Ruthenium Bottom Electrode Prepared by Electroplating for a High Density DRAM Capacitor

Article abstract of DOI:10.1149/1.1637900

The possibility of Ru electroplating for application as the bottom electrode in high density dynamic random access memory (DRAM) capacitors was investigated. Prior to Ru electroplating on a TiN substrate, HF cleaning and Pd activation were performed. Removal of Ti oxide from the TiN substrate by HF treatment enabled Pd activation, which enhanced the nucleation of Ru on TiN substrate. Optimized pretreatments led to a continuous Ru film deposition. The surface roughness was measured to be 4.4 nm at 45 nm Ru film on the bare substrate. Moreover Ru electroplating method was also applied to a capacitor node-type TiN wafer. The deposition rate of Ru on the patterned wafer was the same as that on a bare wafer. The film showed 93% step coverage and good adhesion, comparable to CVD Ru films.

Full text of DOI:10.1149/1.1637900

Journal of The Electrochemical Society, 151 ͑2͒ C127-C132 ͑2004͒
C127
0
013-4651/2004/151͑2͒/C127/6/$7.00 © The Electrochemical Society, Inc.
Ruthenium Bottom Electrode Prepared by Electroplating for a
High Density DRAM Capacitor
Oh Joong Kwon,** Seung Hwan Cha, and Jae Jeong Kim*,z
Research Center for Energy Conversion and Storage, School of Chemical Engineering,
Seoul National University, Kwanak-gu, Seoul 151-742, Korea
The possibility of Ru electroplating for application as the bottom electrode in high density dynamic random access memory
͑DRAM͒ capacitors was investigated. Prior to Ru electroplating on a TiN substrate, HF cleaning and Pd activation were per-
formed. Removal of Ti oxide from the TiN substrate by HF treatment enabled Pd activation, which enhanced the nucleation of Ru
on TiN substrate. Optimized pretreatments led to a continuous Ru film deposition. The surface roughness was measured to be 4.4
nm at 45 nm Ru film on the bare substrate. Moreover Ru electroplating method was also applied to a capacitor node-type TiN
wafer. The deposition rate of Ru on the patterned wafer was the same as that on a bare wafer. The film showed 93% step coverage
and good adhesion, comparable to CVD Ru films.
©
2004 The Electrochemical Society. ͓DOI: 10.1149/1.1637900͔ All rights reserved.
Manuscript submitted March 20, 2003; revised manuscript received August 18, 2003. Available electronically January 9, 2004.
As the device density of dynamic random access memory
Experimental
͑
DRAM͒ increases to gigabyte scale, the unit cell size, consisting of
The substrates used in this study were CVD TiN ͑10 nm͒/PVD Ti
15 nm͒/Si ͑100͒ blanket wafers and CVD TiN ͑20 nm͒/Si ͑50 nm͒/
2
one transistor and one capacitor, decreases to 0.1 ␮m . As unit cell
size decreases, the fabrication of 25 fF capacitor has become a cen-
tral issue in DRAM manufacturing. In the gigabit density era, the
use of new materials with high dielectric constants ͑␧͒ such as
Ta O and (Ba, Sr͒TiO has been proposed as a new solution. In
high dielectric material integration, a bottom electrode with high
oxidation resistance and high work function is required. As the suit-
͑
SiO patterned wafers ͑aspect ratio 2.5:1͒.
2
Prior to electroplating, two surface pretreatments on TiN sub-
strate were performed sequentially. Ti oxide was removed from TiN
substrate by using 1% HF solution for 10 min followed by deionized
͑DI͒ water cleaning. Pd activation was done in the Pd activation
solution of pH 2.57 composed of PdCl2 , 50% HF ͑12.5 mL/L͒, and
HCl ͑3.0 mL/L͒ while PdCl2 concentrations were varied from 0 to
1
,2
2
5
3
3
,4
able metals which satisfy above two requirements, Pt, Pt/Ru, and
Ru were proposed.
0.4 g/L. The oxide free TiN substrate was dipped in Pd activation
solution for 20 s and rinsed with DI water.
Among these metals, Pt has been used as a bottom electrode. But
oxygen diffusion and bad etching property have become problems in
deep submicrometer dimension era. Alternatively, Ru film has re-
cently attracted attention as a substitution of Pt because it can be
easily patterned, has good etching property and forms oxygen diffu-
sion preventing layer of RuO2 on top of Ru film during high ␧
material deposition. Pt/Ru is also considered to be suitable as a
bottom electrode due to high work function and oxygen diffusion
After pretreatment process, Ru electroplating was performed. Ru
source and adjusting salt were W-RU2 ͑Japan Energy͒. The Ru elec-
trolyte was made up of ruthenium sulfate solution ͑5 g/L͒, NaOH
͑40 g/L͒, and adjusting salt ͑100 g/L͒. The adjusting salt was com-
posed of phosphoric acid, sulfamic acid, and ammonium sulfamate.
The pH of Ru electrolyte was 1.3. A constant Ϫ1.3 V potential was
applied by a PAR 263 ͑EG&G Princeton Applied Research Corpo-
ration͒ with respect to saturated calomel electrode ͑SCE͒ at room
2
preventing layer of RuO at interface between Pt and Ru. In spite of
temperature. And the current density was 42 mA/cm with this ap-
2
these advantages, it cannot be applied to deep submicrometer di-
mension era because it is too thick to apply to deep submicrometer
dimension structure. From above description it can be concluded
plied potential. Electroplating was carried out on both bare wafer
and patterned wafer at the same condition.
The Ru film was characterized using field emission scanning
electron microscopy ͑FESEM͒, X-ray photoelectron spectroscopy
͑XPS͒, X-ray diffraction ͑XRD͒ measurement, atomic force micros-
copy ͑AFM͒, and Auger electron spectroscopy ͑AES͒. The resistiv-
ity of ruthenium film was measured using the standard four-point-
probe method.
5-8
that Ru is the most promising material.
Chemical vapor deposition ͑CVD͒8 and physical vapor deposi-
,9
tion ͑PVD͒10 have been widely used for the deposition of Ru films.
These two processes have certain limitations including narrow pro-
cess window, complex apparatuses, and high cost. In addition, rough
surface morphology and poor step coverage are becoming important
obstacles as feature size shrinks. Electroplating represents an alter-
native for the deposition of thin metal films because of good film
quality, simple apparatus, and its low cost. In spite of these advan-
Results and Discussion
To find out an optimum Pd activation condition, PdCl concen-
2
trations were varied from 0.025 to 0.4 g/L. The deposited Pd par-
ticles were analyzed by XPS and FESEM as shown in Fig. 1 and
tages, however, electroplating has rarely been used for depositing
Ru film on semiconductor devices.1
1-17
The chief problems of Ru
Fig. 2. Both Pd and PdCl were found on the activated TiN sub-
2
electroplating on a TiN substrate are Ru nucleation and limitation on
the current distribution associated with high resistivity of the TiN
barrier layer.1
In this study, Pd activation was investigated as a nucleus forma-
tion method to make Ru electroplating possible on TiN substrate.
The size and density of nuclei were optimized, and Ru electroplating
was carried out onto Pd activated TiN substrate. Then the quality of
Ru film was estimated.
strate.
According to AFM analyses, the size of the Pd particle increased
in proportion to PdCl2 concentration. It was smaller than 5 nm be-
low 0.025 g/L PdCl2 concentration, and larger than 10 nm above 0.4
g/L PdCl2 concentration. Pd densities were almost same to be 8.25
8-20
1
0
2
ϫ 10 /cm , independent of PdCl2 concentration of 0 to 0.4 g/L at
1% HF concentration. Figure 3a to 3e show the difference in density
due to the shading of large particles over small ones. Among differ-
ent PdCl concentrations, Pd activation with 0.1 g/L PdCl showed
2
2
the most uniform size distribution and the smallest surface rough-
ness of 2.9 nm as shown in Fig. 2 and 3c.
For variable Pd activation conditions Ru electroplating was per-
formed on TiN substrate as shown in Fig. 4. Below the concentra-
*
Electrochemical Society Active Member.
*
* Electrochemical Society Student Member.
z
Electronic mail: jjkimm@snu.ac.kr
Downloaded on 2015-03-09 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C128
Journal of The Electrochemical Society, 151 ͑2͒ C127-C132 ͑2004͒
Figure 2. Surface FESEM image of TiN substrate activated with 0.1 g/L
PdCl2 .
Figure 1. XPS spectrum of TiN substrate activated with 0.1 g/L PdCl2 .
Figure 3. AFM analyses of TiN sub-
strate activated with different PdCl2
concentrations ͑a͒ 0.025 g/L, ͑b͒ 0.05
g/L, ͑c͒ 0.1 g/L, ͑d͒ 0.2 g/L, and ͑e͒
0.4 g/L.
Downloaded on 2015-03-09 to IP 157.182.150.22 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, 151 ͑2͒ C127-C132 ͑2004͒
C129
Figure 4. Surface FESEM images of Ru
film activated with PdCl2 concentration
of ͑a͒ 0 g/L, ͑b͒ 0.025 g/L, ͑c͒ 0.05 g/L,
͑
d͒ 0.1 g/L, ͑e͒ 0.2 g/L, and ͑f͒ 0.4 g/L.
To investigate the compositional characteristics and crystalline
structures of Ru film, AES, XPS and XRD analyses were performed.
Carbon contamination was checked by using low intense Ru ͑MNN͒
peak because high intense Ru ͑MNN͒ peak overlaps with carbon
tion of 0.1 g/L PdCl , the Ru film was not continuous and in the
2
range of 0 to 0.025 g/L PdCl , Ru cluster size and surface morphol-
2
ogy were almost identical. These phenomena reveal that Pd particles
below 5 nm do not work as the nucleation centers. In the range of
͑KLL͒ peak in AES analysis and by comparing the area ratio of
0
.05 to 0.1 g/L PdCl concentration, Ru clusters became smaller and
2
Ru/RuO2 of 3d5/2 peak with that of 3d3/2 peak in XPS analysis.
Carbon was not detected in both analysis methods. In the AES
analysis, it is found that some oxygen is contained in the Ru film
and Pd particles exist at the interface of Ru and TiN ͑Fig. 6a͒. It can
be inferred from Fig. 6b that the oxygen content in the Ru film
a continuous Ru film was obtained. However, for Ru films activated
above 0.2 g/L PdCl concentration, the Ru clusters were larger and
2
the surface morphology was rougher compared with films activated
in solution with 0.05 to 0.1 g/L PdCl . This is ascribed to the larger
2
Pd particle size associated with the 0.2 g/L PdCl ͑Fig. 4e and f͒.
2
corresponds to RuO . Four overlapping peaks at 280.0, 280.7,
2
These observations indicate that there is an optimum PdCl concen-
2
284.2, and 284.7 eV were observed in XPS data. Among them,
tration for making Ru film continuous and smooth. A near optimum
280.0 and 284.2 eV correspond to the 3d5/2 and 3d3/2 binding energy
5
nm Pd particle size was obtained in a 0.1 g/L PdCl at 1% HF
2
of Ru, and 280.7 and 284.7 eV indicate the 3d5/2 and 3d3/2 binding
solution. The thickness and surface morphology of a Ru film depos-
energy of RuO . Moreover, XRD analysis points out that RuO is
ited from a 0.1 g/L PdCl solution are shown in Fig. 5a and b. Film
2
2
2
amorphous ͑Fig. 7͒, in which a weak and broad peak around 35.16°
thickness was 45 nm and root mean square ͑rms͒ surface roughness
was 4.4 nm. The RMS surface roughness was below 10% of Ru film
thickness and which enabled application as a bottom electrode of a
DRAM capacitor. And the adhesion of Ru film was tested with
American Society for Testing and Materials ͑ASTM͒ tape test
method. Ru film was not peeled off by tape test method.
reveals RuO . Therefore it can be concluded that the electroplated
2
Ru film is the composite film of Ru and RuO2 . The ratio of RuO2 to
Ru is 1/24 in the middle of as-deposited film, which was calculated
with AES data by considering sensitivity factor and stoichiometric
number. And the ratio of Ru/RuO ͑1/3 in AES analysis and 1/2 in
2
Downloaded on 2015-03-09 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C130
Journal of The Electrochemical Society, 151 ͑2͒ C127-C132 ͑2004͒
Figure 5. ͑a͒ Cross sectional FESEM image of Ru film activated with 0.1
g/L PdCl , and ͑b͒ AFM analysis of Ru film activated with 0.1 g/L PdCl .
2
2
Figure 6. ͑a͒ AES depth profile of Ru film activated with 0.1 g/L PdCl and
2
͑
b͒ XPS spectrum of Ru film activated with 0.1 g/L PdCl2 .
Figure 7. XRD patterns of ͑a͒ TiN substrate, ͑b͒ Ru film after electroplating,
and ͑c͒ Ru film after anneal process at 500°C for 10 min in air atmosphere.
Figure 8. Trend of resistivity of electrodeposited Ru film.
Downloaded on 2015-03-09 to IP 157.182.150.22 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, 151 ͑2͒ C127-C132 ͑2004͒
C131
Figure 9. FESEM micrographs of ͑a͒
patterned TiN wafer, ͑b͒ Ru film on pat-
terned TiN wafer, ͑c͒ patterned TiN wa-
fer ͑enlarged͒, and ͑d͒ Ru film on pat-
terned TiN wafer ͑enlarged͒.
XPS analysis͒ is lower at the surface than in the middle of film due
to native oxide. Inclusion of oxygen into the middle of film is hard
to be avoided in solution experiment due to high affinity of Ru to
wafer, showing 93% step coverage and smooth surface morphology.
And it also showed good adhesion.
oxygen.2
1,22
Conclusion
Thus to remove oxygen content from Ru film, post-
treatment such as electrical reduction and anneal were suggested and
showed possibility. These processes are to be studied.
The possibility of Ru electroplating for application as the bottom
electrode fabrication method of a high-density DRAM capacitor was
investigated. Without the Pd activation process, a discontinuous Ru
film was obtained due to the low density of nucleation center. But
As-deposited Ru film showed small cluster size and poor crys-
talline structure by XRD pattern in Fig. 7b. But after anneal process
at 500°C for 10 min in air, ͑101͒ preferred orientation was obtained
and ͑100͒, ͑002͒ peaks appeared during anneal because of grain
growth and recrystallization. By comparing Fig. 7a with 7b, it can
after Pd activation was performed with 0.1 g/L PdCl for 20 s on
2
TiN wafer, a continuous Ru thin film was obtained. After electro-
plating, the Ru film had 45 nm thickness, 7.5 nm/min deposition
rate, and 4.4 nm surface roughness. These are appropriate value for
DRAM application. The resistivity of electroplated Ru film was high
compared with that of MOCVD Ru film, but it could be further
reduced by annealing. A reduced resistivity of 38 ␮⍀ cm was ob-
tained although the optimal annealing conditions have not been es-
tablished. With a capacitor node-type wafer, Ru electroplating also
showed the same thickness as bare TiN wafer, acceptable surface
roughness, 93% step coverage, and good adhesion. These results
indicate that Ru electroplating is comparable to Ru MOCVD
method.
be seen that a Ti N ͑200͒ peak shifted to the left after annealing, so
2
that compressive stress is applied to Ru film.
The trend of resistivity of Ru film is shown in Fig. 8. The resis-
tivity of 300 ␮⍀ cm of as-deposited Ru film in Ru film deposited by
electroplating is relatively higher than that of metallorganic CVD
͑
MOCVD͒ Ru film. It is believed that small cluster size, poor crys-
talline structure and impregnated amorphous RuO contribute to the
2
high resistivity. With annealing process, however, this high resistiv-
ity of Ru film can be further reduced to 38 ␮⍀ cm due to grain
growth and recrystallization of Ru, which is comparable to the re-
sistivity of MOCVD Ru film. Oxidation of Ru in the film did not
Acknowledgment
occur during annealing because thin RuO layer formed on top of
2
3
the film prevented further diffusion of oxygen.
This work was supported by KOSEF through the Research Cen-
ter for Energy Conversion and Storage ͑RCECS͒ and also by Insti-
tute of Chemical Processes ͑ICP͒, and Inter-university Semiconduc-
tor Research Center ͑ISRC͒.
The optimized process condition of Ru electroplating onto bare
wafer was applied on capacitor node-type TiN patterned wafer to
examine the possibility of Ru electroplating to real DRAM fabrica-
tion method. The minimum feature size of TiN wafer used in this
study was 0.15 ␮m with 2.5:1 aspect ratio, and the structure of
Seoul National University assisted in meeting the publication costs of this
article.
substrate was CVD TiN ͑20 nm͒/Si ͑50 nm͒/SiO as shown in Fig.
2
9
. The experimental conditions including 0.1 g/L PdCl activation
References
2
were the same as those in the bare wafer experiment. In patterned
wafer, the film thickness was about 45 nm, which is the desired node
thickness to 0.1 ␮m cell size DRAM application with the consid-
1
. J. W. Lee, K.-M. Kim, H.-S. Song, K.-C. Jeong, J. M. Lee, and J. S. Rho, Jpn. J.
Appl. Phys., Part 1, 40, 5201 ͑2001͒.
2
2. S. Ezhilvalavn and T.-Y. Tseng, Mater. Chem. Phys., 65, 227 ͑2000͒.
3
. H. N. Al-Shareef, X. Chen, D. J. Lichtenwalner, and A. I. Kingon, Thin Solid
Films, 280, 265 ͑1996͒.
eration of the thickness of Ta O and it of top electrode. And the
2
5
deposition rate was 7.5 nm/min, which is the same value in the bare
wafers. The pattern was well covered with Ru film throughout a
4
. K. L. Saenger, G. Costrini, D. E. Kotecki, K. T. Kwietniak, and P. C. Andricacos,
J. Electrochem. Soc., 148, C758 ͑2001͒.
Downloaded on 2015-03-09 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C132
Journal of The Electrochemical Society, 151 ͑2͒ C127-C132 ͑2004͒
5
. Y.-S. Kim, R. H. Rampersad, and G. R. Tynan, Jpn. J. Appl. Phys., Part 2, 37, L502
1998͒.
13. H. A. Gasteiger, N. Markovic, P. N. Ross, and E. J. Cairns, J. Phys. Chem., 97,
͑
12020 ͑1993͒.
6
7
. T. Aoyama and K. Eguchi, Jpn. J. Appl. Phys., Part 2, 38, L1134 ͑1999͒.
. E.-S. Choi, J.-C. Lee, J.-S. Hwang, and S.-G. Yoon, Jpn. J. Appl. Phys., Part 1, 38,
5
14. Y. S. Chang and M. L. Chou, J. Appl. Phys., 69, 7848 ͑1991͒.
15. J. Scherzer and A. M. Weisberg, U.S. Pat. 3,530,049 ͑1970͒.
16. J. N. Croshy, U.S. Pat. 4,082,625 ͑1978͒.
17. T. Jones, Met. Finish., 99, 6, 121 ͑2001͒.
18. A. Radisic, J. G. Long, P. M. Hoffmann, and P. C. Searson, J. Electrochem. Soc.,
148, C41 ͑2001͒.
19. L. Graham, C. Steinbruchel, and D. J. Duquette, J. Electrochem. Soc., 149, C390
͑2002͒.
20. K. M. Takahashi, J. Electrochem. Soc., 147, 1414 ͑2000͒.
21. A. S. Arico, S. Srinivasah, and V. Antonucci, Fuel Cells, 1, 133 ͑2001͒.
22. K. L. Ley, R. Liu, C. Pu, Q. Fan, N. Leyarovska, C. Segre, and E. S. Smotkin, J.
Electrochem. Soc., 144, 1543 ͑1997͒.
317 ͑1999͒.
. T. Aoyama, M. Kiyotoshi, S. Yamazaki, and K. Eguchi, Jpn. J. Appl. Phys., Part 1,
8, 2194 ͑1999͒.
. J. J. Kim, D. H. Jung, M. S. Kim, S. H. Kim, and D. Y. Yoon, Thin Solid Films,
09, 28 ͑2002͒.
0. J. Lin, N. Masaaki, A. Tsukune, and M. Yamada, Appl. Phys. Lett., 74, 2380 ͑1999͒.
8
9
3
4
1
1
ˇ
1. V. Horvat-Rado sˇ evi c´ , K. Kvastek, M. Vukovi c´ , and D. Cukman, J. Electroanal.
Chem., 482, 188 ͑2000͒.
1
2. K. A. Friedrich, K.-P. Geyzers, A. Marmann, U. Stimming, and R. Vogel, Int. J.
Res. Phys. Chem., 208, 137 ͑1999͒.
Downloaded on 2015-03-09 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).