Evaluation of titanium silicon nitride as gate electrodes for complementary
metal-oxide semiconductor
H. Luan, H. N. Alshareef, H. R. Harris, H. C. Wen, K. Choi et al.
Citation: Appl. Phys. Lett. 88, 142113 (2006); doi: 10.1063/1.2188380
View online: http://dx.doi.org/10.1063/1.2188380
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v88/i14
Published by the American Institute of Physics.
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 02 Apr 2012 to 203.237.57.191. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
APPLIED PHYSICS LETTERS 88, 142113 共2006兲
Evaluation of titanium silicon nitride as gate electrodes for complementary
metal-oxide semiconductor
H. Luan,a兲 H. N. Alshareef,b兲 H. R. Harris,c兲 H. C. Wen, K. Choi, Y. Senzaki,
P. Majhi, and B.- H. Leed兲
Sematech, 2706 Montopolis Drive, Austin, Texas 78741
共Received 11 November 2005; accepted 22 February 2006; published online 7 April 2006兲
The effect of deposition temperature and film thickness on the work function of TiSiN gate
electrodes has been studied. It is shown that the work function of TiSiN can be tuned from 4.28–
4.74 eV on SiO2, 4.40–4.79 eV on HfO2, and 4.44–4.83 eV on HfSiOx. For high-k dielectrics, the
work function can be tuned by 200 meV on each side of the band gap, making it a suitable electrode
for fully depleted silicon-on-insulator devices. Furthermore, TiSiN deposition at high temperature
increases the work function to 4.87 eV while Si implantation increases it to 4.93 eV, making TiSiN
a good p-type metal candidate. © 2006 American Institute of Physics. 关DOI: 10.1063/1.2188380兴
Metal gate electrodes have been evaluated as replacements for polycrystalline silicon 共poly-Si兲 gates to continue
scaling the equivalent oxide thickness 共EOT兲 of metal-oxidesemiconductor field-effect transistors 共MOSFETs兲. In part,
this focus is necessary to address the issue of the undesirable
series capacitance associated with poly-Si depletion and to
circumvent performance degradation related to boron penetration.1,2 Consequently, an appreciable research effort is underway to identify the most appropriate n- and p-type metal
electrode materials that exhibit band-edge work functions,
threshold voltage 共Vt兲 stability, and chemical stability with
the underlying dielectric as a function of temperature. Concerns pertaining to materials interactions, impurity diffusivity, minimization of layer stress mismatch effects, and etch
characteristics support the notion that an amorphous metal
electrode structure is preferred. Many pure metals and metal
alloy films have recently been proposed as candidates for
dual metal gate applications, including bimetal alloys,
binary/ternary metal nitrides, metal silicon nitrides, conductive metal oxides, and metal silicide films.3–6 In this letter,
we report on the evaluation of chemical vapor deposited
共CVD兲 TiSiN as a gate electrode and identify the dependence
of the effective work function on deposition temperature and
film thickness.
To investigate the influence of CVD TiSiN electrodes on
the electrical properties of MOS devices, capacitors were
fabricated on heavily doped p-type substrates. Terraced oxide SiO2 was used as the sole dielectric layer to provide a
comprehensive dielectric thickness series for the efficient
and reliable work function extraction from a single wafer for
each electrode material being evaluated.7 The TiSiN films
were deposited by CVD with tetrakis共diethylamido兲 titanium
as the Ti precursor. SiH4 and NH3 were used as the Si and N
sources. A 150 nm poly-Si layer 共phosphorus-doped兲 was
then deposited on top of the TiSiN layer followed by annealing at 1000 ° C for 5 s. Three nm thick 共physical thickness兲
HfSiOx films 共30 mol % SiO2兲 or 2 nm thick HfO2 films
were deposited by atomic layer depositionon the terraced
oxide wafers for comparison with TiSiN/ SiO2. Work functions of the TiSiN electrode were extracted by plotting the
flatband voltage 共Vfb兲 versus the EOT of the capacitors. Vfb
and EOT values were extracted by fitting the capacitancevoltage 共C-V兲 curves to the conserved vector current
program.8 The film stress was calculated by the wafer bow
difference before and after film deposition on the wafers. All
the work function data presented in this letter were calculated after the 1000 ° C for 5 s thermal budget.
Figure 1 shows the dependence on film thickness of the
CVD TiSiN work function. For these gate dielectrics, the
work function increases with film thickness and saturates
when the thickness reaches about 20 nm. The work function
changes from 4.28 to 4.74 eV on SiO2, 4.40 to 4.79 eV on
HfO2, and 4.44 to 4.83 eV on HfSiOx, when the thickness
increases from 2.5 to 20 nm. The C-V curves for 2.5 and
20 nm TiSiN/ HfSiOx are shown in the inset in Fig. 1. Both
C-V curves look well behaved with no evidence of any distortion. The Vfb difference in the C-V curves is consistent
with work function differences for the two samples. The
similar thickness dependence of work function was observed
for the TiN gate electrode case.9 In addition, Fig. 2 shows the
Vfb vs EOT plots for TiSiN/ HfSiOx with different TiSiN
a兲
Author to whom correspondence should be addressed; eletronic mail:
hongfa.luan@sematech.org; Infineon assignee.
b兲
TI assignee.
c兲
AMD assignee.
d兲
IBM assignee.
0003-6951/2006/88共14兲/142113/3/$23.00
FIG. 1. Work function of CVD TiSiN on SiO2, HfO2, and HfSiOx vs film
thickness.
88, 142113-1
© 2006 American Institute of Physics
Downloaded 02 Apr 2012 to 203.237.57.191. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
142113-2
Luan et al.
Appl. Phys. Lett. 88, 142113 共2006兲
FIG. 2. Vfb vs EOT plots of CVD TiSiN on HfSiOx with different thicknesses deposited at 340 ° C.
thicknesses. Note that the data show an excellent linear fit
and small interface fixed charge for all thicknesses, suggesting an accurate work function extraction methodology. There
are several possible mechanisms that must be considered to
explain the thickness dependence of the effective work function of TiSiN. These include 共1兲 fixed or mobile charge
changes in the high-k film, and 共2兲 composition changes in
TiSiN with thickness. The first possibility, charges in the
high-k film increase work function as the TiSiN thickness
increases, can be excluded. An experiment with different
high-k thickness shows that the high-k thickness has little
effect on the work function of TiSiN. The density of bulk
charge in the high-kfilm calculated by modeling the relation
of effective work function with the thickness of high-k film
from this experiment is lower than 1 ⫻ 1018 cm−3.
The second possibility is that the TiSiN film composition
changes with TiSiN film thickness, thereby increasing work
function. To test this possibility, x-ray photoelectron spectroscopy analysis was done to measure the composition of
TiSiN; the results are summarized in Fig. 3. Figure 3 shows
that the Ti/ Si ratio changed as the TiSiN film thickness was
increased even though no major changes in the bonding energies were detected. Specifically, the amount of Ti in the
film decreased and the Si content increased with increasing
film thickness. Both Auger and secondary-ion-mass spectroscopy 共SIMS兲 show that Ti, Si, and N equally distribute
through the film 共see Fig. 4兲. This can be explained by
FIG. 3. Ti and Si concentrations vs film thickness.
FIG. 4. SIMS results from TiSiN film.
1000 ° C annealing after poly-Si implantation. It is possible
that the reduction in Ti metal concentration may have contributed to the work function increase since Ti metal has a
relatively low vacuum work function 共4.3 eV兲.
Figure 5 shows the dependence on deposition temperature of the 20 nm TiSiN work function. The work function increases with deposition temperature on both HfO2
and HfSiOx, but it tends to saturate at ⬃4.9 eV at 400 ° C.
The higher deposition temperature will generate stronger
chemical bonds and higher film density, thereby leading to a
higher work function. The Misra group10,11 and the Kwong
group12,13 also reported the effective work function increase
of metal Ru, TaSiN, TaN, and HfN by high temperature post
metal-deposition anneal. Another approach that was evaluated to increase the TiSiN work function was Si implantationinto the TiSiN film. This was done based on the earlier
result in Fig. 3 that showed that the work function increases
with Si content in the TiSiN electrode. Figure 6 shows the
result of the implant study, where the work function of TiSiN
could be increased to 4.93 eV. The implantation was done by
Si+ with energy of 10 keV and a dose of 5 ⫻ 1015 cm−3 on
25 nm TiSiN. The results show that CVD TiSiN with ex-
FIG. 5. Work function of 20 nm CVD TiSiN on HfO2 and HfSiOx vs deposition temperature.
Downloaded 02 Apr 2012 to 203.237.57.191. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
142113-3
Appl. Phys. Lett. 88, 142113 共2006兲
Luan et al.
The authors thank Melvin Cruz and Kam Hettiaratchi for
technical assistance.
1
FIG. 6. Vfb vs EOT plots of 20 nm CVD TiSiN implanted with Si on
HfSiOx.
pressed structure and composition is a good candidate for
p-type metal on HfSiOx.
In conclusion, the work function of CVD TiSiN can be
tuned by changing the film thickness. The results show CVD
TiSiN can be a good candidate for silicon-on-insulator devices where a tunable midgap work function is needed. In
addition, the work function of CVD TiSiN increases with
deposition temperature, but tends to saturate at 400 ° C. Si
implanted into TiSiN will also increase its work function. All
these results show that CVD TiSiN is a good p-type metal
candidate.
J. K. Schaeffer, S. B. Samavedam, D. C. Gilmer, V. Dhandapani, P. J.
Tobin, J. Mogab, B.-Y. Nguyen, B. E. White, S. Dakshina-Murthy, R. S.
Rai, Z.-X. Jiang, R. Martin, M. V. Raymond, M. Zavala, L. B. La, J. A.
Smith, R. Garcia, D. Roan, M. Kottke, and R. B. Gregory, J. Vac. Sci.
Technol. B 21, 11 共2003兲.
2
P. S. Lysaght, J. J. Peterson, B. Foran, C. Young, G. Bersuker, and H. R.
Huff, Mater. Sci. Semicond. Process. 7, 259 共2004兲.
3
J. H. Lee, Y. S. Suh, G. Heuss, H. Lazar, R. Jha, J. Gurganus, Y. Lin, and
V. Misra, Tech. Dig. - Int. Electron Devices Meet., 2003, 323.
4
C. S. Kang, H. J. Cho, Y. H. Kim, R. Choi, K. Onishi, A. Shahriar, and J.
C. Lee, J. Vac. Sci. Technol. B 21, 2026 共2003兲.
5
Y. S. Suh, G. Heuss, H. Zhong, S. N. Hong, and V. Misra, VLSI Des.,
2001, 47.
6
J. H. Sim, H. C. Wen, J. P. Lu, and D. L. Kwong, IEEE Electron Device
Lett. 24, 631 共2003兲.
7
G. A. Brown, H. C. Wen, G. Smith, J. Saulters, P. Majhi, and B. H. Lee,
IEEE SISC 共2004兲.
8
J. J. Hauser and K. Ahmed, Characterization and Metrology for ULSI
Technology 共AIP, New York, 1998兲.
9
K. Choi, H.-C. Wen, H. Alshareef, R. Harris, P. Lysaght, H. Luan, P.
Majhi, and B. H. Lee, ESSDERC 共2005兲, p. 101.
10
J. H. Lee, H. Zhong, Y. S. Suh, G. Heuss, J. Gurganus, B. Chen, and V.
Misra, IEDM, 2002, 359.
11
Y. S. Suh, G. Heuss, J. H. Lee, and V. Misra, IEEE Electron Device Lett.
24, 439 共2003兲.
12
C. Chen, H. Y. Yu, J. F. Kang, Y. T. Hou, M.-F. Li, W. D. Wang, D. S. H.
Chan, and D. L. Kwong, IEEE Electron Device Lett. 25, 123 共2004兲.
13
H. Y. Yu, M. F. Li, and D. L. Kwong, IEEE Electron Device Lett. 51, 609
共2004兲.
Downloaded 02 Apr 2012 to 203.237.57.191. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions