C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1

.sc iencedi rect .com

Avai lab le at www

ScienceDirect

journal homepage: www.elsev ier .com/ locate /carbon

Enhanced tribological properties of MWCNT/Ni bulkcomposites – Influence of processing on frictionand wear behaviour

0008-6223/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.carbon.2013.08.054

* Corresponding author: Fax: +49 681 70502.E-mail address: c.gachot@mx.uni-saarland.de (C. Gachot).

Sebastian Suarez, Andreas Rosenkranz, Carsten Gachot *, Frank Mucklich

Department of Materials Science, Saarland University, Campus D3.3, D-66123 Saarbrucken, Germany

A R T I C L E I N F O

Article history:

Received 22 May 2013

Accepted 24 August 2013

Available online 5 September 2013

A B S T R A C T

The friction and wear behaviour of MWCNT-reinforced Ni matrix bulk composites is inves-

tigated and compared to the response of a pure bulk reference sample. Several effects are

observed in the composites such as microstructural refinement, leading to improved hard-

ness, coefficient of friction and wear resistance. Said refinement might also be responsible

for a non-trivial behaviour of hot-pressed composites under a 300 mN load, due to higher

oxidation rates. The worn volume is analysed and discussed from a microstructural and

morphological point of view. In this context, a direct correlation is shown between two rel-

evant wear parameters, namely, the wear constant K and the cutting efficiency fAB. In addi-

tion, the tribochemistry is investigated by Raman spectroscopy and linked to the friction

and wear behaviour. Finally, the wear volume reduction peaked at 94.1% in HUP samples.

This clearly indicates the high potential of these composites for tribological applications.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction contrasted with a pure Ni reference in order to determine

In recent years, carbon nanotubes (CNT) have been foreseen

as the ultimate reinforcing phase for composites due to their

outstanding intrinsic physical properties, particularly

mechanical [1] and thermal [2]. The main drawback is their

processing difficulty due to strong van der Waals interactions

that cause agglomerates. This reduces the aforementioned

properties as well as the integration as a second phase in

composites.

Previous work on CNT-reinforced ceramic [3,4], and metal

matrix composites [5–11] has reported improvements in their

mechanical properties and friction performance. To the best

of our knowledge, no direct correlation to the microstructural

refinement produced by the CNTs has been attempted regard-

ing this type of composites.

In the present work, the results obtained for CNT-

reinforced Ni bulk composites manufactured by two different

powder metallurgy routes are shown. These results are

the influence of the CNTs on the friction behaviour. Said addi-

tion is also discussed and analysed with respect to the result-

ing composite microstructure and its subsequent correlation

to the wear and friction behaviour. Finally, the conventional

Archard wear model was adjusted to achieve a more suitable

description of the microstructural influence on the wear rate.

2. Materials and methods

Multiwalled CNTs (MWCNT) commercially available as Baytu-

bes C150P (purity: >95%; outer diameter distribution: 5–20 nm;

agglomerate size: 0.1–1 mm) and Ni dendritic powder (Alfa

Aesar, average particle size: 44 lm) were used as starting

materials. The composite powders were blended by a colloidal

mixing process, which is described elsewhere [12]. The ob-

tained blend was then consolidated with an axial pressure

of 990 MPa to form cylinder-shaped samples 8 mm in

C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1 165

diameter. The green pellets containing 1 wt.% (6.47 vol%) of

MWCNT were then separated into two sets of samples. The

first was densified by pressureless sintering (CPS) under vac-

uum (10�3 Pa) at 1223 K for 2.5 h. The second set was densified

in a hot unaxial press (HUP) under vacuum (10�4 Pa) at 1023 K

also for 2.5 h. The resulting relative densities were deter-

mined by water buoyancy (Archimedes) resulting in 95%,

95.5% and 98% for pure Ni, MWCNT/Ni 1% CPS and

MWCNT/Ni 1% HUP respectively. Raman spectroscopy was

performed on a LabRAM ARAMIS instrument (HORIBA Scien-

tific) with an unfiltered 532 nm laser beam. The optical ele-

ments of the Raman microscope used in this analysis

produced a laser spot size of 5 lm (full width at half maximum)

and depth penetration of approximately 1 lm, whilst a spectral

resolution around 5 cm�1. The microstructural characterisation

was carried out with a focused ion beam/scanning electron dual

beam microscope (FIB/SEM) (Helios NanoLabTM 600, FEI Com-

pany). Furthermore, the grain size distribution was studied with

an electron backscattered diffraction (EBSD) detector by apply-

ing a scanning voltage and current of 20 kV and 22 nA, respec-

tively. The scanned areas were measured with a step size of

0.25 lm. Finally, the microhardness was measured with a Leica

Vickers microindenter under an indentation load of 981 mN

(HV0.1) for 15 s, performing 20 measurements per sample.

The tribological tests were performed with a nanotribom-

eter using a ball on disc configuration in a linear reciprocating

sliding mode (CSM Instruments) with a stroke length of

0.6 mm and a total sliding length of 60 mm. The normal force

was set to four different loads, namely, 50, 100, 200 and

300 mN and the linear sliding speed to 1 mm/s. The counter-

body consisted of a typical Al2O3 ball with a diameter of

3 mm. Temperature and relative humidity were kept constant

at 20 ± 2 �C and 4 ± 1%. Subsequently, the wear volume was

determined by means of white light interferometry (WLI) with

a Mirau configuration.

Fig. 1 – EBSD inverse pole figure maps of the surface prior to the

(a) pure Ni, (b) MWCNT/Ni 1 wt.% CPS and (c) MWCNT/Ni 1 wt.% H

The respective grain distribution histograms. (A colour version

3. Results and discussion

3.1. Microstructural characterisation

Fig. 1 shows EBSD surface scans of the studied samples and the

respective grain distribution histograms. It can be observed that

the size of the grains decreases by adding CNTs to the compos-

ite [13]. This particularity has been already clarified and is gen-

erated by a grain growth stagnation produced by the CNTs.

The main difference between the reference sample and

the MWCNT-reinforced samples is that the grain growth in

the former is only hindered by the porosity. During sintering,

porosity migrates to grain boundaries, reducing the energy

and slowing the growth, whereas in the MWCNT-containing

samples, the nanotubes provide an extra hindering phase

which further reduces the grain boundary mobility [14].

The difference between the two CNT-containing samples

is mainly due to the higher amount of agglomerates in the

CPS composite. With this production technique, a higher

degree of reagglomeration due to poor cohesion between

the matrix and the CNTs has been observed, related to the

absence of pressure during sintering [12,15].

This difference can be noticed in Fig. 2. It shows fracture

surfaces of the MWCNT/Ni 1 wt.% CPS (Fig. 2a) and HUP

(Fig. 2b) samples. The agglomerates and voids in the CPS sam-

ple are highlighted with an ellipse on the figure. The disparity

in the CNT distribution within the samples is due to a reag-

glomeration process, typical of the manufacturing method.

Another difference between CPS and HUP is the densifica-

tion temperature utilised during sintering [16]. However, this

is not the only factor playing a role on the final grain size and

density, but the applied pressure during HUP improves the den-

sification by avoiding the swelling (or gas bloating) of the sam-

ple. Moreover, in HUP there are other densification mechanisms

that do not act in CPS (i.e. plastic deformation and creep).

tribological tests, highlighting the grain size distribution of

UP. The scale bar in (c) differs from those in (a) and (b). (e)–(g)

of this figure can be viewed online.)

Fig. 2 – Fracture surfaces of the CNT-containing samples. (a) MWCNT/Ni 1 wt.% CPS and (b) MWCNT/Ni 1 wt.% HUP. The CNT

agglomerates, voids and dispersed CNTs are highlighted. (A colour version of this figure can be viewed online.)

166 C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1

The increment in the mechanical properties is closely re-

lated to the microstructural refinement. By measuring the

microhardness of the composites and comparing these values

to pure Ni, the increase in the hardness could be related to the

reduction in the final grain size. This translates to an enlarge-

ment of the net grain boundary, hindering the dislocation

movement through the lattice, thus improving the mechani-

cal properties. A summary of the mean grain size and microh-

ardness values for the tested samples is found in Table 1.

3.2. Friction behaviour

The friction coefficient was evaluated for loads ranging from

50 to 300 mN for the three samples sets. Fig. 3 depicts the val-

ues for each load, showing a clear difference between the

CNT-containing samples in comparison to the Ni reference.

In the case of the 1 wt.% HUP sample, the maximum COF

reduction is about 67%, whereas in the CPS sample the reduc-

tion is close to 50%, both of which are above the expected val-

ues reported in the literature [8,17]. This significant difference

could be related to several isolated effects or a combination

thereof.

For the sake of the analysis, it was decided to separate the

discussion of the friction results into two different regions.

The first one covers the lower loads, i.e. 50 and 100 mN. The

second group covers the higher tested load (200 and

300 mN), where a non-trivial behaviour is observed particu-

larly in the HUP sample.

3.2.1. Region I – lower loadsRegarding the first region, it can be noticed that the evolution

of the COF throughout the friction experiment is roughly the

Table 1 – Mean grain size, Vickers microhardness and root mea

Sample Mean grain size (lm) Vicke

Pure Ni 47.54 ± 3.06 92.2 ±MWCNT/Ni 1 wt.% CPS 22.39 ± 0.96 119.3MWCNT/Ni 1 wt.% HUP 7.58 ± 0.62 131.5

same for the three types of samples, showing no particular

influence of the load on the friction behaviour (Fig. 4).

The obtained mean COFs were 0.4, 0.2 and 0.1 for the pure

Ni, 1% CPS and 1% HUP samples, respectively. The reference

value is in good agreement with those reported in the litera-

ture for microcrystalline Ni tested against Al2O3 [18]. Eq. (1)

is used to calculate the Hertzian stresses, where P is the ap-

plied normal load, D is the counterpart diameter and C is

the effective elastic module. The latter can be estimated with

Eq. (2):

rH ¼ 0:918

ffiffiffiffiffiffiffiffiffiffiffiP

D2C2

3

rð1Þ

C ¼ 1� m21

E1

� �þ 1� m2

2

E2

� �ð2Þ

Considering the tribological pair being tested (Ni/Al2O3)

and combining Eqs. (1) and (2) with the data available in the

literature, the initial (Hertzian) elastic contact pressure be-

tween these materials amounts to 1.56 GPa. This value highly

exceeds the estimated yield strength of the samples (obtained

from the relationship ry � hardness/3), which ranges from 0.3

to 0.45 GPa (pure Ni and 1% HUP respectively). Therefore, un-

der equal load, the yield strength is lower, resulting in a great-

er penetration depth and a subsequently increased real

contact area. In other words, the increase in hardness of the

sample hinders the indentation depth of the dynamic coun-

terpart, thus reducing the contact area during the friction

test. It would then be reasonable to state that the main role

in the friction behaviour at the mentioned loads is played

by an indentation-induced plastic deformation mechanism.

This effect was also observed in the work of Shafiei et al.

n square roughness of the samples.

rs microhardness HV0.1 Surface roughness – rms (nm)

2.6 18.8 ± 17.7± 5.3 26.7 ± 7.1± 7.0 46.9 ± 10.2

Fig. 3 – Average coefficient of friction (COF) versus the different experimental loads under dry conditions. (A colour version of

this figure can be viewed online.)

Fig. 4 – Mean COF evolution over the sliding cycles for lower loads. Experimental loads for each are shown on the upper right

corner. (A colour version of this figure can be viewed online.)

C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1 167

[18], where they compare the tribological performance of Ni

with different grain sizes (micro and nanocrystalline) and

mechanical properties. Since all the mean grain sizes in this

work are in the micrometre range, our assumption can be cor-

related to said study. Nevertheless, another important factor

is the presence of a distribution of individual CNTs as well

as clusters, which are believed to act as self-lubricating mate-

rials [6,8]. As mentioned before, the frictional behaviour can

be a combination of effects which are difficult to isolate from

each other.

3.2.2. Region II – higher loadsFor higher loads, the COF evolution of the reference and CPS

samples remain almost unchanged, whereas in the case of

the HUP sample (300 mN), an overlap appears between 50

and 75 cycles (Fig. 5).

After this point, the COF shows a marked increase sur-

passing the COF of the CPS sample. It is clear in this case, that

the major influence is not particularly related to the mechan-

ical properties of the composites, but to an underlying effect.

In order to understand the experimental findings and to elu-

cidate the aforementioned effect, the chemistry of the worn

surfaces was studied by Raman spectroscopy. Focus was put

on the 300 mN tracks, believed to contain higher chemical

activity. Fig. 6 depicts the analysed sectors of each track and

their correspondent Raman spectra. There is a clear transition

of the wear activity from severe (in the case of the reference

sample) to mild wear on the hot-pressed sample. Regarding

the chemistry, nickel oxide (NiO) was detected in all the wear

tracks [19], sporadically in pure Ni and MWCNT/Ni 1%CPS yet

continually in the 1% HUP wear tracks. This may be due to the

fact that the larger grains have lower interfacial strength lead-

ing to a grain peeling mechanism in the most superficial lay-

ers [7]. Moreover, the pure Ni and 1%CPS samples show

spallation of the oxide layer with a visible amount of break-

age, also resulting in oxide depleted regions.

It is well documented that there is a direct relationship be-

tween the microstructural characteristics of a polycrystalline

material and its oxidation kinetics. It means that the lower

the mean grain size, the higher the oxide layer formation rate.

This is attributed to the larger grain boundary area which fur-

ther translates to an improvement in the diffusion paths for

oxygen [18,20]. This formed tribolayer may be both beneficial

and detrimental to friction. The upside is that an oxide layer

may provide a decrease in the friction coefficient and wear

rates in a first stage as studied by Stott et al. [21]. These

so-called glazes, when subjected to low speed reciprocating

sliding conditions, tend to be compacted providing wear

Fig. 5 – Mean COF evolution over the sliding cycles for higher loads. Experimental loads for each are shown on the upper right

corner. (A colour version of this figure can be viewed online.)

Fig. 6 – SEM micrographs of the worn surfaces and their correspondent Raman spectra. From left to right: pure Ni, MWCNT/Ni

1%CPS and MWCNT/Ni 1% HUP tested at 300 mN. S.D. stands for sliding direction. (A colour version of this figure can be

viewed online.)

168 C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1

protection. However, due to the brittle nature of the oxide

scales, the indentation of the dynamic counterpart will break

the glaze and increase the roughness of the track, i.e. in-

creased coefficient of friction [22]. Particularly for NiO [8], it

has been found that it acts as a high shear strength layer

which increases the friction coefficient.

Furthermore, the quality of the CNTs can be estimated

from the spectral data. The intensity of the D band illustrates

the amount of defects and structural modification in the

nanotubes, thus, as is widely accepted, the D to G band ratio

accounts for the quality [8]. In this case, considering a refer-

ence ID/IG mean ratio of 0.89 (state of the CNTs prior to the tri-

bological tests), the corresponding ratios for 1%CPS is

1.09 ± 0.07 and 0.93 ± 0.11 for 1% HUP. This leads to infer that

the CNTs were strongly tribomechanically active in the 1%CPS

during the test as a friction interface which prevents asperity

contact. On the other hand, in the 1% HUP sample the CNTs

were not as operative throughout the process, being the oxide

layer the main interfacial protection.

As mentioned before, it is highly unlikely that only one

type of friction mechanism is acting in the process, rather, a

combination of effects is the more plausible case. Some

authors relate the difference between CNT-reinforced bulk

composites and the plain matrix to an effect exclusively

brought on by the CNTs. In the study carried out by Scharf

et al. [8], they observed the formation of a graphitic film which

presents low shear strength and favours lubrication. On the

other hand, Dong et al. [6] state that the main effect occurring

in the friction behaviour of CNT-reinforced metal matrix

composites is also the formation of a carbon film, which

Table 2 – Wear constant K and cutting efficiency fAB valuescalculated from the experimental data.

Sample Wear constantK (10�5)

Cuttingefficiency fAB

Pure Ni 17.50 ± 13.00 0.60 ± 0.15MWCNT/Ni 1%CPS 8.75 ± 2.39 0.76 ± 0.13MWCNT/Ni 1% HUP 2.03 ± 0.56 0.24 ± 0.15

C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1 169

decreases the friction coefficient and therefore retards severe

wear. Another important mechanism that is frequently as-

sumed in this type of friction is the enhancement of the inter-

facial strength between the CNTs and the metallic matrix [7].

It is believed that the increased interfacial strength retards

the peeling of the matrix grains, thus reducing wear. A further

hypothesis formulated in the literature states that during the

friction tests, the CNTs are released to the surface acting as

spacers which prevent the asperities from contacting and,

to some extent, acting as a lubrication medium [5].

3.3. Wear behaviour

The evolution of the removed volume follows a linear incre-

ment in all three samples, but with different slopes (Fig. 7).

This behaviour can be correlated to the Archard model, which

states that there is an inversely proportional relationship be-

tween the wear rate and the mechanical properties (hardness)

of the tested material. The aforementioned slopes are repre-

sented in the equation by a proportionality constant symbol-

ised by K. For composite materials, this relationship can be

further reformulated by considering the Hall–Petch effect

generated by the different microstructural refinement

brought by the reinforcing phases [23], leading to a modified

Archard model. By combining both relationships, the follow-

ing equation is obtained:

W ¼ KLSH

� �¼ K

LS

H0 þ kd�0:5

� �ð3Þ

where W is the wear rate per unit sliding distance, K is the

wear coefficient, L is the applied load, S is the total sliding dis-

tance, H0 is the hardness of a defect-free single crystal of the

bulk material, k is a proportionality constant and d is the

mean grain size of the composites. From this equation, the

influence of the microstructure on the wear rate during slid-

ing wear can be directly estimated, the wear rate being pro-

portional to the mean grain size of the material.

The estimation of the wear constant K from the

experimental data reveals interesting results for the tested

Fig. 7 – 3D plot of the wear volume as a function of the applied

irregularity (wear track depth and width) of the wear tracks. (A

materials. There is no unanimous agreement about what this

constant physically represents, but Rigney has summarised

the most accepted interpretations [24]. Most of these express

the generation of third bodies and the relationship between

the contact and the asperities. Another explanation states

that this constant relates the removed to the plastically de-

formed volume. Since the author affirms that researchers

should be cautious in interpreting and extrapolating the

experimental wear rates, it would be useful to correlate this

constant to a supplementary factor which might help to iden-

tify the wear mechanisms affecting sample response. A help-

ful tool for the determination of the acting mechanisms in

sliding friction from morphological data is provided by the

Zum Gahr equation [25] (also named cutting efficiency fAB):

fAB ¼P

Abelow �P

AabovePAbelow

¼ B� AB

ð4Þ

where B is the cross-sectional area of the wear groove below

an arbitrary reference plane, A represents the summation of

the cross-sectional areas of the material displaced to the

sides and/or within the track as ridges. The values of fAB range

from 0 for a ploughing mechanism (pure plastic deformation)

to 1 for a cutting mechanism (pure abrasive effect).

Observing both the cutting efficiency and wear constant

results (Table 2), the pure Ni and 1%CPS samples have similar

values, tending to a cutting mechanism. Considering the rel-

atively scattered values, this may be mainly attributed to

irregularities of the analysed wear tracks. Both show a high

K value representing more removed than plastically deformed

load and mean grain size. The error is originated from the

colour version of this figure can be viewed online.)

170 C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1

volume. Furthermore, the fact that CNT agglomerates are lar-

ger in size and quantity in the CPS samples (compared to HUP

samples) [12] makes them tend to roll over the wear surface

until they are removed without being taken apart [26]. On

the other hand, the 1% HUP sample shows a low fAB value,

which indicates that ploughing is the dominant mechanism.

This is consistent with the lower K value for this composite,

which represents a lower amount of removed versus plasti-

cally deformed volume (according to the interpretation given

before). These low values of K and fAB for the HUP samples

could be related to a redeposition of broken oxide particles

that are somehow embedded within the matrix throughout

the process.

Summarising, since it is very complex to interpret either

constant alone, a useful and straightforward way might be

by analysing both together. For this particular case, a good

correlation between both is observed.

The shown results are helpful to approach an elucidation

of certain aspects of the friction and wear behaviour of these

CNT-reinforced bulk composites. It can be concluded that the

influence of the CNTs has two effects:

• A fundamental role in the microstructural refinement

which directly affects the friction, by modifying the oxida-

tion kinetics, thus generating an oxidation wear mecha-

nism; and the wear behaviour, through a grain boundary

strengthening effect, which increases the hardness of the

material.

• At high loads on the samples with higher CNT (friction)

activity (higher CPS ID/IG ratio, i.e. more active CNTs), they

might act as an interfacial layer which performs as a lubri-

cation medium.

4. Conclusions

Tribochemical studies proved the formation of an oxide layer

formation in all samples, and in the CNT-containing samples,

a carbon coating film was also observed. The CNTs showed

activity throughout the friction process, acting as an interfa-

cial layer which may act as a friction-reducing agent. The for-

mation of a uniform Ni oxide layer in the HUP sample proved

to be detrimental to its friction behaviour. This can be traced

back to the reduction of the lubricating effect by keeping the

CNTs away from the counterpart.

Non-trivial friction behaviour is observed in the lower

grain size sample which might be related to the enhanced

oxidation kinetics and the subsequent breakage of the formed

oxide layer.

For low loads (50–100 mN), the main friction mechanism is

related to the indentation depth of the dynamic counterpart,

whereas for higher loads (200–300 mN) oxidation plus interfa-

cial lubrication is more significant.

The main influence of the CNTs in the friction and wear

behaviour is related to the microstructural refinement, which

improves both the oxidation kinetics and the mechanical

properties of the composites.

A modified Archard model (considering the mean grain

size influence) was applied and complemented by a

morphological analysis (cutting efficiency) in order to under-

stand the wear mechanisms acting in the process. It was

found that for the coarse-grained samples, grain peeling is

preponderant, whereas redeposition arises in the refined

samples.

Acknowledgements

S.S. wishes to thank the German Academic Exchange Service

(DAAD) for financial support. All authors wish to acknowledge

the EFRE Funds of the European Commission for support of

activities within the AME-Lab project.

R E F E R E N C E S

[1] Lu Y, Liaw PK. The mechanical properties of nanostructuredmaterials. JOM 2001;53(3):31–5.

[2] Kim P, Shi L, Majumdar A, McEuen P. Thermal transportmeasurements of individual multiwalled nanotubes. PhysRev Lett 2001;87:19–22.

[3] An JW, You DH, Lim DS. Tribological properties of hot-pressedalumina–CNT composites. Wear 2003;255:677–81.

[4] Lim DS, You DH, Choi HJ, Lim SH, Jang H. Effect of CNTdistribution on tribological behavior of alumina–CNTcomposites. Wear 2005;259:539–44.

[5] Chen W, Tu J, Wang L, Gan H, Xu Z. Tribological application ofcarbon nanotubes in a metal-based composite coating andcomposites. Carbon 2003;41:215–22.

[6] Dong SR, Tu JP, Zhang XB. An investigation of the sliding wearbehavior of Cu-matrix composite reinforced by carbonnanotubes. Mater Sci Eng A 2007;313(1–2):83–7.

[7] Kim KT, Cha SI, Hong SH. Hardness and wear resistance ofcarbon nanotube reinforced Cu matrix nanocomposites.Mater Sci Eng A 2007;449–451:46–50.

[8] Scharf TW, Neira A, Hwang JY, Tiley J, Banerjee R. Self-lubricating carbon nanotube reinforced nickel matrixcomposites. J Appl Phys 2009;106:013508.

[9] Tan J, Yu T, Xu B, Yao Q. Microstructure and wear resistanceof nickel–carbon nanotube composite coating from brushplating technique. Tribol Lett 2006;21:107–11.

[10] Guiderdoni C, Estournes C, Peigney A, Weibel A, Turq V,Laurent C. The preparation of double-walled carbonnanotube/Cu composites by spark plasma sintering, andtheir hardness and friction properties. Carbon2011;49:4535–43.

[11] Guiderdoni C, Pavlenko E, Turq V, Weibel A, Puech P,Estournes C, et al. The preparation of carbon nanotube(CNT)/copper composites and the effect of the number ofCNT walls on their hardness, friction and wear properties.Carbon 2013;58:185–97.

[12] Suarez S, Soldera F, Gonzalez Oliver C, Acevedo D, Mucklich F.Thermomechanical behavior of bulk Ni/MWNT compositesproduced via powder metallurgy. Adv Eng Mater2012;14:499–502.

[13] Huang Q, Gao L, Liu Y, Sun J. Sintering and thermal propertiesof multiwalled carbon nanotube–BaTiO3 composites. J MaterChem 2005;15:1995–2001.

[14] Liu Y, Patterson B. Grain growth inhibition by porosity. ActaMetall Mater 1993;41:2651–6.

[15] Yamanaka S, Gonda R, Kawasaki A, Sakamoto H, Mekuchi Y,Kuno M, et al. Fabrication and thermal properties of carbonnanotube/nickel composite by spark plasma sinteringmethod. Mater Trans 2007;48:2506–12.

C A R B O N 6 6 ( 2 0 1 4 ) 1 6 4 – 1 7 1 171

[16] Kang S. Sintering densification, grain growth &microstructure. 1st ed. Oxford: Elsevier Butterworth-Heinemann; 2005.

[17] Chen X, Peng J, Li X, Deng F, Wang J, Li W. Tribologicalbehavior of carbon nanotubes – reinforced nickel matrixcomposite coatings. J Mater Sci Lett 2001;20:2057–60.

[18] Shafiei M, Alpas AT. Friction and wear mechanisms ofnanocrystalline nickel in ambient and inert atmospheres.Metall Mater Trans A 2007;38:1621–31.

[19] Mironova-Ulmane N, Kuzmin A, Steins I, Grabis J, Sildos I,Pars M. Raman scattering in nanosized nickel oxide NiO. JPhys Conf Ser 2007;93:012039.

[20] Singh Raman RK, Khanna AS, Tiwari RK, Gnanamoorthy JB.Influence of grain size on the oxidation resistance of 2 1/4 Cr–1Mo steel. Oxid Met 1992;37:1–12.

[21] Stott F, Wood G. The influence of oxides on the friction andwear of alloys. Tribol Int 1978;11:211–8.

[22] Stott F. The role of oxidation in the wear of alloys. Tribol Int1998;31:61–71.

[23] Farhat ZN, Ding Y, Northwood DO, Alpas AT. Effect of grainsize on friction and wear of nanocrystalline aluminum. MaterSci Eng A 1996;206:302–13.

[24] Rigney D. Some thoughts on sliding wear. Wear1992;152:187–92.

[25] Vilar R, Colaco R. Laser-assisted combinatorial methods forrapid design of wear resistant iron alloys. Surf Coat Technol2009;203:2878–85.

[26] Ni B, Sinnott SB. Tribological properties of carbon nanotubebundles predicted from atomistic simulations. Surf Sci2001;487:87–96.