- Email: [email protected]
Solid particle erosion of Si3N4 materials
115
Wear, 69 (1981) 115 - 129 0 Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
SOLID PARTICLE EROSION OF SisN4 MATERIALS
MARY ELLEN GULDEN Solar Turbines International, An Operating Group of International Pacific Highway, P. 0. Box 80966, San Diego, CA 92138 (U.S.A.) (Received
December
Harvester,
2200
8, 1980)
Summary Four commercially available Si3N4 materials were eroded with SIC and quartz particles over a wide range of particle sizes (8 - 10 pm) at subsonic velocities. The target materials encompassed a broad range of mechanical properties and microstructures. Erosion weight loss was measured and the erosion dependences on particle size and velocity were determined. Single impacts were examined to characterize the damage. All the targets eroded with Sic particles exhibited an elastic-plastic type of impact damage. This type of damage is characterized by a highly deformed impact crater with radial and lateral cracks propagating outward from the impact crater. The damage for quartz particle impact varied from an elastic-plastic type to minor intergranular chipping for the harder higher fracture toughness targets. The type of erosion damage was more dependent on relative target-particle properties than on particle size and velocity. The experimental erosion data are compared with the two models developed for elastic-plastic impact damage which are based on single impacts on isotropic targets under idealized conditions.
1. Introduction Considerable interest is being shown in the use of ceramics for high technology engineering components in such applications as gas turbine engines, bearings, heat exchangers and radomes. All these applications may involve impingement by solid particles. A knowledge of impact and erosion behavior is necessary before ceramics can be used with confidence in these systems. The materials discussed here either are in current use or are being considered for potential use in one or more of the above applications. Recent investigations have shown that a number of erosion mechanisms for ceramics can exist and that erosion and impact is a complex process [ 1 41. Essentially two types of models have been proposed for solid particle impact (single-particle impacts) and erosion (multiple-particle impacts) of brittle materials. The earlier models were based on elastic interaction between
116
target and particle and using these models it was predicted that material removal occurs by the intersection of ring cracks on the target surface. This process has been observed on several materials under static and low velocity impact conditions with relatively large spherical particles [ 3, 41. In a more recent analysis, static and dynamic plastic indentation, which is characterized by plastic deformation of the contact area between the particle and the target with radial cracks propagating outward from the contact zone and with subsurface lateral cracks propagating outward on planes nearly parallel to the surface, was treated. This type of damage, termed “elastic-plastic” damage, is observed for impact with angular particles of generally greater hardnesses than that of the target [ 1, 21. The models are based on single impacts and were developed for isotropic materials under idealized conditions. One objective of this investigation was to assess the validity of these models to predict erosion of engineering ceramics by natural dust environments. The experimental approach for this investigation was to perform single-impact and erosion (multiple-particle impacts) tests in a controlled manner to simulate a service environment in the subsonic velocity regime. This approach advances the understanding of erosion mechanisms of engineering materials as well as providing data of direct value in the application of these materials to engineering structures.
2. Experimental
procedure
2.1. Materials Four commercially available Si3N4 materials, which cover a wide range of structure and properties, were selected for target materials. The target materials and mechanical properties relevant to erosion are shown in Table 1. Four-point bend strength is not specifically related to erosion resistance but is included in the table to illustrate the large variation in strength encompassed by these targets. The properties of the impacting particles are also included in Table 1. The microstructure of the targets is shown in Fig. 1. Hot-pressed Si3N4 is fine grained and fully dense; it contains minor amounts of a glassy phase at the grain boundaries. The pressureless-sintered material is also fine grained but contains significant amounts of a glassy phase at the grain boundaries and isolated regions of low porosity and second-phase particles of silicon and molybdenum silicide. Both of the reaction-bonded materials are approximately. 75% dense. However, the porosity in the NC 350 product is relatively small (1 - 7 pm) and uniformly distributed. The KBI product is very inhomogeneous and contains non-uniform porosity as large as 50 pm, unreacted silicon regions as large as 60 pm and other secondphase particles. The particles selected for impact and erosion testing also provide a variation in properties. Sic is harder than all the targets, while quartz is softer than hot-pressed and pressureless-sintered SisN4 and has a hardness
117 TABLE 1 SiaNa target and particle properties Four-point bend strength
Elastic modulus VW
Fracture toughness (MPa m l/2)
Hardness WW
Hot-pressed SiaN4 (NC 132)
320
5
16
725
Pressureless-sintered SisN4
258
6
12
450
WW
(GTE 3502) Reaction-bonded (NC 350)
SiaN4
170
2.2
7.5
210
Reaction-bonded (KBI)
Si3 N4
147
2.2
5
145
95
0.7
6
Quartz Sic
(4
420
3
23
(d)
Fig. 1. Typical microstructures of SiaN4 targets: (a) hot-pressed SiaN4 (NC 132); (b) pressureless-sintered SisN4 (GTE 3502); (c) reaction-bonded SiaN4 (NC 360); (d) reaction-bonded SiaN4 (KBI).
118
equivalent to those of the two reaction-bonded materials. Quartz is the most erosive constituent of natural dusts [ 51. Six narrow size ranges between 10 and 385 pm for quartz particles and between 8 and 940 pm for Sic particles were used. For each particle size up to five velocities between 24 and 285 m s-l were evaluated. 2.2. Erosion testing Erosion tests were performed on a stationary target impacted by particles accelerated in an air stream. Particles are injected into the stream 3 m from the target to provide sufficient distance for acceleration. High pressure, filtered and chemically dried air is used for the particle carrier gas. The carrier air velocity is measured using standard Pitot tube techniques. The air velocity variation across the nozzle 0.95 cm in diameter is less than 5% and the velocity is varied between 30 and 343 m s-l to achieve the desired particle velocity. The particle velocity is measured using the rotating double-disc technique described in ref. 6. The comparison with calculated velocities based on the two-phase flow theory is good [?] . All erosion tests were performed at a 90” impingement angle at ambient temperature. Perpendicular impingement is at or near that for maximum erosion of brittle materials. The number of particles used per test was varied from a few particles (to examine single-particle impacts) to as many as 10’ particles for measurable weight loss over target area of 0.71 cm2. The particles are fed into the gas stream using a precision feeder at a sufficiently low concentration that particle interactions in the carrier gas stream or on the target surface are negligible. For the long-time large number of particle tests the specimens were weighed at specific intervals to assess any changes in erosion with the number of impacts. A detailed description of the erosion tests apparatus is given in ref. 7. 2.3. Analysis of results
In addition to weight loss measurements the eroded surfaces were examined using optical, replica transmission and scanning electron microscopies. The progression of impact or erosion events was monitored by examining the specimens after various number of impacts for several particle size-velocity combinations ranging from single-particle impacts to deeply eroded surfaces. The surfaces were also examined in cross section to assess the nature of subsurface damage. 3. Experimental results and discussion This section is separated into three parts, In Section 3.1, erosion weight loss as a function of particle size and velocity is discussed, and in Section 3.2 a discussion and examples of the type of impact damage which occurs are given. In Section 3.3 a comparison between erosion behavior in terms of impact models and material properties and structure is made.
119 TABLE 2 Exponential
function for particle size dependence on erosion of Si3 N4
quartz particle velocity
Sic particle velocity
80 m s-l
160 m s-l
250 m s-l
50 m s-l
100 m s-l
150 m s-’
Hot-pressed Si3N4 (NC 132)
3.1
3.1
3.2
4.1
4.0
4.1
Pressurelesssintered Si3N4 (GTE 3502)
-
3.9
3.9
4.0
3.9
4.0
Reactionbonded Si3 N4 (NC 350)
3.4
4.1
4.5
4.1
4.3
4.4
Reactionbonded Si3N4 (KBI)
3.9
4.0
4.2
3.9
3.9
3.9
3.1. Erosion weight loss and particle size and velocity dependences The erosion weight loss at 90” impingement was measured on the four targets for impact with both Sic and natural quartz (SiOz) particles. The erosion dependences on particle size and velocity were determined. The exponential functions are shown in Tables 2 and 3. For impact with Sic particles, erosion per impact of all targets was dependent on the fourth power of the particle radius for all velocities. The radius exponent for quartz particle impact varied between 3 for hot-pressed Si3N4 to approximately 4 for the other targets. The power function for velocity dependence on erosion was not so consistent and varied from 1 for hot-pressed Si3N4 impacted with all particle sizes of quartz to as high as 6 for NC 350 reaction-bonded SisN4 impacted with 385 pm quartz, although the most‘common value was between 3 and 4. Except for hot-pressed Si3N4 impacted with quartz, there was no consistent variation in the velocity exponent with target material, particle size or particle type. To compare erosion weight loss of the four targets the data were plotted as the erosion volume loss per impact against R’, the fourth power of the particle radius, multiplied by V3, the velocity cubed; this is shown in Figs. 2 and 3. As can be seen in Fig. 2, for impact with Sic particles, the data fit reasonably well on two parallel lines of slope unity for all particle sizes and velocities, suggesting that the same mechanism is operative for these ranges of particle size and velocities. The erosion losses for the two reaction-bonded targets are indistinguishable, as are those for the hot-pressed and pressureless-
l-4
0-J
ti
N’
I
t-4
t-l
ti
ti
0
co
m
&
Ri
I
I
I
“.
o-3
u?
d
Ci
In
C4
0)
m
ei
ci
00
0,
4
N’
121
I
103’ 10’7
10.19
I 10’9
I
I
1
10.1’
lo-9
10.7
R4 V3, m7 s.~
Fig. 2. Erosion of four S&N4 materials impacted with Sic particles of various sizes: 0, 940 pm; A, 560 pm; q, 305 pm; 0,102 pm; V, 50 pm; 0, 8 Mm.
sintered targets; they are separated by approximately one order of magnitude of volume loss per impact for the same particle size-velocity test conditions. There is less difference between the targets for impact with 8 I.tm particles. This is thought to be due to the embedding of the small particles in the pores of the reaction-bonded materials which, in effect, reduces the measured weight loss. (Examination of the eroded surfaces does in fact reveal embedding of 8 E.tmparticles, which does not occur significantly with the larger particles.) Erosion for quartz particle impact versus R4V3 is shown in Fig. 3. The data for the reaction-bonded targets fit a straight-line relation of slope unity (on a log-log basis) fairly well. However, there is a separation in the volume losses of approximately one order of magnitude between the two targets for the same impact conditions. The erosion data for hot-pressed and pressureless-sintered Si3N4 do not follow a straight-line relationship; this is due primarily to the lower velocity exponent dependence on erosion (Table 3) for these targets. Again, for a given particle size-velocity test condition the erosion weight loss was the same for hot-pressed and pressureless-sintered Si3N4. At higher velocities there is a difference of several orders of magnitude in erosion between reaction-bonded and sintered (both pressureless-sintered and hot-pressed) Si3N4. The data suggest that there is a change in erosion mechanism for
t 6
10'5
10'4
I lo.13
loR4 V3,m7
I
I lo."
lo-
10-e
lo.8
s-3
Fig. 3. Erosion of four S&N* materials impacted with quartz particles of various sizes: 0,386 pm; A, 273 pm; O, 115 pm; 0,64 I.trn; V, 49 ,f.im;0, 10 pm.
impact with quartz particles on hot-pressed and pressureless-sintered SisN,. The data further suggest that the relative target and/or particle properties are more important with regard to controlling the erosion mechanism than the variations in particle size and velocity for the test conditions investigated. 3.2. Characterization of erosion damage Both heavily eroded surfaces and single impacts were examined to characterize the damage and material removal processes. All targets impacted with SiC particles exhibited an elastic-plastic type of impact damage [l] . Typical examples for each target are shown in Fig. 4. Although all the targets impacted with SIC particles exhibited elastic-plastic impact, there was a minor variation in the damage. The details of impact damage are shown in Fig. 5. Figure 4(a) shows single-impact damage on hot-pressed Si3N4. The radial and lateral cracks are long and tight. None of the laterally cracked material in this example had been removed. The secondary cracks, both radial and lateral, on the pressureless-sintered target are more open and intergranular or through the glassy grain boundary phase (Figs. 4(b) and 5(b)). The radial cracks in the hot-pressed Si3N4 were transgranular, but the lateral cracks were intergranular (Fig. 5(b)). Cracking in NC 350 reaction-bonded
(4
(b)
(cl
(4
Fig. 4. Typical impact damage produced by 300 I.trnSic particles: (a) hot-pressed Si3N4 (NC 132) (velocity, 176 m s-l); (b) pressureless-sintered Si3N4 (GTE 3502) (velocity, 176 m s-l); (c) reaction-bonded SiaN4 (NC 350) (velocity, 79 m s-l); (d) reactionbonded Si3N4 (KBI) (velocity, 79 m s-l).
(4 Fig. 5. Details of elastic-plastic impact damage typical of hot-pressed and pressurelesssintered Si3N4 : (a) apparent plastic flow in impact crater; (b) intergranular lateral cracking.
04
(b)
Fig. 6. Damage produced by quartz particle impact: (a) shallow inter~anular chipping characteristic of hot-pressed and pressureless-sintered S&N4 ; (b) circumferential cracking typical of damage on reaction-bonded Si3N4 (NC 350).
Si3N4 was wide and followed the porosity (Fig. 4(c)). Crack branching was observed. The impact area in KBI reaction-bonded SisN, (Fig. 4(d)) was relatively large and secondary cracks, both lateral and radial, were minimized (Fig. 4(d)). It was difficult to differentiate impact areas from the large pores inherent in this target. The appearance of the actual impact areas for all the targets impacted with Sic was not one of brittle fracture, but rather of plastic flow. No cleavage markings or intergranular facets were observed. A typical example of the bottom of an impact crater is shown in Fig. 5(a). The impact damage produced by quartz particles was quite different from that produced by Sic particles. The damage produced on hot-pressed and pressureless-sir&red Si3N4 was characterized by shallow intergranular chipping, as shown in Fig. 6, and no cracking was observed away from the impact area. The NC 350 reaction-bonded Si3N4 impacted surfaces suggest that a transition in type of impact damage may have occurred with quartz particles. The single-impact damage varied from elastic-plastic type to shallow impact craters bounded by concentric ring cracks (Fig. 6(b)). The concentric ring cracks are considered to be due to the elastic interaction between the particle and target. On the KBI reaction-bonded Si3N4 the quartz particle impacts were not dist~~ishable from the porosity. 3.3. Correlation of erosion data with impact models In view of the fact that the targets exhibited elastic-plastic impact when eroded with Sic particles, it is of interest to plot the data in terms of the impact models which have recently been developed for this type of damage. Two expressions have been derived, eqns. (1) and (2). Details of the derivations are given in the references. Generally, the equations relate erosion to the depth of damage and area of lateral cracking. The equations are as follows:
125
13/12R PP
11/3v P
19/6K,-
4’3H-
P
1’4
(1)
after Evans [ 81 and
Va
(PpPtpdt)2’3 {(PPc(P) l’2 + (ptpt)l’2}8’3
ll’3up22’9K
p,ll’19R
P
-4’3Hl’g c
(2)
after Ruff and Wiederhorn [9]. The major difference8 between the two equations are in the particle density, particle velocity and target hardness dependences. These equations are based on single impacts and were developed for isotropic materials under idealized conditions. The validity of the equations can be assessed by plotting the erosion weight loss normalized by the appropriate particle size and velocity dependences against the functions of particle and target properties predicted by the models. These plots are shown in Figs. 7 and 8. The data points are average values of all data for a single target calculated on a logarithmic basis. The error bar8 correspond to plus or minus one standard deviation, also calculated on a logarithmic basis. A line with a slope of unity is drawn through the data points for those targets which exhibited elastic-plastic erosion damage. In both cases the actual slope through the data points is approximately 1.25. The data for targets which did not exhibit elastic-plastic impact when impacted with quartz particles (hot-pressed, pressureless-sir&red and NC 350 reaction-bonded Si3N4) are in the correct order as predicted by both relationships but fall well below the data line for elastic-plastic impact. The most remarkable result is that the experimental erosion data fit both models equally well although there are two major differences in the equations: the dependences on particle velocity and target hardness The power function of velocity from eqn. (1) is 3.17, which is approximately the experimental value. The power function of velocity in eqn. (2) is lower at 2.44 and is also lower than observed experimentally in this work for elastic-plastic impact conditions. A lower velocity exponent has been observed in other work [lo]. The hardness exponents are -l/4 for eqn. (1) and +1/9 for eqn. (2). Although the targets exhibit a wide range of hardne88e8, the small hardness dependence is not sufficient to separate the two equations. These results indicate that further experimental work is necessary before the model which most accurately describes elastic-plastic erosion can be ascertained and that both model8 provide a reasonable fit to the experimental erosion data. The result8 further suggest that, for the targets investigated, microstructural variation such as grain size and amount and type of second phase are sufficiently accounted for in the fracture toughness and hardness values and do not need to be considered separately in the models.
126
_I
I t
=:
1
3
127
I
A Y
I t
s -i
128
4. Discussion Four commercially available Si3N4 materials were eroded with Sic and quartz particles. Particle sizes ranged between 8 and 1000 pm, and testing was performed in the subsonic velocity regime (less than 300 m s-l). For Sic particle impact the erosion weight loss was dependent on the fourth power of the particle radius for all targets and velocities. However, the velocity dependence on erosion varied, apparently randomly, between approximately the third and the fifth power, i.e. there was no consistent variation in velocity dependence with particle size or target material type. The exponential radius dependence for quartz particle impact varied between 3 for hot-pressed SisN4 to 4 for the reaction-sintered targets, and the exponential velocity dependence varied from 1 for hot-pressed Si3N4 to between 2 and 6 for the other targets. For the stronger targets there was a difference of as much as two orders of magnitude in erosion between Sic and quartz particle impact. Examination of the surfaces revealed that all targets impacted with Sic particles exhibited elastic-plastic impact damage characterized by a highly deformed impact crater and radial and lateral cracks propagating outward from the crater. Damage characteristics produced by quartz particles varied with target. Shallow intergranular chipping with no secondary cracks was typical for hot-pressed and pressureless-sintered Si3N4. The damage on reaction-bonded Si3N4 impacted with quartz particles was characterized by circumferential cracks and short radial cracks with relatively minor damage in the contact area. The results indicate that the type of erosion damage produced is more dependent on relative target-particle properties than on particle size and velocity for a given target-particle combination. The experimental erosion data were plotted in terms of two models developed for elastic-plastic impact. In the models, erosion is related to the depth of damage and the area of lateral cracking and is dependent on target and particle density and shear modulus, on particle size and velocity and on target fracture toughness and hardness. The dependences on particle density and velocity and target hardness vary. The erosion data for SIC particle impact fit-both models reasonably well. An interesting result is that the large variation in microstructure exhibited by these targets is sufficiently accounted for in fracture toughness and hardness values and does not have to be considered uniquely in the expressions for elastic-plastic impact.
Acknowledgment This work was performed in the Solar Research Laboratories under sponsorship of the Office of Naval Research under Contract N0014-73-C0401.
129 Nomenclature H
KC
RP “P
V ClP
I.rt PP
Pt
target hardness target fracture toughness particle radius particle velocity erosion volume loss per impact particle shear modulus target shear modulus particle density target density
References 1 A. G. Evans, M. E. Gulden and M. Rosenblatt, Froc. R. Sot. London, Ser. A, 361 (1978) 343. 2 M. E. Gulden, Solid particle erosion of high technology ceramics, Erosion: prevention and useful applications, ASTM Spec. Tech. Publ. 664, 1979, p. 101. 3 W. F. Adler and G. T. Sha, Analytical modeling of subsonic particle erosion, AFML Tech. Rep. 72-144, 1972 (U.S. Air Force Materials Laboratory, Wright-Patterson Air Force Base, OH). 4 H. L. Oh, K. 0. L. Oh, S. Vaidyanathan and I. Finnie, On the shaping of brittle solids by erosion and ultrasonic cutting, NBS Spec. PubI. 348, 1972, p. 119 (National Bureau of Standards). 5 C. E. Smeltzer, M. E. Gulden, S. .S. McElmury and W. A. Compton, Mechanisms of sand and dust erosion in gas turbine engines, USAA VLABS Tech. Rep. 70-36, 1970 (U.S. Army Aviation Material Laboratories, Fort Eustis, VA). 6 A. W. Ruff and L. K. Ives, Wear, 35 (1975) 195. 7 M. E. Gulden and A. G. Metcalfe, Study of erosion mechanisms of engineering ceramics, STI Sol. Rep. RDR 1778-4, 1976 (Solar Turbines International, San Diego, CA) (Office of Naval Research Contract NOOOl.4-73-C-0401). 8 A. G. Evans, in C. M. Preece (ea.), Treatise on Material Science and Technology, Vol. 16, Erosion, Academic, New York, 1979, p, 1. 9 A. W. Ruff and S. M. Wiederhorn, in C. M. Preece (ed.), Treatise on Materials Science and Technology, Vol. 16, Erosion, Academic, New York, 1979, p. 69. 10 S. M. Wiederhorn, personal communication, 1978.
Wear, 69 (1981) 115 - 129 0 Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
SOLID PARTICLE EROSION OF SisN4 MATERIALS
MARY ELLEN GULDEN Solar Turbines International, An Operating Group of International Pacific Highway, P. 0. Box 80966, San Diego, CA 92138 (U.S.A.) (Received
December
Harvester,
2200
8, 1980)
Summary Four commercially available Si3N4 materials were eroded with SIC and quartz particles over a wide range of particle sizes (8 - 10 pm) at subsonic velocities. The target materials encompassed a broad range of mechanical properties and microstructures. Erosion weight loss was measured and the erosion dependences on particle size and velocity were determined. Single impacts were examined to characterize the damage. All the targets eroded with Sic particles exhibited an elastic-plastic type of impact damage. This type of damage is characterized by a highly deformed impact crater with radial and lateral cracks propagating outward from the impact crater. The damage for quartz particle impact varied from an elastic-plastic type to minor intergranular chipping for the harder higher fracture toughness targets. The type of erosion damage was more dependent on relative target-particle properties than on particle size and velocity. The experimental erosion data are compared with the two models developed for elastic-plastic impact damage which are based on single impacts on isotropic targets under idealized conditions.
1. Introduction Considerable interest is being shown in the use of ceramics for high technology engineering components in such applications as gas turbine engines, bearings, heat exchangers and radomes. All these applications may involve impingement by solid particles. A knowledge of impact and erosion behavior is necessary before ceramics can be used with confidence in these systems. The materials discussed here either are in current use or are being considered for potential use in one or more of the above applications. Recent investigations have shown that a number of erosion mechanisms for ceramics can exist and that erosion and impact is a complex process [ 1 41. Essentially two types of models have been proposed for solid particle impact (single-particle impacts) and erosion (multiple-particle impacts) of brittle materials. The earlier models were based on elastic interaction between
116
target and particle and using these models it was predicted that material removal occurs by the intersection of ring cracks on the target surface. This process has been observed on several materials under static and low velocity impact conditions with relatively large spherical particles [ 3, 41. In a more recent analysis, static and dynamic plastic indentation, which is characterized by plastic deformation of the contact area between the particle and the target with radial cracks propagating outward from the contact zone and with subsurface lateral cracks propagating outward on planes nearly parallel to the surface, was treated. This type of damage, termed “elastic-plastic” damage, is observed for impact with angular particles of generally greater hardnesses than that of the target [ 1, 21. The models are based on single impacts and were developed for isotropic materials under idealized conditions. One objective of this investigation was to assess the validity of these models to predict erosion of engineering ceramics by natural dust environments. The experimental approach for this investigation was to perform single-impact and erosion (multiple-particle impacts) tests in a controlled manner to simulate a service environment in the subsonic velocity regime. This approach advances the understanding of erosion mechanisms of engineering materials as well as providing data of direct value in the application of these materials to engineering structures.
2. Experimental
procedure
2.1. Materials Four commercially available Si3N4 materials, which cover a wide range of structure and properties, were selected for target materials. The target materials and mechanical properties relevant to erosion are shown in Table 1. Four-point bend strength is not specifically related to erosion resistance but is included in the table to illustrate the large variation in strength encompassed by these targets. The properties of the impacting particles are also included in Table 1. The microstructure of the targets is shown in Fig. 1. Hot-pressed Si3N4 is fine grained and fully dense; it contains minor amounts of a glassy phase at the grain boundaries. The pressureless-sintered material is also fine grained but contains significant amounts of a glassy phase at the grain boundaries and isolated regions of low porosity and second-phase particles of silicon and molybdenum silicide. Both of the reaction-bonded materials are approximately. 75% dense. However, the porosity in the NC 350 product is relatively small (1 - 7 pm) and uniformly distributed. The KBI product is very inhomogeneous and contains non-uniform porosity as large as 50 pm, unreacted silicon regions as large as 60 pm and other secondphase particles. The particles selected for impact and erosion testing also provide a variation in properties. Sic is harder than all the targets, while quartz is softer than hot-pressed and pressureless-sintered SisN4 and has a hardness
117 TABLE 1 SiaNa target and particle properties Four-point bend strength
Elastic modulus VW
Fracture toughness (MPa m l/2)
Hardness WW
Hot-pressed SiaN4 (NC 132)
320
5
16
725
Pressureless-sintered SisN4
258
6
12
450
WW
(GTE 3502) Reaction-bonded (NC 350)
SiaN4
170
2.2
7.5
210
Reaction-bonded (KBI)
Si3 N4
147
2.2
5
145
95
0.7
6
Quartz Sic
(4
420
3
23
(d)
Fig. 1. Typical microstructures of SiaN4 targets: (a) hot-pressed SiaN4 (NC 132); (b) pressureless-sintered SisN4 (GTE 3502); (c) reaction-bonded SiaN4 (NC 360); (d) reaction-bonded SiaN4 (KBI).
118
equivalent to those of the two reaction-bonded materials. Quartz is the most erosive constituent of natural dusts [ 51. Six narrow size ranges between 10 and 385 pm for quartz particles and between 8 and 940 pm for Sic particles were used. For each particle size up to five velocities between 24 and 285 m s-l were evaluated. 2.2. Erosion testing Erosion tests were performed on a stationary target impacted by particles accelerated in an air stream. Particles are injected into the stream 3 m from the target to provide sufficient distance for acceleration. High pressure, filtered and chemically dried air is used for the particle carrier gas. The carrier air velocity is measured using standard Pitot tube techniques. The air velocity variation across the nozzle 0.95 cm in diameter is less than 5% and the velocity is varied between 30 and 343 m s-l to achieve the desired particle velocity. The particle velocity is measured using the rotating double-disc technique described in ref. 6. The comparison with calculated velocities based on the two-phase flow theory is good [?] . All erosion tests were performed at a 90” impingement angle at ambient temperature. Perpendicular impingement is at or near that for maximum erosion of brittle materials. The number of particles used per test was varied from a few particles (to examine single-particle impacts) to as many as 10’ particles for measurable weight loss over target area of 0.71 cm2. The particles are fed into the gas stream using a precision feeder at a sufficiently low concentration that particle interactions in the carrier gas stream or on the target surface are negligible. For the long-time large number of particle tests the specimens were weighed at specific intervals to assess any changes in erosion with the number of impacts. A detailed description of the erosion tests apparatus is given in ref. 7. 2.3. Analysis of results
In addition to weight loss measurements the eroded surfaces were examined using optical, replica transmission and scanning electron microscopies. The progression of impact or erosion events was monitored by examining the specimens after various number of impacts for several particle size-velocity combinations ranging from single-particle impacts to deeply eroded surfaces. The surfaces were also examined in cross section to assess the nature of subsurface damage. 3. Experimental results and discussion This section is separated into three parts, In Section 3.1, erosion weight loss as a function of particle size and velocity is discussed, and in Section 3.2 a discussion and examples of the type of impact damage which occurs are given. In Section 3.3 a comparison between erosion behavior in terms of impact models and material properties and structure is made.
119 TABLE 2 Exponential
function for particle size dependence on erosion of Si3 N4
quartz particle velocity
Sic particle velocity
80 m s-l
160 m s-l
250 m s-l
50 m s-l
100 m s-l
150 m s-’
Hot-pressed Si3N4 (NC 132)
3.1
3.1
3.2
4.1
4.0
4.1
Pressurelesssintered Si3N4 (GTE 3502)
-
3.9
3.9
4.0
3.9
4.0
Reactionbonded Si3 N4 (NC 350)
3.4
4.1
4.5
4.1
4.3
4.4
Reactionbonded Si3N4 (KBI)
3.9
4.0
4.2
3.9
3.9
3.9
3.1. Erosion weight loss and particle size and velocity dependences The erosion weight loss at 90” impingement was measured on the four targets for impact with both Sic and natural quartz (SiOz) particles. The erosion dependences on particle size and velocity were determined. The exponential functions are shown in Tables 2 and 3. For impact with Sic particles, erosion per impact of all targets was dependent on the fourth power of the particle radius for all velocities. The radius exponent for quartz particle impact varied between 3 for hot-pressed Si3N4 to approximately 4 for the other targets. The power function for velocity dependence on erosion was not so consistent and varied from 1 for hot-pressed Si3N4 impacted with all particle sizes of quartz to as high as 6 for NC 350 reaction-bonded SisN4 impacted with 385 pm quartz, although the most‘common value was between 3 and 4. Except for hot-pressed Si3N4 impacted with quartz, there was no consistent variation in the velocity exponent with target material, particle size or particle type. To compare erosion weight loss of the four targets the data were plotted as the erosion volume loss per impact against R’, the fourth power of the particle radius, multiplied by V3, the velocity cubed; this is shown in Figs. 2 and 3. As can be seen in Fig. 2, for impact with Sic particles, the data fit reasonably well on two parallel lines of slope unity for all particle sizes and velocities, suggesting that the same mechanism is operative for these ranges of particle size and velocities. The erosion losses for the two reaction-bonded targets are indistinguishable, as are those for the hot-pressed and pressureless-
l-4
0-J
ti
N’
I
t-4
t-l
ti
ti
0
co
m
&
Ri
I
I
I
“.
o-3
u?
d
Ci
In
C4
0)
m
ei
ci
00
0,
4
N’
121
I
103’ 10’7
10.19
I 10’9
I
I
1
10.1’
lo-9
10.7
R4 V3, m7 s.~
Fig. 2. Erosion of four S&N4 materials impacted with Sic particles of various sizes: 0, 940 pm; A, 560 pm; q, 305 pm; 0,102 pm; V, 50 pm; 0, 8 Mm.
sintered targets; they are separated by approximately one order of magnitude of volume loss per impact for the same particle size-velocity test conditions. There is less difference between the targets for impact with 8 I.tm particles. This is thought to be due to the embedding of the small particles in the pores of the reaction-bonded materials which, in effect, reduces the measured weight loss. (Examination of the eroded surfaces does in fact reveal embedding of 8 E.tmparticles, which does not occur significantly with the larger particles.) Erosion for quartz particle impact versus R4V3 is shown in Fig. 3. The data for the reaction-bonded targets fit a straight-line relation of slope unity (on a log-log basis) fairly well. However, there is a separation in the volume losses of approximately one order of magnitude between the two targets for the same impact conditions. The erosion data for hot-pressed and pressureless-sintered Si3N4 do not follow a straight-line relationship; this is due primarily to the lower velocity exponent dependence on erosion (Table 3) for these targets. Again, for a given particle size-velocity test condition the erosion weight loss was the same for hot-pressed and pressureless-sintered Si3N4. At higher velocities there is a difference of several orders of magnitude in erosion between reaction-bonded and sintered (both pressureless-sintered and hot-pressed) Si3N4. The data suggest that there is a change in erosion mechanism for
t 6
10'5
10'4
I lo.13
loR4 V3,m7
I
I lo."
lo-
10-e
lo.8
s-3
Fig. 3. Erosion of four S&N* materials impacted with quartz particles of various sizes: 0,386 pm; A, 273 pm; O, 115 pm; 0,64 I.trn; V, 49 ,f.im;0, 10 pm.
impact with quartz particles on hot-pressed and pressureless-sintered SisN,. The data further suggest that the relative target and/or particle properties are more important with regard to controlling the erosion mechanism than the variations in particle size and velocity for the test conditions investigated. 3.2. Characterization of erosion damage Both heavily eroded surfaces and single impacts were examined to characterize the damage and material removal processes. All targets impacted with SiC particles exhibited an elastic-plastic type of impact damage [l] . Typical examples for each target are shown in Fig. 4. Although all the targets impacted with SIC particles exhibited elastic-plastic impact, there was a minor variation in the damage. The details of impact damage are shown in Fig. 5. Figure 4(a) shows single-impact damage on hot-pressed Si3N4. The radial and lateral cracks are long and tight. None of the laterally cracked material in this example had been removed. The secondary cracks, both radial and lateral, on the pressureless-sintered target are more open and intergranular or through the glassy grain boundary phase (Figs. 4(b) and 5(b)). The radial cracks in the hot-pressed Si3N4 were transgranular, but the lateral cracks were intergranular (Fig. 5(b)). Cracking in NC 350 reaction-bonded
(4
(b)
(cl
(4
Fig. 4. Typical impact damage produced by 300 I.trnSic particles: (a) hot-pressed Si3N4 (NC 132) (velocity, 176 m s-l); (b) pressureless-sintered Si3N4 (GTE 3502) (velocity, 176 m s-l); (c) reaction-bonded SiaN4 (NC 350) (velocity, 79 m s-l); (d) reactionbonded Si3N4 (KBI) (velocity, 79 m s-l).
(4 Fig. 5. Details of elastic-plastic impact damage typical of hot-pressed and pressurelesssintered Si3N4 : (a) apparent plastic flow in impact crater; (b) intergranular lateral cracking.
04
(b)
Fig. 6. Damage produced by quartz particle impact: (a) shallow inter~anular chipping characteristic of hot-pressed and pressureless-sintered S&N4 ; (b) circumferential cracking typical of damage on reaction-bonded Si3N4 (NC 350).
Si3N4 was wide and followed the porosity (Fig. 4(c)). Crack branching was observed. The impact area in KBI reaction-bonded SisN, (Fig. 4(d)) was relatively large and secondary cracks, both lateral and radial, were minimized (Fig. 4(d)). It was difficult to differentiate impact areas from the large pores inherent in this target. The appearance of the actual impact areas for all the targets impacted with Sic was not one of brittle fracture, but rather of plastic flow. No cleavage markings or intergranular facets were observed. A typical example of the bottom of an impact crater is shown in Fig. 5(a). The impact damage produced by quartz particles was quite different from that produced by Sic particles. The damage produced on hot-pressed and pressureless-sir&red Si3N4 was characterized by shallow intergranular chipping, as shown in Fig. 6, and no cracking was observed away from the impact area. The NC 350 reaction-bonded Si3N4 impacted surfaces suggest that a transition in type of impact damage may have occurred with quartz particles. The single-impact damage varied from elastic-plastic type to shallow impact craters bounded by concentric ring cracks (Fig. 6(b)). The concentric ring cracks are considered to be due to the elastic interaction between the particle and target. On the KBI reaction-bonded Si3N4 the quartz particle impacts were not dist~~ishable from the porosity. 3.3. Correlation of erosion data with impact models In view of the fact that the targets exhibited elastic-plastic impact when eroded with Sic particles, it is of interest to plot the data in terms of the impact models which have recently been developed for this type of damage. Two expressions have been derived, eqns. (1) and (2). Details of the derivations are given in the references. Generally, the equations relate erosion to the depth of damage and area of lateral cracking. The equations are as follows:
125
13/12R PP
11/3v P
19/6K,-
4’3H-
P
1’4
(1)
after Evans [ 81 and
Va
(PpPtpdt)2’3 {(PPc(P) l’2 + (ptpt)l’2}8’3
ll’3up22’9K
p,ll’19R
P
-4’3Hl’g c
(2)
after Ruff and Wiederhorn [9]. The major difference8 between the two equations are in the particle density, particle velocity and target hardness dependences. These equations are based on single impacts and were developed for isotropic materials under idealized conditions. The validity of the equations can be assessed by plotting the erosion weight loss normalized by the appropriate particle size and velocity dependences against the functions of particle and target properties predicted by the models. These plots are shown in Figs. 7 and 8. The data points are average values of all data for a single target calculated on a logarithmic basis. The error bar8 correspond to plus or minus one standard deviation, also calculated on a logarithmic basis. A line with a slope of unity is drawn through the data points for those targets which exhibited elastic-plastic erosion damage. In both cases the actual slope through the data points is approximately 1.25. The data for targets which did not exhibit elastic-plastic impact when impacted with quartz particles (hot-pressed, pressureless-sir&red and NC 350 reaction-bonded Si3N4) are in the correct order as predicted by both relationships but fall well below the data line for elastic-plastic impact. The most remarkable result is that the experimental erosion data fit both models equally well although there are two major differences in the equations: the dependences on particle velocity and target hardness The power function of velocity from eqn. (1) is 3.17, which is approximately the experimental value. The power function of velocity in eqn. (2) is lower at 2.44 and is also lower than observed experimentally in this work for elastic-plastic impact conditions. A lower velocity exponent has been observed in other work [lo]. The hardness exponents are -l/4 for eqn. (1) and +1/9 for eqn. (2). Although the targets exhibit a wide range of hardne88e8, the small hardness dependence is not sufficient to separate the two equations. These results indicate that further experimental work is necessary before the model which most accurately describes elastic-plastic erosion can be ascertained and that both model8 provide a reasonable fit to the experimental erosion data. The result8 further suggest that, for the targets investigated, microstructural variation such as grain size and amount and type of second phase are sufficiently accounted for in the fracture toughness and hardness values and do not need to be considered separately in the models.
126
_I
I t
=:
1
3
127
I
A Y
I t
s -i
128
4. Discussion Four commercially available Si3N4 materials were eroded with Sic and quartz particles. Particle sizes ranged between 8 and 1000 pm, and testing was performed in the subsonic velocity regime (less than 300 m s-l). For Sic particle impact the erosion weight loss was dependent on the fourth power of the particle radius for all targets and velocities. However, the velocity dependence on erosion varied, apparently randomly, between approximately the third and the fifth power, i.e. there was no consistent variation in velocity dependence with particle size or target material type. The exponential radius dependence for quartz particle impact varied between 3 for hot-pressed SisN4 to 4 for the reaction-sintered targets, and the exponential velocity dependence varied from 1 for hot-pressed Si3N4 to between 2 and 6 for the other targets. For the stronger targets there was a difference of as much as two orders of magnitude in erosion between Sic and quartz particle impact. Examination of the surfaces revealed that all targets impacted with Sic particles exhibited elastic-plastic impact damage characterized by a highly deformed impact crater and radial and lateral cracks propagating outward from the crater. Damage characteristics produced by quartz particles varied with target. Shallow intergranular chipping with no secondary cracks was typical for hot-pressed and pressureless-sintered Si3N4. The damage on reaction-bonded Si3N4 impacted with quartz particles was characterized by circumferential cracks and short radial cracks with relatively minor damage in the contact area. The results indicate that the type of erosion damage produced is more dependent on relative target-particle properties than on particle size and velocity for a given target-particle combination. The experimental erosion data were plotted in terms of two models developed for elastic-plastic impact. In the models, erosion is related to the depth of damage and the area of lateral cracking and is dependent on target and particle density and shear modulus, on particle size and velocity and on target fracture toughness and hardness. The dependences on particle density and velocity and target hardness vary. The erosion data for SIC particle impact fit-both models reasonably well. An interesting result is that the large variation in microstructure exhibited by these targets is sufficiently accounted for in fracture toughness and hardness values and does not have to be considered uniquely in the expressions for elastic-plastic impact.
Acknowledgment This work was performed in the Solar Research Laboratories under sponsorship of the Office of Naval Research under Contract N0014-73-C0401.
129 Nomenclature H
KC
RP “P
V ClP
I.rt PP
Pt
target hardness target fracture toughness particle radius particle velocity erosion volume loss per impact particle shear modulus target shear modulus particle density target density
References 1 A. G. Evans, M. E. Gulden and M. Rosenblatt, Froc. R. Sot. London, Ser. A, 361 (1978) 343. 2 M. E. Gulden, Solid particle erosion of high technology ceramics, Erosion: prevention and useful applications, ASTM Spec. Tech. Publ. 664, 1979, p. 101. 3 W. F. Adler and G. T. Sha, Analytical modeling of subsonic particle erosion, AFML Tech. Rep. 72-144, 1972 (U.S. Air Force Materials Laboratory, Wright-Patterson Air Force Base, OH). 4 H. L. Oh, K. 0. L. Oh, S. Vaidyanathan and I. Finnie, On the shaping of brittle solids by erosion and ultrasonic cutting, NBS Spec. PubI. 348, 1972, p. 119 (National Bureau of Standards). 5 C. E. Smeltzer, M. E. Gulden, S. .S. McElmury and W. A. Compton, Mechanisms of sand and dust erosion in gas turbine engines, USAA VLABS Tech. Rep. 70-36, 1970 (U.S. Army Aviation Material Laboratories, Fort Eustis, VA). 6 A. W. Ruff and L. K. Ives, Wear, 35 (1975) 195. 7 M. E. Gulden and A. G. Metcalfe, Study of erosion mechanisms of engineering ceramics, STI Sol. Rep. RDR 1778-4, 1976 (Solar Turbines International, San Diego, CA) (Office of Naval Research Contract NOOOl.4-73-C-0401). 8 A. G. Evans, in C. M. Preece (ea.), Treatise on Material Science and Technology, Vol. 16, Erosion, Academic, New York, 1979, p, 1. 9 A. W. Ruff and S. M. Wiederhorn, in C. M. Preece (ed.), Treatise on Materials Science and Technology, Vol. 16, Erosion, Academic, New York, 1979, p. 69. 10 S. M. Wiederhorn, personal communication, 1978.