Large-scale tribological characterisation of eco-friendly basalt and jute fibre reinforced thermoset composites

The present research aims at understanding the tribological behaviour of advanced unsaturated polyester/vinyl ester based thermoset composites reinforced by inorganic (mineral-based) or organic (vegetal) fibres such as basalt and jute. These fibres are non-toxic and widely available in nature. Thermosets have limitations in the formation of a uniform transfer layer during sliding wear. To surpass these limitations, tribo-fillers such as polytetrafluoroethylene, polyoxymethylene or molybdenum disulphide (PTFE/POM/MoS2) are added into the contact surface. The composites developed for the current research are characterised for their friction and wear behaviour, using a large-scale (sample size typically 50 x 50 x 7 mm) linear reciprocating sliding flat-on-flat test configuration. In order to simulate real scale application, 50 mm/s sliding speed and 10 kN normal force which corresponds to 4 MPa contact pressure, are applied under dry contact conditions. In this research work 12 different tribocomposites are developed and tested against AISI 100Cr6 steel counterface. It was evidenced that composites blended with PTFE have the lowest coefficient of friction and longest service life. MoS2 filled tribocomposites have the highest coefficient of friction. The dominant wear mechanisms for the failure of all investigated composites are thermal degradation and delamination, and abrasion for the counter surface.


Introduction
Polymers are widely used in large scale applications such as bearings, linear slides, gears or guideways under both dry and lubricated conditions [1][2][3][4]. Semi-crystalline thermoplastics, such as e.g. polyoxymethylene (POM), polyether ether ketone (PEEK), polyamide 6 (PA6) and polytetrafluoroethylene (PTFE) are beneficial due to their ability to form an adequate transfer layer with uniform thickness on the counter surface [5][6][7][8]. This transfer layer can significantly improve the coefficient of friction and wear resistance on the condition that there is not too high adhesion between the formed transfer layer and the polymer sample [9]. Thermosets have some other advantages: they can withstand heavy loads and shock loading and they have a level. The transfer layer formed on the counter surface was also investigated as a function of the applied tribo-fillers.

Test materials and mechanical characterisation
In this research work twelve different polymer tribocomposites were investigated. They were produced by hand lay-up technique at the Centre for Composite Materials, Kalasalingam University (India). The basalt and jute fabrics were impregnated with resin/curing agent mixture and stacked on each other until reaching the required thickness. Wax was used as a release agent. The laminated composites were compressed by means of dead weights, subsequently cured at room temperature (32 °C) for 24 h and finally post-cured at 70 °C for 3 hours in hot air oven. No further operations were performed on the contact surface of the specimens. All samples for mechanical and tribological tests were cut out by water jet machining. The structure of these composites can be divided into two parts: the bulk material and the top layer, both including fibre reinforcement (textile) layers. This material combination was developed based on the experience from an initial investigation [35]. The bulk material has a purely mechanical role, while the top layer that is in contact with the counterface during wear testing, fulfils mainly a tribological role. Consequently, only the top layer should contain tribofillers. The bulk material does not include these tribo-fillers and as such does not suffer potential negative influence of these lubricants on its mechanical properties. The thickness of the tribological top layer corresponds to the thickness of one applied reinforcement layer, which is around 1 mm. The composites developed in the present research work are divided into four different composite systems (see Table 1). The matrix material is unsaturated polyester (UPE) or vinyl ester (VE) resin, which was reinforced with basalt (B) or jute (J) textile layers. The tribo-fillers were PTFE, POM and MoS2 in 2 or 4 wt% filler content. This range of filler contents is based on literature survey and is expected to be sufficient to promote adequate transfer layer formation. Higher tribo-filler content was not considered because it could reduce the mechanical properties significantly; e.g. 7.5 wt% of PTFE can reduce the compressive strength of epoxy composites with 21%, while 12.5 wt% leads to 37% strength reduction [36]. Similar tendency was reported for the hardness [36]. The tribo-fillers were added only to the top layer. The fibre content of basalt was chosen 40 to 45 wt% and of jute between 25 to 30 wt%. Both unsaturated polyester and vinyl ester thermoset matrices and the different tribo-fillers (PTFE, POM and MoS2) were ordered from Sigma Aldrich Ltd., Bangalore, India. For the curing purpose, both Methyl Ethyl Ketone Peroxide (MEKP) and cobalt naphthenate were used as catalyst and accelerator, respectively. The basalt and jute reinforcement were purchased from Nickunj Eximp Entp Ltd., Chennai, India. Tensile and flexural tests were performed on the tribocomposites according to ASTM-D 3039 and ASTM-D 790 standards using an Instron 3382 universal testing machine. All tests were performed at room temperature with 5 mm/min cross-head speed. Specimen geometry of 200 x 20 x 3 (mm) was used for tensile tests with a grip distance of 50 mm, and 127 x 12.7 x 3 (mm) geormetry for flexural tests with a 90 mm of span distance. The hardness was measured with a Zwick H04.3150.000 digital hardness tester in Shore-D measurement range. The mean and standard deviation values were calculated from at least 5 measurements.

Tribological characterisation
The polymer tribocomposites were tribologically characterized by means of a large-scale linear reciprocating sliding flat-on-flat tribotester with dry contact condition. Figure 1 shows the heart of the tribotester. A central support block at two sides houses the metal counterfaces. The two polymer samples, fixed in appropriate holders, are pressed against the sliding block that moves up and down. A significant benefit of this equipment is that its size is close to real application. More detailed information of this equipment can be found in a previous publication [37]. The 100Cr6 steel counterfaces were polished to 0.2 µm Ra roughness value. The counterface size was 200 x 80 x 20 mm, while the size of the polymer samples was 50 x 50 x 7 mm. Wear tests were performed in a conditioning chamber which maintains a uniform 23 °C temperature and 50% relative humidity (RH). The bulk temperature resulting from frictional heating was measured with thermocouples which are placed in the steel counterface and located at 10 mm distance from the contact surface. Considering heavy duty applications, all specimens were tested with 10 kN normal force, which corresponds to 4 MPa average contact pressure. The applied sliding speed was set to 50 mm/s, the stroke was 100 mm and the stipulated sliding distance was 5000 cycles (corresponding to 1000 m). Independent tests were performed at least three times under identical test conditions to study the uncertainty from the tribotester where a deviation (±1σ) of 10% in coefficient of friction and 20% in wear rate was observed. The online recording was provided with the use of NI 6036E DAQ (National instruments BNC 2100) in a LabVIEW platform. Data are sampled at a frequency of 500 Hz. The static and dynamic coefficient of friction are derived from the results of every logged friction cycle. A typical cycle is shown in Figure 2. The sign switch of the force halfway the cycle indicates the reversal of the sliding motion direction. The maximum of the absolute values in the first and second stroke length are defined as the static coefficient of friction. The dynamic coefficient of friction is the average value of the centre (red marked) region in each stroke. The coefficient of friction is calculated with the following equation: µ symbolises the calculated coefficient of friction [-], FFr is the measured friction force [N] and FN is the applied normal force [N]. The factor 2 in this equation is needed to take into account the two friction faces.

Figure 2. Friction cycle measured during reciprocating tribotesting
To calculate the average friction values, first the static and dynamic coefficient of friction were evaluated in every stroke (half cycle) as shown in Figure 2. The average static and dynamic coefficient of friction values are calculated at an interval between 20% and 80% of the lifetime (number of cycles). The reason of this method is that some of the materials did not reach steady state friction as they failed after a low number of cycles. By defining an interval between 20% and 80% of the lifetime, the effect of the initial and end period are minimised and the materials can be compared in a more precise way. Figure 3 shows an example of a friction curve with this interval indicated. The surface roughness of the steel counter surface was scanned before and after wear testing by means of stylus profilometer (Surfascan 3D roughness tester, Hommel somicronic) with a stylus S6T (radius 2 μm, angle 90°). The average values and the standard deviations were calculated from 5 roughness measurements per sample. The measured roughness values were evaluated according to ISO 4288 standard with an assessment length lt = 4.00 mm and cut-off wavelength λc = 0.80 mm for 0.1 μm < Ra ≤ 2 μm. For the micrographs an Olympus reflected light bright field optical microscope was used. The image acquisition parameters were kept constant (150 μs exposure time and 30% illumination) for monitoring the transfer layer deposition.

Mechanical properties of the tribocomposites
The results of the static mechanical and hardness tests are shown in Table 2. Jute fibre reinforced composites had slightly lower hardness compared to basalt composites with the same 2 wt% tribo-filler content. The highest strength values were reached with the vinyl ester matrix. Jute fibre reinforcement results into significantly lower tensile and flexural strength compared to basalt fibres. This remarkable difference comes from the mechanical properties of jute and basalt fibres. The tensile strength and elastic modulus of jute fibre are 0.3-0.7 GPa and ~26 GPa, while for basalt fibre these values are ~2.8 GPa and ~89 GPa, respectively [21]. Significant difference in tensile and flexural strength was not registered as a function of the applied tribo-fillers. The measured differences were at the range of the standard deviation comparing PTFE, POM and MoS2 fillers. It means that focusing on the tensile and flexural strength of the composites all tribo-fillers show similar performance. This indicates that in mechanical viewpoint all PTFE, POM and MoS2 can be used with the same expectations on the mechanical strength in these composites as tribo-fillers.

Lifetime and coefficient of friction
The operational variables during wear testing were kept constant for all 12 tested materials.
As it can be seen in Table 3 Table 3 shows the total number of cycles for each material, the average static/dynamic coefficient of friction, the maximal (bulk) temperature and the measured bulk temperature at cycle 50. The maximal (bulk) temperature is the highest measured temperature in the steel counterfaces during wear tests. The highest temperature was registered at the last cycles of the tests (Figure 4). The last column of Table 3 introduces the bulk temperature after the same sliding distance (50 cycles corresponding to 10 m). As UPE/B/POM/2 and UPE/B/MoS2/4 did not reach 50 cycles, the bulk temperatures registered at the last cycle are shown. From Table  3 it is clear that the static coefficient of friction was higher than the dynamic coefficient of friction for all tested samples, as expected. It is also shown in Table 3 that some samples reached a relatively low maximal (bulk) temperature during the wear test, which can be explained by the lower heat generation due to the lower lifetime.   All materials failed due to thermal degradation and delamination which comes from the high contact temperature and from the intensive shear stress during testing. As a result of the thermal degradation of the thermoset matrix resin, the mechanical strength and adhesion between the layers decreased and consequently the top layer was delaminated. Delamination is a typical failure method in layered composites [38]. In case of jute reinforced composites more significant delamination was observed than with basalt composites.     It can be seen that the bulk temperature curve moves together with the dynamic coefficient of friction; the higher friction values increased the temperature, which led to thermal degradation of the matrix and simultaneously to the reduction of the mechanical properties of the composites, finally resulting into failure.
(a) (b) Figure 9. Dynamic coefficient of friction (Dynamic CoF) and bulk temperature of UPE/B/PTFE/4 (a) and VE/B/PTFE/4 (b) composites. AISI 100Cr6 steel counterface, 4 MPa contact pressure, 50 mm/s sliding speed, 100 mm stroke. The black curve is the moving average of the dynamic coefficient of friction with a period value of 10.

Transfer layer analysis
One of the key factors in polymer tribology is the formation of a uniform and adequate transfer layer during the wear process. To have a deeper understanding of the transfer layer, the worn steel counterfaces were investigated at both micro-and macroscale. Figure 10 shows some macrographs of samples with different tribo-fillers (PTFE, POM, MoS2). The transfer layer is clearly visible. Literature makes distinction between the primary and the secondary transfer layer [39]. The primary transfer layer is considered as a deposit between the asperities of the metal counterface. The secondary transfer layer is the result of a step by step deposition process, formed on top of the primary transfer layer. The white coloured primary transfer layer can be seen in (a) and (b), generated by friction of UPE/B/PTFE/2 and VE/B/POM/4 respectively. This white colour of PTFE and POM is well reflected in the transfer layer. Figure 10 Figure 11 displays the micrographs of the tested steel counterfaces at x20 magnification. Each micrograph was taken at 30% intensity of Xenon illumination and exposure time of 150 μs in order to keep control of discoloration of the contact surfaces. The micrographs show the worn surfaces with the formed transfer layer and the local wear scars. Figure 11 (a) -(f) shows areas where significant primary transfer layer was supposed. Micrograph (c) and (f) introduce a darker background caused by MoS2 filler. These pictures also confirm the existing of a fillerrich transfer layer. Figure 11 (g) -(l) were taken from the areas with abrasive wear scars, which are parallel with the sliding direction. The perpendicular lines come from the polishing method of the steel counterfaces, previous to wear testing. MPa contact pressure, 50 mm/s sliding speed, 100 mm stroke. Table 4 and 5 introduce the Rz (peak-to-peak) and Rku (Kurtosis) surface roughness values of the steel counterfaces respectively. Rz and Rku were measured in the sliding direction (excluding the areas of scratch marks) to study the influence of the transfer layer formation. In Table 4 and 5 both unworn (before wear) and worn (after wear) surface values are given. From Table 4 can be seen that Rz decreased in all cases, but mostly the difference between the before and after wear tests was in the range of the standard deviation or close to it. In case of jute reinforced samples, a higher difference of Rz can be seen. Rku values are also slightly reduced due to wear (Table 5) in a similar range or also close to the standard deviation. This slight difference in Rz and Rku values between worn and unworn steel surfaces indicates that the formed transfer layer on the steel surface did not influence the original surface pattern significantly (Figure 11 (a)-(f)).

Conclusions
In this research work PTFE/POM/MoS2 filled tribocomposites were investigated. From the results of this characterisation, the following conclusions can be drawn: 1 Focusing on the tensile and flexural strength PTFE, POM and MoS2 tribo-fillers show the same mechanical performance in the investigated tribocomposites. On the other hand in tribological viewpoint significant differences were registered. Semi-crystalline thermoplastic tribo-fillers as PTFE and POM reached a lower coefficient of friction than the lamellar structured MoS2. 2 PTFE filled samples showed the lowest coefficient of friction which comes from the excellent lubricating behaviour of PTFE. Due to the low coefficient of friction the frictional heating was also moderated and consequently PTFE filled tribocomposites had the longest lifetime. 3 Specimens with vinyl ester matrix had higher lifetime compared to the same composites based on unsaturated polyester. This can be attributed to the advanced tensile and flexural strength of vinyl ester composites. The longest lifetime was found for VE/B/PTFE/4 with 1364 cycles. 4 The failure method for all tested tribocomposites was matrix thermal degradation and delamination. This phenomenon comes from the intensive shear stress during wear process and from the high frictional heating which degraded the thermoset matrices decreasing their mechanical properties. 5 Transfer layer formation generated from PTFE/POM/MoS2 tribo-fillers was clearly observed for steel counterfaces both in micro-and macrographs. Primary and secondary transfer layer were registered. The transfer layer formation is beneficial as it enables improved coefficient of friction and increased lifetime. Abrasive wear scars were also observed on the steel counterfaces but in case of PTFE solid lubricant it is less significant. The formed transfer layer did not change the dominating surface characteristic of the steel counter surfaces.
Based on the results of this research work finding the optimal tribo-filler ratio and the hybridization of this composite materials could be considered to further improve the wear properties.

Acknowledgement
This investigation was supported and funded by Fonds Wetenschappelijk Onderzoek -Vlaanderen (FWO) 12S1918N. This research was performed under a cooperative effort between Ghent University (Belgium) and Budapest University of Technology and Economics (Hungary). Laboratory Soete acknowledges the material support received from Centre for Composite Materials, Department of Mechanical Engineering, Kalasalingam University (Krishnankoil-626126, Tamilnadu, India). The authors are grateful to Sam Nowé and Wouter Ost for their support on experimental activities.