H918 Journal of The Electrochemical Society, 155 共11兲 H918-H922 共2008兲

0013-4651/2008/155共11兲/H918/5/$23.00 © The Electrochemical Society

Electrolyte Composition for Cu Electrochemical Mechanical

Planarization
Abhinav Tripathi,a,* Craig Burkhard,b Ian Ivar Suni,a,**,z Yuzhuo Li,b,**
Francois Doniat,c Alex Barajas,c and James McAndrewc,**
a
Department of Chemical and Biomolecular Engineering and bDepartment of Chemistry and Biomolecular
Science, Center for Advanced Materials Processing (CAMP), Clarkson University, Potsdam, New
York 13699, USA
c
American Air Liquide, Delaware Research and Technology Center, Newark, Delaware 19702-2462, USA

Cu electrochemical mechanical planarization 共ECMP兲 is currently being investigated to replace or supplement Cu chemical
mechanical planarization 共CMP兲 due to the introduction of porous low-k dielectric materials, which may not withstand the
mechanical force applied during conventional CMP. Electrolytes for Cu ECMP at pH 3 containing 5-phenyl-1-H-tetrazole 共PTA兲,
hydroxyethylidenediphosphoric acid, and oxalic acid are investigated using electrochemical methods and polishing of Cu-coated
blanket and patterned wafers. The Cu removal rate and the planarization efficiency during Cu ECMP can be approximated using
electrochemical measurements of the Cu removal rate, with and without surface abrasion. These results predict a 500 mV potential
window within which the Cu removal rate is greater than 600 nm/min and the planarization efficiency is greater than 0.90.
However, high planarization efficiencies are only obtained when silica abrasives are included within the ECMP electrolyte. In situ
electrochemical impedance spectroscopy results indicate that the interfacial impedance is increased by the presence of silica,
suggesting that silica is incorporated into the PTA-based passive film and is thus needed for effective planarization. Electrochemi-
cal quartz crystal microbalance experiments indicate that PTA may provide better Cu surface passivation at a high anodic potential
than benzotriazole, which is widely used during Cu CMP.
© 2008 The Electrochemical Society. 关DOI: 10.1149/1.2980439兴 All rights reserved.

Manuscript submitted May 28, 2008; revised manuscript received August 18, 2008. Published September 25, 2008.

During the last decade, chemical mechanical planarization
共CMP兲 technology has been widely employed during dual dama-
scene processing of Cu interconnects to remove excess Cu and re-
duce surface topography. The planned introduction of porous low-k
dielectric materials has highlighted the need for an alternative tech-
nology to replace or augment CMP for Cu planarization.1 CMP is a
dynamic process involving the coupled chemical oxidation and me-
chanical abrasion of Cu in the presence of passive film forming
agents. The need for both high downforce and abrasive particles
may limit the use of CMP for planarization of devices containing
porous low-k dielectric films, which can be damaged by mechanical
stress.1,2 Other limitations of CMP include high cost of consum-
ables, frequent replacement of parts, poor end-point detection, and
dishing and erosion, which often arise from overpolishing.2,3
The materials integration difficulties described above for Cu
CMP with porous low-k dielectrics have led researchers to investi-
gate alternative planarization processes such as electrochemical pol-
ishing 共ECP兲 and electrochemical mechanical planarization
共ECMP兲. ECP, which is a purely electrochemical process, has sev-
eral advantages over CMP, including its noncontact nature, low con-
tamination levels, and easy end-point detection. However, serious
problems such as pattern sensitivity4 and formation of residual Cu
islands5 have limited its viability as an interconnect planarization
technology.2 ECMP appears to be a more viable approach to bulk Cu
removal. High planarization efficiency and high removal rates have
been reported using this process at very low applied pressure
共⬍0.5 psi兲.6,7
During ECMP, an anodic potential supplies the thermodynamic
driving force for the oxidation and removal of Cu. In contrast, an
oxidizing agent supplies the thermodynamic driving force during Cu
CMP.8 Although the precise mechanism of ECMP is unclear, pla-
narization likely occurs due to accelerated electrochemical Cu re-
moval from surface elevations, where the passive film is abraded,
coupled with little or no removal at surface recesses, where the
passive film remains intact.6 Hence, the nature of the passive film
formed during ECMP largely determines the planarization efficiency
and the Cu surface quality. Benzotriazole 共BTA兲 is the most com-
* Electrochemical Society Student Member.
** Electrochemical Society Active Member.
z
E-mail: isuni@clarkson.edu
monly used inhibitor during CMP and has been studied in great
detail.9,10 However, BTA passivation of Cu at higher anodic poten-
tials may be inadequate.11 In addition, BTA is ineffective at low pH,
where Cu removal rates are expected to be high. Recently, several
alternative triazole derivatives have been tested as low pH inhibitors
for Cu CMP,12-16 including 5-phenyl-1-H-tetrazole 共PTA兲, which has
been identified as a promising candidate for Cu CMP at lower pH.17
Besides the inhibitor, the other important ingredients in an
ECMP electrolyte are a chelating agent, whose effectiveness varies
with pH, and a phosphate-containing species, which is involved in
electrochemical passive film formation. Fundamental electrochemi-
cal studies of Cu dissolution in hydroxyethylidenediphosphoric acid
共HEDP兲 have identified this as an interesting candidate for applica-
tions to both ECP and ECMP.18 Unlike the case for the more com-
monly used phosphoric acid, the Cu passive film thickness in HEDP
can be modulated through control of the applied potential.18 Oxalic
acid has been investigated as a chelating agent for application at low
pH, where Cu dissolution rates are elevated.19,20
The present article describes electrochemical studies and blanket
and patterned wafer polishing results for Cu ECMP electrolytes con-
taining HEDP, PTA, and oxalic acid. Voltammetry measurements
with and without abrasion are employed to benchmark this electro-
lyte formulation. The nature of the passive film formed at different
potentials was further investigated by electrochemical impedance
spectroscopy 共EIS兲 and electrochemical quartz crystal microbalance
共EQCM兲.
Experimental
Voltammetry measurements were made with and without abra-
sion using a 99.999% Cu rotating disk electrode 共RDE兲 in a spe-
cially designed 400 mL three-electrode Teflon cell shown in Fig. 1.
Prior to each experiment, the 1.13 cm diameter Cu working elec-
trode was mechanically polished using 1500 grit sandpaper and then
a 5 wt % slurry of 50 nm alumina until a mirror finish was obtained.
During the voltammetry experiments with abrasion, the Cu electrode
rotation was fixed at 850 rpm, while two cases were studied without
abrasion, rotation at 850 rpm, and no rotation. The rotational speed
that most closely matched experimental conditions during ECMP
was expected to lie between these two extreme values, but is diffi-
cult to know given the complexity of ECMP. The downforce of the
RDE on the pad was not calibrated, but repeat experiments with
abrasion showed variations of only about 10%. Fresh electrolyte was

Downloaded 15 Oct 2008 to 128.153.14.49. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
 Journal of The Electrochemical Society, 155 共11兲 H918-H922 共2008兲 H919

Cu RDE 120

Pt Counter Electrode

Reference SCE 100

Portion of RDE in contact during abrasion

80
ECMP PAD

j (mA/cm2)
60

40

20

Figure 1. 共Color online兲 Schematic of the ECMP beaker electrochemistry

0
experiments.
0 0.5 1 1.5 2 2.5 3

Potential (vs SCE, V)

used for each experiment. All voltammetry measurements were ob-
tained using a Princeton Applied Research 263A potentiostat using a
Pt mesh counter electrode and a saturated calomel reference elec-
trode 共SCE兲. EIS was studied by coupling the same potentiostat with
a Solartron 1250 frequency response analyzer using an ac probe
amplitude of 5 mV.
EQCM measurements were made using a research quartz crystal
microbalance from Maxtek Inc. Crystals with a resonant frequency
of 5 MHz were used for all the measurements, with an approximate
mass resolution of 4 ng/cm2. Cu was electrodeposited onto the Au
surface from a 0.5 M CuSO4 and 0.5 M H2SO4 electrolyte deaer-
ated with Ar. Approximately 2 ␮m of Cu was electrodeposited on
the crystal, with an active working area of 1.37 cm2. This corre-
sponds to approximately 2.45 mg of Cu, which is less than 2 wt %
of the crystal, as required for accurate measurements.21-23 The
freshly deposited Cu surface was rinsed thoroughly in double-
distilled deionized water and used immediately for corrosion mea-
surements in the desired electrolyte. Dissolution was assumed to be
uniform across the quartz surface and changes in resonance fre-
quency were attributed only to mass changes.
Commercial patterned and blanket Cu-coated wafers were used
for testing the ECMP electrolyte. Cu ECMP experiments were per-
formed on 100 mm Cu-coated wafers using a pilot ECMP tool. This
tool was donated by Buehler and modified so that an external po-
tential could be applied. The electrical connections were made with
the help of stainless steel slip rings and graphite–silver alloy carbon
brushes. For all polishing experiments with colloidal silica, Ludox-
AM-30 from Sigma Aldrich was used. For all wafer polishing ex-
periments the downforce was 1.5 psi, the platen and wafer rotational
speeds were 30 rpm, and the electrolyte flow rate was 380 mL/min.
The cathode was a stainless steel disk approximately 0.5 cm thick
and 30 cm in diameter. An IC 1570 共Rohm & Haas兲 pad with XY
and K grooves was used for all the experiments. The pad was per-
forated with holes of 4 and 5 mm diameter at regular intervals, with
an overall pad porosity of approximately 20%. To supply the higher
currents required, a Kepco power supply was employed to control
Figure 2. Voltammetry curves at a scan rate of 5 mV/s measured without
abrasion 共no rotation and with rotation兲 and with abrasion in 1.0 mM PTA,
0.29 M HEDP, and 0.065 M oxalic acid at pH 3 with abrasion and rotation
共…兲, with rotation but no abrasion 共—兲, and without abrasion and without
rotation 共--兲.
ferred during Cu oxidation and dissolution.6 Thus, the voltammetry
curve with abrasion shown in Fig. 2 can be used to calculate the Cu
removal rate, as shown in Fig. 3.
Similarly, the Cu removal rate within trenches can be determined
from the voltammetry results without abrasion, where the surface
inhibitor is intact. This allows calculation of the planarization effi-
ciency 共PE兲 for the electrolyte studied in Fig. 2 according to6
jabrasion − jno abrasion
PE = 关1兴
jabrasion
Figure 3 also illustrates the PE obtained from applying Eq. 1 to the
voltammetry results shown in Fig. 2, using the results without rota-
tion for jno abrasion. Thus, the voltammetry measurements with and
without abrasion can be employed to predict the removal rate and
the PE, which are two critical benchmarks for ECMP. Figure 3
nicely illustrates the trade-off between the removal rate and PE. At
modest anodic potentials, the passive film is quite protective at sur-
face recesses, but the Cu removal rate is low. If the anodic potential
is too high, then the removal rate will be substantial, but the passive
film will be compromised at surface recesses. Because ECMP is not
widely used, these benchmarks are not yet rigidly defined, but re-
moval rates of 600 nm/min are desired with PE ⬎ 0.7. By inspec-
tion of Fig. 3, a wide potential window of approximately 500 mV
1 1800
1600
0.8
1400
Cu Removal Rate(nm/min)
Planarization Efficiency

Cu wafer ECMP. The average Cu thickness was measured with an 1200

RS-30 automated four-point probe. Step heights on patterned wafers 0.6
1000
were measured with an Ambios XP-2 optical profilometer before
and after each polishing step. The surface profile was measured 800
using a Burleigh Horizon noncontact optical profilometer. 0.4
600

Results and Discussion 400

0.2

The voltammetry curves shown in Fig. 2, with and without abra- 200
sion, have been found to be useful for semiquantitative benchmark- 0 0
ing of ECMP electrolytes. The removal rate in the field areas in 0 0.5 1 1.5 2
contact with the pad corresponds to the abrasion curve, while the Potential (vs SCE, V)
removal rate within trenches corresponds to the nonabrasion curve.
Similar methods have recently been described with a slightly differ- Figure 3. Calculated Cu removal rate and PE for ECMP electrolyte contain-
ent geometry.6 According to recent reports, the current supplied dur- ing 0.29 M HEDP, 0.065 M oxalic acid, and either 1.0 mM PTA 共—兲 or
ing ECMP can be directly related to the number of electrons trans- 2 mM PTA 共- -兲 at pH 3.

Downloaded 15 Oct 2008 to 128.153.14.49. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
H920 Journal of The Electrochemical Society, 155 共11兲 H918-H922 共2008兲

1 µm)
Distance(µ
2400

100

200

300

400

500

600

700
0
2200 0
0.8 2000

Cu Removal Rate(nm/min)
-100
1800
Planarization Efficiency

1600
0.6 -200

Step Height (nm)

1400
1200 -300

0.4
0.2
0 0
0 0.5 1 1.5
Potential(vs SCE, V)
Figure 4. Calculated Cu removal rate and PE for ECMP electrolyte contain-
ing 0.29 M HEDP, 1 mM PTA, and varying concentrations of oxalic acid at
pH 3. PE 共䊊兲 and Cu removal rate 共쎲兲 for 0.0325 M oxalic acid, PE 共䊐兲 and
Cu removal rate 共䊏兲 for 0.065 M oxalic acid, and PE 共䉭兲 and Cu removal
rate 共䉱兲 for 0.13 M oxalic acid.
exists where both benchmarks can be achieved. Note that this po-
tential window exists regardless of the choice of rotational speed 共0
vs 850 rpm兲, because the voltammetry curves in Fig. 2 are identical
at low potentials, where the passive film remains intact.
This approach of comparing voltammetry measurements with
and without abrasion can be employed to help optimize the compo-
sition of an ECMP electrolyte. Figure 3 also illustrates the changes
in predicted removal rate and predicted PE obtained upon increasing
the PTA concentration from 1.0 to 2.0 mM. Not surprisingly, in-
creasing the PTA concentration increases the thickness of the pas-
sive film, decreasing the removal at a given voltage. In addition,
such a change in PTA concentration does not significantly impact
the PE. This suggests that 1.0 mM PTA already provides effective
inhibition at surface recesses, so further increases in PTA concentra-
tion can only have deleterious effects on the Cu removal rate.
The exact concentrations of the other chemical ingredients con-
tained in the Cu ECMP electrolytes studied in Fig. 2 and 3 are less
critical than that of PTA. Figures 4 and 5 illustrate the variation Cu
removal rate and PE predicted from voltammetry experiments at
several concentrations of oxalic acid and HEDP, with and without
abrasion. At high oxalic acid concentrations, this reagent begins to
1 100 ⫻ 100 and 50 ⫻ 50 ␮m features measured on different dyes on
0.8
1000
-400
800
600 -500
400
-600
200
-700
2
Figure 6. 共Color online兲 Planarization efficiency experiments showing the
surface profile before and after ECMP for 100 ⫻ 100 ␮m featured on a Cu
patterned wafer at j = 20 mA/cm2 in 0.29 M HEDP, 0.065 M oxalic acid,
1.0 mM PTA, and 3 wt % silica.
chemically etch Cu, preventing passive film formation, so the PE is
compromised. As long as the oxalic acid concentration is high
enough to be an effective complexing agent, then both the Cu re-
moval rate and PE are adequate. In the range of HEDP concentra-
tions tested, the main effect is that the Cu removal rate can be
increased at high HEDP concentration, with a concomitant reduction
in PE.
Comparing the predicted Cu removal rate from electrochemistry
experiments with that obtained in the pilot ECMP tool is compli-
cated by the large internal resistance 共iR兲 drop that is known to
occur in the electrical contacts to the Cu-coated wafer. For the
100 mm wafers studied here, a current density of 20 mA/cm2 yields
a total current of approximately 1.6 A. Even for an ohmic connec-
tion with less than 1 ⍀ resistance, the iR drop from the electrical
contacts is quite substantial relative to the voltage scale in Fig. 2 and
3. However, the predicted PE from electrochemistry experiments
can be compared to that obtained on the pilot tool by conducting
experiments at the same current density.
The PE has therefore been measured on Cu patterned wafers.
Some of the Cu overburden was removed in one 45 s polish. The
profile of a 100 ⫻ 100 ␮m feature before and after polishing at a
current density of 20 mA/cm2 is illustrated in Fig. 6 for an electro-
lyte containing 0.29 M HEDP, 0.065 M oxalic acid, 1.0 mM PTA,
and 3 wt % silica. To obtain an absolute comparison, the surface
profile after polishing has been normalized using the average Cu
removal rate. The measured PE was between 0.9 and 1.0 for both
2400
the patterned wafer. This compares well with the predicted value
2200
from Fig. 3.
2000 However, the excellent PE shown in Fig. 6 can only be obtained
Cu Removal Rate(nm/min)

1800
Planarization Efficiency

by the addition of 3 wt % silica to the ECMP electrolyte. Without

1600 silica, no planarization occurs. This is surprising, given that the in-
0.6
1400 troduction of silica into the electrolyte has a negligible effect on the
1200 electrochemistry results, both with or without abrasion, shown in
0.4 1000 Fig. 2. Even though passivation appears to be effective in the results
800 of Fig. 2 without abrasion, this passivation is inadequate for ECMP.
600
These results suggest that the added silica is actually incorporated
0.2 into the PTA film, providing either static or dynamic Cu corrosion
400
inhibition. Static inhibition means that silica fills defects in the PTA
200
film, whereas dynamic inhibition means that silica stabilizes the
0 0
PTA film and prevents its removal by convective mixing during
0 0.5 1 1.5 2
ECMP.
Potential(vs SCE, V)
Incorporation of silica into the passive film is supported by the
Figure 5. Calculated Cu removal rate and PE for ECMP electrolyte contain- electrochemical impedance studies shown in Fig. 7, which are taken
ing 0.065 M oxalic acid, 1 mM PTA, and varying concentrations of HEDP at in a quiescent electrolyte. These impedance results are not fit here,
pH 3. PE 共䊊兲 and Cu removal rate 共쎲兲 for 0.145 M HEDP, PE 共䊐;兲 and Cu because their shape is somewhat unusual, with a straight line at high
removal rate 共䊏兲 for 0.29 M HEDP; and PE 共䉭兲 and Cu removal rate 共䉱兲 for frequency, a highly depressed semicircle at intermediate frequency,
0.58 M HEDP. and a nearly vertical line at low frequency. However, the magnitude

Downloaded 15 Oct 2008 to 128.153.14.49. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
 Journal of The Electrochemical Society, 155 共11兲 H918-H922 共2008兲
600
500
400
-Zimaginary (Ω cm )
2
300
200
100
0 cal.
0 100 200 300 400 500
2 other benchmark for Cu ECMP is the within-wafer nonuniformity,
Zreal (Ω cm )
Figure 7. Nyquist plot of impedance results for Cu at −250 mV 共vs SCE兲 in
0.29 M HEDP, 0.065 M oxalic acid, 1.0 mM PTA, with 共䊏兲 and without 共䉱兲
3 wt % silica.
of the real impedance, which corresponds to the resistive impedance
component, increases dramatically upon the addition of silica to the
ECMP electrolyte. This is consistent with increasing blockage of the
Cu surface due to an increase in either the surface coverage or
thickness of the passive film. Silica incorporation into the passive
film is also supported by energy-dispersive X-ray analysis experi-
ments on Cu foil, because silicon peaks are observed only for em-
ersion from ECMP electrolytes that contain silica.
In addition, effective passivation can only be obtained if the Cu-
coated wafer is immersed into the ECMP electrolyte at a cathodic
potential to form a passive film prior to application of the anodic
potential that initiates Cu removal. This indicates that the formation
of passive film is restricted once Cu removal is initiated. In BTA-
based ECMP electrolytes, unlike PTA-based ECMP electrolytes, Cu
passivation can be initiated at anodic potentials. However, this com-
parison may be misleading, because the concentration of BTA in
those experiments is much higher 共50 mM兲 than the PTA concen-
tration in the current experiments 共1 mM兲.
For completeness, Cu removal rates have also been measured on
blanket Cu wafers in an electrolyte containing 0.29 M HEDP,
0.065 M oxalic acid, and 1.0 mM PTA, with and without 3 wt %
silica. For an applied voltage of 3.5 V, the average current density
during Cu ECMP was 28 and 29 mA/cm2, respectively, with and
without silica. This again illustrates that the presence of silica does
not increase the Cu removal rate. The average Cu removal rates
were measured as 680 and 710 nm/min, respectively, with and with-
out silica. These measured Cu removal rates on blanket wafers are
about 10% higher than predicted by Faraday’s law.
While the removal rate and PE are quite comparable to or supe-
rior to conventional CMP, the surface quality is not. Noncontact
optical profilometry results for blanket Cu-coated wafers after
ECMP in a PTA-based electrolyte are shown in Fig. 8. Blanket
Cu-coated wafers looked slightly dull after polishing, and root-
mean-square 共rms兲 surface roughness values were as high as
7–10 nm in some spots. The higher surface roughness might be due
to the corrosive nature of oxalic acid, or to the absence of a Cu
oxide film at lower pH. The extent to which low surface quality
impacts the first step of Cu removal, where ECMP was first intro-
duced, is unclear. In other words, if the first step of Cu removal by
ECMP, which prepares a planar surface, is followed with a second
Downloaded 15 Oct 2008 to 128.153.14.49. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
H921
Figure 8. 共Color online兲 Surface profile from Horizon noncontact optical
profilometer following Cu blanket wafer ECMP in 1.0 mM PTA, 0.29 M
HEDP, and 0.065 M oxalic acid at pH 3.
step of Cu removal by CMP to clear the remaining Cu, then the
surface quality following the first Cu removal step may not be criti-
600 In addition to the Cu removal rate, PE, and surface quality, an-
which is expected to be less than 10%. Figure 9 illustrates the
within-wafer nonuniformity for a blanket Cu-coated wafer polished
at a current density of 20 mA/cm2, where the within-wafer nonuni-
formity is approximately 10%.
EQCM studies were also conducted to compare the effectiveness
of BTA and PTA for Cu passive film formation at anodic potentials.
The results are shown in Fig. 10 and were obtained with a potential
scan rate of 5 mV/s. The quartz crystal microbalance studies dem-
onstrate that the Cu surface passivation breaks down for an ECMP
electrolyte containing 1.0 mM PTA at an anodic potential of ap-
proximately +900 mV vs SCE. Comparison studies of BTA-
containing electrolytes were planned carefully, because BTA-based
passive films are well known to break down at low pH when they
become protonated.24,25 EQCM studies of BTA-based ECMP elec-
trolytes were conducted at pH 5.8 using the same HEDP concentra-
tion, and the same concentration 共0.065 M兲 of tribasic ammonium
citrate, which is a more appropriate complexing agent in this pH
range than oxalic acid or glycine. The results shown in Fig. 10
illustrate the much poorer passivation obtained at anodic potentials
for ECMP electrolytes containing BTA, with passive film break-
down at approximately +350 mV vs SCE, even with a much higher
BTA concentration of 10 mM. However, as discussed above, the Cu
passivation observed in Fig. 10 is inadequate without the addition of
silica to the ECMP electrolyte.
2000
1800
1600
1400
Cu Thickness(nm)
1200
1000
800
600
400
200
0
-50 -40 -30 -20 -10 0 10 20 30 40 50
Location
Figure 9. Wafer profile from automated four-point probe ResMap showing
共䊏兲 Cu thickness before ECMP, 共ⴱ兲 Cu thickness remaining after ECMP, and
共䉱兲 Cu thickness removed. ECMP electrolyte contains 1.0 mM PTA, 0.29 M
HEDP, and 0.065 M oxalic acid at pH 3.
H922 Journal of The Electrochemical Society, 155 共11兲 H918-H922 共2008兲
Potential (vs SCE, V)
0.2
0.4
0.6
0.8
1.2
1.4
0
1
0 priate for that pH, passive film breakdown occurs at +350 mV vs
-5
(µg/cm )
2
-10
∆m(µ
-15
-20
-25
Figure 10. EQCM studies of mass change with anodic potential in ECMP
electrolyte at a scan rate of 5 mV/s. Each electrolyte contains 0.065 M com-
plexing agent, 0.29 M HEDP, and 共䊏兲 1 mM BTA, 共䊐兲 10 mM BTA or 共ⴱ兲
1 mM PTA.
Conclusions
Electrolytes for Cu ECMP at pH 3 containing PTA, HEDP, and
oxalic acid are investigated using electrochemical methods and pol-
ishing of Cu-coated blanket and patterned wafers. The performance
of such electrolytes was evaluated using the four main ECMP
benchmarks: the Cu removal rate, PE, surface quality, and within-
wafer nonuniformity. The Cu removal rate and the PE during Cu
ECMP can be approximated using electrochemical measurements of
Cu removal rate, with and without surface abrasion. These results
predict a 500 mV potential window within which the Cu removal
rate is greater than 600 nm/min and the PE is greater than 0.90.
However, experiments on patterned wafers indicate that this PE can
only be achieved with the inclusion of silica in the ECMP electro-
lyte. This appears to reflect incorporation of silica within the PTA-
based passive film, as supported by in situ measurements by EIS.
The rms surface roughness after polishing of Cu-coated blanket wa-
fers was 7–10 nm. The within-wafer nonuniformity during polishing
of blanket Cu-coated wafers was approximately 10%.
In addition, EQCM experiments were employed to compare the
Downloaded 15 Oct 2008 to 128.153.14.49. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
passivation provided at high anodic potential by PTA with that pro-
vided by BTA, which is widely used during Cu CMP. At the pH
共5.8兲 where BTA works well, and using a complexing agent appro-
SCE for BTA, whereas it does not occur until +900 mV for PTA.
Acknowledgments
This research was supported by American Air Liquide and by the
Center for Advanced Materials Processing 共CAMP兲.
Clarkson University assisted in meeting the publication costs of this ar-
ticle.
References
1. N. Chandrashekaran, S. Ramarajan, W. Lee, G. M. Shabde, and S. Meikel, J.
Electrochem. Soc., 151, G882 共2004兲.
2. I. I. Suni and B. Du, IEEE Trans. Semicond. Manuf., 18, 341 共2005兲.
3. P. B. Zantye, A. Kumar, and A. K. Sikder, Mater. Sci. Eng., R., 45, 89 共2004兲.
4. B. Du and I. I. Suni, J. Electrochem. Soc., 151, C375 共2004兲.
5. A. C. West, H. Deligianni, and P. C. Andricacos, IBM J. Res. Dev., 49, 37 共2005兲.
6. F. Q. Liu, T. Du, A. Duboust, S. Tsai, and W. Hsu, J. Electrochem. Soc., 153, C377
共2006兲.
7. Y. Oh, G. Park, and C. Chung, J. Electrochem. Soc., 153, G617 共2006兲.
8. S. Sato, Z. Yasuda, M. Ishihara, N. Komai, H. Ohttorii, A. Yoshio, Y. Segawa, H.
Horikishi, Y. Ohoka, K. Tai, et al., Tech. Dig. - Int. Electron Devices Meet., 2001,
4.4.1.
9. V. Brusic, M. A. Frisch, B. N. Eldridge, F. P. Kaufman, B. M. Rush, and G. S.
Frankel, J. Electrochem. Soc., 138, 2253 共1991兲.
10. P. G. Fox, G. Lewis, and P. J. Boden, Corros. Sci., 19, 457 共1979兲.
11. A. Muthukumaran, V. Lowalekar, and S. Raghavan, Mater. Res. Soc. Symp. Proc.,
914, 231 共2006兲.
12. F. Zucchi, G. Trabanelli, and M. Fonsati, Corros. Sci., 38, 2019 共1996兲.
13. D. P. Schweinsberg, S. E. Bottle, V. Otieno-Olego, and T. Notoya, J. Appl. Elec-
trochem., 27, 161 共1997兲.
14. N. Huynh, S. E. Bottle, T. Notoya, and D. P. Schweinsberg, Corros. Sci., 42, 259
共2000兲.
15. D. Q. Zhang, L. X. Gao, and G. D. Zhou, Corros. Sci., 46, 3031 共2004兲.
16. D. Q. Zhang, L. X. Gao, and G. D. Zhou, Appl. Surf. Sci., 252, 4975 共2006兲.
17. S. K. Govindaswamy, A. Tripathi, I. I. Suni, and Y. Li, J. Electrochem. Soc., 155,
H459 共2008兲.
18. J. Huo, R. Solanki, and J. A. McAndrew, J. Appl. Electrochem., 34, 305 共2004兲.
19. V. K. Gorantala, A. Babel, S. Pandija, and S. V. Babu, Electrochem. Solid-State
Lett., 8, G131 共2005兲.
20. S. Aksu, J. Electrochem. Soc., 152, G938 共2005兲.
21. M. D. Ward and D. Buttry, Chem. Rev. (Washington, D.C.), 92, 1355 共1992兲.
22. D. Buttry, in Electroanalytical Chemistry. A Series of Advances, A. J. Bard, Editor,
p. 29, Marcel Dekker, New York 共1991兲.
23. M. R. Deakin and D. A. Buttry, Anal. Chem., 61, 1147A 共1989兲.
24. Y. C. Wu, P. Zhang, H. W. Pickering, and D. L. Allara, J. Electrochem. Soc., 140,
2791 共1993兲.
25. D. Tromans and R. H. Sun, J. Electrochem. Soc., 138, 3235 共1991兲.