Corrosion Science, Vol. 37, No. 9, pp.

1365-1375, 1995
Pergamon Copyright Q 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
001&938X/95 $9.50 + 0.00

0010-938X(95)00036-4

AN ACOUSTIC EMISSION TECHNIQUE FOR MONITORING

PITTING CORROSION OF AUSTENITIC STAINLESS STEELS

H. MAZILLE, R. ROTHEA* and C. TRONEL

Physicochimie Industrielle, Institut National des Sciences Appliquees, 69621 Villeurbanne, France

Abstract-An acoustic emission (AE) technique was used to detect and study the development of pitting
corrosion on AISI 316L austenitic stainless steel in laboratory experiments. Tests were conducted at room
temperature in 3% NaCl solution acidified to pH 2, at the free corrosion potential or with applied anodic
polarization. It appears that AE signals were easily detected during pitting corrosion and a good correlation
was observed between AE activity and pitting rate. The exact nature of AE sources within pits is not
identified clearly but the results demonstrate that an acoustic emission technique can be used to detect and
even monitor the occurrence of such phenomena.

INTRODUCTION

The acoustic emission (AE) phenomenon is the result of a transient elastic wave
propagation generated by a rapid release of energy within a material. This technique is
widely used as a non-destructive evaluation method for testing metallic or reinforced
plastic vessels on-site. Many microscopic deformation or fracture processes have also
been studied with this technique in laboratory experiments throughout the world
during the past two decades. The authors’ recently reported, for example, the
application of AE for continuous monitoring of damage initiation and propagation
due to localized aqueous corrosion. A literature survey underlines that most of these
studies were focused on corrosion modes combining two effects such as anodic
dissolution and mechanical effects, mainly in stress corrosion cracking investigations.
Some published papers also deal with abrasion or erosion corrosion studies. Only a
few attempts have been made to study purely electrochemical corrosion types such as
uniform corrosion24 or pitting corrosion.4’0 In the latter case, the first published
work was written by Parry,’ who claimed to be able to detect pits inside piping systems
by AE. AE was supposed to be generated by secondary effects such as scale cracking or
oxide fracture on corrosion defects when a slow steady hydrostatic pressurization was
applied to each section of tested pipes. In 1976, Rettig and Felsen4 established that the
AE count rate was linear with the rate of hydrogen evolution during the corrosion of
aluminum wire in salt water. They suggested that an explanation for AE would not
only include macroscopic motion of hydrogen bubbles, but also ‘other physical/
chemical mechanims’, as some AE signals were monitored in cases where no bubbles
were detectable even at 10 x magnification. In 1979, Mansfeld and Stocker6 correlated

* Materiaux - Corrosion, CRIT-D, Rhone-Poulenc Industrialisation, 69153 Decines Charpieu, France

(now with Materials and Corrosion Engineering, Rhone-Poulenc Inc., Cranbury Center NJ 08512-7500,
U.S.A.)
Manuscript received 6 January 1995.

1365
1366 H. Mazille et al

qualitatively the rate of pitting corrosion of aluminum alloys in 3% NaCl (by weight-
loss measurements) to the activity (number of events) of AE. Large fluctuations in AE
were observed during tests and attributed to ‘the statistical nature of the pitting
process’ induced by anodic polarization. In a more recent study on anodized
aluminum immersed in 0.05 M NaCl with 20 mg/l Cu2+ , Magaino’ attributed AE to
hydrogen evolution from the growing pits. However, in another paper’ concerning a
similar study on AISI 304 stainless steel, immersed in NaCl solution containing
NaClO, he observed that AE signals were no more associated with gas evolution, but
with ‘instantaneous stress change on a metal surface’. As far as AE generated on
aluminum alloys, the more thorough investigation has been performed by Arora,’ in
various well-controlled experimental conditions. As mentioned by the author, if AE is
to be used for detecting and identifying an active corrosion process such as pitting, it is
absolutely necessary to proceed to careful frequency and waveform analyses of
recorded AE signals.
To our knowledge, the only recent attempt to quantify AE data with pitting
corrosion on stainless steel, has recently been reported by Jones and Friesel.” They
found that AE activity during pitting of 304 stainless steel was significant. As bubbles
formation or evolution was ruled out, they concluded that ‘the exact process
responsible for the AE generation was not delineated, but the rupture of an oxide or
salt cap covering the pits was suggested’. According to them, if such a process occurs,
only 1 - 3% of the pits are involved and only larger pits are acoustically active. That
could be a problem for using AE measurements for pit monitoring.
There is clearly a need to explore this technique further. Our research programme
has been initiated to investigate if there was a quantitative correlation between AE
parameters and pitting corrosion rate on stainless steels; the ultimate goal being to
develop an acoustic emission technique which could detect on-going pitting corrosion
and quantitatively monitor the pitting corrosion rates of various vessels or chemical
equipments in service.

EXPERIMENTAL METHOD
Marrriul and specimen preparation
AISI 3 I6L austenitic stainless steel, the chemical composition of which is given in Table I, wasused for
this study. The material was sliced from an annealed bar and machined to a 65 mm diameter, 4 mm thick
cylindrical plate, in order to use each sample as the bottom part of a cylindrical PTFE corrosion cell. A
slightly conical adaptation, with careful machining, allowed a very good water tightness without gasket and
less risk of developing crevice corrosion during tests, even under anodic polarization.
The exposed surface of the specimens was wet ground up to 1200 grit silicon carbide paper, followed by
polishing with diamond paste to 3 pm finish. After a passivation treatment of 30 min in 20% HN03 at 6o”C,
the specimens were thoroughly rinsed with de-ionized water before dried in a stream of warm air. An
overnight storage in a desiccator prior to test gave more reproducible results as far as the pitting behavior.

Table I

Element C Si Mn Ni Cr MO
(wt%) 0.015 0.62 1.21 11.09 16.68 2.02
Element co cu N S P Fe
(wt%) 0.48 0.1 I 0.018 0.024 0.024 bal
 An acoustic emission technique for monitoring pitting corrosion 1367

Electrochemical environment
All the studies reported here were conducted at room temperature in 3% NaCl solution with the initia1
pH adjusted to 2 with HCI addition. For polarization tests, the electrochemically applied current (or
potential) was controlled with a Sotelem PC STAT. ZI potentiostat, the sample being the working electrode,
a platinum mesh as the counter-electrode and a saturated calomel electrode as a reference. In order to avoid
acquisition of acoustic noise induced by hydrogen evolution from the counter electrode during anodic
polarization of the specimen, the platinum mesh had to be placed in a near-by annex cell connected to the
corrosion cell via a salt bridge.

Acoustic emission monitoring

The two channel AE instrumentation consisted oftransducers, preamplifiers, filters and amplifiers (Briiel
and Kjakr) and a transient recorder (Tektronix 2430A). The transducers are resonant RI5 type from
Physical Acoustic Corp. (piezo-electric disks). They have been selected through preliminary experiments in
other research programs’ because of their high sensitivity in a bandwidth of 10~500 kHz. The two channels
were digitalized at IOMHz sampling rate. The acquisition system was completely computer controlled. The
waveforms were transferred and stored on a hard disk as soon as detected, whereas the electrochemical
parameter (free corrosion potential or resulting potential for a given applied current) was periodically
measured and converted to digital data. The recorded signals were processed after each pitting test
completion to obtain characteristic parameters: amplitude, risetime, energy and frequency components. The
frequency was measured only with the first part (few ps) of the recorded waveforms, which was more
characteristic of the AE source and not of the transducer resonance or cell configuration. Acquisition and
data were processed through adapted softwares from a specially developed program in a previous work on
AE and stress corrosion cracking.”
Details of the sample configuration, electrochemical cell and acoustic emission devices are given in Fig. I.

Surfhceanalysis
After each corrosion test. the pitted surface was observed with optical and electronic microscopy. A
software was developed to take into account the total tested surface using a stereoscope working at low
magnification and equipped with a CCD camera and an image analyser (Matra MS2i - Pericolor 3100). It
allows the various pits (number, size and surface-shape distribution) to be fully characterized.

Complementary analysis
The weight-loss of each sample was accurately measured after each test (k 0.1 mg) and compared to the
amount of dissolved elements into the solution (analysed by chemical spectroscopy) and to the weight-loss
calculated from the total amourit of coulombs supplied by the potentiostat during the anodic polarization
runs.

EXPERIMENTAL RESULTS AND DISCUSSION

In order to increase the reproducibility of pitting nucleation and development on

the sample’s surface and the subsequent AE recording, the following procedure has
been implemented for most of the experiments : first a 15 h stabilization stage at open
circuit potential, then a 7 h polarization stage with a constant anodic current.
Very little AE activity occurs during the first 15 h, most of the signals being of low
amplitude. The AE activity does not initiate immediately after application of the
anodic polarization, but it decreases as soon as the polarization is disconnected.
During the polarization tests, the higher was the current density, the shorter was the
‘initiation’ time before appearance of significant AE signals. In fact, the amount of
applied coulombs necessary to generate AE signals is quite constant whatever the
applied current density (Table 2). AE activity becomes significant only when the size
and/or the number of stable pits is large enough.
The characteristic parameters of the recorded AE signals are the same whatever
the applied current density. Figure 2 illustrates data obtained with an applied current
density of 0.2 mA/cm2. The maximum amplitude distribution ranges between 30 and
 H. Mazille et al.

Pt counter-electrode

4
Transducer 2
Sample Transducer 1

- IEEE bus

- BNC cables

- Digit. cables

Fig. I. Schematic of AE acquisition, processing instrumentation, and electrochemical

control equipment of the corrosion cell.

Table 2. Time necessary for appearance of AE activity

Current density (mA/cm*) 0.1 0.16 0.2 0.27

Initiation time for AE (h) Z3 %2 = 1.5 Zl
Amount of electricity charges (C) Z30.5 ~32.5 230.5 x27.5
 An acoustic emission technique for monitoring pitting corrosion 1369

80 100 120 140 160 180

Amplitude (pV)

2 140
120
100
80
60
40
20
0 1
250 300 350 400 450
Frequency power spectrum (kHz)

450
400
350
300
250
200
150
100
50
0
0,OEO LOES 2.OE5 3.OE-5 4.OE-5 5.OE5 6,OES ‘?.OES 8.OES

Risetime @s)
Fig. 2. Amplitude distribution, frequency power spectrum and risetime distribution for tests
conducted with an anodic current density of 0.2mA/cm2.
1370 H. Mazille et al.

80 pV, i.e. 28-38 dB (cf. 1 PV used at the transducer). Most signals have a peak
amplitude ranging between 30 and 55 ,uV. These data are in good agreement with those
determined by Seah et al.2 during uniform corrosion tests on bright mild steel in dilute
HCl. According to them, the detected AE could be attributed to the release of
hydrogen bubbles during the corrosion process although ‘there was still no conclusive
evidence for this’. From the energy power spectrum calculated with the FFT of the
beginning of each recorded signal, it appears that the frequency range is mainly
between 80 and 350 kHz, the maximum being around 160 kHz. The rise time of most
signals is below 20 ps. As far as pitting and transgranular crack initiation tests on 304
stainless steels, immersed with or without stress in 0.01 NaCl at 0 mV (SCE), Jones and
Friesel” found acoustic emission rise time to be below 5 ps, but in their case, as
previously stated, they ruled out gas evolution as an AE source.
The total number of recorded AE signals for a complete experiment is in very good
correlation with the applied anodic current density as represented in Fig. 3. Moreover,
the total weight-loss measured after test completion is linear with the applied current
density, as well as with the calculated weight-loss from the amount of coulombs
supplied by the potentiostat (Fig. 4). The very good agreement between the two curves
confirms that no selective leaching and no specific element redeposition occur during
pitting tests. It also confirms that the thorough ultrasonic cleaning at the end of each
test was quite effective for eliminating eventually precipitated corrosion products in
the bottom of deep pits.
The metallographic observations of the pitted surface also confirm that the pitting
rate closely depends on the applied current density. But it also indicates that quite a
large number of pits were not fully developed at the end of the test, as evidenced in Fig.
5(a). In such a case, a small central irregular hole appears surrounded by a lot of
secondary smaller pits. The inclusion responsible for the pit initiation disappeared,
but a thin metallic layer is still present and covers most of the hollow occluded cell. A
light pressure with a very fine needle tip (under the microscope) is sufficient to break
down this thin pit cover, and a much larger and regular-shaped (nearly hemispherical)
pit then appears (Fig. 5(b)).
Other authorsI observed the same phenomenon on stainless steels tested in similar
conditions. In the present case, it was necessary to introduce an experimental

0 0.05 0.1 0.15 0.2 0.25 0.3

Anodic current density (mAkm2)

Fig. 3. AE events vs applied anodic current density for a given test duration.
 An acoustic emission technique for monitoring pitting corrosion 1371

60-

3 50.-

ip
7 40.-
s
.s
g 30 --

20 .- +-measured

--X-- calculated
10 -_

0 0.05 0.1 0.15 0.2 0.25 0.3

Anodic current density (mAkm2)
Fig. 4. Measured and calculated weight-loss vs applied anodic current density for a given
test duration.

correction factor to the image processing software, in order to take into account, for
semi-developed pits, the true inner diameter and not the visible one at the central hole.
Such software proved to be very useful, as shown in Fig. 6. It clearly establishes the
direct relationship between the applied anodic current density and the number of fully
or semi-developed pits, for a given test duration. With the use of the correction factor
on the apparent diameter of semi-developed pits, it was also confirmed that the total
pitted area is proportional to the applied current density.

Fig. 5. Micrographs of two main pits before (a) and after (b) mechanical destruction of the
thin remaining metallic layer.
1372 H. Mazille et al

1
20
500
t
0 x 0
0 0.05 0.1 0.15 0.2 025 0.3

Anodic current density (mA/cmZ)

Fig. 6. Total number of pits generated (+) and pitted area ( x ) vs applied anodic current
density for a given test duration.

From Figs 3 and 6, it is possible to establish a direct and quantitative relationship

between AE recorded events during the various tests and the number of pits detected
at the end of the tests. Figure 7 clearly demonstrates that AE technique is a powerful
tool for monitoring the development of pits: the more severe the pitting rate, the
higher is the activity of AE. But the surprisingly good correspondence between the
total number of AE events and the number of detected pits at the end of the tests is
probably a coincidence. It should not be concluded that each pit is responsible for only
one single AE event. The two figures strongly depend on the sensitivity, the accuracy
and throughput rate of both acquisition and processing systems. Furthermore, it is
well knownI that the pitting corrosion proceeds through metastable stages, at least
during the nucleation and even during the early growth steps. Some of the pits will just
die out and repassivate before reaching the stable growth stage. Would they have
sufficient lifetime to be acoustically active?

woor
sc
P 4cKQ --

% 3000 --

2OCKl--

1cOo --

O! I
0 500 loo0 1500 2000 2500 3OC0 3500 4000 4500

Number of pits
Fig. 7. AE events vs number of detected pits
 An acoustic emission technique for monitoring pitting corrosion I313

I I I

n I=0 mAkm2
__ _________________________________________ -_-
tit 0, II n 0 I=O.1 mA/cm2
____________~___~___~~~~~~~~-~~~ ___
I=O.16 mAkm2
-_-

_~_“~__-_~__~__~_~__~__-_~~_-~_- ITIk0.27 mAkm2 -_-

I J

______________________-_-__-

Pit diameter at the end of the test

Fig. 8. Distribution of pit number vs pit size (diameter), for the same duration of anodic
polarization but with various applied current densities.

Figure 8 gives a detailed distribution of pit number vs pit diameter, for the same
duration of anodic polarization but with various current densities between 0.1 and
0.27 mA/cm2. Most of the pits have a diameter smaller than 200 pm after 7 h of tests,
yet some of them reach a 3OWOO pm diameter. In the first small diameter category,
the distribution indicates a slight increase of the pit size with the current density. Such
an increase is not actually significant for diameters ranging above 220 pm, except for
the highest applied current density of 0.27/cm”, where it becomes very significant.
Remembering that AE activity is closely related to the current density and therefore to
the pitting rate (Figs 3 and 7) the previous statements indicate that large pits are not
the only ones to generate acoustic emission. The total number of pits appears more
important than the size of few individual occluded cells.
The use of the image processing software, together with, perhaps, a too simple-
but not unrealistic-hypothesis concerning the hemispherical shape of each individual
pit, allows correlation of AE data and pit size data in Fig. 9. It gives the number of
detected AE events vs the calculated cumulative pit size in terms of total corroded
volume of pit cells, or total area of the inner active surface of the cells. It is worthwhile
to add that the weight-losses measured after the completion of each test as well as the
weight-losses calculated from the charge transfers from the potenstiotat during
polarization tests are in good agreement with the weight-losses calculated from the
volume of corroded material derived from image processing, assuming a density of 7.9
for the stainless steel. For instance, the ratio between two weight-loss measurements
corresponding to the same test is always ranging between 1 and 1.5, whatever the
determination method. Furthermore, and probably more significant, is the obvious
relationship between AE and damage severity due to the pitting process, at least above
a threshold value of corrosion damage. It is not yet possible to conclude on this
apparent threshold for AE generation: is it related to the total number of pits, or to the
size of some of them (or probably to both parameters)? Further specific investigations,
together with the use of a more optimized image processing, are undertaken to shed
1374 H. Mazille et al.

Area (mm2)
100 120 140
4500

4000

3500

3000

2500

2000

1500

1000

500

0 2 4 6 8 10 12
Volume (mm3 )
Fig. 9. Number of detected AE events vs calculated pit size (total corroded volume, or total
area of the inner surface of the cells).

light on those questions, as well as on the exact nature of the AE sources during the
pitting corrosion process of austenitic stainless steels in chloride solutions.

CONCLUSIONS

The present research confirms that there is a good correlation between AE activity
and pitting corrosion damage on austenitic stainless steel in acidified saline solution.
AE signals are of significant amplitude and their specific frequency range enables an
easy detection and processing from the surrounding noises. The results prove that
acoustic emission offers potential advantages over other techniques for basic or
applied research on pitting corrosion. It provides dynamic type information (pitting
rate) and real-time measurements. Moreover, AE is a passive and non-intrusive
technique since sensors are mounted on the surface not exposed to the corrosive
environment, and remote monitoring is possible. For on-site monitoring, detection
and location of on-going pitting corrosion, even at low rates, should be possible by
using suitable transducer arrays.
Further experiments are necessary to achieve sufficient evidence and theoretical
understanding on the exact nature of AE sources within the pitting process. Some new
experimental data will be presented in a later paper.

REFERENCES
I H. Mazille and R. Rothea, Modelling Aqueous Corrosionfrom Individual Pits IO System Management
(ed. K.R. Trethewey and P.R. Roberge), Nato ASI Series - Vol. 266, p. 103. Kluwer Acad. Publish.,
London (1994).
2. K.H.W. Seah, K.B. Lim, C.H. Chew and S.H. Teoh, Corros. Sci. 34, 1707 (1993).
3. R.J. Grossman, IEEE Trans. Sonirs and Ultrasonics SU-31, 25 (1984).
4. T.W. Rettig and M.J. F&en, Corrosion 32, 121 (1976).
 An acoustic emission technique for monitoring pitting corrosion 1375

5. D.L. Parry, 14th Conf: on Non-Destructive Evaluation, San Antonio, U.S.A., p. 388 (1983).
6. F. Mansfeld and P.J. Stocker, Corrosion 35, 541 (1979).
7. S. Magaino, Boshoku Giijutsu 34, 657 (1985).
8. S. Magaino, A. Kawaguchi, A. Hirata and T. Osaka, J. Electrochem. Sot. 134, 2993 (1987).
9. A. Arora, Corrosion 40, 459 (1984).
10. R.H. Jones and M.A. Friesel, Corrosion 48, 751 (1992).
11. A. Proust, P. Fleischmann, H. Mazille and R. Rothea, Corrosion-Deformation-Interaction (ed. T.
Magnin and J.M. Gras), CDI’92, p. 829. Les Editions de Physique, Les Ulis (F) (1993).
12. B. Baroux and D. Gorse, Europ. Symp. on Modifications of Passive Films (ed. P.Marcus, B.Baroux and
M. Keddam), No. 12, p. 300. Institute of Materials for Europ. Fed. of Corrosion (1994).
13. G.T. Burstein, P.C. Pistorius and S.P. Mattin, Corros. Sci. 35, 57 (1993).