Electrochemical and Solid-State Letters, 6 共6兲 C92-C95 共2003兲
C92
0013-4651/2003/6共6兲/C92/4/$7.00 © The Electrochemical Society, Inc.
Preferential Copper Electrodeposition at Submicrometer
Trenches by Consumption of Halide Ion
Masanori Hayase,a,*,z Munemasa Taketani,a Takeshi Hatsuzawa,a
and Keisuke Hayabusab
a
Tokyo Institute of Technology, Precision and Intelligence Laboratory, Yokohama,
Kanagawa 226-8503, Japan
b
Ebara Research Company, Limited, Fujisawa-shi, Kanagawa 251-8502, Japan
Preferential copper electrodeposition at submicrometer trenches in an acid copper sulfate electroplating bath by addition of only
two components 关bromide ion and polyethylene glycol 共PEG兲兴 is observed. Strong suppression by PEG is observed by the addition
of Br⫺ compared with the addition of Cl⫺. It is supposed that halide ions work as an adhesive between PEG and the copper
surface. During electroplating, halide ions are consumed and the concentration of halide ions in submicrometer trenches is reduced
by diffusional limitation. The reduction of halide ion concentration weakens the PEG adsorption and thus preferential copper
electrodeposition at submicrometer trenches is realized. Nonuniform deposition observed by filling experiments is explained by
this model.
© 2003 The Electrochemical Society. 关DOI: 10.1149/1.1568832兴 All rights reserved.
Manuscript submitted January 8, 2003; revised manuscript received February 14, 2003. Available electronically March 25, 2003.
Copper on-chip interconnection is a current topic in the semiconductor industry.1 It is realized by copper superfilling2 of trenches and
vias in the damascene process. The superfilling is achieved by the
presence of additives in the acid copper sulfate electroplating bath.
The function of each additive is not clear and a physical description
of copper superfilling is not yet established. Many studies based on
the diffusion-adsorption theory3-12 have been carried out to understand the superfilling process and recent studies have paid attention
to catalytic additives like bis共3-sulfopropyl兲disulfide 共SPS兲 or
3-mercapto-1-propanesulfonate 共MPSA兲.13-18 Moffat et al. have
succeeded generally in predicting the shape evolution of copper
electrodeposition by curvature enhanced accelerator coverage
mechanisim 共CEAC兲.
Recently, we demonstrated with Chang et al. that the copper
bottom-up electrodeposition can be achieved by addition of only
Cl⫺ and polyethylene glycol 共PEG兲.19,20 Therefore, there are possibilities that other dominant mechanisms also work in the superfilling. Chang et al. showed interest in the wetting effect of PEG, while
we are interested in the breakdown of PEG-Cl suppression and proposed a model for the bottom-up electrodeposition realized in superfilling. In the model, it is assumed that Cl⫺ is consumed on the
plating surface by incorporation into the deposited copper or formation of soluble complex ions such as CuCl⫺
2 . If the initial consumption is uniform over the plating surface, it is reasonable to suppose
that the reduction of Cl⫺ concentration in the trench bottom is larger
than the reduction on the top surface because of diffusional limitation. The suppression weakens at the trench bottoms, then the rapid
deposition at the trench bottoms is prompted. In this study, the suppression of electrodeposition by PEG and halide ions 共not only Cl⫺)
was carefully investigated and the filling experiments were performed.
the bath was spit out from a thin tube connected to a pump for
supplying fresh electrolyte in the hole. To avoid contamination of
Cl⫺, a copper plate in the cover resin was used as a reference electrode which was expected to work as a stable Cu/CuSO4 electrode. The composition of the standard electrolyte was 225 g/dm3
CuSO4 •5H2 O and 55 g/dm3 H2 SO4 . Electrodes were connected to a
potentiostat 共Hokuto Denko, HABF501兲 and constant current was
applied for copper electrodeposition on the WE. All operations were
controlled by a PC and stable iteration of measurements were expected.
Filling experiment.—The cell for the filling experiment was a
200 mL beaker. A patterned wafer with a copper seed layer 共International Sematech兲 served as the working electrode. The reference
electrode and the counter electrode were platinum wires. All electrodes were connected to a potentiostat. To assume a similar condition of the overpotential measurements in the above section, the
wafer chip was covered with a Teflon plate which had a cylindrical
hole. The electroplating was performed by applying a constant current in an electrolyte bath containing PEG and Br⫺. After electroplating, the chips were rinsed by pure water and dried by Ar gas
flow. The chips were cleaved and the cross section was observed by
a scanning electron microscope 共SEM, JEOL JSM6301F兲.
Experimental
Overpotential measurement.—The cell for electroplating experiments was a 500 mL beaker submerged in a water bath at
298 ⫾ 0.5 K. The working electrode 共WE兲 was a polished platinum
disk in an epoxy resin. To assume one-dimensional flow of current,
ions, and additives, the WE was covered by a resin plate which had
a cylindrical hole (␾ ⫽ 3 mm). The WE was set downward to
avoid the unstable convection due to density change. The WE was
preplated with copper at 200 A/m2 for 20 s in the electrolyte of
interest before each experiment. After preplating, the electrolyte of
* Electrochemical Society Active Member.
z
E-mail: [email protected]
Figure 1. Time variation of overpotential. HBr is added to a PEG-containing
electroplating bath. Applied current density is 100 A/m2.
Electrochemical and Solid-State Letters, 6 共6兲 C92-C95 共2003兲
C93
Table I. Halide ion concentration and overpotential when peak
suppression is maintained at each current density. Overpotential
is an average value of 10 s after electroplating starts.
Current density
共A/m2兲
50
100
3
0.2
0.5
Cl⫺ concentration 共mmol/dm 兲
Peak overpotential
共mV兲
⫺210 ⫺234
200
500
0.75
1.5
⫺259
⫺295
3
0.5
0.75
1
Br⫺ concentration 共mmol/dm 兲
Peak overpotential
共mV兲
⫺260 ⫺288
⫺313
1.5
⫺327
Results and Discussion
Suppression by PEG and halide ions.—Suppression by PEG and
Cl⫺ is sensitive to the concentration of Cl⫺ and these features have
been reported.19,21 Figure 1 shows the time response of overpotential
when HBr was added to a PEG 共Mw 3000兲-containing electroplating
bath. The concentration of PEG was constant at 300 mg/dm3 and
measurements were performed at Br⫺ concentration of 0, 0.01, 0.02,
0.035, 0.05, 0.075, 0.1, 0.2, 0.35, 0.5, 0.75, 1, 1.5, 2, and 3
mmol/dm3. Current density of 100A/m2 was applied for 60 s and IR
drop was subtracted assuming that the resistivity of the electrolyte
solution was uniform over space and constant over time. The suppression by PEG is sensitive to the concentration of Br⫺ as well as
Cl⫺.
Although slight unstable behavior can be seen in the initial electroplating period, nearly peak suppression was maintained when
0.75 mmol/dm3 of Br⫺ was added, which was almost the same
amount of addition as in Cl⫺. The reduction of suppression along
time can be seen when the addition of Br⫺ was less than 0.75
mmol/dm3. Halide ion is supposed to work as an adhesive between
PEG and copper plating surface. Therefore, these reductions may be
caused by the reduction of halide ion concentration on the plating
surface. As a cause of halide ion decrease on the surface, consumption may be dominant because migrational effect is expected to be
small by numerical calculation using Nernst-Plank equation. Cl incorporation into the deposited copper was verified by secondary ion
mass spectrometry 共SIMS兲 measurements.19,22,23 Itagaki et al. investigated dissolution of copper electrode in sulfuric acid solution containing halide ion and the formation of CuCl⫺
2 was observed on both
anode and cathode electrodes by channel flow electrode
experiments.24 Therefore, the consumption of halide ion is supposed
to occur on the cathode surface during electrodeposition.
It was also found that an excess amount of Br⫺ lowered the
overpotential. The reduction of overpotential was observed by the
addition of only HCl or HBr without PEG to a standard electrolyte
bath and this effect may appear by the larger amount of halide ion
Figure 3. SEM micrograph of the trench cross section. Copper is electroplated by applying 100 A/m2 for 25 s in a standard electrolyte bath with PEG
300 mg/dm3 and Br⫺ 0.2 mmol/dm3 .
Figure 2. SEM micrograph of the trench cross section. Copper is electroplated by applying 100 A/m2 for 20 s in a standard electrolyte bath with PEG
300 mg/dm3 and Br⫺ 0.1 mmol/dm3 .
addition. Therefore, it is supposed that the suppression by PEG was
maintained when a large amount of halide ion was added, although
the overpotential was reduced.
The overpotential measurements were performed with various
current densities. The concentration of halide ion, when the peak
suppression was maintained, and peak overpotentials are listed in
Table I. A larger current needed a larger concentration of halide ion
to maintain the suppression. Suppression by Br⫺ was stronger than
suppression by Cl⫺. Similar overpotential measurements were performed for I⫺ and the increase of the overpotential by the addition
of I⫺ to PEG containing bath was observed, but I⫺ was oxidized to
I2 under atmospheric condition and reproducible results were not
obtained.
C94
Electrochemical and Solid-State Letters, 6 共6兲 C92-C95 共2003兲
Figure 4. SEM micrograph of the trench cross section. Copper is electroplated by applying 90 A/m2 for 20 s in a standard electrolyte bath with PEG
300 mg/dm3 and Br⫺ 0.2 mmol/dm3 .
Preferential deposition at trenches.—In the above section, Br⫺
had a similar function to Cl⫺. Therefore, a model similar to Cl⫺, in
which the consumption of Br⫺ led the suppression breakdown and
bottom-up deposition in submicrometer trenches was realized, may
be applicable. To verify the assumption, filling experiments were
carried out.
Figures 2-5 show scanning electron microscopy 共SEM兲 images
of the trench cross section after electroplating. In Fig. 2, almost
conformal deposition can be seen, while in Fig. 3a, copper deposition is observed only at the bottom of the trenches and no deposition
can be seen on the top surface. Strong suppression was maintained
on the top surface during electroplating when a larger amount of
Br⫺ was supplied. Figure 3b and c shows the progress of the deposition. The deposition is not uniform over space due to the instability
described below and it was difficult to obtain precise reproducibility
along plating time. Thus the results of the same electroplating time
are shown in Fig. 3a and b. A wide area of the patterned wafer is
shown in Fig. 4 and 5. Preferential copper deposition at the trenches
can be seen. In Fig. 4, a larger amount of deposition is observed at
the center of the trench series surrounded by flat surface. This tendency is often seen in filling experiments with the addition of PEG
and halide ions.
An explanation for this phenomenon can be made using the halide consumption model. Figure 6 shows a schematic view of the
explanation. Initially, halide ion is consumed uniformly over the
plating surface and a superficial consumption rate of halide ion is
large on the trench series 共Fig. 6a兲. Because diffusional supply of
halide ion is larger at the edge of the trench series, nonuniform
distribution of halide ion concentration is formed on the trench series as shown in Fig. 7. In the calculation, a thin boundary layer was
used because we assumed an initial period of electroplating within 1
s and a stir effect by the submergence of specimen chip may exist. A
diffusional equation should be used for calculating the halide ion
concentration distribution, however, at this moment, we do not have
a quantitative model for halide consumption on the copper surface.
As a rough estimation, a simple calculation was performed using the
Laplace equation. A relatively large halide ion consumption rate
2 ⫻ 10⫺6 mol/m2 s was used. We considered that such a high consumption rate at an initial period may occur, because over 500 ppm
Figure 5. SEM micrograph of the trench cross section. Copper is electroplated by applying 100 A/m2 for 20 s in a standard electrolyte bath with PEG
300 mg/dm3 and Br⫺ 0.5 mmol/dm3 .
Figure 6. Schematic view of the nonuniform deposition due to macroscopic
distribution of halide ion concentration.
of Cl was reported in deposited copper25 and the formation rate of
26
Further quantitacomplex ions such as CuCl⫺
2 may be very high.
tive investigations about halide consumption must be performed to
construct a proper model.
According to the above assumption, the reduction of halide ion
concentration is expected to be largest at the center of the trench
series and the initial suppression breakdown starts at the bottom of
the center trench 共Fig. 6b兲. Once the suppression begins to break, the
electrodeposition rate on the breakdown points increases. The
amount of fresh copper surface exposure to halide ion also increases
and it is supposed that the consumption rate of halide ion increases
on the breakdown points. This assumption may be supported by
Table I in which the larger current density needs higher halide ion
concentration to maintain suppression. Therefore, the reduction of
halide ion concentration becomes larger around the center of the
trench series and subsequent suppression breakdowns occur in the
neighboring trench bottoms 共Fig. 6c兲. Because of this unstable halide ion consumption, it is supposed that a small difference of halide
ion concentration is amplified and a large difference of deposition
rate can be observed.
While in Fig. 5, deposits concentrate around only dense trench
series and there are no deposits on sparse trench series. A dense
trench series has a larger true surface area than a sparse trench series
and the consumption of Br⫺ may be larger on the narrower trench
Figure 7. Simple estimation of Br⫺ concentration on the plating surface.
Assuming steady-state diffusion and uniform consumption of Br⫺, the
Laplace equation is solved by a boundary element method. Boundary condition is shown in the left figure and the diffusion coefficient of Br⫺ is
1.5 ⫻ 10⫺9 m2 /s. Br⫺ consumption rate, 2.0 ⫻ 10⫺6 mol/m2 s, is chosen
for an example.
Electrochemical and Solid-State Letters, 6 共6兲 C92-C95 共2003兲
series. Then suppression breakdown occurs only on dense trench
series. Therefore, it is supposed that these nonuniform deposition
rates observed in Figs. 4 and 5 may support the halide ion consumption model.
Conclusion
Suppression of copper electrodeposition by PEG and halide ion
(Cl⫺ and Br⫺) was carefully investigated. Stronger suppression was
observed by the addition of Br⫺ than by the addition of Cl⫺. Preferential copper electrodeposition at submicrometer trenches was observed by the addition of only PEG and Br⫺ as well as by the
addition of PEG and Cl⫺. Nonuniform copper deposition in the
trench series can be explained by the halide ion consumption model.
However, suppression breakdown is unstable and the quality of deposited copper film does not seem good by SEM observations. Further quantitative studies with other additives are needed to fully
understand the mechanism of superfilling.
Acknowledgments
The authors thank Professor Emeritus Kunitsugu Aramaki of
Keio University and Koji Mishima of Ebara Corporation for useful
discussions.
Tokyo Institute of Technology assisted in meeting the publication costs of
this article.
References
1. T. Osaka, T. Yokoshima, S. Yoshihara, S. Miura, H. Honma, and M. Abe, Electrochemistry (Tokyo, Jpn.), 70, 802 共2002兲.
2. P. C. Andricacos, C. Uzoh, J. Dukovic, J. Horkans, and L. Delligianni, IBM J. Res.
Dev., 42, 567 共1998兲.
3. K. G. Jordan and C. W. Tobias, J. Electrochem. Soc., 138, 1251 共1991兲.
4. C. Madore, M. Matlosz, and D. Landolt, J. Electrochem. Soc., 143, 3927 共1996兲.
5. C. Madore and D. Landolt, J. Electrochem. Soc., 143, 3936 共1996兲.
6. J. J. Kelly, C. Tian, and A. C. West, J. Electrochem. Soc., 146, 2540 共1999兲.
C95
7. J. J. Kelly and A. C. West, Electrochem. Solid-State Lett., 2, 561 共1999兲.
8. K. M. Takahashi and M. Gross, J. Electrochem. Soc., 146, 4499 共1999兲.
9. M. Georgiadou, D. Veyret, R. L. Sani, and R. C. Alkire, J. Electrochem. Soc., 148,
C54 共2001兲.
10. Y. Cao, P. C. Searson, and A. C. West, J. Electrochem. Soc., 148, C376 共2001兲.
11. Y. Cao, T. Taephaisitphongse, R. Chalupa, and A. C. West, J. Electrochem. Soc.,
148, C466 共2001兲.
12. T. Taephaisitphongse, Y. Cao, and A. C. West, J. Electrochem. Soc., 148, C492
共2001兲.
13. K. R. Hebert, J. Electrochem. Soc., 146, C726 共2001兲.
14. T. P. Moffat, J. E. Bonevich, W. H. Huber, A. Stanishevsky, D. R. Kelly, G. R.
Stafford, and D. Josell, J. Electrochem. Soc., 147, 4524 共2000兲.
15. D. Josell, D. Wheeler, W. H. Huber, and T. P. Moffat, Phys. Rev. Lett., 87, 016102
共2001兲.
16. D. Josell, D. Wheeler, W. H. Huber, J. E. Bonevich, and T. P. Moffat, J. Electrochem. Soc., 148, C767 共2001兲.
17. A. C. West, S. Mayer, and J. Reid, Electrochem. Solid-State Lett., 4, C50 共2001兲.
18. T. P. Moffat, D. Wheeler, C. Witt, and D. Josell, Electrochem. Solid-State Lett., 4,
C110 共2002兲.
19. M. Hayase, M. Taketani, K. Aizawa, T. Hatsuzawa, and K. Hayabusa, Electrochem.
Solid-State Lett., 5, C98 共2002兲.
20. S. C. Chang, J. M. Shieh, K. C. Lin, B. T. Dai, T. C. Wang, C. F. Chen, M. S. Feng,
Y. H. Li, C. P. Lu, J. Vac. Sci. Technol. B, 20, 1311 共2002兲.
21. J. P. Healy and D. Pletcher, J. Electroanal. Chem., 338, 155 共1992兲.
22. S. G. Malhotra, P. S. Locke, A. H. Simon, J. Fluegel, C. Parks, P. Dehaven, D. G.
Hemmes, R. Jackson, and E. Patton, in Advanced Metallization Conference 1999,
M. E. Gross, T. Gessner, N. Kobayashi, and Y. Yasuda, Editors, p. 77, Materials
Research Society, Warrendale, PA 共2000兲.
23. M. E. Gross, R. Drese, D. Golovin, W. L. Brown, C. Lingk, S. Merchant, and M.
Oh, in Advanced Metallization Conference 1999, M. E. Gross, T. Gessner, N.
Kobayashi, and Y. Yasuda, Editors, p. 85, Materials Research Society, Warrendale,
PA 共2000兲.
24. M. Itagaki, H. Yamamoto, and K. Watanabe, Corrosion Engineering (Zairyo-toKankyo), 49, 546 共2000兲 关in Japanese兴.
25. S. B. Law, S. Loh, H. W. Ng, B. B. Zhou, H. Zhang, W. L. Tan, J. Sudijono, and
L. C. Hsia, Abstract 675, The Electrochemical Society Meeting Abstracts, Vol.
2003-1, Paris, France, April 27-May 2, 2003.
26. M. Hayase, M. Taketani, K. Aizawa, T. Hatsuzawa, and K. Hayabusa, Abstract 677,
The Electrochemical Society Meeting Abstracts, Vol. 2003-1, Paris, France,
April 27-May 2, 2003.