Surface mechanical properties of alumina matrix nanocomposites

Acta mater. Vol. 45, No. 10, pp. 3963-3913, 1997 0 1997 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain

Pergamon

PII: s1359-4654(97)00113-4

SURFACE

1359~6454/97 $17.00+ 0.00

MECHANICAL PROPERTIES OF ALUMINA MATRIX NANOCOMPOSITES

M. STERNITZKE’T, E. DUPAS,

P. TWIGG* and B. DERBY’

‘University of Oxford, Department of Materials, Parks Road, Oxford OX1 3PH, and *University of Leeds, School of Materials, Leeds LS2 9JT, U.K. (Received 13 January 1997; accepted 19 March 1997)

Abstract-We have examined the surface properties of alumina/silicon carbide nanocomposites containing 5 vol.% Sic particles of 4 different size distributions with mean particle size from 12 to 115 nm. In all cases the nanocomposites were more resistant to a severe erosive wear environment than monolithic alumina of similar grain size. The resistance to wear increases with decreasing SIC particle size. We have measured the damage introduced to these ceramics during lapping and polishing by three techniques: counting surface flaws visible to an optical microscope, measuring surface elastic constants using an acoustic microscope, and determining surface crack sizes using a Hertzian indentation technique. Surface observation and Hertzian indentation show the nanocomposites to have both a smaller mean defect size and fewer large defects than similar grain size alumina. This is supported by our acoustic microscope results which show a lower level of surface damage in the nanocomposites after a given lapping or polishing process, 0 1997 Acta Metallurgica Inc.

1. INTRODUCTION The inclusion of small fractions (about 5 vol.%) of Sic particles into a polycrystalline alumina matrix so that a proportion of the particles become inclusions entrapped within alumina grains can result in significant increases in strength over conventional monolithic alumina [ld]. The size of particles used, typically about 100 nm or smaller, has led to this class

of materials being termed ceramic nanocomposites. Despite the increase in strength these materials display over alumina (increase of 50-100% have been reported), fracture studies show a much smaller, if any at all, increase in fracture toughness [l-5]. Thus the mechanism by which this strength increase occurs is not understood. In general, the strength of a brittle material is controlled by two parameters: its toughness and its defect population-in particular the size of the largest stressed defect. Thus two generic mechanisms of strengthening are possible. First, the toughness of a material can be increased by microstructural control which either increases the work of fracture or introduces a steeply rising crack growth resistance (R-curve). Second, the size and population density of the fracture initiating flaws can be reduced. For nanocomposites both types of mechanism have been invoked in the literature to explain the observed increase in strength and a number of microstructural features have been proposed as the causative agent: _ tTo whom all correspondence should be addressed. Present Max-Planckaddress: GKSS-Forschungszentrum, StraDe, D-21 502 Geesthacht, Germany.

dislocation sub-cells introduced by thermal mismatch on cooling from fabrication [l], crack healing on annealing [2], a small mean grain size with no abnormal large grains [3], reduced processing flaw size and distribution [4], and grain boundary strengthening [6]. For a full discussion of this literature the reader is referred to a recent review [7]. This paper is a continuation of our earlier work in which the mechanical properties of four alumina/silicon carbide nanocomposites each containing 5 vol.% Sic of different mean particle size were measured and compared with those of an equivalent grain size alumina [4]. These results are summarised as follows. All of the nanocomposites showed an increased strength over the reference alumina but there was no measurable influence of Sic particle size on strength. The fracture surfaces of the nanocomposites show the characteristic transgranular fracture which has been reported by all other workers [l-4]. Careful study of the fracture surfaces showed that the strength limiting defects were processing flaws but with different origins in the alumina and nanocomposites. In the case of alumina the defects were large volume pores, presumably caused by agglomerations within the alumina powder which were not dispersed prior to processing. In the nanocomposites the defects were smaller and associated with agglomerates of the SiC reinforcement causing crack-like voids. From this we concluded that the strength increase found in the nanocomposites was caused by a reduction in the flaw density which was possibly an effect of the Sic reinforcement refining the alumina powder during the mixing stage which used attrition

3963

3964

STERNITZKE~J~~..SURFACE MECHANICAL PROPERTIES

milling. This was further confirmed by the fact that many workers have reported a decrease in nanocomposite strength when the Sic particle fraction exceeds 5 vol.% [l, 2, 61 which we inferred as indicating increased Sic agglomeration problems. Indeed we have now been able to extend the strengthening effect in these materials to Sic contents up to 10 vol.% by using a novel hybrid of conventional powder processing and organometallic preceramic polymers to reduce the problem of SIC defects [S]. Since our original work, there has been a report of the wear properties of ceramic nanocomposites containing Sic particles of one size range [9]. In this study the nanocomposites were found to have significantly better wear resistance than alumina of equivalent grain size. The data also suggested that whereas the wear rate of alumina is strongly dependent on grain size, with the mechanism of material removal being grain pull out, the wear rate of the nanocomposites was less dependant on matrix grain size with a different mechanism of material removal that is not dominated by grain pull-out. In conjunction with the widely observed transition in fracture mode from intergranular in monolithic alumina to transgranular in nanocomposites, a grain boundary strengthening effect has been invoked to explain the good wear performance of nanocomposites. A number of workers have also reported the ease by which nanocomposites polish [2, 31. Thus, there is evidence that if the strengthening of nanocomposites is a result of a lower defect density, this may be a result of an increased resistance to surface damage as well as a reduced density of processing flaws. In this paper we present a further study of the wet erosive wear of ceramic nanocomposites with the same range of BC particle size as in our previous strength study [4]. We have used three techniques to measure the size distribution of surface flaws after lapping and polishing. These are: optical microscopy to count the surface flaws, a Hertzian indentation technique [lo, 111, and acoustic microscopy which measures the surface elastic properties of a material. These results are used to determine whether

SIC source

2.EXPERIMENTAL In this paper we investigate four different nanocomposites, all containing 5 vol.% Sic but possessing different average Sic particle sizes, together with monolithic alumina. The experimental procedures used to manufacture these materials have been described in detail elsewhere [4. 121 and hence only a brief summary will be given. The materials properties are listed in Table I where UF45 and UF 15 represent commercial sc-Sic powders (Lonza), and the number code indicates their specific surface area in m*/g. The nanocomposites with the UFl5, UF45 and sedimented UF45 Sic powders were processed by conventional powder processing. For sedimentation, 100 g of UF45 Sic powder was ultrasonically dispersed in 1000 ml distilled water and stored in a measuring cylinder for four weeks. Afterwards, 230 ml of the upper fraction was extracted and freeze dried with a yield of approximately 30%. The Sic powders were ultrasonically dispersed in deionised water for 20 min and then added to an attritor mill which used zirconia media, together with a dispersing agent (Dispex A40, Allied Colloids) and the alumina powder. After attrition milling the mixtures were freeze dried and passed through a 150 pm sieve. The Al:OJSiC nanocomposites containing ultrafine Sic particles with an average size of 12 nm [ 121 were manufactured using a polycarbosilane pre-ceramic polymer coated on the A1203 ceramic powder. All ceramics were hot-pressed in a graphite die at 1700°C for 1 h by applying a pressure of 20 MPa under argon. To obtain monolithic alumina with a similar grain size to the nanocomposites, a hot-pressing temperature of 155O’C was chosen. The as-received A&O, powder (Sumitomo, AKP53) was used without any further treatment to manufacture monolithic alumina reference material.

Polymer c

Sic content [vol.%] (d)SiC [nm] (Initial)

1;

5.0

(d)AhOl [pm] P [g/cm’1

3.91 (2)t 399 (4) 0.25 (I) 491 (63) 3.25 (27) 50.8 (45) 16

E [GPa]

rate[nm/s] Flaw density [mm-*] Number of Sic particles

nanocomposites show a resistance to surface damage greater than that of unreinforced alumina of similar grain size.

Wear

oer urn’

tNumbers in brackets represent :Vickers indentation 5 kg load.

the standard

deviation:

3.5 3.86 (2) 395 (4) 0.24 (1) 738 (115) 3.02 (21) 9.2 (12) 9.5 52,262 3.97 (2) = 3.79 + 0.02

UF15

Sed. UF45 5 5; 4.0 3.90 (2) 394 (4) 0.24 (1) 549 (125) 2.16 (31) 15.5 (2) IO 514

90 2.1 3.89 (2) 396 (4) 0.23 (I) 587 (ill) 3.04 (39) 10.1 (10) 13 131

I15 3.5 3.92 (2) 397 (4) 0.25 (I) 689 (80) 3.47 (61) 19.3 (IO) 13 63

STERNITZKE et al.:

SURFACE MECHANICAL PROPERTIES

c_I watertesed

alumina slurry 0.5 -1 mm

Fig. 1. Experimental configuration of the wet erosion wear tests.

Micrographs of polished and thermally etched cross-sections were obtained using a JEOL JEM6300 scanning electron microscope (SEM). The thermal etching was carried out in argon at 1400°C for 1 h. The average matrix grain size was obtained from the SEM micrographs using the intersection method and a correction factor of n/2. Four specimens, each with a different surface finish, were prepared from each of the materials shown in Table 1, respectively. The roughest surface finish was produced by grinding on a flat-bed grinder with a resin bonded diamond wheel (grit size = 76, equivalent to about 200 pm). Then, 150 pm of material was removed with a 12 pm diamond polish to give the second surface, 50 pm with a 3 pm diamond polish for the third and, finally, 20 pm was removed with a 1 pm diamond polish to give the smoothest surface. Wet erosion wear tests were carried out for all materials using a modified high torque attritor mill filled with a water based slurry of crushed fused alumina aggregate of 0.5 to 1 mm in size [13]. Figure 1 shows the experimental configuration for the wear tests. Discs of material 25 mm in diameter and 4 mm thick were clamped in a hard polyurethane holder with - 50% of the disk surface exposed to the slurry. The rotation speed was 8 Hz giving an average linear velocity of 1.9 m/s. Following a standard procedure, the weight loss (Am) was determined by weighing the disks after 2 h and 6 h exposure to the erosion environment. Using the volume loss (Am/p) of material normalised by the exposed surface area (A) and the testing time (t) a wear rate (R) can be defined as follows:

3965

acoustic waves of frequency 225 MHz to a line on the specimen surface. The line geometry allows anisotropic elastic properties to be evaluated but here we used it to measure the property of a given area in different directions to eliminate errors. Deionised water is used as a transmitter between lens and specimen. Surface acoustic waves (Rayleigh waves) are excited and interfere with the reflected waves. The exciting acoustic lens can be used in a pulse-echo mode to detect the reflected waves. Measuring the intensity of the reflected waves as a function of the distance between the lens and the surface (V(z) curves) allows the Rayleigh wave velocity (c.~) to be determined on a basis of a mathematical analysis given by Kushibiki and Chubachi [ 151. The transverse (Q) and longitudinal (0,) acoustic wave velocities are determined from the Rayleigh wave velocity by solving the following equation:

= 0.

(2)

Equation 2 can be solved by an iterative computer program. If the density (p) of the material is known the surface Young’s modulus (E,) and surface Poisson’s ratio (v,) can now be calculated:

2” =

J

Es (1 - vs) P(1 + vs)(l - 2 v,)

(3a)

1”= Recently, Hertzian indentation has been used to determine flaw size distributions of the surface of monolithic alumina [lo, 11, 161. For a better understanding of the method it is necessary to summarise the principles of the Hertzian indentation technique. Hertzian indentation occurs when a spherical ball is pressed into a flat surface. The elastic stress field introduced by the sphere can exactly be calculated in in

out

R=$. . . For each material four wear tests were performed and average wear rates as well as their standard deviations were calculated. The surface elastic properties of the materials were determined using a line focus acoustic microscope [14, 151. Figure 2 shows a schematic cross-section of the instrument used. A cylindrical lens focuses

Fig. 2. Experimental configuration of the line focus acoustic microscope.

3966

STERNITZKE et (11.: SURFACE MECHANICAL

PROPERTIES

3. I. Erosive wear results

r Crack with depth c and maximum K

Fig. 3. The geometry of a Hertzian contact.

terms of the load, sphere radius and elastic properties of the sphere and surface [16]. During the experiment the stress field can interact with a defect present at the surface (see Fig. 3) and a typical ring crack can form. The stress field outside the contact zone shows a tensile radial component near to the surface which decreases with depth eventually becoming compressive. Therefore, the stress intensity factor at a crack normal to the surface (K,) depends on its depth and on its distance from the contact area. Consequently, the stress intensity factor shows a maximum in its dependency on crack position and length [lo]. In an experiment only a crack with a particular length and distance from the contact area will propagate to form a ring crack for a given load (& > K,,). In the case of a diluted flaw density, the test results reflect the present flaw distribution. Hertzian indentations with an alumina ball of radius R = 5 mm were performed on 3 pm polished surfaces of all materials. The testing machine (ET10 Engineering Systems, U.K.) is equipped with an acoustic emission transducer mounted in the loading train to detect the initiation of a characteristic ring- or ring-cone crack. The diameters of the ring cracks were measured with an optical microscope and the flaw densities were determined according to the method described by Warren et al. [ll].

Figure 6 presents the wet erosive wear rates of the nanocomposites with different Sic particle size distributions and monolithic alumina (Table 1) as a function of the matrix grain size. For a better comparison, wear rates for alumina [13] and nanocomposites [9] already available in the literature have been included. All the data were measured on the same experimental equipment used in this study. From Fig. 6 the following conclusions can be drawn. First, monolithic alumina and nanocomposites show a strong dependence of the wear rate on matrix grain size. Second, the wear rates of the nanocomposites are significantly smaller than those of equivalent grain size alumina for grain sizes greater than 1 kern. The dashed line in Fig. 6 was measured from nanocomposites with the same SIC particle size as the UF45 material in this study but with different mean grain sizes. The UF45 nanocomposite shows a smaller matrix grain size than the other nanocomposites which is due to a lower hot-pressing temperature (see [4]). A comparison with the sedimented UF45 nanocomposite is therefore difficult. However, there is an apparent trend in erosion behaviour with the wear rate increasing with increasing Sic particle size. This trend is most apparent when comparing the wear rates of polymer derived and UF15 nanocomposites. which have the same matrix grain size, but very different SIC particle sizes. The sedimented UF45 and UF45 nanocomposites have wear rates intermediate between polymer derived and UF15 nanocomposites.

3. RESULTS AND DISCUSSION

Figure 4 shows a SEM micrograph of a thermally etched surface of a monolithic alumina specimen. The large platelet-like grains are typical for the alumina microstructure resulting in a broad grain size distribution. In contrast, the nanocomposites in Fig. 5 show a much narrower distribution of grain sizes. Although thermal etching was carried out in a protective atmosphere, Sic particles can react with residual oxygen and the alumina leaving pores visible at grain boundaries as well as in alumina matrix grains [ 171.

Fig. 4. Scanning electron micrograph of a thermally etched polished surface of monolithic alumina.

STERNITZKE et al.:

SURFACE MECHANICAL

PROPERTIES

Fig. 5. Scanning electron micrographs of a thermally etched polished surface of nanocomposites 5 vol.% Sic; (a) polymer derived, (b) sedimented UF45, (c) UF45, and (d) UF15.

SEM micrographs of the eroded surfaces are presented in Fig. 7. Monolithic alumina is compared with the polymer derived and UF15 nanocomposites. The eroded surface of the monolithic alumina sample is dominated by intergranular fracture-and grain pull-out as typical for the grain size regime above 1 pm [13]. The UF15 nanocomposite shows a mixed wear mechanism (Fig. 7(b)), again having a considerable intergranular fracture, but with more plastically deformed surface areas than the mono-

with

lithic alumina sample. The polymer derived nanocomposite material wears almost exclusively by a plastic deformation mechanism, with very little evidence of fracture and no grain pull-out (Fig. 7(c)). The transition in the wear mechanism from grain pull out in the case of alumina to plastic deformation or tribochemical wear in nanocomposites is in good agreement within the study by Davidge et al. [9]. Furthermore, the present study shows that the transition in the wear mechanism can directly be

3068

STERNITZKE

cf (/I.:

SURFACE

related to the size of the Sic particles. Whereas tribochemical wear is dominant in the polymer derived nanocomposite with ultrafine SIC particles, the UFl5 material shows a large number of grain pull-outs. 3.2. Acoustic microscopy Figure 8 shows the results of the acoustic microscope study. The Rayleigh wave velocity is plotted as a function of the polishing grade for the different nanocomposite materials as well as for alumina. As described in the experimental section, four different surface states were investigated where the coarsest was made by surface grinding and the others by lapping procedures. The grinding process was rather coarse with a grit size of 76 which is equivalent to a size of the diamond grinding medium of about 200 pm. Each data point in Fig. 8 represents the average value from a set of measurements on the same specimen area. The lens was rotated by 5 degrees and a new measurement was performed until a full circle was reached. The error bars in Fig. 8 represent the standard deviation from such a set of measurements. The errors decrease with a smoother surface finish which is typical for acoustic microscopy techniques. At rough surfaces waves are scattered, thus decreasing the measured signal. In Fig. 8 it can be seen that all the nanocomposites behave in a similar manner with the Rayleigh wave velocity decreasing with a coarser surface finish. No clear dependence of velocity on the SIC particle size can be observed. The Rayleigh wave velocities for alumina with a good surface finish (polishing grade < 3 pm) are similar to those for the nanocomposites. However, alumina is more affected by coarser surface treatments than are nanocomposites. As can be seen by the greater decrease of the Rayleigh wave velocities with polishing grade. On the basis of equations (2) and (3) Young’s moduli were calculated using the Rayleigh velocities in Fig. 8 and the density values determined by Archimedes method and given in Table 1. Due to the

i

3o 1 0

/ 1

2

UF15

3

4

5

6

7

Grain size [pm] Fig. 6. Wear rate grain size dependencies in the wet erosive wear of 5 vol.%

SiC/AI203 nanocomposites alumina.

compared

to

MECHANICAL

PROPERTIES

nature of equation (2) and the iteration process used for its solutions, different solutions can be obtained depending on the start values for the iteration Some of the solutions are physically process. meaningless and were excluded. It was found that for all materials the range of start parameters leading to physically meaningful results are similar. The same set of start parameters was used for all iterations. Obviously, the results have to be treated carefully but they are consistent. Therefore, only the normalised Young’s moduli are plotted in Fig. 9 as a function of polishing grade. The values were normalised by the Young’s modulus obtained at the 1 pm surface finish for each material. Again, all the nanocomposites behave in a similar manner whereas alumina shows a higher drop of the normalised Young’s modulus with a coarser surface finish. It is now necessary to link our acoustic measurements of surface elastic properties to our observations of surface damage during erosive wear. In general, wear can be defined as a process in which two surfaces interact resulting in removal of material. Wet erosive wear is very similar to a grinding or polishing process in which material is removed from a surface by a grinding medium. Thus, the acoustic microscopy study on materials with different ground and polished surface finishes can give information relevant to the interpretation of worn surfaces. Due to the small penetration depth of the acoustic Rayleigh waves (IO-50 pm), the Young’s moduli in Fig. 9 are only representative of the surface region of the specimen. A decrease of the Young’s modulus can be explained either by tensile residual stresses, by a lower density or by a higher surface damage level showing cracks and grain pull-outs. However, the level of internal stress required to significantly reduce the elastic modulus is too high to be a likely mechanism. The acoustic microscopy results are consistent with our observations of the eroded surfaces of alumina and the nanocomposites. After a wear or a grinding process, the surface of the nanocomposites is much smoother than that of alumina showing less damage in the form of grain pull-outs or cracks which is consistent with the reduced surface elastic properties of the alumina. The accuracy of the acoustic microscopy is. however, not sufficient to resolve the differences between the different nanocomposites even for good surface finishes. Chou et al. [I81 reported that both a monolithic alumina ceramic and a pressureless sintered A1201/ 5 Vol.% Sic nanocomposite showed similar surface roughness for the same machining treatment. However, we found lower Rayleigh wave velocities and therefore a higher degree of surface damage for alumina compared to the nanocomposites after grinding and 12 pm polishing. The acoustic microscopy technique is not only sensitive to the surface roughness but also to cracks introduced while machining which cannot be detected by a profilometer.

STERNITZKE et al.:

SURFACE MECHANICAL

PROPERTIES

3969

Fig. 7. Scanning electron micrographs of worn surfaces of (a) monolithic alumina, (b) a 5 vol.% UF15 SiC/AhO, nanocomposite, and (c) a 5 vol.% polymer derived Sic/Also. nanocomposite.

3.3. Visual inspection of surfaces In the case of nanocomposites, the resistance to surface damage is substantially increased over that of monolithic alumina, especially at large matrix grain sizes as demonstrated by the wear tests. In order to investigate this in more detail we conducted a study of the density of surface flaws on the polished surface of the material. Therefore, we investigated defect populations like cracks, grain pull-outs and processing flaws on polished surfaces. Figure 10 shows flaw

size distributions for all materials obtained from 1 pm polished surfaces by optical microscopy. A surface area of 30 mm* was carefully examined and all visible flaws were sized and counted. In the case of the monolithic alumina, flaws up to a diameter of 70 pm are present even after a 1 pm polishing treatment. Typically, these flaws can be identified as large grain pullouts. However, the nanocomposites show much smaller flaw sizes up to a maximum diameter of 30 pm. As shown in previous studies the

STERNITZKE PI al.:

3970

73,

5800

.g

5700

F c .o, a, 2 fY

SURFACE MECHANICAL

PROPERTIES

Alumina shows the largest mean flaw size and also has large flaws of up to about 9 pm. In the case of the nanocomposites, the mean flaw size increases with increasing SIC particle size. The polymer derived material shows the smallest flaws and the UF15 material shows the largest ones with its largest flaws almost the same size as those of monolithic alumina. The flaw densities for alumina presented in this study are in good agreement with data taken from alumina with a similar grain size published by Franc0 PI al.

5600 5500

V61. 5400

1

10

100

Polishing grade [pm] Fig. 8. Average Rayleigh velocities as a function of polishing grade of 5 vol.% Sic/Ah03 nanocomposites compared to monolithic alumina.

flaws can be described as “crack-like voids” due to SIC agglomerations formed during processing [4,8]. Grain-pullouts were only occasionally observed. In Fig. 10 a trend can be recognised within the nanocomposite series: The UFl5 nanocomposite with the largest SIC particle size distribution shows the largest flaws whereas the polymer derived material with ultrafine SIC particles exhibits smaller flaws. The total flaw densities which are given in Table 1 follow this trend too. 3.4. Hertzian indentation The detection of microcracks at polished surfaces with optical microscopy is very difficult. Thus, a Hertzian indentation technique was used to mechanically characterise flaw populations [ 14, 151.Figure 11 shows flaw densities as a function of flaw size in the form of a histogram for each material. The flaws can be interpreted as surface breaking cracks because of the stress pattern of the Hertzian indentation technique used. Again, a clear trend can be seen:

1

10

100

Polishing grade [pm] Fig 9. Normalised surface Young’s modulus as a function of polishing grade of 5 vol.% SiC/Al20, nanocomposites compared to monolithic alumina.

In our previous study of the bulk mechanical properties of nanocomposites we found no influence of SIC particle size [4, 81 and no significant increase in toughness over monolithic alumina. We proposed that any increase in strength was based on a reduction in size and density of processing flaws and this is in agreement with the reduced level of surface defects found in this study. The size and number of grain pull-outs, processing flaws and surface breaking cracks are significantly smaller in the nanocomposites than in monolithic alumina of similar grain size. The major role of the Sic particles in this case seems to be in modifying the processing conditions so as to minimise the defects introduced during powder handling prior to sintering. Any influence the Sic particles may have on the bulk mechanical properties is then masked by the strength still being limited by the size of the critical failure inducing defect. In contrast, wear, erosion and polishing are all phenomena requiring the nucleation of multiple fracture events and are less dependant on the size of the largest defects present, Thus any influence of nanocomposite microstructure masked in the bulk property measurements may be revealed here. From these results there appears to be a clear influence of SIC particle size on the erosion resistance of the nanocomposites. The flaw densities measured visually also show a trend with reduced defect densities and reduced maximum defect size with decreasing SIC particle size. The Hertzian indentation data are less clear but in a good agreement with the other results. Thus our data show convincing evidence for the nanocomposites showing a resistance to surface damage and erosive wear which is considerably greater than that of unreinforced alumina This is accompanied by a suppression of grain pull-out as a significant material removal mechanism. Our results do not help shed light on any of the proposed mechanisms of nanocomposite strengthening because, although we have characterised a lower defect density as a cause for composite strengthening we have not identified a mechanism for this increased damage resistance. Resistance to surface damage is expected to scale with fracture toughness and the toughness of these materials is not expected to increase to a sufficient extent to explain the observed damage resistance. Although the mechanisms for the high damage resistance of nanocomposites and their dependence

STERNITZKE et al.: SURFACE MECHANICAL PROPERTIES

0

20

40

60

3971

80

Flaw size [pm]

0

20

40

60

80

0

26

40

60

40

60

66

Flaw size [WI-I]

Flaw size [pm]

0

26

86

Flaw size [pm]

0

26

40

60

80

Flaw size [WI]

Fig. 10. Flaw size distributions including processing flaws and grain pullouts for monolithic alumina and different 5 vol.% SiC/A1203nanocomposites.

on Sic particle size are not yet fully understood, the results of this work possibly support a grain boundary strengthening mechanism. The radial stress distribution around SIC particles in an alumina matrix is compressive and independent of particle size. As shown previously, local compressive stresses around Sic particles which are close to a grain boundary can have a strengthening effect [7]. Table 1 indicates the number of Sic particles per unit volume assuming spherical particles for each of the nanocomposites. It is clear that the number of SiC particles greatly increases with decreasing SIC particle size for a constant volume

fraction. Furthermore, the SEM micrographs of the eroded surfaces in Fig. 7 demonstrate that the amount of transgranular failure increases with decreasing Sic particle size. This implies that the amount of transgranular failure arises from a grain boundary strengthening effect. The grain boundary strengthening effect, however, might be related to the number of SIC particles which would explain the effect of SIC particle size. It must be pointed out that further work is necessary to confirm the mechanisms proposed in this research for the high damage resistance of nanocomposites.

3972

STERNITZKE

rf it/.:

SURFACE

MECHANICAL

PROPERTIES

40 N” i 30 2

Alumina

c ‘5 20 5 0 jj

10

LL 0 0

2

4

6

8

10

Flaw size [cm]

40

1

30

c r 8

2o

UF15

UF45 E ‘f

20

4 Z ii

10

0 4

6

8

0

10

2

4

6

8

10

Flaw size [pm]

Flaw size [pm]

200

g

Sed. UF45

150

Polymer

g .E

i

100

50

0 I 2

4

6

8

10

0

IT

2

Flaw size [pm]

4

I I I1

I ‘I

“1,

6

8

10

Flaw size [pm]

Fig. 11. Surface flaw size distribution due to machining damage measured by Hertzian indentation 4. SUMMARY

AND CONCLUSIONS

The wet erosive wear tests have shown that nanocomposites exhibit a better wear resistance than monolithic alumina as reported in previous work. However, there is an influence of SIC particle size with the finest Sic particles giving the best wear resistance. Surface damage as investigated by acoustic microscopy for different surface treatments is also reduced in the case of nanocomposites even for a 12 pm polish. This is in good agreement

to the flaw density distributions measured by and Hertzian indentation optical microscopy fracture. However, the most interesting conclusion from this study is that the resistance to surface damage in the nanocomposites shows a clear but weak dependence on mean SIC particle size. Acknowledgements-The

authors

are grateful

to Dr S. G.

Roberts, Professors R. W. Davidge and F. L. Riley for their helpful discussion. This project was supported by the EU (BRITE/EURAM II; Human Capital and Mobility contract No. ERB BRE2 CT94 3093).

STERNITZKE et al.:

SURFACE MECHANICAL

REFERENCES K., J. Gram. Sot. Jpn, 1991, 99, 974. L. C., Harmer, M. P., Chan, H. M., Miller, G. A. and Cook, R. E., J. Am. Ceram. Sot., 1993, 76, 503. Davidge, R. W., Brook, R. J., Cambier, F., Poorteman. M., Leriche, A., O’Sullivan, D., Hampshire, S. and Kennedy, T., British Ceramic Transactions and Journal, in press, 1997. Carroll, L., Sternitzke, M. and Derby, B., Acta. Mater., 1996, 44, 4543. Hoffman, M. J., Sternitzke, M., Riidel, J. and Brook, R. J., Fracture Mechanics of Ceramics, ed. R. C. Bradt, D. P. H. Hasselman, D. Munz, M. Sakai and V. Y. Shevchenko, Vol. 12. Plenum Press, New York, 1996, p. 179. Levin, I., Kaplan, W. D., Brandon, D. G. and Layyous, A. A., J. Am. Ceram. Sot., 1995, 78, 254. Sternitzke, M., J. Eur. Ceram. Sot., 1997, 17, 1061. Sternitzke, M., Derby, B. and Brook, R. J., J. Am. Ceram. Sot., in press, 1997.

1. Niihara,

2. Zhao, J., Stearns,

3.

4. 5.

6. 7. 8.

PROPERTIES

3973

9. Davidge, R. W., Twigg, P. C. and Riley. F. L., J. Eur. Ceram. Sot., 1996, 16, 799. 10. Warren, P. D., J. Eur. Ceram. Sot., 1995, 15, 201. 11. Warren, P. D., Hills, D. A. and Roberts, S. G., J. Mater. Res., 1994, 9, 3194. 12. Su, B. and Sternitzke, M., Fourth Euro. Ceramics, ed. A. Bellosi, Vol. 4, 1995, p. 109. 13. Davidge, R. W. and Riley, F. L., Wear, 1995, 186-187, 45. N. C. and Briggs. 14. Sklar, Z., Mutti, P., Stoodley, G. A. D.. Advances in Acoustic Microscopy, ed. G. A. D. Briggs, Plenum Press, New York, 1995, p. 209. 15. Kushibiki, J.-l. and Chubachi, N., IEEE Transactions on Sonics and Ultrasonics, 1985, SU-32, 189. 16. Franco. A., Roberts, S. G. and Warren, P. D., Acta Mater., 1997, 45, 1009. 17. Borsa, C. E., Jiao, S., Todd, R. I. and Brook, R. J., Journal of Microscopy, 1995, 177, 305. 18. Chou, I. A., Chan, H. M. and Harmer, M. P.. J. Am. Ceram. Sot., 1996. 79, 2403.