Thermal stability of chemical vapor deposition grown W and WNx thin films

in low-k integration structure

Sungho Jeon and Kijung Yonga兲
Surface Chemistry Laboratory of Electronic Materials, Pohang University of Science
and Technology (POSTECH), Pohang 790-784, Korea
Sung-Gyu Park and Shi-Woo Rhee
Laboratory for Advanced Molecular Processing, Department of Chemical Engineering,
Pohang University of Science and Technology (POSTECH), Pohang 790-784, Korea
共Received 23 November 2005; accepted 17 April 2006; published 12 May 2006兲
The Cu/barrier/low-k SiCOH / Si structures were fabricated and their thermal stability was
investigated. SiCOH films were deposited by plasma-enhanced chemical vapor deposition using
divinyldimethylsilane 共DVDMS兲 and O2. As barrier materials, tungsten and tungsten nitride films
were deposited by chemical vapor deposition using W共CO兲6 and NH3 sources at 450 ° C. Variations
of scanning electron microscopy and x-ray diffraction results of Cu/barrier/low-k SiCOH / Si were
examined depending on the annealing temperature. Both results showed that W and W2N film were
stable up to 500 and above 600 ° C, WO3 nanorods were grown from the sample surface. It is
thought that the thermal stability of the Cu/barrier/SiCOH / Si structure is closely related with the
thermal destruction of the W and WNx films. © 2006 American Vacuum Society.
关DOI: 10.1116/1.2203638兴

I. INTRODUCTION mal stability of the integrated structure. SiCOH films were

As the feature size of semiconductor devices such as ran-
dom access memory and microprocessor unit shrinks to
nanoscale dimensions, there has been a pressing demand for
Cu interconnect due to lower resistivity, higher resistance to
electromigration, and stress migration compared to Al.1–4
However, in the Cu interconnect, a reliable diffusion barrier
is required to retard Cu diffusion because copper is easy to
diffuse into silicon well and dielectric layers. A thin barrier
layer should prevent diffusion and reaction between the Cu
overlayer and the low-k dielectric layer in a stable Cu/low-k
contact system. For these reasons, various materials such as
TiN, TiSiN, TaN, Ta, TaCx, TaSiN, WCx, and WNx have
been investigated as diffusion barrier for Cu interconnect.5–10
Most of these films were deposited on SiO2 or Si substrates
to test barrier properties. Compared to other barrier candi-
dates, the WNx film has several good barrier properties such
as thermal reaction inertness with Cu, relatively low resistiv-
ity, possibility of simple CMP process, and availability of
precursors for chemical vapor deposition 共CVD兲 application.
These characteristics make tungsten nitride one of strong
candidates for the Cu diffusion barrier.11
To reduce resistance-capacitance 共RC兲 time delay in Cu
interconnect, another requirement is the integration with low
dielectric constant 共low-k兲 materials.12,13 Though numerous
low-k alternatives exist today, much attention has focused on
CVD-based carbon-doped silicon oxide 共SiCOH兲 composite
films due to their low dielectric constant with good thermal
and mechanical stabilities.14
In this study, we fabricated the integration structure of
Cu/diffusion-barrier/low-k SiCOH / Si and studied the ther-
a兲
Author to whom all correspondence should be addressed; electronic mail:
kyong@postech.ac.kr
deposited on Si using plasma-CVD process. As diffusion bar-
rier material, CVD-grown W and WNx films were tested for
this study. In this particular low-k integration structure, CVD
process integration was also realized for the growth of diffu-
sion barrier and low-k materials.
II. EXPERIMENT
The low-k SiCOH films were deposited on Si共100兲 at
40 W, 100 ° C by plasma-enhanced CVD 共PECVD兲 using
divinyldimethylsilane 共DVDMS兲 and O2. A capacitively
coupled plasma reactor was used where both O2 and
DVDMS were introduced into the reactor through the show-
erhead, which is also an upper electrode with 6 in. diameter.
Both gases were flown vertically toward the substrate on the
bottom electrode. The film was deposited on 共100兲 oriented
p-type silicon substrates. DVDMS flow rate was fixed at
10 SCCM 共SCCM denotes cubic centimeter per minute at
STP兲 and O2 flow rate was varied with total flow rate ad-
justed at 210 SCCM with helium. The chamber pressure re-
mained constant at 1 torr. Finally as-deposited SiCOH films
were annealed at 450 ° C for 1 h in Ar ambient.
The WNx films were deposited on Si共100兲 at 450 ° C by
CVD using W共CO兲6 and NH3 precursors in a vertical-type
cold wall reactor. The tungsten 共W兲 film was also deposited
by CVD using W共CO兲6 in the same system. A typical film
thickness of W and WNx was about 50 nm.
In order to investigate the barrier properties, the
Cu/ barrier/ SiCOH / Si structures were fabricated. A 60 nm
thick Cu film was deposited by evaporation onto the pre-
pared barrier materials. The annealing experiments were per-
formed for 1 h in argon ambient. After each annealing ex-
periment, scanning electron microscopy 共SEM兲 images,
X-ray diffraction 共XRD兲 patterns, and four-point probe were

1428 J. Vac. Sci. Technol. B 24„3…, May/Jun 2006 1071-1023/2006/24„3…/1428/4/$23.00 ©2006 American Vacuum Society 1428
1429 Jeon et al.: Thermal stability of CVD grown W and WNx thin films
FIG. 1. SEM micrographs and XRD patterns of as-deposited Cu/barrier/
SiCOH / Si structure: 共a兲 SEM and 共b兲 XRD of Cu/ W / SiCOH / Si, respec-
tively; 共c兲 SEM and 共d兲 XRD of Cu/ WNx / SiCOH / Si, respectively.
examined. Field emission scanning electron microscopy 共FE-
SEM兲 was used for the film thickness measurements and the
cross-sectional structure imaging. XRD was used for the dif-
fusion barrier property analysis by measuring change of the
crystallographic structure. Four-point probe was used for the
diffusion barrier property analysis by measuring the sheet
resistance, respectively.
III. RESULTS AND DISCUSSION
Figure 1 shows the SEM images and XRD patterns of the
Cu/ barrier/ SiCOH / Si structure, which is fabricated for the
thermal stability test. A uniform, thin barrier film is shown
between Cu and the SiCOH / Si substrate at Figs. 1共a兲 共W兲
and 1共c兲 共WNx兲. Figure 1共b兲 showed the definite Cu共111兲,
Cu共200兲, and W共110兲 peaks. Figure 1共d兲 showed the
Cu共111兲, Cu共200兲, W2N共200兲, W2N共111兲, and W2N共220兲
peaks. The SiCOH films showed a low dielectric constant of
2.2 at an optimum growth and annealing conditions 共not
shown here兲.
Thermal stability of the Cu/ barrier/ SiCOH / Si structure
was investigated by SEM and XRD analysis. Figure 2 shows
the SEM images and XRD patterns of the Cu/ W / SiCOH / Si
films annealed at different annealing temperatures for 1 h at
Ar atmosphere. The SEM images show that the
Cu/ W / SiCOH / Si structure is stable up to 500 ° C. However,
a drastic morphology change of the integration structure was
observed for annealing temperatures above 600 ° C. The
roughness of Cu surface was increased and nanorods ap-
peared on the surface of the 600 ° C. Figure 2共e兲 clearly
shows Cu agglomerates and nanorods produced from the
tungsten film. The XRD analysis results indicated that the
grown nanorods were tungsten oxide nanorods. For the as-
fabricated sample, the definite Cu 共111兲, Cu 共200兲, and
W共110兲 peaks were observed. The intensity of W peaks de-
creased with the increase of annealing temperature up to
600 ° C. On the other hand, the WO3 peak appeared after
JVST B - Microelectronics and Nanometer Structures
1429
FIG. 2. 关共a兲, 共c兲, and 共e兲兴 Variations in SEM image and 关共b兲, 共d兲, and 共f兲兴
XRD pattern of Cu/ W / SiCOH / Si as a function of the annealing tempera-
tures: 关共a兲 and 共b兲兴 400 ° C, 关共c兲 and 共d兲兴 500 ° C, and 关共e兲 and 共f兲兴 600 ° C.
thermal annealing at 500 ° C and the intensity of WO3 peak
increased with increasing annealing temperature up to
600 ° C. As reflected from XRD analysis, thermal annealing
has induced the main structure of W films undergoing a
change from W phase to WO3 phase. Judging from SEM
images, nanorods were generated at 600 ° C annealing tem-
perature and the grain size of Cu film increased to form Cu
agglomerates at this temperature. The XRD analysis results
implied that the grown nanorods were tungsten oxide. It
seems that WO3 nanorods grew up with Cu agglomeration
because nanorods could not grow if Cu film blocked W film.
The W film was annealed in Ar ambient without oxygen.
Since no additional oxygen was introduced into the system,
the origin of oxygen has not been proved. The growth
mechanism of the WO3 nanorods is under investigation.
Similar results were also found for the
Cu/ W2N / SiCOH / Si samples. Figure 3 shows the SEM im-
ages and XRD patterns of the Cu/ WNx/ SiCOH / Si films
annealed at different annealing temperatures for 1 h at Ar
atmosphere. SEM images show that the Cu/ W2N / SiCOH / Si
structure is stable up to 500 ° C and the nanorod formation is
observed for annealing above 600 ° C. These results showed
exactly the same trend as the Cu/ W / SiCOH / Si structure.
The Cu agglomerates and tungsten oxide nanorods were
found on the surface of the 600 ° C annealed sample. The
XRD analysis results indicated that the WO3 peak appeared
after thermal annealing at 500 ° C and the intensity of WO3
peak became stronger and sharper with increasing annealing
temperature up to 600 ° C. The XRD data showed consistent
results with SEM data.
1430 Jeon et al.: Thermal stability of CVD grown W and WNx thin films
FIG. 3. 关共a兲, 共c兲, and 共e兲兴 Variations in SEM image and 关共b兲, 共d兲, and 共f兲兴
XRD pattern of Cu/ WNx / SiCOH / Si as a function of the annealing tem-
perature: 关共a兲 and 共b兲兴 400 ° C, 关共c兲 and 共d兲兴 500 ° C, and 关共e兲 and 共f兲兴
600 ° C.
Sheet resistance measurements were also performed to in-
vestigate the thermal stability of the Cu/ barrier/ SiCOH / Si
structure. Figure 4 shows that the variations in sheet resis-
tances of Cu/ W / SiCOH / Si and Cu/ WNx/ SiCOH / Si struc-
tures as a function of annealing temperature for 1 h at Ar
atmosphere. The sheet resistances of both as-deposited
samples were between 0.8 and 0.9 ⍀ / sq. Both samples
showed that the average sheet resistance values initially de-
creased after annealing 共⬃400 ° C兲 with respect to that of the
as-deposited sample, as shown in the inset of Fig. 4. We
believe that this is due to grain growth and the defect anni-
hilation of the Cu film during annealing. The resistance in-
FIG. 4. Variations in sheet resistance of Cu/barrier/SiCOH / Si as a function
of the annealing temperature. Inset shows magnification of the results in low
temperature region.
J. Vac. Sci. Technol. B, Vol. 24, No. 3, May/Jun 2006
1430
creased abruptly after annealing above 600 ° C. Judging from
SEM images of both samples, nanorods were generated at
600 ° C annealing temperature and the grain size of Cu film
increased to form Cu agglomerates at this temperature. The
results of four-point probe showed consistent trend as in
SEM analysis. The destruction of Cu/ barrier/ SiCOH / Si
structure may induce the increase of sheet resistances. Our
results of W and W2N films showed lower barrier failing
temperatures than the reported values in literature.7 We be-
lieve this discrepancy might be due to the destruction of W
and W2N films.
For better understanding of the nanorod growth mecha-
nism, annealing experiments were performed for W / Si
samples. Figure 5 showed the SEM image and XRD pattern
of the grown tungsten oxide nanorods. These tungsten oxide
nanorods were produced simply by thermal annealing of
CVD-W / Si film for 1 h at 700 ° C in Ar atmosphere. This
result shows that tungsten oxide nanorods could grow up
without Cu and SiCOH. Since annealing was conducted in
Ar ambient, oxygen source may come from residual gas or
oxygen in the films. Recent study of tungsten oxide nanowire
growth showed that they could grow WO3 nanowires by
heating W filament in high vacuum conditions 共10−7 torr兲.15
Also they found that the nanowire length was not extended
by extending growth time, which may support that the oxy-
gen source comes from the film. However, we still cannot
rule out the possible role of residual oxygen gas for the
growth of tungsten oxide nanorods. We also performed an-
nealing experiments in O2 gas flow, but in these cases only
WOx films were obtained without nanorod products. The
same results were observed for W2N / Si samples. Currently,
we are investigating the growth mechanism and further char-
acterization of nanorods grown from the integrated samples.
IV. CONCLUSIONS
Tungsten and tungsten nitride deposited by CVD have
been studied for thermal stability in the
Cu/ barrier/ SiCOH / Si structures. After annealing, SiCOH
films showed a low dielectric constant of 2.2 at an optimum
condition. The low-k integration structure was stable up to
500 ° C for 1 h annealing. After 600 ° C annealing, the Cu
agglomerates and tungsten oxide nanorods were found on the
tungsten and tungsten nitride films. The WO3 XRD peak
appeared after thermal annealing at 500 ° C and the intensity
of WO3 peak became stronger and sharper with the increas-
ing annealing temperature up to 600 ° C. It is thought that the
thermal stability of the Cu/ barrier/ SiCOH / Si structure is
closely related with the thermal destruction of W and W2N
films. W and W2N films grown on Si substrates also pro-
duced tungsten oxide nanorods upon annealing about
600 ° C. Residual oxygen gas or oxygen in films is thought
as an oxygen source to produce tungsten oxide nanorods.
ACKNOWLEDGMENTS
This work was supported by Grant No. B1220-0401-0165
from the University Fundamental Research Program of the
Ministry of Information & Communication on Republic of
1431 Jeon et al.: Thermal stability of CVD grown W and WNx thin films 1431

FIG. 5. SEM micrograph and XRD pattern of WO3 nanorods grown on Si.

7
Korea and Grant No. RTI 04-01-04 from the Regional Tech- B. H. Lee and K. Yong, J. Electrochem. Soc. 151, C594 共2004兲.
8
nology Innovation Program of the Ministry of Commerce, G. Erdelyi et al., Thin Solid Films 459, 303 共2004兲.
9
L. Y. Yang, D. H. Zhang, C. Y. Li, and P. D. Foo, Thin Solid Films 462,
Industry and Energy 共MOCIE兲.
176 共2004兲.
10
1 Y. H. Shin and Y. Shimogaki, Sci. Technol. Adv. Mater. 5, 399 共2004兲.
C. K. Hu, B. Luther, F. B. Kaufman, J. Hummel, C. Uzoh, and D. J. 11
N. Gonohe, Mater. Trans., JIM 43, 1585 共2002兲.
Pearson, Thin Solid Films 262, 84 共1995兲. 12
2 M. Morgen, E. T. Ryan, J. H. Zhao, C. Hu, T. Cho, and P. S. Ho, Annu.
R. J. Gutmann, T. P. Chow, A. E. Kaloyeros, W. A. Lanford, and S. P.
Muraka, Thin Solid Films 262, 177 共1995兲. Rev. Mater. Sci. 30, 645 共2000兲.
13
3
L. Vanasupa, Y.-C. Joo, and P. R. Besser, J. Appl. Phys. 85, 2583 共1999兲. T. Homma, Thin Solid Films 278, 28 共1996兲.
14
4
K. Kim and K. Yong, Electrochem. Solid-State Lett. 6, C106 共2003兲. K. Jeong, S. Park, and S. Rhee, J. Vac. Sci. Technol. B 22, 2799 共2004兲.
5 15
C. H. Winter, Aldrichmica Acta 33, 3 共2000兲. C. Klinke, J. B. Hannon, L. Gignac, K. Reuter, and P. Avouris, J. Phys.
6
S. J. Wang, H. Y. Tsai, and S. C. Sun, Thin Solid Films 394, 180 共2001兲. Chem. B 109, 17787 共2005兲.

JVST B - Microelectronics and Nanometer Structures