Investigation of static corrosion between W Metals and TiNx Barriers in a W chemical-mechanical-polishing slurry

Article abstract of DOI:10.1149/1.2907394

In order to control the polishing qualities of tungsten (W) and titanium nitride (TiN_x) films in W chemical-mechanical-polishing (WCMP) processes, the electrochemical behavior between the W and the TiN_x films deposited at various N2 flow rates was examined in this study. Metrologies, including X-ray diffractrometry, Auger electron spectrometry, and scanning electron microscopy, were used to verify the physical properties of the TiN_x films, while electrochemical analyses, including electrochemical impedance spectroscopy, potential dynamic curves, and potential difference measurements, were used to characterize the mechanism of galvanic corrosion between the W and the TiN_x films deposited at various N2 flow rates. The results show that the N content of the TiN_x films influences not only the physical properties of the TiN_x films but also the chemical activity in the WCMP slurries. The equivalent circuit, including the charge-transfer resistance and the titanium-oxide resistance associated with tantalum-oxide capacitance, was built to characterize the mechanism of the galvanic corrosion between the W and the TiN_x metals.

Full text of DOI:10.1149/1.2907394

Journal of The Electrochemical Society, 155 ͑7͒ H469-H473 ͑2008͒
H469
0013-4651/2008/155͑7͒/H469/5/$23.00 © The Electrochemical Society
Investigation of Static Corrosion Between W Metals and TiNx
Barriers in a W Chemical-Mechanical-Polishing Slurry
Chi-Cheng Hung,a Ying-Lang Wang,b,c,z Wen-Hsi Lee,a, and
*
Shih-Chieh Changb
aDepartment of Electrical Engineering, National Cheng Kung University, Tainan 701, Taiwan
bDepartment of Electrical Engineering, National Chia-yi University, Chia-yi, Taiwan
cDepartment of Material Science, National University of Tainen, Tainan, Taiwan
In order to control the polishing qualities of tungsten ͑W͒ and titanium nitride ͑TiNx͒ films in W chemical-mechanical-polishing
͑WCMP͒ processes, the electrochemical behavior between the W and the TiNx films deposited at various N2 flow rates was
examined in this study. Metrologies, including X-ray diffractrometry, Auger electron spectrometry, and scanning electron micros-
copy, were used to verify the physical properties of the TiNx films, while electrochemical analyses, including electrochemical
impedance spectroscopy, potential dynamic curves, and potential difference measurements, were used to characterize the mecha-
nism of galvanic corrosion between the W and the TiNx films deposited at various N2 flow rates. The results show that the N
content of the TiNx films influences not only the physical properties of the TiNx films but also the chemical activity in the WCMP
slurries. The equivalent circuit, including the charge-transfer resistance and the titanium-oxide resistance associated with tantalum-
oxide capacitance, was built to characterize the mechanism of the galvanic corrosion between the W and the TiNx metals.
© 2008 The Electrochemical Society. ͓DOI: 10.1149/1.2907394͔ All rights reserved.
Manuscript submitted December 30, 2007; revised manuscript received February 11, 2008.
Shrinking transistor sizes and the use of tall, three-dimensional
capacitors in dynamic random access memory requires vertical in-
terconnects ͑also called vias or plugs͒ in silicon integrated
circuits.1,2 Chemical-vapor-deposited ͑CVD͒ tungsten ͑W͒ is the
preferred method to realize the filling of a high-aspect-ratio contact
via or plug due to its superior step-coverage.3-6 W is resistant to
electromigration or stress migration due to its high melting point
among pure metals ͑3387°C͒ and low resistivity ͑5.5 ␮⍀ cm͒.
Prior to W deposition, titanium or its nitride ͑Ti/TiNx͒ film is
typically deposited as a glue layer because of the poor adhesion of
W on SiO2 substrates.7-11 The TiNx film further acts as a diffusion
barrier, which offers protection against the reaction of tungsten
hexafluoride ͑WF6͒ and Ti, avoiding the W “volcano” effect. The
volcano phenomenon occurs when WF6 gas penetrates through the
weak sites of TiNx film and then rapidly reacts with the Ti under-
layer, producing by-products of TiF3 gas and solid W. The chemical
formula for this reaction is as follows10
electrical properties of CVD-TiNx films are highly plasma-treatment
sensitive. The physical and chemical properties of the TiNx films are
dependent on the plasma treatment conditions due to different N/Ti
ratios of the TiNx films. In order to verify the galvanic effect be-
tween the W and the TiNx films with different nitrogen content,
PVD-TiNx films with N/Ti ratios controlled by various nitrogen
͑N2͒ gas-flow rates are investigated with W metals using electro-
chemical analyses. The results demonstrate that nitrogen content is
critical in the formation of titanium-metal oxidation and the galvanic
corrosion between the W and the TiNx films. In order to obtain a
robust W metallization process, the N content of the TiNx barriers
should be optimized to achieve a compromise between the barrier
properties and chemical stability.
Experimental
In this study, blanket wafers were deposited with 50 nm thick
TiNx films using the PVD process and 100 nm thick W films using
the CVD process on 300 nm SiO2/Si͑100͒ substrates. The base
pressure of the PVD system was ϳ4 ϫ 10−6 Torr. The TiNx barri-
ers were deposited from a Ti target ͑99.9995%͒. The Ar flow was
kept at 25 sccm, while the nitrogen flow was varied from
20 to 80 sccm, resulting in a process pressure of ϳ0.2 Torr. The gas
purity was 99.9999%. The gas-flow rate was controlled to within
Ϯ0.1 sccm by mass-flow controllers, which guaranteed reproducible
deposition conditions. In the WCVD process, the nucleation W layer
was deposited using WF6 and SiH4 precursors at 395°C. The
bulk-W film was then in situ deposited using the H2 reduction of
WF6 at 395°C in the same chamber.
In the corrosion analyses, the TiNx films deposited at various N2
flow rates were used as the working electrodes ͑4 cm2͒, W plug
films were used as the counter electrodes ͑4 cm2͒, and Ag/AgCl
was used as a reference electrode in a WCMP slurry, which con-
tained aluminum oxide abrasive, surfactant, and hydrogen peroxide
͑ϳ5% H2O2͒ with a pH of 2–3. The Nyquist plot was measured
using a Princeton Applied Research PARSTAT 2273 to analyze the
ac impedance behavior of the capacitor cells. The impedance mea-
surements were carried out at various potential values with a dc
potential amplitude of 0 mV, associated with open-circuit potential,
and a frequency range of 10 mHz to 1000 kHz. The equivalent cir-
cuit was built and simulated using ZSimpWin version 3.1 with elec-
trical impedance spectroscopy ͑EIS͒ data. The potential difference
between the W and the TiNx metals was measured by electrically
connecting two electrodes to an electrometer, which is a high-
impedance multimeter that can measure electrical voltages, resis-
tance, and current. The N content of the PVD-TiNx films was deter-
WF6͑g͒ + Ti → TiF3͑g͒ + W
͓1͔
͑s͒
͑s͒
The TiNx films are generally deposited using physical vapor
deposition ͑PVD͒ or CVD. Conventional PVD-TiNx films have been
widely used as barrier layers because of their high chemical stability
and low resistivity compared to CVD-TiNx films. However, the step-
coverage of the PVD process for high-aspect-ratio features is ex-
tremely difficult to achieve. A CVD-TiNx film with a smaller over-
hang effect has been proposed as an alternative to the PVD-TiNx
film.12,13 After the glue layer and W deposition, overburdened met-
als are removed using W chemical-mechanical-polishing ͑WCMP͒
to define interconnects.14,15 WCMP is adopted for the redundancy of
defective metal etching and better electrical yields to improve the
overall integration for multilevel interconnects.16 However, the im-
pact of a slurry attack and the influence of corrosion, including
galvanic corrosion and intrinsic corrosion, during the WCMP pro-
cess are significant.17-19 Galvanic corrosion is generated by the dif-
ference in the electrochemical potential at the interface between the
W and the TiNx metals in the slurry environment.
In this study, the galvanic effect between W and TiNx metals in
WCMP slurries is investigated. In order to improve the electrical
properties of CVD-TiNx films, an in situ plasma treatment using a
hydrogen/nitrogen gas mixture is commonly employed to lower the
level of C impurities and to form a crystallized TiNx film.20-25 The
*
Electrochemical Society Active Member.
z E-mail: ylwang@tsmc.com
Downloaded on 2015-06-19 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H470
Journal of The Electrochemical Society, 155 ͑7͒ H469-H473 ͑2008͒
Figure 1. Cross-sectional SEM images
͑before and after dipping͒ of W plugs and
TiNx films deposited at various N2 flow
rates.
mined using an Auger electron spectrometer, which used a 10 kV
E-beam and a 3 kV ion gun, with an ion gun sputter rate of
2.19 Å/s. The O ratio of TiNx films after dipping was obtained using
energy-dispersive X-ray ͑EDX͒. The texture of the TiNx films was
identified using X-ray diffractometry ͑XRD͒ with Cu K␣ radiation
͑wavelength of 1.54 Å͒ at 40 kV and 50 mA. The cross-sectional
profile and the thickness of the TiNx/W films before and after slurry
dipping were reviewed using a field-emission scanning electron mi-
croscope ͑SEM͒ of the KLA-Tencor SEMVision tool.
or TiNx͒ corrodes in the slurry. Galvanic corrosion results from the
potential difference between W metal and TiNx films in the WCMP
slurry.
When the N2 gas-flow rate increased from 20 to 40 sccm, the
corrosion thickness of the TiNx films decreased from ϳ5 to
ϳ 2.1 nm. A further increase in the N2 gas-flow rate, from
40 to 80 sccm, led to an increase in the corrosion thickness of the
TiNx films. However, the corrosion thickness of the W films in-
creased with increasing N2 gas-flow rate. Figure 4 shows the poten-
tial dynamic curves of the W and the TiNx metals obtained from
measurements using Princeton Applied Research PARSTAT 2273. It
was found that the corrosion current density ͑Icorr͒ of the W metal is
much higher than that of the TiNx films, implying that the intrinsic
corrosion rate of the W metals is much higher than that of the TiNx
films.
When the N2 gas-flow rate increased from 20 to 40 sccm, the
TiNx films were less active than W metals in the WCMP slurry.
When the N2 gas-flow rate was larger than 40 sccm, the W films
were less active than the TiNx films in the WCMP slurry. Therefore,
the potential difference between the W metals and the TiNx films
deposited at N2 gas-flow rates from 20 to 80 sccm measured using
an electrometer can be naturally divided into two groups, as seen in
Fig. 5. One group is the positive potential difference. The simple
oxidation half reactions of the W and Ti metals are shown as Eq. 2
and 3, respectively
Results and Discussion
Figure 1 shows cross-sectional SEM images ͑before and after
slurry dipping͒ of the W and the TiNx films deposited at the N2 flow
rates from 20 to 80 sccm. The thicknesses and structures of the TiNx
films clearly change with the N2 flow rate. The relationship of the N
stoichiometry and the deposition rates of PVD-TiNx films with vari-
ous N2 flow rates is shown in Fig. 2. The N stoichiometry increases
with increasing N2 flow rate, while the deposition rate of the
PVD-TiNx films decreases with increasing N2 flow rate. Increasing
the amount of N2 gas in the chamber during the TiNx deposition
increased the number of Ti–N compounds that were formed on the
Ti target, resulting in a higher target resistance. The poison effect
reduces the efficiency of the plasma bombardment to the target and
decreases the deposition rate of the PVD-TiNx films.17,18 The in-
creased formation of the Ti–N compounds also caused an increase in
the N stoichiometry of the PVD-TiNx films. In the corrosion analy-
sis, the corrosion thickness is the difference between the film ͑W or
TiNx͒ thickness before and after slurry dipping. The corrosion thick-
ness of the W and the TiNx films deposited at N2 gas-flow rates from
20 to 80 sccm after dipping the W/TiNx bilayer films in the WCMP
slurry is shown in Fig. 3. The corrosion of bilayer films in the
chemical-mechanical-polishing ͑CMP͒ slurry includes self-corrosion
and galvanic corrosion.17 The self-corrosion means that the film ͑W
Ti → Ti4+ + 4e−
W6+ + 6e− → W
͓2͔
͓3͔
From the electrochemical reactions, the relationship of galvanic cor-
rosion between the W and the TiNx metals is as shown in Eq. 4
Figure 2. Relationship of N element ratio and deposition rates of PVD-TiNx
films at various N2 flow rates.
Figure 3. Corrosion thickness of the W plugs and the TiNx films deposited at
various N2 gas-flow rates after slurry dipping.
Downloaded on 2015-06-19 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑7͒ H469-H473 ͑2008͒
H471
Figure 4. Potential dynamic curves of the W plugs and the TiNx films
deposited at various N2 gas-flow rates in the slurries.
Figure 6. EDX of the O element ratio of the TiNx films deposited at various
N2 gas-flow rates after slurry dipping.
*
WCorr ϳ ␮ TiCorr
͓4͔
Ti4+ + 4e− → Ti
͓6͔
where ␮ = 2MWDTi/3MTiDW Ϸ 0.6, in which MW is the molecular
weight of W Ϸ 183.92 g/mol, MTi is the molecular weight of Ti
Ϸ 47.9 g/mol, DW is the density of W Ϸ 19.3 g/cm3, and DTi is the
density of Ti Ϸ 4.54 g/cm3. The other group is the negative poten-
tial difference. The oxidation half reactions of the W and Ti metals
are the opposite of those above, resulting from the TiNx films being
nobler, as seen in Eq. 5 and 6
The relationship of galvanic corrosion between the W and the TiNx
films is as shown in Eq. 7
*
TiCorr
ϳ
WCorr
͓7͔
v
where = 1.5M D /M D Ϸ 1.66. Table I shows the galvanic
v
W
Ti
Ti
W
and intrinsic corrosion rates of the W and the TiNx films. The results
are consistent with Eq. 4 and 7.
W → W6+ + 6e−
͓5͔
Figure 6 shows the O element content of TiNx films deposited at
various N2 flow rates. It demonstrates that the oxidation of TiNx
films is influenced by N doping, which is proven by the intrinsic
corrosion rates of the TiNx films, as shown in Table I.
Figure 7 shows the XRD patterns for the texture evolution of the
as-deposited TiNx films with N2 flow rates from 20 to 80 sccm.
When the N2 flow rate increases from 20 to 30 sccm, the tetragonal
Ti2N, the face-centered cubic TiN, and the Ti phase of the TiNx
films gradually disappear. After a further increase in the N2 flow rate
from 30 to 80 sccm, the phase of TiNx almost completely trans-
forms into an amorphous phase. This is probably due to the forma-
tion of an amorphous nitrogen-rich film. The change of the TiNx
phase also demonstrates that the TiNx films are nobler than the W
metals in the WCMP slurry when the N2 gas-flow rate is larger than
40 sccm.
The Bode plots of Fig. 8a show the effect of applied frequency
on the corrosion impedance between the W metals and the TiNx
films deposited at N2 gas-flow rates from 20 to 80 sccm. TiNx films
deposited at various N2 flow rates were used as the working elec-
trodes, W films were used as the counter electrodes, and all poten-
tials were measured relative to the Ag/AgCl reference electrodes. In
the high-frequency region ͑Ͼ100 kHz͒, which is the inset of Fig.
Figure 5. Potential difference between the W plugs and the TiNx films
deposited at various N2 gas-flow rates from 20 to 80 sccm.
Table I. Summary of intrinsic and galvanic corrosion between W plug and TiNx films deposited with various N2 flow rates. Units are in nm/min.
TiN
TiN
W
W
self-corrosion rate
galvanic corrosion rate
self-corrosion rate
galvanic corrosion rate
N2 = 20 sccm
N2 = 30 sccm
N2 = 40 sccm
N2 = 60 sccm
N2 = 80 sccm
0.934
0.926
2.356
2.823
3.311
0.711
0.084
−1.651
−1.989
−2.383
8.668
8.647
8.512
8.737
8.521
−0.385a
−0.046
0.717
1.175
1.642
a Minus ͑Ϫ͒ means that TiN/W reduction occurs during galvanic corrosion.
Downloaded on 2015-06-19 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H472
Journal of The Electrochemical Society, 155 ͑7͒ H469-H473 ͑2008͒
Figure 7. XRD patterns of IMP-TiNx films deposited at various N2 flow
rates.
8a, the impedance is divided into two groups. The impedance indi-
cates the same trend for the measurements in the low-frequency
region ͑0.01–0.1 Hz͒, as shown in Fig. 8a. Figure 8b shows Nyquist
plots of the TiNx–W electrochemical system in the CMP slurry. Zim
is the imaginary resistance and Zre is the real resistance. The simu-
lated equivalent circuit of the TiNx–W electrochemical system,
which is the inset of Fig. 8b, was built according to the method used
in our previous reference. In this circuit, Rs is the bulk-solution
resistance, Cdl is the double-layer capacitance of the electricity gen-
erated from surface corrosion of the TiNx films, Cox is the capaci-
tance of the TiNx oxidization layer in the form of constant phase
element ͑CPE͒, Rcorr is the charge-transfer resistance ͑associated
with the double layer͒, and Rox is the resistance of the TiNx oxida-
tion layer.18 Furthermore, Rox and Rcorr are in parallel, because sur-
face oxidation and corrosion of the TiNx films occur at the same
time in the TiNx–W electrochemical system. Two semicircles re-
spectively express the transient surface oxidation of the TiNx films,
as shown in Fig. 8b and c, which is a magnification of the imped-
ance close to zero in the high-frequency region ͑Ͼ100 kHz͒. The
chemical reaction of the TiNx films in the presence of hydrogen
peroxide proceeds in the following intermediate steps
Ti + 2H2O2 ↔ TiO2 + 2H2O
TiO2 ↔ Ti4+ + 2O2−
͓8͔
͓9͔
Hence, the first semicircle ͑Fig. 8c͒ shows the high-frequency region
and indicates that the Ti metal oxidizes in peroxide-based solutions
͑Eq. 8͒, and the second semicircle ͑Fig. 8b͒ shows the low-
frequency region and indicates the continuous corrosion of the TiNx
films ͑Eq. 9͒.
When the N2 flow rate is in the range of 20–30 sccm, the TiNx
film acts as an anode ͑Eq. 2͒ compared to the W metal in the
TiNx–W electrochemical system. The potential difference between
the TiNx and W films is smaller and decreases with increasing N2
flow rate. Therefore, the galvanic corrosion of the TiNx films and the
formation of TiO2 enhanced by the W metal are slight and tend to be
suppressed. The values of circuit components were obtained by fit-
ting the Nyquist spectra of Fig. 8b and are summarized in Table II.
According to our earlier study, the elements of the circuit, including
Rox, Cox, and Rcorr, help explain the mechanism of the oxidation and
corrosion of the TiNx films.18 The tendencies of Rox, Cox, and Rcorr
Figure 8. ͑a͒ Bode plots ͓the inset is a magnification of the high-frequency
region ͑Ͼ100 kHz͔͒, ͑b͒ Nyquist plots of the W plugs and the TiNx films at
N2 flow rates from 20 to 80 sccm, and ͑c͒ a magnification of impedance
close to zero in the high-frequency region.
Downloaded on 2015-06-19 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 155 ͑7͒ H469-H473 ͑2008͒
H473
Table II. Element value of equivalent circuit required for best fitting of impedance spectra in Fig. 8b.
CPEox
TiNx–W
Rs ͑⍀͒
Rox ͑⍀͒
Yox ͑␮⍀−1
͒
␣
Cox ͑␮F͒
Rcorr ͑⍀͒
Cdl ͑nF͒
ox
N = 20
N = 30
N = 40
N = 60
N = 80
24.26
25.84
25.28
24.27
24.02
128.1
122.4
295.5
273.4
257.5
255.9
264.1
168.4
194.7
208.9
0.9058
0.9235
0.9312
0.9296
0.9323
179.35
198.75
134.92
155.92
168.95
1429
1615
489.2
474.9
463.4
2.715
2.875
2.276
1.778
1.371
Acknowledgment
verify that the formation of TiO2 and the galvanic corrosion of the
TiNx films decrease with increasing N2 flow rate because nitrogen
inhibits the electron release from the TiNx film to the W film, as
shown in Table II. The galvanic corrosion rate of the TiNx films,
shown in Table I, is consistent with the results of Fig. 8b. The Cdl
value, corresponding to the generation of electrons, becomes
This work was supported by the National Science Council of
Taiwan ͑grant no. NSC 96-221-E-006-125 and NSC 97-2623-7-006-
012-ET͒. The authors thank the National Cheng Kung University,
Tainan, Taiwan for its technical support.
National Chen Kung University assisted in meeting the publication costs
of this article.
smaller, resulting from the reduction of Icorr
.
After a further increase of the N2 flow rate in the range of
40–80 sccm, the TiNx film acts as a cathode ͑Eq. 6͒ compared to the
W metals in the TiNx–W electrochemical system. The potential dif-
ference between the TiNx and W films is larger and increases with
increasing N2 flow rate; therefore, the galvanic corrosion of the TiNx
films is faster and tends to accelerate. Because of the increasing
reduction of Ti4+ ions to Ti metal, the formation of TiO2 reduces
because it accelerates corrosion to replenish Ti4+ ions ͑Eq. 9͒. Simi-
larly, the increase of Cox and the decrease of Rox and Rcorr verify the
reduction of the TiO2 thickness and the increase in the corrosion rate
of the TiNx films with increasing N2 flow rate, as shown in Table II.
Therefore, the elements of the circuit in the TiNx–W electrochemi-
cal system represent their physical meanings. Using the parameters
of impedance spectroscopy, the mechanism of the galvanic corro-
sion between the W metals and the TiNx barriers can be investi-
gated.
References
1. T. Kaga, M. Ohkura, F. Murai, N. Yokoyama, and E. Takeda, J. Vac. Sci. Technol.
B, 13, 2329 ͑1995͒.
2. S. B. Herner, S. A. Desai, A. Mak, and S. G. Ghanayem, Electrochem. Solid-State
Lett., 2, 398 ͑1999͒.
3. A. C. Westerheim, J. M. Bulger, C. S. Whelan, T. S. Sriram, L. J. Elliott, and J. J.
Maziarz, J. Vac. Sci. Technol. B, 16, 2729 ͑1998͒.
4. C. Faltermeier, C. Goldberg, M. Jones, A. Upham, D. Manger, G. Peterson, J. Lau,
and A. Kaloyeros, J. Electrochem. Soc., 144, 1002 ͑1997͒.
5. P. J. Ireland, Thin Solid Films, 304, 1 ͑1997͒.
6. H. L. Chang and P. R. Jeng, Jpn. J. Appl. Phys., Part 1, 39, 4378 ͑2000͒.
7. S. Parikh, L. Akselrod, J. Gardner, K. Armstrong, and N. Parekh, Thin Solid Films,
320, 26 ͑1998͒.
8. J. Wu, Y. L. Wang, J. Dun, Y. L. Wu, H. Zhang, and A. Wang, J. Vac. Sci. Technol.
B, 17, 2300 ͑1999͒.
9. S. B. Herner, Y. Tanaka, H. Zhang, and S. G. Ghanayem, J. Electrochem. Soc.,
147, 1982 ͑2000͒.
10. H. L. Chang, F. L. Juang, and C. T. Kuo, Jpn. J. Appl. Phys., Part 1, 41, 2906
͑2002͒.
11. S. H. Kim, E. S. Hwang, S. Y. Han, S. H. Pyi, N. Kawk, H. Sohn, J. Kim, and G.
B. Choi, Electrochem. Solid-State Lett., 7, G195 ͑2004͒.
12. Y. Nakasaki, K. Suguro, and M. Kashiwagi, J. Appl. Phys., 64, 3263 ͑1988͒.
13. P. Dekker, P. J. Van der Put, H. J. Veringa, and J. Schoonman, J. Electrochem. Soc.,
141, 787 ͑1994͒.
14. C. Yu, S. Poon, Y. Limb, T. K. Yu, and J. Klein, in Proceedings of the VLSI
Multilevel Interconnection Conference, Santa Clara, CA, p. 144 ͑1994͒.
15. M. Rutten, P. Feeney, R. Cheek, and W. Landers, in Proceedings of the VLSI
Multilevel Interconnection Conference, Santa Clara, CA, p. 491 ͑1995͒.
16. J. Lee, J. Kim, and H. Shin, Thin Solid Films, 320, 15 ͑1998͒.
17. C. C. Hung, Y. S. Wang, W. H. Lee, S. C. Chang, and Y. L. Wang, Electrochem.
Solid-State Lett., 10, H127 ͑2007͒.
Conclusions
In conclusion, the N content of the TiNx films affected not only
the physical film properties but also the chemical activity in the
WCMP process. The mechanism of galvanic corrosion in the
TiNx–W electrochemical system was investigated. An equivalent
circuit simulated using EIS data was built to characterize the surface
reaction between the W metals and the TiNx films. With an increas-
ing N2 flow rate in the PVD-TiNx process, the potential difference
between the W metals and the TiNx films in the WCMP slurry
changes to negative from positive, while the role of the TiNx film
changes to a cathode from an anode. The results demonstrate that
the N element content affects the formation of titanium-metal oxi-
dation and the galvanic corrosion between the W metals and the
TiNx films. In order to have a robust W metallization process, the N
content of the TiNx barriers should be optimized to achieve a com-
promise between the barrier properties and chemical stability.
18. C. C. Hung, W. H. Lee, Y. S. Wang, S. C. Chang, and Y. L. Wang, Electrochem.
Solid-State Lett., 10, D100 ͑2007͒.
19. S. Kondo, N. Sakuma, Y. Homma, and N. Ohashi, Jpn. J. Appl. Phys., Part 1, 39,
6216 ͑2000͒.
20. R. Kröger, M. Eizenberg, C. Marcadal, and L. Chen, J. Appl. Phys., 91, 5149
͑2002͒.
21. S. Ikeda, J. Palleau, J. Torres, B. Chenevier, N. Bourhila, and R. Madar, J. Appl.
Phys., 86, 2300 ͑1999͒.
22. S. Ikeda, J. Palleau, J. Torres, B. Chenevier, N. Bourhila, and R. Madar, Solid-State
Electron., 43, 1063 ͑1999͒.
23. D. H. Kim and B. Y. Kim, Jpn. J. Appl. Phys., Part 2, 38, L461 ͑1999͒.
24. S. H. Whang, J. K. Kim, J. W. Park, D. H. Kim, D. L. Cho, and W. J. Lee, Jpn. J.
Appl. Phys., Part 1, 40, 265 ͑2001͒.
25. B. K. Lim, H. S. Park, A. K. H. See, E. Z. Liu, and S. H. Wu, J. Vac. Sci. Technol.
B, 20, 2219 ͑2002͒.
Downloaded on 2015-06-19 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).