You are on page 1of 12

Tribology

ELSEVIER
SCIENCE?

0301-679X(

International Vol. 29. No. 4, pp. 345-356, 1996


Copyright @ 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0301-679X/96/$15.00
+O.OLl

95)00100-X

Optimum film thickness


of thin
metallic coatings on silicon
substrates
for low load stiding
applications
Dong-Seob

Jang*+

and Dae Eun Kim*

The frictional behaviour of thin metallic films on silicon substrates


sliding against 52100 steel balls is presented. The motivation of
this work is to identify an optimum film thickness that will result
in low friction under relatively low loads for various metallic films.
Dry sliding friction experiments on silicon substrates with soft
metallic coatings (silver, copper, tin and zinc) of various thickness
(I-2000 nm) were conducted using a reciprocating pin-on-flat type
apparatus under a controlled environment. A thermal vapour
deposition technique was used to produce pure and smooth
coatings. The morphology of the films was examined using an
atomic force microscope, a non-contact optical profilometer and a
scanning electron microscope. Following the sliding tests, the
sliding tracks were examined by various surface characterization
techniques and tools. The results indicate that the frictional
characteristics of silicon are improved by coating the surface with
a thin metallic film, and furthermore,
an optimum film thickness
can be identified for silver, copper and zinc coatings. In most
cases ploughing marks could be found on the film which suggests
that plastic deformation of the film is the dominant mode by
which frictional energy dissipation occurred. Based on this
observation, the frictional behaviour of thin metallic coatings under
low loads is discussed and friction coefficients are correlated with
an energy based friction model. Copyright @ 1996 Elsevier
Science Ltd
Keywords:
friction,
vapour deposition,

wear, thin metallic films,


optimum
film thickness

hard substrate,

Introduction
It has long been recognized that soft coatings improve
the tribological
properties of sliding systems under
certain sliding conditions. However, the fundamental
+Currently
at Samsung Electra-Mechanics
*Department
of Mechanical
Engineering,
206 West 18th Avenue,
Columb~,
OH
*Department
of Mechanical
Engineering,
Korea

Co., Suwon, Korea


The Ohio State University,
43210, USA
Yonsei University,
Seoul,

Tribology

thermal

mechanisms behind such behaviour have not been


clearly explained. One basic problem is that friction
interactions
are not clearly understood yet. Even
though much research has been conducted on the
subject of identifying the basic friction mechanism,
the conflicting theories of friction are continuously
and vigorously debated. The view shared by many
researchers and engineers is that friction depends
directly on adhesion phenomena. The adhesion theory
suggests that similar materials will result in high friction
International

Volume

29 Number

4 1996

345

Optimum

film thickness:

D.-S. Jang

and D.E. Kim

due to chemical compatibility.


On the other hand,
there are views which emphasize mechanical interactions at the sliding interface as being the dominant
cause of friction. In this school of thought, surface
roughness, wear particle geometry and the shapes of
asperities are emphasized. This controversy is partly
due to the system-dependent nature of friction and
wear phenomena.
Bowden and Tabor2 first developed a theory of friction
of thin solid films based on the adhesion theory of
friction. They concluded that the effectiveness of thin
solid films in reducing friction results from the low
shear strength of the film material and the high yield
strength of the hard substrate. Their proposition was
modified by incorporating the pressure dependence of
the shear strength of thin solid films3*4. A Hertzian
contact model for thin solid films, which has been
used to explain the decrease in friction coefficient with
increasing load, is derived from this postulatior?.
Even though the adhesion model is widely accepted
for explaining the frictional behaviour of thin soft
films, this model is inadequate for explaining
or
predicting most frictional phenomena, particularly in
certain conditions where the adhesion is not expected
to play a significant role as in unclean environments.
Many researchers have presented convincing evidence
that plastic deformation is commonly involved in most
sliding experiments, even for brittle materials such as
silicon and certain ceramics under very low loads7T8.
The explanation for such behaviour is that asperity
contacts at the sliding interface experience high local
stresses that exceed the flow stress of the material.
The consequence is that frictional energy dissipation
can be attributed to plastic deformation of the material
at and near the sliding interface. Based on this
observation, Heilmann and Rigney9,10 have developed
an energy-based friction model which depends on
certain mechanical parameters and microstructural
features of the materials, where the principal contribution to sliding friction is the near-surface plastic
deformation. In this work, this model has been applied
to model the frictional behaviour of coated systems.
Other studies indicate that elastic contact is possible
under certain conditions, namely, with smooth surfaces
and high hardness, low modulus material@.
Indeed,
a highly desirable sliding condition would be to restrain
all the interfacial interactions to elastic behaviour, so
that wear would be zero and frictional energy dissipation would be minimal.
However, despite many
researchers belief that elastic contact sliding is possible,
it is highly unlikely that such an ideal sliding condition
can be achievedl*. To achieve minimum
friction,
therefore, is to confine plastic deformation to a very
small volume near the contact region. In addition, the
material undergoing plastic deformation needs to be
weak or soft to allow minimal energy dissipation
during asperity interaction. A composite of a very thin
metallic film with low yield strength coated on a
substrate with high hardness may be exploited for this
purpose.
Film thickness has been recognized as a variable of
the frictional behaviour of thin solid films since Bowden
and Tabor* illustrated
the general nature of the
346

Tribology

international

Volume

29 Number

phenomenon. Their work also indicates that there is


an optimum film thickness where a minimum friction
coefficient can be obtained. Similar results have been
obtained with different combinations
of materials,
coating techniques and testing conditions13-15. On the
other hand, certain experimental results do not show
a clear minimum friction coefficient as a function of
film thicknesPT1.
Even though some information
is
available in the literature,
the effects of the film
thickness on the frictional behaviour of solid films
have not been clearly identified. In particular, very
little information is available on the frictional behaviour
of ultra-thin (l-100 nm) films under relatively low
loads.
The present study investigates the optimum
film
thickness to achieve low friction in soft coating systems
for low load sliding applications.
In addition, the
friction mechanism for thin soft coating systems under
low loads is suggested. Ultimately,
such knowledge
may be used for sliding applications in micro-electromechanical systems (MEMS) and other micro devices.
Experimental

procedures

Scope
In the present study dry friction sliding tests were
conducted with bare silicon and silicon substrate coated
with silver, copper, tin or zinc film. These specimens
were slid against 52100 steel balls using a reciprocating
pin-on-flat
type apparatus. The coating materials
were selected based on their relatively low strength
properties. They also have received much attention as
thin film solid lubricants. The sliding speed was kept
low in order to minimize
the effects of thermal
interactions at the sliding interface. In addition, the
environment was purged with nitrogen gas to minimize
chemical changes of the surfaces during the sliding
tests. All experiments were conducted at room temperature with a relative humidity of 30-40% inside
the chamber. A thermal evaporation technique, which
can produce highly pure films with smooth surfaces,
was used to coat the metallic films. After deposition
the morphology of the films was examined using an
atomic force microscope (AFM), a non-contact surface
profilometer
and a scanning electron microscope
(SEM). Following the sliding tests, the specimens
were examined using various surface characterization
tools such as the SEM, energy dispersive spectroscopy
(EDS), a non-contact surface profilometer,
and an
optical microscope.
Apparatus
The custom-built tribotester shown in Fig 1 was used
to conduct the experiments. This apparatus is capable
of providing either reciprocating
or rotary motion
between the two surfaces and monitoring the friction
coefficient in real time for normal loads ranging from
0.05 to 5 gf. The pin is attached to a spring loaded
arm. The back end of the arm is driven with a closedloop micro-actuator
and a motion controller which
controls the vertical motion of the pin with a resolution
of 0.1 pm. This corresponds to a normal load of 0.003
4 1996

Optimum

film thickness:

D.-S. Jang and D.E. Kim

Microactuator
Strain Ga&
(For normal

load)
(For friction
\

force)

Sample
Holder
Rotary
Stage

Iu

Linear Stage

Stage Controller

1 Amplifier

Fig I Schematic of the experimental apparatus

gf. Two sets of full bridge semiconductor strain gauges


(gain factor = 155) are attached to flexible cantilever
arms of appropriate stiffness for the normal load and
friction force measurements. The strain gauge outputs
are fed into an A/D converter board in the microcomputer as well as a chart recorder through two singlechannel signal conditioning
amplifiers. The friction
force and the variations of the applied normal load
are continuously monitored and recorded during the
experiments.
The tribotester is mounted inside a
Plexiglass chamber which allows for the control of the
environmental
conditions. The entire setup is placed
on a vibration isolation table to reduce the effects of
mechanical vibration.
Materials

and preparation

for the deposition. The fihn thichness was controlled


using a thickness monitor during the coating process
and was verifkd by observing the edge of the Hm on
the substrate with a non-contact surface -meter.

RM5:

PROFILE

26.4nm

2.5x
649.8nm

WVLEN:

of coatings

Single crystal (111) n-type silicon wafers were used as


substrates. The hardness was 850 HV (25 g, dwell
time: 10 s) or 8.33 GPa. Pure silver (99.999 w/o),
copper (99.999 w/o), tin (99.9999 w/o) and zinc
(99.9999 w/o) were coated on the silicon wafers
using a thermal evaporation technique. The material
properties of these metals are listed in Table 1. The
base pressure of the vacuum chamber was 0.5 mPa
Tribology

0.05

9.85

Illstance

1.65

on

2.44

surface,

3.24

4.04

mllllmeters

Fig 2 Film thickness measurement using a non-contact


surface profilometer for the 6.5nm thick silver coating
international

Volume

29 Number

4 1996

347

Optimum

film thickness:

Table 1 Material

D.-S. Jang and D.E. Kim

properties

of silver,

copper,

tin and zinc

Material

Youngs
moduIus23 (GPa)

Hardness=
(HV/MPa)

Yield strengthz0-23
IMPa)

Silver
Copper
Tin
Zinc

78
120
44
91

741725
105/1030
6.6164.7
321314

290
344
20.7
110

Measured

with

cold

rolled

metal

flats

using

the microhardness

The controlled film thickness values were 1, 10, 100


and 1000 nm and measured thickness values were 2,
7, 6.5 and 680 nm for silver, and 2, 7, 38 and 690 nm
for copper, respectively. A sample of the thickness
data obtained with the non-contact surface profilometer
for silver (65 nm) is shown in Fig 2. For tin, 13, 92,
and 1080 nm thick coatings were prepared. For zinc,
the controlled film thickness values were 10, 100, 1000
and 2000 nm. The morphology of zinc coatings was
highly discontinuous as shown in SEM micrographs
(Fig 3). For zinc coatings, the film thickness value
may be insufficient to describe the state of zinc deposits
due to the discontinuous morphology. Thus, the term
average film thickness is used to represent the amount
of zinc deposit on the silicon substrate. The state of
the zinc coatings can be clarified with SEM micrographs
showing the density and grain size of the zinc deposits.

Fig 3 SEM micrographs

348

Tribology

of zinc coatings

International

Volume

tester

(25 g, dwell

time:

10 s) after

0.92
1.10
0.57
0.79
polishing

Most zinc grains on the silicon surface have a hexagonal


shape, which is indicative of the hexagonal closepacked (HCP) structure of zinc crystals. Furthermore,
as the deposition amount increases, the size and
density of grains also increase. Despite the discontinuous nature of the zinc coating, it was decided to
investigate its frictional properties for comparison with
continuous films.
The surface roughness of the coated and bare silicon
specimens was measured with a non-contact surface
profilometer.
The obtained values were 4-5 nm rms
for bare silicon and 3-5 nm rms for silver, copper and
tin coated specimens. Surface profiles of the bare and
silver coated (7 nm) silicon specimens are shown in
Fig 4. The values of the surface roughness of zinc
coatings were 2-5 nm rms for 10 and 100 nm thick

(a) t = 10 nm; (b) t = 100 nm; (c) t = 1000 nm; (d) t = 2000 nm

29 Number

4 1996

Optimum

film thickness:

D.-S. Jang and D.E. Kim

used in the experiments. The sliding speed was 1 mm/s


and the stroke was set to 7 mm. The friction forces
and the variations in the applied normal load were
recorded for 100 cycles. The friction force was recorded
from the output of the strain gauges attached to the
vertical flexible cantilever arm. All friction coefficients
presented in this study represent the mean values of
friction coefficients of three or four tests.

rJ..tanc.

-14 ial!
Ll. 13.0nm

LMlCro~~

L.-14.anm

WYKC

Before and after the sliding test, the adhesive force


in the normal direction was measured by monitoring
the vertical force needed to separate the pin from the
slider after contacting the two with predetermined
normal forces. The adhesion coefficient, which is
defined as the ratio of the adhesive force to the
applied normal force, was calculated. For all specimens
tested, the mean values of the adhesion coefficient
were less than 0.01. Particularly,
for the coated
specimens, there was no detectable adhesive force
before and after the sliding tests.
Characterization

of the specimens

The wear tracks of the specimens and worn areas of


the pins were observed using a Nomarski
optical
microscope and an SEM. The SEM was equipped with
an EDS facility which can detect elements heavier
than boron. It was used to check for material transfer
between the pin and the slider. The surface topography
of the wear tracks of the coated specimens was
analysed using a non-contact surface profilometer.

Results and discussion


WYKC
Fig 4 Non-contact surface projilometer surface profiles
for (a) bare silicon, (b) silver coated (7 nm) silicon

coatings, and 1.50-200 nm rms for 1000 and 2000 nm


thick coatings.
Pure bulk metals were also tested for their frictional
behaviour. Flat metal specimens were ground with 600
grit Sic abrasive paper in water and then polished by
15, 6 and 1 pm diamond pastes with oil. Finally, the
flat specimens were polished using 0.05 pm ralumina
to obtain a smooth surface finish. The measured values
of the surface roughness of bulk metal flats were in
the range of 30-60 nm. After polishing with alumina
the specimens were cleaned with trichloroethylene
and
rinsed with methanol using an ultrasonic cleaner.
Sliding

and adhesion

tests

Dry sliding tests were conducted with the reciprocating


tribotester using bare and coated silicon as well as
bulk metal specimens as the flat, and 52100 balls as
the pin. The diameter of each steel ball was 3.2 mm
and its hardness was 890 HV (25 g, dwell time: 10 s).
The measured surface roughness value of the steel
ball was about 10 nm rms. The steel balls were cleaned
with trichloroethylene
and rinsed with methanol using
an ultrasonic cleaner. 0.1 and 1 gf normal loads were
Tribology

The experimental friction coefficient data with respect


to the film thickness are presented for the first and
100th cycle. As sliding proceeds, the film thickness on
the sliding track changes since the film material
undergoes permanent deformation.
Thus, the film
thickness value is valid only at the first cycle. Data
are also given for the 100th cycle to compare the
effectiveness of the deposition material over a given
sliding distance.
Silver

films

on silicon

substrates

The variation of friction coefficient with film thickness


for silver coated silicon is shown in Fig 5. (Error bars,
here and in later figures, represent the range of friction
coefficient about mean values of friction coefficients
of three or four sliding tests.) The left end of the
figure indicates the friction coefficient for bare silicon
and the right end shows the value for bulk silver. For
both 0.1 and 1 gf loads, initial friction coefficients for
bare silicon were relatively high. For the 0.1 gf load,
the friction coefficients for bare silicon increased from
0.41 to 0.66 in 100 cycles. On the other hand,
the friction coefficients for the coated substrates
maintained relatively low values during the tests. In
particular,
for a 7 nm thick coating, the friction
coefficients were reduced significantly. For 65 and
680 nm thick coatings, initial friction coefficients were
very close to those of the bulk silver. It seems that
the 65 and 680 nm thick coatings behaved like bulk
silver under the loads tested. As sliding proceeded, the
International

Volume

29 Number

4 1996

349

Optimum

film thickness:

D.-S. Jang

thick coating is not uniform, which is presumed to be


responsible for the high friction.

.
I

.
1
1 sto/de

0.0

and D.E. Kim

0.1 Qf

- 0.8
j

Microscopic examination of the surfaces showed many


grooves in the sliding tracks of the coated and bulk
flats (Fig 7). It seems clear that severe plastic
deformation of the coating took place during sliding
action. Some differences in wear track features with
the 7 nm thick coating and other thicker coatings or
bulk silver were observed. In contrast to the definite
grooves on the sliding tracks of the 65 and 680 nm
thick coatings, and bulk silver, the sliding tracks of
the 7 nm thick coating were relatively smooth. In
addition, the topography of the wear tracks (Fig 8)
showed that the depth of the wear track (~3 nm) of
the 7 nm thick coating was much smaller than those
of the 65 and 680 nm thick coatings, and bulk silver.
This indicates that the friction and wear can be
controlled by an appropriate choice of coating thickness. SEM micrographs and EDS analysis of the steel
ball indicated that no detectable silver was transferred
to the ball. This observation,
in addition to the
measured adhesion coefficients, suggest that adhesion
did not play a significant role in these tests.

0 IQf

0.6

B 0.4
0.5

e
6

.!I0.7
IA 0.3

0.2

0.1
0 I
I,,,,,,
bare silicon
1

,.,.,I

I,
,.I
looo
IO
100
Film thickness
(nm)

.I

I.,

1
0.9

lOlJthc/de

I
bulk

silver

O.lgf

o.af 0.7-

1Qf

H 0.6- 1
g 0.5.
; 0.4-f
0.3-.

Copper films

0.2-

The friction coefficients of silicon could be reduced


significantly with copper coatings. Figure 9 shows the
friction coefficient values with respect to film thickness
for copper coated silicon. The optimum film thickness
observed is also 7 nm for both normal loads. The
initial friction coefficient with a 2 nm thick coating
was low as with other coatings. The AFM examination
showed the 2 nm thick copper coated surface is
relatively smooth and uniform unlike the 2 nm thick
silver coating. This suggests that the film morphology
is one of the important factors for controlling friction
of thin coating systems.

O.lbati

silicon

i
Film thickness

100
(nm)

loo0

bulk

silver

Fig 5 Friction coeficient vs. film thickness for silicon


coated with silver (a) at first cycle (b) at 100th cycle

friction coefficients of all coated specimens gradually


increased under the 0.1 gf load while the values of
friction coefficients remained constant under the 1 gf
load. Higher friction coefficients observed for the
0.1 gf load are indicative of the load dependent nature
of the soft metallic coating which has been also
observed in a different load ranget3+18. For a 2 nm
thick coating, the initial friction coefficient was high
compared to those for other coatings. The AFM
micrograph (Fig 6) shows that the surface of the 2 nm

8
20.0

on silicon

Bulk copper resulted in lower friction coefficients than


most copper coatings. The low friction of bulk copper
is assumed to be due to relatively high hardness
(= 1.03 GPa) which resulted from work-hardening of
the surface layer during mechanical polishing19. The
hardness observed for cold rolled copper is about
780 MPa which is very close to that for cold rolled
silverzO. Following the friction test, the bulk copper
specimen was annealed at 220C for one hour in
flowing hydrogen and tested again. The hardness of
the annealed copper was 48 HV (25 g, dwell time:
10 s) or 470 MPa. There was no difference in the
surface roughness before and after annealing. However, it should be mentioned that annealing may cause
a change in the content of the oxide in the copper
which in turn may affect the frictional behaviour. The
friction coefficients of the annealed copper were almost
two times higher than those of the bulk copper.
Microscopic examination of the worn areas showed
characteristics similar to those of silver coatings.
Tin films on silicon

Fig 6 AFA4 micrograph


350

Tribology

of a 2 nm thick silver coating

International

Volume

29 Number

substrates

substrates

Figure 10 shows the variation of friction coefficient


with respect to film thickness for tin coated silicon.
For the 1 gf load, higher friction coefficients were
4 1996

Optimum

film thickness:

Fig 7 SEM micrograph and Nomarski


optical micrographs
of the wear tracks
sliding- for 100 cycles against 52100 steel at 1 -_
gf load: (a) t = 7 nm (SEM
t = 680 nm; (d) bulk silier

observed for all coated specimens compared to the


values obtained for bare silicon at both the first and
100th cycle. Nomarski microscopy showed that tin
coatings were easily worn out in a few sliding cycles.
It seems that the higher friction coefficients are caused
by tin wear particles generated during sliding. The
easy wearing of the tin coating is probably related to
the weak adhesion of a tin film to the substrate owing
to the low surface energy of tin (Table l), and the
morphology of the tin coating. Unlike the silver and
copper coatings, tin coatings consisted of small granules
(Fig 11). For the 0.1 gf load, the 1100 nm thick coating
shows lower friction coefficients than the values
observed for bare silicon, but higher values than those
of bulk tin. The minimum friction coefficient of the
tin coating was observed with a 92 nm thick coating
under a 0.1 gf load after 100 cycles. Though the initial
friction coefficient with the 92 nm thick coating was
relatively high, as sliding proceeded, the friction
coefficient decreased to the value of 0.28 and then
remained constant. Such low friction force behaviour
is attributed to the transferred film to the pin. The
SEM (Fig 12) and X-ray mapping analysis showed
that small agglomerates of tin grains formed on the
pin sliding against the 92 nm thick coating. The
conditions responsible for tin transfer were not identified. However, it is speculated that these conditions
involve film thickness, film morphology,
and applied
normal load. For the 1 gf load, the friction coefficients
Tribology

D.-S. Jang and D.E. Kim

on silicon coated with silver after


micrograph);
(b) t = 65 nm; (c)

for bulk tin rapidly increased to 0.9 after a few sliding


cycles. The SEM micrograph (Fig 13) showed many
cylindrical wear debris and rough grooves on the wear
track. It seems that severe ploughing or microcutting
by hard asperities of the steel ball took place. The
existence of an optimum film thickness is not obvious
for the tin coating under 0.1 and 1 gf loads. However,
for the 0.1 gf load, the 1100 nm thick coating seems
to be beneficial for short sliding cycles while the 92 nm
thick coating is better for longer sliding cycles.
Zinc films

on silicon

substrates

The dependence of the friction coefficient on average


film thickness for zinc coated silicon is shown in Fig
14. The minimum
friction coefficient was achieved
with the 10 nm thick coating despite the highly nonuniform morphology of the film (Fig 3). For the 1 gf
load, all the coated specimens maintained lower friction
coefficients than the values obtained for bare silicon
and bulk zinc throughout the test. On the other hand,
for the 0.1 gf load, as sliding proceeded, the friction
for the coated specimen increased rapidly. After sliding
for 100 cycles, most of the film was removed from the
substrate. For the 1 gf load, small agglomerates of
zinc grains formed on the pin (Fig 15) while such
agglomerates could not be found on the pin for the
0.1 gf load. It seems that the formation
of small
agglomerates on the pin is necessary to maintain low
International

Volume

29 Number

4 1996

351

Optimum

film thickness:

D.-S. Jang and D.E. Kim

Is&de

O.lgf

El 0
3 -,e2g-1q
-2.57.

lor

IVV

50
Distance

96
on

143
surface,

109

236

microns
0

1..1..,

bare

silicon

,,,,,,,

.1..,.,

10
Film thickness

.I,.,,(

IIn

100
(nm)

1000

bulk<

pper

lOO(hcyde

I
T

annealedcopper->
,

-11.4
3

50
Distance

96
on

143
surface,

109

236

microns

bare

silicon

lb
Film thickness

4.0

(0

lb0

1600

bulk c&per

(nm)

Fig 9 Friction coefjicient vs. film thickness for silicon


coated with copper (a) at first cycle (b) at 100th cycle
1.17

-4.0

o.o-

-6.0

93

40

138

184

0.6-

229
f

Distance

on

surface,

1 &c/de
I

0.43
0.3%

:
m
z

1!3f

microns

k
+
E

O.lfl

0.7-

B OB- 1.
c 0.5-

v)

11.4

I
m

m
I

0.2-

0.0
tin
-11.4

.
-22.8

51

98

145

192

I-

239

o.oDistance

on

surface,

1OOUlCjdS

microns

Fig 8 Non-contact surface projilometer surface topography of the wear tracks on silicon coated with silver
after sliding for 100 cycles against 52100 steel at 1 gf
load: (a) t = 7 nm; (b) t = 65 nm; (c) t = 680 nm;
(d) bulk silver
0.2-

friction in the case of zinc coated silicon. In addition,


it was noticeable that the size of grains agglomerated
on the pin increased compared to the grains on the
coated surface before the test (Fig 3) and most zinc
grains on the pin were distorted from the perfect
hexagon shape observed prior to the sliding test. This
352

Tribology

International

Volume

29 Number

O.l0
1.1.111,
bare silicon
1

..I,,.,,

,,,,.,.,
,,,.,.I
IO
100
moo
Film thickness
(nm)

,...bulk

tin

Fig 10 Friction coeficient vs. film thickness for silicon


coated with tin (a) at j%st cycle (b) at 100th cycle
4 1996

Optimum

Fig 11 SEM micrograph

of a 92 nm thick tin coating

indicated that the zinc grains were plastically deformed


during the sliding action. Definite ploughing marks
were also observed on the surface of the zinc agglomerate. Even though zinc coatings exhibit discontinuous
morphology, the frictional behaviour shows characteristics similar to those of continuous films such as silver.
It is believed that such behaviour is observed because
the discontinuous
zinc deposit gets smeared and
transferred to the pin as sliding proceeds, thus creating
the effect of a continuous film. As with bulk tin, bulk
zinc showed relatively high friction after a few sliding
cycles at the 1 gf load. The high friction is attributed
to severe plastic deformation
as shown in the SEM
micrograph (Fig 16).
General

discussion

Results of the present study show that the frictional


characteristics of silicon can be improved by coating
the surfaces with thin metallic films for low load sliding
applications. If the film thickness is optimized,
the
friction may be significantly reduced. For silver, copper
and zinc coatings, the optimum film thickness, where
the friction coefficient is minimum,
can be clearly
identified. The optimum film thickness for tin coatings
is not clear. Similar results have been reported for tin

film thickness:

D.-S. Jang and D.E. Kim

Fig 13 SEM micrograph of the wear track on bulk tin


after sliding for 100 cycles against 52100 steel at 1 gf
load

coatings on 52100 steel at low loadsih. The examination


of the morphology of silver and copper coatings shows
that as long as the surface of the film is smooth and
uniform, the optimum thickness can be achieved with
ultra thin coatings. On the other hand, despite
the highly discontinuous
and non-uniform
surface
morphology, zinc coatings exhibit low frictional behaviour probably due to the agglomeration
of zinc grains
transferred to the pin, For tin and zinc coatings, the
films were completely worn out after 100 cycles of
sliding. The easy wearing is assumed to be due to the
weak adhesion strength between the film and substrate
which may result from the low surface energy of film
material and the limitation
of thermal evaporation
coating technique. Compared with silver and copper,
the surface energies of tin and zinc are relatively low
(Table 1). Microscopic examination of worn specimens
and the low values of the adhesion coefficient suggest
that plastic deformation of the film due to ploughing
is the dominant mechanism of friction for thin soft
coatings under low loads. There was no indication of
any damage to the silicon substrate in these experiments
except for the case of bare silicon.

Fig 12 (a) SEM micrograph of the 52100 steel ball after sliding for 100 cycles against a 92 nm thick tin coating at
0.1 gf load; (b) higher magnification of the region indicated by an arrow in Fig 13(a)
Tribology

International

Volume 29 Number 4 1996

353

Optimum

film thickness:

D.-S. Jang and D.E. Kim

1.1

Average

film thidtness

.El

0.9

lOOfho/de

Fig 16 SEM micrograph


of the wear track on bulk
zinc after sliding for 100 cycles against 52100 steel at
1 gf load

O.lQf

0.6f 0.7-

(nm)

Isr

I,
I

Q 0.4H
E
0.3-.
!I'

0.2-

bare silicon

IO
Averege

100
film thickness

1000

bulk zinc

(nm)

Fig 14 Friction coeficient vs. film thickness for silicon


coated with zinc (a) at first cycle (b) at 100th cycle

Various mechanisms have been proposed to explain


frictional behaviour in dry sliding. These include
ploughing, asperity deformation,
and adhesion. It is
generally agreed that friction is due to the combination
of these mechanisms. This limits the effectiveness of
friction analysis in terms of a single model. Models
based on only one mechanism have not yielded
successful results because of the history-dependent
nature of frictional phenomena. Based on the observations in the present study, an energy-based friction

model developed by Heilmann and RigneylO may be


appropriate for correlating the friction coefficient of
thin soft coatings. The contributions of ploughing and
asperity deformation
on the friction force may be
predicted using the energy-based mode121*22. For a
thin soft coating on a hard substrate, as long as there
is no direct contact between the asperities and the
substrate, the minimum friction coefficient is expressed
aslo:

(1)
where pL is the friction coefficient of the bulk layer
material, r,,, , 7i,, are the ultimate shear stresses of
the substrate and bulk layer material, respectively, 7,
is the average surface shear stress of the layer and
F(x) is a monotonic function expressed as:
)Q)

= 1 _ 2 b(l+X)--Xl
ln(-X2)

(2)

Then, the experimental values of the friction coefficient


in the case where the coatings are effective can be

Fig 15 SEM micrographs of the 52100 steel balls after sliding for 100 cycles against: (a) a 10 nm thick zinc coating
at 1 gf load; (b) a 1000 nm thick zinc coating at 1 gf load
354

Tribology

International

Volume

29 Number

4 1996

Optimum

Table 2 Comparison

of experimental

Material
--.~.-~.

Normal

Silver
Copper
Tin
Zinc

___

load (gf)
1
0.1
1
0.1
1
0.1
1
0.1

and theoretical

film thickness:

values of the minimum

PL

p (experimental)

0.25
0.28
0.40
0.43
N/A
0.38
0.33
0.36

correlated with the theoretical value obtained from


the energy-based friction model. In the following
correlation, the ultimate shear stress, T,,,~~, is equated
to the material shear strength which is obtained from
the yield strength values in Table 1 and the minimum
friction coefficient is calculated using Eq. (1) with
et&,,
= 0.9999. Table 2 shows a comparison of the
experimental
and theoretical values of the minimum
friction coefficient.
For the correlation of the minimum friction coefficient
of the copper coating, the yield strength and the
friction coefficient of the annealed copper were used.
The yield strength observed for annealed copper is
about 209 MPaz3. For the tin coating, the friction
coefficient at the 100th cycle is used in the correlation
since the 92 nm thick tin coating is effective only after
some of the film material is transferred to the pin. As
shown in Table 2, the minimum friction coefficients
can be predicted using the energy-based friction model
for silver and tin coatings. On the other hand, the
theoretical values of the friction coefficient are higher
than the experimental
values for copper and zinc
coatings. In part, such a discrepancy may be caused
by the differences in the material properties and
microstructures
of the bulk and film material. In
addition, the relatively thin film can allow the silicon
substrate to influence the mechanical properties of the
film. Despite such discrepancy, the theoretical values
of the friction coefficient are quite reasonable compared
to the experimental values. Even though it is difficult
to obtain the exact prediction for the friction coefficient
using the energy-based friction model, the model may
be applied to predict the frictional behaviour with thin
soft coatings within reasonable limits.

0.21
0.23
0.22
0.20
N/A
0.29
0.18
0.19

D.-S. Jang

friction

and D.E. Kim

coefficient
p (theoretical)
0.18
0.20
0.29
0.31
N/A
0.27
0.24
0.26

thickness are about 7 nm for the silver and copper


coating, and about 10 nm for the zinc coating under
the 0.1 and 1 gf loads. For the tin coating, an optimum
film thickness could not be clearly identified. It was
interesting to note that despite discontinuous
film
morphology, zinc coating showed behaviour that was
comparable to those of continuous films. In particular,
the effectiveness of the zinc coating seems to be
determined by the accumulation
of transferred film
material on the pin.
Microscopic observations of the specimens indicate
that plastic deformation due to ploughing is the main
cause of friction with thin soft metallic coatings. The
energy-based friction model developed by Heilmann
and Rigney may be used to predict the friction
coefficient with thin soft metallic coatings for low load
contact sliding applications.
Acknowledgements
The authors would like to thank Professor D. Rigney
at the Ohio State University, Materials Science and
Engineering Department,
for his helpful comments.
We would also like to thank Professor B. Bhushan at
the Ohio State University,
Mechanical Engineering
Department, for use of the AFM and a non-contact
surface profilometer.
This study was partly supported
by the Center for Materials Research in the Ohio
State University.
References
1. Rabinowicz
York,

E. Friction and Wear of Materials,

Wiley,

New

I965

Bowden F.P. and Tabor D. The Friction and Lubrication of


Solids, Clarendon Press, Oxford, 1954
3. Bowers R.C. and Ziiman W.A. Pressure effects on the friction
coefficient of thin-film solid lubricants. J. Appl. Phys. 1968,
2.

Conclusions
In the present study, the frictional behaviour and
optimum film thickness of the thin soft metallic coating
for low load sliding applications have been investigated.
The results of the experimental work show that the
frictional behaviour of the thin soft coating under
relatively low loads depends on the film thickness,
normal load and coating material. Furthermore,
the
morphology of the film as well as the tendency of the
film material to transfer to the pin also affected the
frictional behaviour of the thin soft coating. It has
also shown that the values of the optimum
film
Tribology

39, 5385-5395

Briscoe B.J., Scmton B. and Willis F.R. The shear strength of


thin lubricant films. Proc. Roy. Sot. Lond. 1973, A333, 99-114
5. Shafei T.S.E., Amell R.D. and Hal&g J. An experimental
study of the Hertzian contact of surfaces covered by soft metal
films. Trans. ASLE, 1983, 26, 481-486
6. Singer LL., Bolster R.N., Wegand J., Fayeulle S. and Stupp
B.C. Hertzian stress contribution to low friction behavior of
thin MoS, coatings. Appl. Phys. Lett. 1990, 57, 995-997
I. Rigney D.A. Sliding Wear of Metals. Ann. Rev. Mater.
Sci.
1988, 18, 141-163

4.

International

Volume 29 Number 4 1996

355

Optimum

film thickness:

D.-S. Jang

and D.E. Kim

8. Kim D.E. and Sub N.P. On microscopic mechanisms of friction


and wear. Wear, 1991, 149, 199-208
9. Rigney D.A. and Hitih J.P. Plastic deformation and sliding
friction of metals. Wear, 1979, 53, 345-370
10. Heilmann P. and Rigney D.A. An energy-based model of friction
and its application to coated systems. Wear 1981, 72, 195-217
11. Greenwood J.A. and Williamson J.B.P. Contact of nominally
flat surfaces. Proc. Roy. Sot. Lond. 1966, A295, 300-319
12. Kim D.E. PhD dissertation. M.I. T., 1991
13. Sherbiney M.A. and Hailing J. Friction and wear of ion-plated
soft metallic films. Wear, 1977, 45, 211-220
14. Spalvins T. and Buzek B. Friction and morphological characteristics of ion-plated soft metallic films. Thin Solid Films, 1981,
84, 267-272
15. Shimura Y., Ito T., Taga Y. and Nakajima K. Friction properties
of sputtered tin films. Wear 1978, 49, 179-193
16. Shih C.Y. and Rigney D.A. Sliding friction and wear of tin-,
indium- and lead-coated 52100 steel. Wear 1989, 134, 165-184

356

Tribology

International

17. Takagi R. and Liu T. The lubrication of steel by electroplated


gold. Trans. ASLE, 1967, 10, 115-123
18. Tsuya Y. and Takagi R. Lubricating properties of lead films on
copper. Wear 1964, I, 131-143
19. Samuels L.E. Metallographic Polishing by Mechanical Methods.
Sir Isaac Pitman & Sons Ltd, London, 1982
20. Tabor D. Hardness of Metals, Clarendon Press, Oxford, 1951
21. Zhang J., Moslehy F.A. and Rice S.L. A model for friction in
quasi-steady-state sliding, Part I: derivation. Wear, 1991, 149,
1-12
22. Zhang J., Moslehy F.A. and Rice S.L. A model for friction
in quasi-steady-state sliding, Part II: numerical results and
discussion. Wear 1991, 149, 13-25
23. ASM Handbook, Metals Handbook, Vol. 2, Properties and
Selection: Nonferrous Alloys and Pure Metals, 9th Edition,
ASM International, Metals Park, Ohio, 1979, 707-803

Volume 29 Number 4 1996

You might also like