Effect of counterface surface roughness and its evolution on the wear and friction of PEEK and PEEK-bonded carbon fibre composites on stainless steel

Effect of counterface surface roughness and its evolution on the wear and friction of PEEK and PEEK-bonded carbon fibre composites on stainless steel

WEAR ELSEVIER Wear 217 (1998) 288-296 Effect of counterface surface roughness and its evolution on the wear and friction of PEEK and PEEK-bonded car...

605KB Sizes 0 Downloads 14 Views

WEAR ELSEVIER

Wear 217 (1998) 288-296

Effect of counterface surface roughness and its evolution on the wear and friction of PEEK and PEEK-bonded carbon fibre composites on stainless steel D.M.

E i l i o t t ~', J . F i s h e r

~'*, D . T . C l a r k

b

l)epartment ~f Mechanical Engineering. University of Leeds, la,ed.v, LS2 9JT, UK hRe.~earc'h Unitfor Surfaces. Transfi~rmsand Inlerfaces. Daresbury Laboraulry, Warringtml. WA4 4AD. UK

Received 22 September 1997; accepted 9 January 1998

Abstract The time-dependent variations in the dry sliding wear rate and friction of PEEK (450G and 100P) and carbon fibre-bonded PEEK (APC2) on stainless steel ( 316S 16) have been investigated and the evolution of wear scratches measured with optical profilometry. A three-pin-ondisk tester was u~d, operating at 0.18 m s ~, with a pressure of I.O MPa on 5 mm diameter pin. Specific wear rates were calculated from ma.,~slost by the pins at 15 km intervals and friction force was monitored by a strain-gauged torque beam. Two series of 9d-kin tests were performed for each pin/disk combination. The stainless steel disks had either a fine-ground or polished surface. The roughness of the ground stainless steel surface was cbo.~n to coincide with the optimum value for low wear rate of the PEEK pins. To maximise pin/disk contact area, the PEEK and APC2 pin surfaces were mn-in on 10do-grit silicon carbide paper and ultrasonically cleaned in acetone before the tint wear test. There was no significant difference in steady-state wear between these pins and those tested with as-moulded surfaces. The disks were analysed at 15 km intervals with the UBM pmfilometer, providing 3D surface scans and 2D profiles for roughness parameters without damaging the wear track. All the surface roughness parameters reported in this work were produced from the UBM non-contacting optical profilometer. A marked difference in specific wear rate was observed between the two types of PEEK. The 450G had the lowest wear of 6 x I0 " mm ~ N ' m J on both the polished and ground counterface, whereas the 100P had wear factors of I × 10 4 mm ~ N " ' m ' on ground "and 2 x I0 ~ mm ~ N ~ m ' on polished stainless steel. However, the friction forces measured between the 450G PEEK 'and the polished disk were up to three times higher than those against the ground disk, (0.35 +0.01 ), decreasing to similar values as the polished surface became scratched. The APC2 pins showed very little difference in wear rate between the ground and polished disks. However, a steady de~.areasefromlXlO ~ ' m l x l O 7mm~N t m 'overadistanceof9Okmwasmeasuredasmorefibresweregroundintoagraphitetransfer film lubricant "andthe coefficient of friction dropped from 0.35 to O. 18. The specific wear rates and coefficients of friction of the APC2 samples were~w~r~hanth~f~rPEEK~bow~v~r~theApC2caus~dm~rescratchd~f~mmtion~fth~po~ish~dstee~counterfac~. © 1998 Elsevier Science S.A. All rights re~rved. Kt'l~'ord.~: Stainless sleel: PEEK: PEEK-Ixmdcdcarlton libre composites: Scratching

I. I n t r o d u c t i o n P E E K has m a n y applications in engineering and medicine b e c a u ~ o f its high strength and high melting point relative to other polymers, as well as its resistance to chemical and biological action. ~ addition o f carbon fibres to a P E E K matrix, to form the composite known as APC2. improves the mechanical properties and w e a r resistance. H o w e v e r . as w e a r • C~.,*respondingauthor. University of Leeds. Dept. Mechanical Engineering. The University. Wotalhous¢ lane, Leeds. LS2 9JT, UK. Tel.: +44. n532-332155: fax: +444)532-t24611. 0tM3.16,48/98/$19.00 ~5 1998 Elsevier Science S.A. All rights ~mn'cd. P I I s t ) n 4 3 - 1 6 4 8 ( 98 ) O0148- 3

is not an intrinsic material property, it is not possible to predict the w e a r response o f either P E E K or A P C 2 in e v e r y situation alter having p e r f o r m e d one series o f w e a r tests. As t h e ~ p o l y m e r based materials are often e m p l o y e d in applications where their w e a r and friction behaviour is critical to the performance o f the system, it is important to have a fuller understanding o f their tribological behaviour under a range o f conditions. Tribological results from three-pin-on-disk w e a r rigs are often different from those o f single pin types. A single pin is more prone to stick-slip p h e n o m e n a and preferential w e a r o f the leading edge o f the pin. The three pin system is particu-

D.M. Ellion el aL / Wear 217 (1998) 288-296

larly suited to studying the effect of surface roughness on wear as the contact area remains reasonably constant, after the initial running-in period. Reciprocating dry sliding wear tests with PEEK 450G on smooth stainless steel have been shown to obey the Amonton-Coulomb laws quite well for small interface temperature increases, but at higher temperatures the friction increased and the abrasive wear mechanism dominated with thick and irregular polymer transfer films [ I I. The wear rate of PEEK depends on the test configuration as PEEK pins sliding against a steel disk wore at tens times the rate of PEEK disks being worn by steel pins [ 2 ]. When stainless steel balls were run against a PEEK flat the main wear mechanism was ploughing with no net loss of material [3], and in tests of PEEK pins against steel disks of different surface finish the wear increased significantly with increase of the roughness oftbe steel [4]. However, this last result referred to initial or short term wear rates. Other experimenters have reported an almost linear increase in wear rate with load from 7 × I0 -~ mm3/Nm at 4 MPa to 1.8× 10 -5 mm3/Nm at 8 MPa for PEEK pins sliding at I m s - t against a steel ring ( 100Cr6. HRC 62, Ra 0.2-0.3 pro). The coefficient of friction of 0.38 + 0.06 was independent of loading under the same test conditions. A combination of tuicrocutting, plastic flow and fatigue-delamination effects were responsible for the wear of the PEEK, with the fatigue mechanism becoming more prominent at higher loads [5]. Graphite fibre PEEK composite cubic pins of 8.9 mm side length have been tested with a contact pressure of 0.6 MPa against a 1018 Steel disk (HRC 20, Ra 0.06/,tm) sliding at 0.5, !.0 and 1.5 m s - ~ for 75 rain at RH 30% and various temperatures. It was found that woven fibres gave better wear resistance by far than unidirectional fibres and that the wear rates increased linearly with p v values and also with temperature. In addition, the friction was independent of orientation but increased slightly with increases in p v values. With normally oriented fibres, steel was scored or cut out of the disc and embedded into the PEEK matrix causing steel-on-steel sliding micro p a i n and severe ,scoring oftbo counterface [ 6 ]. In other experiments the wear rate of APC2 has been shown to be almost independent of coonterface steel roughness [41With a rounded APC2 pin against a sapphire disk a normal fibre orientation (as opposed to parallel and anti-parallel) resulted in the lowest friction except at low velocities where it produced the highest friction. At low velocities the wear process was dominated by fatigue wear, whereas adhesive wear dominated at high velocities. Counterface material was crucial as fatigue wear with the steel disc was not observed [ 7 I. However, the surface finish of the steel disk may not have been as smooth as the sapphire. The effect of initial counterface topography on the wear of PEEK and APC2 using a three-pin-on-disk configuration has been studied. A minima in the wear rate for both materials was found corresponding to a surface roughness of about 0.13/.tm r.m.s for PEEK and 0.23 p,m r.m.s, for APC2. The counterface was mild steel (41 HRB) and the surface rough-

289

hess was in the form of ~ratcbes of random orientation [ 81. The effect of controlled surface topography on the transfer and wear of PEEK has also been investigated, using a surface scribing apparatus to produce longitudinal, transverse and yawed transverse configurations. No correlation between groove angle and PEEK wear rate was found, and coherent, stable, and continuous PEEK transfer films were not observed in any of the scribed-disk wear tests. However, secne discrete agglomerates of highly worked PEEK platelets inside the grooves and on the plateau between grooves being initiated by the trailing lips of the grooves were observed [9]. In the present study the grinding scratches of the rough disk were all in one direction and thus changed through 90 ° with respect to the sliding direction as the pins travelled along the circular track. This allowed investigation of any difference in pin wear debris collection from one region of the disk to another as a result of its surface scratch orientation. The main objectives were to provide comparative results on the dry sliding wear and friction behaviour of PEEK and APC2 on stainless steel and to illustrate the evolution oftbe .surface profile of the steel and how this in turn affected the wear rate.

2. Experimental details 2. I. Material.s

Two grades of PEEK ( Polyetheretherketone ) were te.caed. The first PEEK 100P was moulded from powder and had a relatively low molecular weight and low viscosity, it was provided for this work because it was used to produce AINU2, and because it was made at least two years before the 450G it contained black Slx. k s of degraded PEEK ( up to 500 fan dispersed a: random). The manufacturing process has now been improved and the 450<3 contained a negligible amount of specks. The 450G was moulded from granules and had a higher molecular weight than 100P. PEEK 450G is used for most applications at pre~nt. The APC2 was made of repeated mats of approximately 100 layers of 7 tttu diameter carbon fibres, with each successive mat being at 45 ° to the previous one and bonded with 100P PEEK. A summary of some properties of the materials tested is presented in Table I. Table I Some relativepropertiesof the materialsunder teSl Density

Yceng's

Vickcrs

(g/cm +)

modulus (GPa)

"hardness (MPa)

Fib~ lypc

PEEK l O O P PEEK 450G APC2

1.26-1.32 1.26-1.32 1.6

3.8 3.8 134 (()o~

~ 300

-

600

14crculcsAS4 (61% vol)

Stainlesssteel (316SI6)

7.8

2130

2020

D.M. Ellion et al. / Wear 217 (1098) 288-296

29O

Table 2 Test parametersfor the dry slidlng wear tests

2.2. Apparatus and test pnmedure A three pin-on-disk wear testing machine, designed and built at Leeds University, was used for the dry sliding wear and friction tests (Fig. I). The speed of disk rotation was measured by counting the disk revolutions with a metal sensor detecting the passing o f a bolt screwed into the axle of the disk holder. The pins were fixed in collets mounted in a cylindrical block pin holder, whose central hole located on a post in the centre of the disk holder, (enlarged for the present work), and was free to rotate by the aid o f bearings. Loads were ~ c u r e d to the top of the pin holder with a central bolt. An arm protruded horizontally from the side oftbe block and, as the disk beneath the block turned, the pin holder arm came into contact with a strain-gauged torque beam allowing the friction force to be monitored. A chart recorder was used to plot any changes in friction during the first and last hours o f each 15-km test and at intervals throughout the test. The disk and pin holder were protected from air-borne contamination and sudden temperature changes by a glass dome cover. PEEK l o o P pins were cut and turned to 5.0 mm from injection moulded disks, and those of 450(3 were cut from a 5.0 mm cylindrical rod. When possible, the original asmoulded surface was u ~ d , but the 450G pins were finished against 1000-grit silicon carbide paper. This gave the 450G PEEK pins an Ra o f about 0.35 # m . Initial experiments were performed with each material to a ~ e r t a i n the length of running-in period required before steady-state wear rates could be recorded. Full 24 h ( 15 kin) tests were run over night, followed by weighing o f the pins on a Sartorious mechanical balance and surface analysis o f the disks on a UBM optical profilometer before the next test. Loose debris was removed from the pins and the dis~ with compressed air betbre analysis. Each test provided three mass losses, from which a single average specific wear rate was calculated and the test series for any one mt of parameters was repeated at least once. Tests were often stopped before the chosen maximum o1"90 km if either the pins had worn out. or a reasonably constant wear rate had been recorded after at least 60 km. The test paramete~ are shown in Table 2. A total load of 6 kg produced a pressure of 1.0 MPa per pin and the disk

-- --.-- _ . . . -__ ~ ---?.. _ _ -: ,

+ +~S

+'i++i~ ....

Type/range 3 Pin+on-disk 5 mm diametercylindrical pins. 17.5 mm long 316S16 stainless steel: Ra 0.30/.tm and 0.02 #m 0.18 ms +t 1.0 MPa multiplesof 15 km ( max. 90 km ) ambient (within ranges 18-25°Cand 30c~"-7~/~ respectively)

Sliding counterl:ace Sliding velocity Contact pressure Sliding distance Temperatureand humidity

rotation of one revolution per second produced a sliding speed of 0.18 ms - ~. Temperature and humidity were measured at the beginning and end of each 15 km test using a Kane-May H81 Humidity probe, but no attempt was made to control them, other than avoiding running the test in direct sunlight. The sliding distance was calculated from the total number o f disk revolutions and the circumference of the circle described by the centre o f the pins. 2.3. Wear equation The specific wear rate, W,, used in this work, was calculated by dividing the volume o f material lost ( m m "~) by the apparent normal force ( N ) and the sliding distance ( m ) . This defines the material wear in terms of volume lost and work done. A dimensionless wear rate, W,, is defined as:

W,=m/ALp where m is the mass lost, A is the apparent contact area, L is the sliding distance, and p the density o f the material being wear tested. The specific wear rate is equivalent to the dimensionless wear rate divided by the apparent contact pressure p:

w,=w,/p Specific wear rates of dimensions mm~/Nm are used throughout this work. 2.4. Choice o f disk surface ronghness

.

21+'.

Parameter "lest apparatus Specimen size

_ ~p.P.oU.~.mn.

...... L~,,n. . . . .+_ an,,-.~m

Fig I. Diagramof tl~ 3-pin-on-diskapparatus.

Four surface roughnesses were investigated with PEEK 100P and from these initial friction and wear results ( Figs. 2 and 3), two roughnesses were selected for all the tests on PEEK and APC2. These represented the results of two reasonably standard methods o f surface finish which were easy to reproduce. Fine grinding, with a cylindrical silicon carbide grinding wheel, produced surfaces with Ra values of 0.30+_0.05 p.m measured at 90 ° to the direction of grinding. Whereas, lapping and then polishing with 3 # m diamond paste produced mirror-finish surfaces of Ra value 0.02 +_0.002/.tm. As the majority of published work on sur-

D.M. Elliottet aL I Wear217 ( 19981288-29¢i la [ .--

t o+!

3. Results

o I o

0.0s

0+

01s

02

02s

03

0:~

SurtaozI~uCammRa(win) Fig. 2. The variationof PEEK 100Pwearratewith initialcountcrl~e surlace roughness,with standarddeviationerror bars. oa~

291

ing .scratches as, by experiment, this was found to yield the most meaningful results. A cut-off of 0.8 mm was used, as well as Ibe standard filtering, which minimized the effectthat the larger .scratches had on the calculation of the roughness average.

. . . . . . . . . . . . . . . . . . .

°+t

!

ss I

os: ~ o4s i

~ o.,; 03S ~

0.3 ° . . . . . . . . . . . . . . . . 005

0.1

015

02

__. 02S

0.3

3. I. Effect q f cmmterface rottghness on the friction and wear o f PEEK Fig. 4 shows the difference in wear rate between PEEK lOOP pins against a ground 316 stainless steel disk and PEEK lOOP pins against a polished 316 stainless steel disk. There is almost a ten-fold increase from PEEK wear against the polished surface to PEEK wear against the ground surface. After 45 kin, the pins wearing against the ground surface had worn down to the pin bolder. The coefficient of friction ( Fig. 5) was just below 0.3 for 100P against the ground disk and stared at 0.80 against the polished disk surface, during the initial stages of the first 15 km run, ending at about 0.58. Then it dropped Lhroughout the next 15 km to about 0.35 and steadily increased lbr the next four periods of 15 km to finish at about 0.53 alter 90 km.

0:)5

Surca¢~Rougtm*m Ra ~ )

Fig. 3. The variationof PEEK 100P IYiclioncoetlicicntwith initialcounlcrlace surface roughness,with error bars showing maximumand minimum friction.

14 O0

.

.

.

.

.

~_

e moo.

0

2.5. Microscopy

Fig. 4. The variation of wear rate with ~,liding distance for PEEK 100P against ground and polished 316 ~lainless steel, with slandard deviation error

burs. O7 [=;Ground

oe.

2.6. Non-contacting optical proJilomet~" The disks were scanned using a UBM Optical Profilometer to produce a 3D plot of a section of the wear track measuring 5 mm X 2 mm and a single 5 mm scan was made at a higher resolutio,~ (400 points/mm ) providing data for the various surface roughness parameter calculations. The scan direction w a s at r i g h t a n g l e s t o t h e w e a r t r a c k s c r a t c h e s a n d a n y g r i n d -

:

OPoIi~

face roughness effects use contacting stylus type values, it should he noted that the ground surfaces in this work had corresponding Ra values of approximately O. 15 + 0.05 p m and the polished surfaces had Ra values of approximately 0.007+0.001 /~m measured with a Rank Taylor Hobsou Talysurf.

Wear surfaces were examined by optical and .scanning electron microscopes in order to investigate the mechanisms responsible for the various friction forces and wear ,ares resulting from each test. Samples for physical analysis of the surface by SEM were gold coated and tbose for EDAX chemical analysis were carbon coated. An arrow on the micrograph indicates the direction of sliding of the disk relative to the pin surface.

:_ _-

is

i++ii_._ ;.H._i :3O

45

~

75

gO

Fig. 5. The vurialion c,f fi'iclion ¢(~fficicnt with ",,liding dislam.c for PEEK ] (X)P againsl gnmnd and p~lished 316 ~,luin]ess steel with error burs showing maximum and minimum fiiction.

D.M. Elliottel al. / Wear217 (1908)288-2~

292

PEEK 450G behaved as if it were a completely different material than lOOP. Fig. 6 shows that the wear o f this type of PEEK was almost independent o f the two surface finishes and was half the wear rate experienced by lOOP against the polished disk. The coefficients o f friction (Fig. 7), like the wear rates, were more or less independent of the surface roughness, being within the range 0.3 to 0.35 allowing for errors, The highest friction was measured against the polished surface after 15 km and the lowest friction was measured against the ground surface after 60 kin.

3.2. Effect o f counterface rouglmess on the friction and wear o f APC2 Fig. 8 presents the wear rate results for APC2 with fibres lying parallel to the wear surface, but with random fibre orientation with respect to the sliding direction. The reasons for this are discussed later, in general the wear of APC2 against the ground disk was higher than that against the polished disk by about 50%, with the exception o f the wear rate after 60 km and there was a decrease in wear with sliding distance against both surfaces. Apart from the wear rate against the ground disk during the first 15 km. all the other APC2 wear rates were at least ten times less than those for PEEK 450G against either surface. The friction coefficients o f APC2 (Fig, 9) against the ground disk were around 0.2 for the first 60 km and then decreased to O. 18 for the next two 15-km periods. Friction against the polished disk dropped from 0.25 after 15 km to 0.2 after 45 km and then rose to 0.25 again after 90 kin. This trend in friction was similar to that observed with PEEK l o o P against the polished surface and suggests that the matrix might play a part in controlling the friction of APC2 against the polished stainless steel surface. Carbon debris from the fibres was responsible for the low coefficient of friction measured against both surfaces.

Is

3o 4s sHo~ng t ~ a n c . e (kin)

-

12oorF~- ........

i

I~

~o

4s

Sliding~

6o

03; i] E

.

.

.

.

.

.

.

.

.

DGround- I . ,",Polis,~, [

.

,

. is

r .,

~o

~

T

.

~

4s

¢~

.

?s

to

511d4ng ¢k~tance (kin)

: rapo~ishecl

":i

go

[ is

OGround i

~

(kin)

o,5i

g

~0, i

i

[] Polished

'i

3.3, I. Peek lOOP agabJst polished stainless steel After two hours running-in ( 1583 m ) . the disk surface Ra had increased from 0.019 to 0.026 p.m, and the Rv had 900;

=0o u00 =i

Fig. 8. The variation of wear rate with sliding distance for APC2 against ground and polished 316 stainless steel, with standard deviationerror bars.

3.3. Evohaitm o f surface roughness

Ioo : -'~EEr~ i

eo

Fig. 7. The variation of friction coefficient wilh sliding distance for PEEK 450(5 against ground and polished316 stainless steel, with en'or bars showing maximumand minimum friction.

4S

dimnce (kin)

6O

Fig. 6. The varialitm of wear rate with sliding distance Ibr PEEK 45DG againstground and polished316 stainlesssteel, with standarddee ialionerror bar:..

Fig. 9. The variation of friction coefficient with sliding dislance for APC2 against ground and polished 316 stainless steel, with error bars showing znaximmnand minimum friction. increased by a factor o f five from 0.102 to 0.538 p.m. The skew factor of the surface had changed from - 0.039 to 1.19 and the kurtosis increased from 4.98 to 63.62. After 15 kin, the Ra had increased to 0.O41 p.m, the Skew factor had dropped to - 2.52 and the Kurtosis was down to 45.26, hut the Rv had only increased by a factor of two to 1.054 p.m. This suggests that the main scratches seen after 15 km were already there after the first 1.5 kin, most likely being formed during the initial stages of contact when the pin and disk surfaces were not yet parallel and, therefore, some contact

D.M. E/lionel al. / Wear217 (1998) 288-296

293

points would be under considerably higher stress than during the majority of the wear test, applying differential shearing stresses to the steel surface. Whether this alone was sufficient stress to cause plastic deformation of the steel is unlikely. However, coupled with entrapped PEEK debris and the possibility of other harder panicles from the polishing process (embedded in the steel surface and thus not removed during disk cleaning) a hypothesis to explain how the PEEK apparently .scratches the steel could he made. After 60 km sliding distance, the Ra had increased to 0.088 pro. Rt to 4.58 pin,

and the Skew factor had increased to -0.858 with a kunosis of 48.4 being almost the same as it was after 15 kin. A relatively deep scratch (6.7/tm) in the middle of the wear tr~ck appeared between 15 and 30 km. A similar scratch appeared during a repeat experiment during the first 15 km of the test. 3.3 2. Peek lOOP against ground stainless steel The surface roughness decreased slightly from 0.272 #m to 0.269/.tm after the first 1500 m and then increased to0.342 ~m after 15 km. Similar deep scratches to those found with the polished disk were ob~rved, however, in both tests against a ground disk the wear was dominated by the effect of the black specks of degraded or cross-linked PEEK di~ tributed at random throughout the sample and being roughly spherical of different diameters in the range of 5 to 500/~m. One such speck caused a 300 p,m wide ~ratch nearly 8 ttm deep with 5/,tm of p!le-up which corresponded to a 20/.tm deep scratch on the pin in line with the area of degraded PEEK standing about 12/.tin above the pin surface. Analysis by EDAX revealed that the area was embedded by a mixture of stainless steel wear debris (thin sheets of about 100/,tm diameter) and metal oxides. Fig. i0 shows a scanning electron microscope image, in back-scattered mode, clearly distinguishing the thin sheets of steel debris from the PEEK background. 3.3.3. Peek 450G agabl.vr polished stainless steel After 15 km the disk Ra had changed from 0.022/,tin to 0.089 #m and the Rt from 0.356 ~m to 3.86/,tin. There were three or tour isolated ~ratches with considerable pile-up (e.g., a depth of 1.43 #m and pile-up of 2.69/tin ). After 60 km the Ra had increased to 0.i09 p.m and the Rt reduced slightly to 2.73 p,m. The majority of ~r,,tches, and certainly all the deepest, occurred within the first 15 km of sliding, (Fig. II). 3.3.4. Peek 450G agabtsr ground stabdess steel The surface roughness of the disk changed very little from 0.304/tm initially to 0.316/zm after 15 km and 0.357/tm after 60 km. On the areas of the disk where the wear ~ars were at right angles to the grinding scratches, optical microscopy revealed evidence of deformation of the steel in the form of relatively deep 10-20/zm wide ~ratcbes. These were probably caused by detached work-hardenc-! steel particles becoming embedded in the PEEK surface and cutting the

Fig. IO. Back scaucfedscamningcleclroa microsr4~ of stakless sled wea" debris around a degradedPEEK speck at the .~rl'ac¢ of a PEEK 100Ppin after 60 km wearagain~ u polished 31b stainless.~eeldi~.

klledSL~,.

Elli0tt

a.01.~lllRlM 0Kmz,q ~ m r m m r

oxm

Z.6g Z.O0

100

I L~ ~ .' 'J - -,*r, tr~q

00n -r

Surl'~ t ' t r 24 hr,~: 'L.~I~

!0j ...... jo!

i '~"~"I Surfaceafter M ~rs.-se~ Fig. I I. Comparison of the polished stainlesssteel disk w~sr Ir,v,:kprolilcs after 15 km and 60 km sliding distance against PEEK 4Y~Gpins.

294

D.M. Elliott et aL / Wear 217 (I 9~8) 26'8-296

steel counterface, Back-scattered electron Imaging of the PEEK pin surface showed such panicles, particularly around the edge of the pin. 3.3.5, APC2 agablst polished smbzless ,steel

A few deep scratches occurred during the first 15 kin, but no significant increase in their number or depth was observed throughout the rest of the 90 km test. Two surface profiles are shown in Fig. 12 to illustrate this result. 3.3.6. APC2 against grotmd stahdess steel

Profilometry of the ground disk surface showed no great difference in surface roughness between the unworn and worn (60 km). however, optical microscopy again showed deformation of the steel, in the form of smoothing of the grinding scratch ridges with occasional minor scratches in the direction of sliding. The smoothing effect was most likely caused by the c',abon fibres. p.eemum t

EIIxO??

0z.0~.g? 13glI41"IR

(~(i

",'rr~'llT " '-

~.O0 m ; 400 p/m ~ri'~

~/tct 24hes ;15 km

:,j i

1 ~ 0 0 m* 400 wmm

r:mrftlCelttQP444trs: gore, Fig. 12. Comparisonof the i~dishedstitinlesssteel disk ',+.'eartrack proliles

alter 15 km and 91)km slidingdistanceagainst AIK'2pins,

4. Discussion In tests on both surface types the roughness of the steel disk was modified during the initial stages of contact, but the rate at which it continued to change decreased enormously from then on. There appeared to be an optimum roughness produced and controlled by the material properties and test parameters. However. this control was strongly affected by 'third bodies' such as degraded PEEK specks in the PEEK lOOP and graphite from the carbon fibre wear debris with Arc2. The initial surface roughness of the polished disk would not have affected the number and depth of the scratches it received. With PEEK 450G, as there were few other factors affecting wear, the evolution of the polished surface tended towards that of the ground disk, and thus the long term wear rate was practically independent of the initial counterfac¢ conditions. In comparison to the cumulative effect of the grinding scratches on the ground disk, a single scratch on tbe polished disk. of similar depth, appeared to be almost as efficient at wearing PEEK. The pressure between the PEEK and the pile-up edge of the single scratch on the polished disk would have been many times greater than that between PEEK and any of the grinding scratch asperities. The grinding scratches also changed orientation with respect to the sliding direction as the disk turned, thus diminishing their ability to cut the PEEK. Debris entrapment, leading to partial lubrication, was also more likely to occur with the ground disk, further limiting the wear of the PEEK by the rough surface, Despite a rigorous disk cleaning regime prior to wear testing, the scratches tbrmed on the polished disks could have been made by some 3 #m diamond particles from the polishing stage entrapped in the disk sud'ace. Similarly, the scratches on the ground disk could have been caused by the workhardened steel debris attached as a burr to the edges of the grinding scratches. Although these particles afl;,+ctthe various tribosystems they are representative of real conditions and, as such, do not invalidate the comparative results presented in this paper as long as their presence is acknowledged. The effect of the ground disk against the relatively soft PEEK lOOp was to uncover more specks of degraded PEEK generating deep scratches which kept the wear rate at ten times higher than that for lOOp against the polished disk. The phtte-like PEEK debris formed during the PEEK 100P on polished steel test was squeezed together to lbrm a film of about 5 txm thick covering about 2/3 of one pin surface, ( Figs. 13 and 14). Fig. 15 shows a 3D plot of the edge of this debris film. However. no such film was observed on any of the stainless steel counteriaces. With the polished counterfaces, some isolated PEEK debris was observed in the scratches and on the surface between scratches, but was easily removed by compressed air. With the ground surfaces, similar PEEK debris was observed, attached to the surface, in isolated areas of up to 50/,tin in diameter, between the major wear track scratches and covering no more than 5% of the wear track surface. In these tests, as the PEEK debris was mainly

D.M. Elliott et al. / Wear 217 ~19~8t 288--296

~g. 13 S e M n6crosraph ~ Ff,~K IOOP~

~lm m a PE~K ~..

~J

Fig. 14. SEM micrographshowingplate-likemorphologyof PEEK 100P debris film.

.'295

r:ported significant PEEK transfer films, used higher loads and speeds, larger contact areas, and rougher counterface surfaces, with the grinding .scratches often at a constant orientation with respect to the direction of slidiog. In the pre~nt work. the wear tracks on the polished disks from the 100P PEEK were often discoloured by a very thin brown layer, but this was too thin to measure by profilometry. XPS analysis detected carbon, oxygen and iron in the un-wom surface of the polished disk. But within the wear track, in addition to these elemenLs, there was .sodium, nitrogen, calcium, zinc, phosphorous, lead. boron, and chromium. The carhon:oxygan ratio was greater within the wear track than outside it. but the carbon Is signal from within the wear track ,showed little oxygen functionality. Therefore. the oxygen was mainly associated with the unexpected elements such as phosphorous. Both inside and ouLside the wear track the carbon Is spectrum did not resemble that of PEEK. Electron energy maps (E-X images) of the sample showed .some areas of relatively thick carbon, which blocked electrons from the steel disk. This carbon appeared to correspond to the deepest .scratches in the wear track. The areas between these .scratches were covered by a much thinner carbon layer, thin enough m allow electrons from the disk to pass through it during XPS. Fig. 16 shows evidence of wear by delamination as well as micro-ploughiog of the PEEK 100P by the ground stainless steel disk. The APC2 fibre layer lay-up was 0°.45°.90° which, along with the fact that three pins were wearing in the .same test. made isotropic effects difficult to study. The pins did not always wear at the same rate and often the fibre layer was not exactly parallel to the disk surface leading to wear surfaces composed of two areas of different fibre orientation. However. the angle of the fibre layer to the wear surface meant that there was a decreasing thickness of fibres in one direction ( a wedge shape) and it was observed that fibres did not pull out easily near the thin end of the wedge. Fig. 17 shows the well-defined line between high and low fibre pull-out. The angle of the fibre layer to the surface of this particular pin

-ZOrn

I

~O.n

Fig. 15.3D oplicalprolilescan ~lfPEEKdebris lihn in Fig. 13. attached to the .sections of the wear track that had retained the original grinding ~ratcbes, its ability to lubricate would be limited and it is unlikely that it affected the evolution of the counterface surface roughness. Previous work, that

Fig. 16. SEM micrographshowingdelamhuttionand ploughingwear on PEEK IOOPagainstground316 stainlesssteel

296

D.M, Elliott el al. / Wear 21711998) 288-296

4, APC2 showed 50% lower wear against the polished disk compared to the ground disk. 5. All the polished disks received the majority o f scratch deformation during the first 15 km sliding with little increase in surface roughness from then on. 6. The pile-up at the scratch edge on the polished disks was of the same order of magnitude as the grinding scratches on the ground disk. 7. A few isolated scratches formed on the polished disk produced about the same amount o f PEEK 450(3 wear as the grinding scratches on the ground disk.

soo

i1

t

i

Fig. 17. Ol~ical mierograph showing the change from high to low fibre pallout on the worn surfaceof an APC2 pin.

was about 14° and the depth from the surface to the next fibre layer o f different orientation, at the point where fibre pull-out stopped, was about 215 p,m or about 30 fibres thick. Each layer had at least 100 fibres in it. The effect would suggest that making the composite o f thinner layers would improve its wear resistance.

S. C o n c l u s i o n s I. The PEEK lOOP had a higher wear rate and lower fricdon against ground stainless steel than against polished stainless steel. 2. PEEK 450G had a lower wear rate than PEEK IOOP and its friction and wear rate were both practically independent of initial surface roughness. 3. APC2 had a lower wear rate than both PEEK grades with low friction against both counlerfaces because of graphite lubrication from the carbon fibre debris.

References I I [ A. Scheliing, FI.H. Kausch. Friction hehaviour of poiyetheretherkclone under dry reciprocating movement, Wear 151 ( 1991) 129-142. 121 T.A. Stolarski. Tribo]ogy of polyetheretherketone,Wear 158 (1992)

71-78. 131 J.H. Yon. N.S. Eiss Jr,, Tribological behavior of blends ofpolyether ether ketone and polyether imide, Wear 162-164(1993) 418-4.25. 14l K. Friedrich. J. Kargcr-Kocsis,Z. Lu. Effects of steel eountefface roughnessand temperatureon the frictionand wear of PE(E ) K composites under dry sllding conditions, Wear 148 ( 1991) 235-247. 15 [ M.Q. Zhang. Z.P. Lu, K, Friedrich, On the wear debris of polyetheretherketnne: frdctaldimensions in relation to wear mechanisms.Tribology Int. 30 (2) (1997) 87-102. 161 P.B. Mr~ly, T.W. Chou, K. Friedrich, Effects of testing conditionsand miert~s~roctureon the sliding wear of graphite fibre/PEEK matrix composites,J. Mater. Sci. 23 (1988) 4319--4330. 171 B.S. Tripathy. M.J. Furey. Tribological behavior of unidirectional graphite-epoxyand carbon-PEEKcomposites.Wear 162-164(1993) 385-.396. 181 T.C. Ovaen, H.S. Chcng. Counterface topographical effects on the wear of Ix~lyetheretherketoneand a p,,)lyctheretherketone-carbonfiber co,nposite, Wear 15n ( 1991) 257-287. I ~)I S. Ramaehandra,T.C. Ovaen. The effect of controlled surface topographical features on the unluhricated transfer and wear of PEEK. Wear 206 (1997) 94-¢P).