Tribology International 138 (2019) 99–110

Contents lists available at ScienceDirect

Tribology International
journal homepage: www.elsevier.com/locate/triboint

A study of brake contact pairs under different friction conditions with T

respect to characteristics of brake pad surfaces
L. Wei, Y.S. Choy∗, C.S. Cheung
Department of Mechanical Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong

ARTICLE INFO ABSTRACT

Keywords: In the present study, influences of friction conditions on friction and wear behaviors, as well as characteristics of
Disc brake brake pad surfaces were investigated. Low metallic (LM), semi metallic (SM) and non-asbestos organic (NAO)
Pin-on-disc brake pads sliding against an iron disc were tested using a pin-on-disc tribometer. Results show that the friction
Friction and wear coefficient and specific wear rate decrease with increasing contact pressure and sliding velocity. For the mor-
Fractal dimension
phology of brake pad surface, friction layers are large with few cracks on surfaces of SM and LM brake pads.
While for NAO brake pad, friction layers have a comparatively slender and irregular shape. From fractal analysis,
the fractal dimension of brake pad surface is in the range of 2.38–2.84 for all brake pads.

1. Introduction compacted secondary plateaus leading to a smooth and stable coeffi-

cient of friction (COF). In contrast, less compacted secondary plateaus
Disc brakes are used to slow down or stop vehicles and trains result in unstable evolution of COF [8]. Secondary plateaus could be
through pushing brake pads against a brake disc. Performances of disc classified into type I, with the support of primary plateaus, and type II,
brakes are determined by the friction and wear process between the without support of primary plateaus [10]. Österle and Urban [11]
brake pad and brake disc which are made of friction materials and grey found the thickness of friction layer was 100 nm and its surface was
cast iron respectively. Brake pads contain more than 30 ingredients covered by a thin film of metal sulphides and graphite which could help
which can be categorized as binder, reinforcements, fillers and fric- to fix wear particles. Alemani et al. [2] investigated characteristics of
tional additives [1,2]. Moreover, brake pads can be classified into three the friction layer of a LM brake pad under three contact loads and re-
types: low metallic (LM), semi metallic (SM) and non-asbestos organic ported iron and copper were major elements of secondary plateaus and
(NAO) brake pads according to different contents of ingredients. As iron oxides formed a very thin layer on the surface of secondary pla-
described in a previous study [3], LM brake pads have a high friction teaus. Federici et al. [6] found the primary plateaus of NAO brake pad
and good braking capacity at high temperature, but have a high wear consisted of large ceramic particles, such as ZrO2 and Mg–K-silicate
rate. SM brake pads have a low wear rate but cause high braking noise. particles. From the literature review, little attention is paid on in-
NAO brake pads have a low wear rate and brake noise, but they lose vestigating the change of friction layer of the brake pads when the
braking capacity at high temperature. contact load and sliding velocity are varied. Moreover, most studies
Researchers have proved that the wear process is characterized by related to the characteristics of friction layers were conducted using LM
the formation and disruption of friction layers formed on the contact and NAO brake pads, while for SM brake pads with 30–65% of metal
surfaces of pad and disc [4–6]. Friction layers consist of primary and content [12,13], properties of friction layers have not been studied in
secondary plateaus [7,8]. Primary plateaus are formed by hard com- depth.
ponents, including metal and ceramic fibers, which protrude from the Fractal geometry has been introduced to describe the morphology of
surface. These components had higher hardness than pearlitic grey cast contact surfaces in tribology because of the self-affinity (a small part
iron [9]. Secondary plateaus are made by compacted wear debris which can be considered as a reduced-scale image of the whole) of contact
can move in channels between pad and disc and stack up against the surfaces [14–16]. Moreover, fractal properties of contact surfaces can
primary plateaus. These wear debris consist of copper, graphite, iron be characterized by a non-integer dimension, called fractal dimension,
oxides and some other soft components [6,8]. Previous studies have which value is independent of the resolution of measuring instrument
found that small wear particles, such as copper, can form well [17]. Yan and Komvopoulos [18] investigated the normal contact of

∗
Corresponding author.
E-mail address: mmyschoy@polyu.edu.hk (Y.S. Choy).

https://doi.org/10.1016/j.triboint.2019.05.016
Received 14 February 2019; Received in revised form 17 April 2019; Accepted 12 May 2019
Available online 17 May 2019
0301-679X/ © 2019 Published by Elsevier Ltd.
L. Wei, et al. Tribology International 138 (2019) 99–110

Nomenclature NAO non-asbestos organic

SEM scanning electron microscopy
ANOVA Analysis of variance SI supplementary information
BSE Backscatter electron SM semi metallic
COF coefficient of friction XRD X-ray diffraction
EDXS energy-dispersive X-ray spectroscopy XRF X-ray fluorescence spectroscopy
LM low metallic

elastic-plastic rough surfaces using three-dimensional fractal geometry pin-on-disc tribometer was built up based on the design of Olofsson
and obtained the relationships between total contact force and real et al. [21]. It is considered as a reliable equipment for measuring the
contact area in terms of the mean surface separation distance, fractal wear and friction behaviors of brake pad materials sliding against iron
diameters and material properties. Zhu et al. [19] investigated contacts disc [1,2,8], although it cannot reflect the actual brake condition. Fig. 1
between two carbon steels with hardness of HRC 30 and HRC 24 and shows the schematic diagram of the tribometer that includes a hor-
between a carbon steel with hardness of HRC 30 and Cu–Zn alloy under izontal rotating disc and a deadweight loaded brake pad sample (pin).
the contact pressure of 0.56 MPa and sliding velocity of 0.55 m/s during Braking tests were conducted at room temperature and humidity. The
running-in stage. They reported fractal dimensions of surface profiles tribometer runs under steady conditions with constant normal force and
ranging from 1.11 to 1.20 for the worn surface of steel in steel-steel constant rotating speed. The friction coefficient was calculated as the
contact and 1.10 to 1.29 for the worn surface of Cu–Zn alloy in steel-Cu- measured tangential force divided by the applied normal force. The
Zn alloy contact at different friction times. Wu et al. [20] investigated tangential force was measured by an HBM Z6FC3/20 kg load cell with
the contact between GCr15 steel and C45 steel under different contact 1 Hz data sampling rate. A K-type thermocouple was inserted in a hole
pressures and sliding velocities and reported that the fractal dimensions drilled inside the pin body at a distance of 2 mm from the contact
of profiles for worn surfaces of GCr15 steel were in the range of surface to estimate the brake pad temperature. The brake pad tem-
1.83–1.90. Contact surfaces on brake pads should also be consider as a perature was recorded every 10 s. Specific wear rate (ks) was used to
self-affine fractal, while their fractal dimensions are lack of study. evaluate the wear behavior of brake pad sample and it was calculated
In the present study, characteristics of brake pad surface, including based on the equation in Ref. [1]:
morphology, fractal dimension, elemental composition and phase
composition were investigated using different brake pads sliding m
ks =
against a pearlitic cast iron disc under different friction conditions. The s FN (1)
objectives of the present study are (1) to study the influence of contact
Where ks is the specific wear rate in m2 / N ; m is the mass loss of sample
pressure and sliding velocity on the change of properties of the brake
during the test in g as determined by weighing the sample before and
pad surface; (2) to investigate the characteristics of surface of brake pad
after the test; is the density of sample in g/m3; s is the sliding dis-
with high metal content (larger than 30 wt%). To conduct a broader
tance in m; FN is the applied normal force on the sample in N. The
analysis, SM, LM and NAO brake pads were used in present study.
sample's mass is measured through a VIBRA analytical balance with an
accuracy of 0.01 mg.
2. Experimental set-up Morphologies and elemental compositions of friction layers on
brake pad surface were investigated using a scanning electron micro-
2.1. Test set-up and characterization techniques scopy (SEM) equipped with an energy-dispersive X-ray spectroscopy
(EDXS) system and a backscatter electron (BSE) detector. Before con-
In order to study the friction and wear behaviors of disc brakes, a ducting a SEM measurement, the brake pad sample was coated with a

Fig. 1. Schematic of tribometer.

100
L. Wei, et al. Tribology International 138 (2019) 99–110

thin layer of gold. About 10 measurements were conducted for each velocity and contact pressure on experimental results based on the
brake pad sample to ensure that the results are reliable and repeatable. method in Montgomery [28]. Standard errors, as shown in relevant
To further identify the crystalline phases of friction layers, X-ray dif- tables and figures, were calculated following the method in Moffat et al.
fraction (XRD) measurements were conducted with an IPD 3000 dif- [29].
fractometer using Cu-Kα radiation with Ni-filter. Phase composition
was determined through a full pattern fitting procedure based on
Rietveld's method [22–24]. SEM pictures with low resolution (50 × ) 3. Results
were used to calculate the fractal dimension of brake pad surface. Areas
of contact plateaus were measured and counted by using a commercial 3.1. Friction and wear behavior
picture processing software Image-Pro Plus 6.0. The corresponding SEM
pictures were transformed to white-and-black as shown in Fig. S1 of Fig. 3 shows the average friction coefficients with standard errors at
Supplementary Information (SI). Contact plateaus became white, while steady-state stage for all brake pads and Fig. S3 of SI shows the typical
the surroundings became black after the segmentation process. The variation of friction coefficient with sliding distance at mode 2 test. The
smallest area of contact plateau was about 500 μm2. White points or 32 ANOVA results of friction coefficient are shown in Table S1 of SI. The
areas smaller than 500 μm2 were considered as noise and neglected in friction coefficients are in the ranges of 0.38–0.46, 0.37–0.49 and
counting. 0.4–0.48 respectively for SM, LM and NAO brake pads. The friction
coefficient increased continually from an initial value to a steady-state
2.2. Friction materials value after a running-in stage for all the brake pad samples. Moreover,
the running-in stages of SM and LM brake pads are shorter than that of
The Pin samples were milled from commercial brake pads and NAO brake pad. Similar phenomena were also observed in other test
machined to a cubic shape with 10 ± 0.2 mm in length and 7–8 mm in modes.
thickness. Three types of brake pad including SM, LM and NAO were The friction coefficient decreases with increasing contact pressure
used in the present study which were bought from a major manu- for all the brake pads. With increasing sliding velocity, the friction
facturer of brake pads in China. The exact ingredients in a brake pad coefficient also decreases in most cases. Both contact pressure and
were not known because they are commercial secret. Therefore, an X- sliding velocity have a significant effect on the friction coefficient based
ray fluorescence spectroscopy (XRF) was used to measure the elemental on results of ANOVA analysis. In addition, contact pressure has a higher
composition of each brake pad as shown in Table 1. Since some light effect on friction coefficient than sliding velocity for LM and NAO brake
elements such as carbon and oxygen cannot be detected by XRF, the pads. But for SM brake pad, they have similar effects on friction coef-
sum of element concentrations is not equal to one hundred. BSE-SEM ficient. The very low P-values of interaction for all the brake pads in-
pictures of brake pad samples are shown in Fig. 2. Some components on dicates that the variation of friction coefficient with sliding velocity and
brake pad surfaces were identified using EDXS point spectra as shown contact pressure is independent of each other.
in Fig. S2 of SI. A commercial brake disc with a diameter of 239 mm and Fig. 4 shows the average brake pad temperatures with standard
a thickness of 9 mm, bought from BOSH and made of pearlitic grey cast errors at steady-state stage and Fig. S4 of SI shows the variation of
iron, was used in this study. Before conducting the experiments, pin brake pad temperature with sliding distance at mode 6 test. The 32
samples were cleaned with air and disc was cleaned with ethanol. ANOVA results for brake pad temperature are shown in Table S2 of SI.
The average brake pad temperatures increase with increasing contact
pressure and sliding velocity for all the brake pads. From ANOVA re-
2.3. Design of experiments
sults, both sliding velocity and contact pressure can significantly affect
the brake pad temperature. Moreover, there is a significant interaction
Braking tests were conducted at three sliding velocities (v) of 1.6
on the brake pad temperature between the contact pressure and sliding
(low speed), 2.8 (medium speed) and 4.9 m/s (high speed) and at three
applied loads of 52 (low load), 81 (medium load) and 122 N (high load)
Table 1
which correspond to nominal contact pressures (p) of 0.52, 0.81 and
Element compositions and densities of brake pads.
1.22 MPa respectively. Friction power (P) is used to analyze the re-
lationship between friction/wear results and friction conditions as Element SM brake pad (wt LM brake pad (wt NAO brake pad (wt
shown in the equation below [25,26]. %) %) %)

P = µFN v (2) Mg 0.68 0.80 0.77

Al 1.10 1.38 4.16
Where P is the frictional power in W; µ is the friction coefficient; FN is Si 2.89 3.90 6.28
the applied normal load in N; v is the sliding velocity in m/s. Frictional P 0.05 0.05 0.08
S 4.62 5.02 1.28
power combines the friction force with sliding velocity which re-
Cl 0.11 0.13 0.02
presents the work applied on contact surfaces by the friction force. The K 0.27 0.21 3.14
wear track on the disc had a mean diameter of 220 mm. Detailed test Ca 10.08 9.34 25.57
conditions and frictional powers are listed in Table 2. Ti 0.21 0.10 3.69
The p∙v values range from 0.83 to 5.98 MPa m/s which are adequate Cr 0.05 0.03 –
Mn 0.15 0.20 0.08
to investigate these materials because in real brake systems p∙v values
Fe 34.38 25.94 13.42
are in the range of 0.3–20 MPa m/s [27]. If scaled up to the reference Ni 0.04 0.06 –
vehicle, the sliding velocities correspond to vehicle speeds of Cu 4.09 5.67 1.55
15.2–46.6 km/h. The selected test conditions are appropriate for si- Zn 0.75 1.37 1.06
As – – 0.44
mulating the light braking process of a vehicle under urban driving
Se 0.02 – –
conditions which reduces the vehicle speed but do not stop it. The Sr 0.02 – 0.03
sliding distance was 22050 m in the middle of the wear track for all Zr 0.03 – –
tests. The duration of each test varied from 75 to 230 min. Friction force Mo 0.01 – –
and brake pad temperature were recorded during each test. Experi- Sb 7.41 6.94 4.01
Pb 0.31 0.22 –
ments were repeated three times and average results are shown in the
Densities 2.76 ± 0.04 2.75 ± 0.12 2.12 ± 0.07
present study. Analysis of variance (ANOVA) for 32 factor design using (g/cm3)
three repetitions was also conduced to compare influences of sliding

101
L. Wei, et al. Tribology International 138 (2019) 99–110

Fig. 2. BSE-SEM pictures of brake pad samples. (a: SM brake pad, b: LM brake pad, c: NAO brake pad).

velocity. Sliding velocity has a stronger effect on brake pad temperature 3.2. Characteristics of brake pad surface
than contact pressure. Due to the difference of thermal conductivity
among brake pad samples, the brake pad temperature increases in the 3.2.1. Morphology of friction layers
order of NAO, LM and SM brake pads which is also in line with the The friction layers on the contact surfaces of brake pads after mode
metal contents in these brake pads. The brake pad temperature ranges 4 test (low load test) and mode 6 test (high load test) are shown in
from 45 to 130 °C indicating that the degradation of resin did not occur. Figs. 6 and 7 respectively. It can be observed that most of the pin
Thus, the influence of resin degradation on friction and wear behaviors surface is covered by widely distributed friction layers with irregular
can be neglected. shape. Comparing the friction layers of these two tests, the following
The specific wear rates (ks) with standard errors are shown in Fig. 5. two differences are observed. Firstly, high contact pressure results in
The 32 ANOVA results for specific wear rate are shown in Table S3 of SI. more compacted secondary plateaus. Many coarse particles can be
The specific wear rates are in the range of 10−15 to 10−14 m2/N which observed at the edge of the friction layers after mode 4 test as shown in
are in line with the previous studies [8,30,31]. The specific wear rates squares with dashed line in Fig. 6. Secondly, some parts of the friction
decrease with increasing the contact pressure for all the brake pads and layers were blown off from the bulk material with some cracks on them
the maximum reduction can reach 36%, 75% and 56% for SM, LM and as shown in circles with dashed line in Fig. 6. In addition, abrasive
NAO brake pads respectively when comparing results at the sliding grooves can be identified on the surfaces of all brake pads as shown in
velocity of 4.9 m/s. The specific wear rates also decrease with in- Fig. 7. This phenomenon indicates the effect of third-body abrasion on
creasing sliding velocity and the maximum reductions are 69%, 57% the wear behavior of brake pads sliding against iron disc. Generally, the
and 57% for SM, LM and NAO brake pads respectively when comparing SM and LM brake pads have similar morphology of friction layers with
results at the contact pressure of 1.22 MPa. Both sliding velocity and properties of large and well compacted. In contrast, friction layers of
contact pressure have a significant effect on specific wear rate based on the NAO brake pads are slender and more irregular in shape.
the ANOVA analysis. Moreover, there is a significant interaction on the Typical friction layers on the surfaces of brake pads after mode 7
specific wear rate between sliding velocity and contact pressure. test (high speed test) are shown in Fig. S5 of SI. Compared with the

Table 2
Test conditions and fractal dimensions.
Test mode Sliding velocity (m/s) Nominal contact pressure (MPa) Frictional power (W) Fractal dimension (Ds)

SM LM NAO SM LM NAO

1 1.6 0.52 38.50 40.42 40.10 2.56 2.62 2.53

2 1.6 0.81 56.30 55.72 55.28 2.67 2.61 2.46
3 1.6 1.22 83.41 80.84 82.83 2.59 2.49 2.53
4 2.8 0.52 64.87 68.19 69.43 2.64 2.53 2.46
5 2.8 0.81 93.01 94.08 96.07 2.51 2.52 2.46
6 2.8 1.22 133.61 135.79 138.07 2.42 2.38 2.48
7 4.9 0.52 107.77 115.51 113.92 2.47 2.71 2.59
8 4.9 0.81 152.60 157.98 161.30 2.84 2.71 2.45
9 4.9 1.22 229.11 223.56 242.21 2.76 2.60 2.59

102
L. Wei, et al. Tribology International 138 (2019) 99–110

Fig. 3. Average friction coefficients of SM, LM and NAO brake pads at steady-state stage.

friction layers on brake pad after low speed test, there are more and velocity. In addition, the widths of friction layers of NAO brake pad are
deeper abrasive grooves on the surface of brake pad. With increasing smaller than those of SM and LM brake pads. From ANOVA analysis,
sliding velocity, kinetic energy of wear debris is increased, which en- sliding velocity has significant effect on the width of friction layer,
hances the third-body abrasion on the brake pad surface, leading to while contact pressure cannot significantly affect it. There is no sig-
deeper abrasive grooves. nificant interaction on the width of friction layer between sliding ve-
Widths of friction layers (D), as marked in Fig. 7, are used to de- locity and contact pressure.
scribe the size of friction layer. Four different SEM fields of view were High resolution SEM pictures of friction layers for the LM brake pad
randomly selected and photographed to determine the widths of fric- after mode 6 test (high load test) are shown in Fig. 9. High resolution
tion layers. About 20 friction layers were measured in each SEM pic- SEM pictures for the SM and NAO brake pads are shown in Fig. S6 of SI.
ture. Fig. 8 shows the average widths of friction layers with standard The primary plateaus circled by white dash lines are formed by Fe fi-
errors. The 32 ANOVA results for the width of friction layer are shown bers. The secondary plateaus are made of compacted wear debris which
in Table S4 of SI. The mean widths of friction layers increase with in- commonly have a larger size than the primary plateaus.
creasing contact pressure, while decrease with increasing sliding

Fig. 4. Brake pad temperatures of SM, LM and NAO brake pads sliding against iron disc.

103
L. Wei, et al. Tribology International 138 (2019) 99–110

Fig. 5. Specific wear rates of SM, LM and NAO brake pads sliding against iron disc.

3.2.2. Fractal dimension measurement results of fractal dimensions are shown in Table 2. The average fractal
The distribution of contact plateaus on the contact surface of a brake dimensions with standard deviations for SM, LM and NAO brake pads
pad follows the power-law relationship as shown below [16]. are 2.61 ± 0.13, 2.58 ± 0.11 and 2.51 ± 0.06 respectively. The SM
and LM brake pads have similar values of fractal dimension and stan-
N( A> a) a (Ds 1)/2
(3) dard deviation which are larger than that of the NAO brake pad. But the
difference in fractal dimension among SM, LM and NAO brake pads are
Where N (A > a) is the total number of contact plateaus with areas
not significant at 95% confidence level.
larger than a particular area, a; Ds is the fractal dimension of a surface.
The lowest value of a is 500 μm2, corresponding to the smallest contact
plateau that can be identified. Typical number-area distributions of 3.2.3. Elemental and phase compositions
contact plateaus after mode 4 test are shown in Fig. 10. It can be ob- Fig. 9 shows the EDXS point spectra acquired from the friction
served that the relationship between N (A > a) and a follows the power- layers of LM brake pad surface after mode 6 test. EDXS point spectra of
law very well with R2 larger than 0.99. Then, the fractal dimensions can the SM and NAO brake pad surfaces after mode 6 test are shown in Fig.
be evaluated through curve fitting for N (A > a) versus a. Detailed S6 of SI. Some major elements in the friction material are identified.

Fig. 6. SEM pictures of brake pad surfaces after mode 4 test (a: SM brake pad, b: LM brake pad, c: NAO brake pad; white arrow indicating the sliding direction).

104
L. Wei, et al. Tribology International 138 (2019) 99–110

Fig. 7. SEM pictures of brake pad surfaces after mode 6 test (a: SM brake pad, b: LM brake pad, c: NAO brake pad; white arrow indicating the sliding direction).

The friction layers are found to be rich in Fe for all the tested brake brake pad samples. Mean element concentrations with standard errors
pads. The LM and SM brake pads have similar elemental compositions are shown in Fig. 11 and the relevant 32 ANOVA results are shown in
for the friction layers. Tables S6 and S7 of SI.
The metallic elements, including Fe and Cu, were selected to ana- As shown in Fig. 11a, the mean values of iron concentrations vary
lyze the variation of elemental composition of friction layers with from 54 to 72%, 42–66% and 17–58% for the SM, LM and NAO brake
friction conditions. Ten EDXS measurements were conducted for each pads respectively. Iron contents in the friction layers of SM and LM

Fig. 8. Widths of friction layers on brake pad surfaces.

105
L. Wei, et al. Tribology International 138 (2019) 99–110

Fig. 9. SEM pictures and EDXS point spectra of the LM brake pad surface (Test mode 6).

brake pads are similar and they are higher than that in the friction layer quantified because their concentrations are less than 1% and cannot be
of NAO brake pad. In addition, iron concentrations in the friction layers detected in each brake pad sample. There is no special trend for the
are also much higher than those in the bulk material of brake pads, as phase composition with either the sliding velocity or the contact pres-
shown in Table 1, in most cases. This phenomenon indicates that most sure for each brake pad. As shown in Table 3, mass fractions of graphite
of the iron element is produced from the tribo-oxidation and abrasion of are in the ranges of 37.0–60.5%, 43.9–62.1% and 34.1–53.4% respec-
iron disc, which is in line with the previous studies [2,32]. The influ- tively for SM, LM and NAO brake pads, which are much higher than
ences of contact pressure and sliding velocity are complicated. There that of other elements. It can be observed that the concentration of iron
are no particular trends for the variation of iron concentration with oxides measured by the XRD system is much lower than the con-
either the contact pressure or the sliding velocity. The average con- centration measured by the EDXS in most cases. Another finding is that
centrations of copper with standard errors are shown in Fig. 11b. The the pure copper concentrations in the friction layers are higher than
copper contents range from 4.8 to 10.3%, 2.8–8.0% and 3.1–13.5% those in the friction materials in most cases.
respectively for the SM, LM and NAO brake pads. There are no special
trends for the variation of copper concentration with either the sliding
velocity or the contact pressure. 4. Discussion
Main phase compositions of the friction layers on brake pad surfaces
were measured through XRD analysis. Concentrations of the major The decrease of friction coefficient with increasing contact pressure
phases are quantified following the method in Ref. [22] and shown in can be attributed to the un-proportional increase of contact area on
Table 3. Some minor phases, such as tenorite and zinc, are not asperities which results in a reduction of local pressure on the asperities
and subsequently a lower friction coefficient [33]. The un-proportional
increase of contact area on asperities is mainly due to the viscoelastic
properties of friction materials [34]. The decrease of friction coefficient
with increasing sliding velocity is due to the increment of sliding ve-
locity which reduces the contact time between asperities of brake pad
and disc, then reducing the possibility of forming strong junctions by
adhesion [35].
Shorter running-in stages of the LM and SM brake pads compared
with that of the NAO brake pad can be attributed to higher metal
contents in LM and SM brake pads. As shown in Table 1, metal contents
in the NAO brake pad (Fe + Cu: 14.97%) are much lower than those in
LM (Fe + Cu: 31.61%) and SM brake pads (Fe + Cu: 38.47%). Higher
metal content can generate more primary plateaus which provide the
majority of friction force and more type I secondary plateaus with good
compactness and adhesion strength. Therefore, friction force increases
and reaches a steady-state value quickly.
The reduction of specific wear rate with increasing contact pressure
is mainly due to the higher contact pressure which improves the com-
pactness of secondary plateaus and subsequently reduces their possi-
bility to be detached. The reduction of specific wear rate with in-
Fig. 10. Number-area distributions of contact plateaus on the brake pad surface creasing sliding velocity is due to the lower friction coefficient results in
(Test mode 4). the lower shear force on contact plateaus leading to a lower specific

106
L. Wei, et al. Tribology International 138 (2019) 99–110

Table 3
Main phase compositions (wt%) of the friction layers on brake pad surface.
1.6 m/s 2.8 m/s 4.9 m/s

0.52 MPa 0.81 MPa 1.22 MPa 0.52 MPa 0.81 MPa 1.22 MPa 0.52 MPa 0.81 MPa 1.22 MPa

(a) SM brake pad

Graphite 55.6 43.4 39.5 50.7 41.9 38.2 60.5 37.0 48.6
Copper 12.3 11.2 11.5 13.7 11.3 12.8 10.8 11.6 9.6
Hematite (Fe2O3) 11.2 18.9 17.6 10.1 22.6 23.4 12.4 12.6 14.4
Calcite (CaCO3) 15.9 18.5 25.3 21.1 19.6 20.6 10.7 30.9 16.3
Stibnite (Sb2S3) 4.8 8.0 6.2 4.3 4.5 5.0 5.3 7.9 11.0
(b) LM brake pad
Graphite 62.1 47.4 46.8 47.9 64.1 49.7 46.9 60.0 43.9
Copper 4.6 12.9 16.9 15.3 1.5 24.7 10.5 7.6 9.1
Hematite (Fe2O3) 12.1 13.5 12.3 16.8 10.5 10.5 10.2 13.6 14.0
Calcite (CaCO3) 15.7 22.0 17.5 19.6 19.2 12.5 24.3 12.1 22.4
Stibnite (Sb2S3) 4.9 4.3 6.8 0.3 3.0 2.4 8.2 6.7 11.2
(c) NAO brake pad
Graphite 38.4 45.3 34.3 41.4 34.1 53.4 35.4 34.4 37.2
Copper 10.1 5.6 0.8 9.3 1.1 10.3 10.9 0.5 0.8
Hematite (Fe2O3) 9.3 6.8 10.9 9.6 9.7 3.8 9.0 12.0 9.8
Calcite (CaCO3) 41.5 42.2 52.4 39.7 52.8 32.6 43.2 51.9 51.7
Stibnite (Sb2S3) 0.3 0.8 1.9 0.3 2.0 0.4 0.2 1.7 1.1

wear rate. The relationship between specific wear rate and frictional with either contact pressure or sliding velocity.
power is shown in Fig. 12. A good linear relationship is observed for the Standard deviation of the fractal dimension for the NAO brake pad
NAO brake pad and close to linear relationship is observed for the SM surface is smaller than those for the SM and LM brake pad surfaces. This
brake pad. The fitting parameters are shown in Table S5 of SI. The smaller standard deviation indicates that the variation of rms slope for
friction and wear process are determined by adhesion and abrasion the surface of the NAO brake pad is lower than those for the surfaces of
between contact surfaces. Adhesion varies with sliding velocity and SM and LM brake pads. This phenomenon can be attributed to the
contact pressure [35] while abrasion is almost independent from con- different wear resistant of contact plateaus on the contact surfaces be-
tact pressure and sliding velocity [36]. The high values of adjusted R tween SM/LM brake pads and NAO brake pad. High metal contents in
square, as shown in Table S5 of SI, indicate that the wear mechanism SM and LM brake pads can provide many primary plateaus and type I
for brake pad is mainly determined by the adhesive wear mechanism. secondary plateaus with high wear resistant on contact surfaces. For SM
Different morphologies between SM/LM brake pads and NAO brake and LM brake pads, the high wear resistant of contact plateaus reduces
pads are due to the lower iron content in the NAO brake pad than that their possibility to be cracked, hence contributing to a high fractal di-
in SM and LM brake pads, as shown in Table 1. Effects of iron on friction mension [38]. This phenomenon becomes more pronounced in friction
layers have been introduced in Ref. [37]. With the lack of iron, other conditions with high frictional power, as shown in Fig. S7 of SI, leading
hard elements, such as ceramic particles, quartz and Aluminum oxides, to high standard deviations of fractal dimensions for the SM and LM
can work as primary plateaus. Thus, the different characteristics be- brake pads. For the NAO brake pad, the relatively low wear resistant of
tween steel fibers and these hard particles, including size and shape, contact plateaus on its surface reduces the fractal dimension leading to
result in different shapes of primary and secondary plateaus. Moreover, a lower standard deviation.
comparing NAO brake pad with SM and LM brake pads, more type II Person [38] reported most natural and man-made rough surfaces
secondary plateaus with an irregular shape are formed on the brake pad are three-dimensional fractal surface with a typical dimension ranging
surface because type II secondary plateaus do not have the support of from 2 to 2.3 because when the fractal dimension is larger than 2.3, the
primary plateaus. surface will be very fragile and damaged easily due to the high average
Sliding velocity can significantly affect the width of a friction layer. surface slope. But in the present study, all the fractal dimensions are
It is mainly because the higher sliding velocity leads to higher kinetic higher than 2.3. This phenomenon may be due to the large difference in
energy of the wear debris which reduces the probability of wear debris hardness between the contact plateaus and surrounding matrix of the
being blocked by the primary plateaus and hence reducing the width of brake pad surface. Eriksson and Jacobson [5] reported the hardness of
friction layer. With increasing contact pressure, the space between the secondary plateaus of NAO brake pad ranges from 0.2 to 4 GPa through
brake pad and disc is reduced. Therefore, more wear debris is blocked nanoindentation test, while the hardness of surrounding matrix is about
by the primary plateaus and larger secondary plateaus are formed. In 200 MPa through regular microhardness indentation test. The large
this mechanism, extension of secondary plateaus is mainly along the difference in hardness between the contact plateaus and matrix in-
sliding direction. For the width of a friction layer, which is orthogonal creases the height variation of surface profile, leading to a high fractal
to the sliding direction, contact pressure cannot significantly affect it. dimension.
Therefore, contact pressure cannot significantly affect the width of a Compared with the iron contents measured with EDXS, the lower
friction layer. iron contents measured with XRD are mainly due to the much smaller
There is no special trend for the variation of fractal dimension with penetration depth for X-rays of EDXS system than that for X-rays of XRD
either the contact pressure or sliding velocity for all the brake pads. On system. Iron oxides generated from tribo-oxidation are widely dis-
the one hand, increments of sliding velocity and contact pressure lead tributed in the friction layers on the brake pad surface. But they form a
to a more complicated friction and wear process and subsequently in- thin layer and contributes a limited fraction to the total diffraction
crease the complexity of surface profile which contributes to a higher intensity. Therefore, the mass fraction of iron oxides measured by XRD
fractal dimension [20]. On the other hand, contact plateaus with a high is much smaller than that measured by EDXS.
slope, corresponding to a high fractal dimension, would be very fragile The higher copper contents in friction layers compared with that in
and easily damaged by interactions with other objects [38]. Therefore, bulk materials is mainly due to the copper fibers milled into copper
no special trends are observed for the variation of fractal dimension particles which are widely dispersed in the friction layer. These copper

107
L. Wei, et al. Tribology International 138 (2019) 99–110

Fig. 11. Average mass fractions of Fe and Cu measured on friction layers.

particles can improve the compactness of the friction layer and reduce pad after brake tests with high sliding velocity or high contact pressure
its tendency to be broke down into wear debris [39]. Therefore, higher become very high. For SM and LM brake pads, similar phenomenon
copper content is measured in the friction layers. This phenomenon cannot be observed due to their high iron contents. Most secondary
becomes more pronounced for the NAO brake pad at the sliding velo- plateaus of SM and LM brake pads are supported by primary plateaus
city of 4.9 m/s or at the contact pressure of 1.22 MPa. As shown in which reduces their tendency to be broken down.
Fig. 11b, the copper contents in the friction layers of NAO brake pad at
these heavy friction conditions are significantly higher than those
measured at other friction conditions. For NAO brake pad, there are 5. Conclusions
many type II secondary plateaus on the contact surface where wear
resistance relies more on copper compared with type I secondary pla- Three types of brake pad were tested by sliding against an iron disc
teaus. Therefore, copper contents in the friction layers of the NAO brake to investigate the influences of sliding velocity and contact pressure on
friction and wear behaviors, as well as characteristics of the brake pad

108
L. Wei, et al. Tribology International 138 (2019) 99–110

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://

doi.org/10.1016/j.triboint.2019.05.016.

References

[1] Wahlstrom J, Lyu YZ, Matjeka V, Soderberg A. A pin-on-disc tribometer study of

disc brake contact pairs with respect to wear and airborne particle emissions. Wear
2017;384:124–30.
[2] Alemani M, Gialanella S, Straffelini G, Ciudin R, Olofsson U, Perricone G, et al. Dry
sliding of a low steel friction material against cast iron at different loads: char-
acterization of the friction layer and wear debris. Wear 2017;376–377:1450–9.
[3] Sanders PG, Xu N, Dalka TM, Maricq MM. Airborne brake wear Debris: size dis-
tributions, composition, and a comparison of dynamometer and vehicle tests.
Environ Sci Technol 2003;37:4060–9.
[4] Cho MH, Cho KH, Kim SJ, Kim DH, Jang H. The role of transfer layers on friction
characteristics in the sliding interface between friction materials against gray iron
brake disks. Tribol Lett 2005;20:101–8.
[5] Eriksson M, Jacobson S. Tribological surfaces of organic brake pads. Tribol Int
Fig. 12. Relationship between specific wear rate and frictional power. 2000;33:817–27.
[6] Federici M, Gialanella S, Leonardi M, Perricone G, Straffelini G. A preliminary in-
vestigation on the use of the pin-on-disc test to simulate off-brake friction and wear
surfaces. The following conclusions can be drawn from the present characteristics of friction materials. Wear 2018;410–411:202–9.
study. [7] Österle W, Griepentrog M, Gross T, Urban I. Chemical and microstructural changes
induced by friction and wear of brakes. Wear 2001;251:1469–76.
[8] Chandra Verma P, Menapace L, Bonfanti A, Ciudin R, Gialanella S, Straffelini G.
1. Friction coefficient and specific wear rate decrease with increasing Braking pad-disc system: wear mechanisms and formation of wear fragments. Wear
the contact pressure and sliding velocity. From ANOVA analysis, 2015;322–323:251–8.
[9] Österle W, Prietzel C, Kloß H, Dmitriev AI. On the role of copper in brake friction
both contact pressure and sliding velocity have significant effect on materials. Tribol Int 2010;43:2317–26.
these two parameters. [10] Neis PD, Ferreira NF, Fekete G, Matozo LT, Masotti D. Towards a better under-
1. Friction layers on surfaces of SM and LM brake pads are large with standing of the structures existing on the surface of brake pads. Tribol Int
2017;105:135–47.
few cracks, while the friction layer on surface of NAO brake pad has
[11] Österle W, Urban I. Friction layers and friction films on PMC brake pads. Wear
comparatively slender and irregulated shape. 2004;257:215–26.
2. The number-area distribution of contact plateaus follows the power- [12] Chan D, Stachowiak GW. Review of automotive brake friction materials. Proc Inst
Mech Eng - Part D J Automob Eng 2004;218:953–66.
law very well and fractal dimension of brake pad surface ranges
[13] El-Sonbaty IA, Khashaba UA, Selmy AI, Ali AI. Prediction of surface roughness
from 2.38 to 2.84. With different friction conditions, the variation of profiles for milled surfaces using an artificial neural network and fractal geometry
the fractal dimension for NAO brake pad is smaller than that for SM approach. J Mater Process Technol 2008;200:271–8.
and LM brake pads. [14] Komvopoulos K, Ye N. Three-dimensional contact analysis of elastic-plastic layered
media with fractal surface topographies. J Tribol 2001;123:632–40.
3. There are no particular trends for iron and copper contents and main [15] Thomas TR, Rosén BG, Amini N. Fractal characterisation of the anisotropy of rough
phase compositions with either the sliding velocity or the contact surfaces. Wear 1999;232:41–50.
pressure. Iron oxides are widely dispersed in the friction layer with a [16] Majumdar A, Bhushan B. Role of fractal geometry in roughness characterization and
contact mechanics of surfaces. J Tribol 1990;112:205–16.
shallow depth leading to the different iron concentrations between [17] Mandelbrot B. How long is the coast of britain? Statistical self-similarity and frac-
XRD and EDXS results. tional dimension. Science 1967;156:636–8.
[18] Yan W, Komvopoulos K. Contact analysis of elastic-plastic fractal surfaces. J Appl
Phys 1998;84:3617–24.
Results present a deeper understanding of the role of friction con- [19] Zhu H, Ge S, Huang X, Zhang D, Liu J. Experimental study on the characterization
ditions on the morphologies and elemental compositions of the friction of worn surface topography with characteristic roughness parameter. Wear
layers on brake pad surfaces which helps to understand the variation of 2003;255:309–14.
[20] Wu Z-h, Zhu H, Li G. Research on the relation of fractal dimension of friction
composition of airborne wear particles emitted from disc brakes under
coefficient to load and velocity. TRIBOLOGY-BEIJING- 2007;27:165.
different friction conditions. [21] Olofsson U, Olander L, Jansson A. A study of airborne wear particles generated from
Experiments were conducted under constant sliding velocity and a sliding contact. J Tribol Trans ASME 2009;131:044503–4.
[22] Verma PC, Alemani M, Gialanella S, Lutterotti L, Olofsson U, Straffelini G. Wear
contact pressure. But in real friction conditions, some brakes occur in a
debris from brake system materials: a multi-analytical characterization approach.
short time only including the running-in stage. Further study should be Tribol Int 2016;94:249–59.
conducted in the running-in stage to validate if results obtained under [23] Lutterotti L, Matthies S, Wenk HR, Schultz AS, Richardson JW. Combined texture
steady-state stage are still correct under running-in stage. In the present and structure analysis of deformed limestone from time-of-flight neutron diffraction
spectra. J Appl Phys 1997;81:594–600.
study, the brake pad temperature is lower than the critical temperature [24] Lutterotti L. Total pattern fitting for the combined size–strain–stress–texture de-
of resin degradation (230–475 °C). Thus, resin degradation is not con- termination in thin film diffraction. Nucl Instrum Methods Phys Res, Sect B
sidered. Future studies should investigate the change of properties of 2010;268:334–40.
[25] Alemani M, Wahlström J, Olofsson U. On the influence of car brake system para-
brake pad surface with different friction conditions under the critical meters on particulate matter emissions. Wear 2018;396–397:67–74.
temperature that resin degradation occurs. [26] Bettge D, Starcevic J. Topographic properties of the contact zones of wear surfaces
in disc brakes. Wear 2003;254:195–202.
[27] Straffelini G, Maines L. The relationship between wear of semimetallic friction
materials and pearlitic cast iron in dry sliding. Wear 2013;307:75–80.
Acknowledgement [28] Montgomery DC. Design and analysis of experiments. eighth ed. John wiley & sons;
2013.
[29] Moffat RJ. Describing the uncertainties in experimental results. Exp Therm Fluid Sci
Authors would like to thank Prof. Poon, Chi Sun at the department
1988;1:3–17.
of civil and environment engineering for the XRF test and also thank [30] Wahlström J, Matějka V, Lyu Y, Söderberg A. Contact pressure and sliding velocity
The Hong Kong Polytechnic University for the financial support maps of the friction, wear and emission from a low-metallic/cast-iron disc brake
contact pair. Tribology in Industry 2017;39.
(RUTT).
[31] Hong US, Jung SL, Cho KH, Cho MH, Kim SJ, Jang H. Wear mechanism of multi-
phase friction materials with different phenolic resin matrices. Wear
2009;266:739–44.
[32] Kukutschová J, Roubíček V, Malachová K, Pavlíčková Z, Holuša R, Kubačková J,

109
L. Wei, et al. Tribology International 138 (2019) 99–110

et al. Wear mechanism in automotive brake materials, wear debris and its potential sliding behaviour of Al-based metal-matrix-composites against friction material.
environmental impact. Wear 2009;267:807–17. Wear 2004;256:754–63.
[33] Persson BNJ. Sliding friction : physical principles and applications. second ed. [37] Eriksson M, Bergman F, Jacobson S. On the nature of tribological contact in auto-
Berlin: Springer; 2000. motive brakes. Wear 2002;252:26–36.
[34] Bijwe J, Aranganathan N, Sharma S, Dureja N, Kumar R. Nano-abrasives in friction [38] Persson BNJ. On the fractal dimension of rough surfaces. Tribol Lett
materials-influence on tribological properties. Wear 2012;296:693–701. 2014;54:99–106.
[35] Straffelini G. Friction and wear: methodologies for design and control. Springer; [39] Menapace C, Leonardi M, Perricone G, Bortolotti M, Straffelini G, Gialanella S. Pin-
2015. on-disc study of brake friction materials with ball-milled nanostructured compo-
[36] Straffelini G, Pellizzari M, Molinari A. Influence of load and temperature on the dry nents. Mater Des 2017;115:287–98.

110