The influence from PTFE on surface and sub-surface damages of glass fiber reinforced PPS

Polymer composites are common materials for tribological components, even at relatively high operating temperatures. One advantage is the possibility to add low friction additives, such as PTFE, to the base material. In this paper the influence of such additives was studied, with focus on surface and sub-surface damages. The polymer composites were PPS with glass fibers (PPS-F), PPS with glass fibers and PTFE (PPS-L) and PEEK with carbon fibers, graphite and PTFE (PEEK-L). A reciprocating ball-on-disc test set-up, with ball bearing steel balls as counter material, was used with 5 N and 15 N normal load. The tests were run for 2,000 and 10,000 cycles in room temperature, 80 ◦ C and 120 ◦ C. PPS-F showed high friction ( µ ≈ 0.4 – 0.5) and severe surface damage of both the polymer and the counter surface. Sub-surface cracks, detachment of fibers from the polymer and a deformed surface layer was revealed, when studied in cross section. The load and temperature had negligible effect on the friction, however a higher temperature resulted in more surface damage of the PPS-F. PPS-L and PEEK-L showed relatively low friction ( µ ≈ 0.1) and superficial surface damage. XPS analysis revealed a thin tribofilm of PTFE in the wear track for PPS-L.


Introduction
In this study glass fiber reinforced poly-phenylene-sulphide (PPS) composites are investigated, with a poly-ether-ether-ketone (PEEK) composite for comparision.Both PEEK and PPS are widely used in industrial applications today.Unfilled PPS is cheeper, but PEEK has superior thermal properties.The mechanical properties at room temperature are comparable for the two polymers, although PPS exhibits a more brittle behavior than PEEK, which is more ductile [1][2][3].The performance and behavior of PPS is the main focus of this study and PEEK, known for its high performance in tribological contacts [4][5][6][7][8], is included for comparision.
The demand for materials tolerating higher temperatures makes PEEK more interesting because of its high glass transition temperature T g and melting point.Above T g the tensile modulus decreases [9] which can be devastating for the tribological performance.The T g is around 90 • C for PPS and around 143 • C for PEEK.T g is often referred to as a specific temperature but the transition occurs more gradually over an interval, and varies with crystallinity, fillers and cooling rate, to name a few of the decicive parameters [9,10].
Within the automotive industry, polymers are used both for interior parts and surfaces as well as for some mechanical components, such as in clutch actuators.For mechnical components that operate close to the engine, the ability to maintain mechanical and tribological properties at elevated temperatures is of utmost importance.Operating temperatures around 80 • C have been typical for this type of components, but the development of more energy efficient vehicles has tended to raise the operating temperature up to around 120 • C.This development makes it important to investigate the tribological performance of these materials up to 120 • C. The mechanical strength of unfilled polymers is often too low to be used for mechanical components, but it can be substantially improved by adding reinforcement [11,12].Such reinforcement include different types of fibers, for example glass or carbon fibers [13,14].Fibers contribute to increased strength, but from a tribological perspective they can also cause abrasive wear on the counter surface [15,16].Not only the type of fiber but also the fiber orientation relative to the sliding direction has been found to be important in minimizing wear of the counter surface, with fibers oriented parallel to the sliding direction causing the least wear [17].
In tribological contacts, polymers most often transfer material to the counter surface, which may result in a covering polymer layer, called transfer layer [18].The specific mechanisms and the role of polymer transfer layers are not fully understood.It has been shown that the thickness of transfer layers is not related to the polymer wear rate [18].It has e.g.been shown that PPS has poor adhesion to steel counter surfaces and does not form any uniform transfer layer on such surfaces [15].
Another interesting tribological aspect of polymer composites is the fact that lubricating additives can be added to the matrix, decreasing the need for adding an external lubricant in the application.In this paper this will be referred to as internal lubrication.Such additives, for example PTFE particles, can be mixed with the base polymer in order to achieve low friction properties [18,19].It is known that PTFE as an additive significantly decrease friction and wear in a component.This is explained by its low shear resistance and its high tendency to adhere to the counter surface.This process results in PTFE on both mating surfaces and is associated to very low friction [18,20].The film transferred to the counter surface may subsequently be transferred to other surfaces in the system, so-called secondary transfer, i.e. resulting in PTFE layers also on surfaces without direct contact with the PTFE [21].
Although SEM imaging is common in studies of the tribological performance of polymer materials, the lack of studies involving subsurface analysis is apparent, while cross section microscopy is widely used for other materials [22][23][24][25].We have previously shown the potential of cross section studies, where the deformation mechanisms of PPS with glass fiber at room temperarure were studied [16].
The field of polymer tribology still lacks an understanding of friction and deformation mechanisms, during initial conditions as well as in steady-state.A deeper understanding of these mechanisms needs to be reached in order to bring the science forward.
In this study, cross sections are used to study the subsurface deformation of PPS and PEEK composites at room and elevated temperatures.The effect of internal lubrication on tribological properties is also investigated.This is done by: i) studying how the addition of PTFE to a glass fiber PPS composite affects friction and surface damage, and ii) comparing the tribological behavior of two different polymer composites (PPS and PEEK), both containing internal lubrication.The influence of temperature and load is studied.

Materials
Two types of base polymers, from two manufacturers, with different types of fillers were investigated, see Table 1.
The base polymers were poly-phenylene-sulphide (PPS) and polyether-ether-ketone (PEEK).The fillers consisted of glass fibers (GF), carbon fibers (CF), poly-tetra-fluoro-ethylene (PTFE) and graphite (G).In this paper the low-friction additives, PTFE and G, are referred to as internal lubrication.The materials were produced as plates by injection molding, roughly 100×100×2 mm for the PPS based composites and 150×150×5 mm for the PEEK composite.PPS-F was studied using a micro CT from Bruker (Skyscan 1172).The final image was then reconstructed using software package NRecon, also from Bruker.The flow during injection molding creates zones of different fiber orientation, see Fig. 1.At the surfaces the fibers are aligned in the direction of the flow while they are more random oriented in the middle of the plate.The composite plates were cut into smaller pieces, approximately 60 × 20 mm, to fit in the sample holder.

Experimental set-up and test parameters
A reciprocating sliding ball-on-disc test set up was utilized for the tribological testing.Ball bearing steel balls of 100Cr6, with a diameter of 10 mm, and polymer plates were used.The contact load was 5 N and 15 N.The sliding distance (peak-to-peak) was 5 mm and the frequency was 3 Hz, giving a mean sliding speed of 30 mm/s.The sliding was oriented to follow the alignment of the fibers, giving more parallel oriented fibers during sliding.The friction force was recorded continuously and the tests were run for either 2 000 or 10,000 cycles.The tests are not designed to mimic any particular application but to gain more knowledge and understanding about the initial stages of wear for this type of materials.
Testing was performed at three different surface temperatures of the polymers: room temperature (RT), 80 • C and 120 • C. The heating was achieved using thermoelements mounted inside the disc sample holder, i.e. the whole polymer disc was heated.The temperature was measured at the polymer surface, and the thermoelements were calibrated to ensure a stable temperature over time.Measurements were also conducted both before and after each of the tests, to ensure a stable temperature.The additional increase in contact temperature due to friction was neglected, due to it being fairly low for this ball-on-disc test [26].The ball was loaded onto the polymer disc for 5 min prior to start of the test, to ensure a stable temperature of both contact surfaces prior to the test.

Analysis
The wear marks of the steel balls were studied with a light optical microscope (LOM).The polymers were investigated using scanning electron microscopy (SEM, Zeiss Leo1550) using an accelerating voltage of 5 kV, except for the cross sections of PEEK-L where 10 kV had to be used in order to get a good contrast from the carbon fibers and graphite particles.Prior to the SEM studies, the samples were cleaned with ethanol and coated with a thin conductive layer of gold/palladium, approx.5 nm.
Surface measurements were made for PPS-F and PPS-L using optical

Table 1
The materials investigated, with filler content and some properties specified.The materials are commercially available and the data is from the suppliers.vertical scanning interferometry (VSI, WYKO NT-1100).The topographical images were then reconstructed using software package Mountains Map 7.4.Cross sections of the three materials were prepared using a broad beam ion polisher (Gatan Ilion+ II), see Fig. 2 for a schematical image, following the same procedure as in [16].The samples tested at 120 • C, with 15 N contact load and run for 10,000 cycles, were chosen for these studies.In addition, also cross sections of PPS-F tested similarly but at RT were studied.The cross sections were prepared perpendicular to the sliding direction.The polishing procedure included irradion with 6 keV ions for 2 h and 4 keV ions for 20 h.The sample is rotated during milling and the ion beam is pulsed.This resulted in cross sections approximately 700 µm wide and 200-500 µm deep.
In addition, selected wear marks for the two PPS composites were studied using X-Ray Photoelectron Spectroscopy (XPS, PHI Quantum 2000).The wear marks from tests conducted at RT with a 5 N load and run for 10,000 cycles were analyzed.

Friction
PPS-F shows a high friction coefficient, around 0.4-0.5, while both PPS-L and PEEK-L with internal lubrication show relatively low friction, around 0.05-0.15,see Fig. 3.No obvious or strong influence of contact load, temperature or test duration is observed.
The two PPS composites follow the same trend at RT and 80 • C with a higher friction after 10,000 cycles than after 2 000.There is no major difference between the two temperatures.For PPS-F at 120 • C, the friction is lower after 10 000 cycles than after 2000 cycles.
The PEEK-L has marginally higher friction overall compared to PPS-L, but has the most stable friction over time when comparing friction at the end of the 2 000 and 10,000 cycle tests respectivley.
Typical examples of how the friction coefficient develops with the number of cycles are shown in Fig. 4. For all materials, the friction begins at the same level around 0.1.For PPS-F the friction increases rapidly over the first 1000 cycles, to a more slowly increasing level.The inital increase is slower at higher temperatures (not shown in any figure).For the materials with internal lubrication the friction is stable or even decreases during the test.The tests show good repeatability with the 2000 cycle tests having the same behavior as the 10,000 cycle tests for the first 2000 cycles.

Surface damage and characteristics
This chapter starts with a general description and comparison of the three polymer materials.Then follow separate sections describing each of the three materials.

General description
A general description can be formulated, irrespective of contact load, test temperature, or test duration.For PPS-F, the wear tracks are    damaged and rough, see Fig. 5a.In addition, there is wear debris and deformed material along the edges and/or at the turning positions.For PPS-L and PEEK-L the wear tracks are smooth, often smoother than the unworn surfaces, see Fig. 5b-c.Still, some wear debris is found at the turning positions, much less than for PPS-F.Surface measurements show the significant difference in surface damage between PPS-F and PPS-L, see Fig. 6.PPS-F has cavities in the wear track as well as material displaced to outside of the wear track, while PPS-L barely shows any signs of surface deformation.
After all tests, polymer has been transferred to the steel counter surface, see examples after 2000 cycles in Fig. 7. PPS-F causes mild wear of the counter surface already after 2000 cycles, which is seen as reflective areas in the wear mark and slight polishing of the original surface, see Fig. 7a.The transfer of material to the steel balls is more prominent at the lower temperatures and for longer tests.All wear marks on the steel balls, irrespective of wear, have lines running in the sliding direction due to transfer of polymer to the surface.
Higher load results in wider wear tracks for all materials, as exemplified in Fig. 8.For PEEK-L the wear track is about 350 µm at 5 N load and about 500 µm at 15 N load.For PEEK-L higher load also evens out the voids in the surface, which are due to the injection molding.For all materials and temperatures, the only effect of increased load on surface properties appears to be wider wear tracks.

PPS-F
The apperance of the steel ball counter surface was similar after 2000 cycles for all temperatures and loads.After 10,000 cycles there are still similarities, such as wear of the steel ball, see Fig. 9.However, the wear is worst at 80 ℃ and 5 N load, compared to the other temperatures, and 15 N load.
In the long duration tests, 10,000 cycles, and at all test temperatures, the polymer is subjected to plastic deformation with tongues of polymer in the wear track.There are also steel wear particles embedded in the polymer surface, see Fig. 10, confirmed by EDS analysis.After the tests run for 2000 cycles no steel wear particles could be detected in the polymer surface (results not shown here).The estimated roughness, from the SEM images, of the wear track increases with increasing temperature.At the higher temperatures, most pronounced at 120 • C, tongues of material are present at the turning positions after 2000 cycles, see Fig. 11, and a lot of small wear particles are found along the wear tracks after 10,000 cycles, see Fig. 12a.At 10,000 cycles the polymer surface (120 • C test) is heavily damaged, with exposed glass fibers, cavities and tongues, see Fig. 12.
The cross section after 10,000 cycles at 15 N contact load and 120 • C shows extensive plastic deformation and cracking, see Fig. 13.Some material is also seen on top of the original polymer surface (marked I and II in Fig. 13).To the right of the wear track (II), material has most likely been pushed out of the contact during sliding, comparable to what is seen in Figs. 10 and 12a.The material to the left is a mixture of fibers and polymer (I), and originates from a cavity with exposed fibers, comparable to what is seen in Fig. 12b.The fibers protrude from the polymer surface due to the sample mounting prior to cross sectioning.The fibers have stuck to the glue and been lifted up from their original location.
Repeated deformation of the polymer has also resulted in a 10-15 µm thick surface layer of plastically deformed material, see Fig. 13c.This mixed top layer is porous, which indicates that it consists of small polymer elements, indicating that the deformation occurs in small volumes at a time.This mechanism can also be observed in top view, see Fig. 12b.The surface appearance after the 5 N test is similar to that after

PPS-L
Also PPS-L has caused some wear of the steel ball at RT, see Fig. 16a.For all tests, there is a thin transfer layer in the wear scars on the steel balls.It is not prominent which is expected since PPS is known to form poor transfer layers on steel [15].
PPS-L shows smooth wear tracks, apart from some crevices, see Fig. 17.These crevices are located at the interface between the matrix, fibers and PTFE particles (easily located when using the EDS detector, not shown here), showing the poor adhesion between the different constituents.
The largest part of the surface in the wear tracks of the polymer consists of the polymer matrix, PTFE is only seen as particles.No PTFE film can hence be discerned in these top view SEM images, nor confirmed by EDS.
The cross sections show the distribution of fibers and PTFE particles, see Fig. 18.Neither cracks below the surface nor plastically deformed material can be seen, and no tribofilm could be detected in the cross sections, either by imaging or by EDS analysis.The difference in contrast is between the PTFE particles and fibers is larger here compared to the top view, as they are affected differently by the ion beam.

PEEK-L
All tests with PEEK-L result in transfer of polymer onto the counter surface steel ball, see Fig. 19.Most transfer of material occurs at RT.At  The wear tracks are smooth, as shown in Fig. 5c, and there is no significant influence of temperature or load.The cavities observed in the wear track originate from the manufacturing of the material, so similar features are present also outside of the wear track, see Figs. 5c), 6 and 15.A closer look at the wear tracks reveals the PTFE particles (which are confirmed by EDS analysis, results not shown here).These particles appear to be recessed, either by being pressed into the surface or subjected to more wear than the matrix, see Fig. 20.As for PPS-L some separation between the matrix and PTFE particles is visible.
In the cross section, all fillers are visible, see Fig. 21.A PTFE particle is present in the sliding surface.No surface or subsurface damage is seen.No tribofilm can be detected.
A summary of observed surface characteristics for all materials and temperatures is presented in Table 2.

Surface chemistry
The PPS wear tracks were analyzed using XPS mapping in order to detect any changes in chemistry and possible presence of tribofilms.Three areas were chosen for each wear mark, as illustraded in Fig. 22.No pre-sputter was used before the first analysis in order to preserve any tribofilms.

PPS-F
The analysis of PPS-F clearly shows that the wear mark contains Fe at the surface visible as an increase in intensity of Fe 2p 3/2 , see Fig. 23.The signal is about the same for all three positions.The middle area is also analyzed using a mild pre-sputtering (Ar + , 200 eV) in order to clean the

PPS-L
For PPS-L there is a higher signal from F 1 s in the wear track, than outside, see Fig. 25.This is an indication of a tribofilm of PTFE in the wear track.The middle area was also analyzed after using pre-sputtering (Ar + , 200 eV), resulting in an instant decrease in signal from F 1 s, see Fig. 26.After only 15 s the signal is barely detected in the wear track, and after another 15 s it is even lower.This clearly shows the existens of a very thin PTFE tribofilm, which is removed in a matter of seconds by the ion beam.The bright spots outside of the wear track correspond to PTFE particles located in the surface.As seen in Fig. 26 the intensity from the particles is more or less the same after sputtering due to the fact that they are thicker and not removed entirely by the ion sputtering.In another area nine analyses were conducted after another, in the same spot, showing that the PTFE in fact decomposes solely due to the X-ray radiation, see Fig. 27.This is an important aspect to take into consideration when conducting XPS analysis.

Discussion
Automotive components close to the engine have an expected life in the range of years.The tests performed in this work are rather short and show the initial behavior of candidate polymer materials for such applications.Higher contact loads than what is common in typical applications are used to accelerate the testing and hence get results in a shorter time.Nevertheless, the results are still valuable to reveal trends and to understand the behavior of these materials, which can be important for the final application.
One aim with this investigation was to study the effects of contact load and test temperature on tribological behavior.In general, the different materials show a greater variation between them than what contact load and temperature induce.Higher contact load gives somewhat wider wear tracks but there is no change in deformation mechanisms compared to lower load.The effect of test temperature is more inconsistent and no general conclusion can be drawn, except for PPS-F where higher temperature results in more extensive damages and degradation as well as fatigue of the polymer surface.
The PPS-F composite shows high friction and extensive wear.Several wear mechanisms are observed, including extensive shearing of the material, resulting in plastically deformed tongues exiting the wear tracks, as well as surface fatigue, resulting in large cavities.These surface characteristics are clearly visible in both the SEM images and the surface roughness measurements.The glass fibers strengthen the composite, but also cause abrasive wear of the counter surface, and consequently the increased steel surface roughness increases the wear of the polymer material.This also results in steel wear particles, which become embedded into the polymer surface.From the cross sections of PPS-F, it is clear that the wear tracks are heavily deformed.High friction and wear of the polymer has resulted in a deformed surface layer, as previously observed in [16].This mechanism does not cause much wear in terms of material loss, but it changes the surface morphology and appearance.This is due to polymer material either being transferred to   the counter surface and then back to the polymer or just displaced on the polymer surface.The sub-surface cracks at RT are probably due to the repeated deformation, which seems to cause delamination between the deformed surface layer and the bulk, similar to observations in [16].
Here the characteristics are not as clear as in [16], due to the use of BIB and application of a conductive coating on the sample prior to SEM.The detachment of fibers from the matrix also shows the high stresses the material is exposed to.These sub-surface features are most prominent at 120 • C, a temperature well above T g for PPS.Exceeding the T g means a change in mechanical properties and therefore more plastic deformation and shearing is expected.
The sub-surface crack running under the width of the wear track is an indication of fatigue, which can be correlated to the high friction, and thus high sub-surface stresses for this material.Since this type of cracks

Table 2
Surface characteristics of the wear tracks of the three polymer composites after 10,000 cycles.Numbers indicate the level of occurance: (-) -no observation, 1 -sparse, 2 -some, 3 -common, 4 -dominant.were found at both RT and 120 • C, and also with both 5 N and 15 N load, it is likely that there is crack formation below the surface for all tests for PPS-F.Crack formation finally leads to delamination and exposure of fibers in the wear track, and exposed fibers will lead to an increase in wear of the counter surface.
The presence of wear particles from the counter surface steel ball is confirmed with XPS.The fact that the signal from Fe 2p 3/2 increases after pre-sputtering indicates a thin polymer layer on top of the steel wear particles, which is removed by the ion sputtering.Both top view and cross section images reveal steel wear particles, both at the surface and embedded several µm deep.EDS also reveals steel wear particles in the surface and aids the visual interpretation of the SEM images, while XPS reveals wear particles from the counter surface steel ball on the surface itself.The combination of both techniques, along with cross sectioning, gives valuable information on the presence and location of steel wear particles.
When adding PTFE to the PPS, both friction and wear decreases.PPS-L shows very smooth wear tracks and any wear is negligible in the surface roughness measurements.In the cross section, no sub-surface damage is seen, nor have the fibers detached from the matrix, which was common for PPS-F.PTFE that lowers friction also reduces subsurface stresses and the fiber-matrix interface is intact.PTFE is known to be easily transferred between surfaces [18,20], and to generate a tribofilm in the wear track.This tribofilm is very thin and cannot be detected using EDS, neither in top view nor in cross section.On the other hand, using the more surface sensitive method XPS demonstrably reveals a PTFE film covering the wear track.PTFE is clearly affected by both ion and X-ray radiation.The mild pre-sputtering with Ar ions removes the tribofilm and the X-ray radiation during the XPS analysis breaks down the C-F bonds.Using pre-sputtering is common to remove surface contamination, but the fact that the entire tribofilm is removed with a mild and short pre-sputtering shows that the sputter rate is very high for PTFE.In contrast to PPS-F, the PPS-L has about the same behavior at all temperatures, showing that the PTFE compensates for the changes in mechanical properties when the PPS exceeds its T g .
Although adding PTFE to the PPS reduces friction and wear, the steel ball counter surface shows abrasive wear marks after 10 000 cycles at RT.The wear of the steel ball counter surface decreases with higher temperature, but nevertheless wear particles containing iron are found in all wear tracks of the polymer composite.The amount of steel wear particles in the polymer wear track is however only a fraction compared to in PPS-F.
PEEK-L behave similarly to the PPS-L.The friction is slightly higher but on the other hand there are no signs of wear of the counter surface.PEEK-L contains additional graphite, which is added to reduce the adhesion to metallic counterparts, although the effect of graphite is not investigated in this study [27,28].PEEK-L is below its Tg for all tested temperatures, hence no drastic differences are expected.The carbon fibers are less aggressive to the counter surface and cause no wear of the steel ball even after 10 000 cycles.PEEK-L, as well as PPS-L, shows smooth wear tracks and no sub-surface damages.Neither here, a tribofilm cannot be detected with EDS.Although the behavior with low friction and negligible surface damage is similar to that of PPS-L, the probability of the presence of a tribofilm is high.
From Table 2 it is clear that PPS-F shows a variety of deformation and surface damage mechanisms.There is sheared material, a modified surface layer, worn fibers and wear of the counter surface.PPS-L and PEEK-L rather show smoothing of the polymer surfaces in the wear track, and transfer of polymer to the steel ball.There are still small differences.PPS-L has more worn fibers and some steel wear particles embedded in the wear tracks while PEEK-L is more prone to transfer to the steel surface.The difference in wear is due to the difference in fibers, glass fibers act abrasively on the steel surface while the carbon fibers are    less abrasive [13].The difference in transfer tendency might be due to both the difference in fibers and polymer matrix.PEEK seems to form a more uniform transfer layer to steel surfaces than PPS, but the carbon fibers may also contribute as they are not as abrasive as glass fibers, meaning that they do not remove the polymer that is transferred to the steel counter surface.More studies need to be conducted in order to confirm this.
To sum up this study, PPS-F is a poor choice for tough tribological contacts.But it is necessary to study it in order to understand what role PTFE has in these types of contacts.It also gives information about what mechanisms are present in tribological contacts for these types of materials.PTFE is a good choice of additive to a polymer composite in order to achieve low friction and wear.The difference in performance between PPS-L and PEEK-L is small in this study, but it is known by users that PEEK-L is able to handle more demanding contacts, e.g.where the working temperature is higher or where it is very important not to cause wear on the counter surface [8].

Conclusions
The difference between two glass fiber reinforced PPS composites, one containing PTFE, has been investigated at different temperatures and different loads.In addition, a fiber reinforced PEEK, also containing PTFE and graphite, was used as a reference.The following general conclusions can be drawn: • Neither temperature nor load had a strong influence on the friction.
• PTFE had a very strong influence on friction for PPS, lowering the friction coefficient to around 0.05-0.15compared to 0.4-0.5.• None of the polymer composites containing PTFE showed any subsurface damage.PTFE also affects the surface damage, where the wear track on PPS-L was barely identified while the surface damage of PPS-F was severe.• PTFE formed a thin tribofilm in the wear tracks, revealed by surface sensitive XPS analysis.• PPS-F showed high friction and severe surface damage, with extensive wear of the steel counter surface, and consequently more damage of the polymer.Exceeding the T g further increased the surface damage.SEM studies of the surface and cross section of the wear track revealed severe plastic deformation and fatigue.• Cross sections, produced by BIB and analyzed with SEM, revealed subsurface cracks as well as a modified surface layer for PPS-F, both at RT and 120 • C. • Although PPS-L showed marginally lower friction coefficient, the PEEK-L surface was less damaged and so was the mating steel ball.Nevertheless, PPS-L showed superior properties compared to PPS-F and can be a viable option for many applications.

Statement of originality
As corresponding author, I Jonna Holmgren, hereby confirm on behalf of all authors that: 1) The authors have obtained the necessary authority for publication.2) The paper has not been published previously, that it is not under consideration for publication elsewhere, and that if accepted it will not be published elsewhere in the same form, in English or in any other language, without the written consent of the publisher.3) The paper does not contain material which has been published previously, by the current authors or by others, of which the source is not explicitly cited in the paper.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1. : Micro CT of a polymer plate of PPS-F.The image represents the whole thickness of the polymer plate, which is 2 mm.The bright parts are the glass fibers, while the polymer is transparent in CT (dark contrast).The fibers are aligned at both surfaces, and more randomly oriented in the middle of the plate.The arrow indicates the sliding direction.

Fig. 2 .
Fig. 2. : Schematical image of the broad ion beam set-up, showing the direction of the beam, with geometry and position of the final polished cross section.Picture previously published in[16].

Fig. 4 .
Fig. 4. : Typical examples of friction coefficeint curves for the three materials, tested at 80 • C with 15 N contact load and run for 2000 (filled markers) and 10,000 (unfilled markers) cycles.

Fig. 5 .
Fig. 5. : Typical wear tracks after tests at RT with a contact load of 5 N and 2000 cycles.Arrows indicate sliding direction.a) PPS-F, b) PPS-L and c) PEEK-L.SEM, 5 kV, sample tilted.

Fig. 6 .
Fig. 6. : Surface roughness measurements of a) PPS-F and b) PPS-L after test at RT, 5 N and 10,000 cycles.PPS-F shows significant wear while the wear track for PPS-L is hardly visible.The ellips indicates the placement of the wear track for PPS-L.

Fig. 7 .
Fig. 7. LOM of the steel counter surfaces tested at RT, with 5 N load and run for 2000 cycles.a) PPS-F, b) PPS-L and c) PEEK-L.Compared to Figs. 9a), 16a) and 19a), all run for 10,000 cycles, it clearly shows smaller wear marks and also less polymer transfer to the steel ball.Arrows indicate sliding direction.

Fig. 8 .
Fig. 8. : Typical wear tracks of PEEK-L, after 2000 cycle tests at RT: a) 5 N load (exactly the same region as in Fig. 5c) and b) 15 N load.The black lines show the width of the wear tracks, approx.350 µm at 5 N load and 500 µm at 15 N load.White arrows indicate sliding direction.SEM, 5 kV, sample tilted.

Fig. 9 .
Fig. 9. : Appearance of the wear marks on the steel balls after testing against PPS-F.10,000 cycles and 5 N load.a) RT, b) 80 • C and c) 120 • C. In all cases the steel ball is worn, most prominent at 80 • C. Patches around the wear mark are polymer debris.At higher temperaturs there is less polymer transfer to the steel ball.The bright surface in the wear mark is abrasivley worn, although in a) there is also some polymer covering the worn surface.Arrows indicate sliding direction.LOM.

Fig. 10 .
Fig. 10.Steel wear particles are embedded in the polymer surface of PPS-F at 120 • C, with 5 N load and run for 10,000 cycles.White arrow indicates sliding direction.SEM, 5 kV.

Fig. 11 .
Fig. 11.: Appearance of wear track on PPS-F after testing at 120 • C, 15 N contact load and 2000 cycles.Sheared, more or less loosened tounges of material can be seen along the edges of the wear track.Arrow indicates sliding direction.SEM, 5 kV, sample tilted.

Fig. 12 .
Fig. 12. : Appearance of wear tracks on PPS-F after testing at 120 • C with 15 N contact load and 10,000 cycles.a) Exposed glass fibers, cavities, tongues and wear debris.Small box indicates area that is zoomed in in b).b) Fibers exposed by delamination of the polymer matrix.Right arrow indicates sliding direction.SEM, 5 kV, sample tilted.

Fig. 13 .
Fig. 13.: Cross section perpendicular to the sliding direction under the wear track on the PPS-F after testing at 120 • C, 15 N contact load and 10,000 cycles.SEM, 5 kV.a) Overview of the cross section.The green marker indicates the width and position of the wear track.A subsurface crack runs at a depth of roughly 10-30 µm across the whole wear track, not fully visible in the image.b) Region where fibers have been lifted off the polymer surface by the glue used in the sample preparation before cross sectioning.c) Fibers, a sub-surface crack and deformed material, close to the surface.

Fig. 14 .
Fig. 14. : Cross section of the wear track of PPS-F at RT, 5 N contact load and 10,000 cycles.Also here, fibers with associated cracks are observed under the surface, as well as cracks below the wear track.Steel wear particles are observed embedded in the polymer surface.Cross section is perpendicular to the sliding direction.Dashed box indicates area for Fig. 15.SEM, 5 kV.

Fig. 15 .
Fig. 15.Close-up of the steel wear particles that are embedded into the polymer surface.Cracks can also be seen.PPS-F at RT with 5 N load and run for 10,000 cycles.SEM, 5 kV.

Fig. 16 .
Fig. 16.: LOM of the steel counter surfaces after testing against PPS-L.10,000 cycles and 5 N load.a) RT, b) 80 • C and c) 120 • C. At RT some wear of the steel ball has occurred.Less transfer of polymer material to the steel ball at higher temperatures.Patches around the wear mark are polymer debris.The darker stripes in the wear mark relates to the transfer layer formed on the steel ball surface.Arrows indicate sliding direction.

Fig. 17 .
Fig. 17. : The worn surface of PPS-L after testing at 120 • C, 15 N contact load and 10,000 cycles.The surface shows several cracks and a piece of a fiber sticking out.Arrow indicates sliding direction.SEM, 5 kV, sample tilted.

Fig. 18 .
Fig. 18. : Cross section in th middle of the wear track, of PPS-L at 120 • C, 15 N contact load and 10,000 cycles.Fibers and PTFE particles are indicated.No indication of plastic deformation.Cross section is perpendicular to the sliding direction.SEM, 5 kV.

Fig. 19 .
Fig. 19.: LOM of the steel counter surfaces after testing against PEEK-L.10,000 cycles and 5 N load.a) RT, b) 80 • C and c) 120 • C. No wear of the steel balls.Most transfer of material occurs at RT.At 80 • C and 120 • C the transfer is roughly the same.Black patches are larger debris of polymer.The darker stripes relates to the transfer layer formed on the steel ball surface.Arrows indicate sliding direction.

Fig. 20 .
Fig. 20.: The smooth wear track of PEEK-L, after testing at 80 • C, 15 N contact load and 2000 cycles.PTFE particles are recessed, as indicated by white arrows, and the cracks show the poor adhesion between the matrix and the PTFE particles.The cavities are from the manufacturing process.Black arrow indicates sliding direction.SEM, 5 kV, sample tilted.

Fig. 21 .
Fig. 21.: Cross section in the middle of the wear track of PEEK-L at 120 • C, 15 N contact load and 10,000 cycles.Fibers, PTFE and graphite particles can be seen.Some PTFE particles are exposed at the surface.No indication of plastic deformation, the cracks between the PTFE particles and polymer matrix only show the poor adhesion between the two.Cross section is perpendicular to the sliding direction.SEM, 10 kV.

Fig. 22 .
Fig. 22.Schematic view over the wear mark and the analyzed areas relative to the wear mark.The analyzed area is about 1200 × 50 µm.

Fig. 23 .
Fig. 23.XPS maps, for Fe 2p 3/2 , of the wear track for PPS-F.The wear track contains Fe, showing that worn fragments of the steel ball is present in the whole wear track.No pre-sputtering.

Fig. 24 .
Fig. 24.XPS maps, for Fe2p 3/2 , of the wear track for PPS-F, middle position.The use of pre-sputtering removes contamination of C and O, and the Fe signal is more intense.

Fig. 25 .
Fig. 25.XPS maps, for F1s, of the wear track for PPS-L.The wear track contains F, showing that there is a continous tribofilm of PTFE in the wear track.Also PTFE particles present at the surface give a signal.

Fig. 26 .
Fig. 26.XPS maps, for F1s, of the wear track for PPS-L, middle position.The F1s signal in the wear track decreases with pre-sputtering, while the signal from the PTFE particles in the surface remains about the same.

Fig. 27 .
Fig. 27.XPS analyses of the wear track for PPS-L.Nine succeding measurements were done in the same spot.There is a clear signal loss for C-F with number of analysis cycles, while the C-C signal remains about the same.