CHIN. PHYS. LETT. Vol. 30, No. 3 (2013) 038503
Improvement of Ni Silicide Thermal Stability By Using Vanadium Elements
*
LIU Hai-Long(刘海龙)1 , LIU Yan(刘艳)2 , LIU Min(刘敏)3 , WANG Tao(汪涛)4** , A. Tuya5
1
2
Department of Radiology, Zhejiang Province Tongde Hospital, Hangzhou 310012
Department of Radiology, Second Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou 312000
3
Department of Radiology, The Second Hospital, Shaoxing 312000
4
College of Electrical Engineering, Zhejiang University, Hangzhou 310012
5
Department of Electronics Engineering, Chungnam National University,
220 Gung-dong, Yuseong-gu, Daejeon 305-764, Korea
(Received 6 November 2012)
Ni silicide thermal stability is improved by the use of a Ni-V (nickel vanadium) alloy target. The relationship between the formation temperature and the thermal stability of Ni silicide is investigated. The sheet resistance after
the formation of Ni silicide with the Ni-V shows stable characteristics up to a rapid-thermal-process temperature
of 700∘C, while degradation of sheet resistance starts at that temperature in the case of pure-Ni. Moreover, the
thermal stability improvement is demonstrated by the post-silicidation annealing. It is considered that the thermal robustness of Ni-V silicide is highly dependent on the formation temperature. With the increasing silicidation
temperature (around 700∘C), more thermally stable Ni silicide is formed in comparison to the low-temperature
case using the Ni-V. A Ni-V alloy target is utilized to form Ni silicide. The V and the V-trap complexes are
explained to block the transformation from NiSi to NiSi2 so as to improve the Ni silicide thermal stability.
PACS: 85.30.Tv, 73.40.Gk
DOI: 10.1088/0256-307X/30/3/038503
Self-aligned silicide (SALICIDE) is one of the most
indispensable techniques for high-performance CMOS
logic devices. In real IC processes, it has experienced
the evolution from TiSi2 to CoSi2 , and from CoSi2
to NiSi. NiSi has several advantages over TiSi2 and
CoSi2 for ultra-large scale integration (ULSI) technology: low temperature silicidation process, low silicon
consumption, no bridging failure property, smaller mechanical stress, no narrow line width effect on sheet
resistance, smaller contact resistance for both n- and
p-Si,[1] and higher activation rate of B for SiGe poly
gate electrodes.[2] However, NiSi has poor thermal stability, which turns out to be its barrier for application
in nano-scale MOSFETs. Many attempts have been
made to improve the thermal robustness of Ni silicide, such as by ion implantation, application of alloying target materials, and deposition of inter- and/or
capping-layers.[3−7] The understanding of single metal
metallization is not enough for device metallization
schemes. The investigation of alloyed silicide formation will help to gain a better understanding of multilevel metallization reactions. As the same transition
metal V, there have been few papers about V influence
on process and devices till now.[8−11] In this Letter, V
in NiSi performance will be discussed and trigger its
new potential usage in IC self-alignment silicide processes. The formation temperature dependence of the
thermal stability of Ni silicide with Ni-V alloy is investigated for use in nanoscale CMOS technology. Ni
silicide with Ni-V indicates a more thermally stable
characteristic in comparison with pure-Ni silicide. The
thermal robustness of Ni silicide with Ni-V is highly
dependent on the formation temperature.
Ni/TiN (10/25 nm) and Ni-V/TiN (10/25 nm, V
atomic 5%) layers were deposited on the p-type Si
wafer using an rf magnetron sputtering system in Ar
ambient. During the deposition, the sample holder
was rotated to deposit the metals uniformly. TiN
capping is applied to prevent the abnormal oxidation
of NiSi.[4] Nickel silicide formation was achieved by
using a rapid thermal process (RTP) at five different temperatures (400, 500, 600, 700 and 800∘C) for
30 s to investigate the thermal stability dependence
on the formation temperature. To test the thermal
stability of silicides, samples were furnace annealed in
three different temperatures (600, 650 and 700∘C) for
30 min. Sheet resistance was measured using a conventional four-point probe (FPP). Atomic force microscopy (AFM) and x-ray diffraction (XRD) analysis
were performed to investigate surface roughness and
the silicide phase of the silicides formed with pure-Ni
and Ni-V, respectively.
Nickel silicides were formed using two different target materials of pure-Ni and Ni-V with thickness of
10 nm on p-type silicon wafer. Figure 1 shows the
sheet resistance of nickel silicide as a function of the
formation temperature. It can be seen that a sharp
drop in the sheet resistance is already achieved at
* Supported by the National Natural Science Foundation of China under Grant No 61176101, the Ph.D. Programs Foundation of Ministry of Education of China (20120101120156), and the Fundamental Research Funds For the Central Universities
(2012FZA4016).
** Corresponding author. Email: [email protected]
© 2013 Chinese Physical Society and IOP Publishing Ltd
038503-1
CHIN. PHYS. LETT. Vol. 30, No. 3 (2013) 038503
XRD is used to identify the silicide phases as presented in Fig. 3. Almost identical profiles are obtained
in the pure-Ni and Ni-V cases. However, in the case
of Ni-V, less NiSi2 peaks are observed, reflecting better thermal stability than the pure-Ni case. At RTP
temperature of 700∘C, high resistance of NiSi2 is observed. The existence of V has restrained the change
from NiSi to NiSi2 . Thus fewer NiSi2 peaks are shown
in Fig. 3.
200
40
Ni/TiN
35
Ni-V/TiN
30
700
600
120
500
80
pure-Ni
Sheet resistance (Ω/sq.)
45
160
Ni-V
RTP (C)
Intensity (arb. units)
400∘C, which is the lowest temperature considered
in this study. It is found that the nickel silicide
with pure-Ni has a relatively poor thermal stability
than that with Ni-V during the silicide formation.
Based on the results from previous studies on thermal stability, the observed increase of sheet resistance
is suggested to be of nickel film agglomeration and
phase transition from low resistive nickel mono-silicide
(NiSi) to high resistive nickel di-silicide (NiSi2 ) at high
temperatures.[4] After annealing at 400∘C, the resistance becomes very small, as shown in Fig. 1. The
reason is that Ni-rich Ni3 V alloy has been formed first
at low temperature when the temperature is not high
enough to promote large diffusion.
700
600
40
25
500
0
20
30
40
50
60
Fig. 3. XRD scans for pure-Ni and Ni-V silicide for RTP
temperatures of 500–700∘C (squares for NiSi and circles
for NiSi2 ).
10
5
0
20
2θ (deg)
15
400
500
600
700
800
RTP temperature (C)
(b)
(a)
Fig. 1. Nickel silicide sheet resistance as a function of the
formation temperature for both silicides with pure-Ni and
Ni-V alloy.
45
Sheet resistance (W/sq.)
40
35
30
Ni/TiN (RTP at 800C)
300 nm
300 nm
NiV/TiN (RTP at 600C)
Fig. 4. Cross-sectional FESEM images of nickel silicides
formed with (a) Ni and (b) Ni-V after RTP at 700∘C.
NiV/TiN (RTP at 700C)
NiV/TiN (RTP at 800C)
25
20
15
Increasing RTP temperature
10
5
0
600
650
700
Annealing temperature (C)
Fig. 2.
Nickel silicide sheet resistance after postsilicidation annealing at different temperatures for 30 min.
After the post-silicidation annealing shown in
Fig. 2, the nickel silicide sheet shows the thermal stability characteristic with different annealing temperatures for 30 min. After the furnace annealing, nickel
silicides with Ni-V show quite good thermal robustness while pure-Ni samples all are almost failed to
record reasonable sheet resistance using FPP. Moreover, it is shown that thermal stability of nickel silicides depends strongly on the formation temperature
of silicides. Higher RTA temperature leads to better
thermal stability, as shown in Fig. 2.
Field-emission secondary-electron microscopy
(FESEM, S-4700, Jeonju branch of KBSI) images
for pure-Ni and Ni-V alloy after RTP at 700∘C are
shown in Figs. 4(a) and 4(b), respectively. Pure-Ni
silicide shows agglomeration which leads to degradation of sheet resistance in Fig. 1. Ni-V silicide formed
at the same temperature shows better silicide/silicon
interface profile.
Figure 5 shows the AFM surface roughness for
both silicides formed with pure-Ni and Ni-V alloy,
respectively. Silicide formed with Ni-V (rms roughness 1.2 nm) shows better surface roughness than that
formed with pure-Ni (rms roughness 2.1 nm) after
post-silicidation annealing at 650∘C for 30 min.
For the pure Ni, it is well-known that the increasing temperature results in NiSi transforming into
NiSi2 . In the case of Ni-V alloy, intermediate phase
Ni-rich Ni3 V is formed first during the intermixing
process. With increase temperature, Si is the dominant diffusion atom in the system. Finally, Ni3 V is
decomposed to form Ni(V)Si and finally transferred
038503-2
CHIN. PHYS. LETT. Vol. 30, No. 3 (2013) 038503
into Ni(V)Si2 , since NiSi2 and VSi2 are relatively more
stable than Ni3 V with rich-Si circumstance.
nm
30
(a)
nm
30
20
10
0
4.5
25
20
15
4
3.5 3
mm 2.5 2
1.5
1
0.5 0 0 0.5
2
1 1.5
2.5
3
4
3.5
4.5 10
5
mm
0
nm
(b)
20
nm
20
15
10
5
0
4.5
15
4
3.5
mm
3
2.5
2
1.5
1
0.5
0 0
0.5
1
1.5
2
4
3 3.5
2.5
mm
4.5
10
5
ble silicides in the V-Si system, for instance, VSi2 ,
V3 Si, V5 Si. Our annealing temperature is 700C and
the VSi2 formation temperature is 600∘C and silicides
are stable at high temperature (>900∘C). In our experiments without O, Si-rich circumstances tend to
form VSi2 . It can be deduced from the density of
VSi2 (4.62 g/cm) that the average size of precipitates
is 7 nm in diameter, with about 4000 vanadium atoms
in one pure VSi2 . Thus the post annealing at 650∘C
even with a long time of 30 min also does not influence NiV silicide and the surface of the silicide is still
smooth. If the V and trap concentrations are higher,
the precipitates may grow to a larger size, which may
cause agglomeration. On the other hand, the low V
dosage may provide enough accommodation ability for
V atoms. It is the reason why the lower V concentration in Ni-V alloy, only 5%, is settled in our experiments.
0
10-2
10-3
VD/⊲ V
10-4
VD/⊲ V
10-5
10-6
Solid : RTP
Open : Anneal
10-7
10-8
10-9
(a)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
10-2
LG=80 nm
10-3
Id (A/mm)
In the above experimental results, usage of the V
element is obvious to block the transformation from
NiSi to NiSi2 so that NiSi becomes more stable at
higher temperature. It is reported that the enthalpy
of formation of VSi2 is 3.2 eV.[8,9] The enthalpy values
of formation of NiSi and NiSi2 are 0.928 eV and 3.90 eV
at room temperature, respectively. Thus it is easier to
form VSi2 under the same annealing conditions than
NiSi2 . At the same V and Ni concentrations, the formation of silicide is more likely as the sequence of
NiSi>ViSi2 >NiSi2 .
The V ionic radius is 0.065 nm and the Si ionic
radius is 0.078 nm. The metallic radius of Ni is
0.1244 nm, which is smaller than that 0.1338 nm of V.
In view of atom size, Ni is likely to be easier to move.
However, the size effect is not a dominant effect, since
the electronic structure of the impurities results in a
large variation in diffusion coefficients. The atomic
diffusion of the Ni-V-Si system is not strongly dependent on the relative atomic sizes of the elements, but
rather on the structure of the compound.
V atoms are symmetrically distributed in Ni-V alloy and do not move since Si atoms have a much
higher diffusion constant than V atoms in VSi2 . During the annealing, Si and Ni atoms diffuse within each
other. Si atoms combine with V to form the V-trap
complexes, which are very stable even at temperature
higher then 1000∘C. It is believed that the impurities
are stable and are not mobile when they are trapped
at temperature at least as high as 700∘C. The V-trap
complexes block the diffusion and prevent the transformation from NiSi to NiSi2 . There are several sta-
Id (A/mm)
Fig. 5. AFM surface roughness for (a) Ni and (b) NiV silicides after post-silicidation annealing at 650∘ for 30
min (RTP at 700 ∘ ).
LG=80 nm
VD/֓⊲ V
10-4
10-5
VD/֓⊲ V
10-6
Solid : RTP
Open : Anneal
10-7
10-8
(b)
10-9
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
Vg (V)
Fig. 6. 𝐼𝑑 -𝑉𝑔 characteristic of 80 nm (a) NMOS and (b)
PMOS transistors silicided using Ni(V) alloy.
At low formation temperatures (≤600∘C), both
nickel silicides with pure-Ni and Ni-V show almost
the same thermal stability. Characteristics of silicides
formed with pure-Ni and Ni-V alloy change dramatically before and after the post-silicidation annealing.
Better thermal stability is achieved using vanadiumadded Ni and the thermal robustness of both silicides strongly depends on the formation temperature.
Therefore, nickel silicide with Ni-V alloy could be
promising material for next-generation CMOS technology, which has been demonstrated in our 90 nm
technology node process as shown in Fig. 6. The technology may be promising for sub 45 nm technology
nodes, which challenges the narrower distance from
038503-3
CHIN. PHYS. LETT. Vol. 30, No. 3 (2013) 038503
the polygate to source/drain and the tighter process
margin for Ni silicide piping.[12−16] The enhancement
of Ni silicide thermal stability is considered due to the
V-trap complexes blocking the transformation from
NiSi to NiSi2 . The content of V needs to be optimized
by further experiments.
References
[1] Iwai H, Ohguro T and Ohmi S I 2002 Microelectron. Eng.
60 157
[2] Liu J and Ozturk M C 2005 IEEE Trans. Electron Devices
52 1535
[3] Ohguro T, Nakamura S, Morifuji E, Ono M, Yoshitomi T,
Saito M, Momose H S and Iwai H 1995 Int. Electron. Devices Meeting 453
[4] Wang M Y, Wu C W, Lin C T, Hsieh C H, Shue W S and
Liang M S 2003 Symp. VLSI Tech. Dig. 157
[5] Chang J G, Wu C B, Ji X L, Ma H W, Yan F, Shi Y and
Zhang R 2012 Chin. Phys. Lett. 29 058501
[6] Ding T, Song J Q and Cai Q 2012 Chin. Phys. Lett. 29
036803
[7] Lavoie C, Heurle F, Detavernier C and Cabral C 2003 Microelectron. Eng. 70 144
[8] Andrews J M and Phillips J C 1975 Phys. Rev. Lett. 35 56
[9] Mann R W and Clevenger L A 1995 Properties of Metal
Silicides ed Maex K and Rossum M V (Stevenage: IEEE
INSPEC)
[10] Zhang P, Stevie F, Vanfleet R, Neelakantan R, Klimov M,
Zhou D and Chow L 2004 J. Appl. Phys. 96 1053
[11] Francois H, Anoshkina E, Stevie F, Chow L and Richardson
K 2001 J. Vac. Sci. Technol. B 19 1769
[12] Chen H F, Guo L X and Du H M 2012 Chin. Phys. B 21
088501
[13] Qin J R, Chen S M and Li D W 2012 Chin. Phys. B 21
089401
[14] Zhang J P, Li Z H, Zhang B, Liang B and Liu B W 2012
Chin. Phys. B 21 068504
[15] Ren M, Li Z H, Deng G M, Zhang L X, Zhang M, Liu X L,
Xie J X and Zhang B 2012 Chin. Phys. B 21 048502
[16] Li M, Yu X F and Xue Y G 2012 Acta Phys. Sin. 61 106103
(in Chinese)
038503-4