Journal Pre-proof A pin-on-disc study on the dry sliding behavior of a Cu-free friction material containing different types of natural graphite

This research investigates the influence of graphite’s granulometry on the dry sliding behavior of a copper-free friction material against pearlitic cast iron. Samples were designed and fabricated using three different types of commercial natural graphite. A sample without graphite was also considered as a reference. Tests were carried out with a pin-on-disc tribometer at room temperature (RT), at 400°C, and at RT after the high temperature tests. The results show that both the shape and size of the graphite particles influence the coefficient of friction and the specific wear rate. The friction material featuring a lower particle size and equiaxed grains of natural graphite exhibits a better behavior, as compared to coarser graphite with plate-like grains. The results were obtained comparing specific characteristics (i.e., morphology and chemical composition) of the friction layers formed on each friction material under the different testing conditions.


1.INTRODUCTION
An important research topic regarding brake pads for passenger cars is related to recent regulations aiming at reducing the use of copper. Indeed, the U.S. legislation currently requires copper to be reduced down to 0.5% by 2025 [1]. Nevertheless, copper has important properties, such as high thermal conductivity and the ability to stabilize the friction coefficient, that are beneficial to the brake pad performances [2] [3]. Therefore, in view of the ban of copper, potential alternatives able to counterbalance its replacement need to be investigated.
Organic friction materials used to produce automotive brake pads are made of several constituents [4].
The friction material components, often more than 20, are classified according to their main role as: reinforcements, binders, fillers and friction additives [5] [6]. The friction additives can be classified into two further types: lubricant and abrasive. Lubricants are friction coefficient stabilizer and help regulating the generation of the friction layer, therefore contributing to wear reduction. A smooth sliding behavior, i.e. the stabilization of the friction coefficient, is essential for good brake performances. In these regards, various metal sulfides and graphite are commonly used in the brake pad formulations [7].
Among these, graphite is regarded as one of the strategic constituents in the development of copperfree formulations for different reasons [8]. First, its high thermal conductivity, between 350 and 250 WK −1 m −1 , i.e., is close to the thermal conductivity of copper, although, copper is far less sensitive to temperature than graphite [9]. Moreover, studies on the contact area performed by Österle et al. [10], suggest that soft inclusions, like graphite in the friction layer, as well as copper, result in the stabilization of the friction coefficient, and induce a smoother sliding behavior.
Several studies on graphite are reported in the literature for copper-based friction materials [11] [12][13], in which graphite is one of the most important components. Different types of graphite are used as ingredients, either synthetic or natural, and with different grain morphologies [14]. Graphite is also widely studied in organic friction materials, mainly in those that also contain copper.
In general, it can be stated that not every form of graphite is suitable to improve the friction material performances [15]. Synthetic graphite has been broadly studied, considering different aspects, including its influence on the sensitivity of friction coefficient and wear rate to braking pressure and sliding speed [16], and its noise emission contribution [17]. The studies, however, suggest more than a range of synthetic graphite particle size, both coarse and fine. Also thermal conductivity has been extensively investigated, as concerns especially treated graphite [18] [19], to reach a satisfactory value. Concerning natural graphite, a few studies on its properties as a component in friction materials can be found in the literature, mostly in comparison with a synthetic counterpart. D.K. Kolluri et al. [20] studied the effect of fine particle granulometries on the thermal localization phenomenon in a composition involving also brass, finding that the particle size is the main parameter which influences the friction coefficient. However, wear rate was not taken into consideration for these materials.
Although graphite is regarded as a possible replacement for copper, a few research studies only have been dedicated to study graphite in Cu-free formulations, and just a few of them investigated the influence of graphite on the relevant tribological properties (friction and wear) at different testing temperatures.
The present research used a pin-on-disc tribometer to correlate specific test conditions, including the temperature of the disc, to fundamental tribological mechanisms in Cu-free base formulations, each one of them prepared by adding to a reference masterbatch a given concentration of different types of natural graphite, featuring different particle size and morphology. The tribological performances of the novel formulations were comparatively investigated on a pin-on-disc (PoD) tribometer. Notoriously, this set up cannot replicate real brake system action, for which specific types of experimental equipment, e.g., dyno-bench tests, are available. However, PoD can comparatively investigate in a timely manner the influence that specific composition and test conditions have on the material performances. Tests were performed at room temperature (RT) and 400°C. Concerning RT, the tests were conducted also on the sample already tested at 400°C. This was meant to investigate the recovery capability of the diverse friction materials. Microscopy observations and elemental analyses were conducted on the friction layer that builds upon the worn pin's surface, to obtain a solid interpretation of the tribological results.

MATERIALS
A masterbatch of a low-metallic friction material not containing graphite was used as a reference for this study. The composition of the material, measured by energy dispersive X-ray spectroscopy (EDXS), is shown in Table 1. The main components can be classified according to standard categories as: reinforcing fibers (steel fibers); friction additives (aluminum oxide, tin sulfide, magnesium oxide, zinc oxide, coke); fillers (vermiculite, calcium carbonate); binder (phenolic resin).   Three different types of natural graphite, featuring a diverse particle size, were selected. In Fig. 2 the scanning electron microscopy (SEM) micrographs with the morphology of these three types of graphite are shown. The graphite # 1 is a flake graphite with the particle size in the 300 -425 µm range.
The graphite # 2 has a flake morphology, with the grain size below 100 µm. The graphite # 3 is a fine granular graphite with particle size below 40 µm. The main features of the selected graphite are summarized in Table 2.  The powder mixtures were prepared by adding 9 wt% of graphite to the reference masterbatch, mixermilled for 20 minutes using a Turbula shaker-mixer. The concentration of graphite is representative of concentrations used in other studies on friction material, with a particular focus on graphite [21][17]. In Table 3   Each tribological test specimen was prepared by pressing roughly 2 g of powder mix in a mold of a hot-pressing apparatus with a pressure of 17 MPa. In this way, cylindrical pins with a diameter of 10 mm and a height of 10 mm were obtained. The samples were cured for 15 min at 150°C, and successively post-cured in an oven for 7h at 200°C in air.
The density of the pins, evaluated as the mass of each pin divided by its volume, was 2.29 ± 0.06 g/cm 3 for the Ref. and 2.18 ± 0.04 g/cm 3 for the pins with graphite (N1, N2, N3).
Discs made of pearlitic grey cast iron with a diameter of 60 mm and 6 mm in height, with the hardness of 235 HV10 were used as counterface for the tribological tests.

EXPERIMENTAL PROCEDURE
The friction and wear properties of the samples were firstly characterized running room temperature (RT) PoD tests. The tribometer used for the tests, equipped with a heating system for the disc, was manufactured by Ducom Instruments.
The sliding velocity was 1.5 m/s and the nominal contact pressure was 1 MPa in all cases. These conditions, used in previous investigations, are intended to obtain mild wear sliding conditions [22][23].
The duration of each test was 90 min. and before each test a 30 min. run-in period was performed.
Four tests were conducted for each sample at room temperature.
High temperature tests at 400°C (HT) were carried out using an induction heating apparatus to heat up the disc. A closed loop feedback control system kept the temperature constant throughout the test, using an infrared high temperature sensor. It is important to note that the HT tests were started after the disc temperature reached 400°C. During the heating-up of the disc the pin was lifted, and repositioned only once everything was ready to start the test, including a stable temperature. The duration of these tests was 60 min. only, instead of 90 min., because of the more demanding conditions.
The HT tests were then followed by further room temperature tests, still on the same specimens: RT_after HT test. Two test runs were conducted for the evaluation of the specific wear rate. For this purpose, each pin had to be removed from the testing equipment when passing from the HT to the RT_after HT test, since its weight had to be measured at the beginning and at the end of each test.
Two more tests were conducted for a more reliable observation of the evolution of the friction coefficient. In this latter case, the pin was not removed from the test rig for the whole duration of the test sequence.
The friction coefficient was continuously recorded using a load cell. The wear rate was assessed by weighting the test sample before and after each run with an analytical balance having a 10 -4 g sensitivity. The wear of the pin was estimated using the specific wear coefficient (K a ), given by: 3 ] is the measured wear volume, calculated using the bulk density;  After an initial run-in stage with a duration of approximately 2500 s, all samples reached a steadystate, whose relevant mean friction coefficients are listed in The experimental specific wear coefficients (K a ) are also summarized in Table 4 and displayed for a better comparison in Fig. 4. The Ref. and N1 materials show higher K a values than N2 and N3, and also larger scatters, as evidenced by the standard deviation bars. Material N3 displays the lowest wear rate and steady-state friction coefficient.   increasing again up to a second peak, and then progressively decreases down to the steady-state value (the obtainment of a steady-state after approximately 3000 s of sliding was observed also during longer tests, not reported herewith). This particular behaviour was observed in all the friction materials prepared for the present study.

FRICTION AND WEAR BEHAVIOUR
After cooling the disc down to RT, the PoD test was restarted to check how the HT test had influenced the friction and wear behaviour of these materials (Fig. 5b). In this case, quite a long run-in stage was observed, with a continuous raise in the friction coefficient, again followed by steady-state conditions which were reached close to the end of the test. The experimental values of the friction coefficient once the steady-state was achieved, are indicated in Table 5.
These combined tribological tests, involving a high temperature followed by a room temperature run,

Eq. (3)
The change of µ (%µ HT and %µ RT_afterHT ) are also included in Table 5. The comparative graphs in Fig. 6 show how the graphite influences these changes in the friction coefficients. All tested materials show a similar reduction in µ after the test at HT. The largest decrease, 22%, was found with the N1 sample.
The situation is different for the RT tests carried out after the HT tests. The Ref. sample actually recovers its µ (%µRT_afterHT= 2%), whereas all other samples exhibit an incomplete recovery of the friction coefficient. Among these samples, N3 exhibits the best recovery (%µRT_afterHT = -8%), whereas the worst behavior is displayed by N2 (%µRT_afterHT = -19%).
The specific wear coefficient calculated for all the tested pins is shown in Fig. 7

ANALYSIS OF WORN SURFACES
In Fig. 8   The planar views of the worn pins tested at RT are shown in Fig. 9. The Ref. sample features a large area covered with wear debris, but just small portions of its surface (II), near the primary plateaus (I), are well compacted.
This observation is coherent with the cross-section images in Fig. 8. On the N1 sample surface, the secondary plateaus are definitely wider, but they are also characterised by several cracks. On the surface of N2 sample, the plateaus are smaller and more regularly distributed even if some cracks are present in this case too. The secondary plateaus on the N3 sample are even smaller but without fractures: this indicates good compaction of the constituents and better mechanical properties of the resulting plateaus (Fig. 9). In Table 6 a summary of the main features of the secondary plateaus are given, as obtained from the SEM observations of the cross-section and planar views of the samples tested at RT, i.e. average thickness, compaction degree, extension and presence of cracks.  Table 7 provides the compositional results of the EDXS analyses carried out on the secondary plateaus (regions II in Fig. 9). The majority element is iron, which comes both from the disc and pin.
This element is present in the friction layer mainly in the form of mixed oxides [25][26] [27]. The   (Fig. 9).  In Table 8 the EDXS analyses carried out on the secondary plateaus of the samples tested at RT_after HT are summarized. Also, in this test condition, the main elements detected for all the samples are Fe, O, C and Zn.  respectively) compared to the samples Ref. and N1 (6.6 wt% and 6.8 wt% respectively).

DISCUSSION
In the present work we have investigated the role of graphite as concerns the tribological properties of brake friction materials. Natural graphite reduces the friction coefficient (µ), and, in particular, the friction coefficient decreases with the graphite particle size. The same trend was observed as concerns the wear coefficient (K a ): the samples with a finer graphite (N2, N3) showed K a values lower than the reference material (Ref., no graphite) and also of the N1 sample, featuring larger, plate-like flakes of graphite. The relationship between friction coefficient and wear rate for the tested materials is illustrated in Fig. 13. A second order polynomial function is fitting the experimental data adequately.
The relevant fitting parameters are listed in Table 9.
This trend can be interpreted by considering that the tribological characteristics of the friction materials are influenced by the morphology and chemistry of the friction layer that builds up at the interface between the sliding materials [6] [28]. Friction is generated by the adhesive and abrasive interactions between the pin and the disc. The pin, as well as the disc, are possible sources of carbon. In the pin, many carbonaceous constituents are present, other than the deliberately added graphite, like: phenolic resin, coke, rubber and organic fibers. The presence of a relatively high concentration of graphite on the friction surface may contribute to the formation of a graphite-rich friction layer with dense and well compacted secondary plateaus, leading to a general improvement of the friction performances [29][30].
The graphite with smaller particle size enters the friction layer more easily and becomes more homogeneously distributed. This results in the reduction and stabilization of the friction coefficient. In agreement with this situation, the samples with a finer graphite, N2 and N3, displayed better compacted friction layers (Fig. 8), which, also for their graphite content, are responsible for the reduction in the adhesive interaction between the mating surfaces, resulting in a reduction in the wear rate K a .   In the tests conducted at RT, after approximately 2500 s run-in, the coefficient of friction reaches a steady state, determined by the formation of a friction layer, which stabilizes the contact between the pin and the disc.
In the tests at HT, the friction coefficient is characterised by a peculiar behaviour, observed in all tests and shown by the relevant (HT) curve in Fig. 14. Four main stages can be identified: (A) A friction layer, which formed during the RT test, is already present on the pin's surface at the beginning of the HT test. However, since during the heating up, the disc and the pin are not in contact, at the beginning of the test a sort of running-in takes place anyway. It is related to the re-establishment of the full contact between the two mating surfaces.
(B) The friction curve shows a drastic decrease in µ, due to the removal of the friction layer from the pin surface for the combined action of: adhesion forces and thermal softening of the friction material, that is not sustaining any longer the original surface layer. It has to be considered that during the test at 400°C, a thermal degradation of the phenolic resin occurs [31]. As clearly evidenced in a previous investigation [32], this reduces the capability of the underlaying friction material to sustain the friction layer adequately. Moreover, an abrasion contribution from the oxide forming on the disc surface is certainly contributing to removing the original friction layer from the pin surface.
(C) Disc oxidation and wearing out of the pin surface determine a new raise in the friction coefficient, accompanied by the formation of a new friction layer, made of the debris coming from both mechanisms. A similar behaviour was observed by Vergne et al. [33], studying the high temperature dry sliding of two steels. Specific PoD tests were interrupted after different time intervals in order to define the contribution from the oxides on the evolution of the coefficient of friction, very similar to that displayed in Fig. 14. (D) Once a maximum value of the friction coefficient is achieved, another decrease is observed.
The new reduction in the friction coefficient is due to the combined effect of: the friction material softening and its transfer onto the disc surface, as reported already in similar experiments [32]. This determines the progressive transition from a friction material-cast iron coupling towards a situation in which the friction layer on the pin surface is in sliding contact with the transfer layer on the disc. A steady condition is thus achieved, featuring the stabilization of µ at low values. As depicted by Fig. 10.
The influence of graphite particle size at high temperatures was evaluated with the Eq.2. The decrease in the friction coefficient observed after the HT tests (%µ HT ) showed no significant differences among all samples. It can be concluded that graphite does not significantly influence the behaviour at high temperature. Its role of solid lubricant and friction coefficient stabilizer is jeopardized at high temperature by oxidation phenomena, considering that flash temperatures reached at the contacting asperities of the pin-disc mating surfaces may very easily overcome 400°C [34]. Graphite and other constituent are quickly removed as worn debris [35], and indeed all the samples tested at HT show a drastic increase in the specific wear coefficient (K a ).
In the tests at RT after the high-temperature tests (Fig. 14, RT_after_HT) a friction layer, which formed during the HT tests, is already present on the pin surface. Since it is discontinuously distributed on the pin surface (Fig. 10), a run-in is necessary before reaching the steady state in the new test conditions. The Eq.3 gives an indication of the recovery attitude of the tribological properties by the different materials after the HT tests [36]. Lower values of the friction coefficient with respect to the initial tests at RT were detected in all the samples with graphite additions. The best behaviour in terms of friction coefficient recovery, is observed in the N3 sample, and this is due to the finer size of added graphite, which makes it easier for the particles to enter the friction layer. In this way, graphite together with other constituents of the friction layer, restore surface conditions similar to those attained at the end of the initial RT tests. In this recovery process, it should be considered that the organic components of the friction material (pin), more sensitive to temperature, degraded and were partially emitted from the tribological system as gaseous products. This reduced their contribution to the growth of the friction layer, and in particular, of the secondary plateaus during the RT_after HT tests. The reduction of carbon concentration is confirmed by the pin-pointed EDXS analyses, carried out on the secondary plateaus (Fig. 12).
The Ref. material displayed the higher specific wear coefficient (see Fig. 7) and, at the same time, the better friction coefficient recovery. The relevant scenario emerging from these data is that a rapid restoration of the pin's pristine surface is favoured by the faster wearing out of friction layer, initially present on the pin surface and produced by the HT tests.

5.CONCLUSIONS and FUTURE PERSPECTIVES
The effect of natural graphite with different particle size was investigated in a Cu-free friction material using a PoD test rig. The study regarded in the first place the tribological behavior and the formation of the friction layers during the tribological tests on these new materials against pearlitic cast iron. The main results can be summarized as follows: • The particle size of graphite impacts considerably on the friction, wear properties and friction layer formation. • In the RT test, the larger is the particle size, the larger is the friction coefficient and wear rate.
• The best behavior was observed in the N3 sample, with graphite particle size ˂ 40 µm; in this case the friction coefficient is comparable to the one of the N2 sample (graphite grain size ˂ 100 µm), but with an improvement in the wear performances.
• Graphite particle size ˂ 100 µm render the friction layer more compact and the secondary plateaus more stable, since richer in carbon. • All tested materials exhibit a reduced percentage of recovery of the friction coefficient after HT tests (400°C), which is apparently not affected by the additions of graphite.
• The material without graphite (Ref.) retains the better friction-recovery (%) behavior; this is ascribed to a faster wear rate, resulting in a fast restoration of the initial pin's surface conditions. Two further main investigations on natural graphite are planned to complete the picture.
The first will deal with the estimation of an optimal concentration of graphite. The second research theme concerns the way in which graphite (possibly in different concentrations) affects the particulate matter emissions from the friction materials. Nowadays, this is indeed a major issue in the development of novel friction materials.
Eventually, it would be extremely interesting to perform full dynamometer tests in order to understand the role of graphite particle size on NVH performance of the Cu-free formulations.