Heinrich J.G., Aldinger F. (Eds.) Ceramic Materials and Components for Engines
Подождите немного. Документ загружается.
ACKNOWLEDGEMENT
This work has been performed within the Inorganic
Interfacial Engineering Centre, supported by the
Swedish National Board for Industrial and Technical
Development (NUTEK) and the following industrial
partners: Erasteel Kloster AB, Ericsson Cables
AB,
Hoganas AB, Kanthal
AB,
OFCON Optical Fiber
Consultants AB, Sandvik AB, Seco
Tools
AB
and
Uniroc AB.
REFERENCES
T. Ekstrom and M. Nygren: “Sialon ceramics”,
J. Am Ceram. SOC.
75 (1992) 259-276.
T.
Andersson and
D.J.
Rowcliffe, “Indentation
thermal shock test for ceramics”, J. Am. Ceram.
G.R. Anstis,
P.
Chantikul, B.R. Lawn, and D.B.
Marshall: ”A critical evaluation of indentation
techniques for measuring fracture toughness:
I,
Direct crack measurements”, J. Am. Ceram.
K.E. Johansson,
T.
Palm and P.E. Werner: “An
automatic microdensitometer for X-ray
diffraction photographs”,
J.
Phys. E. Sci.
Instrum.,
13 (1980) 1289-1291.
P.E. Werner: ”A Fortran program for least-
squares refinement of crystal structure cell
dimensions”, Arkiv fdr kemi”,
3 1 (1969)
5
13-
516.
T.
Ekstrom, P.O. Kall, M. Nygren, and P.O.
Olsson: ”Dense single phase p-sialon ceramics
by glass encapsulated hot isostatic pressing”, J.
Mat. Sci.
24 (1989) 1853-1861.
C.M. Hwang and T.-Y. Tien: “Microstructural
development in silicon nitride ceramics”, Mater.
Sci. Forum
47 (1989) 84-109.
H. Mandal,
D.P.
Thompson and T. Ekstrom:
“Reversible
a-P
sialon transformation
in
heat
treated sialon ceramics”, J. Eur. Ceram. SOC.
12
T.
Ekstrom: “Effect of composition, phase
content and microstructure on the performance
of yttrium Si-Al-0-N Ceramics”, Mat. Sci.
Eng.,
A109 (1989) 341-349.
SOC.,
79 (1996) 1509-1514.
SOC.,
64 (1981) 533-538.
(1993) 421-429.
152
CORROSION
OF
NONOXIDE SILICON-BASED CERAMICS
IN A GAS TURBINE ENVIRONMENT
S
Sic
SN
H. Klemm*, Chr. Schubert", Chr. Taut**, A. Schulz***, G,
Wetting****
3.18 430
430 2.8 1.5~10-~
3.53 840 720 8.1 1.7~10-5
"Fraunhofer-Institute for Ceramic Technologies and Sintered Materials,
IKTS Dresden, Winterbergstr. 28, D-01277 Dresden, FRG
**Siemens AG,
KWU,
Wiesenstr. 35,45473 Mulheim, FRG
***Institut fut Thermische Stromungsmaschinen, Uni Karlsruhe,
Kaiserstr. 12,76128 Karlsruhe, FRG
****Ceramics for Industry, CFI GmbH
&
Co KG, Oeslauerstr. 35,96472 Rodental, FRG
SNMo
KY
ABSTRACT
3.69 830
710
8.0 1.6~10-~
3.39
660
530
6.2
~.OXIO-~
In the present study the corrosion behavior of
nonoxide silicon-based ceramic materials was investi-
gated. Hot gas tests were conducted on silicon nitride
and silicon carbide materials in an atmosphere similar to
that in a gas turbine.
While some materials displayed a high degree of
microstructural stability, all materials suffered surface
degradation during the rig test. The oxidation surface
layer of mainly Si02, which is essential for the oxida-
tion protection of nonoxide materials because it induces
a passive, diffusion-controlled oxidation mechanism,
was found to be degraded by evaporation processes in-
volving volatile silicon hydroxides. The partial pressure
of steam was found to be the most important factor gov-
erning these processes.
INTRODUCTION
One of the main requirements necessary for ceramic
materials to be applied confidently in gas turbines is the
long-term stability of all relevant properties at elevated
temperatures. Recently, nonoxide ceramic materials
based on Si3N4 and Sic, which are featured by their
superior mechanical, chemical and thermophysical
properties, have been developed
[
1-41. Additionally,
significant progress has been made in improving the
long-term stability of these materials at temperatures up
to 1500°C for application-relevant times of more than
10
000
h under laboratory conditions [5,6]. The main
feature of these materials, and the key factor in
stabilizing the mechanical properties, is their
microstructural stability, achieved by minimization of
oxidation-enhanced diffusion processes in the bulk of
the material during long-term service at elevated
temperatures.
Nevertheless, employment of the results of these
laboratory tests for gas turbine applications, for
example, is of limited use due to the severe
environmental conditions in a gas turbine combustor.
The high pressure and flow rate water vapore
and
other
corrosive components (Ca2+, V5+) of the hot gas were
found to have a great influence on the oxidation and
corrosion behavior of the ceramic materials [7,8].
The main problem preventing silicon-based
nonoxide ceramic materials from being applied in such
environments was found to be the formation and
evaporation of silicon hydroxides (Si(OH)4) described
by Opila and coworkers, who investigated the corrosion
behavior of CVD Sic in humid environments. These
processes were observed to be enhanced in burner rig
tests with severe environmental conditions, such as
higher gas and vapor pressure and higher flow rate
In the present study nonoxide ceramic materials
based on Si3N4 and Sic were tested in a real hot gas
environment. The hot gas tests were carried out under
various environmental conditions (temperature and
water vapor pressure).
19,101.
EXPERIMENTAL, PROCEDURES
The studies were conducted on a solid-state-sintered
Sic
(S
Sic) with B and C as sintering additives, a
liquid-phase-sintered (hot pressed) Si3N4
(SN),
a
Si3N4-MoSi2 (SNMo) composite with Yb203 as the sin-
tering additive and a Lu203-containing silicon nitride
material (KY) produced by Kyocera (Japan)
SN
281.
Relevant properties of these materials are summarized
in Table 1.
Table 1
:
Mechanical properties of materials investiga-
ted:
oRT
and
oI4:
4-point bending strength at
room temperature and 1400°C; KIc: SENB
F0.15mm;
E:
200 MPa, 1400"C, 100 h.
153
The corrosion resistance of the materials was
investigated using bending bars
in
a high-temperature,
high-pressure burner rig test apparatus developed by the
Institute for Thermal Turbomachinery, ITS, University
Karlsruhe. Details about the test facility are described in
a former report
[
111. The ceramic materials underwent
three burner rig tests for 100 h by changing temperature
and water vapor pressure in the test facility. The test
conditions are summarized in Table 2.
K/mg/
Am
I
cm2h
Table 3:
Weight
loss
and evaporation rate of the mate-
rials tested
in
the high-pressure burner rig.
Mat.
I
1. Test
I
2. Test
I
3. Test
1
Am
K/mg/
Am
K
/
mg/
cmZh cmZh
Table
2:
Test conditions
I
I
1.Test
I
2. Test
I
3. Test
I
Temperature
/
"C
Pressure
/
bar
1400
1300 1400
5
5
5
Flow speed
/
m/s
Part. steam pressure
After the tests, the weight loss of each bending bar
was determined. Changes in the phase composition at
the surface and the bulk region below the oxidation
layer were investigated by
XRD.
To assess the damage
resulting from the high-pressure burner rig tests, the
room-temperature bending strengths of the specimens
were obtained and compared with the strengths of the
as-fabricated samples. Information about the
microstructural alterations was obtained through
observation of the microstructures (polished and CF,
plasma-etched cross sections) and the surfaces after
corrosion in the SEM.
50
50
50
0.9 0.9
0.4
RESULTS
The results of the burner rig tests are summarized
in
Table 3. All materials exhibited a weight loss after the
tests; this was found to
be
dependent on the test param-
eters used.
From a comparison of the data, it can be seen that
the Si3N4 materials behave alike. The surface layer,
consisting of silica and the disilicates usually formed
from the sintering additives during oxidation of silicon-
based ceramic materials, was found to be degraded by
the corrosive environment in the burner rig. The mea-
sured weight loss was due to the evaporation of silicon
hydroxides, mainly Si(OH),, and the R2Si20,
(R
=
Yb
or
Lu)
formed during oxidation of the Si3N4 materials,
independent of the sintering additive.
As
a consequence
of the formation and evaporation of Si(OH),, the protec-
tive SiO, layer was destroyed during the test, leaving a
very rough layer consisting of only disilicate grains
which were partially spalled off due to the low stability
of this surface layer, as shown in Fig. 1.
From an application point of view, the materials'
surface degradation is quite high. The weight
loss
of
about
10
%
observed for the
S
Sic material at 1400°C
and a partial steam pressure of
0.9
bar led to a material
loss
of about
90
pm in 100 h under these conditions.
After a time of about 10
OOO
h
or
higher (relevant times
for application in energy production), an unacceptably
high material loss of about
9
mm would be observed.
0.08
KY
0.09
Fig.
1:
Surface layer Si N material with Yb203 as
the sintering additive after high-pressure
burner rig test No. 1 (100 h, 1400°C,
0.9
bar
partial steam pressure).
34
With respect to the microstructural stability of the
bulk material, the results of the burner rig tests con-
firmed those of oxidation tests performed and reported
previously [12,13]. During the rig test the microstruc-
ture changed due to oxidation processes that were de-
pendent on the diffusion of oxygen into the upper region
of the bulk and reaction of the oxygen with Si3N4
solved
in
the grain boundary phase. As in oxidation tests
performed in previous studies, the microstructures of the
Si3N4 materials and the Si3N4-MoSi, composites were
found to
be
different after the test. In the Si3N4 materi-
als, typical oxidation damage, i.e., surface and grain
boundary phase inhomogeneities and pores up to the
middle of the materials, was observed. This was the
consequence of the reaction of the diffusing oxygen in
the grain boundaries of the Si3N, materials, resulting
in
the formation of SiO, and leading finally to a grain
boundary phase supersaturated with silica in the surface
region of the Si3N4 material. The relaxation of this
silica-rich grain boundary phase was considered to be
the reason for the changed microstructure
in
the Si3N,
material. The Si3N4-MoSi2 behaved
in
a different man-
ner. Similar to the SI~N, materials, some of the oxygen
diffused through the grain boundaries and triple junc-
tions and penetrated into the upper region of the bulk
material to react with the Si3N4 solved
in
the oxynitride
glass. Instead of the SiO, found
in
the Si3N4 materials,
crystalline Si20N2 was found to be the oxidation prod-
uct in the composite material. The relaxation made nec-
essary by the oxidation processes
in
the grain boundary
154
phase
in
the upper bulk region (formation of a silica-
rich grain boundary phase) was achieved by the im-
mediate crystallization of Si20N,. As a consequence,
the grain boundary composition remained nearly con-
stant and the microstructure of the Si3N4-MoSi2 com-
posite was stabilized after the burner rig test.
The
S
Sic material was found to have a lower rate of
oxygen diffusion. In this case, all of the oxygen reacted
at the interface between the oxidation layer and the
bulk, which consequently was not significantly changed
or
damaged.
Due to the microstructural degradation
in
the bulk,
the residual strength of the Si,N, materials was found to
be much more degraded in comparison to the
Si3N4-MoSi2 composite and the
S
SIC material. The
results of the bending tests after the burner rig test are
summarized
in
Table
4.
Mat.
Table
4:
Residual strength of the ceramic materials
after rig exposure; each value represents the
average of at least
4
specimens.
1.
Test
2.
Test
3.
Test
OTest
OT'Oo
OTest
OT/Oo
OTest
OT/Oo
MPa
%
MPa
%
MPa
%
S
Sic
SN
SNMo
KY
350
81
390 90 450
100
400
48
490 58 520 62
660 80 680 82
700
84
320
48
430 65 420
64
As already mentioned, the weight loss observed for
the
S
Sic material was the consequence of the forma-
tion and evaporation of silicon hydroxides from the
sur-
face
of
the ceramic material. As opposed to the surface
layer found after oxidation
in
air with some crystobalite
crystallites
in
a flat oxidation layer of amorphous silica
(Fig.
2A),
the surface layer of the
S
Sic after the rig test
was found to be rough, indicative of the evaporation of
the silicon hydroxides during the test (Fig.
2B).
The
weight loss obtained was dependent on the test condi-
tions. Both temperature (comparison of Tests
1
and
2)
and water vapor pressure (Tests 1 and
3)
influence the
degradation of the materials. As indicated in the litera-
ture, the water vapor pressure was found to have the
highest influence on the evaporation processes at the
surface of the ceramic materials. When the partial steam
pressure was reduced from
0.9
to
0.4
bar, the weight
loss was reduced by three
or
four times for all materials
studied.
Finally the data obtained
in
this study
for
the
S
Sic
material are compared with the data of Robinson and
Opila [9,10]. The materials investigated
in
the studies
performed by Robinson and Opila, solid-state-sintered
Sic with
B
and C and CVD Sic, are not the same as
those investigated here; however, they can be used for
comparison due to their similar oxidation behavior, i.e.,
the formation of an oxidation surface layer of pure SO2.
Regarding the test conditions, however, several differ-
ences should be considered. For that reason, only a
rough comparison of the data is possible. A summary of
1.
Test
Temp./"C 1400
the conditions
of
the high-pressure burner rig tests is
provided
in
Table
5.
2.
Test Ref.
1400
1400
Fig.
2:
Comparison of the oxidation surface of the
S
Sic material after oxidation in air (A) and
corrosion
in
burner rig test
No.
1
(B).
In Table
6
a comparison of the weight loss rates
measured at 1400°C
in
both studies are given. Ad-
ditional weight loss rates calculated by the model pre-
sented by Opila are used for comparison. Assuming
nearly similar test conditions, calculated weight loss
rates of Si(OH), for this study can be obtained by using
the expression proposed by Opila:
pressurehar
fuel
airtfuel
flow speet
/
mls
with
J
as the weight
loss
rate
in
mg/cm2h,
v
the linear
gas velocity, ptotal the total pressure and PH20 the partial
vapor pressure.
5
5
6.3
nat. gas nat. gas
jet fuel
2.0
2.0
1.1
50 50
20
vapor press.
/
bar
0.9 0.4 0.6*
155
Table
6:
Comparison of measured Ky and calculated
&
weight loss rates for Sic of this study
with data published by Robinson [9]
and
Opila
[lo]
at 1400°C.
0.062
In principle, all data are of the same order of magni-
tude. The calculated values were found to be lower than
the measured data in all cases, indicating the very com-
plex character of the corrosion processes, e.g., the pres-
ence of other silicon hydroxides than %(OH),,
as
pro-
posed by Opila. As a consequence
of
the addition
of
steam, which produced the highest partial vapor pres-
sure, the highest weight loss rate was observed in Test
1
of this study. The weight
loss
rates from Test
3
and
from tests in the literature in which a similar partial
steam pressure was used seem to be more comparable.
While the calculated data were found to be similar, the
higher weight loss rate
of
Test 3 in this study may be the
consequence of a higher influence of the flow speed in
the burner rig.
CONCLUSIONS
Silicon-based nonoxide ceramic materials were stud-
ied in a corrosive atmosphere similar to that in a gas
turbine at high temperatures. Under these conditions,
the microstructural stability
of
the bulk material can be
stabilized by e.g. Si3N4-MoSi2 composites by a mecha-
nism similar to the mechanisms found during oxidation
of these materials in air.
Unlike oxidation in air, the tests in the burner rig
caused all materials to suffer surface degradation. The
weight loss measured was the consequence of the for-
mation and evaporation of silicon hydroxides, mainly
Si(OH)4.. The oxidation surface layer
of
mainly SiO,,
which
is
essential for the oxidation protection of
nonoxide materials because it induces the passive,
diffusion-controlled oxidation mechanism,
was
found to
be degraded. As a result of this, the material showed a
higher rate of oxidation. The main factor influencing the
evaporation processes seems to be the partial vapor
pressure. In spite of the different burner rig set-ups and
test conditions, the data obtained in this study were
similar
to
the results for corrosion
of
CVD Sic reported
in the literature. However, additional studies are re-
quired for a quantitative interpretation
of
the influence
of test parameters such as pressure, partial vapor pres-
sure and flow speed.
ACKNOWLEDGMENTS
We acknowledge the interesting discussions with E.
Opila,
R.
Robinson and J. Smialek, NASA Glenn Re-
search Center, Cleveland. The studies were carried out
on behalf of Siemens KWU Mulheim.
REFERENCES
G. Pezzotti, "Si,N,/SiC-Platelet Compo- site
without Sintering Aids: A Candidate for Gas
Turbine Engines", J. Am. Ceram. SOC. 76 [5]
M.N. Menon et al, "Creep and Stress Rupture
Behavior of an Advanced Silicon Nitride",
J.
Am. Ceram. SOC. 77 [5] 1217-41 (1994).
K.
Watanabe et al, "Development of Silicon
Nitride Radial Turbine Rotors", pp. 1009- 10 16 in
Proc. 4th Int. Symp. Ceram. Mater.
&
Engines,
Ed. by R. Carlson, T. Johanson and T. Kahlman,
Elsevier London
(
199 1).
C. J. Gasdaska, "Tensile Creep in an in Situ
Reinforced Silicon Nitride", J. Am. Ceram. SOC.
H. Klemm, M. Herrmann, Chr. Schubert, "High
Temperature Oxidation and Corrosion of
Silicon-Based Nonoxide Ceramics", J. of
Engineering for Gas Turbines and Power, 122
[
11
H. Klemm, Chr. Schubert, "Long-term stability
of
non-oxide ceramics in
an
oxidative
environment at temperatures above 14OO0C", 24
Annual Cocoa Beach Conference and
Exposition, 23.
-
28.01. 2000, Ceram. Eng.
&
Sci. Proc., 21 (3) (2000).
E.J. Opila, R.E.
Hann,
"Paralinear Oxidation of
CVD Sic in Water Vapor", J. Am. Ceram. SOC.,
E.J. Opila,
D.S.
Fox, N.S. Jacobson, "Mass
Spectrometric Identification of Si-0-H
from the Reaction of Silica with Water apor at
Atmospheric pressure", J. Am. Ceram. SOC., 18
R.C. Robinson, J.L. Smialek, "Sic Recession
Caused by SiO, Scale Volatility under
Combustion Conditions:
I,
Experimental Results
and Empirical Model," J. Am. Ceram. SOC., 82
E.J. Opila, J.L. Smialek, R.C. Robinson, D.S.
Fox, N.S. Jacobson, "Sic Recession Caused by
SiO, Scale Volatility under Combustion
Conditions:
11,
Thermodynamics and
Gaseous-Difhsion Model," J. Am. Ceram. SOC.,
D. Filsinger,
A.
Schulz,
S.
Wittig, C. Taut,
H.
Klemm,
G.
Wotting, "Model Combustor
to
Assess the Oxidation Behavior of Ceramic
Materials under Real Engine Conditions," ASME
Turbo Expo '99, Indianapolis, USA 1999,
H.
Klemm, K. Tangermann, Chr. Schubert, W.
Hermel, "Influence of Molybdenum Silicide
Additions on High-Temperature Oxidation
Resistance of Silicon Nitride Materials", J. Am.
Ceram. SOC. 79 [9] 2429-35 (1996).
H.
Klemm, Chr. Schubert, "Si3N4-.MoSi2
Composite with Superior Long-Term Oxidation
Resistance at 1500"C", submitted to J. Am.
Ceram. SOC.
1313-20 (1993).
77 191 2408-18 (1994).
13-18 (2000).
80
[l]
197-205 (1997).
(ptSpecies
[4] 1009- 12
(
1997).
[7] 18 17-25 (1 999).
82 [7] 1826-34 (1 999).
99-GT-349.
156
MECHANICAL PROPERTIES AND WEAR BEHAVIOUR OF
DIFFERENTLY MACHINED SILICON NITRIDE
AND SILICON CARBIDE CERAMIC SURFACES
T.
Hollstein*,
W.
Pfeiffer,
R.
Zeller
Fraunhofer-Institut fir Werkstoffmechanik (IWM)
Wohlerstr.
11,
D-79108
Freiburg, Germany
ABSTRACT
Machining and wear of ceramics lead to specific
surface topographies, and to damage, micro-plastic
deformation, and residual stresses in the surface layers.
Their effects on sliding wear under boundary friction
conditions and rolling wear without lubrication are
investigated. The analysis of these tribological systems
show, that their characterisation and assessment
-
based
on the main failure and wear mechanisms
-
allow in
most cases for a straight-forward optimisation of the
service behaviour. The life time of components operat-
ing with sliding
or
rolling contact under mixed friction
conditions is increased at most by all those parameters,
which generate compressive residual stresses and en-
hance micro-hydrodynamic conditions.
A
consequent
design of the topography and an increased pressure on
the lubricant may be most effective.
INTRODUCTION
Machining and friction influence ceramics in
bear-
ings predominantly in surface-near regions. The load
capacity and the wear behaviour of ceramic roller
or
sliding bearings depend on the machining parameters
and the surface treatments applied during fabrication.
The assessment of the tribological situation and the
optimisation of component performance need the iden-
tification of the main wear effects and the improvement
of the related near-surface characteristics of the compo-
nents. These near-surface characteristics are mainly
microstructure, topography, surface strength, and ma-
chining induced damage and residual stresses.
The effects of the topography on the sliding and
rolling wear behaviour
of
ceramics under mixed
or
fully lubricated conditions have been studied during the
last years [l-31. Less attention has been paid for dam-
age and residual stresses
-
which may
be
summarised
as surface integrity
-
and their effect on the wear
be-
haviour. It is well known that machining introduces
both, damage and residual stresses, in the near surface
layers of ceramics. Machining induced damage reduces
the strength, while residual stresses, which are usually
compressive, leads to an improved strength. Both
strength behaviour can be observed depending on
which effect dominates [4-61.
A
very pronounced effect of near surface damage
and residual stresses should be observable in contact
loading conditions due to rolling
or
sliding, because the
maximum load stresses occur near the surface. Of spe-
cial interest are systems like water-lubricated face-seals
or
ceramic roller bearings designed to operate without
lubrication. Theoretical and experimental investigations
show that the failure behaviour of these bearings is
controlled by peak stresses near the surface [3, 71.
As
a
consequence, one should expect that the load bearing
capacity of a ceramic component can be increased by
introducing compressive residual stresses in the near
surface region. It has been shown that
-
besides ma-
chining
-
this can be achieved by a shot-peening proc-
ess
[8].
Thus, the aim of the investigations was to under-
stand better the wear behaviour of differently machined
ceramics by applying surface sensitive characterisation
methods like advanced X-ray diffraction methods and
ball on plate tests.
EXPERIMENTAL PROCEDURE
MATERIALS AND MACHINING
The materials investigated where a commercial sin-
tered silicon carbide (EKasicBD, Elektroschmelzwerke
Kempten, ESK) and
a
commercial silicon nitride
(GPSN, SN-N3208, Ceramics For Industry, CFI). The
most important material data are given in Table 1.
SN-N3208 EKasicBD
Material parameter
Char. tensile strength
oo
900
MPa 350 MPa
Weibull modulus 18
10
Young’s modulus 310GPa 400GPa
Fracture toughness KI, 4.7 MPadm 3.2 MPadm
(pre-cracked bend test)
Vickers Hardness
(HV10)
15.2 GPa
>
24 GPa
Tab.
1:
Material data
of
the investigated silicon nitride and silicon
carbide.
157
Samples with cylindrical geometry were manufac-
tured by grinding and lapping for the wear tests from
the near-net-shaped sintered workpieces,
see
Figure
1.
For face sliding wear tests with SIC specimens the
relevant plane surfaces were finished by different lap-
ping procedures (F500- and F220-B4C-abrasives), ul-
trasonic-assisted lapping (F70-Sic-abrasives) and pol-
ishing (diamond abrasives, 0.25 pm and
2-4 pm grain
sizes). The surfaces for the Si3N4 specimens were fin-
ished by a laser-beam shaping method. The circumfer-
ence of the silicon nitride rings were finished using
conventional grinding (D64-diamand wheel), grinding
+
partial polishing, and a laser assisted turning process
(LAM)
. The machining of the specimens was per-
formed by the Fraunhofer-Institut fur Produktionstech-
nologie
(IPT),
Aachen.
E
E
0
t
J.
10
mm
Fig.
I
:
Specimen geometry and loading situation used for the sliding
wear tests (left) and the rolling wear tests (right).
NEAR-SURFACE STRENGTH
The near-surface strength of the machined speci-
mens was determined using a ball-on-plate test. During
this test the specimens are placed on a flat stiff area and
loaded by an
A1203
ball with a diameter of
10
mm.
The
load
is
increased stepwise until a typical cone-crack
appears, which follows the maximum tensile stresses
occurring at the surface just at the border of the circular
contact zone. Because of the statistical behaviour of
ceramics, the load at fracture varies from test to test
within a certain scatterband. In this paper, the fracture
load represents a probability of
50%.
For more details
of the ball-on-plate test
see
[7,8].
WEAR TESTS
Face sliding wear tests were performed under
boundary friction conditions using a contact load
of
2
MPa, sliding speeds
of
0.3
m/s
and water
as
a lubri-
cant. The sample geometry and a sketch of the experi-
mental arrangement are shown in Figure 1. More details
of the experimental procedure are given in
[9].
Rolling wear tests were performed in a so-called
>>Amslerc< experimental set-up. Contact loads in the
range of 1-3 GPa and slip ratios in the range of 1-3
%
were used. No lubricant was used. The sample geome-
try and a sketch of the experimental arrangement is
shown in Figure
1.
More details of the experimental
procedure
are
given in
[9].
X-RAY DIFFRACTION ANALYSIS
Residual stresses and micro-plastic deformations
due to machining and wear processes were determined
by X-ray diffraction techniques. The most important
measurement parameters are given in Table
2.
ceramic
lattice plane y-range
Si3N4
(41
11-P
-45"
I
w
I
+45"
Sic
{
21 3)-6ha
-69.6"
5
y
SO"
ceramic
Radiation penetration depth
Si3N4
CrKa zeff
=
8
pm
Sic FeKa
0.6
pm
c
zeff
c
18
pm
Tab.
2:
Measurement parameters for residual stress and micro-strain
evaluations
Due to the curved geometry of the rolling wear
samples only the average values of near-surface resid-
ual stress components could be determined using the
conventional sin$-method
[
10,
1
I]. The flat surfaces
of the sliding wear samples allowed determination of
the depth distributions of micro-plastic deformation and
residual stresses within the first 30pm using the
in-
verse Laplace transformation method
[
1
11.
RESULTS
AND
DISCUSSION
SLIDING WEAR OF SILICON CARBIDE
In systems operating under boundary friction con-
ditions (like water-lubricated face-seals during
start/stop operations) the wear behaviour can be related
to the surface integrity of the components in addition to
micro-hydrodynamic effects. This may be illustrated by
Figs. 2
-
4,
in which the near surface strength, the depth
distributions of residual stresses, and the material re-
moval due to sliding wear of differently machined sin-
tered silicon carbide specimens are shown.
A steep increase in surface strength of nearly a fac-
tor of
2
is obtained by increasing the >>roughness<< of
the machining procedure (Fig.
2):
the lowest surface
strength for the smoothest polishing with finest grains
and the highest surface strength for the roughest ultra-
sonic lapping with coarse grains. Although rough ma-
chining conditions like lapping with coarse grained
abrasives may introduce more damage than polishing
processes, the near-surface strength is increased due to
the machining induced high and relatively deep-
reaching compressive residual stresses (Fig.
3). Obvi-
158
ously, these compressive residual stresses are able to
reduce the wear rate after the roughness
peaks
have
been removed (Fig.
4).
gOOirature
load
in
N
7-
,
700
600
500
400
300
200
100
0
polished lapped US-lapped
0.25pm
F500 F280
polished lapped
2-4prn F220
Machining
Figure
2:
The near surface strength
of
differently machined Sic
specimens measured by the ball on plate test
Residual
stress
in
MPa
loo,
I
0
-100
-200
-300
-400
0
0.005
0.01
0.015
0.02
Depth
in
m
Figure
3:
The depth distributions
of
the residual stresses of differ-
ently machined Sic specimens
Layer removal in
yrn
100
10
1
01
-.
.
polished polished lapped lapped US-lapped
0,25pm 2-4pm F500 F220 F280
Machining
Figure
4:
The wear rates
of
differently machined Sic specimens.
Each bar represents a mean value from
three
specimen pairs.
The polished specimens show very high wear rates
at early stages of wear due to adhesive failure followed
by a continuous material removal. The lapped speci-
mens show significantly lower wear rates. Both effects,
lubrication reservoirs due to higher surface roughness
and especially high compressive residual stresses lead-
ing to a higher wear resistance, lead to this lower wear
rates. During the first stage of test up to
1
km
wear of
US
oscillation lapped surfaces reduces the roughness
peaks
at about lpm. Thereafter only very few wear is
observed during the stages of test up to 10
km.
This can
be attributed to the relatively far reaching compressive
residual stresses for these specimens, which are still
present
after
the removal of about 1.5 pm of the surface
layer (Fig.
3)
and to the existence of static lubrication
reservoirs..
\
m4
52
US
swing
lapped
0
2 4
6
8
10 12 14
16
Sliding distance in
krn
Figure
5:
Comparison
of
the wear
of
laser-beam structured surfaces
with ultrasonic swing lapped surfaces of Sic specimens
A
comparison of the wear of laser-beam structured
surfaces with ultrasonic swing lapped surfaces of Sic
specimens is shown in Figure
5.
The shaping by laser
beams leads additionally to a reduction of wear due to
the lubrication reservoirs in the first stages of test.
But
this beneficial lubrication effect seems to be present
only for a limited period in contrast to the
far
reaching
beneficial effect of the compressive residual stresses.
The corresponding microstructures of an ultrasonic
oscillation lapped surface and a laser-beam shaped
surface of Sic specimens after a sliding distance
of
10
km
are shown in Figure
6.
MACHINING
AND
ROLLING WEAR
OF
SILI-
CON NITRIDE
Silicon nitride ceramics are appropriate materials
for roller bearings designed for an operation without
lubrication, e.g. for vacuum applications. Hence, silicon
nitride
balls
and rings are being investigated under
rolling contact and wear situations. The main effects of
rolling wear on silicon nitride, generation of cracks and
compressive residual stresses, are compiled in Figure
7.
For
different slip ratios the development of the re-
sidual stress components parallel and transverse to the
rolling direction is shown. The test started with
no
significant machining induced residual stresses as the
samples were
in
a fine-ground and partially polished
condition. With increasing number of cycles, high
compressive residual stresses are created by the cyclic
contact loading of the surface. The tangential loading
of
the near-surface layers seems to
be
the most important
driving force
for
the residual stress development. This
can be concluded from the fact, that the amount
of
residual stresses is higher parallel to the rolling direc-
tion and from the higher residual stresses of the sample
tested at a higher slip ratio.
159
Figure
6:
Ultrasonic oscillation lapped surface (top) and laser-beam
structured surface (bottom)
of
Sic
specimens after a
wear distance
of
10
km
n
-
I
LUU
20000
40000
60000
80000
100000
Cycles
of
overrolling
Figure
7:
Development
of
residual
stress
components in silicon
nitride, parallel
(RD)
and transverse
(TD)
to the rolling direction
(contact load
=
3
GPa, slip ratios
1
%
and
3
%,
dry rolling)
and micro-graph showing cracks in the
2
mm
broad
race-
way after
300
cycles.
After about 30,000 cycles the residual stress level
saturates and
is
nearly constant up to 100,000 cycles,
when dramatic failure processes
start
(the sample tested
at
1
%
slip failed early due to a failure of the wear test-
ing machine).
The beneficial effect of the rolling-induced residual
stresses on the life time can be imagined from the
mi-
cro-graph of the sample tested at
3
GPa and
3
%
slip.
Already after
300
cycles large cracks are present. They
are oriented transverse to the rolling direction and ex-
tend across the entire raceway. Apparently they follow
the direction of the maximum tensile residual contact
stresses. Obviously the compressive residual stresses
being simultaneously introduced by the cyclic contact
loading stabilise the surface layers by crack closure
effects and maintain the performance up to
100,OOO
cycles.
The effect of machining induced residual stresses on
the rolling wear behaviour of silicon nitride is shown in
Figure8. Specimens in the ground condition
(D64
wheel) with machining induced compressive residual
stresses in the range of 400MPa and laser assisted
turned
(LAM)
samples with negligible residual stresses
were run at
1.5
GPa and
3
%
slip under dry conditions.
100
200
300
400
500
Thousend
of
cycles
of
overrolling
Fig.
8:
Development
of
residual
stress
components parallel to the
rolling direction in ground and in laser-assisted machined
(LAM)
silicon nitride due
to
rolling wear (contact
load
=
1.5
GPa, slip ratios
3
%,
dry rolling) and micro-graphs showing cracks in the raceway
after
5000
cycles.
Up to
20,000
cycles no significant change of the
grinding induced residual stress state is observed
whereas compressive residual stresses are developed
in
the LAM-sample. The micro-graphs in Figure
8
indi-
cate that in the early stages of the cyclic contact loading
the grinding induced residual stresses obstruct the de-
velopment of cracks in the surface layer.
After
20,000
cycles the residual stresses begin to
diminish in both samples due to abrasive surface layer
removal. At later stages of cyclic loading, the effect of
the layer removal on the residual stress states is com-
pensated by rolling induced residual stresses. This pro-
cess is more evident for the LAM-sample as less sub-
surface damage
is
present
than
in
the ground sample.
160
CONCLUSIONS
The investigations have shown that finishing proc-
esses, abrasive wear and cyclic contact loading intro-
duce substantial micro-plastic deformation and com-
pressive residual stresses into the surface layers of
high-strength ceramics. Compared to rough machining
conditions using high material removal rates, relatively
thin surface layers are affected. Thus, advanced X-ray
diffraction methods for depth probing and surface sen-
sitive strength tests like the ball-on-plate test are needed
to determine and to assess the surface integrity.
A significant beneficial effect of the residual
stresses on the life time is obtained for sliding applica-
tions of silicon carbide under boundary friction condi-
tions. After the removal of the topographic
peaks
dur-
ing wear in the early stages surface layer with compres-
sive residual stresses due to rough machining
are
able
to withstand abrasive wear better than smoothly pol-
ished surfaces containing no significant machining
induced stresses.
The near-surface residual stress investigations per-
formed on cyclic contact loaded silicon nitride have
shown the development of surprisingly high compres-
sive stresses up to the GPa-range. In addition,
the
cyclic
contact loading introduces severe damage in the very
early stages
of
loading. However, the specimens show
acceptable life-times due to the crack closure effect of
the compressive residual stresses. Compressive residual
stresses due to moderate grinding conditions shift the
generation of damage to higher cycle numbers.
ACKNOWLEDGEMENT
The machining of the samples was performed by the
Fraunhofer-Institut fur Produktionstechnologie
(IPT),
Aachen. The investigations were sponsored by the
German Science Foundation (DFG).
REFERENCES
G. Knoll, V. Lagemann, A. Radtke, ‘Stromungs-
mechanische Kennwertbildung zur Charakterisie-
rung mikrohydrodynamischer Effekte sowie zum
Festkorperkontakt keramischer Werkstoffe’, Werk-
stoffe fur die Fertigungstechnik
(F.
Klocke, B.-R.
Hohn, Edts.), Wiley-VCH Verlag GmbH, Wein-
heim,
(
1999) 175-
180.
M. Geiger, W. Konig, H.K. Tonshoff, ‘Einsatz von
Laserstrahlung zur Herstellung tribologisch opti-
mierter
Oberflachentopographien’
,
Werkstoffe fur
die Fertigungstechnik (F. Klocke, B.-R. Hohn,
Edts.), Wiley-VCH Verlag GmbH, Weinheim,
M. Zimmermann, M. Rombach, ‘Microhydrody-
namic Effects in Sliding Contacts of Real Surfaces:
A Numerical Approach’, Ausmb
98
-
Tribology at
(
1999) 157- 162.
Work, The Institution of Engineers Australia,
[41 W. Pfeiffer, E. Sommer, ‘Bewertung der Festig-
keitseigenschaften
oberflachen-bearbeiteter
Kera-
mikbauteile’, Werkstoffkunde, DGM Informati-
onsgesellschaft mbH, Oberursel, (199 1) 575-584.
[51 W. Pfeiffer,
T.
Hollstein, ‘Damage determination
and strength prediction of machined ceramics by
X-ray diffraction techniques’, Machining of Ad-
vanced Materials. NIST Special Publication 847,
United States Department of Commerce, (1993)
[6]
W. Pfeiffer, T. Hollstein, ‘Characterization and
Assessment of Machined Ceramic Surfaces’, 2nd
Intern. Confer. on Machining of Advanced Materi-
als (MAM), VDI-Verlag, Dusseldorf, VDI-
Berichte Nr. 1276,
(196)
587-602.
[7] M. Rombach, ‘Experimental and Theoretical In-
vestigations on Plastic Deformation and Brittle
Cracking in a Ball-on-Plate Contact of Ceramic
Materials’, Ceramics Charting the Future (Edit. P.
Vincenzini, Techna Srl), (1995) 1055-1064.
[8] W. Pfeiffer, M. Rombach, ‘Residual Stresses and
Damage in Ceramics Due to Contact Loading’,
Proc. of the Fifth Int. Conf. on Resid. Stresses,
Linkoping, Univ., Sweden, (1998). 1,302-307.
[9]
R. Zeller,
T.
Hollstein, ‘EinfluO der Oberflkhen-
bearbeitung von Hochleistungskeramiken auf das
GleitverschleiBverhalten
im Grenzreibungsgebiet
und auf das
WiiIzverschleiBverhalten’,
Technische
Keramische Werkstoffe,
(J.
Kriegesmann, Edt.),
51. Erg.-Lfg., (1999) 5.2.2.3.
[
10lP. Muller, E. Macherauch, ‘Das sin$-Verfahren
der rontgenographischen Spannungsmessung’
,
Z.
angew. Physik
13,
(1961) 305-312.
[11]V. Hauk (Edt.) ‘Structural and residual stress
analysis by non-destructive methods’, Elsevier,
(1998) 201
-
206.
235-245.
1997, ISBN 0-444-82476-6.
161