The measurement of sliding friction and wear of ceramics at high temperature

The measurement of sliding friction and wear of ceramics at high temperature

Wear, 138 (1990) 169 - 187 169 THE MEASUREMENT OF SLIDING FRICTION CERAMICS AT HIGH TEMPERATURE* AND WEAR OF M. G. GEE National Physical Laborat...

1MB Sizes 0 Downloads 15 Views

Wear, 138 (1990)

169 - 187

169

THE MEASUREMENT OF SLIDING FRICTION CERAMICS AT HIGH TEMPERATURE*

AND WEAR

OF

M. G. GEE National Physical Laboratory,

Teddington,

Middlesex (U.K.)

C. S. MATHARU Department

of MateriaLs Technology,

Brunel University, &bridge,

Middlesex (U.K.)

E. A. ALMOND National Physical Laborato~,

Teddington,

~iddZesex (U.K.)

T. S. EYRE Department

of Materials Technology,

Brunel University, Uxbridge, Middlesex (U.K.)

Summary

A wear testing system is described which can be used to carry out sliding wear tests at temperatures up to 1500 “C. The tests can be performed in air, vacuum or controlled atmospheres under applied loads in the range of 5 - 100 N and at speeds up to 1.5 m s- I. The relative displacement of the two specimens and the frictional load generated between the two surfaces are monitored throughout the test. Preliminary results are given of wear tests carried out at temperatures up to 1200 “C on reaction bonded silicon nitride, sir&red silicon nitride, sintered silicon carbide and yttria partially stabilized zirconia.

1, Introduction

There is increasing interest in the use of ceramics in gas turbines and internal combustion engines. This is prompted by the attractive properties of ceramics, such as their good high ~rnpe~t~e strength and low thermal conductivity, which offer oppo~~ities for reducing power losses and improving performance. As a result of the possibility of using ceramics for wear resistant components in engines, there is a demand for information about the wear and friction of ceramics at elevated temperatures. However, before results from laboratory wear tests can be confidently used by engineers and designers, it is important to assess the relevance of these tests.

*Paper presented at the International CO, U.S.A., April 8 - 14, 1989.

Conference

on Wear of Materials, Denver,

0043-1648

Elsevier Sequoia/Printed

in The Netherlands

170

There have been a number of studies of the wear and friction of ceramics at moderate temperatures (up to about 800 “C) [l - 41, where, with the addition of heating equipment, conventional wear test machines can often be used. However, there have only been a few reports of experiments at higher temperatures [ 51. It is expected that ceramic components will be subjected to temperatures up to 1000 “C in internal combustion engine applications and up to 1400 “C in gas turbines. However, as the applications are developed there are likely to be increases in the upper temperature limits of components. For this reason, an essential requirement for the test system was that it should have the capability for testing up to 1500 “C. This paper reports the results of an initial set of tests on reaction-bonded silicon nitride (RBSN), sintered silicon nitride (SSN), sintered silicon carbide (Sic) and yttria partially stabilized zirconia (PSZ).

2. Test system The geometry of the test system has a considerable effect on the magnitude and type of wear that is observed [6]. Thus, the results that are obtained using one test geometry and environment cannot immediately be applied to applications where the service conditions are radically different. The approach that was adopted in the current design was to choose a simple continuous sliding geometry which would allow a fundamental understanding to be developed of the processes that occur in high temperature wear of ceramics. This knowledge should then be applicable to other geometries and types of motion. Since oil-based lubricants may be difficult to apply at high temperatures, the unlubricated wear of ceramics was of initial interest. However, this does not exclude the examination of solid selflubricating coatings which have been developed recently [ 71. Some ceramics oxidize in air at high temperatures. This may have a considerable impact on wear and friction processes. In order that the effects of changes in atmospheric composition can be examined, it is important to be able to change the atmosphere surrounding the wear specimens. This was achieved in the current design by enclosing the wear system in a vacuum enclosure, evacuating and back-filling with the appropriate gas composition. This approach meant that tests could also be performed under vacuum. Ceramics are difficult and expensive to machine. In order to reduce the cost of testing it was important to make the specimen geometry as simple as possible. In the test system described here, this was achieved by the use of a basic ring geometry which could easily be produced from most ceramics as unmachined sintered bodies. Machining was then confined to simple grooving and grinding of the wear surfaces, thereby avoiding the machining of holes. The general assembly of the test system is shown in Fig. 1. Most of the system was manufactured from stainless steel, with those components

171

2 1 red

Water cooled bearings Vacuum Load

Frlctlon

chamber pan

load

cell

Displacement transducer

Fig. 1. General arrangement of test system.

subjected to high temperatures fabricated from high purity alumina or RBSN. The control and signal conditioning electronics were housed in a rack mounted to the side of the system. 2.1. Specimen geometry and drive The specimens were rings with the flats of the rings used as the surfaces in contact (Fig. 2). The contact area between the surfaces can be reduced by machining most of the flat away from one of the rings to leave an annular surface in continuous contact (in the tests reported here, the upper specimen was reduced in contact area), and can be machined away even more to leave three small areas in contact (this is broadly equivalent to a “three-pin-ondisc” geometry). The specimen assembly is shown in Fig. 3. The lower specimen was driven by a ceramic drive tube through a coupling piece which fitted into straight slots cut in the lower face of the specimen. The upper specimen was kept stationary and was loaded by a central ceramic rod and driving pins. The system was driven by a 0.37 kW d.c. variable speed motor. The drive was transmitted to the ceramic tube by flexible couplings and a 2:l reduction gear. To enable tests to be carried out in a vacuum or other controlled atmospheres, the drive was taken into the vacuum chamber through a

172

Test

surface

(4 Fig. 2. Specimens:

(a) upper,

(b) lower.

r-

Loadmg

Retalnlng r,ng

speclme” . SPeclmen

Coupling

Fig. 3. Specimen

rod

assembly.

coupling where the seal was achieved using magnetic fluid. The speed of the motor could be varied continuously from 0 to 800 rev min-‘. The motor control incorporated a relay which cut off power to the motor when the frictional torque exceeded a pre-set value (normally equivalent to a friction coefficient of 1.5). 2.2. Specimen heating and environmental control The specimens were heated by a furnace which was mounted on an adjustable pedestal. This was lowered onto the specimen assembly and locked into position. Two different furnaces were used, a wire-wound furnace with Kanthal windings for temperatures up to 1200 “C and an Sic element furnace for temperatures from 1000 to 1500 “C. The specimen temperature was measured by an armoured thermocouple led through the seal with the tip located in contact with the stationary specimen. The thermocouple output was monitored by a digital thermometer. The temperature of the furnace, which was monitored by a thermo-

173

couple buried in the windings of the furnace, was controlled by a proportional-integral-derivative (PID) device which controlled the power supplied by a matched thyristor unit. For tests in vacuum, a closed alumina tube was fitted over the specimen assembly and located in a water cooled flange. Cooling water was provided via flowmeters mounted on one side of the support framework. The whole volume inside the enclosing tube and the gear chamber was evacuated, and could then be backfilled with gas of a known composition. To perform tests in laboratory air the enclosing tube was not fitted. The vacuum system consisted of a high ~roughput diffusion pump backed by a rotary pump giving a vacuum better than 10d2 Pa. 2.3. ~e~u~e~en t and loading The load was applied by adding weights to a loading pan mounted at the base of the central loading rod. A load cell placed at one side of the load pan prevented its rotation and measured the frictional force. The load cell output was calibrated by known weights before it was installed into the sytem and the accuracy of frictional measurements was checked by applying a known torque to the upper specimen with the load cell in position. There was no noticable error. The vertical displacement of the loading tray gave a measure of wear and was monitored by a displacement transducer. The sensitivity of the displacement signal to the specimen temperature was checked by heating a specimen assembly at a controlled rate and observing the d~pla~ement output. This was a linear increase of 0.027 pm K-’ over a temperature range of 20 - 900 “C. The output from the load cell and displacement transducer were continuously recorded on a chart recorder during tests. The normal test procedure was to assemble the specimens and driving train, lower the furnace over the specimen assembly and then heat the specimens to the required temperature before starting the motor. For tests in a vacuum the closed alumina tube was placed over the specimens and the wear system was evacuated before the specimens were heated. This procedure was satisfactory apart from a test which was attempted on RBSN at 1400 “C in air, where the specimens were found to have seized together when the motor was started.

3. Results 3.1, Wear and frict~~ Wear tests were carried out on RBSN, SSN, SIC and PSZ (Table 1). Specimens were prepared with the dimensions given in Fig. 2 and the wear surfaces were lapped and polished to give a finish with a roughness Ra better than 0.1 pm (measured with a 250 pm cut-off length). All tests were carried out with an applied load of 10 N and a speed of 0.2 m s-l_

174 TABLE 1 Materials Material

Composition

(wt.%)

Hardness (HV 0.2)

Density (Mg m-3)

Finish roughness (pm)

PSZ

92.2 ZrOz, 5.28 Yz03, 2.14 HfOz, 0.12 Af203

1390

5.75

< 0.01

RBSN

95.8 SisNq, 1.42 Si, 0.6 SiOz, 0.64 AlzOs, 0.41 Fe203,0.33 CaO

1030

2.56

0.01

SSN

79.5 Si3N4, 11 Y203, 1.8 SiOz, 1.22 Si, 3.94 MgO, 2.19 C&OS

1560

3.28

0.07

SiC

94.89 SIC, 5.02 Si, 0.04 A1203, 0.04 Fez03

3200

3.1

0.025

Slldmg

d6tance.m

Fig. 4. Displacement traces for RBSN tests carried out in air. Numbers by traces are the temperatures of the tests.

As an example of the behaviour often observed, Fig. 4 shows the variation of displacement with sliding distance for tests on RBSN in laboratory air at temperatures from 21 to 1170 “C. There was an inGreasein wear with temperature and sliding distance at temperatures up to 900 “C. However, at 1170 “C almost no change in wear displacement was recorded. Normally there was a relatively high rate of wear at the start of the test, followed by a reduced wear rate which continued for the remainder of the test. The displacement curve for the test at 900 “C showed an additional region of increased wear rate at a sliding distance of about 700 m. A change in wear rates was also observed in a test carried out on RBSN at 600 “C in air (in this test the area of contact had been reduced leaving

175

three contact areas). After 180 m of sliding the wear rate increased sharply, with a corresponding increase in the friction coefficient (Fig. 5). Wear rates were calculated from t\e mass loss of the specimens and are shown in Fig. 6 (Table 2). The wear rates for RBSN tested in air generally reflected the displacement curves, with an increasing wear rate up to 900 “C, but with a mass gain measured for the tests at 1140 and 1170 “C, making the

1.2 -

E

lo-

; ‘, 3

0.6

L,

-

_

1500

_

1000

5 4 r:

06-

0.4

-

02-

__-----

/

-----J 100

/

/

/

/

,

/

/

/

/

/

/

/

/

_) /

E” 3 z 500

I

I 300

200 Slldmg

dstonce,

LOO

200

400

600 Temperature.

800

1000

2

0

m

Fig. 5. Displacement and friction traces for test on RBSN (machined separate contacting areas) at 900 “C in air.

0

5 f

1200

to leave three

1400

lC

Fig. 6. Variation of wear rates calculated from mass losses of upper rings with temperature for tests in air: 0, RBSN;., SSN; m, PSZ; +, Sic.

176 TABLE Results

2 of tests

Temperature (“C)

Material

Wear rate ( lo3 m3 N-l m-‘)

upper RT 300 600 900 1150

RBSN RBSN RBSN RBSN RBSN

RT 300 600 900 1140

RBSNa RBSNa RBSNa RBSNa RBSNa

RT 300 600 900 1180

SSN SSN SSN SSN SSN

RT 300 600 900

PSZ PSZ PSZ PSZ

RT 300 600 900

SIC Sic Sic SIC

7 152 189 1230

796

_b

19

625 815 1150 1660

Roughness after test (Pm)

Duration

0.21 0.16 0.23 0.8 0.76

0.05 0.04 0.42 0.57

432 168 107 18 7

lower

2 62 130 _b

Friction coefficient

0.1

0.5 0.5 0.6 0.74 0.7

4

0.07 0.05 0.05 _c _c

0.26 0.7 0.22

4110 33300

1520 10800

0.42 0.2 0.23 0.7 0.82

62 126 539 150

44 43 547 5070

0.8 0.72 0.76 0.78

_b

_b

1

7.4 -b 2.83 10 _b

_b

24 990 879 1260 8800

_b _b

10 4 4

0.53 0.36 0.2 0.22

- 0.06 - 0.05 - 0.43

0.25 - 0.46 0.27 - 0.66 0.45 _b

_b

0.02 0.27 0.03 0.04

(103 s)

72 20 18 18 11

- 0.1

23 11 11 11 3 18 14 13 19 74 17 16 94

aTested in vacuum. bNot available. ‘Too rough to measure.

calculation of a wear rate impossible for these tests. Similar wear rates were found for the PSZ, again with increasing wear rate with temperature. However, the wear rates for the SSN and the Sic were much lower at temperatures below 900 “C, with little or no wear recorded for the SSN. Above 900 “C the wear of the SSN increases dramatically whilst the wear rate for the Sic remains about the same. It must be noted that the calculation of wear rates for the Sic and SSN was likely to have been affected by oxidation of the wear specimens, and must be treated with caution, since any oxidation products that remained on the specimen would have increased the specimen mass. The wear rates for RBSN tested in a vacuum of low2 Pa are compared with the values for tests in air in Fig. 7. Higher wear rates were observed at temperatures up to 900 “C, where the wear rates are nearly equal.

177

0

200

LOO

600

800

1000

1200

lLO0

Temperoture.'C

Fig. 7. Variation of wear rates calculated from mass losses with temperature tests in air (0, upper; 0, lower) and vacuum (A, upper; A, lower).

for RBSN

12y

1.0 -

E

0.8

u

3 C.6 s

2

I; 0.4

0.2

0

0

I

I

200

LOO

I 600

I

I

I

800

1000

1200

lLo0

Temperature. OC

Fig. 8. Variation of friction coefficients with temperature A, RBSN; +, SSN.

for tests in air: 0, PSZ; l, Sic;

It is interesting that the lower specimen often wore more than the upper specimen. This was always true for RBSN and was particularly noticeable for the tests carried out in air. The variation in friction coefficients (average value after initial few minutes of test) with temperature are shown in Fig. 8. The friction coefficient for RBSN and SSN specimens tested in air was about 0.3 up to a temperature of 600 “C and then increased to about 0.7 at higher temperatures. The friction coefficient for the PSZ was about 0.8 from room temperature to 900 “C. For Sic the friction coefficient decreased from an initial high value of about 0.5 to 0.2 - 0.3 at temperatures above 600 “C.

178

0



I

0

300

L 600

I

I

900

1200

Temperature:C

Fig. 9. Variation of friction coefficients with temperature in a vacuum (A) of 10e4 mPa.

for RBSN tests in air (0) and

In contrast to the friction coefficient for RBSN tested in air, the friction coefficients remained high at 0.6 - 0.8 for RBSN specimens tested in vacuum at temperatures between 21 and 1140 “C (Fig. 9). 3.2. Microstructural examination Examination of polished sections revealed the microstructure of the materials (Fig. 10). The Sic consisted of elongated grains with a grain size of about 4 pm. The RBSN was quite porous and, apart from the Si3N4 grains, there was a small volume fraction of some lighter contrast areas which were thought to be unreacted silicon. The SSN was made up of small elongated grains (a few micrometres in size) with a small volume fraction of an unidentified lighter contrast phase. The PSZ had equiaxed grains of about 0.5 pm in size. The prepared surfaces of the wear specimens were examined using a scanning electron microscopy (SEM) (Fig. 11). There was a large number of pits in all the specimens, except the PSZ, due to grain tear-out introduced in the preparation, and from the intrinsic porosity of the materials. The surface of the PSZ was very smooth, with only a few scratches visible. After testing there was considerable variation in the appearance of the worn areas on the specimens (Table 3). Generally, there was a correlation between the wear rate and the appearance of surfaces, with the surfaces becoming rougher and duller as the wear rate increased. This trend was shown best by the RBSN tested in air, where at room temperature no obvious wear was visible, at 300 “C the wear track seemed smooth and polished, at 600 “C some areas were becoming rough and at 900 “C and above the surface was rough and dull. The RBSN tests made in vacuum were an exception to this tendency where, despite quite high wear rates, the wear surfaces were quite smooth up to a temperature of 900 “C.

(a)

(b)

(cl Fig. 10. Micrographs

(d) of polished

sections:

(a) SIC, (b) SSN, (c) RBSN, (d) PSZ.

Examination of the wear surfaces by SEM showed a number of different types of wear damage. The wear surfaces which had been heavily worn were often covered with particulate material (Fig. 12). This was observed for the SSN tested at 1180 “C, the PSZ tested at 900 “C, and the RBSN tested both in air and in uacuo above 900 “C, with particularly large quantities on the RBSN specimens tested in vacuum. What appeared to be highly deformed material was observed on the wear surface of the PSZ specimen tested at room temperature in air, but there was no particulate material (Fig. 12). The surfaces of the specimens that had worn heavily were coarsely grooved (Fig. 13). Fine scale grooving was also visible on the PSZ specimen tested at 900 “C in air. There was evidence for delamination in some of the specimens (Fig. 14). In the sample of SSN tested at room temperature plates of material were observed which were only loosely attached to the remainder of the surface. Areas where material had broken away from the surface were evident in the specimens of RBSN tested at 300 and 600 “C, with

(a)

(cl Fig. 11. Scanning electron micrographs (b) RBSN, (c) SSN, (d) PSZ.

(b)

Cd) of as-prepared surfaces of specimens:

(a) SE,

approximately half the smooth surface of the specimen tested at 300 “C being lost. An interesting feature observed was ring cracks and the surface of the RBSN specimen tested at room temperature in vacuum (Fig. 14(d)). At high temperatures, oxidation of the non-oxide materials away from the wear interface occurred and was present as layers of oxidized material (confirmed by X-ray diffraction) which grew thicker as the temperature of the test increased (Fig. 15). 3.3. Pro file measurements The profile measurements generally reflect the appearance of the wear surfaces, with a higher roughness value found for the rougher, more heavily worn specimens (Table 2). Thus, for the RBSN tested in air there is a sharp increase in the roughness value at 600 “C which reflects the sudden increase in wear that occurs at this temperature. In the PSZ tests, the roughness of the specimens after testing is very much higher than the initial surface finish. It is also interesting that in most cases the surface finish of the Sic specimens, and the RBSN specimens that were lightly worn, was little changed after testing.

181 TABLE 3 Appearance of wear surfaces

Temperature

Material

Appearance

RT 300 600 900 1150

RBSN RBSN RBSN RBSN RBSN

No obvious wear Smooth and polished Smooth areas/Rough areas Rough and dull Rough and dull

RT 300 600 900 1140

RBSN* RBSNa RBSNa RBSN* RBSNs

Smooth and polished Smooth and dull Smooth and dull Smooth and dull Rough and dull

RT 300 600 900 1180

SSN SSN SSN SSN SSN

Smooth Smooth Smooth Rough, Rough,

RT 300 600 900

PSZ PSZ PSZ PSZ

Rough and Rough and Rough and Very rough

RT 300 600 900 1145

Sic Sic Sic Sic Sic

Smooth Smooth Smooth Smooth Smooth

and polished and polished and polished white deposit white deposit dull dull dull and dull

and and and and and

polished polished polished polished dull

aTested in vacuum.

4. Discussion 4.1. Testing In use, the test system proved to be reasonably easy and simple to operate. However,, in tests where a large amount of wear occurred, a great deal of vibration was generated. As a result of this there was a tendency for the specimen assembly to fall apart during the test. This was cured simply by using high temperature alumina-based cements and platinum wire bindings to make the assembly more’secure. The measurement of wear displacement and frictional torque at the base of the system was also far from ideal, because of the long distance between the interface and the measuring transducers. Nevertheless, the calibration trials for the friction load cell and displacement transducer gave confidence in the results, particularly with the low temperature

(a)

(c)

(b)

(d)

Fig. 12. Scanning electron micrographs: (a) SSN specimen tested at 1180 “C in air, (b) PSZ specimen tested at 900 “C in air, (c) RBSN specimen tested at 900 “C!in vacuum, and (d) PSZ specimen tested at room temperature in air.

coefficient of the vertical displacement measurements of 0.027 pm K-l. This was probably due to the fact that the central loading rod and driving tube were of similar lengths and made from identical materials, so that they formed a balanced arm extensometer with approximately equal thermal expansions in the two arms (central rod and drive tube). 4.2. Effect of geometry and system It is known from wear tests on alumina at room temperature [6] that the geometry of the specimens can have an important effect on the results. This can be due to many factors such as whether the surfaces are in conformal contact or not, whether there is trapping of debris between the surfaces and whether there are differences in the heat flow away from the wear interface. In the tests reported in this paper, it is quite interesting that the lower specimen often wore more than the upper specimen. This is probably linked to the different geometry of the upper and lower specimens, with a raised annular ridge on the upper specimen pressed into a flat surface on the lower

(a)

(b)

(cl Fig. 13. Micrographs: (a) SSN specimen tested at 1180 “c in air (optical), (b) PSZ specimen tested at 900 “c in air (SEM), and (c) PSZ specimen tested at 900 “C in air (optical).

specimen. The effect of this ridge will be made clearer in tests which are in progress where specimens have been prepared with a flat surface on the upper specimen, and a ridge on the lower specimen. It should be noted that the initial tests on RBSN were carried out with a planar contact between the two wear surfaces. As there is quite a large clearance between the loading rod and the hole through the lower specimen, there is a possibility that the two specimens might not have remained concentric. In practice this was not found to be true, with the final wear track on the lower specimen only a little wider than the ridge on the upper specimen. To examine the effect of the continuity of the contact between the surfaces, further tests will be carried out on specimens where the continuous ring has been machined away to leave three equidistant raised areas in contact with the lower specimens (one initial observation from this type of test has been reported in this paper). With the continuous upper ring, debris is likely to be trapped between the two specimens so that the load is carried partly by debris. With an interrupted upper ring, this debris will be brushed away. The effect of changes in load, speed and surface finish will also be examined. It is expected that the dynamics of the test system will have a large effect on the results. An indication of the sensitivity of the system was the large amount of vibration that was generated in some tests, which could be

(b)

(cl

Id)

Fig. 14. SEM micrographs: (a) SSN specimen tested at room temperature, (b) RBSN specimen tested at 300 “C in air, (c) RBSN specimen tested at 600 “C in air, and (d) RBSN specimen tested at room temperature in vacuum.

reduced or eliminated simply by touching the central loading rod quite lightly. This indicated that a major source is likely to be the bending and twisting of the central loading rod under the rapidly changing loads generated at the local contacts between the two surfaces. Further tests will be needed to investigate the detailed nature of the erection of system dynamics with specimens.

Whilst a fully satisfactory interpretation of the results must await a more complete examination of the wear specimens and debris, a preliminary interpretation of the results can be made, where it is clear that the oxidation of the non-oxide materials plays an important part in the wear that was observed. RBSN specimens tested in a vacuum exhibited a higher wear rate than those tested in air at temperatures up to 900 “C, where the wear rates were almost equal. A large volume of particles was generated, possibly as a product of the abrasion of one wear surface by asperities on the other

Fig. 15. SEM micrographs of surfaces away from wear contacts: 1140 “C, (b) SiC tested at 1145 “C.

(a) RBSN tested at

surface or, in the later stages of wear, by the wear debris trapped between the wear surfaces. The wear of RBSN in air is reduced by the formation of masses of oxidized Si3N4 wear debris which separates the two surfaces, reducing the wear rates. As the temperature increases to 900 “C, direct oxidation of the wear surface increases, with the sudden increase in wear rate part of the way through the test was probably due to breakaway of oxidized material from the wear surface, leaving a fresh unoxidized surface. As the oxidation of the wear surface increases still further, a thick layer of soft oxidized material forms directly separating the contacting surfaces and giving the low apparent wear rates observed at 1140 “C. Thus, the wear of RBSN is thought to be governed by a balance between oxidation of Si3N4 debris at low temperatures, and increasing oxidation of the contacting surfaces at high temperatures. It is also likely that mechanical damage to the wear surfaces was occurring. Thus, the sudden increases in wear observed in Figs. 4 and 5 for some tests may be due to the accumulation of surface damage in the surface during wear until sufficient damage has occurred to cause breakaway of surface layers with a consequent increase in wear [8]. There are similar trends in wear behaviour for the SSN, where it is thought that the low wear rates below 900 “C are due to the separation of the wear surfaces by masses of oxidized debris. Above this temperature this gross oxidation of the contacting surfaces occurs, leading to increased wear through a combination of oxidation of fresh material and removal of the resultant loosely adherent surface layers. As might be expected, the hardest material (at room temperature), Sic, had quite reasonable wear resistance at most temperatures, and the material with the lowest hardness, PSZ, wore heavily in all the tests. Thus, the hardness of the material does seem to be important for resistance to wear, with the low wear of Sic due mainly to the high hardness of the

186

Sic, and the high wear of the PSZ due to the low hardness of the PSZ. However, the value of room temperature hardness measurements as a way of predicting the wear resistance of ceramics is debatable, particularly for wear at high temperature where the hardness of materials can decrease quite rapidly with increasing temperature, and the ranking of hardness can change. There are also other major differences between the materials. Thus, the low wear of the Sic is also likely to be a function of the good oxidation resistance of this material. Wear will also be affected by the toughness of the materials, since if crack growth is involved in the wear processes, then a higher resistance to crack growth might be expected to reduce wear rates. There is a trend for high wear rates to be associated with both high friction coefficients and with rough wear surfaces. This may be simply due to the fact that high wear rates generally occur by mechanisms which lead to rough surfaces. The difficulty of moving these rough surfaces over one another then leads to higher friction. Another factor may be the reduction in friction from the reduced interaction of the wear surfaces because of the formation of intermediate layers of oxidized material.

5. Summary and conclusions A high temperature wear testing system has been designed and built for wear tests on ceramics up to 1500 “C. The specimen geometry is ring-onring and continuous sliding. Tests can be performed in air, under a vacuum or in a controlled atmosphere. The wear and friction of the non-oxide ceramics depends on the oxidation resistance of the materials and the coherence of the oxide product that is formed. For RBSN, there was an increasing wear rate with temperature for specimens tested in air up to 900 “C!,but only a small amount of wear at 1170 “C in air. When a test was attempted at 1400 “C in air the specimens seized together before the test could be started. Specimens tested in a vacuum gave higher wear rates than the specimens tested in air below a temperature of 900 “C, but the wear rates were nearly equal at this temperature. For SSN in air there was very low wear at temperatures up to 600 “C, with a sharp rise in wear rates by more than three orders of magnitude at 900 “C. A relatively low wear rate was found for Sic at all temperatures tested. PSZ wore heavily at all temperatures. High friction coefficients of about 0.6 - 0.8 were observed for RBSN specimens tested in vacuum at all temperatures, and above 600 “C for specimens tested in air. Lower friction coefficients of 0.3 were observed for tests carried out in air below a temperature of 600 “C. Low friction coefficients of 0.2 - 0.4 were observed for SSN at temperatures up to 600 “C; above this temperature the friction coefficient increased to 0.7 - 0.8. For Sic the friction coefficient was reasonably low at all temperatures at 0.2 - 0.5, but was high at about 0.7 - 0.8 for the PSZ.

187

References 1 V. Aronov, Wear resistance anomaly of magnesia partially stabilised zirconia, ASLE Trans., 30 (1) (1987) 100. 2 J. Lankford, W. Wei and R. Kossowsky, Friction and wear of ion-beam modified ceramics, J. Mater. Sci., 22 (1987) 2069. 3 B. Longson, Lubrication of high temperature ceramic materials, Tribal. ht. 16 (4) (1983) 221. 4 H. Tomizawa and T. E. Fischer, Friction and wear of silicon nitride at 150 “C to 800 “C, ASLE Trans., 29 (4) (1986) 481. 5 A. P. Semenov and A. A. Kataura, Investigation of friction and wear of corundum ceramics at temperatures to 1500 “C, Proc. 2nd Int. Conf. on Wear of Materials, April 1979, American Society of Mechanical Engineers, New York, pp. 551 - 555. 6 M. G. Gee and E. A. Almond, Wear testing for design, Proc. Conf. on Materials and Design, May 1988, Institute of Metals, London, pp. 23.1 - 23.6. 7 C. D. Dellacorte and H. E. Sliney, Composition optimization of self-lubricating chromium-carbide based composite coatings for use to 760 “C, ASLE Trans., 30 (1) (1987) 77. 8 D. W. Richerson, L. J. Lindberg, W. D. Carruthers and J. Dahn, Contact stress effects on SisN4 and SIC interfaces, Ceram. Eng. Sci. hoc., 2 (1981) 578.