Vous êtes sur la page 1sur 6

Tribology Letters Vol. 11, No.

2, 2001 127

Structural characterization of wear debris produced during friction


between two austenitic stainless steel antagonists
C. Brin, J.-P. Rivière ∗ , J.-P. Eymery and J.-P. Villain
Laboratoire de Métallurgie Physique, UMR 6630 CNRS, Université de Poitiers, Bâtiment SP2MI, Téléport 2, Boulevard Marie et Pierre Curie,
BP 30179, F-86962 Futuroscope-Chasseneuil Cedex, France
E-mail: jpriviere@univ-poitiers.fr

Received 19 December 2000; accepted 6 June 2001

Ball-on-flat friction experiments between two austenitic stainless steel antagonists in sliding contact are carried out under very low
loads and in two liquid environments, namely demineralized water and methanol. The wear debris produced during long-duration tests are
characterized by different techniques such as X-ray diffraction, transmission electron microscopy and Mössbauer spectroscopy. Martensite
is observed as a dominant phase in the wear debris but other phases have been also identified. In the debris produced under demineralized
water, there is significant contribution of a second phase, which is poorly crystallized and which was identified as the hydrous iron oxide
called ferrihydrite. In the debris produced under methanol, there is a remaining amount of austenite that is not transformed into martensite,
and the presence of ferrihydrite has been also detected in very small quantities. The formation of the martensitic debris which occurs from
the beginning of the wear tests supports the fact that under our experimental conditions the main damaging mode is abrasive wear by hard
particles.
KEY WORDS: AISI 304 stainless steel; X-ray diffraction; Mössbauer spectroscopy; wear debris; ferrihydrite

1. Introduction The objective of this paper is to present the results of a


detailed microstructural analysis of the wear debris formed
Austenitic stainless steels are widely used materials be- during those mild friction tests of AISI 304 stainless steel
cause of their very good corrosion resistance in various ag- samples in order to obtain a more complete understanding
gressive environments. However, they exhibit a relatively of their role in wear processes.
low hardness (∼2–3 GPa) and poor tribological properties.
One of the most representative alloys of this category is cer-
tainly AISI 304 stainless steel. Its wear behaviour has been 2. Experimental procedure
extensively studied [1–3] and it is now well established that,
2.1. Friction experiments
whatever the friction conditions, it has a very low wear re-
sistance both in dry sliding and under lubricated conditions. Wear tests were carried out at room temperature using
Severe adhesive and abrasive wear has been observed, as- a conventional pin-on-disc machine. For the experiments,
sociated with significant mass losses. Even after short du- AISI 304 stainless steel discs, 32 mm in diameter and 8 mm
ration tests, Fayeulle et al. [4,5] have observed severe abra- in thickness, rotate in a horizontal plane, while the immobile
sive wear and plastic flow of the material during friction ex- counterpart pin is a cylindrical sample (AISI 304L), 10 mm
periments, while strain-hardening, twinning and martensitic in diameter and 35 mm in length, exhibiting a spherical-
transformation were identified by transmission electron mi- ended extremity of large radius (∼120 mm). The chemical
croscopy. As a result, a hard superficial layer is formed but compositions of both disc and pin are given in table 1. The
its consequences on the wear rate strongly depend on differ- pins and discs were mechanically polished before the tests,
ent factors, such as the nature of both counterparts, applied until a mean roughness of 0.02 µm was reached. Hardness
load, sliding velocity and environmental conditions. measurements performed before testing gave values in the
Most of the previous friction experiments with AISI 304 2.2–2.5 GPa range. The rotational speed selected for these
stainless steels have been carried out under high applied experiments was 60 turns per minute, which corresponds to
loads producing elevated initial Hertzian pressures exceed-
Table 1
ing the elastic limit of the material (∼280 MPa). In recent
Chemical composition (wt%) of disc and pin alloys (304 and 304L stainless
works [6,7], we have investigated the sliding-wear behav- steels, respectively).
iour of AISI 304 stainless steel in a ball-on-flat configura-
C Si Mn S P Ni Cr Fe
tion under Hertzian pressures at least four times lower than
the elastic limit of the alloy. Disc 0.042 0.39 1.16 0.0016 0.021 8.74 18.53 Balance
Pin 0.016 0.65 1.69 <0.003 0.019 10.46 18.54 Balance
∗ To whom correspondence should be addressed.

1023-8883/01/0800-0127$19.50/0  2001 Plenum Publishing Corporation


128 C. Brin et al. / Wear debris between austenitic steel antagonists

a linear velocity of 0.065 m s−1 . The wear debris were pro-


duced during friction tests under applied loads of 4 and 8 N.
The corresponding calculated initial Hertzian pressures, 58
and 75 MPa, respectively, are much lower than the elastic
limit of the starting stainless steel (∼280 MPa). Under our
experimental conditions, it must be noticed that there is a
significant increase in contact area during the wear test due
to the wearing of the spherical pin extremity. Consequently
there is also a considerable decrease in the pressure applied
by the pin onto the disc. For instance, under a 8 N normal
load and after a sliding distance of 0.6 km, the calculated
pressure falls to 0.5 MPa. A special stainless steel vessel
has been designed to perform friction experiments in liq-
uid media such as demineralized water and methanol [6],
and the debris generated during sliding were collected for
structural analyses. The production of the amount of debris
necessary for X-ray diffraction and Mössbauer spectroscopy
experiments has required very long tests (∼2–4 weeks) cor-
responding to sliding distances of the order of 100 km.

2.2. Characterization methods

The wear debris were first analysed by X-ray diffrac-


tion (XRD) using Cu Kα radiation (λ = 0.15406 nm) and
a Siemens D5000 diffractometer equipped with a graphite
monochromator. Mössbauer spectra were recorded at room
temperature in transmission mode using a constant acceler-
ation spectrometer. As a source of γ -rays, the 57 Co in Rh
matrix was used. Isomer shifts are given relative to the cen-
troid of the spectrum of α-Fe at 293 K. The spectra were
fitted to Lorentzian lines by using the least-squares mini-
mization method. Transmission electron microscopy (TEM)
observations were performed at 200 kV in a Jeol 200CX mi-
croscope after depositing the wear debris onto a carbon grid.
The morphology of the wear tracks was also monitored by a
JSM 5300 scanning electron microscope. Figure 1. Scanning electron micrographs showing the worn surface near
the main wear track, after a sliding distance of 3 km under a 4 N load
in water environment. The initial and final extremities of a lateral groove
are indicated by A and B, respectively. The arrow indicates the sliding
3. Experimental results direction.

3.1. Morphology and structure of the worn surfaces formation of new lateral grooves during the tests, as is illus-
trated in figure 1 for a test performed in water environment
The analysis of the worn surfaces [6,7] suggests that, at under a 4 N normal load and after a sliding distance of 3 km.
the very beginning of our tests, the initial wear mechanism In figure 2 we present at a higher magnification the initial
is mainly due to adhesion. Effectively, even if the roughness and final extremities of the A–B groove shown in figure 1.
of both surfaces in contact is very low, there are always sur- This groove exhibits highly deformed lateral regions, indi-
face asperities at which the real contact pressure is very high vidual debris, and at its final extremity a significant agglom-
and then adhesion takes place between them. The relative eration of wear debris. This process of agglomeration could
motion of the two surfaces induces severe plastic deforma- then explain the formation of large grooves by an abrasive
tion and rupture of these deformed junctions, thus producing mechanism. When the tests are performed in methanol envi-
hard wear debris. After this transient period, scanning elec- ronment, the morphology of the wear tracks remains nearly
tron microscopy observations show that the wear tracks are the same, although the friction coefficient is lowered by a
formed of very fine, smooth and parallel grooves [6,7]; they factor of two and the wear rate is also reduced.
are typical of an abrasive wear process of a ductile mater- A typical XRD pattern of the employed stainless steel,
ial caused by the ploughing action of hard particles. During taken before starting the friction tests, is presented in fig-
the experiments, the width of the main wear track progres- ure 3. It only exhibits the diffraction peaks of the γ fcc
sively increases, so it is possible to continuously observe the austenite, which are located at 2θ = 43.6◦, 50.7◦, 74.6◦,
C. Brin et al. / Wear debris between austenitic steel antagonists 129

Figure 4. X-ray diffraction pattern taken from a steel pin after a friction test
performed under a 8 N load in water environment. Note the appearance of
the α  martensite phase.

90.5◦ and 95.9◦, respectively, as expected from the JCPDS-


ICDD powder diffraction database (PDF-2). The struc-
tural characterization of both worn counterpart surfaces has
clearly shown that, whatever the applied load or environ-
ment and since the beginning of the friction tests, austenite is
transformed into martensite in the surface layers. The XRD
pattern of the worn pin surface after a test in water environ-
ment under a 8 N load is shown in figure 4 and illustrates
this behaviour. It is important to emphasize that there are no
additional diffraction peaks that might indicate the presence
of other phases. The pattern in figure 4 was used to calculate
the lattice parameter of the martensite phase. The derived
value, a = 0.2878 nm, is in good agreement with that ob-
tained from neutron scattering experiments (a = 0.2875 nm)
by Herjo et al. [8]. The crystallite size (D) of both γ and α 
phases was also estimated by using the method described
Figure 2. Scanning electron micrographs showing at a higher magnification
in [9]. This method is well suited, since it takes into ac-
the extremities of the A–B groove in figure 1: (a) initial extremity A and
(b) final extremity B. The arrow indicates the sliding direction. count the line broadening caused by the lattice strain. Before
the friction tests, the D value for the austenitic phase along
the [111] direction is measured to be ∼40 nm. The structural
changes brought about by the tests resulted in lowering the
latter D value by a factor of two, for both γ and α  phases.

3.2. X-ray diffraction experiments on wear debris

The XRD pattern of the wear debris formed during test-


ing under demineralized water is presented in figure 5. The
most intense peaks can be unambiguously identified to α 
martensite, but the presence of two additional peaks, which
are broad and weak, can be noticed at 2θ = 35.6◦ and 62.7◦;
the associated distances are d = 0.252 and 148 nm, respec-
tively. The corresponding phase is not well crystallized, and
it is not possible to identify it only on the basis of the latter
XRD pattern. Mössbauer experiments were carried out in
order to identify this second phase.
Figure 3. X-ray diffraction pattern taken from the starting 304 stainless In the case of the debris formed in methanol environ-
steel, before the friction tests. Only the peaks from the γ austenite phase ment under a 8 N load, the XRD pattern presented in fig-
are visible. ure 6 is different, since the contribution of three different
130 C. Brin et al. / Wear debris between austenitic steel antagonists

Figure 5. X-ray diffraction pattern taken from wear debris produced in


water environment under a 4 N normal load. The arrows indicate the peak
positions for ferrihydrite.

Figure 7. TEM micrograph of wear debris produced in water environment


under a 4 N load (upper image) and corresponding electron diffraction pat-
tern (lower image).

produced in a water environment after a long sliding distance


of 100 km under a 4 N load. The diagram is formed of con-
Figure 6. X-ray diffraction pattern taken from wear debris produced in centric rings of weak intensity which are characteristic of
methanol environment under a 4 N normal load. The arrow indicates the the crystalline structure of α  martensite with fine and dis-
position of the main ferrihydrite peak. oriented grains. The diffuse background in the central part
of the diagram is the contribution of the amorphous carbon
phases can be noticed. The first and dominant phase still grid supporting the debris. The bright field image in figure 7
remains α  martensite and the diffraction peaks attributed indicates that most of the debris range from 50 to 500 nm in
to the unidentified phase are here practically not detectable; size. However, some of the observed particles are probably
only the 2θ = 35.6◦ peak slightly emerges from the back- agglomerates formed by clustering of smaller particles.
ground. Moreover, the presence of the (111) diffraction
peak of austenite is visible, superimposed to the α  marten- 3.4. Mössbauer spectroscopy experiments
site (110) peak. These results suggest firstly that the debris
formed under methanol are not completely transformed into Mössbauer spectroscopy can provide valuable informa-
martensite, and secondly that there is a smaller contribution tion on the nature of iron compounds because it is very sen-
of a new phase which is to identify. sitive to their magnetic properties. It is well known that
martensite is ferromagnetic at room temperature and that
3.3. TEM observations austenite is not; as a consequence these compounds exhibit
quite different Mössbauer spectra. The spectra from the de-
The TEM observations of the debris collected either un- bris formed in a water environment under 4 and 8 N loads
der water or under methanol have shown that they are es- were recorded in the −8 to +8 mm/s velocity range, and
sentially formed of martensite since the contribution of the the former is presented in figure 8. In this spectrum, there
other phases could not be detected in the electron diffraction is a major contribution of a ferromagnetic sextuplet with
patterns. In figure 7 we present a TEM micrograph and the average hyperfine field Hhf = 26.9 ± 0.2 T and isomer
corresponding diffraction pattern of the ultimate wear debris shift δ = −0.03 ± 0.02 mm/s. These values are in very
C. Brin et al. / Wear debris between austenitic steel antagonists 131

Figure 8. Transmission Mössbauer spectrum for wear debris produced in Figure 9. Transmission Mössbauer spectrum for wear debris produced in
water environment under a 4 N load. The S and D subspectra are attributed methanol environment under a 8 N load. The S, D and L subspectra are
to martensite and ferrihydrite, respectively. attributed to martensite, ferrihydrite and austenite, respectively.

good agreement with the Mössbauer data corresponding to the relative contributions of the two phases are different, i.e.,
martensite [10]. The second contribution to the experimen- 48% of iron atoms belong to the martensite and 52% to the
tal spectrum is very different since it consists of a doublet ferrihydrite.
with equal components, which is the signature of a para- The Mössbauer spectrum of the debris formed in a
magnetic compound. The values derived for the isomer shift methanol environment under a 8 N load is presented in fig-
and quadrupole splitting are δ = 0.345 ± 0.02 mm/s and ure 9. Its shape differs from that of the two previous spectra
= 0.83 ± 0.01 mm/s, respectively. In order to identify since the components of the central doublet do not have the
the corresponding compound and since the friction exper- same intensity, thus suggesting the possible contribution of a
iments were performed under water during long duration third component with a single line. We have fitted this spec-
tests, we have focussed the investigations towards iron ox- trum by the superimposition of three contributions, namely
ides, hydroxides and hydroxyoxides. We have compared a single central peak, a doublet and a sextuplet. The single-
our data with the Mössbauer parameters of the above com- line component can be ascribed to γ austenite which is para-
pounds [11] and found a very good agreement with the pa- magnetic with an isomer shift δ = −0.05 ± 0.02 mm/s.
rameters of a paramagnetic hydrous iron oxide known as The Mössbauer parameters of the two other components, at-
ferrihydrite and whose chemical formula can be written as tributed to martensite and ferrihydrite, respectively, have the
Fe5 O7 (OH)·4H2 O. This compound is either a mineral or a same values as for the previous spectra. The percentage of
synthetic product, but it often exhibits a poorly crystallized iron atoms in the three different phases are 73% for marten-
structure; its X-ray diffraction pattern is composed of either site, 13.5% for ferrihydrite and 13.5% for austenite. This
two or six broad peaks, depending on the degree of crystal- result is qualitatively in good agreement with the XRD pat-
lization [12,13]. In the case of the poorly ordered forms, two tern in figure 6, suggesting the minor contribution of both
peaks are generally observed with Cu Kα radiation, appear- ferrihydrite and austenite phases. The Mössbauer parame-
ing at 2θ ≈ 35.5◦ ((100) reflection) and 2θ ≈ 62◦ ((115) ters of the different components deduced from the analyses
and (106) reflections) [13], which fairly well corresponds to are listed in table 2.
the positions of the unidentified peaks in our diffraction pat-
terns (see figures 5 and 6). The values of the quadrupole
splitting found in previous studies for ferrihydrite with two 4. Discussion and conclusion
XRD peaks are in the 0.83–0.85 mm/s range [13], and the
isomer shift is 0.35 mm/s, which is in very good agreement The above structural analysis has shown that marten-
with our results. From the ratio of the sextuplet area over site is the dominant phase in the wear debris formed dur-
the total spectrum area, we have estimated the percentage of ing long duration friction tests under demineralized water or
iron atoms involved in the martensitic phase as 61.5 ± 2%, methanol. This result is not surprising, since it is well known
so the remaining Fe atoms (38.5 ± 2%) may be ascribed to that martensitic transformation of 304 stainless steel can be
the ferrihydrite phase. induced by plastic deformation [14]. Under the present ex-
The spectrum taken from the debris produced in a water perimental conditions, the average Hertzian pressure is much
environment under a higher load of 8 N (not shown here) is lower than the elastic limit; however the local stress inten-
similar to that in figure 8, but here the component lines of sities at the level of microasperities are high enough to pro-
the sextuplet are less resolved, thus indicating that the size duce plastic deformation that could result in strain-hardening
of the martensite particles is smaller. It is still possible to and martensitic transformation.
obtain a good fit with a ferromagnetic sextuplet and para- For the friction experiments performed under water, there
magnetic doublet having the same Mössbauer parameters as is still austenite in the surface layer of both antagonists in
those measured from the spectrum in figure 8. In that case, contact, but this phase is rapidly and completely transformed
132 C. Brin et al. / Wear debris between austenitic steel antagonists

Table 2
Mössbauer parameters from the wear debris formed during friction tests performed in water and methanol environments under
4 and 8 N loads. Hhf = magnetic hyperfine field, δ = isomer shift relative to iron metal, = quadrupole splitting, r = relative
area.
Martensite Ferrihydrite Austenite
r Hhf δ r δ r δ
(%) (T) (mm/s) (%) (mm/s) (mm/s) (%) (mm/s)

Debris formed under water


4N 61.5 26.9 −0.03 38.5 0.345 0.83 – –
8N 48 26.9 −0.03 52 0.34 0.83 – –

Debris formed under methanol


8N 73 26.2 – 13.5 0.34 0.83 13.5 −0.08

into martensite during plastic deformation and removal of fully acknowledge Dr. R. Cauvin for many helpful discus-
the debris. In the case of experiments under methanol, it sions.
is clear that the transformation of austenite into martensite
in the debris is not complete, and this result can be corre-
lated with the role of methanol as lubricant. The significant References
reduction of the friction coefficient (factor of two) can ex-
plain the reduction of the intensity of the associated tangen- [1] S. Fayeulle and D. Treheux, Nucl. Instrum. Methods B 19/20 (1987)
tial shear stresses responsible for the debris removal. As a 216.
[2] R. Wei, P.J. Wilbur, W.S. Sampath, D.L. Williamson, Y. Qu and L.
consequence, the amount of debris produced under methanol
Wang, J. Tribol. 112 (1990) 27.
is less abundant and the plastic deformation less severe, thus [3] A. Van Herpen, B. Reynier and C. Phalippou, J. Phys. IV (France) 8
explaining the presence of untransformed austenite in the (1998) 181.
wear debris. Under these conditions, there is a reduction [4] S. Fayeulle, in: Wear of Materials, ASME Proc., ed. K.C. Ludema
in both abrasive effects and wear rate. (New York, 1987) p. 3.
[5] S. Fayeulle, in: Defect and Diffusion Forum, Vol. 57–58, ed. F.H.
The absence of ferrihydrite at the surface of both antago-
Wöhlbier (Aedermannsdorf, 1988) p. 327.
nists strongly suggests that this phase is formed in the debris [6] C. Brin, Thesis, University of Poitiers, France (1999).
only either at the moment of their removal (being promoted [7] J.P. Rivière, C. Brin, J.P. Villain and R. Cauvin, J. Phys. IV (France),
by a local and short temperature increase) or after the debris accepted.
ejection during the prolonged stay in water or methanol. Let [8] S. Herjo, Y. Tomata and M. Ono, in: The 5th Int. Conf. on Resid-
ual Stresses, Vol. 1, eds. T. Ericsson, M. Odén and A. Andersson
us mention that ferrihydrite has been recently detected as a
(Linköping, 1997) p. 46.
corrosion product formed on carbon steel after long expo- [9] P. Goudeau, K.F. Badawi, A. Naudon, M. Jaulin, N. Durand, L.
sure to the open atmosphere in both polar and subtropical Bimbault and V. Branger, J. Phys. IV (France) 6 (1996) 187.
climates [15]. Under the present conditions, there is how- [10] J.P. Eymery, J. Phys. IV (France) 2 (1992) 211.
ever no experimental evidence, from the morphological and [11] G.W. Simmons and H. Leidheiser, in: Applications of Mössbauer
microstructural characteristics of the worn surfaces, which Spectrometry, Vol. 1, ed. R.L. Cohen (New York, 1976) p. 85;
E. Murad and J.H. Johnson, in: Mössbauer Spectroscopy Applied
indicates that this new phase plays a role in the wear be- to Inorganic Chemistry, Vol. 2, ed. G.L. Long (Plenum, New York,
haviour. The principal mechanism by which the degradation 1987) p. 507.
of the two stainless steel surfaces occurs, appears to be of [12] F.V. Chukhov, B.B. Zvyagrin, A.I. Gorshkov, L.P. Yernilova and V.V.
mechanical origin and related to abrasion produced by hard Balaskova, Int. Geology Rev. 16 (1974) 1131.
martensite particles. [13] B. Cho, K. Fujita, K. Oda and H. Ino, Nucl. Instrum. Methods B 76
(1993) 415;
E. Murad and U. Schwertmann, Am. Mineral. 65 (1980) 1044.
[14] N. Bergeon, C. Esnouf, G. Guénin, Y. Robach and L. Porte, J. Mater.
Acknowledgement Sci. Lett. 16 (1997) 1135;
W. Shiao and W. Shunhua, Prog. Natural Sci. 7 (1997) 469.
This research was supported by Electricité de France un- [15] J.F. Marco, M. Gracia, J.R. Gancedo, M.A. Martin-Luengo and G.
der contract No. 780336. The authors would like to grate- Joseph, Corros. Sci. 42 (2000) 753.

Vous aimerez peut-être aussi