You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/298418583

On the Laws and Theories of Sliding Friction

Conference Paper · November 2015


DOI: 10.1115/IMECE2015-51470

CITATIONS READS
0 817

1 author:

Nannaji Saka
Massachusetts Institute of Technology
113 PUBLICATIONS   1,973 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Tribology: Friction, wear,and lubrication View project

All content following this page was uploaded by Nannaji Saka on 11 November 2018.

The user has requested enhancement of the downloaded file.


International Mechanical Engineering Conference & Exposition
IMECE 2015
November 13-19, Houston, Texas, USA

51470
On the Laws and Theories of Sliding Friction
Nannaji Saka
Department of Mechanical Engineering
Massachusetts Institute of Technology
Cambridge, MA 02139, USA
Email: nsaka@mit.edu

ABSTRACT
The many seminal contributions made by Professor Nam P.
Suh to the theories of wear, such as the delamination wear and
the solution wear, are well known. The contributions made by
him and his associates to the theories of friction, however, are
less known; but they are equally significant. In this article, I
first briefly survey, to provide an historical context, the laws
and theories of sliding friction as proposed over the past
centuries and decades. Then the contributions of Prof. Suh and
his associates in recent decades are reviewed. Specifically, the
role of wear particles in the frictional phenomena of dry and
boundary-lubricated sliding is examined. A novel concept of
undulating, or patterned, surfaces has been advanced to
minimize friction in both dry and boundary-lubricated sliding. Fig. 1 Types of friction. Friction of dry and boundary-
The undulating surfaces trap wear debris and thus minimize lubricated surfaces may be classified as Coulomb
plowing friction in dry sliding, above the transition temperature friction.
in boundary-lubricated sliding, and even in hydrodynamic
bearings during start/stop operations. The concept is especially there is no viscous medium between the sliding surfaces or the
appropriate for heavily loaded tribological systems with tighter sliding surfaces themselves do not deform viscously. Because
clearances in which the likelihood of seizure is imminent. boundary-lubricant films behave more or less as elastic-plastic
Keywords: adhesion, plowing, seizure, undulating surfaces. solids, however, friction phenomenon in these systems is akin
to that in dry sliding and hence it will also be included.
1 INTRODUCTION The primary object in reviewing the friction theories is to
examine the essential characteristics, similarities, and
In this presentation, first the laws of friction proposed
differences among the various friction mechanisms or models,
several centuries ago are introduced and then the theories of dry
and to examine to what extent a mechanism follows the laws of
sliding friction of the past several decades are discussed. The
friction. The theories of friction considered address only the
friction laws are idealized generalizations of empirical
micro-scale phenomena. The nano-scale aspects of friction and
observations by engineers; friction theories are in-depth
related tribological phenomena are addressed in other
analyses of physical phenomena by scientists. The demarcation
contributions in the conference. Furthermore, it is not the object
between engineers and scientists, however, is less sharp. As
of this presentation to conduct critical literature review. Rather,
shown in Fig. 1, there are basically two types of friction: dry
the intent is to review mostly the work of Suh and his co-
and viscous. In both cases friction depends on whether the
workers at MIT and elsewhere. Works by several other authors
mating bodies slide past or roll over each other. Friction in
are cited to provide the necessary contextual background. No
hydrodynamic and elasto-hydrodynamic systems is due to
effort will be made to critique or evaluate their work. The
viscous shear of the lubricant at the interface of the mating pair.
historical and general aspects are from the first five references
These systems violate almost all the laws of friction. Even in
listed [1-5] and the experimental data are from the various
dry rolling, if the bodies in contact deform by viscous flow, or
sources cited after.
internal friction, at least the third law is violated. Therefore we
are concerned here only with dry sliding friction, that is when

1 Copyright © 2015 by ASME


2 THE LAWS OF FRICTION 2.1 On the Validity of the Laws of Friction. It may be
The first two laws of friction were enunciated by Leonardo noted immediately, however, that the laws of friction are not as
da Vinci in the middle of the 15th century [1,2], two hundred fundamental and universal as the laws of motion, the law of
years before Sir Isaac Newton laid down the laws of motion gravity, the laws of thermodynamics, electromagnetism, and so
and provided a clear concept of force. The first law of friction on. Of course, even the laws of physics have been amended in
states that the frictional force is proportional to the applied the past century as new phenomena of nature came to light. The
normal load. Based on Leonardo’s observations the second law laws of friction are primarily based on the experimental
can be stated as: The friction force is independent of the findings of practical engineers centuries ago. Naturally, they
nominal contact area. It may be noted, however, that Leonardo are expected to be valid only in a small range of experimental
himself did not investigate the causes of friction. or service conditions.
Two centuries later, Guillaume Amontons rediscovered the That the friction force is proportional to the normal force is
laws of friction, now called Amontons’ laws. As we shall see almost a necessity. Otherwise it would not be possible to define
later, these laws are good approximations, but there are serious a friction coefficient. If the two forces were not proportional,
exceptions, too. He, however, took a giant step forward, for in their ratio would be different for different normal forces and
addition to the laws of friction he also proposed several hence would be just a parameter. In such a case it would be
mechanisms of friction. He postulated that friction arises better to call the ratio traction coefficient, as is done in
because of surface roughness, thus the “roughness theory” of hydrodynamic lubrication. Nevertheless, it has been observed
friction was born. Moreover, he further speculated that friction for a fairly large range of loads that the frictional force is
arises not only because of the need to raise one rough surface indeed proportional to the normal force. As will be seen
over another but also by the bending of roughnesses (which are shortly, it is in fact a consequence of the second law.
now called asperities), or their breaking. Thus in effect he According to the second law, friction force is independent
proposed several mechanisms at once and hence several of the nominal contact area. We now know that the real contact
theories, if qualitatively, of friction: roughness, deformation, area depends on and is proportional to the applied normal load,
and fracture. Like Leonardo he did not model these and that the tangential force is proportional to the real contact
mechanisms. area. Consequently the friction force is expected to be
Thirty years after Amontons’ contribution, John Theophilus proportional to the normal force, which is the first law. It may
Desaguliers proposed a very different view of the mechanism be noted that the nominal area does not appear in these
of friction. He attributed the observed high friction of smooth considerations, hence the second law. However, if the real area
surfaces to adhesion. Thus was born the “adhesion theory” of of contact approaches the nominal area, the law is expected to
friction, but he could not explain Amontons’ laws by the be invalid. In such an event, junction growth, seizure, etc. are
adhesion mechanism. expected to make the second law invalid.
Perhaps the first quantitative theory of friction was that due The third law of friction essentially defines dry sliding
to Charles Augustin Coulomb. His theory addresses the friction conditions and friction. If the friction force is a strong function
of rough surfaces by the mechanics of rigid bodies to derive a of the relative velocity, then viscous deformation would be
simple equation for the friction coefficient, µ, in terms of the responsible for friction. The deformation behavior of most
asperity slope as: µ = tan θ . Unfortunately, the mechanics of materials, especially metals, at moderate temperatures,
rigid bodies is alone is not of much help in calculating friction however, is strain-rate insensitive and hence the third law will
force, for rigid body contact and motion are not dissipative. He be valid. Such friction is aptly named Coulomb friction.
could not have analyzed the problem, however, because at his The validity of the laws can be qualitatively explained by
time the mechanics of deformable (plastic) bodies was not the asperity-contact schematics shown in Fig. 2. The general
sufficiently developed. Today, of course, this deficiency is requirements for the validity of the first two laws of friction are
easily corrected by allowing the asperities to deform plastically. that the real area of contact should be much smaller than the
He is, however, credited with the third law of friction: The nominal contact area, and randomly distributed, which
friction force is independent of the relative velocity of the generally is the case. For example, if the ratio of the real to
contacting bodies. nominal contact areas, Ar/An, is 0.01, the ratio of the asperity
Before the theories of friction are presented, perhaps it is size and the asperity spacing, a/λ, is about 0.1. Then the
well to recapitulate the laws of friction and examine their asperity contacts are sparsely distributed and thus they behave
validity. The Amontons-Coulomb laws of friction may be as independently deforming regions. As a result, both the first
summarized as: and second laws are valid. So these laws are not quite
independent, because one law is related to the other.
Law 1: Friction force is proportional to the applied normal As the normal load is increased, the real area approaches
load. the nominal area and the asperity contacts coalesce, as shown in
Law 2: Friction force is independent of the nominal contact the schematics, into larger areas and thus the asperity
area. deformation may not be quite independent. Then both the first
Law 3: Friction force is independent of the relative sliding and second laws of friction may be violated. Moreover, at high
velocity. loads and high velocities of sliding, the interface temperatures

2 Copyright © 2015 by ASME


will be high and new mechanisms of friction may come into 3 THE THEORIES OF FRICTION
play. For example, at high temperatures the surfaces may soften As has been noted earlier, the first quantitate theory of
and flow viscously, oxidize, or even melt. When a liquid layer friction was due to Coulomb. It will be shown shortly, however,
is present, the laws of dry friction are no longer valid. In nano- that it is not a theory of friction at all. Realistic theories of
scale systems where the size of such systems is the size of the friction have been advanced only in the 20th century, starting
asperity, the real area and apparent areas of contacts are about with the pioneering work of Tomlinson [3,6] and marching
the same. In such systems, accordingly, the laws of sliding toward the present-day theories of nano-scale friction. It must
friction may not be valid at all. be emphasized, however, although the classical theories are
based on different physical aspects, the problem is basically
one: that of plastic deformation (or fracture) of contacting
asperities during sliding. It is a boundary value problem in
continuum mechanics of solids with a multitude of attributes:
geometry (asperity height and slope), interface boundary
conditions (strength of adhesion), surface traction/displacement
boundary conditions, “constitutive” equations (yield strength,
hardness, fracture energy, etc.) and so on. The task then is to
solve the field equations and determine the friction force and
0.01 0.02 the friction coefficient for a given normal force (or
displacements). Because several variables enter the analysis, it
is perhaps better to designate the friction coefficient as a system
property than as surface or material property. This viewpoint
emerges as the various theories are examined in detail later.
Moreover, the calculated friction coefficients are generally
for the onset of motion, and are usually small compared with
experimental observations. In continued sliding, however, wear
particles are copiously generated and they increase friction
greatly. Long plowing grooves are formed on the sliding
0.05 0.1 surfaces both in dry and boundary lubricated surfaces. This
mechanism of friction, dubbed plowing, is the most dominant,
at least in ductile materials. Surprisingly, this mechanism is
quite prevalent even in boundary lubricated sliding. Suh and his
associates have studied the plowing phenomenon extensively
and have presented innovative solutions for minimizing sliding
friction. Specifically, undulating surfaces have been developed
to trap the wear debris and thus minimize friction and wear.
Their investigations are chronicled later.

3.1 Roughness Theory. It has been long known that rough


0.2 0.5
surfaces generally give high friction, though there are serious
exceptions. Thus, Coulomb had examined the friction of rough
surfaces by the mechanics of rigid bodies, of which he was an
expert, to derive a simple equation for the friction coefficient,
µ, as shown in Fig. 3. As the top body is loaded against the
bottom body with a force N, to initiate and sustain motion of
the top body a horizontal force F is required. By resolving the
forces appropriately, the friction coefficient µ can be expressed
in terms of the asperity slope, θ, as: µ = tan θ. It is surprising,
and exhilarating, to note that the friction coefficient could be
0.8 1.0 expressed by a simple trigonometric function. Upon closer
examination, however, the assumptions hidden will be
Fig. 2 Schematics of asperity contacts of random apparent. Generally, two criticisms are leveled against the
rough surfaces. Each schematic comprises 50x50 theory. First, because the average slope of the asperities is zero,
pixels and each black pixel represents an asperity- the friction coefficient should be zero, or at least extremely
asperity contact. The numbers represent the ratio of small. Second, because the asperity slopes are about 10-20
real-to-nominal contact areas, Ar/An. degrees, the friction should be between 0.18 and 0.36, whereas

3 Copyright © 2015 by ASME


3.2 Adhesion Theory. This theory was proposed in the
middle of the 20th century by Ernest and Merchant [7] and by
Bowden, Moore and Tabor [8]. The original concept of
adhesion contributing to friction was due to Desaguliers, as
noted earlier. According to this theory, when asperities come
together they form strong adhesive junctions, Fig. 4. The
normal load is supported by the weaker asperity by plastic flow.
When one asperity is translated relative to the other, the weaker
asperity is sheared.
(a) The normal load is the product of the asperity contact area
and the hardness of the material, and the tangential force is the
product of the same asperity area and the shear strength. The
ratio of the tangential and normal forces then is the friction
coefficient. Because the shear strength is a sixth of the
hardness, the friction coefficient should be about 1/6 or 0.16.
Much has been written on the pros and cons of the theory [2,3]
and so we need not repeat the arguments here. The strong point
of this theory is that both normal loading and tangential motion
result in plastic deformation and hence dissipation, certainly an

(b)

Fig. 3 (a) Coulomb’s sketches of multiple asperities


in contact [1,2] and (b) resolution of the forces acting
at the interface of a two-asperity contact.

the experimental values are much higher. To these two, a third


may be added: the asperity height does not appear but the slope
does in the friction equation. Thus both rough and smooth
surfaces with the same asperity slope should have the same (a)
friction coefficient, contrary to the spirit of the theory.
A fundamental objection to this theory, however, should be
that the two bodies are implicitly assumed to be rigid and hence
there is no dissipation (i.e., conversion of mechanical energy to
thermal energy). Thus the mechanics of rigid bodies will not
be of much help in calculating friction because rigid bodies do
not dissipate. In fact, the theory is all about the resolution of
forces acting at a point or on a surface, rather than an analysis
of friction processes –– such as plastic flow and fracture. To be
fair, however, Coulomb could not have analyzed the problem
because at his time the mechanics of plastically deformable
bodies was not sufficiently developed. Today, this deficiency
can easily be overcome by analyzing plastically deforming
asperities. It will be seen later how this is done by the slip-line
field analysis. (b)
Fig. 4 Friction due to adhesion of mating asperities:
(a) single adhesive junction and (b) multiple adhesive
junctions.

4 Copyright © 2015 by ASME


improvement beyond Coulomb’s theory. The weakest point,
however, is that even with the strongest adhesion at the
interface the friction coefficient is as low as 0.16. It has been
observed that under vacuum conditions the friction coefficients
of metals is two orders magnitude greater. Apart from this
discrepancy, the weak point of the analysis is that plastic
deformation due to the normal load and that due to shearing are
made additive. While in elasticity problems superposition is
valid for linear elastic solids, in plasticity problems, however,
the deformation fields, stresses, etc. cannot be superposed. Put
it alternatively, the shear force is calculated by a lower-bound
(a)
analysis and the normal force by an upper-bound analysis. The
plasticity analysis of the problem cannot be decomposed into
two problems, as can be done in elastic analysis. This seems to
be the primary reason why the calculated friction coefficient is
so low even with perfect adhesion.

3.3 Deformation Theory. By deformation theory we


mean plastic deformation theory. As has been noted earlier,
plastic or viscous deformation in at least one of the mating
asperities is a basic requirement for a viable theory of friction.
Elastic deformation alone will not be sufficient.
After the adhesion theory of friction was proposed, it was
immediately recognized that the analysis was flawed because
the stresses due to normal loading and tangential displacement
are calculated independently. In the deformation theory of
friction the asperity deformation is treated as a unified problem (b)
by Green [9,10], Challen and Oxley [11], and Suh and Sin [12].
In fact, the problem is properly posed for the first time with
well defined geometry, field equations for stresses, boundary
conditions, constitutive equations for material behavior, and so
on. Unfortunately, however, because the algebraic expressions
are so lengthy that it is easy to miss the central point.
In Fig. 5 are shown two slip-line fields [11,12]. The slip
lines are planes, or their traces, on which the shear deformation
takes place. The blocks themselves are rigid, though their
boundaries are deformable. Motion of the rigid blocks is
possible by shear along the boundaries, but the slip-lines should
be so drawn that deformation is compatible. Otherwise wedge-
shaped openings may develop and material continuity is lost.
Additionally, the problem is two-dimensional. To solve the
slip-line field problems, first a kinematically admissible slip-
line field is constructed. Then the stresses in the deforming
region are calculated, followed by the forces at the interface in
the normal and tangential directions. The ratio of tangential
force to normal force is the friction coefficient. It may be
noted, however, that the displacements of the blocks are
virtually zero but the relative velocities are not.
While the slip-line field analysis is rigorous, the solution is (c)
not unique. A number of slip-line fields, all satisfying the basic
requirements, can be drawn and the friction coefficients Fig. 5 Slip-line field analysis of friction: (a) adhesive
calculated. Different fields give different coefficients. In a junction, (b) non-adhesive junction and (c) calculated
sense friction coefficient is a property of the filed because the friction coefficients based on the slip-line field shown
material properties such as the hardness and shear strength in (b) [11,12].
cancel out. Since the asperity slopes are about 10-20 degrees,
the calculated friction coefficients are 0.5-1.0, Fig. 5(c).

5 Copyright © 2015 by ASME


3.4 Plowing Theory. Plowing is negative rake angle
cutting, as in grinding, abrasion, and polishing. As a material
removal process plowing is inefficient, but in frictional
phenomena it is very significant. The friction coefficient can
be extremely large, depending on the shape and size of the
particles as demonstrated by Sin, Saka and Suh [13], Suh,
Tohkai and Saka [14], and Suh, Sin and Saka [15].
Fig. 6 shows three differently shaped hard particles
plowing a softer surface. As the particle plows, the plowed
medium gets permanently displaced and thus plowing is a
dissipative process. The particle advances leaving a groove on
its wake. Thus the presence of long plowing grooves on a
sliding surface is an indication that plowing is taking place, and
that it is even the dominant mechanism of friction.
(a)
For simplicity, the normal load and frictional forces are
estimated here as in the adhesion theory of friction. The normal
load is the product of the projected contact are in the horizontal
plane and the hardness of the material being plowed. Similarly,
the frictional force is the product of the projected contact area
in the plane normal to the sliding direction and the hardness.
The friction coefficient is the ratio of the frictional force and
the normal load. Additionally, if the “interfacial friction” is
strong it too should be included in the calculation of the normal
and tangential forces. It may be noted that the problem of
calculating the normal and tangential forces is divided up into
two upper-bound problems.
Plowing by a wedge is a 2D problem, but plowing by cone
and sphere are 3D problems. Thus the wedge problem can be
solved more rigorously by the slip-line fields, but the cone and
sphere problems cannot be, and hence the simplifications. The (b)
wedge and cone give similar expressions for the friction
coefficients. The friction coefficient µ = tan θ, where θ is the
angle made by the interface with the horizontal. Surprisingly,
perhaps not, the expression is the same as that of Coulomb’s
theory. As θ increases, the friction coefficient increases rapidly.
The spherical particle, on the other hand, gives a small
coefficient for small depths of penetration, but it greatly
increases as the indentation depth is increased. Thus the
frictional force and the normal load are not proportional to each
other. For the simpler analysis the expression for the friction
coefficients of wedge, cone, and sphere are shown in Fig.6. For
a more accurate estimate of the friction coefficient, the finite
element analysis should be used.

3.5 Friction Space Diagram. Now we are in a position to


collect all the concepts together to construct what is called the
friction space diagram [12]. The diagram represents the
coefficient of friction as a function of adhesion, plowing, and (c)
roughness parameters, as shown in Fig. 7.
The adhesion component is expressed in terms of the Fig. 6 Plowing by hard particles: (a) wedge-shaped
dimensionless shear strength, f, the ratio of the shear strength of hard particle, (b) conical particle, and (c) spherical
the interface and the shear strength of the bulk asperity particle. The wedge-shaped and conical particles give
material. The plowing intensity is represented by the friction force proportional to the normal load. The
dimensionless variable, w/2r, where w is the width of the relationship is non-linear in the case of spherical
plowing groove and r is the radius of a spherical wear particle. particles.
Roughness is represented by the slope of the asperity, θ.

6 Copyright © 2015 by ASME


3.6 Friction in Boundary-Lubricated Sliding. A friction
space diagram may also be constructed for boundary-lubricated
sliding because friction coefficients are essentially independent
of the sliding speed. The mechanisms and intensity of friction,
however, are somewhat different. It was proposed by Hardy in
the early part of the past century that lubricant molecules
adsorb on solid surfaces and form thin tenacious films, Hardy
and Hardy [16] and Hardy and Doubleday [17]. The
deformation behavior of the films is solid-like. Friction
coefficient then is the ratio of the shear strength the film and
the hardness of the substrate on which the film has formed.
Because the film shear strength is extremely small compared
with the hardness of the substrate, the friction coefficients can
be extremely small.
The above model assumes, however, that the coverage of
the sliding surfaces by the lubricant is complete. It has been
recognized later that the lubricant coverage on the surface may
not be complete and that there are regions or patches on the
surface in which the lubricant is not present. In that case
friction force will be large, partly due to lubricant shear and
partly due to adhesion of the asperity junctions, Jahanmir and
Beltzer [18,19].
Fig. 7 Friction space diagram for dry sliding showing Recent experimental and theoretical investigations by
the relative contributions of adhesion, deformation Komvopoulos, Saka and Suh [20-23] have shown, however,
and plowing components of friction [4,12]. that plowing is also present, or even prevalent, in boundary-
lubricated sliding. The dimensions of the plowing grooves,
In the diagram, the x-axis represents f, the y-axis w/2r, the depth and width, however are much smaller than in the case of
z-axis the friction coefficient, µ; the slope of the asperities, θ, is dry sliding. This is so for two reasons. Fist, because of the
a parameter. The interfacial conditions in a sliding contact are adsorption of lubricant molecules adhesion is expected to be
ever changing and so the friction coefficient too will be low and hence the size of the particles will be small. Second,
changing. Unless the values of f, w/2r, and θ are precisely and more importantly, because of the presence of the lubricant
known, it would be difficult to know what the friction agglomeration of wear particles is mitigated and hence the
coefficient, µ, is. Nevertheless, the trends can be easily depth and width of plowing grooves will be small.
established. For example, even if adhesion is very good, i.e., f Consequently the friction coefficient will be low, but the
= 1, the friction coefficient is small, about 0.2. Similarly, the primary mechanism of friction is still plowing. Indeed, it has
effect of asperity slope θ is also small, especially because the been shown that the wear of boundary-lubricated surfaces is
range is only 10 to 20 degrees. By contrast, the effect of w/2r primarily due to the plowing mechanism. An important issue in
is substantial, and thus the plowing coefficient of friction is the boundary lubrication is desorption of lubricants at elevated
most dominant, and its value can exceed 2.0. temperatures. Above a certain temperature, the so-called
It may be noted, however, that the friction coefficients transition temperature, lubricants desorb from the surface and
calculated using the above models, except the plowing models, friction is very high; hence plowing could be the dominant
are for the onset of motion, and are usually small because the mechanism of friction at elevated temperatures.
surfaces are fairly smooth and there are no wear particles. In
continuous sliding, however, wear particles are copiously 4 UNDULATING SURFACES
generated, the topography is vastly different from the initial A characteristic feature of dry sliding friction is that at the
topography, and the friction force is much higher. It has been onset of sliding motion the friction coefficient is low, about 0.2,
observed that long grooves are invariably formed on the sliding but as sliding continues, the friction coefficient gradually
surface, a telltale sign of plowing. Thus plowing mechanism of reaches a steady-state value, as high as 1.0 or even higher. The
friction is the most dominant, at least in ductile materials. steady-state value depends on the material pair, load, speed,
Surprisingly, plowing grooves have been observed even in environment, and so on. It has been hypothesized by the MIT
boundary-lubricated sliding. In fact, plowing adequately group that the increase in friction is primarily due to the
accounts for both friction and wear in boundary-lubricated generation and agglomeration of wear particles, the so-called
sliding. These phenomena have been extensively investigated at third bodies. It is possible that the surfaces get cleaned and
MIT under the guidance of Prof. Suh, and innovative surface adhesion increases, but it is an indirect effect in that increased
topologies have been proposed to minimize friction. adhesion leads to more wear, wear particle agglomeration, and

7 Copyright © 2015 by ASME


thus higher friction. Therefore, friction can be minimized if
agglomeration and entrapment of wear particles at the interface
could be mitigated.
One way of removing the generated wear particles is by
continuously cleaning the interface with a jet of air or inert gas
(such as argon). Unfortunately, such a cleaning procedure is
impractical because the cleaning system would be cumbersome
and the sliding surfaces may not even be accessible. A far more
efficient and elegant means of removing the particles is by
trapping them, immediately after they are created, in pockets on
the surfaces as shown in Fig. 8. Such surfaces are termed
undulating (structured, textured, patterned, etc.) surfaces. A
variation the concept is to provide load-bearing area as blocks
or pads as shown in Fig. 9 with enough space around them to
trap the wear debris. The topology of the surface structure Fig. 9 Micro-patterned surfaces of silicon. The line-
could be widely different, but the critical functionality is width of (a) and (b) is 5 µm, and of (c) and (d) 50 µm
entrapment of wear debris. Fig. 10 schematically illustrates the [33].
mechanism of particle entrapment in the pockets.
Extensive experimental and theoretical investigations at
MIT under a wide variety of experimental conditions––
materials, dry and lubricated sliding, loads, speeds, and so on––
show that the undulating surfaces indeed exhibit low friction
for a long time. Studies on dry surfaces for electrical contacts
by Saka, Liu and Suh [24], and on the role of metallurgical
compatibility by Suh and Saka [25] have shown that friction
can be maintained at a low value. Experiments were also
conducted on boundary-lubricated surfaces at room temperature
by Tian, Saka and Suh [26], and at elevated temperatures by
Saka, Tian and Suh [27]. In situ observations by Oktay and
Suh [28] reveal that the wear debris is swept away and is
trapped in the pockets. Additional work by Mosleh and Suh
[29] on high-vacuum undulated sliding bearings, by Mosleh,
Laube and Suh on coated surfaces [30] indeed support the
hypothesis. The work by Mosleh, Saka and Suh [31] on sliding Fig. 10 Schematic of wear particle generation and
bearings with tight clearances is especially noteworthy. Further agglomeration on the tip of a pin (a,b,c) and the
investigations by Hwang, Kim and Lee [32] and Kim, Cha and prevention of agglomeration by undulating surfaces
Sung [33] at Yonsei University, Korea, corroborate the MIT (d,e,f) [28].
findings. Figs.11 through 17 summarize the results.

Fig. 8 Scanning electron micrograph of an Fig. 11 Friction Coefficient of Cu-on-Cu: (a) without
undulating, or patterned, Cu surface. The Cu surface and (b) with undulations [25].
was coated with 3.5 µm Ni and 1 µm Sn-34 Pb over Ni
for electric contact applications [24].

8 Copyright © 2015 by ASME


Fig. 12 Friction Coefficient of Ti surface, with and Fig. 15 Pin-on-disk experiments with uncoated
without undulations, sliding on AISI 52100 steel stainless steel 304 sliding against M50 steel ball
lubricated with oleic acid [26]. It may be noted that under a normal load of 200 g [29-31].
oleic acid is not a good lubricant for Ti.

Fig. 16 Si N coated flat surface and the undulated


3 4
surface. The undulated surface did not wear much
and the wear particles were trapped in the pockets
Fig. 13 Effectiveness of mineral oil as a lubricant for [33].
AISI steel slid against OFHC Cu, with and without
undulations. Friction coefficient is low even above
the transition temperature with undulated surface
[27].

Fig. 17 Friction coefficients of aluminum coated flat


Fig.14 Pin-on-disk experiments with uncoated surface and micro-scale patterned undulated
stainless steel 304 sliding against M50 steel ball surfaces at two different loads [33].
under a normal load of 50 g [29-31].

9 Copyright © 2015 by ASME


It is amply apparent that the above results conclusively microscopic theories of friction can be extended well into the
confirm the hypothesis that agglomerated wear particles are sub-micrometer or nano-scale. But if the asperity-contact size is
primarily responsible for high friction in dry sliding. only a few nanometers or so, the classical continuum-scale
Consequently, by trapping the wear debris in pockets, friction theories should be abandoned in favor of the modern
can be minimized. The coefficient of friction of undulated molecular-scale models.
surfaces even in dry sliding can be as low as those of boundary- In closing, it is perhaps humbling to note that while the
lubricated surfaces. The role of boundary lubrication is to definition and measurement of friction coefficient are quite
prevent the wear particle agglomeration at the sliding interface. elementary, construction of the theories of friction requires, by
The effectiveness of boundary lubricants ceases above a critical contrast, the the entire repertoire of mechanics and mechanical
temperature (about 100 to 250 oC, depending on the lubricant). behavior of materials. Since Leonardo’s time many eminent
Above the transition temperature a lubricant cannot prevent researchers have determined friction coefficients using such
agglomeration of the wear particles because it does not wet the simple devices as the inclined plane, belt-over-pulley, pin-on-
surface. In such instances the undulating surfaces are optimal. disk, and so on. But the theory and calculations, however, have
In geometrically confined spaces (e.g., a steel shat in a bushing) been another matter. There are no closed form solutions and the
with tight clearances, especially, undulating surfaces with computer calculations are only approximate. Many inputs,
grooves is an effective way of lowering friction and preventing such as the interfacial boundary conditions, are not even known
seizure. As of now, though, the geometry of the surface features with certainty. Accordingly, the friction theories and
is not optimized. The scale of undulations is of the order of 100 calculations will remain challenging for sometime to come. In
µm. By optimizing the geometry, materials, etc. it should be the meantime, however, several practical solutions, such as the
possible to lower both friction and wear. patterned surfaces, thin soft and hard films, boundary
lubrication, and their combinations, can be used to minimize
5 SUMMARY friction and wear. These, however, are suboptimal solutions.
In this article, the laws and theories of friction are
reviewed and Prof. Nam P. Suh’s contributions to the theories ACKNOWLEDGMENTS
of friction are highlighted. The classical laws of friction are The many helpful suggestions on the manuscript by
valid only in a small range of the experimental or service Drs. Said Jahanmir, Kyriakos Komvopoulos, and Dae-Eun Kim
conditions. As the real area of contact approaches the nominal are acknowledged with pleasure. I am deeply indebted to Prof.
contact area, the laws are likely to breakdown, especially if the Suh for introducing me to the field of Tribology and for
sliding speeds are high. In extreme cases, such as in high according the opportunity to collaborate with him and his
vacuum, the surfaces are likely to seize, and the friction students over the decades.
coefficient would be extremely large. For the laws of friction
to be valid, the asperity contacts should be sparsely distributed REFERENCES
and the relative velocity should be small. [1] Dowson, D., 1979, History of Tribology, Longman Group
As to the theories of friction, several seemingly different Limited, London.
theories have been proposed over the decades. They include: [2] Bowden, F.P., and Tabor, D.,1973, Friction: An
roughness, adhesion, deformation, and plowing. Among them, Introduction to Tribology, Anchor Press/Doubleday,
plowing is the most severe. In any viable theory of friction on Garden City, NY.
micro-scale, permanent deformation of the surface is a [3] Rabinowicz, E., 1965, Friction and Wear of Materials,
necessary requirement. While it may seem that there are several John Wiley & Sons, New York.
theories of friction, they all can be unified into a single grand [4] Suh, N.P., 1986, Tribophysics, Prentice Hall, Englewood
problem: the problem of plastic deformation of asperity (or Cliffs, NJ.
asperity/surface) contacts under specified force and [5] Suh, N.P., and Saka, N., 1980, Fundamentals of
displacement boundary conditions, interface conditions, and Tribology, The MIT Press, Cambridge, MA.
“constitutive “ equations. It is a boundary value problem in the [6] Tomlinson, G.A., 1929, “Molecular Theory of Friction,”
mechanics of deformable bodies. Depending on the values of Phil. Mag., 7th Series, 7, pp. 905-939.
the various parameters different mechanisms may prevail. [7] Ernest, H., and Merchant, M.E., 1940, “Surface Friction
In continuous sliding, however, wear particles are largely Between Metals––a Basic Factor in Metal Cutting
responsible for the high friction. The wear particles Processes,” Proc. Special Summer Conf. Friction and
agglomerate and plow the surface. The long plowing grooves Surface Finish, pp. 76-101, The MIT Press, Cambridge,
invariably present on the sliding surfaces are the telltale signs MA.
of the plowing phenomenon. In such systems both friction and [8] Bowden, F.P., Moore, A.J.W., and Tabor, D., 1943, “The
wear can be minimized by patterning the surface of the softer Ploughing and Adhesion of Sliding Metals,” J. Appl.
body to trap the wear debris. It is also possible to reduce Phys., 14, pp. 80-91.
friction by boundary lubrication, wherein adsorbed lubricant [9] Green, A.P., 1954, “The Plastic Yielding of Metal
molecules on sliding surfaces prevent wear particle Junctions Due to Combined Shear and Pressure,” J. Mech.
agglomeration. Furthermore, it appears that the classical Phys. Solids, 2, pp. 197-211.

10 Copyright © 2015 by ASME


[10] Green, A.P., 1955, “Friction Between Unlubricated [22] Komvopoulos, K., Saka, N., and Suh, N.P., 1986, "The
Metals: A Theoretical Analysis of the Junction Model,” Significance of Oxide Layers in Boundary Lubrication,"
Proc. Roy. Soc., A, 228, pp. 181-204. Journal of Tribology, Trans. ASME, 108, pp. 502-513.
[11] Challen, J.M., and Oxley, P.L.B., “An Explanation of the [23] Komvopoulos, K., Saka, N., and Suh, N.P., 1987, “The
Different Regimes of Friction and Wear Using Asperity Role of Hard Layers in Lubricated and Dry Sliding,"
Deformation Models,” 1979, Wear, 53, pp. 229-243. Journal of Tribology, Trans. ASME, 109, pp. 223-231.
[12] Suh, N.P., and Sin, H.-C., 1981, “The Genesis of [24] Saka, N., Liou, M.-J., and Suh, N.P., 1984, “The Role of
Friction,” Wear, 69, pp. 91-114. Tribology in Electrical Contact Phenomena," Wear, 100,
[13] Sin, H.-C., Saka, N., and Suh, N.P., 1979, “Abrasive pp. 77-109.
Wear Mechanisms and the Grit Size Effect," Wear, 55, [25] Suh, N.P., and Saka, N., 1987, “Surface Engineering,"
pp. 163-190. Annals of CIRP, 36, pp. 403-408.
[14] Suh, N.P., Sin, H.-C., Tohkai, M., and Saka, N., 1980, [26] Tian, H., Saka, N., and Suh, N.P., 1989, "Boundary
"Surface Topography and Functional Requirements for Lubrication Studies on Undulated Titanium Surfaces,"
Dry Sliding Surfaces," Annals of the CIRP, 29, pp. 413- STLE Tribology Transactions, 32 (2), pp. 289-296.
418. [27] Saka, N., Tian, H., and Suh, N.P., 1989, "Boundary
[15] Suh, N.P., Sin, H.-C., and Saka, N., 1980, “Fundamental Lubrication of Undulated Metal Surfaces at Elevated
Aspects of Abrasive Wear," Fundamentals of Tribology, Temperatures," Tribology Transactions, 32(3), pp. 389-
Edited by Suh, N.P., and Saka, N., The MIT Press, 395.
Cambridge, MA, pp. 493-518. [28] Oktay, S.T., and Suh, N.P., 1992, “Wear Debris
[16] Hardy, W.B., and Hardy, J.K., 1919, “Note on Static Formation and Agglomeration,” ASME Journal of
Friction and on the Lubricating Properties of Certain Tribology, 114, pp. 379-393.
Chemical Substances,” Phil. Mag, 6th Series, 38, pp. 32- [29] Mosleh, M. and Suh, N.P., 1995, “High Vacuum
48. Undulated Sliding Bearings,“ STLE Tribology
[17] Hardy, W., and Doubleday, I., 1922, “Boundary Transactions, 38, pp. 277-284.
Lubrication –– The Paraffin Series,” Proc. Roy. Soc. A, [30] Mosleh, M., Laube, S.J.P., and Suh, N.P., 1999, ”Friction
102, pp. 550-574. of Undulated Surfaces Coated with MoS2 by Pulsed Laser
[18] Jahanmir, S., and Beltzer, M., 1985,“An Adsorption Deposition,” STLE Tribology Transactions, 43(2), pp.
Model of Friction in Boundary Lubrication,” ASLE 495-502.
Transactions, 29 (3), pp. 423-430. [31] Mosleh, M., Saka, N. and Suh, N.P., 2002, "The
[19] Jahanmir, S., and Beltzer, M., 1985, “Effect of Additive Mechanism of High Friction in Dry Sliding Bearings,"
Molecular Structure on Friction Coefficient and Wear, 252, pp. 1-8.
Adsorption,” Journal of Tribology, Transactions of the [32] Hwang, D.H., Kim, D.E., and Lee, S.J., 1999, “Influence
ASME, pp. 1-8. of Wear Particle Interaction in the Sliding Interface on
[20] Komvopoulos, K., Saka, N., and Suh, N.P., 1985, "The Friction of Metals,” Wear, 225-229, pp. 427-439.
Mechanism of Friction in Boundary Lubrication," Journal [33] Kim, D.E., Cha, K.H., and Sung, I.H., 2002, “Design of
of Tribology, Trans. ASME, 107, pp. 452-462. Surface Micro-structures for Friction Control in Micro-
[21] Komvopoulos, K., Saka, N., and Suh, N.P., 1986, systems Applications,” Annals of CIRP, 51/1, pp. 495-
"Plowing Friction in Dry and Lubricated Sliding of 498.
Metals," Journal of Tribology, Trans. ASME, 108, pp.
301-313.

11 Copyright © 2015 by ASME

View publication stats

You might also like