A Cu electroplating solution for porous low-k/Cu damascene interconnects

Article abstract of DOI:10.1149/1.2794885

A Cu electroplating solution applicable to porous silica ultralow- k films (k=2.1) without pore sealing was investigated. A suppressor which causes permeation of Cu electroplating solution was replaced by polyethylene glycol (PEG) with specific molecular weight (Mw). Transmission electron microscopy observation revealed that permeation by the Cu solution into the porous silica layer can be suppressed by decreasing the molecular weight of the PEG suppressor in the electroplating solution. A Cu electroplating solution using PEG with Mw=600 was examined for the low- k porous silica/ Cu single-damascene integration process of 300 mm wafer. The filling characteristics in trenches and the uniformity of Cu film thickness were investigated. Interline leakage current on low- kCu damascene interconnects was successfully reduced by six orders of magnitude using this Cu plating solution compared with the conventional solution.

Full text of DOI:10.1149/1.2794885

D692
Journal of The Electrochemical Society, 154 ͑12͒ D692-D696 ͑2007͒
0013-4651/2007/154͑12͒/D692/5/$20.00 © The Electrochemical Society
A Cu Electroplating Solution for Porous Low-k/Cu Damascene
Interconnects
Masashi Shimoyama,a,z Shinichi Chikaki,a Ryotaro Yagi,a Kazuo Kohmura,a
Hirofumi Tanaka,a Nobutoshi Fujii,a Takahiro Nakayama,a Tetsuo Ono,a
Akira Ishikawa,a Hisanori Matsuo,a Keizo Kinoshita,a and Takamaro Kikkawab,c,
*
aMillenium Research for Advanced Information Technology Report, Association of Super-Advanced
Electronics Technology, and bAdvanced Semiconductor Research Center, National Institute of Advanced
Industrial Science and Technology, Tsukuba, Ibaraki 305-8569, Japan
cResearch Center for Nanodevices and Systems, Hiroshima University, Higashi-Hiroshima,
Hiroshima 739-8527, Japan
A Cu electroplating solution applicable to porous silica ultralow-k films ͑k = 2.1͒ without pore sealing was investigated. A
suppressor which causes permeation of Cu electroplating solution was replaced by polyethylene glycol ͑PEG͒ with specific
molecular weight ͑Mw͒. Transmission electron microscopy observation revealed that permeation by the Cu solution into the
porous silica layer can be suppressed by decreasing the molecular weight of the PEG suppressor in the electroplating solution. A
Cu electroplating solution using PEG with Mw = 600 was examined for the low-k porous silica/Cu single-damascene integration
process of 300 mm wafer. The filling characteristics in trenches and the uniformity of Cu film thickness were investigated.
Interline leakage current on low-k/Cu damascene interconnects was successfully reduced by six orders of magnitude using this Cu
plating solution compared with the conventional solution.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2794885͔ All rights reserved.
Manuscript submitted May 31, 2007; revised manuscript received August 28, 2007. Available electronically October 24, 2007.
Scaling of interconnects in ultralarge-scale integrated ͑ULSI͒ cir-
vestigated. Performance of the Cu electroplating solution in terms of
superfilling and thickness uniformity on a 300 mm wafer were ex-
amined. Interline leakage current in porous silica/Cu damascene
structures was also investigated.
cuits requires interlayer materials with a low dielectric constant k for
reducing resistance-capacitance ͑RC͒ signal delay. For achieving
ultralow-k ͑ULK͒ of less than 2.3, introduction of pores into the
dielectric film with increasing porosity is necessary. One failure
mode of porous low-k films is voiding, which often occurs after
processing. This could be attributed to reactions of permeated Cu
electroplating solution with fluoride residues remaining after plasma
etching.1 Several attempts have been made to avoid this failure. A
pore-sealing layer on the porous low-k and a defect-free barrier
metal layer formed by chemical vapor deposition ͑CVD͒ were at-
tempted to prevent such low-k film voiding. However, the CVD
barrier material penetrated into the porous low-k films, resulting in
the rise in k value of the porous low-k films.2 As the thickness of the
barrier metal becomes thinner and thinner on the sidewalls of
trenches and vias, pinholes may not be avoided. Therefore, a new
Cu solution must be developed which does not permeate into the
low-k porous silica films through pinholes in the barrier metals.
We have investigated the effect of Cu electroplating solutions on
the properties of porous silica films.3 There was no change in leak-
age current of the porous silica film with hydrophobic treatment
after dipping it in deionized water. However, the leakage current of
the film increased significantly after dipping it in a Cu electroplating
solution. In general, a Cu electroplating solution is composed of
cupric sulfate solution and three component additives ͑suppressor,
accelerator, and leveler͒. Thus, to identify which one of the compo-
nents is responsible for the increase of leakage current, the influence
of these additives was examined separately. We found that only the
suppressor affects the permeation of the Cu plating solution into the
porous silica films, resulting in increased leakage current.3 In con-
trast, the cupric sulfate solution, containing additives other than the
suppressor, does not increase the leakage current of porous silica
films. Permeation characteristics of the electroplating solution were
examined using polyethylene glycol ͑PEG͒ as a suppressor. It was
found that the permeation of the solution into the porous silica film
depends on the molecular weight ͑Mw͒ of PEG.
Experimental
Fabrication of trench structure in porous silica dielectric
film.— A single-damascene process flow is shown in Fig. 1. First,
25 nm thick SiCN was deposited by plasma chemical vapor deposi-
tion ͑PCVD͒ on a 300 mm Si͑100͒ wafer covered with high-density
plasma ͑HDP͒ SiO2. A porous silica low-k film with a thickness of
140 nm was formed on the wafer by spin-coating of the precursor
solution, which is derived from tetraethoxysilane ͑TEOS͒ and etha-
nol with water, having nonionic surfactant of triblock copolymer
͑EO͒ ͑PO͒ ͑EO͒x.4-6 The film was calcined to burn off the surfac-
x
y
tant and to stabilize the chemical structure of the matrix. The film
was treated with 1,3,5,7-tetramethylcyclotetrasiloxane ͑TMCTS͒ for
enhancing hydrophobicity. The dielectric constant of the film was
2.1. The average pore size of the film was approximately 3 nm, as
measured by small-angle X-ray scattering.5 The porosity of the film
used in the present experiment was 43 5%.7 A 20 nm thick SiOC
or 50 nm thick SiO2 cap was deposited by PCVD on the porous-
silica low-k film. Next, the trench pattern was transferred using a
resist mask with ArF lithography and C5F8/O2-based plasma etch-
ing. After removing the resist by O2 ashing and subsequent wet
cleaning, the trench pattern was transferred to the dielectric stack.
In this paper, the influence of Cu electroplating solutions having
PEG with various molecular weights as a suppressor on the charac-
teristics of porous silica low-k/Cu damascene interconnect was in-
Figure 1. Single-damascene process flow using a resist mask for trench
patterning.
*
Electrochemical Society Active Member.
z E-mail: m-shimoyama@mirai.aist.go.jp
Downloaded on 2015-06-14 to IP 128.218.248.200 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, 154 ͑12͒ D692-D696 ͑2007͒
Finally, a 10 nm thick Ta/TaN barrier layer and 20 nm thick Cu
D693
seed layer were deposited by physical vapor deposition ͑PVD͒ with-
out pore sealing.
Cu electroplating
.— A commercial Cu electroplating solution
͑Ebatronfill II: Ebara-Udylite Co., Ltd͒ based on standard cupric-
sulfate aqueous solution was used together with three additives ͑sup-
pressor, accelerator, and leveler͒. The mixing ratios of suppressor
͑Ebatronfill II-A͒, accelerator ͑Ebatronfill II-B͒, and leveler ͑Eba-
tronfill II-C͒ for standard cupric-sulfate aqueous solution ͑Ebatron-
fill II-KH͒ were 2 ϫ 10−2, 7 ϫ 10−3, and 5 ϫ 10−3 in volume con-
centration, respectively, as the condition of
a conventional
electroplating solution. The commercial suppressor was replaced by
PEG for detailed consideration in this experiment. The molecular
weights of PEG were changed from 62 to 20,000. The standard
concentration of each PEG was 0.3 g/L, which was the same mass
concentration, in order to standardize comparatively the number of
ethylene oxide ͑-CH2–CH2–O-͒ per unit volume of the solution,
because ethylene oxide of PEG was considered to affect the adsorp-
tion of PEG onto a hydrophobic surface8,9 and the adsorption of
water molecule on PEG.10 For accomplishing uniformity of the film
thickness and integrating a Cu/low-k damascene structure, Cu elec-
troplating was performed with equipment in which a porous resis-
tive plate is inserted between anode and cathode for a 300 mm
wafer.11
Evaluation method
.— Permeation of Cu electroplating solution
was evaluated by observing cross sections of the low-k/Cu dama-
scene structures with transmission electron microscope ͑TEM͒ with
energy-dispersive X-ray analysis ͑EDX͒. The lateral resolution of
EDX was about 10 nm. To investigate the influence of additives in
the electroplating solution on the permeation of the solution into the
porous silica film, the contact angle of the solution to the surface of
porous silica blanket film and the leakage current of the film dipped
in the solution were measured by a pendant drop shape analysis and
a mercury probe, respectively. Uniformity of the Cu film thickness
on a 300 mm wafer was determined by sheet resistances. Leakage
current was measured in N2 atmosphere after degassing moisture for
30 min at 150°C of a single-damascene structure with the porous
silica low-k film after chemical mechanical polishing ͑CMP͒.
Figure 2. Cross-sectional TEM images of trenches with porous silica as
interlayer dielectric: ͑a͒ before electroplating and ͑b͒ after electroplating.
Note that a conventional Cu electroplating solution was used. ͑c͒ Z-contrast
image of the sidewall in damascene interconnects. The Ta/TaN barrier layer
appears bright.
Results and Discussion
Permeation of Cu electroplating in
a single-damascene
structure.— Figure 2 shows cross-sectional TEM images of a dama-
scene trench with porous silica before and after the Cu electroplating
process. The thickness of Ta/TaN barrier metal layer on the sidewall
of trenches was 5 nm as shown in Fig. 2c. There is no voiding or
defects in the porous silica layer before Cu electroplating ͑Fig. 2a͒,
while after Cu electroplating with a conventional plating solution,
Cu penetration was observed in the porous silica layer ͑Fig. 2b͒. The
gray region in the porous silica layer was analyzed by TEM-EDX
for elemental composition. The EDX energy spectra for the porous
silica layer before and after electroplating are compared in Fig. 3.
Only Si, O, and C were detected before electroplating ͑Fig. 3a͒. In
contrast, new peaks corresponding to Cu were detected at the gray
region of the porous silica layer after plating ͑Fig. 3b͒. There is no
influence of electron scattering from Cu filled in trenches of the
damascene structure on analysis of the middle of the layer, because
the lateral resolution of TEM-EDX was about 10 nm and no Cu
peak was detected for a similar structure with a dense SiOC film as
interlayer dielectric.
with the suppressor, compared with electroplating solution without
the suppressor. This Cu electroplating solution contained additives
͑accelerator and leveler͒ other than suppressor in conventional con-
dition. The leakage-current characteristics of porous silica films
dipped into the electroplating solution without the suppressor or
deionized water ͑DIW͒ were the same as those of the control. The
contact angle of these solutions on porous silica was measured.
Time change of the contact angle is shown in Fig. 4b. For the Cu
Effect of molecular weight of PEG as suppressor.— We have
reported the influence of Cu electroplating solution on the charac-
teristics of porous silica blanket film.3 An increase of leakage cur-
rent in the porous silica film due to permeation by the Cu plating
solution was found when the porous silica film was dipped in the
solution. As shown in Fig. 4a, the leakage current increased by two
orders of magnitude after dipping into the Cu electroplating solution
Figure 3. EDX spectra of porous silica layer ͑a͒ before electroplating, and
͑b͒ after electroplating.
Downloaded on 2015-06-14 to IP 128.218.248.200 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D694
Journal of The Electrochemical Society, 154 ͑12͒ D692-D696 ͑2007͒
Figure 4. ͑a͒ Leakage current of porous silica film after dipping into the Cu
electroplating solution with or without suppressor. ͑b͒ Contact angle of the
Cu electroplating solution with or without suppressor. The results of DIW are
included for comparison.
electroplating solution with suppressor, the contact angles were
smaller than 90°, and decreased with time. In contrast, the contact
angles of the solution without suppressor were greater than 100°
which was equivalent to DIW, and did not change with time. These
results of leakage current and contact angle indicate that the sup-
pressor in the Cu electroplating solution caused permeation into the
porous silica film. Next, the permeation characteristics of the elec-
troplating solution were examined using PEG as a suppressor. It was
found that the permeation of the solution into the porous silica film
depends on the molecular weight of PEG. Figure 5 shows the rela-
tionship between the leakage current of porous silica films and Mw
Figure 6. Cross sections of trenches with porous silica as interlayer dielec-
tric after Cu electroplating using PEG with various Mws: ͑a͒ 3400, ͑b͒ 1000,
and ͑c͒ 600.
of PEG used as a suppressor in the electroplating solution. The
leakage current increased from 10−9 to 10−7 A/cm2 when the mean
Mw of PEG was larger than 1000. In other words, the leakage cur-
rent was suppressed when the Mw of PEG was less than 1000. From
the experimental result using the porous silica blanket films, three
kinds of PEG with different molecular weights, 600, 1000, and
3400, were selected as suppressors of the electroplating solution
because of the expected difference of the permeation characteristics
into the porous silica. After Cu plating was carried out in trench
structure in porous silica using electroplating solutions with differ-
ent Mws of PEG, the cross-sectional TEM of the trenches were
observed, as shown in Fig. 6. In the case of Mw = 3400 ͑Fig. 6a͒,
Cu penetration occurred, resulting from the permeation of the elec-
troplating solution into the porous silica layer. The appearance of Cu
in the porous silica film was similar to that in the case of a conven-
tional plating solution. In the case of Mw = 1000 ͑Fig. 6b͒, the
permeation of the solution was improved in comparison with the
case of Mw = 3400. The Cu penetration could not be suppressed
completely. In contrast, for Mw = 600 ͑Fig. 6c͒, the permeation was
suppressed completely. These results support our study on blanket
films of porous silica. However, there is a remaining issue; the plat-
ing solution with PEG of Mw = 600 is inadequate for Cu filling
because some voids were observed in the Cu interconnects in the
trenches ͑Fig. 6c͒.
Figure 5. Leakage current as a function of mean molecular weight of PEG in
Cu plating solutions.
Downloaded on 2015-06-14 to IP 128.218.248.200 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, 154 ͑12͒ D692-D696 ͑2007͒
D695
Figure 8. Distribution of Cu thickness across 300 mm wafer using PEG
͑Mw = 600͒ in 0.6 g/L.
aggressive geometries of a trench width of 60 nm ͑AR = 3.4͒ while
maintaining a good uniformity across 300 mm wafers using PVD
Cu seed layers with a thickness of 20 nm.
Evaluation of line-to-line leakage current
.— Cu electroplating,
using the plating solution containing 0.6 g/L PEG ͑Mw = 600͒, was
performed on 300 mm wafers. After Cu electroplating, a single-
damascene structure was fabricated by CMP. The line-to-line leak-
age current without passivation was measured under nitrogen ambi-
ent after 150°C annealing for degassing of moisture. Figure 9 shows
the leakage current of the comb pattern with lines and spaces of 160
and 300 nm, respectively. In the case of the conventional Cu plating
solution with the commercial suppressor, the leakage current in-
creased up to 10−3 A/cm2. In contrast, in the case of the new elec-
troplating solution with 0.6 g/L PEG ͑Mw = 600͒ as a suppressor,
the leakage current was suppressed to the order of 10−9 A/cm2.
Therefore, the new Cu electroplating solution improved the leakage
current by six orders of magnitude.
Figure 7. Cross sections of trenches with porous silica as interlayer dielec-
tric after Cu electroplating using PEG ͑Mw = 600͒ in 0.6 g/L.
Improvement of Cu-filling performance and uniformity of film
thickness.— A suppressor plays an important role in the electroplat-
ing process. The suppressor affects the Cu filling of trenches and the
field-thickness uniformity of Cu film on the wafer. Because the abil-
ity of suppressing Cu deposition on the planar field substrate is
weakened with decreasing molecular weight of PEG,12 the failure of
Cu filling occurred in deep trenches due to overhanging when the
electroplating solution contained PEG with Mw = 600 as a suppres-
sor. One of the methods to suppress the deposition of Cu is to
increase the concentration of PEG.12 Cu electroplating was carried
out using a solution containing PEG whose concentration was two
times ͑0.6 g/L͒ and eight times ͑2.4 g/L͒ higher than the standard
condition ͑0.3 g/L͒. Figure 7 shows cross-sectional TEM images of
Cu damascene trenches ͓linewidth = 60 nm, analytical reagent
͑AR͒ = 3.4͔ in porous silica after Cu electroplating using an elec-
troplating solution containing higher concentrations of PEG. No
penetration of Cu in the porous silica layer or failure of Cu filling
inside the trenches were observed in either condition.
At the same time, the in-plane uniformity of Cu thickness on the
300 mm wafer was measured. No noticeable terminal effect was
observed in both cases in spite of the ultrathin seed layer of 20 nm.
When the conventional electroplating solution was used with a com-
mercial suppressor, the uniformity across 300 mm wafers was
1.53%. In contrast, the uniformities were 2.33% for electroplating
solution with 0.6 g/L PEG and 1.93% for electroplating solution
with 2.4 g/L PEG, respectively. Little difference of thickness uni-
formity across 300 mm wafers was found among these three condi-
tions. The thickness distribution of Cu plating film from the electro-
plating solution with 0.6 g/L PEG is shown in Fig. 8.
Figure 9. Leakage current of a single-damascene interconnect fabricated
using the resist mask scheme. Open circle and filled triangle designate the
case of the electroplating solution using 0.6 g/L PEG ͑Mw = 600͒ as sup-
pressor and the case using conventional suppressor, respectively.
By controlling the concentration of PEG with Mw = 600 in the
Cu electroplating solution, we have achieved efficient filling for
Downloaded on 2015-06-14 to IP 128.218.248.200 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D696
Journal of The Electrochemical Society, 154 ͑12͒ D692-D696 ͑2007͒
Conclusion References
1. T. Jacobs, K. Brennan, R. Carpio, K. Mosig, J. C. Lin, H. Cox, W. Mlynko, J.
Fourcher, J. Bennett, J. Wolf, R. Augur, and P. Gillespie, in Proceedings of the
International Interconnect Technology Conference, IEEE p. 11-2 ͑2002͒.
2. L. Chen, G. Book, S. Smith, M. Rasco, W. M. Li, J. Kostamo, M. Tuominen, K.
Pfeifer, and J. Iacoponi, in Proceedings of the Advanced Metallization Conference
XIX, Materials Research Society, p. 729 ͑2003͒.
3. M. Shimoyama, R. Yagi, S. Chikaki, N. Fujii, T. Nakayama, K. Kohmura, H.
Tanaka, K. Kinoshita, and T. Kikkawa, in Proceedings of the Advanced Metalliza-
tion Conference (AMC), Materials Research Society, Colorado Springs, CO, Sept
27–29, 2005, p. 131.
4. Y. Oku, K. Yamada, T. Goto, Y. Seino, A. Ishikawa, T. Ogata, K. Kohmura, N.
Fujii, N. Hata, R. Ichikawa, T. Yoshino, C. Negoro, A. Nakano, Y. Sonoda, S.
Takada, H. Miyoshi, S. Oike, H. Tanaka, H. Matsuo, K. Kinoshita, and T. Kikkawa,
Tech. Dig. - Int. Electron Devices Meet., 2003, 139.
5. K. Kohmura, S. Oike, M. Murakami, H. Tanaka, S. Takada, Y. Seino, and T.
Kikkawa, Mater. Res. Soc. Symp. Proc., 812, 85 ͑2004͒.
6. N. Fujii, K. Yamada, Y. Oku, N. Hata, Y. Seino, C. Negoro, and T. Kikkawa, Mater.
Res. Soc. Symp. Proc., 812, F4.10 ͑2004͒.
7. N. Hata, C. Negoro, K. Yamada, H. S. Zhou, Y. Oku, and T. Kikkawa, Extended
Abstracts of the 2002 International Conference on Solid State Devices and Mate-
rials, p. 467, Nagoya, Sept 17–19, 2002.
The influence of Cu plating solution containing PEG as a sup-
pressor on porous silica low-k/Cu damascene interconnects without
pore sealing was investigated. Cross-sectional TEM images of
damascene interconnects revealed that Cu penetration occurs in the
porous-silica-interlayer dielectric film when the thicknesses of PVD
barrier metal layer at the sidewall of the trench is less than 5 nm. As
a result of permeation of the Cu electroplating solution, the line-to-
line leakage current increases. The permeation of the Cu electroplat-
ing solution into the porous silica layer can be mitigated by decreas-
ing the molecular weight of PEG to less than 1000. Consequently,
the characteristic of line-to-line leakage current was improved by
using a new suppressor with a molecular weight of 600. The Cu
filling inside trenches and the uniformity of film thickness across a
300 mm wafer were acceptable. These results suggest that this Cu
electroplating solution can be applicable to future 45 nm technology
node interconnects.
8. R. Hogg, in ACS Symposium Series 240, Polymer Adsorption and Dispersion Sta-
bility, E. D. Goddard and B. Vincent, Editors, p. 24, American Chemical Society,
Washington, DC ͑1984͒.
9. S. O. Nielsen, G. Srinivas, C. F. Lopez, and M. L. Klein, Phys. Rev. Lett., 94,
228301 ͑2005͒.
10. Y. Aray, M. Marquez, J. Rodriguez, D. Vega, Y. Simon-Manso, S. Coll, C. Gonza-
Acknowledgments
lez, and D. A. Weitz, J. Phys. Chem. B, 108, 2418 ͑2004͒.
This work was supported by the New Energy and Industrial
Technology Development Organization ͑NEDO͒.
11. M. Hodai, T. Nakada, and M. Tsujimura, in Proceedings of the ADMETA 2005
Asian Session, The Japan Society of Applied Physics, Silicon Technology Division,
Tokyo, Oct 12–14, 2005, p. 4,5.
12. M. Yokoi, S. Konishi, and T. Hayashi, Denki Kagaku oyobi Kogyo Butsuri Kagaku,
52, 218 ͑1984͒.
Association of Super-Advanced Electronics Technologies assisted in
meeting the publication costs of this article.
Downloaded on 2015-06-14 to IP 128.218.248.200 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).