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

net/publication/239258944

Novel Polymeric Surfactants for Improving Chemical Mechanical Polishing

Performance of Silicon Oxide

Article  in  Electrochemical and Solid-State Letters · May 2001

DOI: 10.1149/1.1357698

CITATIONS READS
9 162

5 authors, including:

Wei-tsu Tseng
GlobalFoundries Inc.
73 PUBLICATIONS   949 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Wei-tsu Tseng on 11 November 2014.

The user has requested enhancement of the downloaded file.

G42 Electrochemical and Solid-State Letters, 4 共5兲 G42-G45 共2001兲
1099-0062/2001/4共5兲/G42/4/$7.00 © The Electrochemical Society, Inc.

Novel Polymeric Surfactants for Improving Chemical

Mechanical Polishing Performance of Silicon Oxide
Wei-Tsu Tseng,a,*,z Ping-Lin Kuo,b Chin-Lung Liao,b Rick Lu,c and Jen-Fin Lind
a
WaferTech, Camas, Washington 98607, USA
b
Department of Chemical Engineering, National Cheng-Kung University, Tainan 701, Taiwan
c
Department of Materials Science and Engineering, National Cheng-Kung University, Tainan 701, Taiwan
d
Department of Mechanical Engineering, National Cheng-Kung University, Tainan 701, Taiwan

We investigated the impacts of novel surfactants on oxide chemical mechanical polishing 共CMP兲 performance. Silica-based
potassium hydroxide was formulated for this study. Two polymeric surfactants, methyl methacrylate 共MMA兲-based CHE and
siloxane-based SHE, were added to the slurry for evaluation. Particle size, viscosity, surface tension, and contact angle of slurries
were characterized. Without surfactant, the silica-based slurry withstood up to 8.5 wt % of solid loading before the silica abrasives
segregate and settle. With the addition of surfactant, however, the slurry held up to 15 wt % of solids without segregation. The
CHE-added slurry yielded higher viscosity and higher oxide CMP removal rate than SHE-added slurry, while the latter exhibited
better colloidal dispersion characteristics and lower within-wafer nonuniformity.
© 2001 The Electrochemical Society. 关DOI: 10.1149/1.1357698兴 All rights reserved.

Manuscript submitted November 13, 2000; revised manuscript received January 24, 2001. Available electronically March 5, 2001.

Chemical mechanical polish 共CMP兲 has become the standard ul-
tralarge scale integrated circuit 共ULSI兲 manufacturing process for
sub-0.25 ␮m Si devices. Despite the proliferation of CMP, however,
much of the basic mechanism involved remains obscure for this
technology. In the open literature, controversy and ambiguity persist
and the basic mechanisms are not yet well understood. For example,
for most of the commercially available slurries, the active chemical
components, such as surfactant 共dispersion agent兲, are usually con-
sidered proprietary and not disclosed to the users. Although it is
generally agreed upon that the dispersion characteristics of slurry is
critical to the polishing performance, in the open literature little is
available regarding the roles of surfactants and their effects on dis-
persion characteristics of CMP slurries. One study1 identified the
type of surfactants used in alumina-based slurries for W-CMP.
However, the basic mechanism of surfactants in CMP slurries has
not yet been investigated systematically.
In this investigation, we synthesized two different surfactants
and evaluated their roles in slurries and impacts on CMP perfor-
mance. Dispersion characteristics of slurries with and without sur-
factants were investigated by measuring the particle size in the slur-
ries. Wetting agent was also added to the slurries to improve the
contact along the slurry/wafer interface. Viscosity and contact angle
of slurries were monitored and slurry pH was adjusted to examine
the influence on CMP performance. CMP removal rate and within-
wafer nonuniformity 共WIWNU兲 were compared between CHE- and
SHE-added slurries. The underlying mechanisms responsible for the
differences in dispersion characteristics and CMP performances are
also discussed.
added as pH buffer, 0.24 or 0.43 wt % KOH solution was added to
adjust the pH to 10.0 or 10.5, respectively. Two types of surfactants
were synthesized and evaluated in this study: the siloxane-based
SHE and MMA-based 共methyl methacrylate兲 CHE. Their chemical
formula are shown in Fig. 1.2 Both are solution-derived polymers
with weight-average molecular weight in the range of 2000 to
3000.2
Particle size in the slurry was measured by an Autosize 4700
laser beam scattering system at a wavelength of 488 nm. Dynamic
contact angle between slurry and wafer surface was calculated from
weight difference during the advancing and recession of a wafer
sample 共measured 20 ⫻ 10 mm).
For oxide CMP tests, the oxide thin films were prepared by
plasma-enhanced chemical vapor deposition 共PECVD兲 on 150 mm
p-type Si wafers, employing SiH4⫹N2O gas chemistry. CMP ex-
periments were conducted on an IPEC372M polish platform with
IC1400 pad and the experimental slurries formulated in our labs.
The applied down force pressure was 5 psi 共32.8 kN/m2兲 throughout
the study. The platen and carrier rotation were fixed at 20 and 25
rpm, respectively. The CMP removal rate was recorded as the five
wafer average taken at nine points on each wafer. Within-wafer
nonuniformity 共WIWNU兲 was defined as the difference between
maximum and minimum removal rates divided by two times the
average removal rate.

Experimental Results
The slurries formulated in this study contained 15 wt % fumed
The measured mean particle size in the slurries is plotted as a
silica abrasives 共Aerosil70 from Degussa兲, 0.15 wt % surfactant, and
0.15 wt % wetting agent. In addition, 0.005 wt % citric acid was function of abrasive concentration 共in wt %兲 in Fig. 2 for slurries
with and without surfactants. The slurry with no added surfactant
共i.e., slurry A兲 had a significantly larger particle size than those with
surfactants 共slurries B and C兲. In addition, particle size of slurry A
* Electrochemical Society Active Member.
z
E-mail: weitsuts@us.ibm.com increased rapidly for silica abrasive concentration greater than 7.5

Figure 1. Chemical formula for 共left兲 SHE and 共right兲 CHE.

Downloaded 14 Apr 2010 to 140.116.208.53. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
 Electrochemical and Solid-State Letters, 4 共5兲 G42-G45 共2001兲 G43

Figure 2. Mean particle size as a function of solid abrasive concentration for Figure 3. Surface tension of slurries as a concentration of surfactant 共CHE
three different slurries. Slurry A: No surfactant added. Slurry B: Slurry with or SHE兲 concentration. No abrasive particles or wetting agent is added.
0.15 wt % CHE as surfactant. Slurry C: Slurry with 0.15 wt % SHE as Slurry pH 10.0.
surfactant. pH was maintained at 10.0 in all three cases.

wt %, and, finally, agglomeration and gellation occurred and the

dispersion characteristics were severely degraded, when solid con-
centration exceeded 8.5 wt %.
On the other hand, slurry B 共with CHE as surfactant兲 and slurry
C 共with SHE as surfactant兲 both exhibited stable dispersion charac-
teristics over the solid concentration range investigated. The latter
also yielded smaller particle size, suggesting a better dispersion ca-
pability.
Surface tension of SHE-added and CHE-added slurries is plotted
against the surfactant concentration in Fig. 3. In this case, no abra-
sive particle or wetting agent was added and the pH was maintained
at 10.0. At concentration less than 5 ⫻ 10⫺4 g/L, surface tension
was basically the same for the two slurries. As surfactant concentra-
tion increased beyond 10⫺3 g/L, however, surface tension of SHE-
added slurry dropped much more rapidly than that of CHE-added
slurry. This result provides further support for the hypothesis that
SHE-added has a better dispersion capability than CHE-added
slurry, because the lower surface tension of the former suggests
better adsorption of the slurry on particle surface and hence better
dispersion capability.
Basic characteristics of experimental CMP slurries formulated
for this study are summarized in Table I. Slurries 1, 2, 3, and 4
contain SHE as surfactant while their respective counterparts, slur-
ries 5, 6, 7, and 8 comprise CHE as surfactant. As shown in this
Table, addition of CHE in the silica-based slurries gave rise to
higher viscosity and larger contact angle in general, compared with
SHE. Also, the median particle size in both groups of slurries was
about the same, despite the size variation. The addition of wetting
agent brings considerable improvement to the wettability of slurries
on the oxide surface, as evidenced from the dynamic contact angle
measurements.
CMP removal rates 共RR兲 with the experimental slurries are
shown in Fig. 4A. Cross comparison between this figure and Table I
indicates that CHE-added slurries give rise to higher removal rates
consistently higher than SHE-added slurries with the same pH and
chemical additives 共e.g., slurry 1 vs. slurry 5兲. The addition of wet-
ting agent also increased the removal rates significantly 共e.g., slurry
1 vs. slurry 2兲. As to WIWNU, SHE-added slurries were lower than
CHE-added slurries, as shown in Fig. 4B. This can be attributed to
the better dispersion capability of SHE as discussed previously. Ad-
dition of wetting agent into the slurry also helps reduce WIWNU.
Note that the removal rates achieved in this study are rather low
for practical industrial applications. This is due mainly to the rela-
tively low down force and velocities used. In a separate test adopt-
ing higher down force 共7 psi兲 and velocities 共carrier/platen
⫽60/45 rpm兲, an average oxide removal rate of 2245 Å/min was
achieved with slurry no. 6, which is comparable to common CMP
practice.

Table I. Composition and basic properties of the slurries formulated in this study. All slurries contain 15 wt % silica abrasive particles.

Wetting Contact
Viscosity Particle size KOH Citric acid Surfactant agent angle
pH 共cP s兲 共nm兲 共wt %兲 共wt %兲 共wt %兲 共wt %兲 共Adv./Rec.兲
Slurry 1 9.96 8.22 211.0 0.24 0.005 SHE 0 52.7/53.6
0.15
Slurry 2 9.95 7.60 207.0 0.24 0.005 SHE 0.15 76.0/70.0
0.15
Slurry 3 10.52 9.22 249.0 0.43 0.005 SHE 0 53.4/59.3
0.15
Slurry 4 10.45 8.82 258.8 0.43 0.005 SHE 0.15 68.0/69.3
0.15
Slurry 5 10.05 19.1 285.3 0.24 0.005 CHE 0 44.7/40.7
0.15
Slurry 6 10.01 22.0 261.8 0.24 0.005 CHE 0.15 67.8/66.7
0.15
Slurry 7 10.46 21.3 273.8 0.43 0.005 CHE 0 49.3/45.8
0.15
Slurry 8 10.52 19.5 264.2 0.43 0.005 CHE 0.15 67.0/67.6
0.15

Downloaded 14 Apr 2010 to 140.116.208.53. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
G44 Electrochemical and Solid-State Letters, 4 共5兲 G42-G45 共2001兲
Figure 4. 共A兲 CMP removal rate and 共B兲 WIWNU of oxides with experi-
mental slurries used for this study. Slurry sample numbers correspond to
those listed in Table I. 3␴ standard deviation is included in the error bar.
Discussion
When dissolved in high pH 共⬎10兲 solution, silica abrasive par-
ticles react with hydroxyls to form Si-OH and Si-O⫺ bonds on the
surface. These partially hydrated particles can be further catalyzed
by OH⫺ and cluster together to form silica aggregate. Time permit-
ting, these aggregates will chain together to form network agglom-
erates or even gellation, leading to sedimentation of abrasives in the
slurry.3 Surfactants are added into slurries to provide either electro-
static barriers or steric barriers between particles for the prevention
of aggregation.
As shown in Fig. 1, SHE and CHE can be classified as nonionic
surfactants with the same hydrophilic tail that carries no charge.
They differ, however, in their hydrophobic groups, siloxane and
MMA, respectively. Because siloxane is lighter in weight than
MMA, SHE will have a higher hydrophile-lipophile balance 共HLB兲
number,4 suggesting a higher water solubility and hence better solu-
bilizing capability than CHE.
From another standpoint, the Si-OH bonds on abrasive silica
surface would act as proton donors and form hydrogen bonds with
the hydrophobic groups of the surfactants. Because silicon has a
lower electronegativity than carbon 共2.20 vs. 2.55兲,5 the siloxane
groups of SHE are attracted more strongly to the Si-OH bonds on
silica surface than the groups of CHE. As a consequence, the hy-
drophobic group of SHE 共e.g., siloxane兲 orient and adsorb more
readily on a silica surface than its counterpart in CHE 共e.g., MMA兲,
leading to better dispersion characteristics.
Downloaded 14 Apr 2010 to 140.116.208.53. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
Based on the arguments above, the underlying mechanism for the
differences in dispersion characteristics between the two surfactants
can be described as follows. While the hydrophilic groups, being the
same in the two cases, orient toward the water 共slurry兲 molecules,
the stronger yet lighter hydrophobic groups in SHE are oriented and
adsorbed more strongly onto silica surface than those in CHE. This
results in better ‘‘isolation’’ between particles, yet higher solubility
of the colloidal dispersion of the SHE-added slurry, as manifested
by its lower surface tension 共Fig. 2兲, lower viscosity and larger
contact angle 共Table I兲.
Regarding CMP performance, the differences between the two
slurries may also originate from their dispersion characteristics. Ad-
dition of surfactant into the slurry has been shown to decrease the
removal rates.1 This is attributed to the reduction in the effective
size of abrasives in the presence of surfactant, hence decreasing the
mechanical removal rate. In another study by the same group, how-
ever, the W-CMP removal rate was found to increase with decreas-
ing particle size and increasing solid loading.6 This contradiction is
explained by the competition between contact-area dominated pol-
ishing and indentation-based polish mechanism.7 At a fixed solid
concentration, the removal rate is proportional to the total contact
area between abrasive particles and wafer surface. As a conse-
quence, smaller particles correspond to larger contact area and hence
higher RR. In the indentation polish mechanism, RR is proportional
to the total indentation volume, which in turn is directly related to
the particle size.
In the current study, however, solid concentration was the same
in both cases and particle size in CHE-added slurry was only mar-
ginally larger than that in SHE-added slurry. Therefore, it is doubt-
ful that this subtle difference in particle size could account for the
significant removal rate difference 共up to ⬃40%兲 between the two
slurries, assuming that polishing operated on the indentation mecha-
nism. More likely, the higher removal rate of CHE-added slurry
originated from accelerated chemical process 共e.g., hydrolysis of
oxide in the slurry兲 or other nonmechanical processes. Further ex-
periments are ongoing to verify this hypothesis. Despite this ambi-
guity, WIWNU of SHE-added slurry can be attributed in part to its
better dispersion characteristics, which give better slurry/wafer con-
tact and hence more uniform polishing.
In this study, the addition of wetting agent into the slurries
marked a significant increase in removal rate. This can be attributed
to the enlarged contact area caused by the wetting agent between the
wafer and abrasive particles, rendering a more hydrophilic oxide
surface in favor of a more rapid hydrolysis reaction of oxides with
the OH⫺ radicals in the slurries.
Also worth noting in this study is the effect of pH on dispersion
characteristics and polish performance. Because of the lack of
charge groups in the polar Si-OH surface of silica, pH was expected
to have a less pronounced effect upon adsorption. This is reflected in
the data in Table I, where viscosity and contact angle are virtually
the same between, for example, slurries 1 and 3, and 2 and 4. The
higher removal rate of higher-pH slurries was simply the conse-
quence of enhanced hydration reaction of oxide wafer surface under
stronger ionic strength of 关 OH⫺兴 , as found in many previous
studies.8
Conclusion
In this preliminary study, we investigated the influence of two
polymeric surfactants, SHE and CHE, on the dispersion characteris-
tics and CMP performance of silica-based alkaline slurries. At a
fixed abrasive particle concentration, SHE-added slurries exhibit
lower viscosity, lower surface tension, higher contact angles, and
slightly smaller particle size than CHE-added slurries. The superior
dispersion characteristics of SHE-added slurry is also reflected by its
lower CMP WIWNU. CHE-added slurry, on the other hand, gives
rise to higher CMP removal rate despite its relatively inferior col-
loidal suspension.
Wafer Tech assisted in meeting the publication costs of this
article.
 Electrochemical and Solid-State Letters, 4 共5兲 G42-G45 共2001兲 G45

References
1. M. Bielmann, U. Mahajan, R. K. Singh, D. O. Shah, and B. J. Palla, Electrochem.
Solid-State Lett., 2, 148 共1999兲.
2. C.-L. Liao, M.S. Thesis, Department of Chemical Engineering, National Cheng-
Kung University, Tainan, Taiwan 共2000兲.
3. R. K. Iler, The Chemistry of Silica, John Wiley & Sons, New York 共1979兲.
4. P. Becher, in Nonionic Surfactants, M. J. Schick, Editor, Marcel Dekker Inc., New
York 共1967兲.
5. W. Gordy and W. J. O. Thomas, J. Chem. Phys., 24, 439 共1956兲.
6. M. Bielmann, U. Mahajan, and R. K. Singh, Electrochem. Solid-State Lett., 2, 401
共1999兲.
7. G. B. Basim, J. J. Adler, U. Mahajan, R. K. Singh, and B. M. Moudgil, J. Elec-
trochem. Soc., 147, 3523 共2000兲.
8. L. M. Cook, J. Non-Cryst. Solids, 120, 152 共1990兲.

View publication stats

Downloaded 14 Apr 2010 to 140.116.208.53. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp