Single solid particle impact erosion damage on polypropylene

Wear,

100

(1984)

263

263

- 280

SINGLE SOLID PARTICLE POLYPROPYLENE S. M. WALLEY,

J. E. FIELD

IMPACT EROSION

DAMAGE

ON

and P. YENNADHIOU

Physics and Chemistry of Solids, Covendish Laboratory, ingley Road, Cambridge CB3 OHE (Gt. Britain)

University

of Cambridge,

Mad-

Summary The literature on the erosion of polymers is reviewed. Single-particle impacts by spheres of 4 mm diameter have been carried out at various angles and speeds on polypropylene. The crater types are classified and plotted on a “deformation map”. High speed photography was used to record the impacts and any material loss. Both ductile (the drawing out of filaments) and brittle (the fracturing of blocks of polymer) erosion processes were observed. The surface finish of the specimens is an important variable in the latter mechanism.

1. Introduction Compared with metals and ceramics, little work has been published on solid particle erosion damage to bulk organic polymers [ 11. Tilly [ 2, 31 carried out a multiparticle study using quartz sand (60 - 125 pm) propelled by an air-blast rig. The bulk polymers he studied include Nylon 66 (filled and unfilled with carbon fibre) and polypropylene (PP). He found that the angular dependence of erosion in these plastics was similar to that for ductile metals, peaking at around 30”. At this angle, nylon suffered about three times the volumetric erosion E, per unit mass of erodent that PP did. Both polymers exhibited long incubation times during which erodent deposited on the surface, producing a mass gain. PP was never observed to start losing mass at 90” (normal impact) in his experiments, although his published graphs indicate that, had erosion been continued at this angle, it is likely that mass loss would have occurred. Neilson and Gilchrist [4] found the angular dependence of erosion in polymethylmethacrylate (PMMA) to be intermediate in character between that of a ductile material and that of a brittle material. The specific erosion parameter E (mass eroded per mass of erodent) peaked at about 40”, but it declined only slightly for angles of impact greater than this. 0043-1648/%4/$3.00

@ Elsevier Sequoia/Printed

in The Netherlands

264

Uuemiiis and Kleis [ 51 showed that particle flux rates had the opposite effect on rubber compared with metals. Increasing the flux rate increased rubber erosion, whereas in metals E decreases with flux rate. It is thought that this effect in metals is due to the collision of rebounding with incoming particles, which scatters the latter into trajectories with a range of impact angles [6]. Clearly some other process must override this effect in rubber. Uuemois and Kleis [5] suggested that the abrasive penetrates deeper into this material than into steel, “breaking down” the rubber layer between embedded particles. Another possible explanation is the build-up of heat due to the work done deforming the surface in each impact [7]. The eroding surface will rise in temperature if the rate of work done on it exceeds the rate at which heat can flow away through conduction, radiation or forced convection. Polymers, being low conductivity materials, would appear to be much more susceptible to “thermal runaway” than metals, although the flux rates needed to achieve this have been calculated to be very high, e.g. 35 kg rn-’ s-i or greater for polyethylene [ 81. Tilly and Sage [ 91 eroded Nylon 66 in a whirling-arm rig under vacuum. This eliminated aerodynamic effects. They found that the erosive efficiency of quartz sand was constant for particle sizes greater than about 100 pm, but fell off steeply below this value. Both sand and glass beads were comminuted by impacts on nylon. High speed photographic sequences were obtained of disintegration occurring. Nylon was found to have the same velocity dependence of erosion as metals, namely a power law of V2.3. Work on the ranking of polymers for erosion resistance is reported in two recent papers [lo, 111. Siiderberg et al. [lo] used a centrifugal particle accelerator (erofuge) to bombard a range of materials. They identified seven distinct mechanisms on the basis of crater morphology. Different polymers belonged to different classes. They also exhibited a range of volumetric erosion rates spanning an order of magnitude. PMMA belonged to the same class as glass (although its erosion rate was less), whereas nylon belonged to the same class as many metals (although its erosion rate was higher). Except for rubbers, the performance of polymers was much worse than that of metals (by about a factor of 10 in eV). Rao et al. [ll] investigated three polymers using a commercial bead blaster operating at normal incidence. In decreasing erosion resistance, they ranked them in the order polytetrafluoroethylene (PTFE), polycarbonate and PMMA, the last named being the worst of the three by about a factor of 10 in E,. The morphology of the material in the pits produced in PMMA they interpreted as giving evidence for melting. PTFE was observed to discolour and, like Tilly [2] who saw blackening of nylon and PP, they attributed this to heating caused by repeated impact. An application of polymeric materials where erosive failure has been reported is the pipeline industry. Increasingly, low pressure bulk transport conveyors of fuel gas or solid-liquid slurries are being constructed of plastics: high density polyethylene (HDPE), PP and unplasticized poly(viny1 chloride) in particular [ 121. These are very satisfactory for slurry transport: Meldt

[12] reported that HDPE pipes had a service lifetime 40% greater than steel pipes in ~~aysian tin mines. Bragaw 1133, however, has reported erosive failure in polyethylene (PE) gas piping systems due to contaminant grits. There were two mechanisms: (i) mechanical damage at constrictions; (ii) small holes in the pipe wall produced by electrostatic discharges. Electrostatic potentials up to 70 kV can be produced by dust-laden gas travelling down a pipe [13]; see also ref. 14. Agarwal et al. [15] found that pipe bends made of rubber performed better than steel bends in the pneumatic conveying of sand, but worse when alumina or coke was carried. Clearly then, erosion is not just a property of the material alone, but may depend on the physical and chemical properties of the erodent. A connection between measurable mechanical properties and abrasion resistance of polymers was made by Ratner et al. [16]. They analysed the abrasion process into three parts and assumed that the wear was proportional to the product of the probability of completing each stage. For a given pressure applied to the abrading body, they derived the relation

where cc is the coefficient of friction, H the polymer hardness, u the breaking stress in tension and E the breaking strain in tension. All the terms in this relation are functions of temperature, so the wear also will bear a (complicated) relationship to temperature. The minimum in wear is predicted to be around the glass transition temperature TB, but c~st~linity and filling agents will introduce other minima. Evans and Lancaster [17] found that the wear resistance was well correlated with the product UE for a range of polymers slid against rough mild steel. However, care needs to be exercised in transferring these ideas to erosion. Briscoe [18] lists a number of points that need to be considered, even for abrasion: (i) the correlation was for single-pass abrasion; (ii) u and E are usually measured in low strain rate uniaxial tension, whereas in wear the strain rates may be high and the stress state complex; (iii) wear damage is cumulative in nature, akin to fatigue processes. These warnings are borne out by the work of Marei and Izvozchikov [ 191 who found the following for vulcanized rubbers: (i) erosion was increased if fillers were included (compare with sliding wear); (ii) erosion decreased the greater the difference between the experimental temperature and Tg; (iii) erosion was least for low modulus highly elastic rubbers. In the papers mentioned so far, little has been said about the basic mechanisms of erosion, i.e. how it is that a succession of particle impacts on a particular spot leads to the fo~ation and breaking away of a wear particle. Electron micrographs published by Engel et al. [ZO] show that lips and fibrils are formed in ductile polymers (such as PE) which eventually are stretched to their breaking point. This phenomenon was also observed using high speed photography by Walley and Field [21] in PE. Following on this work, we decided to extend the study of erosion to other polymers, starting with PP. We chose to investigate single-particle impacts, although we agree

266

with Sarkar [22] that erosion rates cannot be predicted from such studies. Nevertheless, we feel that useful information about various mechanisms of erosive material damage can be obtained with this technique, especially in conjunction with high speed photography and electron microscopy.

2. Experimental technique We obtained PP in the form of extruded rods 26 mm in diameter. Various relevant properties of PP (some that we measured, some obtained from handbooks) are given in Table 1. Specimens about 12 mm in thickness were parted from the rod in a lathe and wet polished using 800 grade abrasive paper (particle size, 19 pm) to remove machining marks. The abrasion was carried out in one direction so as to leave a set of parallel polishing lines on the surface. This allows material displacement to be observed. The specimens were impacted at various angles and speeds-by spheres 4 mm in diameter fired from a laboratory gas gun [25]. The full experimental apparatus is shown diagrammatically in Fig. 1. The projectile is mounted on the front of a solid PE cylinder (called a sabot) which is a sliding fit in the barrel. The valve between the barrel and the gas reservoir consists of two metal diaphragms separated by an independently pressurizable space. This arrangement allows reproducible firing pressures to be used lying between P and 2P, where P is the bursting pressure of the diaTABLE 1 Properties of polypropylene Property

Value

Source

Density

905 + 5 kg rnp3

Heat of fusion

93 f 3 J g-’

DS.?

Crystallinity

49% + 2%

DSC, assuming 190 J g-l for 100% crystalline sample

Bulk melting temperature

165 “C

DSC

Thermal conductivity

0.12 W m-r K-l

Ref. 23

Flow stress in uniaxial compression at a strain rate of 2 X lo4 s-l

70 f 5 MPa

Direct impact Kolsky bar

Tensile strength

29 - 38 MPa

Ref. 24

Maximum tensile strain

50% - 600%

Ref. 24

3.8 x lo5 5.0 x 104

Gel permeation chromatography performed at the Rubber and Plastics Research Association

aDifferential scanning calorimetry

267 Two than nei delay generator

2

1 Specimen 0 A

Beam

spiittef

r

I

7

II

-Gun

Fig.

1.

I-P

Schematic

es

barrel

reservoir

diagram of the experimentaf apparatus.

phragms. The gun is fired by rapidly letting the interdiaphragm space down to atmospheric pressure. The diaphragms then burst in quick succession (opening times are about 250 ~.ls [26]), which allows the compressed gas in the reservoir to accelerate the sabot down the barrel. The sabot is brought to rest at the end by a massive end-stop which has a hole in it to allow the projectile to go on and strike the specimen. The impact speed is measured by arranging for two laser beams to traverse the flight path. These beams fall on two photodiodes whose outputs are passed to an exclusive-OR gate to form a square pulse for timing and triggering purposes. High speed photographs of impacts were taken with a Hadland image converter camera (Imacon) using a 17 @s in~rfr~e time unit. The synchronization of the camera and the flash with the event is achieved using a delay multiplier designed by P. H. Pope of this laboratory. This device sends a pulse to the two-channel delay generator at a time (accurate to 1 ps) calculated from information about the speed of the projectile and the distance from the timing beams to the specimen. This method was designed to overcome the problems caused by the variation in firing speed which the gun produces for a given driving pressure, although in practice the finite rise time (30 &s) of the flash used (a Braun 2000) means that the compensation cannot be perfect [8]. However, photographic success rates better than 70% can be achieved, compared with 30% for single-beam triggering (if the beam has to be remote from the specimen). The specimens are held in a massive steel holder which has a stage that can be rotated in the range from 0” (glancing) to 90” (normal impact). The impact angle is set using a protractor to *lo. Most of the shots were performed with steel spheres, but other materials were used as well. These are

268 TABLE

2

Properties

of impacting

Material

spheres Mass

Density

Sphericity

Surface

WI

(kg me3 1

(Pm

Mm )

Steel

260

7880

WC

495

14880

1

0.4

0.04

0.7

0.05

finish

Duralumin

89.5

2730

25

Semipolished

Semiborosilicate glass

81.5

2450

12

1.2

listed in Table 2. They were all precision spheres. After impact, the specimens had a film of aluminium deposited on them by evaporation from a hot filament under vacuum. This was to allow scanning electron microscopy to be performed. The instrument used was a Cambridge Stereoscan 250 mark II.

3. Results As for PE [ 211, it was found that the impacts made by steel spheres could be plotted on a “deformation map” (Fig. 2). The crater types lie in zones which depend on speed and angle of impact. Micrographs of the various types are presented in subsequent figures. “Smooth” craters are those where there was no microscopic surface modification, e.g. Fig. 3. Such craters were formed at all angles (Y for impact speeds Vi 5 70 m s-l but, for (Y= lo”, they were still being formed up to Vi = 160 m s-l. Above this speed, the end of the crater takes on a roughened appearance, with transverse cracks and bands being formed (Fig. 4). This region extends further and further back as the speed is increased. Higher magnification reveals fibrils stretched across the cracks and scratches (Fig. 5). As for PE [ 211 and metals [ 271, material was displaced into raised lips for certain angles and speeds (Fig. 6). Again high magnification revealed filaments, but this time running forward from the lip to the surface in front. This suggests that the lip has drawn back from its original position during impact. The ductility of PP is emphasized by Fig. 7 which is an example of an Imacon sequence showing a filament being drawn out to a very large strain before breaking. The remains of a filament can be seen in Fig. 8. A different erosion mechanism is seen operating in Fig. 9 where material is ejected ahead of the sphere together with several other fragments (total mass,

269

Embedment

Penetration

Bands

Smooth

Angle/

Fig. 2. 4 mm spheres kinetic

degs

Deformation map for single-impact crater types in PP formed by steel spheres in diameter at various speeds and angles. Individual points are plotted for WC 4 mm in diameter (the speed coordinate is that for a steel sphere with the same energy): B, bands; L, lips; P, penetration.

Fig. 3. A smooth crater. (Vl=143ms-‘;o=30”).

Impact

from right to left by a steel sphere 4 mm in diameter

Fig. 4. A banded crater. (Vi=220ms~‘;ol=lO”).

Impact

from

left to right

by a steel sphere

4 mm in diameter

Fig. 5. Enlargement of part of the banded region in the crater of Fig. 4. The fibtils stretched across the cracks should be noted. The micrograph has the same orientation as Fig. 4.

271

Fig. 6. Crater with a raised lip. Impact from right to left by a steel sphere 4 mm in diameter (Vi = 190 m s-l ; (x = 60”).

Fig. 7. Imacon sequence of an impact by a steel sphere 4 mm in diameter ( Vi = 152 m s-’ ; LY= 30”). The filament f being drawn out in frame 5 should be noted. This has broken by frame 6, and material m is seen adhering to the sphere in frame 7. (Interframe time, 17 p.)

Fig. 8. Filament that has detached from a sphere and remained on the surface. Impact from right to left by a steel sphere 4 mm in diameter (Vi = 300 m s-i ; a = 60”).

Fig. 9. Imacon sequence of an impact by a steel sphere 4 mm in diameter (Vi= 220 m s-l ; CY= 30”). The extrusion of material ahead of the sphere in frame 3 and the extensive mass loss that occurs should be noted. (Interframe time, 1’7 ps.)

2.4 mg). Under these sorts of conditions (CY5 40”; Vi 2 200 m s-l) the craters showed a heavily fractured morphology (Fig. 10). The arrowed regions show loosely held fragments which almost detached. The pieces that came away were typically a few millimetres in size. The surface finish is seen to be important here for blocks of PP have formed by fracture along and transverse to the polishing scratches (Fig. 11). Near the exit end, fine

273

Fig. 10. Heavily fractured crater. Impact from left to right by a steel sylhere 4 mm in diameter (Vi = 316 m s-r; (Y = 30”). Arrows point to loosely held fragmen Its that almc1st detached.

Fig. 11. Enlargement of fractured blocks of polymer at the edge of a crate r. Impact frcDrn right to left by a steel sphere 4 mm in diameter (Vi = 330 m s-r ; CY= 40” ). Cracks alemg and transverse to the polishing scratches should be noted.

Fig. 12. Enlargement of the space between two filaments at the exit end of a crater. Impact from right to left by a steel sphere 4 mm in diameter (Vi = 273 m s-l ; a! = 30”). The fine cobweb-like structure should be noted.

cobweb-like networks could be seen spanning cracks (Fig. 12). The filaments in these can be as small as 0.1 @cmin diameter. We think these networks are formed as sheets of highly sheared material which contract under the action of surface forces. At high impact speeds (greater than about 200 m s-i) pronounced shear banding can be seen in the lip material (Fig. 13(a)). Also pre-existing surface scratches running parallel to the impact direction are opened up (Fig. 13(b)). As in PE [ 211 embedment of single particles (Fig. 14) required very high speeds (greater than about 370 m s-l at ff = 60”). Even when a sphere penetrated to a depth equal to its diameter, it usually was expelled by the relaxing elastic stress field. An Imacon sequence sho~ng this is given in Fig. 15. Circumferential cracking inside a normal impact crater can be seen in Fig. 16. Higher magnification (Fig. 17) shows polymer filaments spanning the cracks. The last zone on the deformation map (Fig. 2) is that of penetration. This is where the sphere went in to a depth greater than or equal to its radius but did not embed. A few impacts were performed with tungsten carbide (WC) spheres to see whether pVi2 is a good scaling parameter for the damage caused by oblique impact. It is not at all obvious that it should be [ 281. However, in this ref. 28, Hutchings found that changing the density of the sphere

275

(b) Fig. 13. (a) Enlargement of one side of a crater exit to show shear banding. Impact from left to right by a steel sphere 4 mm in diameter (Vi = 240 m s-l; (Y= 60”). (b) ErIlargement of central region of the lip (same crater as in (a)) to show the opening UPofF lolishing marks parallel to the impact direction.

276

Fig. 14. An example of an impact where embedment occurred. Impact was from bottom right to top left by a steel sphere 4 mm in diameter (Vi = 370 m s-l ; (Y= 60”).

Fig. 15. Imacon sequence of an impact by a WCsphere 4 mm in diameter (Vi = 250 m s-l ; (Y= 90”). It should be noted that the sphere is almost completely submerged in frames 5 and 6 but still rebounds. Some filaments can be seen adhered below the sphere in frames 12 and 13. (Interframe time, 17 ps.)

277

Fig. 16. A normal impact crater formed by a steel sphere 4 mm in diameter (Vi =:231 m s-l).

Fig. X7. Enlargement of the cracked region of the crater in Fig. 16. The filaments spanning the cracks should be noted.

whilst keeping the kinetic energy constant had very little effect on the gross features of the crater and lip, although small scale features were different. This may have been due to the different speeds of sliding and the different thermal conductivities of the spheres used. In Fig. 2, the map was produced using steel spheres, and the individual data points for them are not plotted (there were 58). The data for WC spheres are plotted, but scaled as pVi2 for a given angle. That is, their speed coordinate is that for a steel sphere with the same kinetic energy. They can be seen to fit quite well. One WC impact has already been presented (Fig. 15); the equivalent steel speed is 345 m s-l. This sequence shows that material removal at normal incidence is possible. However, Duralumin and glass spheres did not fit this pattern. These two types were chosen because they have similar densities but very different thermal conductivities. Differences in behaviour were observed: the Duralumin produced brittle behaviour at the crater end whereas glass did not. However, comparatively few impacts were performed so that definite conclusions are difficult to draw. It is an area where future work would be useful, and we plan to do this.

4. Conclusions (i) A deformation map (Fig. 2) has been constructed for impacts of steel spheres 4 mm in diameter on PP. This shows the crater type to be expected for a given speed and angle of attack. It was shown to be applicable to WC spheres if the speed were scaled by the parameter pVi2. (ii) Significant material loss can occur by single-particle impact on PP. The mechanisms for this are as follows: (a) the drawing-out of fihunents adhering to the spheres; (b) the adhesion of a thin film of PP to the spheres; (c) the combined fracture and extrusion of blocks of polymer when impacted at high speeds and low angles. (iii) Surface finish was found to be an important factor in the lastmentioned mechanism. We suggest that impacts should be carried out on injection-moulded specimens or cylinders bored from cast sheet. Alternatively, the effect of surface scratches could be investigated, especially whether annealing close to the softening point might remove polishing marks. (iv) More work needs to be carried out on the effect of the nature of the impacting particle, especially its size, density and thermal conductivity.

Acknowledgments We would like to thank Dr. M. M. Chaudhri for useful comments on this work. This research was supported in part by a grant from the Ministry of Defence (Procurement Executive).

279

References 1 G. F. Schmitt, Jr., Liquid and solid particle impact erosion. In M. B. Peterson and W. 0. Winer (eds.), Wear Control Handbook, American Society of Mechanical Engineers, New York, 1980, p, 247. 2 G. P. Tilly, Erosion caused by airborne particles, Wear, 14 (1969) 63 - 79. 3 G. P. Tilly, Sand erosion of metals and plastics: a brief review, Wear, 14 (1969) 241 248. 4 J. H. Neilson and A. Gilchrist, Erosion by a stream of solid particles, Wear, 11 (1968) 111 - 122. 5 H. Uuemiiis and I. Kleis, A critical analysis of erosion problems which have been little studied, Wear, 31 (1975) 359 - 371. 6 D. R. Andrews and N. Horsfield, Particle collisions in the vicinity of an eroding surface, J. Phys. D, 16 (1983) 525 - 538. 7 D. R. Andrews, An analysis of solid particle erosion mechanisms, J. Phys. D, 14 (1981) 1979 - 1991. 8 S. M. Walley, Erosion of polyethylene by solid particle impact, Ph.D. Thesis, Cambridge University, 1983. 9 G. P. Tilly and W. Sage, The influence of particle and material behaviour on erosion processes, Wear, 16 (1970) 447 - 465. 10 S. Soderberg, S. Hogmark, U. Engman and H. Swahn, Erosion classification of materials using a centrifugal erosion tester, Tribal. Ink., 14 (1981) 333 - 343. 11 P. V. Rao, S. G. Young and D. H. Buckley, Solid spherical glass particle impingement studies of ptastic materials, NASA Tech. Paper 2161, 1983 (National Aeronautics and Space Administration). 12 R. Meldt, Transportation of solids through plastic pipes, Proc. 5th lnt. Conf. on Plastics Pipes, York, 1982, Plastics and Rubber Institute, London, Paper 29. 13 C. G. Bragaw, Fracture modes in medium density PE gas piping systems, Proc. 4th Int. Conf. on Plastics Pipes, University of Sussex, 1979, Plastics and Rubber Institute, London, Paper 23. 14 W. R. Harper, Contact and ~r~ctjona~Electrification, Oxford University Press, London, 1967. 15 V. K. Agarwal, D. Mills and J. S. Mason, A comparison of the erosive wear of steel and rubber bends in pneumatic conveying system pipelines. In J. E. Field and N. S. Corney (eds.), Proc. 6th Znt. Conf. on Erosion by Liquid and Solid Impact, Cambridge, 1983, Cavendish Laboratory, Cambridge, Cambs., Paper 60. 16 S. B. Ratner, I. I. Farberova, 0. V. Radyukevi~h and E. G. Lur’e, Connection between wear-resistance of phstics and other mechanical properties. In D. I. James (ed.), Abrasion of Rubber, MacLaren, London, 1967, pp. 145 - 154. 17 D. C. Evans and J. K. Lancaster, The wear of polymers, Treatise Mater. Sci, Technol., 13 (1979) 85 - 139. 18 B. J. Briscoe, Wear of polymers: an essay on fundamental aspects, Tribol. Int., 14 (1981) 231 - 243. 19 A. 1. Marei and P. V. Izvozchikov, Determination of the wear of rubbers in a stream of abrasive particles. In D. I. James (ed.), Abrasion of Rubber, MacLaren, London, 1967, pp. 274 * 280. 20 L. Engel, H. Klingele, G. W. Ehrenstein and H. Schaper, An Atlas of Polymer Damage, Wolfe, London, 1981, pp. 94, 95,102, 103. 21 S. M. Walley and J. E, Field, The impact erosion of polyethylene. In J. E. Field and N. S. Corney (eds.), Proc. 6th Znt. Conf on Erosion by Liquid and Solid Impact, Cambridge, 1983, Cavendish Laboratory, Cambridge, Cambs., Paper 44. 22 A. D. Sarkar, A study of crater vohrmes produced by single particles at low impact velocities, Wear, 87 (1983) 173 - 180. 23 G. W. C. Kaye and T. II. Laby, Tables of Physical and Chemical Constants, Longman, Harlow, 14th edn., 1973.

280 24

Properties of plastics, in Physical and Chemical Properties of Polymers and Moulding Compounds, Shell Chemicals. 25 I. M. Hutchings and R. E. Winter, A simple small bore laboratory gas-gun,J. Phys. E, 8 (1975) 84. 26 D. R. Andrews, The bursting diaphragm as a fast-acting valve, J. Phys. E, 16 (1983) 192. 2’7 I. M. Hutchings, R. E. Winter and J. E. Field, Solid particle erosion of metals: the removal of surface material by spherical projectiles, Proc. R. Sot. London, Ser. A, 348 (1976) 379 - 392. 28 I. M. Hutchings, The erosion of ductile metals, Ph.D. Thesis, Cambridge University, 1974.