Sie sind auf Seite 1von 10

Journal of Materials Processing Technology 214 (2014) 285294

Contents lists available at ScienceDirect


Journal of Materials Processing Technology
j our nal homepage: www. el sevi er . com/ l ocat e/ j mat pr ot ec
Modeling and analysis of the material removal prole for free abrasive
polishing with sub-aperture pad
Cheng Fan
a,b
, Ji Zhao
a
, Lei Zhang
a,
, Yoke San Wong
b
, Geok Soon Hong
b
, Wansong Zhou
a
a
College of Mechanical Science and Engineering, Jilin University, Changchun 130022, China
b
Department of Mechanical Engineering, National University of Singapore, Singapore 117576, Singapore
a r t i c l e i n f o
Article history:
Received 16 February 2013
Received in revised form3 September 2013
Accepted 9 September 2013
Available online 19 September 2013
Keywords:
Free abrasive polishing
Material removal prole
Sub-aperture pad
Material removal index
a b s t r a c t
This paper addresses the problem of material removal in free abrasive polishing (FAP) with the sub-
aperture pad both theoretically and experimentally. The effects of some polishing conditions upon the
material removal are analyzed, including not only the process parameters, which refer to the normal
force, angular spindle velocity and angular feed rate, but also the abrasive grain size, polishing slurry
properties, topographical parameters of the sub-aperture pad, as well as tool path curvature. Based on
the analysis, a model of material removal prole is proposed to facilitate more accurate polishing. First,
by analyzing the contact among polishing pad, abrasive grain and workpiece surface in the micro level,
the removal depth per unit length of the polishing path is derived, which is dened as the material
removal index. Then, the distribution of this removal index can be obtained via modeling the pressure
and relative sliding velocity in the contact region of polishing pad and workpiece. After that, the material
removal prole can be calculated by integrating the material removal index along the tool path in the
tool-workpiece contact region. To verify the effectiveness of the proposed model, a series of polishing
experiments have been conducted. Experimental results well demonstrate that our model can accurately
predict the material removal depth during the FAP.
2013 Elsevier B.V. All rights reserved.
1. Introduction
Free-form surfaces with high surface nish, integrity and form
accuracy, such as precise aspheric and non-geometric optical lens
and molds, have become key features of components increasingly
required by optoelectronic and communication industries (Cheng
et al., 2005). The polishing process, which is usually the nal step
of fabrication, is essential to the quality and duration of the part
surfaces. Various types of polishing are used for optical fabrication,
such as free-abrasive polishing (FAP), magneto-rheological nish-
ing, abrasive slurry jet polishing, etc. FAP is considered as one of
the most common and important operations in optical nishing.
In a typical FAP process, a rotational sub-aperture polishing pad in
the presence of liquid slurry which contains hard abrasive grains is
pressed against the workpiece. Between the sub-aperture pad and
the workpiece there forms an area of contact and the interactions
between polishing pad, abrasive grains and workpiece result in the
material removal in the contact area.

Corresponding author at: College of Mechanical Science and Engineering, Nan-


ling Campus, Jilin University, No. 5988, Renmin Street, Changchun 130022, China.
Tel.: +86 18946686195.
E-mail address: zhanglei@jlu.edu.cn (L. Zhang).
The FAP process with the sub-aperture pad is usually used as an
error compensation method to improve the form accuracy of the
part surfaces if the material removal in the contact area is control-
lable. Jones (1977) rst proposed to use the FAP process to fabricate
the aspheric optical surface. By controlling the amount of material
to be removed precisely, the mid-spatial frequency errors left after
grinding or turning can be removed by sub-aperture polishing. Su
et al. (1996) and Su and Sheen (1999) developed a planning strat-
egy for the polishing process to remove an arbitrary prole with
the precondition that the material removal depth in the contact
area can be represented by a linear function of machining time.
Therefore, it is essential to model and analyze the material removal
of polishing in a quantitative manner. However, the fundamen-
tal mechanisms of material removal in FAP involve micro-fatigue,
micro-crack, ploughing and cutting mainly depending on the state
of theinteractions betweenabrasives, polishingpadandworkpiece,
which have not been well understood (Brinksmeier et al., 2006).
Thus, the research on the material removal rate (MRR) of FAP is
still at an empirical level. The Prestons equation is widely used to
predict the MRR during the polishing process, which states that the
MRR is determined by the product of the contact pressure, relative
velocity, machining time and Preston coefcient which is related
to the workpiece material, properties of polishing slurry and pol-
ishing pad, size of abrasive grains, etc. (Preston, 1927). Besides,
some researchers tried to explain the mechanism and establish
0924-0136/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jmatprotec.2013.09.010
286 C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294
their own MRR models fromthe perspective of descriptive model-
ing. Yu et al. (1993) modeled the dynamic interaction between pad
and workpiece. In this model, the polishing pad is characterized by
a statistical asperity model and the effect of polishing pad surface
roughness onthe MRR is included. The contact behavior of a polish-
ing pad with the workpiece surface was also investigated by Klocke
and Zunke (2009) using an FE model. The predicted results show
that the material removal in FAP can be described by the penetrat-
ing of the abrasive particles into the workpiece, and also the sliding
or yielding after the penetration. A predictive model of MRR dur-
ing the chemical-mechanical polishing was developed by Xie and
Bhushan(1996) topredict theeffects of grainsize, polishingpadand
contact pressure. Zhao and Chang (2002) presented a MRR model
for chemical-mechanical FAP based on the elastic-plastic micro-
contact mechanics. In this model, the indentation depth of a grain
into the workpiece is determined by the force equilibrium equa-
tion of the single grain. A similar MRR model was also developed
by Lin (2007), which takes the contact deformation of the abrasive
grains into consideration. Jin and Zhang (2012) proposed a statis-
tical MRR model by assuming that there are two types of abrasive
grains contributing to the material removal process, i.e., the grains
rotating between the pad and workpiece and those embedded in
the pad without rigid body motion. The previous research on the
MRR focuses on the modeling of mechanical contact action among
the pad, abrasive grains and workpiece surface. Due to the lack of
basic understanding of the mechanism of the FAP process, these
models are still regarded as half-empirical.
The MRR distribution is not uniform in the contact region,
mainly depending on distributions of the contact pressure and the
relative sliding velocity between polishing pad and workpiece sur-
face (Lee et al., 2013). The contact pressure distribution during the
polishing was modeled by Roswell et al. (2006) based on the Her-
tizian contact theory. According to the investigation by Yang and
Lee (2001), the contact area and the pressure distribution were
affected by the equivalent radius of polishing tool and workpiece
surface. In their work, the MRR distribution in the contact region
was modeled based on the Prestons equation, and the material
removal depth in the contact area was calculated by the integral
of local MRR. The MRR distribution was dened as the tool inu-
ence function (TIF) by Kimand Kim(2005). Based on the Prestons
equation, the TIF was modeledbyassumingthat the pressure distri-
butioninthe contact regionis Gaussian. The Prestons equationwas
also used by Cheung et al. (2011) to model the TIF, and the model-
based simulation was used to predict the generation of structured
surfaces. These predictive models of MRR distribution and material
removal depthareall basedonthePrestons equation. However, the
Prestons coefcient has to be determined by experiments before
each polishing process, because the changes of polishing slurry,
abrasive grains and polishing pad may result in different values
of the Preston coefcient. Since the MRR models and the MRR dis-
tribution models have been studied independently, there has been
no study directly linking the abrasive grain size and pad properties
to the material removal depth yet.
In this paper, a novel model is proposed to predict the mate-
rial removal depth of the workpiece surface. The material removal
prole, which is dened as the material removal depth orthogonal
to the polishing path, is used to describe the material removal in
the FAP process. The effects of some polishing conditions upon
the material removal are analyzed, including not only the process
parameters, which refer to the normal force, angular spindle veloc-
ity and angular feed rate, but also the abrasive grain size, polishing
slurry properties, topographical parameters of the sub-aperture
pad, as well as tool path curvature. The remainder of this paper
is organized as follows. In Section 2, the material removal index
for FAP, dened as the material removal depth per unit length of
the polishing path, is derived. In Section 3, the distributions of con-
tact pressure and relative velocity are discussed and an approach is
proposed to model the material removal prole. The validity of the
proposed model for the removal prole is veried by experiments
in Section 4. Findings are summarized in the concluding section.
2. Material removal index for free abrasive polishing
In this paper, it is assumed that the material removal from the
workpiece is primarily caused by abrasion wear by the abrasive
grains in the slurry. Besides, the abrasive grains are spherical, with
an average radius of R
abr
and uniformly distributed in the polishing
slurry. These assumptions have been substantiated by Zhao and
Chang (2002), Xie and Bhushan (1996) and Lin (2007). The material
removal index is dened as the material removal depth per unit
length of the polishing path, which is different fromthe denition
of MRR.
2.1. Contact between the polishing pad and workpiece surface
Since the polishing padis muchrougher thanthe workpiece sur-
face, the contact between themcan be regarded as that between a
rough surface and a smooth one, as is shown in Fig. 1. When the
polishing pad is pressed on the workpiece surface under a constant
load, the distance between the reference plane and the workpiece
surface is h
s
. Yu et al. (1993) proposed a statistical model to char-
acterize the polishing pad with random roughness. In this model,
the Gaussian function was used to describe the distribution of the
surface height of the polishing pad, which is given by
(h) =
1
o

2
exp
_

h
2
2o
2
_
(1)
where o is the standard deviation of the distribution of the pad
surface.
In the actual polishing process, the abrasive grains are rarely
on the contact interface. Thus it is reasonable to assume that the
Fig. 1. Interactions between polishing tool, abrasive grains and workpiece.
C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294 287
Fig. 2. Schematic of single abrasive abrasion.
entire contact load is sustained by the contact between pad and
workpiece surface. The GW model (Greenwood and Williamson,
1966) is used to describe the contact between pad and workpiece
surface. According to this model, the pressure between pad and
workpiece is given by
P
c
=
4
3
N
p
R
0.5
p

pw
_
hmax
h
(h h
s
)
1.5
(h) dh (2)
where R
p
is the average radius of the asperities, N
p
is the number
of asperities per unit area, h
max
is the maximumheight of the pad
surface and E
*
pw
is the contact modulus of the polishing pad and
the workpiece given by 1]

pw
= (1 D
2
p
)]
p
+(1 D
2
w
)]
w
. E
p
and
E
w
are the Youngs modulus of the polishing pad and the workpiece
respectively, and D
p
and D
w
are their Poissons ratio respectively.
According to Eq. (2), the pressure P
c
is a function of the separa-
tion distance h
s
. Thus, if the polishing pressure P
c
is known, the
separation distance h
s
can be calculated fromEq. (2).
2.2. The material removal of a single abrasive grain
As shown in Fig. 2, the abrasive grain is embedded between
the pad and the workpiece. R
abr
is the radius of the abrasive grain,
and z
w
and z
p
correspond to the interference between grain and
workpiece and that between grain and polishing pad, respectively.
The volume of the material removed by a single abrasive grain
is approximately equal to the product of its cross-sectional area
immersed in the workpiece and the sliding distance, which gives
zv = /L (3)
where A is the cross-sectional area and L is the sliding distance.
Assuming that the hard abrasive grain maintains its spherical
shape, A can be expressed by
/ = R
2
ubr
arcsin
u
R
ubr
u(R
ubr
z
w
), (4)
where a is the radius of the circular indentation on the workpiece
surface. Compared with the grain radius R
abr
, z
w
is very small, and
so is a/R
abr
, thus Eq. (4) can be further linearized about a/R
abr
as
/ uz
w
_
1 +O
_
z
w
R
ubr
__
, (5)
where a can be expressed as
u =
_
R
2
ubr
(R
ubr
z
w
)
2
=
_
2R
ubr
z
w
z
2
w

_
2R
ubrw
(6)
Assuming that the contacts between a grain and workpiece and
between the grain and polishing pad are plastic, the interference
z
w
is given by Xie and Bhushan (1996) as
z
w
=
H
p
H
w
+H
p
(2R
ubr
h
s
+h) (7)
Substituting Eqs. (6) and (7) into Eq. (5) yields
/ =
_
2R
ubr
_
H
p
H
w
+H
p
(2R
cbr
h
s
+h)
_
1.5
(8)
2.3. Number of abrasive grains participating in material removal
The number of the free-abrasive grains in the slurry directly
participating in the abrasive wear process is one of the most impor-
tant variables inuencing the material removal. Previous studies
assumed that the abrasive grains embedded in the workpiece sur-
face and the asperities of the polishing pad participate in the
material removal (Zhao and Chang, 2002; Jin and Zhang, 2012).
Thus, the number of active grains is the product of the area den-
sity of the abrasive grains in the slurry and the real area of contact
between polishing pad and workpiece. However, in the actual
polishing process, the abrasive grains are rarely on the contact
interface of the pad asperities and workpiece. Moreover, if the area
density of grains is xed, the MMR is independent of the abrasive
grain size in these models, which is unreasonable. In order to over-
come the deciency of these models, the critical condition for the
grains participating in material removal is described and derived.
Furthermore, the effects of the pad properties on the number of
active grains are considered, which is ignored by Xie and Bhushan
(1996).
Let the slurry concentration per unit volume be
d
, and the line
density of the grain in the slurry be l. Then
4
3
R
3
ubr
l
3
=
d
(9)
The relationship between area density and the line density l is
given by
l
2
. (10)
Rearranging Eq. (9) and substituting it into Eq. (10) gives
=
_
3
d
4R
3
ubr
_
2]3
(11)
When the applied load on the abrasive grain is small, the contact
between abrasive grain and workpiece is elastic. According to the
Hertizian contact theory, the indentation depth can be expressed
as
z
w
=
_
3p
4

_
2
R
ubr
, (12)
where p is the average pressure acting on the grain and

uw
is
the contact modulus of abrasive grain and workpiece given by
1]

uw
= (1 D
2
u
)]
u
+(1 D
2
w
)]
w
. E
a
and D
a
are Youngs modulus
and Poissons ratio of the grains, respectively. As the applied pres-
sure increases, the elastic contact will change into plastic contact,
and the critical value of the pressure p is determined by the hard-
ness of the workpiece as p=H
w
. Thus, the maximum interference
z
w0
of the elastic contact is given byD
z
w0
=
_
3H
w
4

_
2
R
ubr
. (13)
288 C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294
Fig. 3. Schematic of polishing along a curved path.
The material is removed from the workpiece surface in a plastic
manner. By substituting Eq. (13) into Eq. (7), the critical separation
distance h
e
leading to plastic contact can be expressed by
h
c
= h
s

_
2
H
w
+H
p
H
p
_
3H
w
4

_
2
_
R
ubr
. (14)
In the polishing process, the pad can only hold a certain portion
of abrasive grains in the slurry. The material of the workpiece is
assumed to be primarily removed by the abrasive grains attached
to the polishing pad with the height of the pad surface higher than
h
e
. As the surface height distribution density of the polishing pad is
(h) (see Eq. (1)), the number of abrasive grains per unit area at a
height between h and h+dh is (h). Thus, the number of grains
participating in the material removal can be expressed as
N
0
=
_
hmax
h
(h)dh (15)
where (h), and h
e
are expressed by Eqs. (1), (11) and (14), and
is the constant between 0 and 1 related the polishing pads ability
to hold the abrasive grains.
2.4. Material removal index
A contact patch in a certain shape is formed between the pol-
ishing pad and the workpiece when a normal load is applied on the
polishing pad and the material removal takes place within the con-
tact region. As shown in Fig. 3, point o is the center of the contact
patch on the path and H is an arbitrary point on the x axis with the
innitesimal area dxdy. The direction of dy is along the tool path.
For aninnitesimal time dt during whichthe tool is contacting with
H,
dt =
dy
v
u
, (16)
where v
a
is the feed rate of the tool moving along the tool path.
During the time interval dt, the real contact area between polishing
tool and H can be expressed as
S = v
s
dx dt =
v
s
v
u
dx dy (17)
where v
s
is the relative sliding velocity between the tool and H.
The number of grains per unit area participating in the material
removal can be calculated by Eq. (15). Then, the total number of
abrasive grains participating in the material removal of polishing
during the time interval dt is
N
1
= SN
0
=
v
s
dx dy
v
u
_
hmax
hc
(h) dh (18)
According to Eq. (8), the volume of material removed by a single
grain across H is
v
0
=
_
2R
ubr
_
H
p
H
w
+H
p
_
1.5
(2R
ubr
h
s
+h)
1.5
dy. (19)
Within the time interval dt, the overall volume of material
removal at H can be expressed as
v = dx dy dz = N
1
v
0
, (20)
where dz is the depth of material removal at H. Substituting Eqs.
(18) and (19) into Eq. (20) yields
w
l
=

_
2R
ubr
v
s

v
u
_
H
p
H
w
+H
p
_
1.5
_
hmax
h
2
(2R
ubr
h
s
+h)
1.5
(h)dh,
(21)
where
w
l

dz
dy
(22)
is the material removal index for free-abrasive polishing dened as
the depth of material removal per unit length along the polishing
path. Eq. (21) can be further modied as
w
l
=

_
2R
ubr
v
s

v
u
_
H
p
H
w
+H
p
_
1.5

__
hs
hc
(2R
ubr
h
s
+h)
1.5
(h)dh +
_
hmax
hs
(2R
ubr
)
1.5
(h) dh
_
(23)
3. Modeling the material removal prole
3.1. Contact pressure
It is assumed that process parameters, such as the normal pol-
ishing force, the angular spindle velocity, the feed rate and the
curvature of the tool path, vary slightly along the tool path, and
the curvature radius of the surface at the point o(P) is very large
compared with the size of the sub-aperture pad. Thus, the con-
tact region is approximately a circle with radius R
t
and the average
contact pressure can be expressed by
P
c
=
F
n
R
2
t
(24)
where F
n
is the normal force on the pad.
3.2. Relative sliding velocity
As shown in Fig. 4, the frame xoy is xed at the point P and the
plane xoy is the tangent plane of the surface at point P. The point
O is the geodesic curvature center of the path, and cO = R which
gives
R =
1
|k
g
|
, (25)
where k
g
is the geodesic curvature of the path at point P. The frame
XOY is set up at point O and the Y direction is parallel to the y
direction. For the convenience of analysis, an arbitrary point in the
frame xoy is described in the frame XOY by the polar coordinate
form, and the corresponding coordinate transformation is given by
_
x = cos +R
y = sin
(26)
C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294 289
Fig. 4. Distribution of relative velocity when polishing along a curved path.
For the arbitrary point M in the contact circle, the velocity due
to the tool rotation v
m
can be expressed as
v
m
= w
p
_
x
2
+y
2
= w
p
_

2
+2R cos +R
2
(27)
where w
p
is the angular spindle velocity. The velocity due to the
feed rate along the curved tool path is given by
v
u
= w
q
(28)
where w
q
is angular feed rate around the point O. The relative slid-
ing velocity v
s
in the contact region is the composition of v
m
and v
a
.
Thus, the relative sliding velocity at M(x, y) can be expressed by
v
s
=
_
v
2
u
+v
2
m
2v
u
v
m
cos( 0) =
_
w
2
q

2
+w
2
p
(
2
+2R cos +R
2
) 2w
q
w
p

2
+2R cos +R
2
cos( 0) (29)
3.3. Material removal prole orthogonal to the tool path
As shown in Fig. 4, when the polishing pad moves across point
H along the curved tool path, the contact length between the pad
and the workpiece at point H can be represented by the arc length
L
1
L
2
centered at O with radius . The removal depth at point H on
the surface can be calculated by integrating w
l
along the arc L
1
L
2
,
which gives
h() =
_

1

2
w
1
d (30)
where
1
and
2
correspond to the starting angle and the ending
angle of the arc L
1
L
2
. The contact circle can be expressed in the
frame xoy by
x
2
+y
2
= R
2
t
(31)
Substituting Eq. (26) into Eq. (31) yields
R
2
t
( cos +R
2
)
2
sin
2
= 0. (32)
And solution of cos can be obtained fromEq. (32) as
cos =
R
2
t

2
R
2
2R
. (33)
And there is one solution of within [/2, ] which gives

1
() = arccos
_
R
2
t

2
R
2
2R
_
(34)
As illustrated in Fig. 4, for R>R
t
, the bounding values of are

min
=RR
t
and
max
=R+R
t
. By substituting Eq. (34) into Eq. (30),
the material removal prole can be calculated as
h() = 2
_

1
()
0
w
1
d R R
t
R +R
t
. (35)
Substituting Eq. (11), (23) and (29) into Eq. (35) yields
h() =
2
_
2
w
p
R
1.5
ubr
_
3
d
4
_
2]3
_
H
p
H
w
+H
p
_
1.5
__
hs
hc
(2R
ubr
h
s
+h)
1.5
(h) dh +
_
hmax
hs
(2R
ubr
)
1.5
(h) dh
_

_

1
()
0
_
w
2
q

2
+w
2
p
(
2
+2R cos +R
2
) 2w
q
w
q

2
+2R cos +R
2
cos( 0) d (36)
In Eq. (36), h
s
can be solved by Eq. (2) and Eq. (24) and h
e
can be
calculated by Eq. (14). The detailed calculation procedure for the
material removal prole is shown in Fig. 5.
4. Experimental verication
4.1. Experimental conditions
To verify the effectiveness of the theoretical model for the
removal prole, a series of polishing experiments have been con-
ducted on a 5-axis polishing machine which is composed of three
290 C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294
Fig. 5. Calculation steps for material removal prole.
translational axes (X-axis, Y-axis and Z-axis) and two rotational
axes (B-axis and C-axis). The structure of the machine tool is shown
in Fig. 6a. As shown in Fig. 6b, the piezoelectric force sensor is
mounted on the end effector of the polishing spindle. It is used
as a feedback sensor to control the normal polishing force. The
workpiece is xed on the rotational table. The set-up enables the
polishing tool to move along a curved path with the normal pol-
ishing force varying from1N to 10N and a specied feed rate. The
Table 1
Polishing conditions.
Item Condition
Workpiece Material: Glass ceramics, Hw =8.3GPa, Ew =85.16GPa,
vw =0.272
Slurry Material: CeO
2
,
d
=0.5% and 0.2%, R
abr
=0.83mand
1.6m, Ea =249GPa, va =0.297, =0.2
Polishing tool Polishing pad: Hp =90Mpa, Ep =250MPa, vp =0.5,
Rp =0.03mm, Np =1.210
3
mm
2
, o =0.025mm,
hmax =0.58mm, Rt =4mm
polishing spindle is mounted on the B-axis, with the angular speed
adjustable from209rad/s to 2090rad/s. The material removal pro-
le is measured by FormTalysurf PGI 810 (Taylor-Hobson Ltd., UK)
with a 0.3mm tip radius stylus along a line perpendicular to the
polishing path. The measurements are conducted under the wavi-
ness mode to suppress the high-frequency noise of the proles, and
the measured data are illustrated fromFigs. 712.
The workpiece material used in the experiments is glass ceram-
ics. In order to get the physical properties of the workpiece, the
nano-indentation experiments are conducted, where H
w
=8.3GPa,
E
w
=85.16GPa and
w
=0.272. An elastic polishing tool which is
made of polyurethane is used as the sub-aperture polisher. The
radius of the polyurethane sub-aperture pad R
t
is xed as 4mm
and the mechanical properties are H
p
=90MPa, E
p
=250MPa and

p
=0.5 (Zhao and Chang, 2002). The topographical parameters
of the polyurethane sub-aperture pad are adopted according to
the experimental measurements of Yu et al. (1993) for typical
polyurethane pad surfaces, in which the average radius of asper-
ity R
p
=0.03mm, the asperity density N
p
=1.210
3
mm
2
and the
standard deviation of the distribution o =0.025mm. The value of
which describes the ability of the pad to hold abrasives is set
according to Xie and Bhushan (1996), in whose work =0.2 for the
pad with mediumhardness. The CeO
2
grains are used as free abra-
sives in the slurry. The mean diameters of the grains selected are
0.83m and 1.6m, and the grain volume concentrations of the
slurry are xed as 0.5% and 0.2%. For a CeO
2
grain in the slurry,
E
a
=249GPa and
a
=0.297, as indicated by Kanchana et al. (2006).
The detailed polishing conditions for the polishing experiments are
Fig. 6. Structure of experimental machine and polishing spindle. (a) Machine tool with ve axes and (b) polishing spindle with force control.
C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294 291
Fig. 7. Comparison of material removal proles for various normal polishing forces
(wp =209rad/s, wq =0.0015rad/s, R=6mm, R
abr
=0.83m,
d
=0.5%). (a) Experi-
mental proles vs. simulation proles and (b) theoretical proles.
summarized in Table 1. The planning of process parameters in the
experiments is shown in Table 2.
4.2. Results and discussion
The mechanism of the FAP process is complicated and the
inuencing factors of the material removal depth involve the pro-
cess parameters, properties of abrasive grains/slurry and polishing
sub-aperture pad, and tool path curvature. Figs. 712 show the
proles of the material removal in different polishing conditions.
The material removal proles show good agreement between the
experimentally measured results and the simulation results calcu-
lated by Eq. (36). The removal proles are not symmetrical, as can
be seen in both experiments and simulation results, which show
deviation toward the center of the tool path curvature. This devia-
tion is attributed to the longer dwelling time in the contact region
near the curvature center of the tool path when polishing along a
curved path.
The experimental material removal proles of the work-
piece surface for various normal polishing forces F
n
are shown
Table 2
Process parameters in the experiments.
No. Fn (N) wp (rad/s) wp (rad/s) R (mm) R
abr
(m)
d
(%)
1 1.5 209 0.0015 6 0.83 0.5
2 6 209 0.0015 6 0.83 0.5
3 1.5 418 0.0015 6 0.83 0.5
4 1.5 209 0.0003 6 0.83 0.5
5 1.5 209 0.0015 5 0.83 0.5
6 1.5 209 0.0015 6 1.60 0.5
7 1.5 209 0.0015 6 0.83 0.2
Fig. 8. Comparison of material removal proles for various angular spindle
velocities (Fn =1.5N, wq =0.0015rad/s, R=6mm, R
abr
=0.83m,
d
=0.5%). (a)
Experimental proles vs. simulation proles and (b) theoretical proles.
in Fig. 7. Other experimental settings include w
p
=209rad/s,
w
q
=0.0015rad/s, R=6mm, R
abr
=0.83m and
d
=0.5%. Fig. 7a
shows the removal proles obtained both experimentally and the-
oretically for F
n
=1.5 and 6N, and Fig. 7b shows simulation results
of the proles for F
n
varying from1.5N to 6N. Both the simulation
and experimental results showthat the material polished increases
as F
n
grows. The effect of the normal polishing force F
n
is reected
in the separation distance h
s
. A large F
n
leads to a small h
s
, which
causes more material to be removed by more abrasive grains par-
ticipating in the process. According to the experimental results and
theoretical predictions in this paper, the removal depth is approx-
imately proportional to (F
n
)
0.65
. This result is consistent with the
experimental nding by Shi and Zhao (1998). Their measurements
indicate that the material removal rate is proportional to (F
n
)
2/3
.
The angular spindle velocity w
p
also plays an important role
in the removal prole. In Fig. 8a, the experimental settings are
F
n
=1.5N, R=6mm, w
q
=0.0015mm/s, R
abr
=0.83m,
d
=0.5%, and
w
p
=209rad/s or 418rad/s. The comparison of material removal
proles also conrms that the experimental proles are consis-
tent with the theoretical ones. The theoretical material removal
prole is calculated by varying the w
p
from209rad/s to 523rad/s.
These experimental and theoretical results indicate that the mate-
rial removal depth increases as the angular spindle velocity w
p
increases. The effect of the angular feed rate w
q
on the mate-
rial removal prole is shown in Fig. 9. And polishing settings are
F
n
=1.5N, R=6mm, w
p
=209rad/s, R
abr
=0.83m,
d
=0.5% and
w
q
=0.0015rad/s or 0.0003rad/s. Both the experimental and theo-
retical results in Fig. 9 indicate that the removal depth decreases as
a result of the increasing angular feed velocity w
q
. A smaller angu-
lar feed rate leads to longer dwelling time at the points along the
path, resulting in a larger amount of material removed from the
workpiece surface. In the real polishing process, w
p
>>w
q
. Thus,
292 C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294
Fig. 9. Comparison of material removal proles for various angular feed rates
(Fn =1.5N, wp =209rad/s, R=6mm, R
abr
=0.83m,
d
=0.5%). (a) Experimental pro-
les vs. simulation proles and (b) theoretical proles.
according to Eq. (36), the polished depth is proportional to w
p
,
and inversely proportional to w
q
, which is also conrmed by the
experimental results in Figs. 8 and 9.
Fig. 10 shows the inuence of the curvature radius of the polish-
ing path R on the removal prole. In the experiments, the curvature
radius R varies from5mmto 6mm, while the other process param-
eters are xed as F
n
=1.5N, w
p
=209rad/s, w
q
=0.0015mm/s, R
abr
=0.83m and
d
=0.5%. As shown in Fig. 10a, when the radius
R decreases from 6mm to 5mm, the material removal depths
increase from5.93mto 9.21mby experimental data, and from
5.76mto8.16mintheoretical calculation. Thetheoretical mate-
rial removal proles (see Fig. 10b) also indicate that a smaller
curvature radius of the tool pathcorresponds to deeper removal. As
indicated in Eq. (28), the feed velocity is the product of the angular
feed velocity and the radius of the polishing path. As the angular
feed velocity is kept constant, the feed velocity increases with the
increasing of the radius of the polishing path. For the sub-aperture
pad polishing along a tool path, a large feed rate leads to the reduc-
tion of processing time for a certain point on the tool path, which
results in a reduction in the material removal depth. As shown in
Fig. 4, the proposed model is developed when the curvature center
O outside the contact circle. Thus, the boundary condition of the
proposed model for the polishing path curvature is R>R
t
.
According to Eq. (36), the material removal depth is inuenced
by the radius of grains in the slurry. Experimental settings include
F
n
=1.5N, R=6mm, w
q
=209rad/s, w
q
=0.0015rad/s,
d
=0.5% and
R
abr
=0.83m or 1.6m. Fig. 11a shows the material removal
proles obtained both experimentally and theoretically for R
abr
=0.83mand 1.6m, and the material removal proles are consis-
tent withthose calculatedbyEq. (36). Fig. 11bshows the theoretical
material removal proles as R
abr
varies from 0.4m to 3.2m,
Fig. 10. Comparison of material removal proles for various tool path curvatures
(Fn =1.5N, wp =209rad/s, wq =0.0015rad/s, R
abr
=0.83m,
d
=0.5%). (a) Experi-
mental proles vs. simulation proles and (b) theoretical proles.
which indicates the removal depth increases with the increasing
R
abr
. According to simulation results in Fig. 13, the removal depth is
approximately proportional to the mean radius of abrasive grains
R
abr
. This relation holds as it is the abrasive grains embedded in
the pad/workpiece that abrade the surface. With the volume con-
centration xed at a certain value, if the radius of abrasive grains
is small, there are more grains in per unit volume of the slurry.
However, a critical condition needs to be satised for the grains
participating in the material removal, as indicated in Eq. (19). Thus,
although there are more abrasive grains in the slurry, only a small
portion of them have the ability to remove the material. On the
other hand, although a big radius of abrasive grains corresponds
to a small number of abrasive grains in the slurry, a large portion
of abrasive grains satisfy the critical condition of material removal.
Furthermore, the volume of material removed by a large grain is
bigger than that by a small grain.
As shown in Figs. 12 and 13, the volume concentration of the
slurry can also inuence the material removal. Moreover, accord-
ing to Eq. (36), the depth of the removal prole is proportional to
the (volume concentration)
2/3
. The experimental results for various
volume concentration,
d
=0.5% and 0.2%, are shown in Fig. 12a,
which are consistent with the theoretical ones.
The polishing pad is one of the most signicant components in
the polishing system. The model developed in this paper reveals
some insights into the mechanism of FAP in terms of the work-
ing principle of a polishing pad. In this study, two parameters of
the polishing pad are discussed, namely, the standard deviation of
pad asperity height o and the average radius of pad asperities R
p
.
The effects of o and R
p
on the removal depth are shown in Fig. 14.
The standard deviation of pad asperity height o is a description of
the pad roughness. A smaller o means a smoother pad, which is
C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294 293
Fig. 11. Comparison of material removal proles for various mean radii of abrasive
grains (Fn =1.5N, wp =209rad/s, wq =0.0015rad/s, R=6mm,
d
=0.5%). (a) Experi-
mental proles vs. simulation proles and (b) theoretical proles.
capable of holding a greater number of abrasive grains. Thus it can
be observed from Fig. 14 that the removal depth h decreases as o
increases, which is similar as described in the work on material
removal rate by Jin and Zhang (2012). A similar nding is also pre-
sented by Xie and Bhushan (1996), which states that the material
removal rate in CMP is proportional to the (o)
0.3
. Compared with
o, the effect of R
p
on the removal depth is much smaller, as shown
in Fig. 14. As R
p
increases from 0.01mm to 0.1mm, the removal
depth only decreases a little.
Fig. 12. Comparison of material removal proles for various volume concentrations
(Fn =1.5N, wp =209rad/s, wq =0.0015rad/s, R=6mm, R
abr
=0.83m). (a) Experi-
mental proles vs. simulation proles and (b) theoretical proles.
The proposed model assumed that the abrasive grains are held
by the polishing pad, and the GWmodel was used to characterize
the contact between pad and workpiece, thus it cannot be used
to predict the material removal of the traditional hard polishing
disk, such as pitch disk. Another limitation is that the model and
the planning of the experiments in this paper ignored the chemical
Fig. 13. Effects of the properties of slurry on the removal depth.
294 C. Fan et al. / Journal of Materials Processing Technology 214 (2014) 285294
Fig. 14. Effects of the properties of polishing pad on the removal.
effect of the polishing slurry on the material removal, which should
be further discussed.
5. Conclusions
Inthis paper, the material removal prole is usedtocharacterize
the material removal during the FAP process with the sub-aperture
pad. A novel mathematical model of the material removal prole
for the FAP process is developed, which successfully explains the
effects of the properties of abrasive grains/slurry and topograph-
ical parameters of sub-aperture pad on the prole. According to
the model, the removal depth is proportional to (normal polishing
force)
0.65
, the angular spindle veloctiy (volume concentration)
2/3
and the radius of abrasive grains, and inversely proportional to the
angular feed rate. Moreover, the removal depth decreases with the
increasing deviation of pad asperity height and curvature radius of
the path.
Polishing experiments are conducted in which the normal pol-
ishing force, tool path curvature, angular spindle velocity, angular
feed rate, abrasive grain size and volume concentration are varied.
The experimental results agree with the theoretical calculations,
which conrms the validity of the model in predicting the mate-
rial removal prole and depth for the FAP process. The proposed
model canbe usedto optimize the process parameters andimprove
the stability and determinacy of the polishing process. By control-
ling the inuential parameters during the FAP process, the polished
depth can be controlled to improve the formaccuracy of the work-
piece surface.
Acknowledgements
The authors are grateful to the nancial support from Chinese
National Program on Key Basic Research Project (973 Program)
[Grant No. 2011CB706702] and Innovation Program of Jilin Uni-
versity for graduate student [Grant No. 20121078].
References
Brinksmeier, E., Riemer, O., Gessenharter, A., 2006. Finishing of structured surfaces
by abrasive polishing. Precision Engineering 30, 325336.
Cheung, C.F., Kong, L.B., Ho, L.T., To, S., 2011. Modeling and simulation of struc-
ture surface generation using computer controlled ultra-precision polishing.
Precision Engineering 35, 574590.
Cheng, H.B., Feng, Z.J., Cheng, K., Wang, Y.W., 2005. Design of a six-axis high preci-
sion machine tool and its application in machining aspherical optical mirrors.
International Journal of Machine Tools & Manufacture 45, 10851094.
Greenwood, J.A., Williamson, J.B.P., 1966. Contact of nominally at surfaces.
Proceedings of the Royal Society of London. Series A. Mathematical and Physical
Sciences 295 (1442), 300319.
Jin, X.L., Zhang, L.C., 2012. A statistical model for material removal prediction in
polishing. Wear 274275, 203211.
Jones, R.A., 1977. Optimization of computer controlled polishing. Applied Optics 16,
218244.
Kanchana, V., Vaitheeswaran, G., Svane, A., Delin, A., 2006. First-principles study of
elastic properties of CeO
2
ThO
2
and PoO
2
. Journal of Physics: Condensed Matter
18, 96159624.
Kim, D.W., Kim, S.W., 2005. Static tool inuence function for fabrication simulation
of hexagonal mirror segments for extremely large telescopes. Optics Express 13,
910917.
Klocke, F., Zunke, R., 2009. Removal mechanisms in polishing of silicon
based advanced ceramics. CIRP Annals Manufacturing Technology 58,
491494.
Lee, H.S., Jeong, H.D., Dornfeld, D.A., 2013. Semi-empirical material removal rate
distribution model for SiO
2
chemical mechanical polishing (CMP) processes.
Precision Engineering 37, 483490.
Lin, T.R., 2007. An analytical model of the material removal rate between elastic
and elastic-plastic deformation for a polishing process. International Journal of
Advanced Manufacturing Technology 32 (78), 675681.
Preston, F.W., 1927. The theory and design of plate glass nishing machines. Journal
of the SOC of Glass Technology 11, 214256.
Roswell, A., Xi, F.J., Liu, G., 2006. Modelling and analysis of contact stress for auto-
matedpolishing. International Journal of Machine Tools andManufacture 46 (3),
424435.
Shi, F.G., Zhao, B., 1998. Modeling of chemicalmechanical polishing with soft pads.
Applied Physics 67, 249252.
Su, Y.T., Horng, C.C., Sheen, J.Y., Hsiau, J.S., 1996. A process planning strategy for
removing an arbitrary prole by hydrodynamic polishing process. International
Journal of Machine Tools & Manufacture 36 (I1), 12271245.
Su, Y.T., Sheen, J.Y., 1999. Aprocess planning strategy for removing arbitrary andaxi-
ally symmetric prole by a polishing process. International Journal of Machine
Tools & Manufacture 39, 187207.
Xie, Y.S., Bhushan, B., 1996. Effects of particlesize, polishingpadandcontact pressure
in free abrasive polishing. Wear 200, 281295.
Yang, M.Y., Lee, H.C., 2001. Local material removal mechanism considering curva-
ture effect in the polishing process of the small aspherical lens die. Journal of
Materials Processing Technology 116, 298304.
Yu, Y., Yu, C.C., Orlowski, M., 1993. A statistical polishing pad model for
chemicalmechanical polishing. In: December 58, IEEE EDM. Washington, DC,
pp. 865868.
Zhao, Y.W., Chang, L., 2002. A micro-contact and wear model for
chemicalmechanical polishing of silicon wafers. Wear 252, 220226.

Das könnte Ihnen auch gefallen