Вы находитесь на странице: 1из 5

Materials Letters 57 (2003) 2965 – 2969

www.elsevier.com/locate/matlet

Effect of in situ H2-plasma cleaning on TiSi2 film properties in


plasma enhanced chemical vapor deposition
Osama A. Fouad a,*, Masaaki Yamazato b, Hiroshi Ahagon b, Masamitsu Nagano b
a
Central Metallurgical Research and Development Institute (CMRDI), P.O. Box 87, Helwan, El-Felezat Street, Cairo, Egypt
b
Department of Applied Chemistry, Faculty of Science and Engineering, Saga University, Honjo 1, Saga 840-8502, Japan
Received 25 April 2002; received in revised form 15 November 2002; accepted 29 November 2002

Abstract

The effect of in situ H2-plasma cleaning on the properties of TiSi2 film was investigated. At high RF plasma power (350 W),
the film surface was rough and had voids due to damage of Si surface. At low RF power (30 W), the film surface was rough due
to residual silicon oxide on the Si substrate surface.
D 2002 Elsevier Science B.V. All rights reserved.

Keywords: Silicides; H2-plasma cleaning; Crystal structure; Electrical properties; Deposition; Surfaces

1. Introduction sity and improved the surface morphology of titanium


silicide films in an LPCVD process. Ilderem and Reif
Recently, a great interest in preparation of the low [6] deposited a very thin polysilicon layer immediately
resistivity TiSi2 thin film for the fabrication of ad- prior to silicide deposition to promote the uniform
vanced integrated circuits (ICs) via low-pressure nucleation of titanium silicide films.
chemical vapor deposition (LPCVD) has arisen [1,2]. Alternatively, few works had been done on dep-
However, LPCVD has shortcomings such as nuclea- osition of titanium silicide films using PECVD [5,7 –
tion, which causes high sheet resistance in thin films, 9] to improve the film properties and to overcome
large silicon consumption and large grain size of the the LPCVD problems. PECVD has improved the
resulting films [2,3]. The main reason for nucleation film surface morphology by forming fine silicide
problem was the native oxide layer on Si substrate grains [7]. However, the effect of in situ H2-plasma
surface [4,5]. Without Si substrate cleaning, the film cleaning conditions on silicide film growth from
either does not grow or grows only sparse nuclei [3]. TiCl4/H2 gas mixture and Si substrate has not been
Saito et al. [3] reported that in situ silane (SiH4) clarified yet.
cleaning of Si substrate greatly increased nuclei den- In our previous work, we investigated the prepa-
ration of TiSi2 film by PECVD using TiCl4/H2 gas
mixture without using the hazardous SiH4 gas [9]. By
* Corresponding author. Tel.: +20-2-5010643; fax: +20-2-
H2-plasma cleaning prior to silicide deposition, the
5010639. deposited film grew uniformly with homogenous sur-
E-mail address: oafouad@yahoo.com (O.A. Fouad). face and smooth interface. In this paper, the effect of

0167-577X/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0167-577X(02)01405-2
2966 O.A. Fouad et al. / Materials Letters 57 (2003) 2965–2969

in situ H2-plasma cleaning on the properties of the deionized water and dried in a stream of nitrogen gas
deposited TiSi2 film is explored in detail. prior to loading into the reaction tube. The time from
the substrate cleaning until loading in the reaction
tube was about 5 min. During that period, it was
2. Experimental expected that a thin oxide layer is formed on the
silicon substrate surface. After loading the substrate,
The PECVD system was described in detail else- the system was pumped down to a base pressure of
where [9]. In brief, the reactor was a quartz tube to 1  10 5 Torr. Prior to deposition, the substrate sur-
which H2 (99.99%) and TiCl4 (99.99%) were intro- face was in situ cleaned by hydrogen plasma. The H2-
duced and controlled by mass flow controllers (MFC) plasma cleaning was done at an H2 flow rate of 30
and needle valves. The plasma source was RF gen- sccm, a substrate temperature of 850 jC, a total
erator of 1 kW, 13.56 MHz. An electric resistance pressure of 5  10 3 Torr, a duration time of 1– 60
furnace was used to heat the substrate and its temper- min and at different RF powers from 30 to 350 W. The
ature was monitored by a Pt/Pt-13% Rh thermocou- TiSi2 films were deposited at an H2/TiCl4 flow rate
ple. The exhausted gases were pumped out using a ratio of 250, a substrate temperature of 850 jC, a
combination of rotary pump and oil diffusion pump. deposition time of 10 min, a total pressure of 0.5 Torr
The pressure was monitored by a Penning gauge and a and an RF power of 130 W.
capacitance manometer. Films were examined by atomic force microscopy
Titanium disilicide films were deposited on silicon (AFM), scanning electron microscopy (SEM), X-ray
(100) substrate. All substrates were first cleaned by diffraction (XRD) and X-ray photoelectron spectro-
dipping in HF solution for 20 s to remove the native scopy (XPS). The film resistance was measured using
oxide layer on Si substrate surface then rinsed with the DC four-probe method.

Fig. 1. AFM profiles of titanium silicide films deposited on Si (100) substrate after in situ H2-plasma cleaning at different RF powers. (a) 0 W,
(b) 30 W, (c) 100 W and (d) 350 W.
O.A. Fouad et al. / Materials Letters 57 (2003) 2965–2969 2967

3. Results and discussion

AFM profiles of titanium silicide films deposited


on Si substrate after cleaning at different RF powers
(0– 350 W) are shown in Fig. 1. At a low RF power
(0– 30), the deposited film had a rough surface and the
grain density was higher and the grain size was in the
range from about 0.2 to 0.5 Am and formed groups of
cone-shaped grains. As the RF power increased (100
W), the film surface improved and became smoother
and the grains became platelet elongated shaped. At a
high RF power (350 W), the film surface became
rough and had voids and the grain density increased.
This behavior could be attributed to the cleanliness of
the silicon substrate surface. At a low RF power (0–
30 W), the oxide layer on the silicon substrate surface Fig. 3. Dependence of XRD peak intensity of TiSi2 (311) plane on
was not completely removed, and the deposition was RF power. The peak intensity was normalized by Si (400) peak
not uniform. As the RF power increased (100 W), the intensity.
oxide layer on the silicon substrate surface was
effectively removed and the deposited film became was z 40 min. Also at lower substrate temperatures
uniform and smoother. At a high RF power (350 W), < 800 jC, the removal of the oxide layer was not
the silicon substrate surface was damaged by the effective and the deposited film grew in the form of
dense plasma as was observed from SEM images, separated particles.
and thus, the deposited film was not uniform and had XRD patterns of titanium silicide films deposited
voids. It is worth mentioning that voids were observed after substrate cleaning at different RF powers are
on Si substrate surface when the duration of cleaning shown in Fig. 2. We clearly see for all films the

Fig. 2. XRD patterns of the as-deposited TiSi2 films deposited on Si (100) substrate after in situ H2-plasma cleaning at different RF powers. (a) 0 W,
(b) 30 W, (c) 100 W and (d) 350 W.
2968 O.A. Fouad et al. / Materials Letters 57 (2003) 2965–2969

TiSi2 has two crystallographic structures, the ortho-


rhombic base-centered C49 of high resistivity
( f 80 –100 AV cm) and the orthorhombic face-cen-
tered C54 of low resistivity ( f 15 –25 AV cm). Fig. 4
shows the variation of the TiSi2 film resistivity as a
function of the RF power. As the RF power increased
from 0 to 100 W, the electrical resistivity of the TiSi2
film decreased from about 100 AV cm to reach a
minimum value of f 25 AV cm, respectively. This
resistivity is in agreement to the reported one of the
C54 phase [10]. The film resistivity is intimately
related to its morphology and thickness. The high
resistivity at low and high RF powers might be due to
Fig. 4. Dependence of electrical resistivity of TiSi2 film on RF rough surface, discontinuity of the film and uneven
power. thickness. At an RF power of 100 W, the film surface
was smooth and homogenous resulting in lowering
the electrical resistivity.
diffraction peaks corresponding to (111), (220), (311),
(040), (022) and (331) diffraction planes of C54 –
TiSi2 phase, respectively. It is obvious that the depos- 4. Conclusion
ited films are polycrystalline with the preferred (311)
orientation. The effect of in situ H2-plasma cleaning prior to
The XRD intensity of the (311) plane was normal- deposition on the structural and electrical properties of
ized by the intensity of Si (400) and shown as a TiSi2 films was investigated. It was found that H2-
function of RF power in Fig. 3. It is obvious that as plasma cleaning was an effective route to remove the
the RF power of H2-plasma cleaning increased, the oxide layer from the silicon substrate surface. How-
intensity of the (311) plane increased to reach a ever, careful change of H2-plasma condition was
maximum at an RF power of 100 W and then recommended to achieve the desired cleaning, avoid-
decreased as the RF power increased up to 350 W. ing surface damage and indeed rough film surface
This behavior could be attributed to the nature of the with high resistivity. The electrical resistivity could be
silicon substrate surface as observed from the AFM lowered to an acceptable value by depositing film
profiles. At a low RF power, the deposited film was not with smooth and homogenous surface.
uniform and discontinuous due to the lack of reaction
between TiCl4 and the residual SiO2. This discontinu-
ity led to low-intensity XRD peaks. As the RF power Acknowledgements
increased (100 W), the oxide layer was effectively
removed and the film deposited uniformly resulted in This work was partly supported by the Sasakawa
increasing the intensity of the XRD peak. At a high RF Scientific Research Grant (No. 13-346) from the
power (350 W), the film surface had voids that might Japan Science Society.
have affected the XRD peak intensities.
XPS analysis showed that the films were composed
mainly of Si and Ti. Oxygen and carbon contamina- References
tion was found on the film surfaces, which were easily
etched by Ar+ ion, ruling out the formation of any [1] D. Maury, J.L. Regolini, P. Gayet, Mater. Res. Soc. Symp.
oxide phase. This observation also suggested that Proc. 402 (1996) 283.
[2] J.S. Byun, S.J. Choi, J.J. Kim, J.T. Choi, J. Swenberg, R.
oxygen and carbon were only adsorbed on the film Achutharaman, J. Electrochem. Soc. 145 (1998) 3941.
surface due to exposure to air for a period of time [3] K. Saito, Y. Higashi, T. Amazawa, Y. Arita, J. Electrochem.
before XPS measurements were performed. Soc. 141 (1994) 1879.
O.A. Fouad et al. / Materials Letters 57 (2003) 2965–2969 2969

[4] G.J. Reynolds, C.B. Cooper III, P.J. Gaczi, J. Appl. Phys. 65 [8] Y. Ohshita, M. Oshida, M. Seki, K. Watanabe, J. Cryst.
(1989) 8. Growth 193 (1998) 322.
[5] J. Lee, R. Reif, J. Electrochem. Soc. 139 (1992) 1159. [9] O.A. Fouad, M. Yamazato, M. Era, M. Nagano, T. Hirai, I.
[6] V. Ilderem, R. Reif, Appl. Phys. Lett. 53 (1988) 687. Usui, J. Cryst. Growth 234 (2001) 440.
[7] K. Saito, Y. Arita, J. Electrochem. Soc. 143 (1996) 3778. [10] J. Lee, R. Reif, J. Electrochem. Soc. 139 (1992) 1166.

Вам также может понравиться