Sunteți pe pagina 1din 6

1. Wear mechanisms review significant.

It is important to remark that a worn


surface never exhibits solely one phenomenon. [4]
Since 1968 wear has been defined as ‘the
progressive loss of substance from the operating This literary review intends to showcase how
body surface, which is resultant of relative motion abrasive and adhesive wear are modeled at a
at the surface’. In 1980 Godfrey stated about macro scale as well as at the micro one.
twelve kinds of wear which are categorized based
on the wear mechanisms present. Burwell listed 2. Wear Theories
four major mechanisms: adhesive wear, abrasive
wear, surface fatigue, corrosive wear, while all 2.1 Adhesion Theory
other mechanisms were listed as ‘minor types of
wear’. [1] This literary review will focus mainly on Originally proposed by Bowden and Tabor;
the first two, adhesive and abrasive. and formulated semi-empirically by Archard in
1953 the wear theory of adhesion is based on
First, adhesive wear can be visualized as a form of the adhesion theory of friction. It states that
cold welding, where bonding of surfaces prevents two surfaces that experience relative
further relative motion. Adhesion mechanism can movement will form junctions between their
be based on electron transfer between the asperities that will be sheared for sliding to
surfaces making it rather immediate and continue. This cycle of formation an shearing
spontaneous, regardless of the interface will lead to the wearing of the surfaces in an
temperature or it can be based on diffusion even manner if these are similar or unevenly if
between the surfaces where the adhesion process dissimilar. Formation of junctions, however, is
is extremely slow and highly dependent on dependent of surface finish and material
temperature. [2] Under atmospheric conditions, microstructure, see figure 1.
adhesive wear is less likely to happen than
abrasion, nevertheless, dry sliding friction between
similar materials can cause it. Worn surface would
exhibit material transfer, crests and scaling as a
result of adhesive wear.[3] Abrasive wear, on the
other hand, is the most likely to happen between
dissimilar materials in relative motion. It involves
fatigue, grain detachment and brittle/ductile
Figure 1. True Contact Between Surfaces[2]
fractures. Usually a soft surface is worn by a
particle that is hard enough to plastically deform The equation governing wear, as formulated
the surface, and even plough a groove in it if a by Archard, relates the wear volume V to the
sufficient load is provided. In spite of having normal force P, the penetration hardness p,
severe stages, abrasive wear can be easily the sliding distance S and an empirical
eradicated by using an adequate combination of wearing coefficient as follows [4]
materials or a lubricant. [2]
kP S
V = (1)
3p
Other generic wear classification are single-cycle
deformation, which are those where deformation, Archard’s equation suggests three laws: the
displacement or fracture occur after a single volume of material worn is proportional to the
engagement; and repeated-cycle deformation, sliding distance (1st law) and also to the load
which require repeated contact for wear to be (2nd law); and it is inversely proportional to the
manifested as delamination, ratcheting or fatigue. hardness of the softer material (3rd law). The
Another mechanism that is always present is the first law is held true for a wide range of
atomic one which involves the migration of conditions, meanwhile, the second one is true
individual atoms between surfaces. At an atomic for very low loads to approximately one third
scale both thermal mechanisms, those associated of the bulk yield stress of the softer material.
with a rise in temperature, and chemical ones, The third one is supported by much
those that involve the formation of oxides or experimental work but the worn volume is
chemical products on the contacting surfaces, are
always less than this theoretical approximation cracks
…\)g_gQ™that shear to the surface or as wear

often by degrees of magnitude. debris in the form of flakes. [8]

When abrasive wear occurs, wear volume V


removed by abrasive wear is related to sliding
distance S, penetration hardness p, normal
force P, angle of abrasion θ and the wearing
coefficient k according to [4]
2k tanθ P S
V = (2)
πp

2.2 Diffusion Theory Figure 3. Schematic of delamination wear


  ‘u“‹ 2YAb!\2 gJ :A_!b\d!\gd —A!… Jg…b!\gd !d: !d A˜!bp_A gJ :A_!
It states that two surfaces in relative motion ‡\_–A…[p_!A: 2gppA… 2gd!2‡u
The depth h of the subsurface dislocation pile-
will develop an interface via diffusion, up can be defined in terms of the shear
particularly at the asperities whereas junctions YA 2gd!2\dQ bA!_‡u ‡ ! …A‡”_8 YA pY!‡A )g”d:!…™ ‡p!2\dQ :A2…A!‡A‡ ”d
modulus G, Poisson’s ration ν, the stress σf
would be formed in the adhesion theory. For 2…™‡!__\dA b\˜”…A A˜2AA:‡ YA J…AA AdA…Q™ gJ YA !bg…pYg”‡ pY!‡Au Y
preventing the dislocation motion and the
)…A!^‡ :g—d !d: …A_A!‡A‡ ! b”_\”:A gJ :\‡_g2!\gd‡u ”…YA… Jg…b!\gd gJ !
atoms to jump from one surface to the other, Burger2!dvector b Y…g”QY
as
bA!_‡ g22”… YA ‡g_\: ‡!A d”2_A!\gd gd YA …Abd!d 2…™
high localized temperature is needed. The \Q”…A ‘u“B \__”‡…!A‡ YA pY!‡A
Gb )g”d:!…™ A–g_”\gd ! ‡YA!… Y! p…gbgA
surface with the highest diffusion flux or the h=
pY!‡A …!d‡\\gd‡u (3)
less favorable solubility conditions will have a 4π (1 − ν) σf
YA bg:A_ gJ :A_!b\d!\gd \‡ gdA A˜!bp_A gJ YA !…!2\–A pg—A… g
depleted surface, evidently a wear scar. In spitegJ 2gbp_A˜
d!\gd‡ of its popularity,
pYAdgbAd! this
! YAtheory is gJonly
)g”d:!…™ b!A…\!_‡ ‡2\Ad2A !d: bA
Diffusive wear is likely to occur when there is \‡ dg !pp_\2!)_A g bg‡ gJ YA d”bA…g”‡ 2!‡A‡ gJ —A!…8 \ Y!‡ dgdAYA_A‡‡ p
valid at low speeds, where the localized rise in
true contact between the atoms and a high ‡”:\A‡ \d ‡…”2”…!_ …!d‡Jg…b!\gd‡ gJ b!A…\!_‡ \d J…\2\gdu YA KA_: gJ A_A
temperature is insignificant so that neither
temperature at the interface, see figure 2. !JJA2A: )™ bA2Y!d\2!_8 A_A2…\2!_8 !d: YA! pYAdgbAd! 2gd\d”A‡ g )A
diffusion nor phase transformation are
significant.

3. Experiments

3.1 Slurry abrasivity test

Figure 2. Schematic of diffusive wear The purpose of this test was to rank the
abrasivity of slurries and to provide a ranking
2.3 Delamination Theory capability for wear resistance against a
particular slurry. Experimental setup consists
Elaborated by Suh in 1974, it intends to explain of a flat rectangular wear specimen, mounted
wear phenomena taking into detailed on an arm, sliding back and forth across a
considerations subsurface dislocation rubber lap, flooded by a slurry, see figure 4.
interactions. [1] Barwell summarized the
physical basis of the theory in the following
four stages, see figure 3. First, plastic loading   ‘u“Š  b\2…gQ…!pY ‡Yg—\dQ p!… gJ ! _gdQ\”:\d!_ ‡A2\gd gJ !d 

flattens the surface asperities of the softer r)_g2^[gd[…\dQ QAgbA…™su YA :AJg…b!\gd ‡”)‡…”2”…A –!…\A‡ —\Y :ApY )A_g—
2”…–A: _\dA \‡ ! Q…!\d )g”d:!…™‰ \ Y!‡ )Ad \d YA ‡_\:\dQ :\…A2\gdu ……g— ‡Yg—
material between mating surfaces in relative 2g”dA…J!2Au !…^ b!A…\!_ dA!… !……g— \‡ bA2Y!d\2!__™ b\˜A: b!A…\!_u ˜pA…\bAd!
motion, so that at each flattened asperity there ‹Š  r‹uB ^Qs‰ jœœ …A–g_”\gd‡ gJ ™pA OOœ ‡AA_ …\dQ‰ j“ b g!_ ‡_\:\dQ :\‡!d2A‰ j
is a cyclic shear loading in the subsurface. !bg‡pYA…Au r…gb \QdA™8 u u8 …\)g_gQ™ \d YA  !d: g…bA… g–\A d\gd
Subsequently, nucleation of voids and cracks ™‡Y^\d8 u u :‡u L‘E‹L8 __A…gd …A‡‡8 A— g…^8 jeeOu \Y pA…b\‡‡\gdus
Figure 4. Slurry Abrasivity Test Setup [6]
begins as loading continues, being crack
growth partially prevented by the surface The arm lifts the specimen at the end of each
© 2007 by Taylor & Francis Group, LLC
compressive stresses. Further cycling, causes stroke to allow the slurry to come between the
dislocation pile-ups that favor voids and specimen and the lap. Whenever the
cracks growth via linking and at a critical crack specimen is moved, mixing and circulation of
length failure occurs. Failure is visible as the slurry occurs in the tray. The wear
specimen is removed periodically and weight
loss is determined. By varying the slurry and
keeping the wear specimen constant, a
ranking of abrasivity of the slurries can be
obtained, known as the Miller Number. The
higher this number, the more abrasive the
slurry is. If the slurry is kept constant instead
the wear resistance of different materials to
that slurry can be ranked, using SAR Number.
The wear rate from the wear curve is Figure 7. Flat-Ended Pin Alignment [6]
multiplied by a factor of 18.18 h/mg to obtain
the dimensionless Miller Number. The SAR ASTM recommends to report volume loss
Number is defined as instead of mass or geometrical loss data. For
SAR = (M illerN o.)
7.58
(4) mass loss, dividing by the density yields the
specimen specif ic weight volume loss. For geometrical loss, see figure
Experiments have shown that the Miller 8, the wear volume V is related to the width of
Number becomes less sensitive to the the wear track W on the disk, the spherical
concentration of the slurry as the concentration radius R of the pin and the radius of the wear
increases, and that oil-based slurries tend to track D according to the following formulas
result in lower wear than water-based slurries,
among others.

3.2 Pin-on-disk test

This test is used to evaluate the sliding wear


behavior of material parts. The setup consists Figure 8. Wear Scars on Pin(A) and Disk(B) [6]
of a radius-tipped or flat-ended pin pressed � �
against a flat disk, see figure 6. The relative πW 4
Vpin = (5)
motion between the two is such that a 64R
circumferential wear path on the disk surface is � �
generated. Either the pin or the disk can be πW 3 D
Vdisk = (6)
moving[6]. The test can be done in a controlled 6R
atmosphere on dry or lubricated surfaces.
3.3 Block-on-ring test

The Block-on-ring test is used to study sliding


wear and to rank materials in terms of
resistance to sliding wear [6]. The test
measures the wear on the block material, and
not the wear on the ring. The ring material is
fixed, while the block material is varied. The
block is pressed against the rotating ring, see
figure 9. Wear on both the block and the ring is
Figure 6. Pin-on-disk setup and various pins[6]
measured after a fixed number of revolutions.
To compare materials, the wear on both
members is measured after a set number of
revolutions. The relationship between wear and
sliding duration tends to be nonlinear because
the stress levels in the rounded pin change
due to wear. When using flat-ended pins, a
wear-in process is necessary until the surface
of the pin aligns with the surface of the other
member, see figure 7.
lubrication induces a strong shear stress that
removes material at the wafer surface. First slurry
chemicals are diffused and adsorbed onto the
wafer surface, reacting with surface atoms and
molecules to form a weakened protective layer.
This modified layer is then sheared by the non-
newtonian flow of the slurry particles. The
strength of the shear results from the high shear
rate at local removal points and the non-
Newtonian viscosity being more effective than its
bulk value. Finally, removed atoms are
transported away downstream by the slurry flow,
see schematic in figure 11.

Figure 9. Block-on-Ring Test Setup [6]

Wear volume for the block is used to rank


materials in terms of their wear resistance. It is
calculated using the following geometrical
relationships, see wear track in figure 10.

Figure 10. Wear on Block [6]


� �
θ
Scar W idth = b = Dsin (7)
2 Figure 11. Schematic diagram for the unified removal
� � mechanism: (a) react: newly created molecules (filled-in gray
b circles) generate an altered layer; (b) shear: slurry particles
θ = 2sin−1 (8) (filled-in black circles) pull off molecules from the altered layer;
D
(c) transport: removed molecules are carried away downstream
d2 t and atoms (open circles) under the altered layer are exposed.
Scar V olume = [θ − sin(θ)] (9)
8 This mechanism is bolstered by various
experimental studies [12,13,14], where researchers
4. Numerical Results - CMP used a polishing pad with characteristic pores of
about 25 microns. None of these experiments,
Chemical Mechanical Planarization (CMP) is the exhibited direct participation slurry particles in the
preferred planarization step utilized in deep sub- process of material removal. Their abrasive
micron IC manufacturing[9], to remove roughness plowing effect was observed in the first stage of
from silicon oxide, metal and polysilicon surfaces. chemical reactions. Thus, it can be assumed that
Material removal mechanisms present in the active particles are embedded on the pad surface
CMP process will be discussed from an unified as part of the trapped fluid inside the pores and at
perspective rather than Kaufman’s[10] chemistry- least some of them do not roll with respect to the
aided mechanical abrasion or Cook’s chemical pad. These particles that slide with respect to the
tooth mechanism[11]. The non-Newtonian slurry wafer surface will generate heat that will lead to a
flow in the regime of mixed or boundary temperature rise. Stein[15] has shown that during
the CMP process the temperature rise in the where C N is the constant of proportionality.
system can be calculated Substituting (12) into the expression (11), the maximum
Ws number of molecules to be removed simultaneously
[10]
∆T = (N*) by a single slurry particle can be deduced as
Fslurry Cp �� ��
� 2τ �
where Ws is the“shaft work” due to friction, Fslurry is

N = � � � [13]
πγR �
the molar flow rate of the slurry, and Cp is the m

average heat capacity of the slurry. The slurry The expression states that the material removal
increased temperature of the slurry will accelerate rate from the wafer surface is independent of the
chemical reactions on the wafer surface, thus a slurry particle size. Such prediction is consistent
higher material removal rate can be obtained. with the study of the effect of abrasive size on the
Indeed a strong correlation between material material removal rate [22]. It is generally accepted
removal rate and slurry temperature has been that the material removal rate does not change
found [15]. It has also been reported by Sugimoto[16] significantly with the abrasive size if the abrasive
that the removal rate of an oxide layer is sensitive particle size is less than 300nm [22]. However,
to the wafer temperature, and that the relation variations of material removal rate with respect to
between them is linear. slurry particles of different sizes have been
reported [17,19]. According to the study by Luo et al.
A mathematical model is obtained de Rosas [19], such variations may be ascribed to differences
Further it is assumed that the shear stress is in the morphology with abrasive particles rather
uniformly distributed around the slurry particle. The than different sizes. However, Luo and Dornfeld
shape of molecules in the altered layer on the studies [17], were on abrasive sizes of about a
wafer surface is modeled as a sphere with micron, thus much larger than 300 nm. The
diameter Rm. The slurry particle is regarded as a material removal rate in volume is qualitatively
perfect ball with diameter Rp. Since the size of a proportional to the volume of the molecule in the
molecule to be removed is usually on the scale of altered layer, to the shearing frequency by the
several angstroms, a slurry particle is generally slurry particle, and to the maximum number N∗
much larger than a molecule[17,18,19], i.e., Rp >> Rm. and it can be approximated to
Thus, it is highly probable for a slurry particle to
remove more than one molecule at a time. To τ 2
Ẋ ≈ R fτ [14]
remove N molecules simultaneously from the 3γ m
altered layer on the wafer surface, the inequality
The shearing frequency fτ depends on many
�π � �π � process parameters, particularly on the weight
2 2
τ R Rm ≥ 2γ R N [11] concentration of particles in the slurry and on the
4 p 4 m
relative sliding speed between the wafer surface
and the pad. For the weight concentration to be
has to be satisfied, where γ is the surface energy
between the altered layer and sub-layer molecules small, the simplest linear dependence of fτ is
[20]. The inequality suggests that in order to move
reasonable, i.e.,
simultaneously N molecules from the wafer surface fτ = c1α α + c0alpha [15]
to a distance of Rm, the energy transferred through
the slurry particle by the shear stress must be Various independent studies[15,17,18,19] confirm
greater enough to overcome the surface energy of such dependence. When the weight concentration
newly created surface if these N molecules are increases beyond a certain level, the critical
moved. It is implicitly assumed that once a weight fraction c at the shearing spots on the
molecule in the modified layer is moved off a altered layer may become saturated, making the
distance of its diameter or larger, it will be removed material removal rate X independent of the weight
permanently from the wafer surface. An fraction. Conversely, the shear stress depends on
experimental model based on surface topography all the process parameters such as the applied
[21] shows that the number of molecules removed pressure on the wafer surface and the relative
simultaneously by a single slurry particle is sliding velocity u, it is not clear how to establish
proportional to the contact area, and can be the qualitative relation between the material
expressed as � � removal rate and the relative sliding velocity u.
π 2 Additionally, in the experimental studies the sliding
N= R CN [12]
4 p
velocity is usually on the order of several m/s
while in a typical CMP process, the relative sliding 6. References
velocity is about 1m/s. Nevertheless, the shear
stress can be computed quantitatively by 1. R. D. Arnell, P.B. Davies, J. Halling and T.L. Whomes. Tribology:
Principles and Design Applications. Springer-Verlag New York. 1991
combining contact mechanics of pad asperities 2. Andrew William Batchelor, Loh Nee Lam and Margam
and the wafer surface with fluid mechanics of the Chandrasekaran. Materials Degradation and its Control by Surface
slurry flow[25]. Since the fluid as a continuum in- Engineering. Imperial College Press. 1999
3. Theo Mang, Kirsten Bobzin and Thorsten Bartels. Industrial
between the wafer surface and the pad is Tribology: Tribosystems, Friction, Wear and Surface Engineering,
separated by asperities at each discrete contact Lubrication. Wiley-VCH 2011
spot, it is only meaningful to solve the fluid 4. R. G. Bayer. Wear Analysis for Engineers. HNB Pub. NY, 2002
5. Bayer, R.G. Engineering Design for Wear: Design Perspective of
problem in full three dimensions, i.e. Wear Behavior, 2nd Ed, CRC Press 2004 pp 45–135
d 6. Bayer, Raymond G. Mechanical Wear Fundamentals and Testing,
ρv = − � ρvv − �π + ρg [16] Revised and Expanded CRC Press 2004
dt 7. Booser, Richard E. Theory and Practice of Tribology, CRC
Handbook of Lubrication Volume II: Theory and Design CRC Press
where ρ is the fluid density, v is the fluid velocity, 1988 pp 201–208
and π is the momentum flux tensor and its 8. Ludema, K.C. Friction, Wear, Lubrication. A Textbook in Tribology
connection with the shear stress CRC Press 1996
9. J. M. Steigerwald, Shyam P. Murarka and R.J. Gurmann, Chemical
π = pδ + τ [17] Mechanical Planarization of Microelectronic Material, Wiley , 1997
10. F.B. Kaufman, D.B. Thompson, R.E. Broadie, M.A. Jaso, W.L.
where p is the fluid pressure. Since the material Guthrie, D.J. Pearson, M.B. Small, Chemical–mechanical polishing
for fabricating patterned W metal features as chip interconnects, J.
removal occurs at the wafer surface, it is Electrochem. Soc. 138 (1991) 3460–3464.
appropriate to focus on the flow behavior at the 11. L.M. Cook, Chemical processes in glass polishing, J. Non-Cryst.
interface between the slurry and the wafer surface. Solids 120 (1990)152–171.
12. W. Lortz, F. Menzel, R. Brandes, F. Klaessig, T. Knothe, T.
Hence, instead of solving equation 16, the liquid Shibasaki, News from the M in CMP – viscosity of CMP slurries a
crystal model or the porous media model for thin constant? Mater. Res. Soc. Symp. Proc. 767 (2003) F1.7.1–
film lubrication are two favorable alternatives[24]. F1710.1710.1710.
13. A.S. Lawing, Pad conditioning and pad surface characterization in
Note that expression (14) can explain the edge oxide chemical mechanical polishing, Mater. Res. Soc. Symp. Proc.
effect for a blanket wafer or the doming effect for a 732 (2002) I5.3.1–I536.536.536.
patterned one[25]. Such effects may damage wafer 14. L. Shan, J. Levert, L. Meade, J. Tichy, S. Danyluk, Interfacial fluid
mechanics and pressure prediction in chemical mechanical
in the microelectronics industry. These are due to polishing, ASME J. Tribol. 122 (2000) 539–543.
the non-uniform material removal at the wafer 15. D.J. Stein, D.L. Hetherington, J.L. Cecchi, Investigation of the
edge or along the circumference of a feature. kinetics of tungsten chemical mechanical polishing in potassium
Slurry turbulent flow may cause non-uniformity. iodate-based slurries. I. Role of alumina and potassium iodate, J.
Electrochem. Soc. 146 (1999) 376–381.
16. F. Sugimoto, Y. Arimoto, T. Ito, Simultaneous temperature
5. Conclusions measurement of wafers in chemical mechanical polishing of silicon
dioxide layer, Jpn. J. Appl. Phys. 34 (1995) 6314–6320.
17. J. Luo, D.A. Dornfeld, Effects of abrasive size distribution in
Being wear, the progressive loss of material at a chemical mechanical planarization: modeling and verification, IEEE
surface due to relative motion between two Trans. Semiconduct. Manuf.(2003) 469–476.
surfaces, it can be desired as in CMP processes or 18. J. Luo, D.A. Dornfeld, Material removal regions in in chemical
mechanical planarization for submicron integrated circuit fabrication:
undesired causing an annual billionaire loss. coupling effects of slurry chemicals, abrasive size distribution, and
Though wear is visible at the macro scale, it is at wafer–pad contact area, IEEE Trans. Semiconduct. Manuf. 16
the micro one where roughness and particle (2003) 45–56.
19. Q. Luo, S. Ramarajan, S.V. Babu, Modification of the preston
distribution determine wear mechanism evolution. equation for the chemical–mechanical polishing of copper, Thin
Several theories and models, besides Archard’s, Solid Films 335 (1998) 160–167.
have been developed and compared to widely 20. Y.-T. Su, S.-Y. Wang, J.-S. Hsiau, On machining rate of
hydrodynamic polishing process, Wear 188 (1995) 77–87.
used experiments that characterize materials wear 21. N. Gane, P.F. Pfaelzer, D. Tabor, Adhesion between clean surfaces
resistance or simulate common operating at light loads, Proc. R. Soc. Lond. A 340 (1974) 495–517.
conditions i.e. pin-on-disk. Tendencies between 22. I. Ali, S.R. Roy, G. Shinn, Chemical–mechanical polishing of
interlayer dielectric: a review, Solid State Technol. 37 (1994) 63–70.
experimental data and models’s predictions are 23. K.L. Johnson, J.A. Greenwood, S.Y. Poon, A simple theory of
congruent though wear coefficients may differ by asperity contact in elastohydrodynamic lubrication, Wear 19 (1972)
orders of magnitude. Nanotribology research on 91–108.
24. J.A. Tichy, Modeling of thin film lubrication, Tribol. Trans. 38 (1995)
thin film lubrication has been valuable to explain 108–118.
non-uniformity of worn surfaces by CMP processes 25. A.R. Baker, The origin of edge effect in chemical mechanical
versus continuum and fluid mechanics approach. planarization, in: I. Ali, S. Raghavan (Eds.), Chemical Mechanical
Planarization, I, Pennington, NJ, 1997, pp. 228–238.

S-ar putea să vă placă și