0

1

2

3

4

5

6

7

II

I

L-

TIME IHRl

6 0.5

f I

i

a

Fig. 6. Effect of pollutant nature and concentration on ACRM

L

I”.

‘5%

response (CaCl, pretrea~m~1). (a) SOr, (b) NOI, (cf OX.

New atmospheric Table 2. Integrated

corrosion corrosion

1185

rate monitor loss

(IO” s ohm- ‘)

ACRM No. PT CaC12 C&12 Cac12

Pollutant SO2 NO2 0,

Total

I

2

3

4

5

6

7

0.27 0.06 0.10

0.81 0.24 0.37

2.32 0.60 0.63

0.67 0.20

0.09

i.4t

3.55

0.87

0.38 0.18 --0.56

0.08

0.43

4.89 0.63 0.45 --5.97

0.08

0.09

0.72

7.36

6.59

NaCI

so2

0.30

NaCi NaCl

NO2 03

0.53 0.69 3.08 4.59 0.40 OS0 2.56 2.43 ---Y-v-

Total

I.25

1.91

13.00

13.61

8

3.90 3.34 2.47 2.27 2.58

3.38

6.37

3.38

5.61

2.58

PT: pretreatment.

CPM lowers the corrosion loss for CaCfl and for NaCf pretreatments (sensors Nos. 1 and 2 compared with Nos. 3 and 4). A comparison of sensors Nos. 7 and 8 with Nos. 5 and 6 and with Nos. 3 and 4 shows that exposure to SO1 produces a larger corrosion loss than exposure to Not, while O3 exposure induces the lowest corrosion damage. Pretreatment with NaCf leads to higher corrosion rates than CaCft for all three pollutants. A comparison of sensors Nos. 3 and 4 with Nos. 5 and 6 shows no large effect of prior exposure to SOz on corrosion losses during exposure to NO1. Similarly, sensors Nos. 3-8 show very similar corrosion losses during exposure to Oj, especially for the NaCf pretreatment. This lack of an effect of prior exposure to a different pollutant is somewhat surprising, since it would be expected that the resulting changes of surface chemistry also change t, and corrosion rates. However, the short exposure periods which had to be used in the present tests might have precluded significant changes of surface chemistry.

3.2. E$ect oj exposure to pollutants, simulated dew cycles and U.V.light Eight sensors of different history were exposed to SOz, NO1 and 0~ with cooling and warming of the sensors to simulate dew with exposure to U.V.fight during the drying cycles. The eight sensors were first exposed to r.h. = 90% for 2 h in the presence of SO2 + NO2 + OS at I ppm each. The sensors were then cooled for 2 h, which fed to conden~tion, followed by 2 h at room temperature in the presence of U.V.fight. The pollutant flow was then stopped. After the first day the ACRMs were exposed overnight at r.h. = 90”/, while after the second day they were kept at r.h. = 45%. Before the first test alf eight sensors were put through the r.h. cycle. Sensors Nos. 1, 3 and 5 for which experimental results are shown in Figs 7a-c were pretreated with CaCfz and CPM. Sensor NO. I had been used in the experiments shown in Figs 3 and 4, while sensor No. 3 was used in theexperiment of Fig. 4. Sensor No, 3 was a new sensor. Cooling of the sensors leads to an almost jns~nt~~us drop of I$, and an increase of corrosion rates [Equation (I) (Fig. 7)]. R, increases slightly

during further cooling and decreases again during warming under U.V.fight. During the warm cycle, the r.h. increases causing an increase in the flux of SO1 to the surface (lower exit value), which is consistent with a higher corrosion rate. A value of R, = 1 x l(r ohmcm2 corresponds to a corrosion rate of 79 pm y - ’ for B = 26 mV [Equation (i)]. A comparison of the corrosion rate data obtained by integration of the RP’ -time curves shows that previous exposure did not change the response to a given environment very much as can be seen from a comparison of the &-values for sensors Nos. I and 3 with those for No. 5 (Fig. 7). The absence of CPM reduced the response of a sensor.

4. CORRELATION D~ERIORATION

OF SENSOR RESPONSE WITH OF DIFFERENT

MATERXA~

An experiment was performed to determine whether the response of the new zinc ACRM can be used as a proxy for the deterioration rate of other materials. An attempt was made to answer this question by determining the corrosion behavior of weathering steel, galvanized steel, marble, latex house paint and oil base house paint (on stainless steel) by the weight loss method. The corrosion behavior of these materials in outdoor exposure had been determined earlier in the St. Louis Study (Mansfefd, 1978,198O).An ACRM was attached to the surface of one of the triplicate materials, and the ACRM output was recorded continuously with the ACRMDL. The experiment was conducted in a special test ceil which allowed the interaction of the pollutants and cooling of the exposed samples which were coated with a polymer coating on their back and attached vertically to the test cell wall. After the 31-day test, the weight loss was determined using specific solutions and taking into aazount the weight loss of a blank. The weathering steel had developed a very uniform layer of corrosion products. The exposure test consisted of a coofjng-warmjng cycle of I.5 h coofing followed by 4.5 h at a higher temperature as shown in Fig. 8 for a 26-h period. When

1186

F. MANSFELD

TIME

8 I^

7

2 3

6

et nt.

IHOURS)

FIRST

DAY

DARK.

NO POLLUTANTS

c 2

5

4

3 0

1

2

3

4 TIME

THIRD

TIME

6

7

8

DAY

(HOURS)

THIRD

TIME

5

(HOURSI

DAY

(HOURS)

Fig. 7. PoWant concentration and polarization rcshnce fu~tion of time and environment conditions for three msors 3 successive days.

R, exposed

as a on

IIS?

New atmospheric corrosion rate monitor

the temperature was reduced, the SO2 concentration, as measured at the outlet of the test cell, dropped due to dissolution of the SOs in the surface electrolyte. Also shown in Fig. 8 is the r.h. as calculated from the temperature of the test cell and the dew point of the air coming out of the test cell. The ACRM response (Fig. 8) follows closely this cycle of environmental conditions. In the first cycle R, changes by a factor of two between the ‘cool’ and the ‘warm’ condition of the ACRM attached to one of the specimens. Figure 9 shows an 8&h cycle in the early and middle part of the test period, respectively, for the six ACRMs. The cyclic response of the ACRMs is clearly seen in both figures. Between 70 and 150 h exposure time there is a gradual decrease of R, corresponding to a slow increase of corrosion rates, while between 310 and 380 h the high and low values in a given cycle do not change much with time. It becomes apparent that sensors Nos. 1,2,5 and 6 show very similar values of R,, while sensors Nos. 3 and 4 have higher values which are more or less the same for these two sensors. These differences can be traced back to the different locations of the ACRMs in the test cell. Sensors Nos. 1 and 2 were located on the west wall of the cell, while sensors Nos. 5 and 6 were on the opposite east wall in an equivalent position. The sensors Nos. 3 and 4 were located next to each other on the south wall. Figure 10 gives a schematic of the experiment arrangement for test coupons and ACRMs. Since the air flow came out of the tubes which were parallel to the east and west walls, it is possible that the pollutant and wet air flow were different for these two walls as compared to the south wall. Figure 11 shows the time dependence of the polarization resistance R, for all six sensors. The SO2

4.01 1 t 312 316

*

I 320

1

I 324

I

concentration during the dry and wet periods of each cycle and the corres&ding maxima and minima of r.h. appear in Fig. 12. After 31 days r.h. was reduced to 45 % and the SO2 flow was stopped. ACRM data were recorded in this condition over the weekend. A slow trend to somewhat higher corrosion rates with increasing exposure time was accompanied by a decrease of the difference between high and low values of R, during a cool-warm cycle. The gap in the data in the beginning of the test is due to a problem with the computer program; the gap around 400 h was caused by a power failure which affected the back-up disk in the VAX computer to which the data had been transferred. The gradual increase of corrosion rates with exposure time is reflected in Fig. 13, in which the average inverse polarization resistance and the calculated reduction in thickness Ad for a given time element (the horizontal bar through the data of sensors No. 1) are plotted. For sensors Nos. 1,2,5 and 6 corrosion rates increase by about a factor of two to three, while for sensors Nos. 3 and 4 this increase is about a factor of six. However, the corrosion rate at the end of the test is still much less for these two sensors than for the other four sensors for the reasons outlined above. Only the weight loss data for the galvanized steel and the weathering steel were sufficiently accurate to warrant the calculation of a conversion factor CF. Due to the relatively short exposure time, the weight loss for the paint and the marble was very low. Table 3 compares the weight loss Am and ACRM data normalized to the exposed area of test coupons or ACRMs. For galvanized steel the average CF = 9.94. 10-3mg*R~s-1 agrees very well with the

I 328

L

I 332

I

I 336

I

TIME (hours)-

Fig. 8. Environmental changes and response of an ACRM.

F.

1188

MANSFELD

et ol

z 0 .E

520

d

a” 0

4.80

s

4.001 .I 310 TIME

320

330

340

350

TIME

BiOURSI

2

0 .E

d 5.60

$ = -a 0

5.20

5

4.80

360

fc)

6.00

s

370

IHOURS

(al

_

360

5.20

!t P :

4.80

2

80

so

100

110

120

TIME

130

340

150

Ihour,)

TIME

Id)

lb) Fig.

9.

Two

80

h test cycles for six ACRMs.

calculated value of 8.83’ 10e3 mg*Q*s-’ obtained from the Stearn-Geary equation [Equation (l)] for zinc with b, = 60 mV according to: Am(mgcm-2~d-‘)

163.4 = R. P

With R, = r/81.5 = 32864 ohm .cm2 one obtains: Am = 2.323, 10-2mgcm-*~d-’

or0.720mgcm-*

and 0.72 wG=~

IHOURSI

=O.l3gmg~ohm*s-‘. A CF, calculated in the as iron equation for Stearn-Geary 7.48. lo- 3 mg -ohm. s- ’ does nat apply in this case, since it compares the corroskm behavior of iron with that of a zinc sensor. It has a&, to be recognized that the corrosion rate of weathtring steel decreases with time according to r co* = &’ where b = 0.5. The experimenti CF is therefore timedependent considering that in the case of zinc b 2 I.

= 8.83. 10m3 mg.ohm,s-’ 5. SUMMARY

AND CXWCLUSIONS

CF,,,=“~~~=9.94.,0-3mg~ohm~s-i. For wenthcring

steel the corrosion

factor CF,,,

A new instantaneous atmaqpheric corrosion rak monitor (ACRM) has been dewloped and tested. The

1189

New atmospheric corrosion rate monitor

1

1

GS tr

1 u

GS

1 1

tfm

OS +s2 t/c

1

1

tr

v

GAS INLET

M

GAS FtOWS

GAS OUTLET

GS: WS: M: LP: OP:

ws

VERTICALLY

AND TOWARDS

GALV. STEEL WEATHERING STEEL MARBLE LATEX BASE HOUSE PAINT OIL EASE HOUSE PAINT +

9

Fig. 10. Schematic of ACRM and test coupon location.

f E d .c

7 -

SENSOR #l

6.

!f P5oz

“:ii 0

0

200

400 tfHOURS)

600

400

600

800

twlOURS)

800

SC”..?**,,

8

1 SENSOR 14 P E c 0 E

7 *

8

!f a % -I

4

0 t(HOURS)

400 twIOURSI

Fig. 11. IE 20:5-H

200

600

800

0

200

400

response of the ACRM to single pollutants and combinations of S02, O3 and NO2 within the range of 0. I - 1 ppm and simulated dew cycfes has been characterized. A data logging system-the ACRMDL-has been designed and tested with the ACRM. The ACRMDL is battery operated and can control up to six ACRMs simultaneously. The use of log converters for the recording of the current transient allows accurate measurements over four decades. A number of experiments have been performed to determine the effects of sensor pretreatment. In experiments in which the ACRMs were exposed to single pollutants without formation of liquid layers for short time periods, pretreatment with NaCl produced higher corrosion rates than CaCI, and increased the reproducibility of the effect of 0,. The ACRMs showed almost instantaneous changes of the polarization resistance when r.h. and the pollutant level, especiahy of SC&, were changed. The sensors are so sensitive in their response to environmental conditions that a less than optimum placement of some sensors in the test cel1 during an exposure test became quite obvious. The response of the ACRMs to SOz, O3 and NO, has been studied as a function of concentration between Oand 1 ppm. A very pronounced ekt of SO2 and NO2 concentrations was observed especially when a liquid electrolyte film was present on the ACRM surface. The fact that this effect is reversible shows that the dissolved pollutant not only changes the electrolyte conductivity, but produces as its main effect a change in corrosivity.

800

600

tlHOURSI *,

ttHoUl?SI

Fig- 11. Tie

dependence of R, for six ACRMs during

entire test period. 1.0'

I

1

I

DRY

SO2

0.8 -

\

J

0

200

400

600

800

TIME lhoursl

I 200 TIME

I

t

400

600

800

(hours1

Fig. 12. SO, concentration at test cell outlet during’wet’and 'dry' periods.Also shown are the corresponding maxima and minima values of r.h. The drop of the SO,-concentration is due to a change to a new tank of SO*.

1191

New atmospheric corrosion rate monitor

SENSOR NO. :

01 0

INTEGRATION TIME

:

150

,-

I 100

I

I

400

I

600

t(h)

Fig 13. Average inverse polarization resistancefor a given time period and calculated corrosion rate as a function of exposure time for the six ACRMs. Table 3. Comparison of weight loss and ACRM data for 31-day exposure test Weight loss ACRM Average (mgcm-*) (sR_t. cm-‘) (sn-‘.cm-r) Galvanized steel Weathering steel Oil base paint Latex base paint Marble

0.81 11.25 I%; (0:002)

71.3 90.5 78.1 86.1

CF

(mg.R.s-‘) 9.94 x 10-J 1.38x 10-l

81.Sf8.5

CF = conversion factor.

From the results of the exposure test conversion factors have been calculated which can be used to convert ACRM data into corrosion losses of galvanized steel. It seems desirable to obtain conversion factors also in exposure to natural environments of different corrosivity in order to ascertain the validity of the conversion factors obtained in a laboratory test. The ACRM in its present form consisting of solid zinc is a very sensitive sensor for the corrosivity of the atmosphere and its changes and for the time-ofwetness. Tbe ACRMDL. provides a very convenient way to control the sensor operation and collect the corrosion data. Since it is battery operated, it can be used unattended for long time periods. Data transfer to a larger computer allows analysis and dispiay of the ACRM data in a more detailed manner.

REFERENCES

Haagenrud S. E. (1980) Investigation of the atmospheric corrosivity by means of an electrochemicaltechnique. Ext.

Abstr. 154th Electrochemical Society Meeting, Vol. 78-2, pp. 327-328. Kendig M., Mansfeld F. and Jeanjaquet S. (1983) Computerized corrosion monito~ng using the current response of multiple sensors, 164th Electrochem. Sot. Meeting, Washington, DC, October 1983,paper No. 74. Kendig M, Jeanjaquet S. and Mansfeld F. (1985)Computer controlled acquisition of corrosion rate data from multiple sensors. In Prof. Symp. Cornpurer A&i Acquisition and AnuIysis of Corrosion Bnrq The Electrochemical Society (1985). Kucera V. and GuBman J. (1981)Practical experiencewith an electrochemical technique for atmospheric corrosion ~nito~n~ ASlM STP 72?. 238-255, Kucera V. and Mattson E. (1974)Ekctrochemical technique for determination of the instantaneous rate of atmospheric corrosion. ASTM STP 558,239-260. Mansfeld F. (1976)Tire polarization resi&mee technique for measuring corrosion currents. Ads. Cm. Sci Tech. 6, i 63. Mansfeld F. (19~)R~uI~ ofthiiy monthsexposurestudy in St Louis, MO. CORROSION/78 Paper No. 88; Proe. 7th Int. Comg.Metoll&Corrosion,Rio de Janeiro, Brazil,Paper No. 193,October 1978. Mansfeki F. (1979) Atmospheric corrosion rates, time-ofwetness and relative humidity, Birrlmrofi und Rorrosion g&38-42.

1192

F.

MANSELD

Man&Id F. (198ad) Regional air pollution study, effects of airborne sulfur pollutants on materials. EPA-600/6-8&007, U.S. Environmental Protection Agency. Research Triangle Park, NC 27711, January 1980. Mansfeld F. (198Ob)Electrochemical techniques for monitoring of atmospheric corrosion phenomena. Proc. Srh European Symp. Corrosion Inhibition, Ferrara, Italy, pp. 191-216. Mansfeld F. (I 98&z)Proc. 5th European Symp. on Corrosion Inhibition, Ferrara, Italy, pp. 191,~216,September 1980. Mansfeld F. (198&d) Electrochemical methods for atmospheric corrosion studies. From. Symp. on Atmospheric Corrosion. The Electrochemical Society, Hollywood, Florida, October 1980, pp. 139-160. Mansfeld F. (1981) Evaluation of electrochemicaI techniques for modelling and monitoring of atmospheric corrosion phenomena. Final Report to National Science Foundation, Grant No. DMR-7923965, May 1981. Mansfeld F. (1981) Evaluation of electrochemical techniques for monitoring of atmospheric corrosion phenomena. ASTM STP 727. 215-237.

Mansfeld F. (1982) New approaches to atmospheric corrosion research using electrochemical techniques. In Corrosion Processes (edited by Parkins R. N.) Appl. Sci. Pub]., England, pp. l-76. Mansfeld F. (1985) Development and evaluation of an instantaneous atmospheric corrosion rate monitor. Final Report. Contract No. 86-02-3741. Rockwell International

et at.

Science Center, March 1985. Mansfeld F. and Kenkel J. V. (1976) Electrochemical monttoring of atmospheric corrosion phenomena. Corr. Sci. 16, 111-122. Mansfeld F. and Kenkel J. V. (1977) Electrochemical measurements of the time-of-wetness and atmospheric corrosion rates. Corrosion 33, 13. Mansfeld F. and Tsai S. (1980) Laborafory studies of atmosphericcorrosion-I. Weight lossand electrochemical measurements. Corr. Sci. 20, 853-872. Mansfeld F., Tsai S., Jeanjaquet S., Meyer M. E., Fertig K. and Ogden C. (1982) Reproducibility of electrochemical me~urements of atmospheric corrosion phenomena. ASTM STP 767, p 309338. M~hailovskii Yu. N., Shubakhina L. A. and Hong Wan8 7. (1973a) Proc. Metals 9, 135. Mikhaifovskii Yu. N.. Klark G. B., Shuvakhina L. A.. Agafonov V. V. and Zhuraveleva N. 1. (1973b) Pro!. Merds 9, 240. Mikhailovskii Yu. N., Strekalov P. V. and Agafonov V. V (1980) Prof. Metals 16, 308. Sereda P. 1. (1960) Atmospheric factors affecting the corro. sion of steel. Iruf. Engng Chem. 52, 157-160. Sereda P. 1. (1974) Weather factors affecting corrosion of metals. ASTM STP 558, 7-22. Sereda P. J., Croll S. G. and Slade H. F. (1982)Measurement of the time-of-wetness by moisture sensors and their calibration. ASTM STP 676, 267-284.