Journal of Minerals and Materials Characterization and Engineering, 2012, 11, 735-743
Published Online July 2012 (http://www.SciRP.org/journal/jmmce)
Effect of Surface Treatment on Tribological Behavior of
Ti-6Al-4V Implant Alloy
B. K. C. Ganesh1*, N. Ramanaiah1, P. V. Chandrasekhar Rao2
1Department of Mechanical Engineering, College of Engineering (Autonomous), Andhra University, Visakhapatnam, India
2Department of Mechanical Engineering, L. B. R. College of Engineering, Vijayawada, India
Received May 20, 2012; revised June 30, 2012; accepted July 15, 2012
ABSTRACT
Titanium alloys are extensively used in various fields of engineering, medicine, aerospace, marine due to its excellent
mechanical properties. Its usage is more pronounced today in the field of biomedical implants due to its superior bio-
compatibility, corrosive resistance and high strength to weight ratio. It has poor abrasive wear resistance due to high
coefficient of friction and low thermal conductivity. Poor abrasive wear resistance results in the formation of wear de-
bris at the implant area causing toxicity, inflammation and pain. Surface treatment of the implant alloy through heat
treatment, application of protective coatings, introduction of compressive residual stresses by shotpeening and shot
blasting are some of the methods to mitigate wear of the implant alloy. In this work Ti-6Al-4V implant alloy is treated
under various conditions of heat treatment, shotpeening and shot blasting operations on a pin on disc wear testing ma-
chine. Shotpeening and Shot blasting are the operations usually performed on this alloy to improve fatigue strength and
surface roughness. In this work the effect of above surface treatments were studied on the wear behavior of Ti-6Al-4V
implant alloy and an improvement in the wear resistance of the alloy is reported. Scanning Electron micrograph (SEM)
along with Energy Dispersive Spectrometry analysis (EDS) is done to authenticate the experimental results obtained
during the wear testing procedure.
Keywords: Heat Treatment; Shotpeening; Wear Resistance; Sliding Velocity
1. Introduction
The bulk properties of biomaterials such as non-toxicity,
corrosion resistance or controlled degradability, modulus
of elasticity and fatigue strength have long been recog-
nized to be highly relevant in terms of the selection of the
right implant alloy. The events after implantation include
interactions between the biological environment and arti-
ficial material surfaces, onset of biological reactions as
well as particular response paths chosen by the body. The
material surface plays an extremely important role in the
response of biological environment to the artificial me-
dical devices. The proper surface modification tech-
niques not only retain the excellent bulk attributes of
titanium and its alloys, such as low modulus, good fa-
tigue strength, formability and machinability, but also
improve specific surface properties such as wear and
corrosive resistance [1].
Titanium has a good biocompatibility but it has a high
coefficient of friction which results in poor abrasive wear
resistance. The property of poor abrasive resistance is
important in generating a wear debris, when the artificial
implant is in contact with the healthy and natural joint.
The accumulation of wear debris causes inflammation,
pain and finally loosening of the joint. Implant wear is a
common phenomenon which is resulted due to high fric-
tion between artificial materials which is higher than that
of healthy and natural joints. This high friction is attrib-
uted to the high rigidity of artificial materials leading to
non-recoverable wear [2].
Wear resistance plays an important role whenever a
material such as a bone plate for fracture fixation is at-
tached to fractured bones of a different stiffness values
and modules of elasticity. Relative movements between
the two different materials such as bone, implant material
and between the parts of multi-component systems will
occur when a cyclic load is applied to the system. These
relative movements causes a wear stress on the attach-
ment devices (bone screws as well as on the eyelets).
Therefore high wear resistance is required for orthopedic
implants to obtain biocompatibility and acceptability [3].
The drawback of extensive use of titanium alloys in
hip replacement and other artificial joints is poor tri-
bological properties such as poor abrasive, poor fretting
behavior and high coefficient of friction. The improve-
*Corresponding author.

B. K. C. GANESH ET AL.
736
ment of the above properties can be done with the help of
four main mechanisms as suggested by Zhecheva et al.
[4]. They are as follows:
1) To induce a compressive residual stress;
2) To decrease the coefficient of friction;
3) To increase the hardness;
4) To increase the surface roughness.
Mitsuo [5] reported that the application of stress by
rapid quenching results in the formation of martensitic
structure in steels which contain residual austenite in
their microstructures. This phenomenon is called stress
or strain induced martensitic transformation which en-
hances ductility or fracture toughness of steel. Deforma-
tion induced martensitic transformation also occurs in
titanium where unstable beta phase is retained at room
temperature by rapid cooling such as water quenching
from a high temperature near the beta transus tempera-
ture. The formation of this martensitic structure or trans-
formed beta is important in improving the hardness of the
alloy by which the implant wear can also be minimized.
The surface modification and change in microstructure
can be obtained by heat treating the various samples at
beta transus temperature, (transformation temperature)
where primary α changes from hexagonally closely packed
crystallographic structure (α) to body centered cubic crys-
tallographic structure (β). These specimens were subse-
quently cooled by rapid quenching by water, cooling in
the furnace and air, followed by aging of these alloys at
500˚C for a period of one hour. The objective of con-
ducting the heat treatment cycle is to improve the hard-
ness of the alloy which is beneficial in increasing the
wear resistance of the alloy by retaining the transformed
beta at the room temperature. The introduction of com-
pressive residual stresses in the surface layer by surface
modification technique such as shotpeening, shot blasting
and ion-beam enhanced deposition etc. also mitigate the
wear of implant material [6].
Wear rate calculation by pin and rotating disc machine
using weight loss method is one of the common tech-
niques to evaluate dry sliding wear behavior of titanium
implant materials. A. Molinari et al. [7] investigated on
dry sliding wear mechanism of Ti-6Al-4V alloy. In their
experimental work it has been found that the wear vol-
ume of the rotating specimens is reported as a function as
sliding speed and the load applied on the pin. An increase
in wear volume is resulted with an increase in applied
load. O. Alam et al. [8] in their experimental work of dry
sliding wear on Ti-6Al-4V alloy have reported that under
a constant load of 45 N applied on pin, an increase of
wear rate was identified up to a sliding distance of 500
meters. There after a steady state is attained during which
no appreciable change in the wear rate behavior of the
alloy was observed. S. J. Li et al. [9] had conducted ex-
perimental analysis on wear testing of Ti-6Al-4V alloy
on various disc materials such as steel, Ultra high mo-
lecular weight polyethylene (UHMWPE) and bone of pig
to represent actual working of the pin material in realistic
operating environment. His experimental results indicate
a decreasing trend in wear volume of the implant, while
working on steel disc to UHMWPE and bone of pig. The
results further concluded that irrespective of the disc ma-
terial used, wear volumes reported are similar with re-
spect to a change in various process parameters such as
type of heat treatment, increase in the sliding velocity etc.
Therefore in this experimental set up steel disc is only
used as disc material and the results are reported.
2. Materials and Methods
The implant material is procured from South Asia Metal
Corporation, Mumbai. The chemical composition by
weight of the metals are as follows: 89.6% Titanium,
6.29% Aluminium, 3.95% Vanadium, 0.09% Iron, 0.029%
Carbon.
The implant alloys are heat treated at above transfor-
mation temperature (950˚C) for one hour in an argon
controlled atmosphere. One set of specimens were sub-
jected to shotpeening operation at 4.5 and 3.5 bar pres-
sure respectively. Another set of specimens were sub-
jected to shot blasting for duration of 5 to 20 min at a
pressure of 4.5 bar. The pins were cut according to the
standard dimensions as shown in the Figure 1 for the
evaluation of wear behavior. Scanning Electron Micro-
graph (SEM) analysis is also conducted to study the wear
45˚
Ø3
30
(a)
8
Ø165
(b)
Figure 1. Dimensions of the (a) friction pin; (b) rotating disc
in millimeters.

B. K. C. GANESH ET AL. 737
track behavior of various heat treated specimens. Hard-
ness values are measured by Vickers micro hardness
testing machine at a constant load of 0.5 kg. Average
surface roughness (Ra) value of the surface treated
specimens was measured by Mitutayo surface roughness
testing instrument.
The shot peening operation was performed according
to the MIL-13165 standards. The various shot peening
parameters are given under: type of shot: S230, material
of shot: steel shot, angle of projection: 90˚, diameter of
shot: 0.6 mm, duration of peening: 30 min, coverage area:
100%. The various shot blasting parameters are: type of
grit: G80, material of the grit: steel grit, distance from
nozzle to work piece: 30 mm, angle of projection: 90˚.
All the treated specimens were tested for wear behavior
by TR-20LE Ducom wear testing machine at a constant
velocity of 1 m/sec and with a sliding distance of 500
metres. Wear rate is calculated on the basis of volume of
the material removed before and after conducting the
wear test.
Loads acting on human joint vary considerably from
joint to joint. For a particular joint, it varies with time
during the loading cycle [10]. It has been reported that
stresses in the living area are of the order of 1 MPa. Gis-
pert et al. [11] used a normal pressure of 0.88 MPa. In
this work a load of 50 N with a pin diameter of 10 mm is
used to obtain a pressure of 0.7 MPa which is considered
to be safe stress acting on the joint during the loading
conditions.
3. Results and Discussion
3.1. Wear Track Analysis
The following Figure 2 shows the wear track micro-
graphs of alloy when subjected to heat treatment and shot
peening operations. It is evident from the microstructures
that the alloy Ti-6Al-4V (hereafter referred as Ti64) has
the presence of a protective layer in its water quenched
and air cooled specimens. The presence of this layer is
only seen when the alloy is cooled at a faster rate such as
water quenching and air cooling. This layer cannot be
seen in the specimen which is subjected to slow cooling
of the alloy in a furnace. Figure 3 shows the scanning
electron micrograph (SEM) of water quenched specimen
shown in Figure 2(b). The energy dispersive spectrome-
try (EDS) of the base and water quenched specimen is
shown in Figure 4.
From the EDS analysis of the base wear specimen it is
clear that the oxygen and carbon content are limited to
0.56 and 2.34 percentage by weight respectively. But
when the specimen are heat treated above the beta tran-
sus temperature and quenched with water there is a for-
mation of a protective layer as can be seen from the Fig-
ure 3. The EDS analysis of the protective layer shows
(a) (b)
(c) (d)
(e) (f)
Figure 2. Wear Track of Ti-6Al-4V subjected to various
heat treatment processes and shotpeening operations at 20×
magnification. (a) Base material; (b) Water quenched; (c)
Air cooled; (d) Furnace cooled; (e) Shotpeened at 4.5 bar; (f)
Shotpeened at 3.5 bar.
protective layer
Figure 3. Scanning electron micrograph of water quenched
wear specimen.

B. K. C. GANESH ET AL.
738
Element Weight% Atomic%
C K 0.56 2.10
O K 2.34 6.53
Al K 1.79 2.96
Ti K 90.36 84.13
V K 4.41 3.86
Fe L 0.53 0.83
Total 100
(a)
Element Weight% Atomic%
C K 1.11 3.49
O K 12.14 28.57
Al K 1.60 2.24
Ti K 71.47 56.19
V K 4.30 3.18
Fe L 9.38 6.32
Total 100
(b)
Figure 4. (a) EDS analysis of base wear specimen; (b) EDS
analysis of protective layer.
high amount oxidation with its weight% at 12.14 and
carbon percentage of 1.11% as shown in Figure 4(b).
Further the EDS analysis of the some portion of the pro-
tective layer also indicates a higher amount of metal to
metal contact with increasing percentage of iron content
when in contact with the steel disc. It is evident from the
literature that when titanium and its alloys are exposed to
oxygen containing atmosphere it results in the formation
of an oxide layer on its surface with an oxygen diffusion
zone beneath it [12]. This formation is more pronounced
when the alloy is heat treated above transformation tem-
perature and cooled rapidly in air or quenching by water.
This formation further plays an important role in devel-
oping a protective layer which promotes remarkable ad-
vantage of the alloy while working in a friction and wear
environment. The presence of protective layer is crucial
in improving corrosive and wear resistance. Further the
scanning electron micrograph of quenched specimen
given in Figure 5 show the presence of acicular marten-
sitic structure (ά) in white globular primary α and β ma-
trix. The presence of needle like acicular alpha greatly
improves the hardness values of the quenched specimen.
The microstructures have been consistent with respect to
the work done on this alloy by A. K. Jha et al. [13] and A.
Molinari et al. [6].
On the other hand the hardness of both the shotpeened
and shot blasted materials have increased drastically
during these surface treatment operations due to which
the wear resistance of the treated specimens had in-
creased twice to that of the untreated base specimen. But,
it does not in any way correlate to the wear resistance
obtained by the heat treated specimens. A very high
amount of wear resistance is only possible with the de-
velopment of a protective layer as can be seen in water
quenched and air cooled specimens. It is evident from the
graphical data presented in Figure 6 that the friction coef-
ficient values of the shotpeened specimens have reduced
Figure 5. Scanning electron micrograph of quenched speci-
en. m

B. K. C. GANESH ET AL.
739
(a)
(b)
(c)
Figure 6. Graphs showing values of coefficient of friction vs time at various conditions. (a) Heat treated at 950˚C; (b) Shot-
peened at 3.5 bar; (c) Shotpeened at 4.5 bar.

B. K. C. GANESH ET AL.
740
to a large extent limiting the value to around 0.39 when
compared to heat treated specimen with a value of 0.42.
This clearly indicates that the introduction of compres-
sive residual stress into the base specimens by virtue of
shot peening increased its hardness on the surface layer
as can be seen from Table 1. Shot blasting for a lower
duration had also decreased the friction coefficient to
0.37. This is due to the formation of a thin superficial
layer which had developed a greater amount of resistance
to indentation.
3.2. Weight Loss and Wear Rate Analysis
Maximum amount of weight loss has been reported from
as received material as shown in Figure 7, where as less
amount wear is reported from both water quenched and
air cooled specimens of Ti64 alloy. The results com-
mensurate with the hardness values and microstructure
behavior of all the heat treated specimens. Similarly
shorter duration of blasting and higher peening pressures
in peening had resulted in obtaining higher wear resis-
tance and surface roughness of the alloy while compared
to the base material. Increase in the surface roughness of
the alloy results in improved bio-adhesion and better
Osseo-integration between the bone and implanted mate-
rial [14].
Presence of high amount of wear is reported from the
wear testing of as received material Ti64 alloy as shown
in Figure 8. This corresponds to low hardness value
where the value of vickers hardness number is only VHN
311 as compared to quenched specimen which is having
a hardness value of VHN 380. High amount of hardness
values in quenched specimen is due to the presence of
acicular ά (martensitic structure) which has been formed
due to the heat treatment above beta transus temperature
followed by water quenching and aging. The wear rate as
shown in Table 1 is also high as in the case of as re-
ceived material when compared to air cooling or water
quenched specimens. The wear rate of furnace cooling
specimen is greater than the water quenched and air
Figure 7. Weight loss of Ti-6Al-4V under various surface
treated conditions.
cooled specimens due to the formation of lamellar α plate
like structure where there is no presence acicular α or
retained beta. This is due to slow cooling of the specimen
where complete transformation of body centered cubic
crystallographic (β) structure to hexagonally closely
packed (α) structure has taken place, which indicates
high strength due to its lamellar microstructure and also
due to high aspect (c/a) ratio of hexagonally packed
crystallographic structure.
An improvement in the wear resistance the alloy when
subjected to shot peening and blasting is reported in Fig-
ure 9. Shot peening is basically an operation intended to
improve fatigue strength of the implant alloy by the in-
duction of compressive residual stresses. In this experi-
mental work it had been proved that application of higher
amount of peening pressure had resulted in obtaining
higher amount of surface hardness as well as a higher
surface roughness value. This improvement in hardness
had played an important role in minimizing the wear rate
and improving the wear resistance of the implant alloy.
Similarly shot blasting for a shorter duration of time had
resulted in obtaining higher wear resistance of the alloy.
Figure 8. Wear behaviour of Ti-6Al-4V subjected to solu-
tion treatment and aging.
Figure 9. Wear behaviour of Ti-6Al-4V alloy subjected to
hot peening and shot blasting operations. s

B. K. C. GANESH ET AL. 741
Table 1. Wear properties of Ti-6Al-4V subjected to various surface treatments.
Type of Surface Treatment Surface Roughness (Ra) Micro Hardness (HV 0.5) Wear Rate × 10−11 m3/m
Ti-6Al-4V heat treated
As received (base metal) 0.014 311 1.954
Furnace cooled and aged 1.29 351 0.93
Air cooled and aged 0.858 340 0.186
Water quenched and aged 1.411 380 0.139
Ti-6Al-4V shotpeened
Shotpeened at 4.5 bar 1.634 380 0.837
Shotpeened at 3.5 bar 1.62 383 0.976
Ti-6Al-4V shotblasted
Shotblasted for 5 min 0.015 359 0.814
Shotblasted for 20 min 0.013 350 0.956
Blasting for a shorter duration had developed a superfi-
cial layer on the specimen which had inhibited the proc-
ess of wear. Continuation of blasting for longer duration
had resulted the merging of the superficial layer with
substrate layer due to which there is a lesser amount of
resistance to indentation and hence a comparatively
higher wear rate. Therefore it can be summarized that
though the surface treatment operations like shot peening
and blasting might not yield a higher wear resistance as
compared to that of the heat treated specimens, but they
have their own importance in enhancing some of the im-
portant properties like fatigue strength, Osseo-integration
and better wear resistance during its actual loading envi-
ronment.
The improvement of wear resistance of all the surface
treated specimens correlates the archard adhesive wear
theory which states that if all junctions have the same
diameter and if the real area of contact is given by the
normal load “L” divided by the hardness “H”, the total
volume of the material “V” removed during sliding
through a distance “S” is given by
LS
Vk
3H

where “k” is the wear constant which gives the probabil-
ity that a wear particle will be formed from an asperity
junction. It is understood from the above equation that
the hardness of the specimen is inversely proportional to
the weight loss or volume of the material removed.
Therefore Ti64 specimens show a proportional decrease
in their wear rate with a significant increase in their
hardness values.
3.3. Effect of Sliding Velocity
High amount of wear resistance is obtained when the
sliding velocity of the alloy pin is subjected to an in-
creased sliding velocity of 2 m/sec instead of 1 m/sec,
with all the remaining parameters kept constant. The
wear resistance almost improved in parallel to the
quenched specimen. This improvement is due to delami-
nation wear of the material when the sliding velocity is
increased. Type of wear at lower sliding velocity happens
to be oxidative wear as shown in Figure 10(a), where as
ploughing or severe plastic deformation has taken place
in the form of delamination as shown in Figure 10(b).
The delamination of the alloy has taken place due to
following reseaons:
During wear, dislocations are generated in the material
due to the plastic deformation of the surface by the slider.
If an oxide layer is present it is broken by the passage of
the slider thus exposing a fresh clean surface. This then
allows those dislocations nearly parlell to the surface to
be eliminated due to the action of stress at the free sur-
face. With continued sliding there will be pile up of dis-
locations at finite distance from the surface which will
lead to the formation of voids. When these voids coalese
crack formation takes place. when the crack reaches a
critical length the material between the crack and the
surface will shear producing a sheet like particle.
This eventual presence of delaminated regions play an
important role in minimising the loss of material due to
coldworking or plastic deformation which had occurred
when a soft material such as alloy pin material is incon-
tact with the hard disc material.
4. Conclusions
1) The wear rate of quenched specimen is very low
due to the presence of protective oxide coating layer
formed during heat treatment and also due to the pres-
ence of acicular martensitic structure (retained beta) in its

B. K. C. GANESH ET AL.
742
(a)
(b)
Figure 10. SEM images of wear tracks at various sliding
velocities. (a) Wear track at 1 m/sec; (b) Wear track at 2
m/sec.
Microstructure;
2) Finer wear tracks were observed without any pro-
tective oxide layer when the specimens were heat treated
below the transformation temperature followed by rapid
quenching and air cooling of the specimen;
3) Formation of protective oxide layer has taken place
in the specimens which were heat treated above the trans-
formation temperature followed by faster rate of cooling
the alloy;
4) Shot peening and shot blasting of the specimens had
resulted in obtaining higher surface roughness values
which were beneficial for improving properties like bio-
adhesion, Osseo-integration and better fatigue strength of
the implant alloys;
5) The improvement in the surface hardness of the
shotpeened and shot blasted specimens has resulted in
improving wear resistance of the specimen;
6) Oxidative wear has resulted in specimens tested at
lower sliding speeds, where as delaminative wear has
occurred at higher sliding velocities;
7) Delamination of the wear specimens has resulted in
drastic improvement of hardness up to 450 HV(0.5),which
has occurred due to resistance offered to plastic deforma-
tion by delaminative layers.
REFERENCES
[1] X. Y. Liu, P. K. Chu and C. X. Ding, “Surface Modifica-
tion of Titanium, Titanium Alloys and Related Materials
for Biomedical Applications,” Journal of Materials Sci-
ence and Engineering: R: Reports, Vol. 47, No. 3-4, 2004,
pp. 49-121.
[2] C. J. Boehlart, C. J. Cowen, J. P. Quast, T. Akahori and
M. Niinomi, “Fatigue and Wear Evaluation of Ti-Al-Nb
Alloys for Biomedical Applications,” Journal of Material
Science & Engineering: C, Vol. 28, No. 3, 2008, pp. 323-
330.
[3] U. I. Thomann and P. J. Uggowitzer, “Wear Corrosion
Behavior of Biocompatible Austenitic Stainless Steel,”
Journal of Wear, Vol. 239, No. 1, 2000, pp. 48-58.
[4] A. Zhecheva, W. Sha, S. Malinov and A. Long, “En-
hancing the Microstructure and Properties of Titanium
through Nitriding and Other Surface Engineering Meth-
ods,” Surface Coatings and Technology, Vol. 200, No. 7,
2005, pp. 2192-2207. doi:10.1016/j.surfcoat.2004.07.115
[5] M. Niinomi, “Mechanical Biocompatibilities of Titanium
Alloys for Biomedical Applications,” Journal of Me-
chanical Behavior of Biomedical Materials, Vol. 1, No. 1,
2008, pp. 30-42. doi:10.1016/j.jmbbm.2007.07.001
[6] Y. Q. Fu, N. L. Loh and A. W. Batchelor, “Improvement
in Fretting and Fatigue Resistance by Application of Sev-
eral Surface Treatments and Coatings,” Journal of Sur-
face and Coatings Technology, Vol. 106, No. 2-3, 1998,
pp. 193-197. doi:10.1016/S0257-8972(98)00528-3
[7] A. Molinari, G. Straffelini, B. Tesi and T. Bacci, “Dry
Sliding Wear Mechanisms of the Ti-6Al-4V Alloy,”
Journal of Wear, Vol. 208, No. 1-2, 1997, pp. 105-112.
doi:10.1016/S0043-1648(96)07454-6
[8] M. O. Alam and A. S. M. A. Haseeb, “Response of
Ti-6Al-4V and Ti-24Al-11Nb Alloys to Dry Sliding Wear
against Hardened Steel,” Journal of Tribology Interna-
tional, Vol. 35, No. 6, 2002, pp. 357-362.
[9] S. J. Li, R. Yang and S. Li, “Wear Characteristics of
Ti-Nb-Ta-Zr and Ti-6Al-4V Alloys for Biomedical Ap-
plications,” Journal of Wear, Vol. 257, No. 9-10, 2004,
pp. 869-876. doi:10.1016/j.wear.2004.04.001
[10] P. Majumdar, S. B. Singh and M. Chakraborty “Wear
Response of Heat Treated Ti-13-Zr-13Nb Alloy in Dry
Condition and Simulated Body Fluid,” Journal of Wear,
Vol. 264, No. 11-12, 2008, pp. 1015-1025.
doi:10.1016/j.wear.2007.08.005
[11] M. P. Gispert, A. P. Serro, R. Colaco, A. M. Botelho and
D. Rego, “Tribological Behavior of Ci-Implanted Tin
Coatings for Biomedical Applications,” Journal of Wear,
Vol. 262, No. 11-12, pp. 1337-1345.
[12] H. Guleryuz and H. Cimenglu, “Oxidation of Ti-6Al-4V

B. K. C. GANESH ET AL.
743
Alloy,” Journal of Alloys and Compounds, Vol. 472, No.
1-2, 2009, pp. 241-246.
[13] A. K. Jha, S. K. Singh and M. S. Kiranmayee, “Failure
Analysis of Titanium Alloy (Ti-6Al-4V) Fastener in
Aerospace Application,” Journal of Engineering Failure
Analysis, Vol. 17, No. 6, 2010, pp. 1457-1465.
doi:10.1016/j.engfailanal.2010.05.007
[14] Z. Mohammadi, A. A. Ziaei-Moayyed and A. S.-M. Mes-
gar, “Grit Blasting of Ti-6Al-4V Alloy: Optimization and
Its Effect on Adhesion Strength of Plasma-Sprayed Hy-
droxyapatite Coatings,” Journal of Materials Processing
Technology, Vol. 194, No. 1-3, 2007, pp. 15-23.
doi:10.1016/j.jmatprotec.2007.03.119