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

net/publication/224448807

A simulation model for thickness profile of the film deposited using planar
circular type magnetron sputtering sources

Article  in  Journal of Vacuum Science & Technology A Vacuum Surfaces and Films · October 1996
DOI: 10.1116/1.580193 · Source: IEEE Xplore

CITATIONS READS

12 168

7 authors, including:

Seungbum Hong Eunah Kim

Korea Advanced Institute of Science and Technology Hanyang University
193 PUBLICATIONS   3,509 CITATIONS    25 PUBLICATIONS   578 CITATIONS   

SEE PROFILE SEE PROFILE

Byeong-Soo Bae KS No
Korea Advanced Institute of Science and Technology Korea Advanced Institute of Science and Technology
272 PUBLICATIONS   4,868 CITATIONS    303 PUBLICATIONS   4,221 CITATIONS   

SEE PROFILE SEE PROFILE

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

Lithium-oxygen batteries View project

Workforce Development Program by MKE & KIAT and Basic Science Reserch Program by NRF View project

All content following this page was uploaded by Seungbum Hong on 29 May 2014.

The user has requested enhancement of the downloaded file.

A simulation model for thickness profile of the film deposited using planar
circular type magnetron sputtering sources
Seungbum Hong,a) Eunah Kim, Byeong-Soo Bae, and Kwangsoo No
Department of Material Science and Engineering, KAIST, Kuseong-dong, Yusong-gu, Taejon, South Korea
Sung-Chul Lim, Sang-Gyun Woo, and Young-Bum Koh
Samsung Electronics Co., Ltd., Yongin-gun, Kyongki-do, South Korea
~Received 31 October 1995; accepted 21 June 1996!
The thickness profile of the film deposited by planar circular-type magnetron is simulated
considering the relationship between magnetic field profile and target erosion rate. The model is
confirmed by the measurement of the film thickness profiles of the Cr films deposited in this study.
It is found that the model is applicable to the magnetron sputtering process at low gas pressure.
Furthermore, it can be used to predict the thickness profile of the films deposited by magnetron
sputtering with various shapes of magnets. © 1996 American Vacuum Society.

I. INTRODUCTION gime that occurs after thermalization. Fan et al.3 and Swann
et al.4 assumed no gas scattering, cosine emitting source, and
Recently, magnetron sputtering has been widely used in calculated the erosion rate from the erosion depth profile.
thin film fabrication since it offers high deposition rate and Most of the authors have calculated the target erosion rate
operates at lower pressure than the conventional one without from the target erosion depth profile. Though it has been
magnetron. The manifestation of the relationship between mentioned by many authors that the target erosion rate is
the main film quality ~mechanical, electrical, optical proper- dependent on the magnetic field profile, there is almost no
ties, etc.! and the deposition parameters ~total chamber pres- quantitative work on their relationship. Therefore, in this
sure, gas flow rate, target-substrate distance, etc.! is required study the role of the magnetic field profile and its relation-
to control the process more efficiently. Especially, among the ship with the erosion rate has been considered in simulating
main film quality, the thickness uniformity of the film is a the thickness profile of the film deposited by the planar
critical factor in optical applications because it is directly circular-type magnetron sputtering system. Then, the validity
relevant to the optical properties of the film such as transmit- of the model was confirmed by the measurement of the Cr
tance, absorption, and refractive index, etc. film thickness profiles after sputter deposition.
Much theoretical and experimental research has been per-
formed to investigate the film thickness distribution.1–4 He- II. MODEL DEVELOPMENT
berlein et al.1 used the Monte-Carlo method to simulate the In this study, several assumptions have been taken to sim-
film thickness, as well as angular and energy distributions. plify the model. These are ~1! no gas phase collisions, ~2!
They divided the magnetron sputtering process into three unit sticking coefficient, ~3! cosine-emitting source,5 and ~4!
parts; particle emission from target, gas scattering, and film linear proportionality between loop current density and etch
growth. Furthermore, they took the cosine distribution func- rate.
tion for emission characteristics, Thompson’s relation for
emission probability, erosion depth profile for erosion rate, A. Cosine-emitting source and plane parallel receiver
elastic gas scattering between sputtered particles and back- (Ref. 6)
ground gas particles, and particle losses by chamber wall Let F 0 denote the flux of the sputtered atoms from the
deposition or thermalization. Fursenko et al.2 also used the source per steradian in the direction perpendicular to the
Monte-Carlo method, but they added some hydrodynamic source plane. It is assumed that Knudsen’s cosine law of
concept into their simulation. They took the emission char- evaporation is valid for each elemental area, DA S , of a pla-
acteristics to be a cosine-emitting distribution function and nar emitting surface and that the source is planar, circular,
assumed the background gas to be a continuum media with and uniform. Then the flux of sputtered atoms from the area
hyposonic flow, focusing on the transport of gases at low of the source element, DA S , to the area of the receiver ele-
pressure and at medium pressure. In the low-pressure region, ment, DA R , should be
the linear Boltzmann equation and the Poisson distribution
F 0 DA R d 2
were used to describe the transport process of the particles dDH5 3rd f dr ~ mol/s! ,
and the collision between the Monte Carlo test particles and ~ r 2 22rx cos f 1x 2 1d 2 ! 2
background gases. In the medium pressure region, the test ~1!
particle Monte Carlo method was utilized for the kinetic re- where DH is the total flux from the source to the area of the
gime and convective diffusion equation for the diffusion re- receiver element DA R , r is the radial distance of the source
element from the origin on its plane, x is the projected length
a!
Electronic mail: [email protected] to the x axis of the line connecting the origin on the source

2721 J. Vac. Sci. Technol. A 14(5), Sep/Oct 1996 0734-2101/96/14(5)/2721/7/$10.00 ©1996 American Vacuum Society 2721
2722 Hong et al.: A simulation model for thickness profile 2722

The loop current density, as defined in Eq. ~2!, is linear in

ion/electron density, which in turn shows a linear relation-
ship with the local sputter rate, thus the loop current density
is proportional to the etch rate. Current density along the y
axis in the configuration shown in Fig. 2 can be expressed as
the following:

J y 5n e e v y , ~2!

where n e is electron concentration, e is the electron charge,

and v y is the electron velocity along the y axis. In this case,
the source plane is taken to be the xy plane, and the surface
normal of the source plane to be the z direction. The hopping
direction of the electrons is assumed to be parallel to the y
FIG. 1. Plane source with parallel, plane receiver ~see Ref. 6!. axis. We also assume that the electric field in the dark space
decreases linearly across the dark space thickness, L.7 Let-
ting the target surface be z50 gives

S D
plane and the central point of the receiver element, d is the
source to receiver distance, and f is the angular coordinate z
E z 52E 0 12 , ~3!
of the source element6 ~Fig. 1!. L

where E 0 is the absolute value of the field at the target.

B. Nonuniform source and plane parallel receiver Moreover, if we assume that the electron emission velocity
In Eq. ~1!, it was assumed that a uniform source is at the surface of the source plane is zero, the equation of
present, but modification of Eq. ~1! is necessary for the non- motion becomes
uniform magnetron sputtering system. The coexistence of the
dz
magnetic field and the local electric field causes electrons to 2 eB x
hop around the source plane, which induces the local current d y dt
52 , ~4!
of electrons, as shown in Fig. 2. This is the reason that a high dt 2 me
plasma density region is formed in a loop shape on the
source plane. The plasma density in this region called etch which yields
track is so high that the etch rate caused by only the electric
field can be ignored in comparison to that done by the com- dy eB x z
5 v y 52 . ~5!
bination of the electric field and the magnetic field. dt me

Considered the electron motion along the z axis

d 2z dy
me 2 52eE z 1eB x . ~6!
dt dt

Solving the differentiated equation in combination with Eqs.

~3!, ~4!, ~5!, and ~6!, we may solve the velocity

e 2E 0B x
v y 52 ~ 12cos v t ! , ~7!
m 2e v 2

where

eE 0 e 2 B 2x
v 25 1 2 . ~8!
m eL me

In this study, the maximum B field at the surface of the

source plane was 176.5 G. The dark space thickness at the
condition of discharge voltage of 500 V and total chamber
pressure of 1.3 Pa was 0.376 mm.8 Using the Langmuir
probe, the plasma potential was measured to be 10 V. There-
FIG. 2. Schematic diagram of ~a! planar circular magnetron and ~b! its cross- fore, under this condition, the ratio of the second term to the
sectional view along A-A 8. first term in Eq. ~8! is given by

J. Vac. Sci. Technol. A, Vol. 14, No. 5, Sep/Oct 1996

2723 Hong et al.: A simulation model for thickness profile 2723

e 2 B 2x
m 2e LeB 2 L 2 eB 2
ratio5 5 5
eE 0 m e E 0 m e ~ V1V p !
m eL
~ 0.37631023 ! 2 31.63102193 ~ 1.76531022 ! 2
5
9.113102313 ~ 500110!
50.0152.
Since the value of the second term is found to be about 1.5%
of that of the first term, the second term can be ignored in
Eq. ~8!. This makes v to be constant independent of the B
field. Therefore, the time-averaged mean velocity becomes
e 2E 0B x
v y 52 5aBx , ~9!
m 2e v 2
where a proportional constant, a52e 2 E 0 /m 2e v 2 .
Thus, the mean current density becomes
J y 5n e e v y 5n e e a B x 5 b B x , ~10! FIG. 3. Distribution profile of x axis component of the magnetic field.
where a proportional constant, b5n e e a .
According to the assumption that the etching rate of the
source plane is proportional to the loop current density, it is A combination of Eqs. ~13! and ~14! will yield the following
found that the etching rate of the source plane increases lin- equation:
early with the magnetic field component being parallel to the DH
source plane. In the case of the planar and circular magne- 522 p
x DA R d 2
tron sputtering system, it can be inferred that
J u5 b B r ~11! 3 E 22.5

0
1.3031024 ~ r 2 222.5r !~ r 2 1x 2 1d 2 ! rdr
@ r 4 22r 2 ~ x 2 2d 2 ! 1 ~ x 2 1d 2 ! 2 # 3/2
.
and the modified form of Eq. ~1! is given by
~15!
B r DA R d 2
dDH5 x 2 3rd f dr, ~12! This can be divided into two parts
~ r 22rx cos f 1x 2 1d 2 ! 2
where x is a proportional constant.
DH
x DA R d 2
522 p E 0
22.5 1.3031024 r 2 ~ r 2 1x 2 1d 2 ! rdr
@ r 4 22r 2 ~ x 2 2d 2 ! 1 ~ x 2 1d 2 ! 2 # 3/2

pE
C. Magnetic field measurement and polynomial 22.5 1.3031024 322.5r ~ r 2 1x 2 1d 2 ! rdr
approximation of the magnetic field 12 .
0 @ r 4 22r 2 ~ x 2 2d 2 ! 1 ~ x 2 1d 2 ! 2 # 3/2
In this study, magnetic field was measured using a Gauss-
~16!
meter ~Fig. 3!. We have adopted the polynomial approxima-
tion method to fit the radial component of the magnetic field Letting j5r , the first term can be integrated in two parts9 to
2

H
at the surface of the source. Thus, the approximated qua- give
j ~ j 2C ! 1
dratic function for the region of interest is 21.3031024 p 1/22
~ C2D ! R ~ C2D !
B r 521.3031024 3 ~ r 2 222.5r ! ~ Tesla! , ~13!
as shown in Fig. 3. 3 @ AR1 ~ D2C ! ln~ 2 AR12 j 22D !# J 22.52

0
, ~17!

D. Indefinite integral and finite elemental method where C is x 2 1d 2 , D is x 2 2d 2 , and R is j22 j D1C 2.
The circular symmetry of the source makes the evaluation The second term in Eq. ~16! is found to be not solved by an
along one projected source radius describing completely the analytical method. Therefore, we used a finite element
distribution over the receiver surface. Thus, Eq. ~12! is suit- method to evaluate the second term.
able for our purpose. The integration of Eq. ~12! over f from
0 to 2p, exclusive of the coefficient x DA R d 2 , gives9 III. EXPERIMENT
DH
x DA R d 2
52 p E B r ~ r 2 1x 2 1d 2 ! rdr
@ r 4 22r 2 ~ x 2 2d 2 ! 1 ~ x 2 1d 2 ! 2 # 3/2
.
Cr films were prepared using a planar circular type mag-
netron sputtering system with a ring-shaped 2 in. target. The
~14! outside diameter of the Cr target was 50.8 mm and the inside

JVST A - Vacuum, Surfaces, and Films

2724 Hong et al.: A simulation model for thickness profile 2724

FIG. 4. Schematic diagram of the sputtering system.

diameter was 45.0 mm with the thickness of 6.2 mm ~Fig. 4!.

The films were deposited at the condition of discharge volt-
age of 500 V, discharge current of 0.1 A, and total chamber
pressure of 1.3 Pa. The reason the total chamber pressure
was chosen to be 1.3 Pa is that the lowest total chamber
pressure is required to minimize the collision effect of the
gas phase and it was found to be 1.3 Pa in our system. The
target to substrate distance was varied with 50, 80, and 100
mm. The deposition was conducted at the ambient tempera-
ture. The thickness profiles of the films were measured by
Tencor alpha-step instrument. Four data points were taken
and averaged for each sampling point to provide credibility
of the measured data. Moreover, all the films were deposited FIG. 5. Simulated and measured thickness profile for different target to
in sufficient time to yield the thickness of over 5000 Å be- substrate distance of ~a! 50 mm, ~b! 80 mm, and ~c! 100 mm.
cause the variation of thickness was usually under 50 Å. The
measured thickness profile was normalized with the highest
thickness value to be compared with the simulated profile. condition should be large to ensure the assumption. To
evaluate the Knudsen number of each condition the mean
IV. RESULTS AND DISCUSSIONS free path for sputtered Cr atoms must be calculated. From the
well known equation for the mean free path,11 one can cal-
Uniformity of the film thickness over an area as large as culate it if the collision cross section and the background gas
possible is a basic demand in thin film technology. One of density are given. Consequently, taking the gas phase colli-
the major factors which influence the uniformity is the varia- sion as being elastic, we will calculate the elastic cross sec-
tion of the target-substrate distance.6 The assumptions such tion and the background gas density to evaluate the mean
as no gas phase collisions, unit sticking coefficient, cosine- free path of sputtered Cr atoms. Moreover, we will present
emitting source, and linear proportionality between loop cur- the pressure ranges in terms of Knudsen number so that the
rent density and target erosion rate were taken into account. model can be applied more reliably.
While deriving the film thickness distribution equation, we
correlated the magnetic field and the erosion rate quantita-
1. Calculation of the elastic cross section
tively. This enabled us to predict the film thickness profile of
magnetron sputtering equipped with various shapes of mag- Assuming Thompson distribution12 of the sputtered Cr at-
nets. oms, the average sputtered energy is calculated to be 11.93
eV with the surface binding energy of 4.2 eV.12 This value of
A. Model verification and applicability
average sputtered energy for Cr is very close to 11 eV mea-
The verification of the model can be done by comparing sured by Thornton et al.13 Since the elastic cross section var-
the experimental data and theoretical values of thickness pro- ies with the log scale of the sputtered energy,14 a relatively
file at some different conditions. Since no gas phase scatter- rough estimation would be sufficient to calculate the cross
ing is assumed in the model, the Knudsen number2,10 of each section. With a similar approach performed by Rossnagel,15

J. Vac. Sci. Technol. A, Vol. 14, No. 5, Sep/Oct 1996

2725 Hong et al.: A simulation model for thickness profile 2725

FIG. 6. The absolute deviation of the predicted values from the measured
data vs. Knudsen number. R, the relative flux, d, the maximum absolute
deviation among sampling points, h, the average absolute deviation over
the sampling points.

FIG. 7. Schematic diagram of ~a! rectangular magnetron sputtering configu-

the cross section for Cr atoms of 11.93 eV is found to be ration and ~b! its overview with magnified pictures of triangular regions of
~c! segment II and ~d! segment III.
6.0310216 cm2 from the work of Robinson.14

2. Calculation of the background gas density for verification and application of our model. However, it
Before calculating the background gas density, the gas was very difficult either to lower the pressure below 1.3 Pa
rarefaction effect should be taken into account. While the or to reduce the substrate-target distance below 5 cm in our
magnetron discharge current of 0.1 A is not expected to con- system so that there was no method to perform experiments
tribute to a significant change in the background gas density, where the Knudsen number reaches far beyond 1. Recently,
it is found to reduce the background gas density to 83% of there have been developed many magnetron sputtering de-
the original value at 1.3 Pa for Cu sputtering in Ar.15 Con- vices which operate at very low pressure ~less than 0.1 Pa!17
sidering the average sputtered energy, momentum transfer and this model may be more suitable for such equipment.
cross section,14 and sputtering yield of Cr,16 we can evaluate Nonetheless, though not perfectly matched, the simulated
the reduction factor of the background gas density to be 0.90 profile was matched to the experimental results quite well, as
from the work of Rossnagel,15 which is a bit higher value shown in Fig. 5. It is found that as the target to substrate
than that of Cu at the condition of interest. It is equivalent to distance increases, the discrepancy between the experimental
say that the background gas density is reduced to and the simulated values increases. This trend is shown in
331014/cm3 from 3.331014/cm3 at 1.3 Pa. Fig. 6 in terms of the deviation versus the Knudsen number.
The deviations between the predicted values and the experi-
3. Verification and applicability of the model mental data showed a linear dependence with the Knudsen
number. Moreover, the points far from the center are found
It is found that the mean free path for 11.93 eV Cr atoms to have more deviated values from the simulated ones. These
in Ar at 1.3 Pa with 0.1 A of magnetron discharge current is may be attributed to the fact that the model is less valid as
about 5.6 cm from the above calculations. This value is com- the collision frequency increases. If we take 3% as a toler-
parable to 4.5 cm at 1.3 Pa ~inferred from the 1.5 cm at 4 Pa! able limit for maximum deviation in the film thickness pro-
for Cu atoms calculated by Rossnagel15 without considering file, it is found that the Knudsen number should be at least 1
the gas rarefaction effect ~it will increase up to 5.4 cm if the for application of this model as is expected from the above
gas rarefaction is taken into account!. Eventually, the Knud- discussion.
sen number will vary from 0.56 to 1.12 with target-substrate
distance variation from 10 to 5 cm at the above given depo-
B. Application for other magnetron geometry
sition condition.
A large Knudsen number over 1 which ensures molecular As mentioned above, this model may be applied not only
flow of the background gas would be the primary condition to circular magnetron systems but also to noncircular mag-

JVST A - Vacuum, Surfaces, and Films

2726 Hong et al.: A simulation model for thickness profile 2726

netron systems such as rectangular ones. Rectangular mag- integration in the case of segments I and III. However, only
netron can be thought of as a combination of linear magne- the light gray triangle in Fig. 7~b! will be discussed in detail,
trons of parallel-type and perpendicular-type with respect to and then the total flux from segments I and III will be given
the sampling direction of the thickness profile ~Fig. 7!. The without further details. First of all, the upper limit for r in the
deposition profile can be obtained by summing the fluxes integration of Eq. ~12! is a function of f and is equal to
contributed by each segment divided, as shown in Fig. 7. b/cos f, hence the limits of the integration over r is 0 to
Segments I and III represent the areas where the magnetic b/cos f. Also, it can be easily shown that the limits for f in
field is a function of both x and y coordinates of the point of the integration is 0 to p/4 from Fig. 7~d!. Secondly, the x
interest, while segment II represents the area where it is an axis is shifted by a value of ‘‘a,’’ thus the ‘‘x’’ in Eq. ~12!
only function of its y coordinate. The dimension of each should be replaced by ‘‘x2a.’’ However, the x axis shift is
segment can be given, as shown in Fig. 7~a!, because the in the reverse direction for segment I and, accordingly, the
width of the open space @denoted by arrows in Fig. 7~a!# ‘‘x’’ should be replaced by ‘‘x1a.’’ Thirdly, assuming that
between magnetrons is usually the same. the maximum value of the B field, B max , in the r axis is
Since the magnetrons beneath the target have a mirror constant with a small variation of the distance between the
symmetry with respect to the x axis, the total flux from the ‘‘N’’ pole and ‘‘S’’ pole of the magnetrons, the equation for
target can be calculated by doubling the flux from the upper the B r will be
half part of the target @Fig. 7~b!#. The flux from each segment
will be treated in detail, and the equation for the total flux
will be given in the form of an integral which necessitates
the use of the finite elemental method. B r 52 ~ const! r r2 S b
cos f
. D ~18!

1. The sputtered flux from segments I and III in Fig. 7

It is necessary to divide the area of integration into two Therefore, the flux from the light gray triangular component
triangular regions to use the circular coordinate ~r,f! for in Fig. 7 will be

DH5 E E
p /4 b/cos f
x 2
r r2S
b
cos f
d2 D rdrd f . ~19!
0 0 @ r 22r ~ x2a ! cos f 1 ~ x2a ! 2 1d 2 # 2

If we denote the denominator of the integrand in Eq. ~12! as f (r,x,d, f ) then the total flux from segments I and III will be

DH5DH I1DH III52 x FE E H

p /4

0
b/cos f

0
S
r r2
b
cos f D
d2

f ~ r,x1a,d, f !
J rdrd f 1 E E
p /4
p /2 b/sin f

0
H S
r r2
b
sin f
d2

f ~ r,x1a,d, f !
D J rdrd f

1 E E
0
p /4
H Sb/cos f

0
r r2
b
cos fD d2

f ~ r,x2a,d, f !
J rdrd f 1 E E p /2

p /4
b/sin f

0
H S
r r2
b
sin f
d2

f ~ r,x2a,d, f !
D J G
rdrd f . ~20!

2. The sputtered flux from segment II in Fig. 7 Hence, the flux from the dark triangular region in Fig. 7 is
The derivation of the equation for the sputtered flux from
segment II in Fig. 7 will be done in the similar way to that of
Sec. IV B 1. In this section, only the dark gray triangular
region will be discussed in detail, and then the total flux will DH5 E tan21 ~ b/a !

0
E
0
a/cos f
x
r sin f ~ r sin f 2b ! d 2
~ r 2 22rx cos f 1x 2 1d 2 ! 2
be derived. First, the limits of the integration of Eq. ~12! over
r in this area are 0 to a/cos f, and those of integration over 3rdrd f . ~22!
f are 0 to tan21(b/a). Secondly, since the B field is an only
function of the y coordinate, the B r will be If we use the same notation for the denominator in Eq. ~22!
with Eq. ~20!, the total flux from segment II can be expressed
B r 52 ~ const! r sin f ~ r sin f 2b ! . ~21! as follows:

J. Vac. Sci. Technol. A, Vol. 14, No. 5, Sep/Oct 1996

2727 Hong et al.: A simulation model for thickness profile 2727

DH II52 x 3 FE tan21 ~ b/a !

0
E
0
a/cos f r sin f 3 ~ r sin f 2b ! 3d 2
f ~ r,x,d, f !
rdrd f

1 E p /2

tan21 ~ b/a !
E 0
b/sin f r sin f 3 ~ r sin f 2b ! 3d 2
f ~ r,x,d, f !
rdrd f 1 E
0
tan21 ~ b/a !
E
0
a/cos f r sin f 3 ~ r sin f 2b ! 3d 2
f ~ r,2x,d, f !
rdrd f

1 E p /2

tan21 ~ b/a !
E 0
b/sin f r sin f 3 ~ r sin f 2b ! 3d
f ~ r,2x,d, f !
2
rdrd f . G ~23!

3. Total flux from rectangular magnetron sputtering ACKNOWLEDGMENTS

This work was sponsored by a grant from Samsung Elec-
From the discussion in the Secs. IV B 1 and IV B 2, we tronics, Co., Ltd. The authors appreciate the cooperation of
can obtain an expression for the total flux from the rectan- Jang Byungtak, a Ph.D course student of KAIST, with the
gular magnetron shown in Fig. 7. Since it is found to be thickness measurement.
difficult to solve the equation analytically, some computer
works with finite elemental method would be necessary.
1
T. Heberlein, G. Krautheim, and W. Wuttke, Vacuum 42, 47 ~1991!.
2
A. A. Fursenko, A. O. Galjukov, Yu. N. Makarov, D. S. Lutovinov, and
However, the equation only contains algebraic terms inside M. S. Ramm, J. Cryst. Growth 148, 155 ~1995!.
itself, so it would not take much time to do such a work. 3
Q. Fan, X. Chen, and Y. Zhang, Vacuum 46, 229 ~1995!.
4
S. Swann, Vacuum 38, 791 ~1988!.
5
M. Knudsen, Ann. Phys. ~Leipzig! 28, 999 ~1909!.
6
R. J. Gnaedinger, Jr., J. Vac. Sci. Technol. 6, 355 ~1969!.
7
B. Chapman, Glow Discharge Processes ~Wiley, New York, 1980!, pp.
260–270.
8
V. CONCLUSIONS S. M. Rossnagel, Handbook of Plasma Processing Technology, edited by
S. M. Rossnagel, J. J. Cuomo, and W. D. Westwood ~Noyes, Park Ridge,
1990!, pp. 166–169.
A simple film thickness uniformity model, which takes 9
I. S. Gredshteyn and I. M. Ryzhik, Table of Integrals, Series and Prod-
the role of magnetic field into account, has been presented. ucts ~Academic, New York, 1980!, pp. 80–83.
10
The good agreement between simulated profile and an ex- S. Dushman and J. M. Lafferty, Scientific Foundations of Vacuum Tech-
perimentally measured one was observed. It has been dem- nique ~Wiley, New York, 1962!, pp. 80–81.
11
M. A. Liberman and A. J. Lichtenberg, Principles of Plasma Discharges
onstrated that the model has a wide range of applicability to and Materials Processing ~Wiley, New York, 1994!, pp. 47.
magnetron sputtering systems utilizing a low-pressure gas 12
M. W. Thompson, Philos. Mag. 18, 377 ~1968!.
during the deposition. It may also be found that this simula-
13
J. A. Thornton, Thin Solid Films 119, 87 ~1984!.
14
R. S. Robinson, J. Vac. Sci. Technol. 16, 185 ~1979!.
tion can predict the film thickness profile of a magnetron 15
S. M. Rossnagel, J. Vac. Sci. Technol. A 6, 19 ~1988!.
sputtering system with various shapes of magnets before 16
B. Chapman, in Ref. 7, p. 377.
manufacturing of the system with relative ease. 17
S. Kadlec and J. Musil, J. Vac. Sci. Technol. A 13, 389 ~1995!.

JVST A - Vacuum, Surfaces, and Films

View publication stats